British Journal of Dermatology 2007 156, Pp1–10 1 2 Diversity of Mycosis Fungoides, D

January 2007 - Vol. 156 Issue 1,Page.1-201

Snippets

RESEARCH SNIPPETS

pages xiii–xiii

Review article

Mycosis fungoides: a dermatological masquerader

D. Nashan, D. Faulhaber, S. Ständer, T.A. Luger and R. Stadler pages 1–10

Guidelines

Guidelines for management of Bowen's disease: 2006 update

N.H. Cox, D.J. Eedy and C.A. Morton on behalf of the British Association of Dermatologists Therapy Guidelines and Audit Subcommittee pages 11–21

Original articles

Clinical and laboratory investigations

Temporal changes in sebum excretion and propionibacterial colonization in preadolescent children with and without acne

K. Mourelatos, E.A. Eady, W.J. Cunliffe, S.M. Clark and J.H. Cove pages 22–31 The role of heat shock protein 60, vascular endothelial growth factor and antiphospholipid antibodies in Behçet disease

O. Shaker, M.A. Ay El-Deen, H. El Hadidi, B.D. Grace, H. El Sherif and A. Abdel Halim pages 32–37

Assessment of the presence of mucosal human papillomaviruses in malignant melanomas using combined fluorescent in situ hybridization and chemiluminescent immunohistochemistry

S. Ambretti, S. Venturoli, M. Mirasoli, M. La Placa, F. Bonvicini, M. Cricca, M. Zerbini, A. Roda and M. Musiani pages 38–44

Skin surveillance of a U.K. paediatric transplant population

M.A. Thomson, N.R. Suggett, P.G. Nightingale, D.V. Milford, U. Baumann, D.A. Kelly, C. Moss and V.A. Hill pages 45–50

Altered innate and adaptive immune responses in patients with hidradenitis suppurativa

E.J. Giamarellos-Bourboulis, A. Antonopoulou, C. Petropoulou, M. Mouktaroudi, E. Spyridaki, F. Baziaka, A. Pelekanou, H. Giamarellou and N.G. Stavrianeas pages 51–56

Does adjuvant alpha-interferon improve outcome when combined with total skin irradiation for mycosis fungoides?

D. Roberge, T. Muanza, G. Blake, C. Shustik, T. Vuong and C.R. Freeman pages 57–61

Transcriptional profiles in melanocytes from clinically unaffected skin distinguish the neoplastic growth pattern in patients with melanoma

C. Magnoni, E. Tenedini, F. Ferrari, L. Benassi, C. Bernardi, G. Gualdi, G. Bertazzoni, E. Roncaglia, L. Fantoni, R. Manfredini, S. Bicciato, S. Ferrari, A. Giannetti and E. Tagliafico pages 62–71

Melanomas arising from naevi and de novo melanomas — does origin matter?

S.C. Weatherhead, M. Haniffa and C.M. Lawrence pages 72–76

Original articles

Contact dermatitis

Severity of hand eczema assessed by patients and dermatologist using a photographic guide

M. Hald, N.K. Veien, G. Laurberg and J.D. Johansen pages 77–80

Original articles

Dermatological surgery and lasers

The use of confocal laser-scanning microscopy in microsurgery for invasive squamous cell carcinoma

M. Horn, A. Gerger, S. Koller, W. Weger, U. Langsenlehner, P. Krippl, H. Kerl, H. Samonigg and J. Smolle pages 81–84

Original articles

Dermatopathology

Effect of smoking on skin elastic fibres: morphometric and immunohistochemical analysis

M. Just, M. Ribera, E. Monsó, J.C. Lorenzo and C. Ferrándiz pages 85–91

Characterization of the expression and activation of the epidermal growth factor receptor in squamous cell carcinoma of the skin

G.B. Fogarty, N.M. Conus, J. Chu and G. McArthur pages 92–98

Original articles

Epidemiology and health services research

The risk for cutaneous malignant melanoma, melanoma in situ and intraocular malignant melanoma in relation to tobacco use and body mass index

Å. Odenbro, P. Gillgren, R. Bellocco, P. Boffetta, N. Håkansson and J. Adami pages 99–105

Characteristics and completeness of clinical trial registrations in psoriasis and atopic dermatitis

R.D. Quain and K.A. Katz pages 106–110

Validation of the U.K. Working Party diagnostic criteria for atopic eczema in a Xhosa-speaking African population

D.A. Chalmers, G. Todd, N. Saxe, J.T. Milne, S. Tolosana, P.N. Ngcelwane, B.N. Hlaba, L.N. Mngomeni, T.G. Nonxuba and H.C. Williams pages 111–116

Original articles

Photobiology

Photodegradation of folic acid during extracorporeal photopheresis

M. Der-Petrossian, M. Födinger, R. Knobler, H. Hönigsmann and F. Trautinger pages 117–121

Effects of psoralen plus ultraviolet A irradiation on cultured epidermal cells in vitro and patients with vitiligo in vivo

C-S. Wu, C-C.E. Lan, L-F. Wang, G-S. Chen, C-S. Wu and H-S. Yu pages 122–129

Original articles

Therapeutics

A double-blind, randomized quantitative comparison of calcitriol ointment and calcipotriol ointment on epidermal cell populations, proliferation and differentiation

J.E.M. Körver, W.H.P.M. Vissers, D.W.A. Van Rens, M.C. Pasch, P.E.J. Van Erp, J.B.M. Boezeman and P.C.M. Van De Kerkhof pages 130–137

Comparison of clinical and pharmacokinetic profiles of etanercept 25 mg twice weekly and 50 mg once weekly in patients with psoriasis

B. Elewski, C. Leonardi, A.B. Gottlieb, B.E. Strober, M.A. Simiens, M. Dunn and A. Jahreis pages 138–142

Case reports

Nephro-urological complications of epidermolysis bullosa in paediatric patients

S.M.H. Chan, M.J. Dillon, P.G. Duffy and D.J. Atherton pages 143–147

CD30+ large cell transformation of mycosis fungoides after psoralen plus ultraviolet A photochemotherapy

J. Ogino, K. Saga, M. Kagaya, A. Kamada, K. Hirosaki, R. Kaneko and K. Jimbow pages 148–151

A failure of mucocutaneous lymphangiogenesis may underlie the clinical features of lipoid proteinosis

T. Uchida, H. Hayashi, M. Inaoki, T. Miyamoto and W. Fujimoto pages 152–157

Gene corner

MSH6 mutation in Muir–Torre syndrome: could this be a rare finding?

E. Mangold N. Rahner N. Friedrichs R. Buettner C. Pagenstecher S. Aretz W. Friedl T. Ruzicka P. Propping A. Rütten and R. Kruse pages 158–162

Correspondence

Painless periungual pyogenic granulomata associated with reverse transcriptase inhibitor therapy in a patient with human immunodeficiency virus infection

L.H. Williams and P. Fleckman pages 163–164

Melanomas in renal transplant recipients: the London experience, and invitation to participate in a European study*

V.L. Brown, R.N. Matin, R. Cerio, M.E. Leedham-Green, C.M. Proby and C.A. Harwood pages 165–167

Melanomas in renal transplant recipients: the London experience, and invitation to participate in a European study: reply from authors

B. Imko-Walczuk, A. Lally, L. Le Mire, D. Casabonne, K. Hollowood, C. Bordea, F. Wojnarowska pages 167–169 Clinical utility of an interferon-γ-based assay for mycobacterial detection in papulonecrotic tuberculid

R. Tanaka, H. Matsuura, Y. Kobashi and W. Fujimoto pages 169–171

T cell/keratinocyte interactions in psoriasis: where is the trigger?

T. Simonart and M. Heenen pages 171–172

Dapsone in the management of 'insect bite-like reaction' in a patient with chronic lymphocytic leukaemia

A. Ulmer, G. Metzler, S. Schanz and G. Fierlbeck pages 172–174

Indurated nodules and plaques showing a dense plasma cell infiltrate as a cutaneous manifestation of Castleman's disease

R. Okuyama, H. Harigae, T. Moriya, S. Kagatani, H. Tagami, R. Ichinohasama and S. Aiba pages 174–176

Selenium intoxication: undesirable effect of a fasting cure

B. Schuh and U. Jappe pages 177–178

Differential clinical response to alefacept in combination with methotrexate in two patients with refractory palmar psoriasis

A. Jacobi, G. Schuler and M. Hertl pages 178–180

Cutaneous horn can be a clinical manifestation of underlying sebaceous carcinoma

H. Kitagawa, M. Mizuno, Y. Nakamura, I. Kurokawa and H. Mizutani pages 180–182

Pseudosystemic sclerosis as a complication of recombinant human interleukin 2 (aldesleukin) therapy

I. Marie, P. Joly, P. Courville and H. Levesqsue pages 182–183

A case of sporadic Bazex–Dupré–Christol syndrome presenting with scarring folliculitis of the scalp

T. Gambichler, S. Hoffjan, P. Altmeyer and F.G. Bechara pages 184–186

Moral and cost dilemma of a psoriasis patient

S. Abdul Ghaffar and A.Y. Finlay pages 186–187

Reiter syndrome triggered by adalimumab (Humira®) and leflunomide (Arava®) in a patient with ankylosing spondylarthropathy and Crohn disease

A.M. Thielen, C. Barde, V. Janer, L. Borradori and J.H. Saurat pages 188–189

Efalizumab-induced aseptic meningitis

N. Kluger, C. Girard, V. Gonzalez, B. Guillot and D. Bessis pages 189–191

Tacrolimus vs. clobetasol propionate in the treatment of facial cutaneous lupus erythematosus: a randomized, double-blind, bilateral comparison study

T-Y. Tzung, Y-S. Liu, H-W. Chang pages 191–192

'Mechanic's hands': a misleading cutaneous sign of the antisynthetase syndrome

C. Bachmeyer, I. Tillie-Leblond, A. Lacert, J. Cadranel and S. Aractingi pages 192–194

A successful therapeutic trial of rituximab in the treatment of a patient with recalcitrant, high-titre epidermolysis bullosa acquisita

S.M. Crichlow, N.J. Mortimer and K.E. Harman pages 194–196

Spotted and rippled reticulate hypermelanosis: a possible variant of Dowling–Degos disease

N. Oiso, D. Tsuruta, T. Ota, H. Kobayashi and A. Kawada pages 196–198

Primary cutaneous follicular B-cell lymphoma arising at the site of radiotherapy for breast cancer

C. Bachmeyer K. Khosrotehrani P. Moguelet and S. Aractingi pages 198–199

Additional dermoscopic presentation of haemosiderotic dermatofibroma

R. Cardoso C. Massone H.P. Soyer and R. Hofmann-Wellenhof pages 199–200

News and Notices

News and Notices pages 201–201

REVIEW ARTICLE DOI 10.1111/j.1365-2133.2006.07526.x Mycosis fungoides: a dermatological masquerader D. Nashan, D. Faulhaber,* S. Sta¨nder,* T.A. Luger* and R. Stadler Department of Dermatology, University of Freiburg, Hautstr. 7, 79104 Freiburg, Germany *Department of Dermatology, University of Mu¨nster, Mu¨nster, Germany Department of Dermatology, Klinikum Minden, Minden, Germany

Summary

Correspondence Mycosis fungoides (MF), a low-grade lymphoproliferative disorder, is the most D. Nashan. common type of cutaneous T-cell lymphoma. Typically, neoplastic T cells localize E-mail: [email protected] to the skin and produce patches, plaques, tumours or erythroderma. Diagnosis of MF can be difficult due to highly variable presentations and the sometimes non- Accepted for publication 8 June 2006 specific nature of histological findings. Molecular biology has improved the diagnostic accuracy. Nevertheless, clinical experience is of substantial importance as Key words MF can resemble a wide variety of skin diseases. We performed a literature clinical subtypes, differential diagnoses, mycosis review and found that MF can mimic >50 different clinical entities. We present fungoides, overview a structured framework of clinical variations of classical, unusual and distinct Conflicts of interest forms of MF. Distinct subforms such as ichthyotic MF, adnexotropic (including None declared. syringotropic and folliculotropic) MF, MF with follicular mucinosis, granulomatous MF with granulomatous slack skin and papuloerythroderma of Ofuji are delineated in more detail.

Mycosis fungoides (MF), a low-grade lymphoproliferative dis- fungoides’ with ‘differential diagnosis’ and ‘clinical picture’, order, is the most common type of cutaneous T-cell lymph- and ‘mycosis fungoides’ and ‘cutaneous T-cell lymphoma’ in oma. Typically, neoplastic T cells localize to the skin and conjunction with clinically descriptive adjectives. From extrac- produce patches, plaques, tumours or erythroderma. Diagnosis ted articles the related articles and publishing authors were of MF can be difficult due to highly variable presentations and also screened. Further original articles were extracted from ref- the sometimes nonspecific nature of histological findings. erence lists. Molecular biology has improved the diagnostic accuracy. Nev- We thereby present a review of all currently published clin- ertheless, clinical experience is of substantial importance as ical pictures of MF simulating other dermatoses and distinctive MF can resemble a wide variety of skin diseases. clinicopathological features which are in part considered sep- Diagnosis of MF is based on a combination of clinical pre- arately in the new World Health Organization (WHO)—Euro- sentation, histopathology and gene rearrangement.1 None of pean Organization for Research and Treatment of Cancer these factors exclusively determines the diagnosis. Histologi- (EORTC) classification.6 The illustration of MF variants follows cally, MF is characterized by the presence of large atypical in Tables according to clinical signs and under the headings of lymphocytes, a lymphocytic infiltrate in the papillary dermis more distinct subtypes. and thickened collagen fibres. However, in early MF not all of these pathological findings are present and distinction from an 2 Clinically and morphologically unusual inflammatory infiltrate is often difficult. Detection of a mono- variations of mycosis fungoides clonal T-cell infiltrate is not lymphoma specific. Positive polymerase chain reaction (PCR) results are also found in diseases In early stages of MF (T1N0M0 or T2N0M0) characteristic le- such as psoriasis, pityriasis lichenoides et varioliformis acuta sions consist of erythematous macules or papules, which are (PLEVA) and lichen ruber.3,4 Thus clinical presentation is a primarily superficial and resemble an ‘eczema’ with sharply major factor determining the diagnosis.5 defined borders. Often some degree of scaling is observed, similar to psoriasis. The edges of the lesions might exhibit in- Search criteria creased scaling, corresponding to a growing infiltrate. The configuration can be arciform, annular, semiannular, serpingi- This article emerged from Medline searches, manual searches nous or polycyclic.7 The skin surface can be atrophic, exhibit- in dermatological journals and textbooks, and from personal ing wrinkles. The colour can be orange to bright red or experiences. Electronic key word searches included ‘mycosis can present livid or brown-red components. Spontaneous

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 1 2 Diversity of mycosis fungoides, D. Nashan et al.

Table 1 Differential diagnoses of classical mycosis fungoides (MF) are The descriptive term ‘mycosis fungoides’, chosen in 1806 assembled based on a predominant clinical sign. A representative by Alibert, already suggests the first differential diagnosis of publication with a clinically imitative MF is given for each differential tinea corporis for a typical MF lesion. The coincidence of der- diagnosis matophyte infections and MF has been described.9,10 Further descriptive terms provide hints for the differential diagnosis: Clinical sign Differential diagnosis First author lichenification, lichenoid, eczematous, seborrhoeic, urticarial, Eczematous Seborrhoeic eczema Figure 1 erythematous, hypopigmented, pityriasis-like.11 Thus the dif- 82 Perioral dermatitis Wolf 1992 ferential diagnosis includes nonspecific eczema, nummular Palmoplantar eczema Spieth 200219 eczema, seborrhoeic eczema, contact dermatitis, atopic derma- Dyshidrotic eczema Kempf 200580 Contact dermatitis Spieth 200219 titis, psoriasis, drug reaction, lichen simplex chronicus and 12,13 Atopic eczema Kazakov 200412 lichen planus (Table 1). Scaling Psoriasis Zackheim 200283 Limited involvement of the skin, especially as unilesional Psoriasis palmaris Spieth 200219 typical MF, is not uncommon.14–17 A well-documented, Psoriasis plantaris Figure 2 although not universally accepted, solitary form is the paget- 84 Parapsoriasis Ackermann 1996 oid reticulosis type Woringer–Kolopp, characterized by a sin- Tinea corporis Chaves 200285 86 gle, scaling and erythematous MF lesion with acral Tinea pedis Resnik 1995 18 Erythematous Erythema multiforme Krebs 197887 localization. A challenge for the clinician might be the Kazakov 200412 appearance of typical MF in an atypical localization, e.g. MF Annular erythema Lim 200388 simulating palmoplantar or periorificial eczema, isolated alope- Cogrel 200589 cia or affecting the mucosa (Figs 1 and 2).19,20 45 Alopecia Alopecia areata Burg 1992 At the tumour stage IIB (T3N0/1M0) nodules of various sizes are found. They can be flat or dome-shaped. Their colour For a further overview publications of Zackheim and McCal- mont13—with 23 differential diagnoses—and Kazakov et al.12 is yellow-red or red-blue to brown. The lesions are more can be recommended. prominent and deeper than plaques. MF tumours mainly develop in the course of the disease or in conjunction with eczematous lesions. The surface is soft on palpation, which regression of lesions can occur; this is sometimes limited to distinguishes the lesions from metastases of solid carcinomas. the centre of the lesion. Alopecia may develop in lesional sites Ulcerations are frequently seen, which can develop secondary or even in clinically unaffected skin. The lesions are asymmet- infections. Tumours develop in either pre-existing MF lesions rically distributed and are predominantly located in a ‘swim- or de novo. The location and configuration of these tumours are suit’ distribution, i.e. preferentially on the abdomen, hips, comparable with B-cell lymphoma. Another tumorous lymph- buttocks and breasts. Lesions are also seen on the medial sides oma and possible differential diagnosis is lymphomatoid papu- of proximal extremities. The mucosa can be affected at any losis (LyP). In 1968, Macaulay21 summarized LyP as a stage of disease.8 It is still under discussion whether parapso- chronically recurring papulonodular dermatosis, sometimes riasis with its scattered finger-like striped lesions on the flanks self-limiting. It can be excluded by its brownish-reddish cen- is an entity on its own or is a precursor of MF. trally necrotic papules, spontaneous healing within weeks and

Fig 1. Mycosis fungoides imitating seborrhoeic eczema.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 Diversity of mycosis fungoides, D. Nashan et al. 3

Table 3 Differential diagnoses of erythrodermic mycosis fungoides

Clinical sign Differential diagnosis Erythrodermic Se´zary syndrome Adult T-cell leukaemia Actinic reticuloid Atopic dermatitis Drug reaction Psoriasis Age-dependent erythroderma

leukaemic involvement’ describe the difficulty of either delineation. SS was defined by Se´zary and Bouvrain in 193823 as a triad of erythroderma, lymphadenopathy and atypically large mononuclear blood cells. It is clinically associated with leonine facies, ectropion, alopecia of the scalp, palmoplantar hyperkeratosis and fissures and dystrophy of the nails. Similar but less frequent monosymptomatic forms or new phenomena such as a vitiligo-like leucoderma with MF and SS hamper the diagnosis.24–26 A diagnostic discrimination is presented by the Interna- tional Society for Cutaneous Lymphomas, specifying in addition to erythrodermic MF and SS, ‘other erythrodermic cutaneous T-cell lymphoma, not otherwise defined’.27 Fur- ther differential diagnoses include primary and secondary erythrodermas caused by psoriasis, atopic dermatitis, drug eruption or age-dependent erythroderma (Table 3). Actinic reticuloid sometimes exacerbates to erythroderma; however, Fig 2. Mycosis fungoides imitating palmoplantar psoriasis. it becomes aggravated in sunlight-exposed skin. It is considered as a form of a cutaneous T-cell pseudolymphoma by most authors.28 Table 2 Differential diagnoses of tumorous mycosis fungoides lesions A real clinical challenge is provided by unusual and newly described variants of MF comprising hyperpigmented, hypopigmented, urticarial, bullous, solely papular, pustular and Clinical sign Differential diagnosis hyperkeratotic variants for which diagnosis is easier in con- Tumorous Lupus erythematosus junction with typical MF lesions or a positive history of MF B-cell lymphoma 29–31 Lymphomatoid papulosis (Table 4). Hypopigmentation and a vitiligo-like outcome CD30+ lymphoma are predominantly described in individuals with dark skin. Hodgkin disease Mainly asymptomatic, irregularly confined, white macules are Sarcoma seen.32,33 A critical review suggested that only 19 of 106 published cases with hypopigmentation were truly MF.34 The bullous form of MF was first described by Kaposi in 1887. Flaccid or tense, often multiple or even generalized blisters a residual stage of hyperpigmented macules or atrophic scars. appear on normal skin or within plaques. A positive Nikolsky Awareness is required as the coexistence of LyP and MF is not sign is observed. Diagnosis is urgent as bullous lesions of MF rare.22 MF tumours have to be differentiated from secondary indicate a poor prognosis.35 lymphomatoid diseases such as CD30+ anaplastic T-cell Some case reports point out single atypical lesions as a clue lymphoma and Hodgkin disease (Table 2). for a false diagnosis, e.g. a warty lesion misdiagnosed as a se- Erythrodermic MF, stage III (T4N0/1M0), is defined by borrhoeic keratosis or Bowen’s disease.36,37 Other clinical erythroderma, a red to red-bluish colour, affecting the whole impressions of atypical MF have stimulated diagnoses such as body surface sometimes sparing small areas of normal skin. purpura pigmentosa, vasculitis and pyoderma gangrenosum Scaling can range from little exfoliation to severe desquama- (Table 5).38 Especially purpuric eruptions are repeatedly de- tion. scribed with petechial patches and varying degrees of epider- The erythroderma of Se´zary syndrome (SS) is closely mal changes—scaling, vesiculation, lichenification—and related. Definitions such as ‘SS preceded by MF’ or ‘MF with brownish pigmentation from haemosiderin accumulation.33

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 4 Diversity of mycosis fungoides, D. Nashan et al.

Table 4 Differential diagnoses of clinical Clinical sign Differential diagnosis First author variants of mycosis fungoides (MF) with a 12 lower incidence than the classical pictures. Hypopigmented Pityriasis versicolor Kazakov 2004 Authors who published MF cases with the Piyriasis alba Whitmore 199491 11 according clinical differential diagnosis are Vitiligo Ardigo 2003 given in column 3. Clinical pictures are 12 Leprosy Kazakov 2004 76 12 published in overviews by Goerdt et al. , Postinﬂammatory hypopigmentation Kazakov 2004 Howard and Smoller90, Zackheim and 92 Hyperpigmented Acanthosis nigricans Willemze 1985 McCalmont13, Kazakov et al.12 and Kotz et al.81 Ashy dermatosis Kazakov 200412 Hypo-hyperpigmented Poikiloderma (vasculare atrophicans) Wain 200593 Bullous/vesicular Bullous autoimmune dermatosis Ho 200094 Pemphigus vulgaris Roenigk 197195 Hyperkeratotic Palmoplantar hyperkeratosis Goldberg 199796 Verrucae vulgaris Goldberg 199796 Keratosis lichenoides chronica Bahadoran 199897 Porokeratosis of Mibelli Breneman 199398 Papular Drug-induced pseudolymphoma Marzano 199914 Lymphomatoid papulosis Kodama 200531 Pustular Palmoplantar pustulosis Moreno 199099 Generalized pustulosis Camisa 1994100 Single lesions Seborrhoeic keratosis Bazza 200236 Bowen’s disease Yoo 200337 Eryipelas Brill 2005101

Table 5 Mycosis fungoides (MF) can imitate more complex Table 6 Distinct subtypes of mycosis fungoides (MF) dermatological disorders which are listed under differential diagnosis. Authors debating these differential diagnoses in conjunction with MF Distinct MF entity First author are named Ichthyotic MF Eisman 2003110 Ku¨tting 199641 Classification Differential diagnosis First author Badawy 200242 Mucous membrane Mucositis Wain 20038 Adnexotropic MF van Doorn 200248 Vessels Lichen aureus, capillaritis Ameen 2000102 Granulomatous MF Goerdt 199676 Pyoderma gangrenosum Ho 200094 Granulomatous slack skin LeBoit 199465 Carbia 200238 von Haselen 1998111 Gangrene Lund 1990103 Topar 2001112 Purpura pigmentosa Barnhill 198833 MF with follicular mucinosis Vollmer 200252 Cather 1998104 Papuloerythroderma of Ofuji Hur 200258 Lymphoma, Actinic reticuloid Neil 198528 Pereiro 200373 leukaemia, Sarcoma Machler 1994105 tumour Others Ischaemic foot Goldstein 1999106 ‘Invisible dermatosis’ Pujol 2000107 Ichthyotic mycosis fungoides 108 Hwong 2001 MF may be mistaken for an ichthyosis. Clinical aspects of an Dissecting cellulitis Gilliam 1997109 ichthyotic MF more often reveal extensive skin involvement of the scalp (Fig. 3).40 Typical dry scaling affects the trunk and extremities. Sometimes a mild erythroderma-like condition may develop.41,42 Clinically obvious follicular keratosis, comedo- like lesions and epidermal cysts are mentioned as additional More definite clinicopathological findings distinguishing cuta- manifestations.40,42 A coexpression of ichthyotic and follicular neous lymphoma from pigmented purpuric dermatosis are MF has been published.43 In MF, in contrast to other lympho- outlined by Martıńez et al.39 mas, the ichthyosis is not paraneoplastic but is self-defining and has a fairly good prognosis.43 Distinct entities of mycosis fungoides Adnexotropic mycosis fungoides More clear-cut features lead to the diagnosis of ichthyotic MF, different adnexotropic forms of MF, granulomatous forms of This is mainly misdiagnosed as acneiform lesions and alopecia. MF, MF with follicular mucinosis and latterly also papulo- Adnexotropic MF can be itemized into subtypes for which dif- erythroderma of Ofuji (Table 6). ferential diagnoses are given (Table 7).

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 Diversity of mycosis fungoides, D. Nashan et al. 5

hyperaesthesia of the affected skin area and anhidrosis are possible adjuncts. Eight of 15 published cases of SLHA were clas- siﬁed as syringotropic cutaneous lymphoma, for which 13 further cases are now known.46 Besides clinical similarities to SLHA, erythema punctata, sometimes with follicular accentu- ation and milia, is described for the syringotropic form. The facial localization is reminiscent of discoid lupus erythematosus. The typical histology shows a dense syringotropic lymphocytic inﬁltrate sometimes surrounding hyperplastic eccrine glands and eccrine ducts. Generally considered a variant of MF, both have a similarly good prognosis.47

Folliculotropic mycosis fungoides

Folliculotropic MF, or follicular MF, is predominantly localized on the head and neck (Fig. 4). Clinically it presents follicular papules (often grouped), alopecia and acneiform lesions.48,49 Patients often complain about severe pruritus.48 Even pseudotumorous forms with nodules have been described, with a marked follicular hyperplasia rather than a lymphocytic proliferation.50 Otherwise perifollicular pleomorphic infiltrates show various degrees of folliculotropism without epidermotropism. Follicular MF is sometimes is associated with follicular mucinosis. In 2002, 51 cases of follicular MF were recorded in the Netherlands.48 Diagnostic accuracy and adap- ted therapeutic regimens seem necessary as follicular MF might be less responsive to standard treatment modalities: this is a possible explanation for disease progression in the Dutch collection.6,48,51 Fig 3. Ichthyotic form of mycosis fungoides. Most cases of folliculotropic MF show mucinous degeneration of the hair follicles and are traditionally designated as MF-associated follicular mucinosis. However, this is not a pre- Syringolymphoid hyperplasia (SLHA) was first described by requisite and, with or without associated follicular mucinosis, Sarkany in 1969.44 Published as lymphomatoid granulomato- these forms are named ‘follicular MF’ or ‘folliculotropic MF’. sis, its characteristics were alopecia, anhidrosis, hypertrophy The difficulty of a diagnostic assignment of follicular mucino- of eccrine glands and vasculitis. In 1992 Burg and Schmo¨c- sis towards MF or of delineation as a primary and exclusive kel45 revealed an association with cutaneous T-cell lymphoma. clinical manifestation is reflected in ongoing discussions.52–57 The clinical presentation is variable, and includes solitary or The differentiation cannot be assisted by PCR of the T-cell re- multiple patches and plaques or skin-coloured to reddish pa- ceptor, as the primary or idiopathic follicular mucinosis might pules, with more extensive skin involvement resembling fol- also exhibit monoclonal T-cell populations. Furthermore, follicular hyperkeratosis, poikiloderma atrophicans vasculare, licular mucinosis has been observed in association with erythroderma and even palmoplantar hyperkeratosis. Hair loss, other MF subforms, such as the adnexotropic form, already

Table 7 Differential diagnoses of adnexotropic forms of mycosis fungoides Clinical sign Differential diagnosis First author Adnexotropic Rosacea Sherertz 1986113 Eczema Bonta 2000114 Seborrhoeic dermatitis van Doorn 200248 Comedones, cysts Oliwiecki 1992115 Epidermal cysts Lacour 1993116 Comedones, cysts, alopecia Peris 1999117 Follicular hyperkeratosis Klempke 199951 Poikiloderma vasculare atrophicans Brecher 2002118 Idiopathic syringolymphoid hyperplasia Kazakov 200412 Discoid lupus erythematosus Thein 200446

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 6 Diversity of mycosis fungoides, D. Nashan et al.

Fig 4. Adnexotropic form of mycosis fungoides.

Fig 5. Mycosis fungoides with granulomatous slack skin. characterized, and papuloerythroderma of Ofuji for which was included in the EORTC64 and then the WHO–EORTC clas- clinical description and differential diagnoses will follow.58 sification6 as a provisional entity. So far up to 42 cases have The coexistence of syringotropic and folliculotropic cutane- been reported. Granulomatous T-cell infiltrates and loss of ous lymphomas also results in acneiform manifestations with elastic fibres lead to asymptomatic skin wrinkles, mainly in follicular papules, comedones, milia, cysts, dry skin and alope- the flexures (Fig. 5).65 Sometimes pruritus and erythema are cia.47 Some authors suggest inclusion of these clinical variants, described. Granulomatous MF may be preceded by erythema- as well as other pilotropic forms of MF, under the term ‘ad- tous scaling patches or macules, and may coexist with classical nexotropic cutaneous lymphomas’.59,60 MF lesions. Benign as well as progressive courses were described. In 50% of cases coexistence with Hodgkin disease or Granulomatous mycosis fungoides and nodal non-Hodgkin disease is observed. granulomatous slack skin From a histological point of view it is discussed whether granulomatous MF and granulomatous slack skin belong Ackerman and Flaxman61 proposed the term ‘granulomatous together.66,67 The pathogenesis of granuloma formation in MF’ in 1970. In 1978 Ackerman62 added the term ‘granulo- lymphoma is unknown and its occurrence is not specific for matous slack skin disease’. In 1968 Bazex et al.63 described a MF. Similar histological formations have been described in clinical form similar to cutis laxa as ‘Besnier–Boeck–Schau- pleomorphic T-cell lymphoma, panniculitis-like T-cell lymph- mann disease’. In 1997 and 2005 ‘granulomatous slack skin’ oma and in B-cell lymphomas.68 A case report of a CD30+

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 Diversity of mycosis fungoides, D. Nashan et al. 7

Table 8 Differential diagnosis of granulomatous forms of mycosis the diagnosis can only be confirmed during the course of dis- fungoides ease.4 Whole-body inspection often allows identification of typical lesions besides atypical lesions, thus guiding towards Clinical sign Differential diagnosis First author the correct diagnosis. Taking all information together and add- Granulomatous Granuloma annulare Jouary 200270 ing histology, differential diagnoses are no longer a problem. Rosacea Sherertz 1986113 Once a lymphoma has been diagnosed a consequent follow- 71 Sarcoidosis Bessis 1996 up is also required to detect possible recurrence or manifest- 119 Necrobiosis Woollons 1999 ation of other associated conditions. Skin atrophy Cutis laxa van Haselen 1998111 We end with a statement by Zackheim and McCalmont13 who called MF ‘the great imitator’. They compared the cha- meleon-like diversity in clinical presentation of MF with the variability of clinical conditions seen in syphilis. lymphoma with granulomatous slack skin exists, so not even granulomatous slack skin is a specific subform of MF.69 There- Acknowledgments fore clinically and histologically granulomatous MF variants have to be differentiated from other cutaneous T-cell lympho- We thank Professor Rein Willemze for revision and constructive mas and from sarcoidosis, granuloma annulare and infectious advice. granulomas (Table 8).70,71 The so-called ‘sarcoidosis–lymphoma syndrome’ includes the association of sarcoidosis and References lymphoma, among other entities described in association with MF.72 1 Willemze R, Meijer CJ. Rationale of a new classification for the group of primary cutaneous lymphomas. Semin Cutan Med Surg 2000; 19:71–7. Papuloerythroderma of Ofuji 2 Naraghi ZS, Seirafi H, Vakikhani M et al. Assessment of histologic criteria in the diagnosis of mycosis fungoides. Int J Dermatol 2003; Is papuloerythroderma of Ofuji no more a differential diagno- 42:45–52. 58,73 sis but a precursor of MF? Papuloerythroderma of Ofuji 3 Dereure O, Levi E, Kadin ME. T-cell clonality in pityriasis li- 74 was first described in 1984. It is characterized by red- chenoides et varioliformis acuta: a heteroduplex analysis of 20 brownish, partly confluent flat papules; the flexors remain cases. Arch Dermatol 2000; 136:1483–6. unaffected, and this is described as a ‘deckchair distribution’. 4 Burg G, Kempf W, Dummer R. Diagnostic signs of cutaneous Reports describe the association of papuloerythroderma of Of- lymphomas. J Eur Acad Dermatol Venereol 2001; 15:358–9. 5 Stadler R. Treatment of cutaneous T cell lymphoma. Skin Pharmacol uji and cutaneous lymphoma as well as the development of a Appl Skin Physiol 2002; 15:139–6. cutaneous lymphoma after primary diagnosis of papuloeryth- 6 Willemze R, Jaffe ES, Burg G et al. WHO–EORTC classification for 75 roderma of Ofuji. Case reports demonstrate the similarity to cutaneous lymphomas. Blood 2005; 105:3768–85. MF with eosinophilia, with elevated serum IgE and similar 7 Saada D, Lami MD, Vabres P et al. Mycosis fungoides presenting histological and molecular biological criteria. as annular erythema. Ann Dermatol Venereol 2005; 132:35–7. 8 Wain EM, Setterfield J, Judge MR et al. Mycosis fungoides involving the oral mucosa in a child. Clin Exp Dermatol 2003; 28:499–501. Distinguishing diagnoses 9 Hubert JN, Callen JP. Recalcitrant tinea corporis as the presenting manifestation of patch-stage mycosis fungoides. Cutis 2003; Based on clinical evaluation, the progress of MF and trans- 71:59–61. formation into other lymphoma subtypes can be assessed. The 10 Capella GL, Altomare GF. Mycosis on mycosis fungoides: zoophi- coexistence of different primary cutaneous lymphomas and lic dermatophytosis selectively superimposed on pre-existing cu- secondary cutaneous lymphomas presents a challenge for dif- taneous T-cell lymphoma (mycosis fungoides) plaques. Mycoses ferential diagnosis.76 Coexisting dermatological conditions 2003; 46:67–70. such as psoriasis or neurofibromatosis and atopic dermatitis 11 Ardigo M, Borroni G, Muscardin L et al. Hypopigmented mycosis fungoides in Caucasian patients: a clinicopathologic study of 7 also have to be taken into account.77–79 Kempf et al. published cases. J Am Acad Dermatol 2003; 49:264–70. a review on paraneoplastic skin reactions including the sign of 12 Kazakov DV, Burg G, Kempf W. Clinicopathological spectrum of 80 Leser–Tre´lat or PLEVA with MF. mycosis fungoides. J Eur Acad Dermatol Venereol 2004; 18:397–415. 13 Zackheim HS, McCalmont TH. Mycosis fungoides: the great imitator. J Am Acad Dermatol 2002; 47:914–18. Conclusion 14 Marzano AV, Berti E, Loredana L, Alessi E. Unilesional follicular Knowledge of cutaneous lymphoma is advancing.64,81 Accord- mycosis fungoides. Dermatology 1999; 199:174–6. 15 Oliver GF, Winkelmann RK. Unilesional mycosis fungoides: a dis- ing to Kotz et al.,81 ‘in every chronic disease resistant to treat- tinct entity. J Am Acad Dermatol 1989; 20:63–70. ment lymphoma has to be considered’. Therefore, if 16 Heald PW, Glusac E. Unilesional cutaneous T-cell lymphoma: lymphoma is suspected then the diagnosis must be confirmed clinical features, therapy, and follow-up of 10 patients with a by clinical, histological and molecular examinations perhaps treatment-responsive mycosis fungoides variant. J Am Acad Dermatol repeated in the course of the disease. In about 10% of cases 2000; 42:283–5.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 8 Diversity of mycosis fungoides, D. Nashan et al.

17 Hodak E, Phenig E, Amichai B et al. Unilesional mycosis fungoid- 40 Marzano AV, Borghi A, Facchetti M, Alessi E. Ichthyosiform my- es. Dermatology 2000; 201:300–6. cosis fungoides. Dermatology 2002; 204:124–9. 18 Steffen C. Ketron–Goodman disease, Woringer–Kolopp disease, 41 Ku¨tting B, Metze D, Luger TA, Bonsmann G. Mycosis fungoides and pagetoid reticulosis. Am J Dermatopathol 2005; 27:68–85. presenting as acquired ichthyosis. J Am Acad Dermatol 1996; 19 Spieth K, Grundmann-Kollmann M, Runne U et al. Mycosis-fun- 34:887–9. goides-type cutaneous T cell lymphoma of the hands and soles: a 42 Badawy E, D’Incan M, Majjaoui SE et al. Ichthyosiform mycosis variant causing delay in diagnosis and adequate treatment of fungoides. Eur J Dermatol 2002; 12:594–6. patients with palmoplantar eczema. Dermatology 2002; 205:239–44. 43 Hodak E, Amitay I, Feinmesser M et al. Ichthyosiform mycosis 20 Lippert BM, Hoft S, Teymoortash A et al. Mycosis fungoides of fungoides: an atypical variant of cutaneous T-cell lymphoma. J the larynx: case report and review of the literature. Otolaryngol Pol Am Acad Dermatol 2004; 50:368–74. 2002; 56:661–7. 44 Sarkany I. Patchy alopecia, anhidrosis, eccrine gland wall hyper- 21 Macaulay WL. Lymphomatoid papulosis. A continuing self-heal- trophy and vasculitis. Proc R Soc Med 1969; 62:157–9. ing eruption, clinically benign—histologically malignant. Arch Der- 45 Burg G, Schmo¨ckel C. Syringolymphoid hyperplasia with alope- matol 1968; 97:23–30. cia—a syringotropic cutaneous T-cell lymphoma? Dermatology 22 Gallardo F, Costa C, Bellosillo B et al. Lymphomatoid papulosis as- 1992; 184:306–7. sociated with mycosis fungoides: clinicopathological and molecu- 46 Thein M, Ravat F, Orchard G et al. Syringotropic cutaneous T-cell lar studies of 12 cases. Acta Derm Venereol (Stockh) 2004; 84:463–8. lymphoma: an immunophenotypic and genotypic study of five 23 Se´zary A, Bouvrain Y. E´rythrodermie avec pre´sence de cellules cases. Br J Dermatol 2004; 151:216–26. monstreuses dans la derme et le sang circulant. Bull Soc Fr Dermatol 47 Weng AAH, Howatson SR, Goodlad JR et al. Erythrodermic syringo- Syphiligr 1938; 45:254. tropic cutaneous T-cell lymphoma. Br J Dermatol 2003; 148:349–52. 24 Vocks E, Hertenberger B, Worret WI et al. An unusual cutaneous 48 van Doorn R, Scheffer E, Willemze R. Follicular mycosis fungoid- T cell lymphoma presenting as leonine facies. Eur J Dermatol 2000; es, a distinct disease entity with or without associated follicular 10:309–12. mucinosis: a clinicopathologic and follow-up study of 51 25 Tang M, Tan SH, Lim LC. Leukoderma associated with Se´zary syn- patients. Arch Dermatol 2002; 138:191–8. drome: a rare presentation. J Am Acad Dermatol 2003; 49:S247–9. 49 Roeschm A, Schleyer V, Landthaler M, Vogt T. Follicular mycosis 26 Bouloc A, Grange F, Delfau-Larue MH et al. Leucoderma associ- fungoides: variability of a rare entity. Skinmed 2005; 4:12–17. ated with flares of erythrodermic cutaneous T-cell lymphomas: 50 Kossard S, Weller P. Pseudotumorous folliculotropic mycosis fun- four cases. Br J Dermatol 2000; 143:832–6. goides. Am J Dermatopathol 2005; 27:224–7. 27 Vonderheid EC, Bernengo MG, Burg G et al. Update on erythro- 51 Klempke CD, Dippel E, Assaf C et al. Follicular mycosis fungoides. dermic cutaneous T-cell lymphoma: report of the International Br J Dermatol 1999; 141:137–40. Society for Cutaneous Lymphomas. J Am Acad Dermatol 2002; 52 Vollmer RT. Mycosis fungoides and follicular mucinosis. Arch Der- 46:95–106. matol 2002; 138:1613–14. 28 Neil SM, Du Vivier A. A case of mycosis fungoides mimicking 53 Cerroni L, Fink-Puches R, Ba¨ck B, Kerl H. Follicular mucinosis. A actinic reticuloid. Br J Dermatol 1985; 113:497–500. critical reappraisal of clinicopathologic features and association 29 Kikuchi A, Shimizu H, Nishikawa T. Mycosis fungoides with with mycosis fungoides and Se´zary syndrome. Arch Dermatol 2002; marked hyperpigmentation. Dermatology 1996; 192:360–3. 138:182–9. 30 Co´rdoba S, Fernańdez-Herrera J, Sańchez-Perez J et al. Vesiculo- 54 Brown HA, Gibson LE, Pujol RM et al. Primary follicular mucino- bullous mycosis fungoides. Br J Dermatol 1999; 141:164–5. sis: long-term follow-up of patients younger than 40 years with 31 Kodama K, Fink-Puches R, Massone C et al. Papular mycosis fun- and without clonal T-cell receptor gene rearrangement. J Am Acad goides: a new clinical variant of early mycosis fungoides. JAm Dermatol 2002; 47:856–62. Acad Dermatol 2005; 52:694–8. 55 LeBoit PE. Alopecia mucinosa, inflammatory disease or mycosis 32 Cordel N, Duval-Modeste AB, Courville P, Joly P. Hypopigmented fungoides: must we choose? And are there other choices? Am J mycosis fungoides. Ann Dermatol Venereol 2005; 132:455–7. Dermatopathol 2004; 26:167–70. 33 Barnhill R, Braverman IM. Progression of pigmented purpura-like 56 Boer A, Guo Y, Ackerman AB. Alopecia mucinosa is mycosis fun- eruptions to mycosis fungoides: report of three cases. J Am Acad goides. Am J Dermatopathol 2004; 26:33–52. Dermatol 1988; 19:25–31. 57 Hodak E, David M. Follicular mycosis fungoides. J Cutan Pathol 34 Werner B, Brown S, Ackerman AB. ‘Hypopigmented mycosis fun- 2002; 29:625. goides’ is not always mycosis fungoides. Am J Dermatopathol 2005; 58 Hur J, Seong JY, Choi TS et al. Mycosis fungoides presenting as 27:56–67. Ofuji’s papuloerythroderma. J Eur Acad Dermatol Venereol 2002; 35 McBride SR, Dahl MG, Slater DN et al. Vesicular mycosis fungoid- 16:393–6. es. Br J Dermatol 1998; 138:141–4. 59 Garcovich A, Garcovich S, Massi G. An unusual variant of gra- 36 Bazza MA, Ryatt KS, Dharmagunawardena PVB. Mycosis fungoid- nulomatous adnexotropic cutaneous T-cell lymphoma. Br J Derma- es masquerading as seborrhoeic keratosis. Br J Dermatol 2002; tol 2003; 148:363–4. 147:1264–81. 60 Fraser-Andrews E, Ashton R, Russell-Jones R. Pilotropic mycosis 37 Yoo SS, Viglione M, Moresi M, Vonderheid E. Unilesional myco- fungoides presenting with multiple cysts, comedones and alope- sis fungoides mimicking Bowen’s disease. J Dermatol 2003; cia. Br J Dermatol 1999; 140:141–4. 30:417–19. 61 Ackerman AB, Flaxman BA. Granulomatous mycosis fungoides. Br 38 Carbia SG, Hochman A, Chain M et al. Mycosis fungoides present- J Dermatol 1970; 82:397–401. ing with extensive pyoderma gangrenosum-like ulcers. J Eur Acad 62 Ackerman AB. Granulomatous slack skin. In: Histologic Diagnosis of Dermatol Venereol 2002; 16:401–4. Inflammatory Skin Diseases (Ackerman AB, ed.), Philadelphia: Lea & 39 Martıńez W, del Pozo J, Va´zquez J et al. Cutaneous T-cell lym- Febiger, 1978; 483–5. phoma presenting as disseminated, pigmented, purpura-like erup- 63 Bazex A, Dupre´ A, Christol B. Besnier–Boeck–Schaumann chalazo- tion. Int J Dermatol 2001; 40:140–4. dermique? Bull Soc Fr Dermatol Syphiligr 1968; 75:448–50.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 Diversity of mycosis fungoides, D. Nashan et al. 9

64 Willemze R, Kerl H, Sterry W et al. EORTC classiﬁcation for pri- 88 Lim DS, Murphy GM, Egan CA. Mycosis fungoides presenting as mary cutaneous lymphomas: a proposal from the Cutaneous annular erythema. Br J Dermatol 2003; 148:591. Lymphoma Study Group of the European Organization for 89 Cogrel O, Boralevi F, Lepreux S et al. Lymphomatoid annular ery- Research and Treatment of Cancer. Blood 1997; 90:354–71. thema: a new form of juvenile mycosis fungoides. Br J Dermatol 65 LeBoit PE. Granulomatous slack skin. Dermatol Clin 1994; 12:375– 2005; 152:565–7. 89. 90 Howard MS, Smoller BR. Mycosis fungoides: classic disease and 66 LeBoit PE, Zackheim HS, White CR Jr. Granulomatous variants of variant presentations. Semin Cutan Med Surg 2000; 19:91–9. cutaneous T-cell lymphoma: the histopathology of granulomatous 91 Whitmore E, Simmons-O’Brien E, Rotter FS. Hypopigmented my- mycosis fungoides and granulomatous slack skin syndrome. Am J cosis fungoides. Arch Dermatol 1994; 130:476–80. Surg Pathol 1988; 12:83–95. 92 Willemze R, Scheffer E, van Vloten WA. Mycosis fungoides simu- 67 Gadzia J, Kestenbaum T. Granulomatous slack skin without evi- lating acanthosis nigricans. Am J Dermatopathol 1985; 7:367–71. dence of a clonal T-cell proliferation. J Am Acad Dermatol 2004; 93 Wain EM, Orchard GE, Mayou S et al. Mycosis fungoides with a 50:S4–S8. CD56+ immunophenotype. J Am Acad Dermatol 2005; 53:158–63. 68 Cerroni L. Cutaneous granulomas and malignant lymphomas. Der- 94 Ho KKL, Browne A, Fitzgibbons J et al. Mycosis fungoides bullosa matology 2003; 206:78–80. simulating pyoderma gangrenosum. Br J Dermatol 2000; 142:124–7. 69 Wada K, Maesawa C, Satoh T et al. A case of primary cutaneous 95 Roenigk HH, Castrovinci AJ. Mycosis fungoides bullosa. Arch Der- CD30+ T-cell lymphoproliferative disorder with features of matol 1971; 104:402–6. granulomatous slack skin disease. Br J Dermatol 2002; 147:998– 96 Goldberg DJ, Stampien TM, Schwartz RA. Mycosis fungoides pal- 1002. maris et plantaris: successful treatment with the carbon dioxide 70 Jouary T, Beylot-Barry M, Vergier B et al. Mycosis fungoides laser. Br J Dermatol 1997; 136:617–19. mimicking granuloma annulare. Br J Dermatol 2002; 146:1102–4. 97 Bahadoran P, Wechsler J, Delfau-Larue MH et al. Mycosis fungoid- 71 Bessis D, Sotto A, Farcet JP et al. Granulomatous mycosis fungoides presenting as keratosis lichenoides chronica. Br J Dermatol 1998; es presenting as sarcoidosis. Dermatology 1996; 193:330–2. 138:1067–9. 72 Schmuth M, Prior C, Illersperger B et al. Systemic sarcoidosis and 98 Breneman DL, Breneman JC. Cutaneous T-cell lymphoma mim- cutaneous lymphoma: is the association fortuitous? Br J Dermatol icking porokeratosis of Mibelli. J Am Acad Dermatol 1993; 1999; 140:925–7. 29:1046–8. 73 Pereiro M, Sanchez-Aguilar D, Pereiro Ferreiros MM et al. Cutane- 99 Moreno JC, Ortega M, Conejo-Mir JS, Sanchez-Pedreno P. Palmo- ous T-cell lymphoma: an expression of papuloerythroderma of plantar pustulosis as a manifestation of cutaneous T cell lym- Ofuji. J Eur Acad Dermatol Venereol 2003; 17:227–49. phoma (mycosis fungoides). J Am Acad Dermatol 1990; 23:758–9. 74 Ofuji S, Furukawa F, Miyachi Y, Ohno S. Papuloerythroderma. 100 Camisa C, Aulisio A. Pustular mycosis fungoides. Cutis 1994; Dermatologica 1984; 169:125–30. 54:202–4. 75 Tay YK, Tan KC, Onhg BH. Papuloerythroderma of Ofuji and 101 Brill TJ, Ludwig RJ, Wolter M et al. Complicated mycosis fungoid- cutaneous T-cell lymphoma. Br J Dermatol 1997; 137:160–1. es mimicking facial erysipelas. Br J Dermatol 2005; 152:1381–3. 76 Goerdt S, Trautmann C, Ku¨tting B et al. Rare variants of cutaneous 102 Ameen M, Darva A, Black MM et al. CD8-positive mycosis fun- T cell lymphoma. Hautarzt 1996; 47:96–105. goides presenting as capillaritis. Br J Dermatol 2000, 142:560–7. 77 Gelfand JM, Berling J, van Voorhees A, Margolis DJ. Lymphoma 103 Lund KA, Parker CM, Norins AL, Tejada E. Vesicular cutaneous T rates are low but increased in patients with psoriasis. Arch Dermatol cell lymphoma presenting with gangrene. J Am Acad Dermatol 2003; 139:1425–9. 1990; 23:1169–71. 78 Herbert CR, McBurney EI. Cutaneous T-cell lymphoma in a 104 Cather JC, Farmer A, Jackow C et al. Unusual presentation of my- patient with neuroﬁbromatosis type 1. Cutis 2003; 72:27–30. cosis fungoides as pigmented purpura with malignant thymoma. 79 Mehrany K, El-Azhary RA, Bouwhuis SA, Pittelkow MR. Cutane- J Am Acad Dermatol 1998; 39:858–63. ous T-cell lymphoma and atopy: is there an association? Br J Der- 105 Machler BC, Elgart GW, Kerdel FA. Extracutaneous mycosis fun- matol 2003; 149:1013–17. goides of the gastrocnemius muscle mimicking sarcoma. JAm 80 Kempf W, Kutzner H, Kettelhack N et al. Paraneoplastic pityriasis Acad Dermatol 1994; 31:673–5. lichenoides in cutaneous lymphoma: case report and review of 106 Goldstein LJ, Williams JD, Zackheim HS, Helfend LK. Mycosis the literature on paraneoplastic reactions of the skin in lymphoma fungoides masquerading as an ischemic foot. Ann Vasc Surg 1999; and leukaemia. Br J Dermatol 2005; 152:1327–31. 13:305–7. 81 Kotz EA, Anderson D, Thierst BH. Cutaneous T-cell lymphoma. J 107 Pujol RM, Ballardo F, Llistosella E et al. Invisible mycosis fungoid- Eur Acad Dermatol Venereol 2003; 17:131–7. es: a diagnostic challenge. J Am Acad Dermatol 2000; 43:324–8. 82 Wolf P, Cerroni L, Kerl H. Mycosis fungoides mimicking perioral 108 Hwong H, Nichols T, Duvic M. ‘Invisible’ mycosis fungoides? J dermatitis. Clin Exp Dermatol 1992; 17:132–4. Am Acad Dermatol 2001; 45:318. 83 Zackheim HS, Koo J, LeBoit PE et al. Psoriasiform mycosis fun- 109 Gilliam AC, Lessin SR, Wilson DM, Salhany KE. Folliculotropic goides with fatal outcome after treatment with cyclosporine. JAm mycosis fungoides with large-cell transformation presenting as Acad Dermatol 2002; 47:155–6. dissecting cellulitis of the scalp. J Cutan Pathol 1997; 24:169–75. 84 Ackerman AB. If small plaque (digitate) parapsoriasis is a cutane- 110 Eisman S, O’Toole EA, Jones A, Whittaker SJ. Granulomatous my- ous T-cell lymphoma, even an ‘abortive’ one, it must be mycosis cosis fungoides presenting as an acquired ichthyosis. Clin Exp Der- fungoides! Arch Dermatol 1996; 132:562–6. matol 2003; 28:174–6. 85 Chaves TA, Graham-Brown RAC. Mycosis fungoides masquer- 111 van Haselen CS, Toonstra J, van der Putte SJC et al. Granulomatous ading as tinea of the axilla. Clin Exp Dermatol 2002; 27:66–7. slack skin. Report of three patients with an updated review of the 86 Resnik KS, Kantor GR, Lessin SR et al. Mycosis fungoides palmaris literature. Dermatology 1998; 96:382–91. et plantaris. Arch Dermatol 1995; 131:1052–6. 112 Topar G, Zelger B, Schmuth M et al. Granulomatous slack skin: a 87 Krebs A, Zala A, Graber W et al. Mycosis fungoides. Dermatologica distinct disorder or a variant of mycosis fungoides? Acta Derm Ven- 1978; 157:312–15. ereol (Stockh) 2001; 81:42–4.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 10 Diversity of mycosis fungoides, D. Nashan et al.

113 Sherertz EF, Westwick TJ, Flowers FP. Sarcoidal reactions to lym- 116 Lacour JP, Castanet J, Perrin C, Ortonne JP. Follicular mycosis phoma presenting as granulomatous rosacea. Arch Dermatol 1986; fungoides. J Am Acad Dermatol 1993; 29:330–4. 122:1303–5. 117 Peris K, Chimenti S, Sacerdoti G et al. Pilotropic mycosis fungoid- 114 Bonta MD, Tannous ZS, Demierre MF et al. Rapidly progressing es. Dermatology 1999; 199:192–4. mycosis fungoides presenting as follicular mucinosis. J Am Acad 118 Brecher A. Mycosis fungoides. Dermatol Online J 2002; 9:23–5. Dermatol 2000; 43:635–40. 119 Woollons A, Darvay A, Khorshid SM et al. Necrobiotic cutaneous 115 Oliwiecki S, Ashworth J. Mycosis fungoides with a widespread T-cell lymphoma. J Am Acad Dermatol 1999; 41:815–19. follicular eruption, comedones and cysts. Br J Dermatol 1992; 127:54–6.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp1–10 GUIDELINES DOI 10.1111/j.1365-2133.2006.07610.x Guidelines for management of Bowen’s disease: 2006 update N.H. Cox, D.J. Eedy* and C.A. Morton on behalf of the British Association of Dermatologists Therapy Guidelines and Audit Subcommittee Department of Dermatology, Cumberland Inﬁrmary, Carlisle CA2 7HY, U.K. *Craigavon Area Hospital, Craigavon BT63 5QQ, U.K. Stirling Royal Inﬁrmary, Stirling FK8 2AU, U.K.

Summary

Correspondence This article represents a planned regular updating of the previous British Associ- Neil Cox. ation of Dermatologists (BAD) guidelines for management of Bowen’s disease. E-mail: [email protected] They have been prepared for dermatologists on behalf of the BAD. They present evidence-based guidance for treatment, with identiﬁcation of the strength of evi- Accepted for publication 5 June 2006 dence available at the time of preparation of the guidelines.

Key words Bowen’s disease, guidelines, treatment

Conﬂicts of interest A statement is provided at the end of these guidelines.

Members of the British Association of Dermatologists Therapy Guidelines and Audit committee are A.D. Ormerod (Chairman), D.J. Eedy, D. Mitchell, F. Humphreys, J. Peters, R. Bull, H. Bell, M. Kouimtzi and S. Jones.

3 Disclaimer (BD). Those guidelines included discussion of epidemiology, predisposing factors, disease associations and risk of malig- These guidelines have been prepared for dermatologists on nancy as well as the local treatment options for the disease behalf of the British Association of Dermatologists (BAD) and itself. New information in these areas, other than that pertain- are based on the best data available at the time the report was ing to issues that influence therapeutic decisions, will only be prepared. Caution should be exercised when interpreting data briefly summarized in this update. Similarly, detailed evidence where there is a limited evidence base; the results of future for therapies discussed in the previous paper will not be studies may require alteration of the conclusions or recom- repeated except where comparison with other evidence is mendations in this report. It may be necessary or even desir- necessary. It should be recognized that this may entail a dis- able to depart from the guidelines in the interests of specific proportionate weight being given to referencing newer ther- patients and special circumstances. Just as adherence to guide- apies. Where there are direct comparisons between therapies, lines may not constitute defence against a claim of negligence, these are generally discussed in the section relating to those so deviation from them should not necessarily be deemed deemed to be most efficacious. Recommendations take into negligent. account simplicity, cost and healing as well as the type and validity of the published evidence base; for any treatment, Definition and introduction to the guideline there may be site-specific differences in the recommended option. The abbreviation RCT is used for randomized con- The scope, aims and methodology of the BAD guidelines pro- trolled trial throughout. cess have been published elsewhere;1,2 these references should BD is a form of intraepidermal (in situ) squamous cell carci- be consulted for guideline validation purposes. noma (SCC), originally described in 1912.4 Genital lesions This article represents a planned regular updating of the which have the histology of BD include erythroplasia of Quey- previous BAD guidelines for management of Bowen’s disease rat (males), some vulval intraepithelial neoplasia (VIN) and

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 11 12 Guidelines for management of BD: 2006 update, N.H. Cox et al. bowenoid papulosis (either sex). There is a strong association lesions on hands and feet (eight of 28 cases) and these accoun- of genital or perianal intraepithelial neoplasia in either sex ted for 50% of the positive results.12 A further study of HPV in with human papillomavirus (HPV) infection, although many extragenital cutaneous BD detected HPV DNA in 58% of 69 such cases do not have the clinical morphology of BD. The as- samples from 50 patients, the percentage of HPV detection sociation between BD and HPV is brieﬂy discussed and we being similar in exposed (55%) and unexposed areas (65%), have included some brief comments in relation to therapy and and also between immunosuppressed and immunocompetent outcomes (especially for erythroplasia of Queyrat, as this is patients.13 A study of the cell proliferation activity between often referred to as penile BD), but a detailed therapeutic HPV-positive and HPV-negative BD showed similar results in review of HPV-related epidermal dysplasia, VIN, vaginal intra- each, suggesting that HPV infection alone does not induce cell epithelial neoplasia and penile intraepithelial neoplasia (PIN) proliferation in those lesions.14 HPV 16 has been implicated in is beyond the scope of this guideline. Perianal BD is not com- 60% of palmoplantar and periungual lesions.15 Multiple lesions monly treated by dermatologists but is brieﬂy discussed as its of BD may occur on the distal digits (‘polydactylous BD’), con- therapy and outcome also often differ from those of BD at sistent with aetiological involvement of HPV. This has thera- extragenital sites. peutic implications, as HPV-induced BD should be responsive to agents that have a combined antiviral and antitumour effect. Clinical description, demographics and 5 Others – chronic injury or dermatoses, pre-existing skin variants lesions such as seborrhoeic keratoses (rarely).

Typical BD presents as a gradually enlarging well-demarcated Association with other malignancy erythematous plaque with an irregular border and surface crusting or scaling.3,5,6 An annual incidence of 15 per Internal neoplasia 100 000 has been suggested in the U.K.;6 however, this figure was derived from an American study that primarily examined Several larger studies and meta-analysis of the association internal neoplasia associated with BD,7 and was the annual in- between BD and internal cancers were summarized in our pre- cidence rate for the 1980 U.S. white population – this may vious guideline.3 A further study of 1147 patients found the not be the same as in the less sun-exposed U.K. population. overall incidence of internal cancers in patients with BD to be In the U.K., the peak age group for BD is the seventh dec- slightly increased [115 cancers vs. 103 expected, the standard- ade, it occurs predominantly in women (70–85% of cases), ized incidence ratio (SIR) of 1Æ1 not being statistically signifi- and about three-quarters of patients have lesions on the lower cant].16 However, there was an SIR of 3Æ2 for leukaemia in leg (60–85%).8,9 Lesions are usually solitary but are multiple men and of 4Æ6 for lung cancer in men with BD before age in 10–20% of patients. Less common sites or variants include 60 years (the overall lung cancer SIR for both sexes and all pigmented BD, subungual/periungual, palmar, genital and ages was 1Æ3). There are various sources of potential bias in perianal (see above) and verrucous BD. many studies of this type, and available evidence would still The age group, number and size of lesions, and site(s) suggest that routine investigation for internal malignancy in affected may all influence therapeutic choice. patients with BD is not justified (Strength of recommendation E, Quality of evidence I; Appendix 1). Aetiology Skin malignancy Reported relevant aetiological factors were discussed in the 1999 guidelines,3 but are briefly listed here in order to update Previous studies suggested that about 30–50% of subjects with the roles of HPV and immunosuppression. Aetiologies include: BD may have previous or subsequent nonmelanoma skin can- 1 Irradiation – solar, photochemotherapy, radiotherapy. cer (NMSC), mainly basal cell carcinoma.17,18 The NMSC risk 2 Carcinogens – arsenic. after an index BD is probably similar to the overall risk of 3 Immunosuppression – therapeutic,10,11 AIDS. For example, NMSC following any index NMSC (overall 35–60% 3-year one study demonstrated that 23% of skin cancers in renal risk19). In the study by Jaeger et al., NMSC had an SIR of 4Æ3, transplant recipients were BD.11 This would suggest that edu- and lip cancer of 8Æ2, in patients with BD.16 These increased cating immunosuppressed patients about sun exposure is im- risks of further BD or of other NMSC probably reflect a com- portant. mon solar aetiology. 4 Viral – oncogenic HPV types such as HPV 16 are strongly implicated in the aetiology of VIN, but are also common in Risk of progression to squamous cell perianal BD, and in PIN. However, HPV DNA has also been carcinoma demonstrated in some extragenital BD. A report of 28 biopsies from extragenital sites demonstrated in situ hybridization evi- There is no new literature to inform this debate in terms of dence of oncogenic HPV types in eight of 28 (29%); all had the overall risk – ex vivo research studies to identify individual HPV 16/18 and two of these also had HPV 31/33/51. Of lesional risk and differentiation from other NMSC are beyond note, this study had a higher than average proportion of the scope of this guideline.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 Guidelines for management of BD: 2006 update, N.H. Cox et al. 13

Most studies suggest a risk of invasive carcinoma of about modalities, the regimen for the established treatment or the 3–5% for ‘ordinary’ BD20,21 and perhaps 10% for erythropla- site at which it is applied may not concur with the approach sia of Queyrat.3 Perianal BD also has higher risk of invasion used by all dermatologists. and recurrence (Quality of evidence II-ii), and an association with Other than a small number of anecdotal or single series cervical and vulval dysplasia.22–24 However, these estimates reports considered at the end of the therapeutic list, the thera- are drawn from retrospective case series, may be biased by peutic options have been listed in a sequence to include obser- different referral patterns of lesions to different disciplines vation alone, topical treatments and surgical treatments; within (dermatologists, surgeons etc.), and do not take account of this sequence, longer-established or less interventional treat- subjects with BD who have either not requested medical ments are considered first. This sequence does not necessarily advice or who have been treated in primary care. Indeed, it is reflect the frequency of use, importance, availability or strength unlikely that this question can be accurately answered as any of evidence for any treatment option – a summary of advice group of patients who could be followed up without interven- incorporating these issues and related to lesion sites and sizes is tion are likely to be unrepresentative individuals (for example, provided in Table 1. Current U.K. product licences for many elderly patients with small lesions). Risks of cervical intraepit- drugs listed do not include treatment of BD; all recommenda- helial carcinoma in affected women or in female sex partners tions in this guideline are extrapolated from literature on BD of affected men with bowenoid papulosis, and of oral papillo- and knowledge of other neoplastic skin lesions, and are presen- mas and tumours in association with HPV 16-positive bowe- ted on the understanding that neither the authors nor the BAD noid papulosis, were discussed previously.3 can formally recommend an unlicensed treatment. Treatments are presented in a sequence that discusses the Treatments least invasive and topical therapies first, surgical approaches, and finally treatments that require more complex or expensive Evaluation of treatment studies of BD is difficult due to poten- equipment or that are not as widely available. tial selection bias to specific forms of treatment. Similarly, healing and success rate may vary with body site. Earlier stud- No treatment ies used clinical appearance rather than histological assessment to determine the end-point of lesion clearance. Even for the In some patients with slowly progressive thin lesions, especial- same treatment modality, there is difficulty in directly com- ly on the elderly lower leg where healing is poor, there is an paring studies due to different lesion sites, sizes of lesions, argument for observation rather than intervention. and use/availability of different types of equipment and treatment regimens.25 Retrospective studies in particular may have 5-Fluorouracil (Strength of recommendation B, Quality of several inherent problems – in ‘real world’ treatment of BD, evidence II-i) dermatologists may select several different types of treatment,26 decisions potentially being influenced by several fac- 5-Fluorouracil (5-FU) has been used topically for treatment of tors such as lesion size and thickness, equipment available, BD in several studies as previously summarized.3 Most of these and the perceived potential for poor wound healing (e.g. at are open trials or small case series; several used concentrations sites such as the lower leg27). Even in recent controlled trials that are not commercially available in the U.K., and some do in which older treatments are compared with newer not specify the concentration or schedule. These suggest cure

Table 1 Summary of the main treatment options for Bowen’s disease. The suggested scoring of the treatments listed takes into account the evidence for beneﬁt, ease of application or time required for the procedure, wound healing, cosmetic result and current availability/costs of the method or facilities required. Evidence for interventions based on single studies or purely anecdotal cases is not included

Lesion characteristics Topical 5-FU Topical imiqumoda Cryotherapy Curettage Excision PDT Radiotherapy Laserb Small, single/few, good healing sitec 43 213354 Large, single, good healing sitec 33 355247 Multiple, good healing sitec 34 235344 Small, single/few, poor healing sitec 2 3 3 2 2 1–2 5 7 Large, single, poor healing sitec 3 2–3 5 4 5 1 6 7 Facial 4 7 2 2 4d 34 7 Digital 3 7 3 5 2d 33 3 Perianal 6 6 6 6 1e 7 2–3 6 Penile 3 3 3 5 4d 3 2–3 3

5-FU, 5-fluorouracil; PDT, photodynamic therapy; 1, probably treatment of choice; 2, generally good choice; 3, generally fair choice; 4, reasonable but not usually required; 5, generally poor choice; 6, probably should not be used; 7, insufficient evidence available. aDoes not have a product licence for Bowen’s disease. bDepends on site. cRefers to the clinician’s perceived potential for good or poor healing at the affected site. dConsider micrographic surgery for tissue sparing or if poorly defined/recurrent. eWide excision recommended.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 14 Guidelines for management of BD: 2006 update, N.H. Cox et al. rates of 90–100%. In current clinical use, 5-FU is usually effects, and is therefore potentially useful for HPV-associated applied once or twice daily as a 5% cream for a variable period BD/bowenoid papulosis as well as for non-HPV-associated BD. of time (between 1 week and 2 months in most studies using The evidence base consists of a single small RCT, one open this concentration) to achieve disease control, and repeated if study plus some small case series (most two or three patients) required at intervals. Lower concentrations are less effective. and individual case reports.33–40 The regimen used varies Efficacy may be increased by application under occlusion, use between reports. Such reports are not referenced at length as it of dinitrochlorobenzene as a vehicle (both previously refer- is felt that stronger evidence is required before firm conclusions enced3), iontophoresis28 (to improve follicular penetration) or can be drawn. At the time of writing, the product licence for pretreatment with a laser (to ablate the stratum corneum and topical imiquimod in the U.K. is for small superficial basal cell thereby enhance penetration of 5-FU).29 In the study of ionto- carcinomas; it is not currently licensed for use in BD. phoresis,28 only one of 26 patients had histological evidence of The best evidence currently available is a single small RCT residual disease at 3 months after eight treatments. More recent- that demonstrated 73% histologically proven resolution with ly, the erbium:YAG laser was used as a pretreatment measure imiquimod (once daily for 16 weeks; lesions untreated for at on half of each lesion in three lesions from a patient with mul- least a month) vs. zero response in the placebo group.33 This tiple BD, who was subsequently treated with twice-daily appli- study acknowledged that the ideal dosing regimen and cost-ef- cation of 5-FU cream to both sides. The response (clinical and fectiveness require further investigation. An earlier 16-patient histological) was accelerated on the side pretreated with laser.29 open study (15 having lower leg lesions; once daily application Few studies provide details of the success rate for the cur- for up to 16 weeks; previously untreated lesions) documented rently available preparation in the U.K. (5% cream to be used that 14 of 15 patients (93%) who completed the study had once or twice daily for 3–4 weeks) as a first-line option for un- clinical and pathological resolution of BD 6 weeks after the selected lesions. However, in an RCT comparing 5-FU with treatment period (one patient died of unrelated causes and was photodynamic therapy (PDT) the initial response rate in the 5- not analysed).34 Five lesions had an area of 5 cm2 or greater. FU limb, after one (or two if required) cycles of once-daily ap- Single cases or small case series suggest that different regi- plication for 1 week then twice daily for 3 weeks, repeated at mens such as cyclical treatment35 might be useful, and also 6 weeks if clinically indicated, was 67%, with only 48% that imiquimod may be useful for large facial lesions.36 The remaining clear at 12 months.30 (The comparative results are latter, together with lower leg lesions,34 are typically those discussed in the section on PDT, below). By contrast, in a fol- that pose the greatest therapeutic challenge. Some studies sug- low-up study (26 patients, clinical follow up of 2Æ4– gest that shorter treatment periods may be adequate.34,35 In 204 months), recurrences had occurred at some point in just the open study discussed above,34 six of 16 patients discontin- two patients (8%).31 This study used 5% 5-FU twice daily for a ued treatment early due to side-effects but still had lesion planned 9 weeks (actual 3–13 weeks), with a repeat cycle for clearance, and in the placebo-controlled trial, three of 15 in recurrences, and biopsy to confirm clearance in most cases, but the active limb dropped out (two being withdrawn by the is a small number collected given the 10-year overall period. investigators due to local side-effects).33 A few anecdotal As 5-FU can be very irritant, less aggressive regimens have reports and small open-label case series suggest that there may been used for disease control rather than cure. Two applica- be a role for imiquimod in treatment of erythroplasia of tions of 5% 5-FU on a single day of each week for 3 months Queyrat37 and in basaloid VIN. improved lesions in 24 of 26 patients (92%) with BD of the Benefit has also been reported in the treatment of BD in im- leg (lesions were flat with less or no erythema, and with munosuppressed patients,38–40 although combining it with minor or absent scaling), although long-term clearance was other modalities such as oral sulindac39 or 5-FU40 makes in- achieved only in a minority with this regimen.32 terpretation of the relative roles of the pairs of agents used Formal comparison with other modalities is limited to RCTs difficult. Five renal transplant patients with BD have been trea- of PDT vs. 5-FU, only one of which is currently published ted with a combination of a local immune therapy, imiqui- and which showed that PDT was the more effective30 (see sec- mod cream, and a topical chemotherapeutic agent, 5% 5-FU, tion on PDT below). with clearing of the areas of SCC in situ. In addition, there is In erythroplasia of Queyrat, application of 5% 5-FU cream evidence that cytokines induced by imiquimod may improve twice daily for 4–5 weeks has been recommended but inflam- the therapeutic efficacy of topical 5% 5-FU in BD.40 mation frequently limits this treatment regimen. Cryotherapy (Strength of recommendation B, Quality of Imiquimod (Strength of recommendation B, Quality of evidence II-i) evidence I) The results reported vary, probably reflecting differences Imiquimod is a topical immunomodulatory heterocyclic imida- between studies in the techniques and regimens used. As previ- zoquinoline amide that has become available since the earlier ously summarized,3,6 the failure rate varies from zero to about BAD guideline.3 It has been used as a 5% cream to treat BD, in- 35%, the larger series suggesting a failure or recurrence rate in cluding larger diameter lesions, lower leg lesions and ery- the order of 5–10% provided that adequate cryotherapy is throplasia of Queyrat. It has both anti-HPV and antitumour used [e.g. liquid nitrogen (LN2) cryotherapy, using a single

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 Guidelines for management of BD: 2006 update, N.H. Cox et al. 15 freeze-thaw cycle (FTC) of 30 s, two FTCs of 20 s with a thaw smaller studies, and at sites such as the perianal region. While period, or up to three single treatments of 20 s at intervals of it is logical that excision should be an effective treatment, the several weeks].41–44 Such doses do, however, cause discomfort evidence base is limited. Additionally, lower leg excision and may cause ulceration, especially on the lower leg. wounds may be associated with considerable morbidity.26 In an RCT of PDT vs. cryotherapy,44 the latter produced A retrospective study of 47 cases of perianal BD24 found a 100% clearance in 20 patients with one to three treatments of lower recurrence rate for wide excision (six of 26, 23%) than

LN2 using one FTC of 20 s on each occasion (50% success for local excision (eight of 15, 53%) or laser therapy (four of after a single treatment). Ulceration was observed following five, 80%) although this series did not include patients treated cryotherapy in 25% of lesions. There were two (10%) recur- with radiotherapy (which has been recommended as a first- rences following cryotherapy in the 1-year follow-up period. line treatment,47 discussed below). Wide surgical excision is A single treatment of PDT was significantly more effective the most commonly used treatment for perianal BD;48 a sur- than cryotherapy. vey of American colorectal surgeons found that most are per- A prospective, nonrandomized study comparing curettage forming wide local excision for both small and large perianal vs. cryotherapy found better healing, less discomfort and a BD lesions (96% for patients with small lesions and 87% for lower recurrence rate with curettage,45 and cryotherapy had patients with large lesions). Prolonged follow up is recom- more complications (discussed below). mended as late recurrences are common at this site (see the Cryotherapy appears to have a good success rate with ade- previous guideline3). quate treatment (recurrences less than 10% at 12 months) but Mohs micrographic surgery has become the recommended healing may be slow for broad lesions and discomfort may treatment for digital BD and for some cases of genital (espe- limit treatment of multiple lesions. Curettage and PDT both cially penile) BD for its tissue-sparing benefits. A large retro- have higher success rates and less discomfort overall, but are spective series of 270 patients has reported on micrographic more time-consuming and/or expensive to perform. surgery for tissue sparing at head and neck sites (this site comprised 252 of 270 patients).49 This study included 128 cases of previously treated head and neck BD. Among the 270 Curettage with cautery/electrocautery (Strength of cases analysed, 94 had had previous cryotherapy, 18 curettage recommendation A, Quality of evidence II-ii) and cautery, 44 excision (10 incomplete) and one radiother- Previously summarized studies3 suggested a wide range of apy (some had been treated with more than one modality); cure rates without recurrence, larger series suggesting a recur- nearly all referrals cited poorly defined tumour, recurrent or rence rate of 20%.17 These studies do not give details of the incompletely excised tumour, or tumour site as the rationale treatment regimens or equipment used (cautery vs. electrodes- for micrographic surgery, so it cannot be assumed to be rou- iccation, number of cycles etc.). tinely necessary or cost-effective (Strength of recommendation B, In a prospective but nonrandomized trial of curettage and Quality of evidence II-iii). The mean and median number of exci- cautery (44 lesions) compared with cryotherapy (36 lesions) sion levels for clearance was 2, range 1–7. Of 95 patients who involving 67 patients, curettage was preferable in terms of had 5-year follow-up data there were six (6%) recurrences. pain, healing and recurrence rate.45 Seventy-four percent of lesions were on the lower leg. Median time to healing with Photodynamic therapy (Strength of recommendation A, cryotherapy was 46 days (90 days on the lower leg) com- Quality of evidence I) pared with 35 days (lower leg 39 days) for curetted lesions, and reported pain was significantly greater with cryotherapy. This modality requires the activation of a photosensitizer, usu- Recurrences were more likely following cryotherapy (13 of ally a porphyrin derivative, by visible light. Systemic photosen- 44) compared with curettage (four of 36) during a median sitization, with various photosensitizers, was used with 2 years’ follow up, although the cryotherapy regimen was less excellent results in early studies summarized previously.3 A aggressive than that used by authors in most studies of this recent report using systemic verteporfin, which has a much technique (see above and comment in this Journal25). shorter duration of photosensitivity than agents used in earlier Curettage followed by cryotherapy has also been used, but studies, has confirmed the efficacy of this approach.50 It has a reports are anecdotal and it is impossible to determine the particular role in patients with multiple BD lesions, in whom relative contribution of the two treatments or whether the use of topical porphyrin derivatives may be expensive and time- combination is better than either alone. consuming, although topical PDT is more practical for most BD. These guidelines refer mainly to topical PDT using topical 5-aminolaevulinic acid (ALA) or its ester, methyl aminolaevu- Excision (Strength of recommendation A, Quality of linate (MAL). Studies have included various illumination evidence II-iii) sources (e.g. filtered xenon arc, diode, halogen, laser), wave- There are no substantive new data on simple excision since lengths (red, green, blue/violet light) and dosing schedules the last guidelines.3 In retrospective studies of 65 and 155 (both in duration of ALA application and total light energy patients, the reported recurrence rates were 4Æ5%17 and delivered), hence comparisons between studies may be 19%,46 respectively. Even higher rates are recorded in some difficult to interpret. Most studies have used one or two

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 16 Guidelines for management of BD: 2006 update, N.H. Cox et al. treatments, depending on response (usually repeated at about approved in many countries for the treatment of actinic kera- 6 weeks if clinically necessary). Issues such as use of analgesia, toses, basal cell carcinomas and BD. and fluorescence detection, are not addressed here but details may be found in the British Photodermatology Group guide- Comparison between wavelengths lines for topical PDT.51 The previously summarized studies3 suggested an initial Green light (29 patients) was compared with red light (32 clinical clearance rate for ALA-PDT of 80–100% (most around patients) in an RCT using ALA-PDT for BD but was inferior in 90%) with one or two treatments, and a recurrence rate of terms of initial clearance (94% vs. 72%) and 12-month clear- about 0–10% at 12 months. These figures remain valid. In a ance (88% vs. 48%) and had no advantages in terms of pain prospective open study, 44 of 50 lesions (88%) cleared after (which was the rationale for the investigation).57 Violet light ) two treatments (30 of 50 after one treatment, 60%) although irradiation (10–20 J cm 2, after application of ALA for 8 h) two patients failed to clear after four treatments; this study, was used in six patients with BD, including one with multiple which used a halogen red light source, had a 31% 12-month lesions involving 50% of the scalp, the rationale being the recurrence rate in the 48 initially responsive lesions.52 Simi- lower light dose required for production of phototoxicity.58 larly, a departmental review documented that 117 of 129 le- Despite the theoretical risk of reduced light penetration com- sions (91%) were cleared,53 and a trial vs. 5-FU found that pared with red light PDT, the solitary lesions responded in all 29 of 33 lesions (88%) cleared after one or two treatments.30 four evaluable patients (one dropped out for unrelated rea- An open study using ALA-PDT specifically for large diameter sons) and the large area of scalp involvement showed a 90% and multiple BD lesions showed that 35 (88%) of 40 large pat- response, 50% of the remaining area responding to retreat- ches of BD, all with a maximum diameter > 20 mm, cleared ment. However, there has been no direct comparison with following one to three treatments, although four patches other wavelengths. recurred within 12 months. In 10 further patients with multiple PDT has been used specifically in immunosuppressed sub- (three or more) patches of BD, 44 (98%) of 45 patches cleared, jects, in an open trial involving 20 transplant recipients and although four lesions recurred over 12 months.54 20 immunocompetent controls with histologically confirmed Digital BD was treated with PDT in four patients, with good actinic keratoses or BD (one or two treatments, 20% ALA for ) ) cosmetic and functional results (one recurrence at 8 months 5h,75Jcm 2 of visible light delivered at 80 mW cm 2). responded to retreatment);55 the schedule was different from The cure rates in both patient groups were comparable at that in most studies (2% ALA solution, occluded for 16 h, 4 weeks (86%) but were significantly lower in transplant ) two treatments of 240 J cm 2 90 min apart using a 630-nm recipients than in controls at 12 and 48 weeks (below 50%). diode source). There are additional single case reports. Despite the poor long-term response, the authors concluded There are several comparative studies involving PDT, as fol- that PDT is particularly useful in transplant recipients because lows. of the possibility for repeated treatment of large lesional areas (although subsequent responsiveness was not confirmed).59 Successful use of PDT has also been reported in two cases Comparison with other treatments of bowenoid papulosis using ALA-PDT with a diode red light ALA-PDT for BD has been compared with cryotherapy45 and source. with 5-FU,30 each in an RCT involving 40 patients. PDT proved superior in terms of efficacy and adverse events in Radiotherapy (Strength of recommendation overall B, comparison with 5-FU, as well as being less painful than cryo- Quality of evidence II-iii for most sites; Strength of therapy. Both studies used a PDT schedule of 20% ALA applied recommendation D, Quality of evidence II-iii for lower 4 h before irradiation with narrowband red light. The cryo- leg) therapy study is discussed above and was summarized in the 1999 guideline.3 In the comparison with 5-FU, this was Various radiotherapy techniques and regimens have been used applied as 5% cream once daily for a week and then twice to treat BD. The larger studies have suggested a complete daily for 3 weeks; either treatment was repeated at 6 weeks if response rate to X-irradiation of 100%, for example in 77 le- necessary. Initial clearance rates (PDT vs. 5-FU) were 88% vs. sions treated by Blank and Schnyder60 (in this study, two 67%, and 12-month rates were 82% vs. 48%, with more patients with genital lesions relapsed at 8 and 16 months) and short-term side-effects in the 5-FU group.30 in 59 patients treated by Cox and Dyson.43 Topical MAL-PDT has been compared with clinician’s choice The patients in the latter series all had lower leg lesions; of cryotherapy or 5-FU in a 40-centre European trial of 225 poor healing, related to age, diameter of field and radiother- patients with 275 lesions of BD:56 MAL was applied for 3 h apy dose, was a feature in 12 of 59 (20%) of cases. Poor heal- and sites illuminated with a broadband red light. Lesion coming of lower legs was supported by a more recent but smaller plete response rates 3 months after last treatment were similar retrospective series of 11 patients with 16 lower leg lesions in with MAL-PDT (107 of 124; 86%), cryotherapy (75 of 91; which 100% cure was obtained but with 25% failure to heal 82%) and 5-FU (30 of 36; 83%). PDT gave superior cosmetic (median follow up 27Æ5 months, minimum 9 months), even results compared with cryotherapy or 5-FU. MAL-PDT is now though the fraction sizes used were relatively low.61 Thus the

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 Guidelines for management of BD: 2006 update, N.H. Cox et al. 17 high cure rate of radiotherapy may be offset by impaired heal- been considered inappropriate.66 The rationale is that the ing on the lower leg, and it is best avoided for BD at this site aspirator removes epidermis but not dermis. It has a 2-mm (Strength of recommendation D, Quality of evidence II-iii). diameter probe, and up to 300 lm oscillation at 28 kHz. An To overcome some of the disadvantages of external beam area of about 1 cm of normal skin was included in the treat- X-irradiation, a skin patch coated with high-energy b-emitter ment field, treatment taking 5–10 min under local anaesthesia. holmium-166 (166Ho patch) was used to treat 29 biopsy-con- Follow up was monthly for 12 months, 3-monthly thereafter, firmed BD lesions in eight patients [one with 22 sites, one for 12–26 (mean 20) months with no recurrences. Large le- with three sites but only one (palmar) treated with this sions, lower leg lesions and lesions over joints were included. method, the others solitary].62 All lesions were 3 cm or larger Hyperthermic treatment was performed using disposable (up to 7Æ2 cm); most lesions were on buttocks or thighs, or chemical pocket warmers applied under pressure each day were acral in the patient with multiple lesions (no lower leg throughout the patient’s waking hours for 4–5 months.67 lesions were specifically identified in the report). The patches There was initial complete clinical remission in six of eight were applied to the surface of lesions for 30–60 min for a patients (and partial remission in one) but absence of residual total radiation dose of 35 Gy. Acute radiation reactions healed histological evidence of BD was achieved only in three of within a month with mild fibrosis; there were good functional eight. Although of some benefit, this response compares and cosmetic results with confirmed histological clearance at poorly with other standard therapies (Strength of recommendation E, 5 months and without any late (10 months–2 years) recur- Quality of evidence IV). rences or complications. This treatment may therefore be use- Acitretin has been used alone or in conjunction with 5-FU ful, at least at non-lower leg sites (Strength of recommendation B, in anecdotal cases but the relative merits of each are unclear Quality of evidence II-ii). in the combination approach. The same applies to combin-

Radiotherapy has been advocated as the treatment of choice ations of topical bleomycin with LN2 cryotherapy in a case of for anal margin epidermoid cancers, although without any digital BD and isotretinoin with interferon-a in a patient with strong evidence to support this viewpoint.47 multiple lesions.

Laser (Strength of recommendation overall B, Quality of Summary of treatment modalities evidence II-iii but may vary according to site) All of the above treatments have some advantages and disad- Lasers have mainly been used to treat lesions at difficult sites vantages, which are dictated by lesional factors (size, number, such as the finger or genitalia. Results are generally stated to site, potential for healing or functional impairment), general be good (Strength of recommendation B, Quality of evidence II-iii), but health issues, availability and costs (both of the equipment or most published results are based on small numbers, or are agent, and of the time to deliver the treatment or its after- considered with other epidermal neoplasia and are difficult to care). A cost-minimization analysis showed that, at the time, analyse. One retrospective review included six cases of digital curettage or excision were the cheapest options, and PDT the

BD treated with CO2 laser, and reported good cosmetic results, most expensive (other treatments considered in this study no functional deficit and no recurrences (follow up 0Æ5– were cryotherapy, 5-FU and laser).68 However, changing costs 7Æ7 years),63 although some failures for digital BD are repor- of PDT and laser, the likely use of (relatively expensive) imiq- ted by others (one of five cases).64 uimod in the future, and the fact that all of these therapies A study of 16 patients with 25 lower leg BD lesions treated may not have equivalent efficacy, means that it is difficult to with CO2 laser demonstrated 100% healing at 2 months and make a strong and currently applicable single recommendation no recurrences at 6 months. However, there was a 12% pro- on the basis of this study. gression to invasive carcinoma within 12 months of discharge The relative status of the available treatment options is sum- from follow up. This suggests that the depth of laser eradica- marized in Table 1. This takes into account the evidence for tion may have been inadequate, and there are currently some benefit, ease of application and time required for the tech- reservations about use of this modality at this site (Strength of nique, wound healing, cosmetic result and availability of the recommendation C, Quality of evidence II-iii).65 method or facilities required. Results for perianal BD are poor48 (Strength of recommendation D, Quality of evidence II-iii). CO2 laser has been recommended for Follow up erythroplasia of Queyrat (Strength of recommendation B, Quality of evidence II-iii) but there is inadequate evidence to comment on The required duration of follow up remains uncertain. Some other sites (Quality of evidence IV). treatments may need to be repeated, for example a second PDT treatment is typically needed in about 20–30% of patients, and a second cycle of 5-FU is needed in a similar Other treatments proportion (although the latter may potentially be instituted An ultrasonic surgical aspirator was used initially in an animal by the patient or the primary care physician). Formal studies model (grafted areas of BD onto immunodeficient mice) and have generally used 12-month follow-up periods and clinical subsequently for 20 human lesions of BD where surgery had assessment for detection of recurrences.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 18 Guidelines for management of BD: 2006 update, N.H. Cox et al.

On the basis that most of the treatments have about a 10% 5 Topical PDT has been shown to be equivalent or superior recurrence risk, a follow-up check for possible recurrence at to cryotherapy and 5-FU, either in efficacy and/or in healing, 6–12 months is recommended. Different arrangements may in RCTs (Strength of recommendation A, Quality of evidence I). It may be dictated in the shorter term by a likely need for a second be of particular benefit for lesions that are large, on the lower cycle of treatment or to check on healing (in which case leg or at otherwise difficult sites, but it is costly. PDT for review at about 2–3 months to confirm clearance and healing NMSC and premalignant skin lesions has now been approved may be more appropriate). The requirement for subsequent as an interventional procedure by the National Institute for review should take into account the presence of multiple le- Health and Clinical Excellence in the U.K.,69 and MAL-PDT sions, previous recurrence, high-risk lesions, other skin neopl- has been approved by the European Medicines Authority for asia, background risk factors such as immunosuppression, the treatment of BD. reliability of the patient and the degree of primary care sup- 6 Curettage has good evidence of efficacy, and time to heal- port. For treatments that are novel, outside product licence, or ing is faster than with cryotherapy (Strength of recommendation A, have a small evidence base, we suggest that follow up should Quality of evidence II-ii). be more frequent and should continue for at least 12– 7 Cryotherapy has good evidence of efficacy (Strength of recom- 24 months in order to compare results with current literature mendation B, Quality of evidence II-i), but discomfort and time to on other therapies. healing are inferior to PDT (Strength of recommendation A, Quality of As a specific follow-up issue, the higher risk and the late evidence I) or curettage (Strength of recommendation A, Quality of evi- timing of recurrences of perianal BD should be noted. In a dence II-ii). series of 19 patients with perianal BD23 the recurrence rate in- 8 Excision should be an effective treatment with low recur- creased from 16% at 1 year to 31% at 5 years; in another ser- rence rates, but the evidence base is limited and for the most ies, the median time to recurrence for 26 radically excised part does not allow comment on specific sites of lesions lesions was 41Æ5 months.24 Longer follow up may therefore (Strength of recommendation overall A, Quality of evidence II-iii). Lower be appropriate for BD at less common and less visible sites, or leg excision may be limited by lack of skin mobility. For peri- where HPV infection is likely to have been relevant. anal BD treated surgically, wide excision is recommended rather than narrow excision or laser treatment (Strength of recom- Tools for guideline users mendation A, Quality of evidence II-iii). Micrographic surgery is logical at sites such as digits or penis where it is important to We have presented here: limit removal of unaffected skin (Strength of recommendation B, 1 A summary of the evidence and relevant aspects of the Quality of evidence III) and is useful for poorly defined or recur- main management options. rent head and neck BD (Strength of recommendation B, Quality of evi- 2 A tabular summary of appropriate treatment for different dence II-iii). lesional types, sizes and situations. 9 Radiotherapy has good evidence of efficacy but poor heal- 3 Suggestions for audit. ing on the lower leg suggests that it should be avoided at this site (Strength of recommendation generally B, Quality of evidence II-iii; for lower leg lesions Strength of recommendation D, Quality of evidence II- Summary of the main management recommendations iii). 1 Routine investigation for internal malignancy in patients 10 There is limited evidence on laser treatment, suggesting with BD is not justified (Strength of recommendation E, Quality of that it is a reasonable option for digital or genital lesions evidence I). (Strength of recommendation B, Quality of evidence II-iii) but probably 2 The risk of progression to invasive cancer is about 3%. This not for other sites (Strength of recommendation mostly C or D, Quality risk is greater in genital BD, and particularly in perianal BD. A of evidence II-iii to IV); specifically, results for perianal BD are high risk of recurrence, including late recurrence, is a particu- worse than those using wide surgical excision. lar feature of perianal BD and prolonged follow up is recommended for this variant (Strength of recommendation A, Quality of All therapeutic options have failure and recurrence rates at evidence II-ii). least in the order of 5–10%, and no treatment modality 3 There is reasonable evidence to support use of 5-FU (Strength appears to be superior for all clinical situations. Direct com- of recommendation B, Quality of evidence II-i) but its use may be limited parison between treatment modalities is difficult as there are by irritancy and it was less effective than PDT in an RCT. It is few randomized clinical trials with comparable patient sub- more practical than surgery for large lesions, especially at poten- groups. There is now increased choice for patients between tially poor healing sites, and has been used for ‘control’ rather clinic-based and home-applied therapies. For individual than cure in some patients with multiple lesions. patients, factors such as treatment-related morbidity and the 4 Topical imiquimod is likely to be used for BD (Strength of ease and availability of the treatment options may be a greater recommendation B, Quality of evidence I), especially for larger lesions issue than the cure rate. As previously, we still feel that it is or difficult/poor healing sites. However, it is costly, currently important that our BAD therapeutic guidelines reflect the fact unlicensed for this indication, and the optimum regimen has that there is no single definite ‘right way’ to treat all patients yet to be determined. with BD.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 Guidelines for management of BD: 2006 update, N.H. Cox et al. 19

Summary of appropriate treatment for different lesional 11 Bordea C, Wojnarowska F, Millard PR et al. Skin cancers in types, sizes and situations renal-transplant recipients occur more frequently than previously recognized in a temperate climate. Transplantation 2004; See Table 1. 77:574–9. 12 Derancourt C, Mougin C, Chopard-Lallier M et al. Oncogenic human papillomaviruses in extra-genital Bowen disease revealed by Possible audit points in situ hybridization. Ann Dermatol Venereol 2001; 128:715–8. 13 Que´reux G, N’Guyen JM, Dreńo B. Human papillomavirus and Is a measure of patient acceptability linked with treatment? (may extragenital in situ carcinoma. Dermatology 2004; 209:40–5. be indirect, e.g. willingness to repeat treatment if necessary) 14 Mitsuishi T, Kawana S, Kato T, Kawashima M. Human papillomavi- For novel, unlicensed or small evidence-base treatments, has rus infection in actinic keratosis and Bowen’s disease: comparative clinical cure rate been extended to 12 months? (and what are study with expression of cell-cycle regulatory proteins p21(Waf1/ the results?) Cip1), p53, PCNA, Ki-67, and Bcl-2 in positive and negative le- For destructive therapies, has the dose, frequency etc. been sions. Hum Pathol 2003; 34:886–92. recorded where applicable? 15 McGregor JM, Proby CM. The role of papillomaviruses in human non-melanoma skin cancer. Cancer Surv 1996; 26:219–46. 16 Jaeger AB, Gramkow A, Hjalgrim H et al. Bowen disease and risk of Conflicts of interest subsequent malignant neoplasms: a population-based cohort study of 1147 patients. Arch Dermatol 1999; 135:790–3. Relevant product details are given in brackets for the first citation of 17 Thestrup-Pedersen K, Ravnborg L, Reymann F. Morbus Bowen. Acta any pharmaceutical company below. Derm Venereol (Stockh) 1988; 68:236–9. N.H.C. has received support for attendance at non-product-related 18 Reizner GT, Chuang TY, Elpern DJ et al. Bowen’s disease (squa- educational meetings from Valeant Pharmaceuticals (5-fluorouracil mous cell carcinoma in situ) in Kauai, Hawaii. A population-based cream); has acted as an advisor regarding development of pathways of incidence report. J Am Acad Dermatol 1994; 31:596–600. care for basal cell carcinoma, sponsored by 3M Pharmaceuticals (imiq- 19 Marcil I, Stern RS. Risk of developing a subsequent nonmelanoma uimod); and has a performed a sponsored clinical trial for Photocure skin cancer in patients with a history of nonmelanoma skin cancer: of photodynamic therapy (PDT) using methyl aminolaevulinic acid for a critical review of the literature and meta-analysis. Arch Dermatol Bowen’s disease. He has performed unsponsored research on cryother- 2000; 136:1524–30. apy and radiotherapy for Bowen’s disease. 20 Peterka ES, Lynch FW, Goltz RW. An association between Bowen’s D.J.E. has received fees for speaking and chairing meetings for 3M disease and cancer. Arch Dermatol 1961; 84:623–9. Pharmaceuticals, travelling expenses from Leo Pharmaceuticals and is 21 Kao GF. Carcinoma arising in Bowen’s disease. Arch Dermatol 1986; an advisor to Novartis (U.K.). 122:1124–6. C.A.M. has received sponsorship for speaking and chairing meetings 22 Beck DE, Fazio VW, Jagelman DG, Lavery IC. Perianal Bowen’s dis- from Galderma (Metvix , a brand of methyl aminolaevulinic acid), ease. Dis Colon Rectum 1988; 31:419–22. and sponsorship from 3M and Leo Pharmaceuticals. He has performed 23 Sarmiento JM, Wolff BG, Burgart LJ et al. Perianal Bowen’s disease: sponsored as well as unsponsored research to evaluate the potential of associated tumors, human papillomavirus, surgery, and other con- topical PDT using various photosensitizers in dermatological indica- troversies. Dis Colon Rectum 1997; 40:912–8. tions. 24 Marchesa P, Fazio VW, Oliart S et al. Perianal Bowen’s disease: a clinicopathological study of 47 patients. Dis Colon Rectum 1997; 40:1286–93. References 25 Cox NH. Bowen’s disease: where now with therapeutic trials? Br J 1 Griffiths CEM. The British Association of Dermatologists guidelines Dermatol 2000; 143:699–700. for the management of skin disease. Br J Dermatol 1999; 141:396–7. 26 Bell HK, Rhodes LE. Bowen’s disease – a retrospective review of 2 Cox NH, Williams HC. The British Association of Dermatologists clinical management. Clin Exp Dermatol 1999; 24:338–9. therapeutic guidelines: can we AGREE? Br J Dermatol 2003; 148:621–5. 27 Ball SB, Dawber RPR. Treatment of cutaneous Bowen’s disease with 3 Cox NH, Eedy DJ, Morton CA. Guidelines for management of particular emphasis on the problem of lower leg lesions. Australas J Bowen’s disease. Br J Dermatol 1999; 141:633–41. Dermatol 1998; 39:63–70. 4 Bowen JT. Precancerous dermatoses: a study of two cases of chron- 28 Welch ML, Grabski WJ, McCollough ML et al. 5-Fluorouracil iont- ic atypical epithelial proliferation. J Cutan Dis 1912; 30:241–55. ophoretic therapy for Bowen’s disease. J Am Acad Dermatol 1997; 5 Lee M-M, Wick MM. Bowen’s disease. CA Cancer J Clin 1990; 36:956–8. 40:237–42. 29 Wang KH, Fang JY, Hu CH, Lee WR. Erbium:YAG laser pretreat- 6 Anonymous. Management of Bowen’s disease of the skin. Drug Ther ment accelerates the response of Bowen’s disease treated by topical Bull 2004; 42:13–16. 5-fluorouracil. Dermatol Surg 2004; 30:441–5. 7 Chute CG, Chuang TY, Bergstralh EJ, Su WP. The subsequent risk 30 Salim A, Leman JA, McColl JH et al. Randomized comparison of of internal cancer with Bowen’s disease. A population-based study. photodynamic therapy with topical 5-fluorouracil in Bowen’s dis- JAMA 1991; 266:816–9. ease. Br J Dermatol 2003; 148:539–43. 8 Eedy DJ, Gavin AT. Thirteen-year retrospective study of Bowen’s 31 Bargman H, Hochman J. Topical treatment of Bowen’s disease with disease in Northern Ireland. Br J Dermatol 1987; 117:715–20. 5-fluorouracil. J Cutan Med Surg 2003; 7:101–5. 9 Cox NH. Body site distribution of Bowen’s disease. Br J Dermatol 32 Stone N, Burge S. Bowen’s disease of the leg treated with 1994; 130:714–6. weekly pulses of 5% fluorouracil cream. Br J Dermatol 1999; 10 Eedy DJ. Summary of inaugural meeting of the Skin Care in Organ 140:987–8. Recipients Group, UK, held at the Royal Society of Medicine, 7 33 Patel GK, Goodwin R, Chawla M et al. Imiquimod 5% cream October 2004. Br J Dermatol 2005; 153:6–10. monotherapy for cutaneous squamous cell carcinoma in situ

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 20 Guidelines for management of BD: 2006 update, N.H. Cox et al.

(Bowen’s disease): a randomised, double-blind, placebo-controlled 54 Morton CA, Whitehurst C, McColl JH et al. Photodynamic therapy trial. J Am Acad Dermatol 2006; 54:1025–32. for large or multiple patches of Bowen disease and basal cell carci- 34 Mackenzie-Wood A, Kossard S, de Launey J et al. Imiquimod 5% noma. Arch Dermatol 2001; 137:319–24. cream in the treatment of Bowen’s disease. J Am Acad Dermatol 2001; 55 Wong TW, Sheu HM, Lee JY, Fletcher RJ. Photodynamic therapy 44:462–70. for Bowen’s disease (squamous cell carcinoma in situ) of the digit. 35 Chen K, Shumack S. Treatment of Bowen’s disease using a cycle Dermatol Surg 2001; 27:452–6. regimen of imiquimod 5% cream. Clin Exp Dermatol 2003; 28 56 Morton CA, Horn M, Leman J et al. Comparison of topical methyl- (Suppl. 1):10–12. aminolevulinate photodynamic therapy with cryotherapy or fluor- 36 Kossard S. Treatment of large facial Bowen’s disease: case report. ouracil for treatment of squamous cell carcinoma in situ: results of Clin Exp Dermatol 2003; 28 (Suppl. 1):13–15. a multicenter randomized trial. Arch Dermatol 2006; 142:729–735. 37 Arlette JP. Treatment of Bowen’s disease and erythroplasia of 57 Morton CA, Whitehurst C, Moore JV, MacKie RM. Comparison of Queyrat. Br J Dermatol 2003; 149 (Suppl. 66):43–7. red and green light in the treatment of Bowen’s disease by photo- 38 Prinz BM, Hafner J, Dummer R et al. Treatment of Bowen’s disease dynamic therapy. Br J Dermatol 2000; 143:767–72. with imiquimod 5% cream in transplant recipients. Transplantation 58 Dijkstra AT, Majoie IM, van Dongen JW et al. Photodynamic ther- 2004; 77:790–1. apy with violet light and topical d-aminolaevulinic acid in the 39 Smith KJ, Germain M, Skelton H. Bowen’s disease (squamous cell treatment of actinic keratosis, Bowen’s disease and basal cell carci- carcinoma in situ) in immunosuppressed patients treated with imiq- noma. J Eur Acad Dermatol Venereol 2001; 15:550–4. uimod 5% cream and a COX inhibitor, sulindac: potential applica- 59 Dragieva G, Hafner J, Dummer R et al. Topical photodynamic ther- tions for this combination of immunotherapy. Dermatol Surg 2001; apy in the treatment of actinic keratoses and Bowen’s disease in 27:143–6. transplant recipients. Transplantation 2004; 77:115–21. 40 Smith KJ, Germain M, Skelton H. Squamous cell carcinoma in situ 60 Blank AA, Schnyder UW. Soft-X-ray therapy in Bowen’s disease (Bowen’s disease) in renal transplant patients treated with 5% imiq- and erythroplasia of Queyrat. Dermatologica 1985; 171:89–94. uimod and 5% fluorouracil therapy. Dermatol Surg 2001; 27:561–4. 61 Dupree MT, Kiteley RA, Weismantle K et al. Radiation therapy for 41 Plaza de Lanza M, Ralphs I, Dawber RPR. Cryosurgery for Bowen’s Bowen’s disease: lessons for lesions of the lower extremity. JAm disease of the skin. Br J Dermatol 1980; 103 (Suppl. 18):14. Acad Dermatol 2001; 45:401–4. 42 Holt PJ. Cryotherapy for skin cancer: results over a 5-year period 62 Chung YL, Lee JD, Bang D et al. Treatment of Bowen’s disease with using liquid nitrogen spray cryosurgery. Br J Dermatol 1988; a specially designed radioactive skin patch. Eur J Nucl Med 2000; 119:231–40. 27:842–6. 43 Cox NH, Dyson P. Wound healing on the lower leg after radio- 63 Tantikun N. Treatment of Bowen’s disease of the digit with carbon therapy or cryotherapy of Bowen’s disease and other malignant dioxide laser. J Am Acad Dermatol 2000; 43:1080–3. skin lesions. Br J Dermatol 1995; 133:60–5. 64 Gordon KB, Garden JM, Robinson JK. Bowen’s disease of the distal 44 Morton CA, Whitehurst C, Moseley H et al. Comparison of photo- digit. Outcome of treatment with carbon dioxide laser vaporiza- dynamic therapy with cryotherapy in the treatment of Bowen’s dis- tion. Dermatol Surg 1996; 22:723–8. ease. Br J Dermatol 1995; 135:766–71. 65 Dave R, Monk B, Mahaffey P. Treatment of Bowen’s disease with 45 Ahmed I, Berth-Jones J, Charles-Holmes S et al. Comparison of cry- carbon dioxide laser. Lasers Surg Med 2003; 32:335. otherapy with curettage in the treatment of Bowen’s disease: a 66 Otani K, Ito Y, Sumiya N et al. Treatment of Bowen disease using prospective study. Br J Dermatol 2000; 143:759–66. the ultrasonic surgical aspirator. Plast Reconstr Surg 2001; 108:68–72. 46 Graham JH, Helwig EB. Bowen’s disease and its relationship to sys- 67 Hiruma M, Kawada A. Hyperthermic treatment of Bowen’s disease temic cancer. Arch Dermatol 1961; 83:76–96. with disposable chemical pocket warmers: a report of 8 cases. JAm 47 Papillon J, Chassard JL. Respective roles of radiotherapy and sur- Acad Dermatol 2000; 43:1070–5. gery in the management of epidermoid carcinoma of the anal mar- 68 Ramrakha-Jones VS, Herd RM. Treating Bowen’s disease: a cost- gin. Dis Colon Rectum 1992; 35:422–9. minimization study. Br J Dermatol 2003; 148:1167–72. 48 Cleary RK, Schaldenbrand JD, Fowler JJ et al. Treatment options for 69 National Institute for Health and Clinical Excellence. IPG155 Photo- perianal Bowen’s disease: survery of American Society of Colon dynamic therapy for non-melanoma skin tumours – guidance and Rectal Surgeons Members. Am Surg 2000; 66:686–8. (http://www.nice.org.uk/page.aspx?o¼IPG155guidance) (last 49 Leibovitch I, Huilgol S, Selva D et al. Cutaneous squamous cell car- accessed 10 July 2006). cinoma in situ (Bowen’s disease): treatment with Mohs’ micro- 70 Ormerod AD. Recommendations in British Association of Derma- graphic surgery. J Am Acad Dermatol 2005; 52:997–1002. tologists guidelines. Br J Dermatol 2005; 153:477–8. 50 Lui H, Hobbs L, Tope WD et al. Photodynamic therapy of multiple nonmelanoma skin cancers with verteporfin and red light-emitting diodes: two-year results evaluating tumor response and cosmetic Appendix 1 Strength of recommendations and a outcomes. Arch Dermatol 2004; 140:26–32. quality of evidence 51 Morton CA, Brown SB, Collins S et al. Guidelines for topical photodynamic therapy: report of a workshop of the British Photoderma- Strength of recommendations tology Group. Br J Dermatol 2002; 146:552–67. A There is good evidence to support the use of the procedure 52 Varma S, Wilson H, Kurwa HA et al. Bowen’s disease, solar keratos- B There is fair evidence to support the use of the procedure es and superficial basal cell carcinomas treated by photodynamic C There is poor evidence to support the use of the procedure therapy using a large-field incoherent light source. Br J Dermatol D There is fair evidence to support the rejection of the use of 2001; 144:567–74. the procedure 53 Clark C, Bryden A, Dawe R et al. Topical 5-aminolaevulinic acid E There is good evidence to support the rejection of the use photodynamic therapy for cutaneous lesions: outcome and comparison of light sources. Photodermatol Photoimmunol Photomed of the procedure 2003; 19:134–41.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 Guidelines for management of BD: 2006 update, N.H. Cox et al. 21

Quality of evidence lin treatment in the 1940s) could also be regarded as this type I Evidence obtained from at least one properly designed, ran- of evidence domized controlled trial III Opinions of respected authorities based on clinical experi- II-i Evidence obtained from well-designed controlled trials ence, descriptive studies or reports of expert committees without randomization IV Evidence inadequate owing to problems of methodology II-ii Evidence obtained from well-designed cohort or case– (e.g. sample size, or length of comprehensiveness of follow control analytical studies, preferably from more than one cen- up, or conﬂicts in evidence) tre or research group aA different system of evidence grading and recommendations II-iii Evidence obtained from multiple time series with or has been adopted for new guidelines,70 but the Therapy without the intervention. Dramatic results in uncontrolled Guidelines and Audit Committee has recommended use of the experiments (such as the results of the introduction of penicil- original grading system in this guideline update.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp11–21 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07517.x Temporal changes in sebum excretion and propionibacterial colonization in preadolescent children with and without acne K. Mourelatos, E.A. Eady, W.J. Cunliffe,* S.M. Clark* and J.H. Cove The Skin Research Centre, Research Institute of Molecular and Cellular Biology, The University of Leeds, Leeds LS2 9JT, U.K. *Department of Dermatology, Leeds General Inﬁrmary, Leeds LS1 3EX, U.K.

Summary

Correspondence Background It is generally accepted that the onset of sebum secretion occurs before J.H. Cove. puberty in boys and girls as a result of increasing androgen output during the E-mail: [email protected] adrenarche. Propionibacteria are part of the commensal skin flora and, in adults, are found in highest numbers in sebum-rich areas of skin such as the face and Accepted for publication 15 December 2005 upper trunk. Previous studies investigating the association between sebum output and propionibacterial population densities have been cross-sectional and have Key words been carried out mainly in adults. acne, longitudinal, propionibacteria, puberty, sebum Objectives The purpose of this study was to examine the association between the onset of sebum secretion and expansion of the propionibacterial flora in a popu- Conflicts of interest lation of early adolescent children aged between 5Æ5 and 12 years, and to evalu- E.A. Eady has acted as a paid consultant for Laboratories Galderma. J.H. Cove received funding ate the temporal relation between the two factors longitudinally. In addition, the for the research carried out in this work from study aimed to evaluate the change with age in sebaceous gland activity and Laboratories Galderma. W.J. Cunliffe has received propionibacterial colonization on the skin and in the nares between children professional fees and travelling expenses from most who developed acne and those who did not. of the companies involved in acne in the UK, in Methods Biannual examinations of volunteers included age, pubertal (Tanner) particular Galderma. The Leeds Foundation for Dermatological Research has also received similar stage, weight and height, lesion counting on the face, propionibacterial coloniza- financial support. tion on the skin surface and in the nares and sebum secretion. A longitudinal analysis based on all observations of each subject throughout the study was applied to examine the change of sebaceous gland activity and propionibacterial colonization with age and pubertal stage. A generalized estimating equation was used with a 0Æ05 level of significance. Results The commencement of sebum production was asynchronous, with only a small number of follicles initially starting to secrete sebum onto the skin surface. The number of secreting follicles and the area of sebum increased with age and pubertal stage (P <0Æ0001, P <0Æ05, respectively). Numbers of propionibacteria on the skin tended to increase after the age of 9 years, but not significantly so. In contrast, numbers of propionibacteria in the nares increased significantly with age (P <0Æ0001) but not with pubertal maturation. Children who developed acne had higher sebum output and propionibacterial densities with increasing age than children who did not develop acne. This effect was significant for the increase of total sebum area with age in pubertal children (P ¼ 0Æ0023), the increase in number of secreting follicles with age (P ¼ 0Æ020) in prepubertal children, and the increase in propionibacteria densities in the nares with age (P ¼ 0Æ0005) in pubertal children. Sebaceous gland activity and propionibacterial numbers on the skin surface remained unchanged with increasing age in children who did not develop acne. Propionibacterial population densities in the nares increased with age regardless of the development of acne.

22 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 Sebum and propionibacteria at onset of acne, K. Mourelatos et al. 23

Conclusions Onset of sebum secretion and consequently expansion of the propionibacterial skin ﬂora occur earlier in children who develop acne than in children of the same age and pubertal status who do not develop acne. These observations suggest that postponing the onset of sebum production or the expansion of the propionibacterial skin ﬂora until after puberty may represent ways of preventing the disease or minimizing its severity. Determinants of propionibacterial colonization on the skin and in the nares may be different.

Acne is a multifactorial disease affecting mainly the piloseba- Kearney et al.11 discovered a linear relationship between the ceous follicles of the face and upper trunk. Acne lesions can lower threshold of propionibacterial population density and be present before any physical signs of puberty, with come- sebum excretion rate (SER) so that a low SER was associated dones detectable in children as young as 7 years.1 It is gener- with a low viable count of propionibacteria whereas a high ally accepted that the vast majority of adolescents suffer from SER was associated with high or low propionibacterial num- acne to some extent. In many the disease is mild, with come- bers. Taken together, these observations suggest that sebum dones only, or comedones plus a few inflammatory lesions. In composition as well as sebum volume may be an important a large survey of Australian school children, the prevalence of determinant of propionibacterial densities. In vitro, propioni- acne (comedonal and/or inflammatory) was zero before age bacteria are not lipid dependent.12 6 years, 3Æ0% among 7–9 year olds, 29Æ6% among 10– Previous studies investigating the association between sebum 12 year olds, 77Æ6% among 13–15 year olds rising to 93Æ2% output and propionibacterial colonization have been cross-sec- among 16–18 year olds.2 tional and have been carried out mainly in adults. This study The onset of acne typically occurs during adrenarche when aimed to investigate the temporal relations between the onset of increased production of androgenic hormones including de- sebum production, adrenal hormone output and propionibacte- hydroepiandrosterone sulphate (DHEAS) by the adrenal cortex rial numbers in preadolescent children with and without acne. stimulates sebum production in both sexes.3,4 Sebum produc- Here we report our findings on the relationship between sebum tion is usually considered to be a prerequisite for the subse- secretion, propionibacterial population densities and acne status. quent hypercornification of the pilosebaceous duct leading to ‘Cutaneous’ propionibacteria are not confined to skin and are micro-comedone and then visible comedone formation. Cuta- ubiquitous members of the resident flora of the anterior nares, neous propionibacteria are probably not involved in comedo- where population densities typically exceed those on skin.10,13 genesis but multiple mechanisms implicate them in the Therefore, the skin and nares were both sampled for propioni- genesis of visible inflammation.5 Propionibacteria are part of bacteria to determine whether the expansion of the nasal flora the normal skin flora and are found in higher numbers in occurs before, concomitantly with, or after the skin flora. sebum-rich areas such as the face and trunk.6 Mean propionibacterial densities from birth to 5 years were found to be 1Æ5 )2 7 Methods log10 CFU cm [95% confidence interval (CI) 1–2Æ25]. In this same longitudinal study of American children, densities Subjects and recruitment fell in early childhood (5–10 years) to a mean of < 1 log10 C- ) FU cm 2 (95% CI 0Æ6–1Æ25) before starting to rise again dur- Children were recruited from three main sources: (i) family ing early puberty and continuing to rise to reach mean members of patients attending acne clinics at the Dermatology )2 densities of 5Æ25 log10 CFU cm (95% CI 4Æ5–6Æ0) in young Outpatient Department of the Leeds General Infirmary (LGI), adulthood (20–25 years). By the age of 11 years, children (ii) local schools, and (iii) advertising in local newspapers. with acne had propionibacterial densities consistent with those Ethical approval was obtained from the West Leeds Research of adult skin, whereas children without acne of the same age Ethics Committee for the recruitment and examination of the )2 8 maintained densities of < 1 log10 CFU cm . volunteer children. Biannual examinations included age, The marked increase in propionibacterial numbers during pubertal stage, weight and height, lesion counting on the face, adolescence,7 a time when sebaceous glands enlarge due to propionibacterial colonization on the skin surface and in the androgenic stimulation, suggests that the presence of sebum is nares and sebum secretion. important in the distribution and density of propionibacteria on human skin. Oral isotretinoin markedly suppresses sebum Inclusion criteria secretion and this suppression is typically followed by a 90– 99% decrease in the number of viable propionibacteria on the To be included in the study, volunteers were required to fulfil skin surface suggesting the dependency of the organisms on the following criteria: (i) they had attained Tanner stage 1 or sebum as a nutrient source.9 Site-to-site variation in propioni- 214 at the time of their first appointment, and (ii) they were bacterial population densities and correlation with the amount not on treatment for acne at the time of their first of sebum produced at each supports this interpretation.10 appointment or throughout the study.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 24 Sebum and propionibacteria at onset of acne, K. Mourelatos et al.

light source was used to illuminate Sebutapes during pho- Age tography. The Image Pro-Plus program was used to analyse Age was recorded twice yearly for each child. Each assessment the Sebutapes. Measurements were made of the number of was carried out either in the 6 months before the child’s sebum dots and area of each dot in units of mm2 in a defined birthday or in the 6 months following their birthday. area of the Sebutape and reported per cm2. The computer program traced around the area of each sebum dot, determined by the intensity range of pixel chosen and the mini- Pubertal stage mum area of interest. An intensity range of 0 (black) to 215 Pubertal stage was defined using self-evaluation. Each child (227 was white) pixels was used. was presented with a diagram as described by Morris and Udry15 and asked to select the picture on the diagram that Statistical analyses best described their stage of development. The five pictures on the diagram corresponded to the five Tanner stages. For all the Statistical analyses were performed using STATA 8Æ0. The analyses pertaining to puberty, the cohort of children was study was longitudinal; therefore repeated observations within divided into prepubertal and pubertal. Children at Tanner each subject were likely to be dependent. In order to account stage 1 were categorized as prepubertal, children at Tanner for this dependency a longitudinal analysis based on all obser- stage 2 and above were categorized as pubertal. vations of each subject throughout the study was applied to examine the change of sebaceous gland activity and propionibacterial colonization on the skin surface and in the nares with Lesion counting age in the entire cohort of children. Longitudinal analysis was Inflamed and noninflamed lesions were counted on the whole also used to evaluate the change of sebaceous gland activity face according to the Leeds technique16 using a Brighton and propionibacterial colonization on the skin surface and in 1001 circular fluorescent lamp with a circular Sylvania warm the nares with age between children who developed acne and white tube, placed approximately 30 cm from the patient. those who did not. This analysis was carried out separately for One trained investigator (K.M.) carried out all acne assess- prepubertal and pubertal children. A generalized estimating ments and lesion counts throughout the study. Acne was arbi- equation (GEE) was used. A 0Æ05 level of significance was trarily defined as the presence of five or more inflamed used for all the analyses. lesions, as used previously by Poli et al.17 Diagnosis of acne was coded as a dichotomous independent variable (0 and 1). The total number and total area of active sebum follicles and the total viable counts on the skin surface Microbiology and in the nares independently served as continuous variables Propionibacteria were obtained from the skin surface (right for sebaceous gland activity and propionibacterial colonization. cheek) by a detergent scrub technique18 and from the nares In GEE analysis, the exchangeable correlation matrix for repea- with a moistened swab as described by Coates et al.13 Viable ted measures was assumed and robust estimator of variance of counts were obtained by plating decimal dilutions of each estimated parameters was applied. A log link and gamma fam- sample onto agar plates containing 2% tryptone, 1% yeast ily specification was assumed to model the GEE analysis. An extract agar (both from Oxoid, Basingstoke, Hants, U.K.), interaction term between acne status and age was defined to ) 0Æ5% glucose and 2 lgmL 1 furazolidone to inhibit the examine that sebaceous gland activity and propionibacterial growth of staphylococci. After 7 days of anaerobic incubation colonization varied by the acne status as children became at 37 C, colonies were counted to obtain numbers of viable older. Pubertal stage-related differences in the longitudinal )2 organisms. The limit of detection was 0Æ6 log10 CFU cm study were evaluated at ages 9–11 years using the nonpara- skin and 1 log10 CFU per nasal sample. metric Mann–Whitney U-test. The ratio of pubertal children to prepubertal children at ages 5Æ5–8Æ5 and 11Æ5–14 was very small or large, respectively, for comparisons between the Sebum secretion groups to be carried out at these ages. Sebum secretion was assessed using a lipid-absorbent tape (Se- butape). The tape was applied for 1 h to the forehead of Results each volunteer, after degreasing with an isopropanol-impreg- nated swab. Parents were asked to perform this test at home Volunteer characteristics as per the manufacturer’s instructions after a demonstration during their initial appointment. All Sebutapes were photo- One hundred and thirty-three children were included in the graphed using a Nikon Coolpix 995 digital camera (3Æ34 study and followed up for 2Æ5 years; 82 were male and 51 megapixels). The images were imported into Image Pro Plus were female. The age range of the cohort at baseline was 5Æ5– for image analysis. All Sebutapes were captured without 12 years. Demographic information on the cohort and details flash, in black and white, in fine mode, and aperture priority of the number of children at each developmental stage are with matrix metering. A SCHOTT KL 1500, 150-W, halogen shown in Table 1.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 Sebum and propionibacteria at onset of acne, K. Mourelatos et al. 25

Table 1 Demographic data for children entering the study children with ‡100 sebum-secreting follicles but no detectable propionibacteria. In contrast, 25Æ5% of all observations were

All children (n ¼ 133) Prepubertal (n) Pubertal (n) from children in whom up to 4Æ0 log10 CFU propionibacte- )2 Boys 82 73 7 ria cm skin were detected in the complete absence of sebum Girls 51 42 9 as detected by Sebutape. At all ages, mean population densities were low and well below expected adult values. n is the number of children in each group. The number of chil- In the nares, there was a signiﬁcant association between dren in the individual subgroups did not always add up to the propionibacterial density and age (P <0Æ001, Fig. 1d). With total number of children because of missing data. Two males were missing pubertal stage data. the exception of certain outliers, population densities in children younger than 8Æ5 years were always low (< 2 log10 CFU per nasal sample). Older children showed densities that were

either high or low, reaching densities of 6 log10 CFU per nasal sample. The significant association was observed in both Age-related changes in the number of sebum-secreting males and females when the analysis was carried out separate- follicles and area of sebum ly for each sex (data not shown). Above the age of 10 years, Activation of sebaceous glands was a progressive process with mean population densities in the nares were always higher the number of actively secreting follicles and the total volume than on skin. of sebum secreted increasing with increasing age. Sebum output was always low until the age of 8 years whereas above Pubertal stage-related differences in the number of that age it was either low or high (Fig. 1a,b). The association active sebaceous follicles and the area of sebum of both total area and total number of active sebaceous folli- secreted cles with age was significant (P <0Æ0001) when a longitudinal analysis was employed, and this significance was observed Significant differences in the number of sebaceous follicles in both males and females when the analysis was carried out and the total area of sebum secreted were detected between separately for each sex (data not shown). prepubertal and pubertal children at ages 10Æ5 and 11 years (P <0Æ05) (Fig. 2a,b, respectively). When stratifying for sex, it was revealed that only males contributed to the differences Age-related changes in propionibacterial numbers on the observed in the total number of active sebum follicles at age skin and in the nares 10Æ5 years and to the differences observed in the number of There was extensive variation in the population densities of active sebum follicles at age 11 years (data not shown). The propionibacteria on skin at all ages (Fig. 1c). At every age difference observed in the total area of active sebum follicles there were always some children from whom no viable prop- at age 11 years was not apparent when analyses were carried ionibacteria were recovered; 9% of all observations were from out separately in males and females (data not shown).

(a) (b)

Fig 1. The association with age of (a) the total area of active sebum follicles, (b) the total number of active sebum follicles, (c) the propionibacterial densities on the skin surface, and (d) the propionibacterial densities in the (c) (d) nares. Analyses of associations with age were carried out using a longitudinal analysis (generalized estimating equation) and the analyses included all observations from children throughout the study. The mean value for each measurement at each age is displayed (red line). The numbers displayed on the graphs are the numbers of observations at each corresponding age. No summary data were plotted at the ages when the number of observations was less than ﬁve.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 26 Sebum and propionibacteria at onset of acne, K. Mourelatos et al.

(a) (b)

(c)

Fig 2. Puberty-related differences in (a) the total number of active sebum follicles, (b) the total area of active sebum follicles, (c) the propionibacterial densities on the skin surface, and (d) the propionibacterial densities in the nares. The analyses were carried out at ages 9, 9Æ5, 10, 10Æ5 and 11 years and differences between (0) prepubertal and (1) pubertal children were evaluated using the Mann–Whitney U-test. The box plots display the median value for each assessment with quartiles. N is the number of children at each pubertal stage at each age.

We then examined the effect of acne on the changes in sebum Pubertal stage-related differences in the population secretion and propionibacterial colonization separately in pre- densities of propionibacteria on the skin and in the pubertal and pubertal children in order to determine whether nares earlier onset of puberty was responsible for the observed dif- No significant difference was found at any age between pre- ferences. Prepubertal (Fig. 4) and pubertal children (Fig. 5) pubertal and pubertal children in population densities of prop- who developed acne had higher mean values for sebum out- ionibacteria on the skin or in the nares (Fig. 2c,d). put and propionibacterial densities compared with children who did not develop acne. This was confirmed by the positive coefficients produced by the longitudinal analysis (GEE) for Comparison between children who developed acne and the term Acne*Age, indicating that the effect of acne on the those who did not change in sebaceous gland activity and propionibacterial col- When changes with age were analysed separately for children onization was a positive one, i.e. sebaceous gland activity and who developed acne and children who did not we found that propionibacterial colonization on the skin surface and in the the observed age-related increases in sebum output and prop- nares was always higher in the group of children who devel- ionibacterial colonization were accounted for mainly by chil- oped acne compared with those who did not. This effect was dren who developed five or more inflamed lesions by the end not always significant (Table 2). of the study (Fig. 3a–d). In children who had not developed acne by the end of the study, there was virtually no increase Comparison of the number of sebum-secreting follicles in sebum output up to the age of 11Æ5 years. On the skin, split by acne and pubertal status there was no increase in propionibacterial numbers in non- acne children but in those who did develop acne, numbers in- We compared the number of secreting follicles in the children creased from the age of 9Æ5 years. In the nares, propionibacte- split by acne status and pubertal status (Table 3). Prepubertal rial numbers increased from age 8 years in children who children had a lower mean total number of sebum-secreting developed acne and from age 10 years in those who did not. follicles than pubertal children. In addition, children who

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 Sebum and propionibacteria at onset of acne, K. Mourelatos et al. 27

(a) (b)

Fig 3. Age-related differences in acne and non-acne children. The association with age of (a) the total area of active sebum follicles, (b) the total number of active sebum follicles, (c) the propionibacterial densities on the skin surface, and (d) the propionibacterial densities in the nares, in children who developed acne in the duration of the study (pink), and in children who did not develop acne in the duration of the study (blue). Analyses of associations with age were carried out using a longitudinal analysis (generalized estimating equation) and the analyses included all observations from children throughout the study. The effect of acne status on the association of each variable with age was demonstrated by introducing an interaction term in the analysis as described in the Methods section. Mean values and SEM are displayed in plots (a–d). The numbers displayed on the graphs are the numbers of observations at each corresponding age for children who developed acne (pink) and those who did not (blue). No summary data were plotted at the ages when the number of observations was less than ﬁve.

(a) (b)

Fig 4. Age-related differences in prepubertal children for those who developed acne (pink) and those who did not (blue). A longitudinal analysis (generalized estimating equation) was carried out to evaluate the effect of the development of acne on the change of the dependent variable (a) total area of active sebum follicles, (b) total number of active sebum follicles, (c) propionibacterial densities on the skin, and (d) propionibacterial densities in the nares, with increasing age. The P-value indicates if the development of acne had a significant effect on age-related differences (P <0Æ05, significance). Mean values and SEM are displayed in plots (a–d). No summary data were plotted at the ages when the number of observations was less than five. developed acne during the study had a higher mean of had a lower mean number of sebum-secreting follicles com- sebum-secreting follicles than children who had not developed pared with pubertal children who had developed acne. acne (Table 3). When the data were analysed further we found that prepubertal children who developed acne had a Temporal relations between sebum output and higher mean total number of sebum-secreting follicles than propionibacterial colonization prepubertal children who had not developed acne (Table 3). Similarly, further analysis of the data for pubertal children Although colonization of the nares was observed in some revealed that pubertal children who have not developed acne individuals before the detection of sebum output, the nasal

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 28 Sebum and propionibacteria at onset of acne, K. Mourelatos et al.

(a) (b)

Fig 5. Age-related differences in pubertal children for those who developed acne (pink) and those who did not (blue). A longitudinal analysis (generalized estimating equation) was carried out to evaluate the effect of the development of acne on the change of the dependent variable (a) total area of active sebum follicles, (b) total number of active sebum follicles, (c) propionibacterial densities on the skin, and (d) propionibacterial densities in the nares, with increasing age. The P-value indicates if the development of acne had a significant effect on age-related differences (P <0Æ05, significance). Mean values and SEM are displayed in plots (a–d). No summary data were plotted at the ages when the number of observations was less than five. propionibacterial population increased in parallel to sebum tors in the vestibular epithelium does not seem to have been output once sebaceous gland activity had started (Fig. 1a,b,d). investigated.20 Factors determining propionibacterial numbers There was an apparent time lag before skin propionibacteria in the nares are presently unknown but must differ to an started to increase in number and densities remained low at extent from those on skin. One difference may be that sebum all ages (Fig. 1c). However, when data were re-analysed for production is triggered by lower levels of DHEAS in nasal as children with and without acne, the rise in propionibacterial opposed to cutaneous sebaceous glands or that the nasal popu- numbers on skin and in the nares occurred in parallel with lation is dependent upon some other nutrient source that the increasing output of sebum. Paradoxically, a small number changes during the adrenarche or early puberty. For the pur- of children aged 7 years or under carried up to 3 log10 CFU poses of this study it was necessary to define acne using a ) propionibacteria cm 2 skin before the onset of detectable number of lesions small enough to detect the onset of the dis- sebaceous gland activity (Fig. 1a,c). ease in a population of children who were mainly prepubertal or early pubertal and thus had a smaller number of inflamed Discussion lesions than what would be found in a population of adolescent children or teenagers. A number of five inflamed lesions This study has shown for the first time that maturity-related also minimized the possibility of a false-positive classification. increases in sebum output and propionibacterial population No further acne severity classifications (mild, moderate, densities in children who develop acne occur in parallel from severe) were made because of the potential small numbers of the age of 8 years. The onset of sebum production that pre- children in the moderate and severe classes. Important varia- sumably triggers the expansion of the propionibacterial flora tions exist in the literature for the definition of acne.21 Lucky occurs sooner in children who develop acne than in those et al.1 distinguished between comedonal and inflammatory who do not. Children who had not developed acne by the acne in young adolescent girls and defined mild, moderate end of the study showed no increase in sebum output or and severe comedonal and inflammatory acne as up to 10, 25 propionibacterial numbers on skin, but did show a slight and > 25 lesions, respectively. increase in propionibacterial numbers in the anterior nares. Regardless of pubertal status, individuals with acne presen- The nares are lined with a stratified squamous epithelium that, ted with higher sebaceous gland activity and higher propioni- like skin, contains sebaceous and sweat glands.19 In children bacterial densities than individuals without acne. The who developed acne, propionibacterial population densities differences were more marked in the pubertal group. This in- rose more sharply with age in the nares than on skin suggest- dicated that pubertal status as determined by self-examination ing that the nasal epithelium also responds to rising androgen was not the main determinant of the differences observed levels. In animals, it is well recognized that the nasal mucosa between individuals with and without acne. contains receptors involved in mediating responses to steroid Why might sebum output start to rise sooner in acne-prone pheromones, but in humans the presence of androgen recep- than non-acne-prone children? Several years ago, Lucky et al.1

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 Sebum and propionibacteria at onset of acne, K. Mourelatos et al. 29

) Table 2 Longitudinal analysis of the effect of acne on the assessed Table 3 Mean number of sebum-secreting follicles (cm 2) split by variables in prepubertal and pubertal children with increasing age pubertal and acne status with coefficients and statistical significance Groups of volunteers Mean number of sebum-secreting ) Variable (n) Coefficient Statistical significance (P) investigated (n) follicles cm 2 skin (SEM) Total area of active sebum follicles Prepubertal (358) 4Æ19 (0Æ83) Prepubertals (336) Early pubertal (99) 27Æ9(4Æ18) Acnea 15Æ04 0Æ197 Acne (82) 26Æ8(4Æ18) Ageb 1Æ90 < 0Æ0001* Non-acne (408) 4Æ41 (1Æ00) Acne*Agec 0Æ83 0Æ396 Prepubertal/acne (27) 8Æ1(1Æ84) Pubertals (145) Prepubertal/non-acne (328) 3Æ35 (1Æ06) Acnea 187 691Æ6<0Æ0001* Early pubertal/acne (39) 40Æ7 (11Æ73) Ageb 5Æ16 < 0Æ0001* Early pubertal/non-acne (60) 8Æ93 (2Æ70) Acne*Agec 0Æ37 < 0Æ0001* Total number of active sebum follicles n, the number of observations included in each group. Prepubertals (333) Acnea 736Æ02 0Æ001* Ageb 2Æ24 < 0Æ0001* Acne*Agec 0Æ53 0Æ002* critical concentration may differ between children. Alternat- Pubertals (137) ively, DHEAS may act in concert with another mediator, the Acnea 4493Æ316 0Æ002* concentration of which also changes during the adrenarche or Ageb 3Æ41 < 0Æ0001* puberty. In our study, we were unable to collect blood sam- c Acne*Age 0Æ511 0Æ006* ples but we did collect saliva and this has been used for the Skin propionibacteria estimation of DHEAS. The results of this analysis will be pub- Prepubertals (352) lished separately. Essentially they show that sebum output rises Acnea 0Æ048 0Æ266 Ageb 0Æ86 0Æ008 with age in parallel with increases in salivary DHEAS concen- Acne*Agec 1Æ38 0Æ233 tration and that DHEAS concentration was significantly associ- Pubertals (149) ated with acne status and with pubertal status in both sexes. Acnea 39Æ11 0Æ132 Whatever the reason, the absence of acne in children in the b Age 1Æ48 0Æ058 study with little or no sebum was not surprising. The group c Acne*Age 0Æ75 0Æ189 of children who had not developed acne by the end of the Nares propionibacteria study may have contained some who will go on to develop Prepubertals (348) Acnea 0Æ135 0Æ103 the disease at a later age or developmental stage. Thus this Ageb 1Æ17 0Æ061* group contains a mixed population of children whereas the Acne*Agec 1Æ31 0Æ043* acne group is homogeneous with respect to their disease sta- Pubertals (152) tus. a ) Acne 512Æ48 0Æ004* Comparing the number of secreting follicles cm 2 in the b Age 1Æ86 < 0Æ0001* children, split by acne status and pubertal stage, showed that Acne*Agec 0Æ599 0Æ006* the highest mean density of such follicles was present in early *Significant P-values; n, the number of observations included in pubertal children with acne (40Æ7±11Æ7) and the lowest was each analysis. in prepubertal children without acne (3Æ4±1Æ1). According aThe effect of acne on the assessed variable when age was kept to Pierard et al.,22 the mean density of sebum-secreting folli- ) constant. cles in adults is 196 ± 63 cm 2 on the forehead, compared b The effect of age on the assessed variable when acne status was with the mean density of pilosebaceous follicles of 292– kept constant. ) 455 cm 2 at the same site.23–25 Thus, even in early pubertal cThe effect of acne on the change in the assessed variable with increasing age. children with acne, only a minority of follicles were apparently switched on by the end of the study and, even in adults, not all follicles appear to be secreting sebum at any showed that girls who develop severe comedonal acne have one time. Extrapolating from our observations, it appears that higher levels of serum DHEAS but similar levels of testosterone it may take a decade or more before all follicles capable of before and after the menarche compared with girls with mild producing sebum are actually doing so. This fits very well comedonal acne or no acne. Levels in both groups were with studies that examined changes in the gross amount of ) ) within the normal age-adjusted range. The biggest difference sebum secreted cm 2 h 1 and which found that the sebum between the two groups was 2 years before menarche sug- excretion increased from the age of 8 or 9 years for approxi- gesting that the presence of high levels of DHEAS at an earlier mately 10 years. The difference is how these data can now be developmental stage may be associated with earlier onset of interpreted. Rather than a model in which all follicles turn on sebum secretion. Perhaps DHEAS has to reach a critical con- at once and produce more sebum over time, we now have an centration before triggering the production of sebum; this alternative scenario in which follicles switch on sequentially

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 30 Sebum and propionibacteria at onset of acne, K. Mourelatos et al. throughout the acne-prone years. All follicles apparently that in the axilla, and in men before hair on the torso.28 Acne become capable of secreting sebum at or around the age of often begins on the forehead and mid-face so that differences 20 years,26 probably with considerable person-to-person vari- in androgen sensitivity between areas of the face and between ation. This time line may be extended even further in individ- the face and trunk may predispose to earlier sebum produc- uals without acne. This model would predict that follicles may tion and earlier acne at some sites. be particularly acne-prone when sebum secretion is switching Although we followed each child for only 2Æ5 years, it is on or switching off, a process that will occur even in adults unlikely that we could obtain meaningful data from many of unless the discrepancy between the number of secreting folli- the children at subsequent visits. As we have shown, a propor- cles and the total number of follicles is entirely due to the co- tion have now developed inflammatory acne and will inevit- alescence of sebum output from adjacent follicles.27 ably wish to seek treatment for this. Although our data show that the presence of propionibacte- In summary, we have shown that acne-prone children ria is not entirely dependent on sebum secretion, sebum output begin to secrete sebum earlier than children without acne and and propionibacterial numbers increased in parallel. This obser- that this results in earlier expansion of the propionibacterial vation supports earlier studies showing an association between flora, first in the nares, then on skin. These changes were not the two variables. Just as sebum secretion had not reached adult explained by earlier onset of puberty. Individual follicles do values by the end of the study, neither had mean propionibac- not start to produce sebum simultaneously and it may take up terial population densities on skin or in the nares, although to 10 years for an adult pattern of sebum secretion to be adult levels may have been attained by some children at either reached. These observations suggest that one way of prevent- or both sites. At age 12Æ5 years, the mean number of propioni- ing acne may be to delay the onset of sebum secretion or, )2 bacteria was 1Æ0 log10 cm skin compared with an expected more simply, to postpone the increase in propionibacterial )2 value in excess of 5Æ0 log10 cm for a sample of adults (acne numbers until Tanner stage 5 has been attained. and non-acne combined).6 In the nares, the situation was sim- )2 ilar, a mean value of 2Æ5 log10 cm in children compared with )2 12 Acknowledgments expected adult values in excess of 5Æ0 log10 cm . In both cases, values represent < 1% of adult levels. The presence The authors wish to thank Miss Faye Turner and Dr Yaojun Li within the cohort of some children with high numbers of pro- for statistical advice, and the families of all the children who pionibacteria but little or no sebum suggests that the composi- participated in the study. The study was financially supported tion as well as the amount of sebum may be important but also by Laboratoires Galderma and the Leeds Foundation for Der- that skin can maintain a self-sustaining resident population of matological Research. propionibacteria in the absence of sebum, at least in some individuals. Epidermal lipids may represent an alternative nutrient References source, boosted by sebum. This fits with the obvious fact that pilosebaceous follicles must be colonized initially from the 1 Lucky AW, Biro FM, Simbartl LA et al. Predictors of severity of acne population living on the skin surface. vulgaris in young adolescent girls: results of a five-year longitud- It is almost 30 years since Leyden et al.8 showed that expan- inal study. J Pediatr 1997; 130:30–9. 2 Kilkenny M, Merlin K, Plunkett A, Marks R. The prevalence of sion of the propionibacterial flora occurs much earlier in common skin conditions in Australian school students. 3. Acne acne-prone subjects than in subjects without acne so that adult vulgaris. Br J Dermatol 1998; 139:840–5. levels are reached in the former group by age 15 years but 3 Stewart ME, Downing DT, Cook JS et al. Sebaceous gland activity not until age 25 years in the latter group. We have now con- and serum dehydroepiandrosterone sulphate levels in boys and firmed that propionibacterial numbers do indeed rise earlier in girls. Arch Dermatol 1992; 128:1345–8. children with acne, and have shown that this is due to the 4 Yamamoto A, Ito M. Sebaceous gland activity and urinary andro- earlier onset of sebum production. This offset of both sebum gen levels in children. Dermatol Sci 1992; 4:98–104. 5 Farrar MD, Ingham E. Acne: inflammation. Clin Dermatol 2004; production and the increase in propionibacterial population 22:380–4. density between acne-prone and non acne-prone children sug- 6 McGinley KJ, Webster GF, Ruggieri MR, Leyden JJ. Regional varia- gests that the timing of these events may be critical in influen- tions in density of cutaneous propionibacteria: correlation of Prop- cing acne susceptibility and raises the obvious question as to ionibacterium acnes population with sebaceous secretion. J Clin Microbiol whether delaying either until after puberty may prevent acne 1980; 12:672–5. or markedly reduce its severity. 7 Leyden JJ, McGinley KJ, Mills OH, Kligman AM. Age-related In the case of propionibacteria, the fact that the population changes in the resident bacterial flora of the human skin. J Invest Dermatol 1975; 65:379–81. expands in the nares earlier than it does on forehead skin sug- 8 Leyden JJ, McGinley KJ, Mills OH, Kligman AM. Propionibacterium gests that there may also be a difference between the face and levels in patients with and without acne vulgaris. J Invest Dermatol the trunk, possibly explaining why truncal acne often starts 1975; 65:382–4. later. It would be interesting to assess whether sebum secre- 9 King K, Jones DH, Daltrey DC, Cunliffe WJ. A double-blind study tion begins simultaneously on the face and trunk. In the case of the effects of 13-cis-retinoic acid on acne, sebum excretion rate of sexual hair, there is a known gradient of androgen sensitiv- and microbial population. Br J Dermatol 1982; 107:583–90. ity, determined in utero, whereby pubic hair appears before

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 Sebum and propionibacteria at onset of acne, K. Mourelatos et al. 31

10 McGinley KJ, Webster GF, Leyden JJ. Regional variations of cutane- 20 Barni T, Maggi M, Fantoni G et al. Sex steroids and odorants modu- ous propionibacteria. Appl Environ Microbiol 1978; 35:62–6. late gonadotrophin-releasing hormone secretion in primary cul- 11 Kearney JN, Ingham E, Cunliffe WJ, Holland KT. Correlations tures of human olfactory cells. J Clin Endocrinol Metab 1999; between human skin bacteria and skin lipids. Br J Dermatol 1984; 84:4266–73. 110:593–9. 21 Dreno B, Poli F. Epidemiology of acne. Dermatology 2003; 206:7– 12 Gribbon EM, Shoesmith JG, Cunliffe WJ, Holland KT. The micro- 10. aerophily and photosensitivity of Propionibacterium acnes. J Appl Bacteriol 22 Pierard GE, Pierard-Franchimont C, Le T. Patterns of follicular 1994; 77:583–90. sebum excretion rate during lifetime. Arch Dermatol Res 1987; 13 Coates P, Vyakrnam S, Eady EA et al. Prevalence of antibiotic-resist- 279:S104–7. ant propionibacteria on the skin of acne patients: 10-year surveil- 23 Otberg N, Richter H, Schaefer H et al. Variations of hair follicle size lance data and snapshot distribution study. Br J Dermatol 2002; and distribution in different body sites. J Invest Dermatol 2004; 146:840–8. 122:14–19. 14 Tanner JM. Growth at Adolescence, 2nd edn. Oxford: Blackwell Scienti- 24 Pagnoni A, Kligman AM, Gammal S, Stoudemayer T. Determination ﬁc Publications, 1962. of density of follicles on various regions of the face by cyanoacry- 15 Morris NM, Udry RJ. Validation of a self-administered instrument late biopsy: correlation with sebum output. Br J Dermatol 1994; to assess stage of adolescent development. J Youth Adolesc 1980; 131:862–5. 9:271–80. 25 Blume U, Verschoore M, Poncet M et al. The vellus hair follicle in 16 Burke BM, Cunliffe WJ. The assessment of acne vulgaris—the Leeds acne: hair growth and sebum excretion. Br J Dermatol 1993; technique. Br J Dermatol 1984; 111:83–92. 129:23–7. 17 Poli F, Dreno B, Verschoore M. An epidemiological study of acne 26 Pochi PE, Strauss JS, Downing DT. Skin surface lipid composition, in female adults: results of a survey conducted in France. J Eur Acad acne, pubertal development, and urinary excretion of testosterone Dermatol Venerol 2001; 15:541–5. and 17-ketosteroids in children. J Invest Dermatol 1977; 69:485–9. 18 Williamson P, Kligman AM. A new method for the quantitative in- 27 Pierard GE. Rate and topography of follicular sebum excretion. Der- vestigation of cutaneous bacteria. J Invest Dermatol 1965; 45:498– matologica 1987; 175:280–3. 503. 28 Rosenﬁeld RL. Pilosebaceous physiology in relation to hirsutism 19 Kluytmans J, van Belkum A, Verbrugh H. Nasal carriage of Staphylo- and acne. Clin Endocrinol Metab 1986; 15:341–62. coccus aureus: epidemiology, underlying mechanisms, and associated risks. Clin Microbiol Rev 1997; 10:505–20.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp22–31 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07536.x The role of heat shock protein 60, vascular endothelial growth factor and antiphospholipid antibodies in Behc¸et disease O. Shaker, M.A. Ay El-Deen,* H. El Hadidi, B.D. Grace, H. El Sherif§ and A. Abdel Halim* Departments of Medical Biochemistry *Vascular Surgery, Dermatology, Internal Medicine and §Rheumatology, Faculty of Medicine, Kasr El Aini Hospital, Cairo University, Cairo, Egypt

Summary

Correspondence Background Behçet disease is a systemic inflammatory disease of unknown aetiol- O. Shaker. ogy. T cells in this disease proliferate vigorously in response to a specific peptide E-mail: [email protected] of heat shock protein (HSP) 60 in an antigen-specific fashion. Vascular endothelial cell growth factor (VEGF) is a cytokine participating in the inflammatory pro- Accepted for publication 12 May 2006 cess. One of the prominent features of Behçet disease is vasculitis as a result of endothelial dysfunction. Antiphospholipid antibodies (APA) may play a role in Key words the development of thrombosis by inhibiting production of prostacyclin by antiphospholipid antibodies, Behçet disease, heat endothelial cells. shock protein 60, vascular endothelial cell growth Objectives To investigate the role of HSP60, VEGF and APA in Behçet disease and factor their relation to clinical manifestations and disease activity. Conflicts of interest Methods Thirty patients with Behçet disease were included; 17 were in the active None declared. stage and 13 were in the inactive. Fifteen age- and sex-matched healthy subjects served as controls. Complete clinical examination and Doppler examination were done. Serum levels of HSP60, VEGF and APA were performed. Results Serum levels of HSP60, VEGF and APA were significantly higher in patients than in controls; however, their level did not correlate with disease activity. The serum level of VEGF correlated significantly with the presence of vascular manifestations and ocular involvement. The serum level of APA was greater in patients with thrombosis. HSP60 has an important role in aetiopathogenesis of Behçet disease, which sheds new light on its autoimmune nature. Conclusions An elevated serum level of VEGF may be a risk factor for the development of ocular disease contributing to poor visual outcome.

Behçet disease is a systemic disorder of recurrent acute inflam- from mammalian/bacterial 60/65-kDa HSP. Sequence homol- mation characterized by major symptoms: genital ulcers, aph- ogy and cross-reactivity between microbial and human HSP thous ulcers, uveitis and skin lesions. Involvement of the led to the concept that HSP might be involved in the aetio- intestines, vessels and central nervous system sometimes leads pathogenesis of Behçet disease.6 to poor prognosis.1 Besides genetic, environmental and immu- Vascular endothelial growth factor (VEGF) is a 34- to 45- nological factors, microbial agents have been implicated in its kDa glycoprotein with similarity to platelet-derived growth aetiology.2,3 factor. It is a potent cytokine that modulates angiogenesis and Patients with Behçet disease have been shown to have anti- vasculogenesis by acting as an essential mitogen for vascular bodies against heat shock protein (HSP) 60, which is one of endothelial cells.7,8 Because VEGF is a potent mitogen for der- the streptococcal proteins.4 HSPs are widely distributed in nat- mal and ocular microvascular endothelial cells,9,10 its expres- ure; they perform important functions in the folding and sion may be important in the vascular bed of Behçet patients. unfolding or translocation of protein as well as in the assem- Antiphospholipid antibodies (APA) may have a pathogenetic bly and disassembly of protein complexes. Because of these role in the development of vascular complications in Behçet helper functions, HSPs have been termed molecular chaper- disease. APA represent a heterogenous group of immunoglob- ones.5 HSPs are present in low concentration in normal condi- ulins that includes lupus anticoagulant and anticardiolipin tions while in stressed cells they accumulate at high levels.5 antibodies. They react with negatively charged and rarely HSP60/65 is an immunodominant antigen that is derived neutral phospholipids. They play an important role in the

2006 The Authors 32 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp32–37 Role of HSP60, VEGF and APA in Behçet disease, O. Shaker et al. 33 development of thrombosis by inhibiting production of pro- Both clinical and laboratory findings [erythrocyte sedimen- stacyclin by endothelial cells.11 tation rate (ESR), neutrophil count] were used to classify the patients as having active (n ¼ 17) or inactive (n ¼ 13) dis- Materials and methods ease. Clinically, the presence of three of five major findings on admission (skin lesions, pathergy test, uveitis, oral and genital The study was conducted on 30 patients with Behçet disease ulcers) was considered to indicate the active stage of the dis- (satisfying the criteria of the International Study Group for Be- ease.13 Patients showing no clinical or laboratory disorders hçet Disease),12 recruited from the Dermatology and Rheuma- related to Behçet disease for at least 1 month were considered tology outpatient clinics and the Vascular Surgery department, as having inactive disease.14 Kasr El Aini Hospital, Cairo, Egypt, during the period March Fifteen patients (50%) had vascular complications and all 2004 to April 2005. Fifteen healthy volunteers served as con- were men. They suffered more venous than arterial events. trols. There was a high prevalence of deep venous thrombosis (11 Patients ranging in age from 17 to 50 years (mean patients, 36Æ7%). Caval thrombosis involving different seg- 32Æ6±9Æ14) and a male to female ratio of 4 : 1 were inclu- ments of either superior or inferior vena cava was present in ded in this study. The disease duration ranged from 1 to 25 four patients (13Æ3%). Superficial thrombophlebitis occurred (mean 6Æ92 ± 6Æ65) years. Fifteen healthy volunteers served as in four patients. Arterial events were evident in four patients controls. Their ages ranged from 18 to 47 years (mean (13Æ3%); arteritis associated with aneurysm formation in three 30Æ13 ± 12Æ32) with a male to female ratio of 4 : 1. The clin- patients (10%); pulmonary aneurysms complicated with ical manifestations of patients with Behçet disease are shown haemoptysis and pulmonary embolism in two patients; saccu- in Table 1. lar aortic aneurysm leading to occlusion of the left external iliac artery and proximal part of the common femoral artery in one patient (3Æ3%); and arterial thrombosis in one patient Clinical assessment (3Æ3%). Clinical assessment was done for all patients. Dermatological examination for all skin lesions was performed, and ocular Investigations and vascular examinations were done. Routine laboratory investigations were done including com- Table 1 Clinical manifestations of Behçet disease in patients studied plete blood analysis, ESR, liver and kidney function, and urine analysis. Vascular involvement was assessed by Doppler ultra- Item Patients, sound. The following parameters were measured in the sera n ¼ 30 (%) of all subjects. Oral ulcers 30 (100) Genital ulcers 28 (93Æ3) Quantitation of heat shock protein 60 in serum by heat shock protein 60 Skin rashes 16 (53Æ3) enzyme-linked immunosorbent assay kit. The Stressgen Stress Xpress Papulopustular, pseudofolliculitis 16 (53Æ3) HSP60 ELISA (enzyme-linked immunosorbent assay) (Stress- Erythema nodosum 6 (20) gen, Nventa Biopharmaceuticals Corporation, San Diego, CA, Palpable purpura 9 (30) Skin pathergy test 16 (53Æ3) U.S.A.) provides a method to detect and quantitate HSP60 in Eye findings 15 (50) samples from humans. Stressgen’s HSP60 ELISA is a quantita- Anterior uveitis 9 (30) tive sandwich immunoassay. A mouse monoclonal antibody Posterior uveitis 11 (36Æ7) specific for HSP60 is precoated on the wells of the anti- Retinal vasculitis 3 (10) HSP60 immunoassay plate provided with the kit. HSP60 is Optic neuropathy 3 (10) captured by immobilized antibody and is detected with an Papillo-oedema 2 (6Æ7) HSP60-specific, goat polyclonal antibody. The goat polyclonal Vascular findings 15 (50) Venous: superficial thrombosis, 11 (36Æ7) antibody is subsequently bound by a horseradish peroxidase- deep vein thrombosis conjugated antigoat IgG secondary antibody. The assay is Arterial: aneurysms, thrombosis 4 (13Æ3) developed with tetramethylbenzidine substrate and blue col- Cardiac manifestations 6 (20) our develops in proportion to the amount of captured Chest manifestations 8 (26Æ6) HSP60. The intensity of the colour is measured in a micro- Neurological manifestations 14 (46Æ6) plate reader at 450 nm. HSP60 concentrations from the sam- Ataxia 1 (3Æ3) ple are quantitated by interpolating absorbance readings from Migraine 13 (43Æ3) Stroke 4 (13Æ3) a standard curve generated with the calibrated HSP60 protein Pyramidal tract signs 6 (20) standard provided. Cranial nerve 5 (16Æ7) Aseptic meningitis 1 (3Æ3) Quantitation of vascular endothelial cell growth factor by the ACCUCYTE Articular manifestation 12 (40) enzyme-linked immunosorbent assay kit. ACCUCYTE Human (Cytim- mune Sciences, Inc., Rockville, MD, U.S.A.) is a competitive

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp32–37 34 Role of HSP60, VEGF and APA in Behçet disease, O. Shaker et al. enzyme immunoassay. This kit measures natural and recom- comparing qualitative data. Correlation and regression analysis binant forms of cytokine human VEGF. Precoated goat and (r) were also used. antirabbit antibodies are used to capture a specific VEGF complex in each sample. Biotinylated VEGF conjugates (competit- Results ive ligands) and sample or standard form a competition reaction for VEGF-specific antibody-binding sites. The amount The serum levels of HSP60, VEGF and APA in all patients were of biotinylated VEGF is detected by the addition of streptavi- significantly higher than that of controls (P <0Æ0001, din-conjugated alkaline phosphate, which binds only to the <0Æ0001 and < 0Æ01, respectively) (Table 2). biotinylated VEGF, followed by the addition of the colour rea- Seventeen patients were in the active stage of the disease gent solution. whereas the remaining 13 patients were in the inactive stage. The neutrophil count and ESR concentration were significantly Quantitation of phospholipid antibody in serum. This is an enzyme im- higher in patients with active disease than inactive (P <0Æ01, munoassay for the quantitative determination of the sum of <0Æ005, respectively) (Table 3). autoantibodies against cardiolipin, phosphatidyl serine, phos- The serum levels of HSP60, VEGF and APA were higher in phatidyl inositol and phosphatitic acid (IgG class). The anti- the active group than in the inactive one but the results were phospholipid screen is an indirect solid-phase ELISA. The not statistically significant (P >0Æ05). Comparing serum levels microplate is coated with a mixture of highly purified nega- of HSP60, VEGF and APA according to presence or absence of tively charged phospholipids. The kit was provided by BL Di- clinical manifestations showed that the serum level of VEGF agnostika, Mainz, Germany. was significantly higher in patients with vascular manifestations and in those with ocular involvement (P <0Æ01, <0Æ005, respectively) (Table 4). In addition, the serum level Statistical analysis of APA was significantly higher in patients with vascular ) The statistical analysis was done using a Macintosh LCIII com- involvement and thrombosis (15Æ49 GpL mL 1) compared ) puter and Statview statistical package (SAS Institute Inc., Cary, with those patients without vascular findings (9Æ87 GpL mL 1) NC, U.S.A.). Statistical analyses were done according to the (P <0Æ005). method of Knapp and Miller.15 Tests used included Student’s There was no correlation between serum levels of HSP60 t-test for comparing means of two groups and the v2 test for and VEGF nor with APA, but a positive correlation was found

Table 2 Comparison between patients with Patients Control Student’s Behc¸et disease and control (n ¼ 30) ± mean (n ¼ 15) ± mean t-test P-value ) HSP60 (ng mL 1)14Æ81 ± 8Æ88 5Æ13 ± 1Æ73 4Æ16 < 0Æ0001* ) VEGF (pg mL 1) 430Æ54 ± 99Æ33 84Æ07 ± 30Æ22 13Æ14 < 0Æ0001* ) APA (GpL mL 1)12Æ62 ± 7Æ24 4Æ18 ± 2Æ36 2Æ44 < 0Æ01*

HSP, heat shock protein; VEGF, vascular endothelial cell growth factor; APA, antiphospholipid antibodies; *, signiﬁcant.

Table 3 Comparison between patients with Inactive (n ¼ 13) Active (n ¼ 17) Student’s t-test P-value active and inactive stage of the disease with regard to laboratory parameters Hb 12Æ11 ± 2Æ13 11Æ78 ± 1Æ62 0Æ48 ns Neutrophil count 6Æ59 ± 1Æ63 8Æ24 ± 1Æ99 2Æ42 < 0Æ01* ESR 23Æ31 ± 16Æ32 52Æ76 ± 30Æ82 3Æ12 < 0Æ005* Platelet 267Æ15 ± 72Æ22 344Æ65 ± 137Æ12 1Æ85 < 0Æ05* HSP60 16Æ72 ± 13Æ22 13Æ35 ± 2Æ58 1Æ03 ns VEGF 424Æ52 ± 88Æ32 435Æ14 ± 109Æ45 0Æ29 ns APA 11Æ45 ± 9Æ11 13Æ52 ± 8Æ30Æ42 ns

Hb, haemoglobin; ESR, erythrocyte sedimentation rate; HSP, heat shock protein; VEGF, vascular endothelial cell growth factor; APA, antiphospholipid antibodies; *, signiﬁcant; ns, nonsigniﬁcant.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp32–37 Role of HSP60, VEGF and APA in Behc¸et disease, O. Shaker et al. 35

Table 4 Comparison of mean vascular endothelial cell growth factor level in Behc¸et Presence Absence Student’s t-test P-value disease patients according to presence or absence of some clinical manifestations Genital ulcers 431Æ03 ± 97Æ1 426Æ07 ± 142Æ95 0Æ08 ns Erythema nodosum 448Æ32 ± 81Æ97 426Æ09 ± 104Æ28 0Æ48 ns Eye 485Æ45 ± 98Æ17 388Æ55 ± 79Æ56 2Æ99 < 0Æ005* Arthritis 430Æ01 ± 103Æ46 430Æ8±99Æ93 0Æ02 ns Vascular 476Æ49 ± 92Æ27 390Æ32 ± 89Æ44 2Æ59 < 0Æ01*

*, Signiﬁcant; ns, nonsigniﬁcant.

between serum levels of VEGF and APA (r ¼ 0Æ78, angiogenesis, microvascular hyperpermeability and endothe- P <0Æ0001). lium-dependent vasodilatation, as well as endothelium nitric oxide (NO) production.20,21 Discussion Increased levels of VEGF have been reported in many autoimmune and infectious inflammatory diseases.22 The most HSPs are highly conserved molecules inducible by any form of prominent feature of Behçet syndrome is systemic dermal and cellular stress. They can bind cellular proteins to act as chaper- ocular vasculitis with endothelial cell dysfunction.8 As VEGF ones, stabilizing protein and preventing denaturation.6 How- has a direct effect on endothelial cells and it is produced by ever, because of great homology between bacterial and human cells participating in the pathophysiology of Behçet syndrome, HSPs, they have been claimed to cause autoimmunity.16 such as neutrophils, macrophages and endothelial cells, we Recently, several mycobacterial HSP65 peptides and their expected elevated plasma VEGF levels in patients with Behçet human analogues, HSP60, have been shown to stimulate a syndrome. In this study, we demonstrated that plasma VEGF specific lymphoproliferative response in patients with Behçet levels were higher in patients with Behçet syndrome than in disease.4 age- and sex-matched healthy control subjects. Moreover, its In this study, we have found significantly elevated serum level correlated with the presence of vascular and ocular mani- levels of HSP60 in patients with Behçet disease compared with festations in Behçet patients. This is in agreement with Cek- controls but the elevation did not correlate with disease activ- men et al.,9 who found elevated levels of VEGF in Behçet ity. This is in agreement with Kibaroglu et al.3 and Direskeneli disease that correlated with ocular manifestations.9 et al.,6 who reported increased serum levels of HSP60 in Be- Many possible mechanisms or factors may be responsible hçet disease. for the increased plasma VEGF concentrations found in this Our results were supported by Imamura et al.,16 who found study. Vascular thrombogenesis occurs in the course of Behçet HSP60 to be expressed in peripheral blood lymphocytes and syndrome; thus, platelets may contribute to higher plasma intestinal tissues of patients with Behçet disease. They stated VEGF levels and hydrogen peroxide released by activated neu- that T-helper (Th) 1-dominant immune responses and HSP60 trophils causes VEGF production by macrophages.23 Proinflam- expression may induce the inflammatory response and thus be matory cytokines, which are known to participate in the associated with the pathogenesis of intestinal Behçet disease. course of Behçet syndrome (tumour necrosis factor-a, soluble It has been shown that selected peptides derived from the interleukin (IL)-2 receptor, IL-6, IL-8 and NO), upregulate sequences of HSP60 induce significant proliferation of T cells endothelial cells as well as VEGF production.9,24 in patients with Behçet disease. Hu et al.17 have shown that VEGF promotes inflammatory processes by mobilizing leu- both oral and nasal administration of HSP60 peptide induce cocytes25 or, more recently, it has been shown that VEGF acti- uveitis, which is a common feature of Behçet disease.18 vates endothelial cell exocytosis of Weibel–Palade bodies, A molecular mimicry mechanism induces and/or exacer- releasing vasoactive substances capable of causing vascular bates Behçet disease; self-HSP60 and/or microbial HSP hom- thrombosis and inflammation.26 Bozoglu et al.27 estimated the ologous to the self-HSP60 activates self-reactive T cells specific levels of vascular endothelial growth factor and monocyte to the HSP peptides.19 Furthermore, HSP peptides stimulated chemoattractant protein (MCP)-1 in Behçet patients with ven- oligoclonal T-cell expansion producing Th1 cytokines. Ergun ous thrombosis. They found increased levels of VEGF and et al.4 have found increased T-cell receptor (TCR) cd cells that MCP-1 in Behçet disease thrombosis suggesting the possible proliferate with overlapping mycobacterial–human HSP pep- role of those angiogenic cytokines in the pathogenesis of Be- tides in Behçet disease and regulate ab T cells.4 hçet disease.27 Cytokines are the major mediators of immunological and It is also known that VEGF production upregulates NO syn- inflammatory reactions. They mediate delayed-type hypersen- thase expression in endothelial cells and increases endothelial sitivity, macrophage activation and the activation and/or release of NO.10 NO could damage host cells and tissues either recruitment of neutrophils. Thus, exacerbation of acute directly and/or following reaction with other free radicals. It inflammation in skin lesions occurs.19 VEGF is a potent endo- plays a critical role in the development of thrombotic events thelium-specific cytokine that stimulates inflammation, or other pathophysiological changes in patients with Behçet

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp32–37 36 Role of HSP60, VEGF and APA in Behçet disease, O. Shaker et al. disease.28 APA exerts a procoagulant effect resulting in throm- 4 Ergun T, Ince U, Eksioglu-Demiralp E et al. HSP 60 expression in bosis mainly of the larger veins and arteries. Indeed, one of mucocutaneous lesions of Behçet’s disease. J Am Acad Dermatol 2001; the key features in antiphospholipid syndrome is vascular 45:904–9. 5 Ranford JC, Henderson B. Chaperonins in disease: mechanisms, thrombosis occurring in 23–58% of patients with APA.29 models, and treatments. Mol Pathol 2002; 55:209–13. Previous reports have failed to find an association between 6 Direskeneli H, Eksioglu-Demiralp E, Yavuz S et al. T cell responses 30 31 APA and thrombosis. However, Ermakova et al. have found to 60/65 kDa heat shock protein derived peptides in Turkish an association between the presence of APA and retinal vascu- patients with Behçet’s disease. J Rheumatol 2000; 27:708–13. lar thrombosis in a group of patients with Behçet disease. 7 Keyt B, Berleau L, Nguyen H. The carboxy-terminal domain of vas- Moreover, Kang et al.32 have found an elevated APA level in cular endothelial growth factor is critical for its mitogenic potency. patients with Behçet disease. Differences in results could be at- J Biol Chem 1996; 271:7788–95. 8 Webb NJ, Myers CR, Watson CJ. Activated human neutrophils tributed to regional determinants, whether environmental or express vascular endothelial growth factor. Cytokine 1998; 10:254– genetic, which may affect the presence of APA as they do for 7. 33 other antibodies. In this study, a significantly elevated level 9 Cekmen M, Evereklioglu C, Er H et al. Vascular endothelial growth of APA was found in patients compared with controls, and the factor levels are increased and associated with disease activity in elevation correlated with the presence of vascular thrombosis. patients with Behçet’s syndrome. Int J Dermatol 2003; 42:870–5. APA play an important role in the development of thrombosis 10 Detmar M, Yeo KT, Nagy JA et al. Keratinocyte-derived vascular by inhibiting production of prostacyclin by endothelial cells. permeability factor (vascular endothelial growth factor) is a potent mitogen for dermal microvascular endothelial cells. J Invest Dermatol Others have found that APA may activate endothelial cells, 1995; 105:44–50. thus creating a hypercoagulable state that precedes and contri- 11 Bang D, Ji HG, Choi YS, Lee S. Absence of lupus anticoagulants in 34 butes to thrombosis. Thus, APA may serve as an additional Behçet’s disease. Yonsei Med J 1991; 32:326–9. marker for a risk of development of vasculitis and thrombo- 12 Anonymous. Criteria for diagnosis of Behçet’s disease. International sis.31 Study Group for Behçet’s Disease. Lancet 1990; 335:1078–80. A positive correlation was found in this study between 13 Evereklioglu C, Turkoz Y, Er H et al. Increased nitric oxide produc- VEGF and APA, both of which are related to the presence of tion in patients with Behçet’s disease: is it a new activity marker? J Am Acad Dermatol 2002; 46:50–4. vascular manifestations. This correlation was not previously 35 14 Orem A, Vanizor B, Cimsit G et al. Decreased nitric oxide produc- reported. However, Williams et al. have found elevated levels tion in patients with Behçet’s disease. Dermatology 1999; 198:33–6. of VEGF in antiphospholipid syndrome and suggested its role 15 Knapp RG, Miller MC. Clinical Epidemiology and Biostatistics. Malvern, in the pathogenesis of thrombosis in antiphospholipid syn- PA: Harwal Publishing Co., 1992. drome.35 16 Imamura Y, Kurokawa MS, Yoshikawa H et al. Involvement of Th1 In conclusion, increased HSP60 has pathological significance cells and heat shock protein 60 in the pathogenesis of intestinal in patients with Behçet disease. The specificity and primary Behçet’s disease. Clin Exp Immunol 2005; 139:371–8. 17 Hu W, Hasan A, Wilson A. Experimental mucosal induction of function of HSP60 as a potential pathogenic factor needs to be uveitis with 60-kDa heat shock protein-derived peptide 336–351. clarified because this may lead to the emergence of new treat- Eur J Immunol 1998; 28:2444–55. ment modalities such as its use in oral vaccination to induce 18 Direskeneli H, Saruhan-Direskeneli G. The role of heat shock pro- tolerance and prevent relapses. teins in Behçet’s disease. Clin Exp Rheumatol 2003; 21 (4 Suppl. High plasma levels of VEGF in patients with Behçet syn- 30):S44–8. drome suggests a role for VEGF in dermal and ocular vascular 19 Sakane T, Suzuki N, Nagafuchi H. Etiopathology of Behçet’s dis- events in the course of the disease. ease: immunological aspects. Yonsei Medical J 1997; 38:350–8. 20 Keyt BA, Berleau LT, Nguyen HV et al. The carboxyl-terminal As ocular Behçet syndrome is essentially diagnosed on the domain (111–165) of vascular endothelial growth factor is critical basis of clinical observation, VEGF could be used as a promis- for its mitogenic potency. J Biol Chem 1996; 271:7788–95. ing marker of ocular vaso-occlusive disease with neovasculari- 21 Han SW, Kim GW, Seo JS et al. VEGF gene polymorphisms and sus- zation. Further investigations of the precise role of VEGF in ceptibility to rheumatoid arthritis. Rheumatology (Oxford) 2004; the pathogenesis of Behçet syndrome may lead to novel ther- 43:1173–7. apies with antibodies or other inhibitors of VEGF. Moreover, 22 Chambers JC, Haskard DO, Kooner JS. Vascular endothelial func- APA may serve as an additional marker for a risk of develop- tion and oxidative stress mechanisms in patients with Behçet’s syndrome. J Am Coll Cardiol 2001; 37:517–20. ment of thrombosis. 23 Cho M, Hunt TK, Hussain MZ. Hydrogen peroxide stimulates macrophage vascular endothelial growth factor release. Am J Physiol References Heart Circ Physiol 2001; 280:H2357–63. 24 Evereklioglu C, Er H, Turkoz Y, Cekmen M. Serum levels of TNF- 1 Suzuki KM, Suzuki N. Behçet’s disease. Clin Exp Med 2004; 4:10–20. a, sIL-2R, IL-6, and IL-8 are increased and associated with elevated 2 Yazici H, Yurdakul S, Hamaryudan V. Behçet’s syndrome. Curr Opin lipid peroxidation in patients with Behçet’s disease. Mediators Inflamm Rheumatol 1999; 11:52–7. 2002; 11:87–93. 3 Kibaroglu A, Eksioglu-Demiralp E, Akoglu T et al. T and NK cell 25 Xiong M, Elson G, Legarda D et al. Production of vascular endothel- subset changes with microbial extracts and human HSP60-derived ial growth factor by murine macrophages: regulation by hypoxia, peptides in Behçet’s disease. Clin Exp Rheumatol 2004; 22 (4 Suppl. lactate and the inducible nitric oxide synthase pathway. Am J Pathol 34):S59–63. 1998; 153:587–98.

26 Matsushita K, Yamakuchi M, Morrell CN et al. Vascular endothelial 31 Ermakova NA, Alekberova ZS, Nasonov EL et al. Role of antiphosp- growth factor regulation of Weibel–Palade–body exocytosis. Blood holipid antibodies in occlusion of retinal vessels in various vascular 2005; 105:207–14. eye diseases. Vestn Oftalmol 2002; 118:29–32 (article in Russian). 27 Bozoglu E, Dinc A, Erdem H et al. Vascular endothelial growth fac- 32 Kang HJ, Lee YW, Han SH et al. Anticardiolipin and anti-b2-glyco- tor and monocyte chemoattractant protein-1 in Behçet’s patients protein I antibodies in Behçet’s disease. J Korean Med Sci 1998; with venous thrombosis. Clin Exp Rheumatol 2005; 23 (4 Suppl. 13:400–4. 38):S42–8. 33 Tokay S, Direskeneli H, Yurdakul S et al. Anticardiolipin antibodies 28 Sancak B, Onder M, Oztas MO et al. Nitric oxide levels in Behçet’s in Behçet’s disease: a reassessment. Rheumatology (Oxford) 2001; disease. J Eur Acad Dermatol Venereol 2003; 17:7–9. 40:192–5. 29 Hughes JR, Davies JA. Anticardiolipin antibodies in clinical condi- 34 Pierangli SS, Harris EN. Probing antiphospholipid mediated throm- tions associated with a risk of thrombotic events. Thromb Res 1998; bosis: the interplay between anticardiolipin and endothelial cells. 89:101–6. Lupus 2003; 12:539–45. 30 El-Ageb EM, Al-Maini MH, Al-Shukaily AK et al. Clinical features of 35 Williams FM, Parmar K, Hughes GR et al. Systemic endothelial cell Behçet’s disease in patients in the Sultanate of Oman; the signifi- markers in primary antiphospholipid syndrome. Thromb Haemost cance of antiphospholipid antibodies? Rheumatol Int 2002; 21:176– 2000; 84:742–6. 81.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp32–37 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07541.x Assessment of the presence of mucosal human papillomaviruses in malignant melanomas using combined ﬂuorescent in situ hybridization and chemiluminescent immunohistochemistry S. Ambretti, S. Venturoli, M. Mirasoli*, M. La Placa, F. Bonvicini, M. Cricca, M. Zerbini, A. Roda* and M. Musiani Sections of Microbiology and Dermatology, Department of Clinical and Experimental Medicine, University of Bologna, Bologna, Italy *Department of Pharmaceutical Sciences, University of Bologna, Via Massarenti, 9-40138 Bologna, Italy

Summary

Correspondence Background The vast majority of studies aimed at detecting human papillomavirus Monica Musiani. (HPV) DNA in skin cancer have used sensitive polymerase chain reaction (PCR) E-mail: [email protected] methods but the PCR technique, despite its high sensitivity, is not suitable to ascertain whether (i) the presence of HPV can be related only to few cells har- Accepted for publication 25 May 2006 bouring the virus, (ii) the presence of HPV is due to a tumour surface contamination and (iii) the presence of HPV is localized in cancer cells, rather than in Key words normal keratinocytes present in the tumour biopsy. In a recent work we have chemiluminescence, combined in situ hybridization found mucosal high-risk (HR) HPV genotypes in primary melanoma by PCR. and immunohistochemistry, fluorescence, human Objectives To localize mucosal HR-HPV nucleic acids and tumoural melanocytic papillomavirus, melanoma marker in the same sections of primary melanoma samples in order to under- Conflicts of interest stand the relationship between HPVs and melanoma cells. None declared. Methods We have developed a very sensitive method that combines an enzyme- amplified fluorescent in situ hybridization (ISH) for the detection of HPV nucleic acids (types 16 and 18) with a chemiluminescent immunohistochemistry (IHC) method for the detection of the tumoural melanocytic marker HMB-45 sequentially in the same section. Digital images of fluorescent ISH and chemiluminescent IHC were separately recorded, assigned different colours and merged using specific software for image analysis. Results The combined fluorescent ISH and chemiluminescent IHC demonstrated a sharp colocalization (in the range 60–80%) of HPV nucleic acids and melanoma marker inside the same sections of melanoma biopsies, with a strong specificity and sensitivity. Conclusions The strong colocalization of mucosal HR-HPV nucleic acids and HMB- 45 melanocytic marker emphasized that viral nucleic acids were specifically present in melanoma cells and supported a possible active role of HPV in malignant melanoma.

Human papillomaviruses (HPVs) are small DNA viruses that The relationship between human HPV, basal cell carcinoma induce a wide variety of hyperproliferative lesions in both cu- and squamous cell carcinoma has been frequently reported, taneous and mucosal epithelia and some types have been although epidemiological interpretation has become increas- recognized as important carcinogens in humans. HPVs are the ingly complicated by data emerging from the investigation causal agents in cervical cancer and have been postulated as of normal skin, hair follicles and benign hyperproliferative carcinogens in a range of other epithelial malignancies includ- dermatoses, in which nucleic acids of cutaneous HPVs has ing those of all the genital areas, oral cavity, tongue, hypo- also been frequently detected. The ubiquity of skin papillo- pharynx, larynx, oesophagus, conjunctiva, bladder, urethra maviruses revealed in many studies2,3 may indicate that the and skin.1 cutaneous HPV genotypes have no relevance in the genesis

38 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 Papillomaviruses in malignant melanoma, S. Ambretti et al. 39 of malignancy. Moreover, a recent paper by Forslund et al.4 maturation while many melanomas will express HMB-45 addressed the possibility that HPV positivity in skin tumour throughout. However HMB-45 is not expressed by all mela- biopsies may merely reflect contamination of the tumour noma cells and thus in some tumours the staining can be lim- surface by viral particles harboured in cells shed from infec- ited.14 ted healthy skin. As regards cutaneous melanomas, only a few studies have Materials and methods examined the presence of HPVs. Previous reports on the association between HPVs and melanoma lesions include single Samples cases of nodular melanoma,5,6 two cases of vulvar melanoma,7 one cohort of primary melanoma in non-sun-exposed mucosal Paraffin-embedded biopsy specimens from eight primary membrane,8 and three cohorts of metastatizing melanomas.9– melanomas, which had been previously diagnosed as positive 11 One of these studies9 provided evidence that the presence for HPV16 (seven samples) and HPV18 (one sample) by two of HPV, found in 58% of the biopsy specimens obtained from different polymerase chain reaction (PCR)-enzyme-linked im- 12 patients with stage III and IV melanoma, correlated with munosorbent assay (ELISA) methods (PCR-ELISA method to rapid melanoma progression, suggesting that HPV may modu- detect HPV with MY09/MY11 primer sets [MY-PCR ELISA] late a more aggressive phenotype of the tumour. In a recent and PCR-ELISA method to detect gene rearrangement [GP-PCR work12 we have found the presence and type of high-risk mu- ELISA]),12 were analysed in the study. In all patients the diag- cosal HPV genotypes in primary melanomas and in acquired nosis of primary melanoma was confirmed by histological an- dysplastic melanocytic naevi (ADMN) detecting high-risk alysis and Breslow’s microstaging to fulfil the American Joint HPVs in 24% of precursor lesions (ADMN) and in 27% of pri- Committee on Cancer (AJCC) criteria (Table 1). mary melanomas with the HPV 16 type most frequently As negative controls, paraffin-embedded biopsy specimens found. from three samples of normal skin, which had been previous- The presence of mucosal high-risk HPV genotypes in nor- ly proved negative for HPV both by PCR and by ISH with co- mal skin has rarely been reported.13 The detection of mucosal lorimetric detection, were also analysed. high-risk HPVs in primary melanomas and atypical naevi may Sections of 5 lm thickness were cut from paraffin-embed- therefore be relevant in the genesis of malignancy. ded tissue blocks, were placed on silanated slides and stored at The vast majority of studies6–8,10–12 aimed at detecting HPV 4 C until use. The human cervical carcinoma cell lines CaSki in skin cancer has used sensitive PCR methods, able to detect and HeLa 229 were used as positive controls as all contain sta- HPV genomes at a level of about one genome. Since PCR pro- bly integrated and trascriptionally active papillomavirus ge- cedures have always been performed on total DNA extracted nomes. In fact the CaSki cell line contains 500–600 integrated from tissue samples, the results obtained were not conclusive copies of the HPV16 DNA sequence in each cell and the HeLa in demonstrating that the detected HPV DNA was derived cell line contains 20–50 HPV18 DNA copies per cell.15 Cells from cancer cells. Thus, despite its high sensitivity, the PCR smeared on silanated glass slides were fixed with 4% parafor- technique is not suitable for ascertaining whether (i) the pres- maldehyde in PBS for 10 min. After fixation, cells were sub- ence of HPV can be related to only a few cells harbouring the jected to three 5-min washes in PBS, and then dehydrated virus, (ii) the presence of HPV is due to a tumour surface with 5-min ethanol washes (30%, 60% and 95%). Cell smears contamination and (iii) the presence of HPV is localized in were then air-dried and stored at 4 C until use. cancer cells, rather than in normal keratinocytes present in the tumour biopsy. Combined fluorescent in situ hybridization and All these data prompted us to study the presence of mucosal chemiluminescent immunohistochemistry high-risk HPV nucleic acids in primary melanomas and its colocalization with a tumoural melanocytic marker in the In situ hybridization step for the detection of HPV nucleic acids. Paraffin- same section, using a very sensitive method that combines an embedded tissue sections were dewaxed by two 10-min incu- enzyme-amplified fluorescent in situ hybridization (ISH) for bations in xylene and then washed in absolute ethanol for HPV nucleic acids with a chemiluminescent immunohisto- 5 min. To permeabilize the tissue, sections were incubated for ) chemistry (IHC) method for the detection of the tumoural 30 min at 37 C with pepsin (250 mg mL 1) diluted 1 : 100 ) melanocytic marker HMB-45. The anti-melanoma monoclonal in prewarmed 0Æ1 mol L 1 HCl. Cell smears were digested ) antibody HMB-45, used in the chemiluminescent IHC, is with pepsin diluted 1 : 25 000 in prewarmed 0Æ01 mol L 1 widely used in diagnostic pathology. Although numerous HCl. Permeabilization of the specimens was stopped with investigators have confirmed the high sensitivity and specifi- three 5-min washes in PBS; then biopsy samples and cell city of HMB-45 monoclonal antibody, there have been scat- smears were dehydrated in 30% and 60% and absolute ethanol tered reports of HMB-45 immunoreactivity in normal for 2 min, then air-dried. melanocytes (benign, in naevi), within the epidermis and Samples were overlaid with 30 lL of hybridization solution superficial papillary dermis. A positive reaction with HMB-45 (DAKO) containing 20 ng of digoxigenin (DIG)-labelled HPV indicates active melanosome formation and, therefore, melan- 16 E6-DNA probe (nt 424–553) or HPV18 E6-DNA probe (nt ocytic differentiation. This reactivity is lost with melanosome 422–539). The HPV16 and HPV18 E6-probes detect both E6

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 40 Papillomaviruses in malignant melanoma, S. Ambretti et al.

Table 1 Primary melanomas clinical data, HPV genotypes detected and co-presence of HPV positive signal and HMB-45 positive signal on total luminescence areas

Year of Thickness AJCC HPV Co-localization Patient Age/sex diagnosis Site (mm) stage type HPV+/HMB45+ (%) 1 48/F 2001 Leg 0Æ8IB1660Æ8 2 53/M 1999 Arm 0Æ7IA1659Æ1 3 61/F 1994 Leg 0Æ5IA1680Æ6 4 27/F 1994 Trunk 1Æ9 IIA 16 61Æ9 5 45/F 1994 Trunk 1Æ5 IIA 16 75Æ4 6 69/M 1996 Arm 9Æ0 IIB 16 79Æ5 7 35/M 1996 Foot 5Æ3 IIB 16 61Æ1 8 49/M 1994 Trunk 5Æ1 IIB 18 81Æ3

DNA and E6 spliced (E6I*) and unspliced transcripts. These After washing twice with wash buffer for 10 min at RT, genotype-specific probes were labelled in PCR by using type sections were overlaid with biotinylated secondary antibody specific primer pairs and a DIG-labelling mix (Roche). Each (Lab Vision, Fremont, CA, U.S.A.) and incubated for 30 min slide was covered with a glass coverslip and the edges were at RT. Then slides were washed as described above and incu- sealed with nail polish to prevent loss of the mixture during bated for 30 min at RT with streptavidin solution (Molecular denaturation and hybridization. After a simultaneous denatura- Probes) diluted 1 : 250 in wash buffer. After washing, tissue tion of cellular target DNA and HPV DNA probe on a heating sections were incubated for 30 min at RT with biotin conju- block for 5 min at 95 C, hybridization was performed by in- gate of horseradish peroxidase (HRP) (Molecular Probes) dilu- cubating the samples at 37 C overnight in a humidified ted 1 : 250 in wash buffer, then washed before performing chamber. After hybridization, the coverslips were carefully the chemiluminescent imaging acquisition, as described in the removed and the slides were washed twice for 10 min at following paragraph. 37 C in washing buffer (50% formamide in 2 · SSC ) ) (0Æ3 mol L 1 NaCl, 0Æ03 mol L 1 Na citrate pH 7Æ0) and then Fluorescent in situ hybridization and chemiluminescent twice for 10 min at room temperature (RT) in 2 · SSC. Sam- immunohistochemistry detection. ples were then processed for detection of the DIG-labelled DNA probes, by incubating for 30 min at 37 C with alkaline The fluorescent signal due to the hybrid formation and the phosphatase (AP)-conjugated anti-DIG Fab fragments (Roche, chemiluminescent signal due to the IHC reaction were detec- Monza, Italy) diluted 1 : 1000 in conjugate buffer (Roche). ted and analysed at the end of the whole procedure by means After washing twice for 10 min in pre-reaction buffer of a highly sensitive imaging technique. An epifluorescence ) ) (30 mmol L 1 Tris, 150 mmol L 1 NaCl, pH 7Æ5), samples microscope (BX 60, Olympus Optical, Tokyo, Japan) was were treated with freshly prepared development solution con- used. The microscope was enclosed in a dark box to prevent taining ELF-97 substrate (Molecular Probes, Eugene, OR, interference from ambient light and connected to a slow-scan U.S.A) diluted 1 : 20 in reaction buffer. After a 10-min incu- ultrasensitive CCD camera (LN/CCD Princeton Instruments, bation at RT, the reaction was stopped by washing the slides Roper Scientific, Trenton, NJ, USA) cooled to )100 Cto 5 min in ELF-97 stop buffer (Molecular Probes). reduce background noise. The microscope was also equipped with a computer-controlled motorized stage (OptiScan ES103, Prior Scientific Instruments Ltd, Fulbourn, England) allowing Immunohistochemistry step for the detection of tumoural reproducible positioning of the slides after processing. Fluor- melanocytic marker HMB-45. escence measurements were performed using an USH-102D After performing the ISH procedure, the same section was re- 100 W mercury excitation lamp (Ushio Inc., Tokyo, Japan) hydrated by 2-min incubations in 95%, 60% and 30% eth- and a self-assembled excitation cube containing a wide-band anol, air-dried, then incubated with blocking buffer UV excitation filter (330–385 nm), a 400 nm dichromatic ) ) (30 mmol L 1 Tris, 150 mmol L 1 NaCl, 1% BSA, 0Æ5% Tri- mirror (both from a Olympus U-MWU excitation cube), and ton X-100, pH 7Æ5) for 30 min at RT to avoid nonspecific a fluorescein specific emission filter (510–550 nm) from an binding. Sections were then overlaid with the primary anti- Olympus U-MWIBA2 excitation cube. All images were body directed against the tumoural melanocytic marker HMB- acquired at a 10 · objective magnification. First, a fluores- ) 45 (DAKO) diluted 1 : 20 in wash buffer (30 mmol L 1 Tris, cence measurement was performed using a 500 ms acquisition ) 150 mmol L 1 NaCl, 1% BSA, 0Æ05% Triton X-100, pH 7Æ5) time, and then the chemiluminescent image was acquired on and incubated for 45 min at RT. Endogenous peroxidase activ- the same field using a 5 min exposure time. In particular, for ity was blocked by incubating tissue sections for 10 min at RT chemiluminescence imaging, ECL (a chemiluminescent sub- with peroxidase blocking solution (3% H2O2 in PBS). strate for HRP based on the luminol/H2O2/enhancer system,

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 Papillomaviruses in malignant melanoma, S. Ambretti et al. 41

Amersham Bioscience, Amersham, U.K.) was added to the tion, proved completely negative at fluorescent ISH with the sample (volume varying in the range 30–50 lL, in order to HPV specific probes. cover the whole tissue section) and the chemiluminescent HPV positive CaSki cells (containing 500–600 copies of image was acquired after 2 min. Finally, the image of the HPV 16 integrated genomes), and HeLa cells (containing 10– same field was acquired under transmitted light (live image) 50 copies of HPV 18 integrated genomes), used as positive to allow localization of the analytes on the tissue section. Ima- controls for each run, proved positive at fluorescent ISH with ges were then digitally combined, as shown in Figure 1. In HPV16 and HPV18 probes respectively, confirming the specifi- particular, the live image was used as a ‘background’ image, city and the sensitivity of the method. The HPV staining in while the fluorescent and the chemiluminescent images were the CaSki and HeLa cell lines had both nuclear and cytoplas- assigned a different colour, and then combined in a final over- mic localization (Fig. 2). To assess the reproducibility of the lay image. fluorescent ISH assay, positive and negative samples were analysed in triplicate in different runs and the results obtained Results were concordant with expected data. All biopsy specimens from eight primary melanomas proved As preliminary experiments to assess the sensitivity and speci- positive to the chemiluminescent IHC procedure specific for ficity of enzyme-amplified fluorescent ISH for HPV nucleic the tumoural melanocytic marker HMB-45. All specimens acids and chemiluminescent IHC for the melanocytic marker showed a diffuse staining for HMB-45 in all epidermal cell HMB-45, and to avoid cross reactions, the two techniques layers, moreover the staining was more intense in the epider- were performed separately. mal basal layer and in the surrounding tissues (Figs 1,3). All In the analysis of the eight biopsy specimens from eight samples of normal skin biopsy specimens proved completely primary melanomas which had previously proved positive for negative. HPV 16 (seven samples) and HPV 18 (one sample), the fluorescent ISH using the respective E6-probes gave clearly positive Positive and negative samples were analysed in signals without any background. As shown in Figure 1(b,e), triplicate in different runs and the results were the fluorescent ISH clearly localized the HPV nucleic acids in concordant with the expected data the neoplastic cells (melanosomes) in different epidermal cell layers of the primary melanomas, comprising the lower layers Once assessed the sensitivity and specificity of the two tech- of the melanomas, as demonstrated by the histopathological niques, the double localization of HPV and of the tumoural findings of haematoxylin and eosin staining (Fig. 1f). melanocytic marker HMB-45 was evaluated. The ISH step for As negative controls, three samples of normal skin biopsy the detection of HPV nucleic acids and the IHC step for the specimens, which had previously proved negative for HPV detection of tumoural melanocytic melanoma marker HMB-45 nucleic acids both at PCR and at ISH with colorimetric detec- were then performed sequentially in the same sections of the

a be f

c d

Fig 1. Co-localization of mucosal high-risk HPV nucleic acids and of the tumoural melanocytic marker HMB-45 in the same section of primary melanoma by means of the combined enzyme-amplified fluorescent ISH and chemiluminescent IHC method. All images were acquired at a 10 · objective magnification. (a) Fluorescent signal obtained for the localization of HPV nucleic acids by means of the ISH procedure. (b) Overlay of the fluorescent signal, which was assigned the yellow colour, and the brightfield transmitted-light image. (c) Chemiluminescent signal obtained for localization of HMB-45 by means of the IHC procedure. (d) Overlay of the chemiluminescent signal, which was assigned the red colour, and the brightfield transmitted-light image. (e) Final overlay image of the colour processed fluorescent signal, the colour processed chemiluminescent signal and the live image; co-localization of HPV DNA and HMB-45 marker is evidenced by the colour combination, yielding an orange hue. (f) Histopathological findings of the tumour (haematoxylin and eosin stain, original magnification · 20).

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 42 Papillomaviruses in malignant melanoma, S. Ambretti et al.

The copresence of HPV positive signal and of HMB-45 positive signal was detectable in a mean of 70Æ0% (range 59Æ1– 81Æ3%) of the total luminescent areas in all the eight samples analysed. The presence of HPV positive signal without HMB- 45 signal was present in a mean of 9Æ5% (range 5Æ8–18Æ4%) of total luminescent areas. The presence of HMB-45 signal without HPV positive signal was present in a mean of 20Æ5% (range 12Æ8–27Æ4%) of total luminescent areas. The specificity of the newly developed combined fluorescent ISH and chemiluminescent IHC was confirmed by the following experiments: (i) no fluorescent positive signal was revealed when melanoma biopsy specimens were singularly treated with chemiluminescent IHC and, similarly, no chemiluminescent positive signal was detected after fluorescent ISH, (ii) primary melanoma biopsies and cells were completely negative after double ISH/IHC omitting HPV DNA probes and primary antibody against melanoma marker, (iii) no chemiluminescent and fluorescent signals were observed when melanoma sections were treated with the labelled HPV DNA probes and the antibody against melanoma marker omitting Fig 2. Localization of HPV 16 nucleic acids in CaSki cells by means of the incubation with conjugated detector systems, and (iv) no enzyme–amplified fluorescent ISH. The images were acquired at a positive signals were revealed when normal skin sections were · 40 objective magnification. Overlay of the fluorescent signal, which treated with double ISH/IHC. was assigned the yellow colour, and the brightfield transmitted-light In our assays, fluorescent ISH and chemiluminescent IHC image. were performed sequentially but consistent results were also obtained when chemiluminescent IHC was performed before eight melanoma specimens. Digital images of fluorescent ISH fluorescent ISH. and chemiluminescent IHC were separately recorded, assigned different colours, and merged using the specific software for Discussion image analysis. Positive signals for the presence of both HPV nucleic acids and melanoma marker were detected in the same High risk mucosal HPVs (especially HPV 16 and 18) display a sections of melanoma biopsies showing a sharp colocalization. strong association with cancer development. They have been

Fig 3. Representative results obtained in primary melanoma tissue sections for the colocalization of mucosal high-risk HPV nucleic acids and of the tumoural melanocytic marker HMB-45. Images were acquired at a · 10 objective magnification. (a) The fluorescent signal obtained for the localization of HPV nucleic acids by means of the ISH procedure was assigned the yellow colour, while the chemiluminescent signal obtained for localization of HMB-45 by means of the IHC procedure was assigned the red colour. (b) The final overlay images of the colour processed fluorescent signal, the colour processed chemiluminescent signal and the brightfield transmitted-light image are reported. Colocalization of HPV DNA and HMB-45 marker is shown by the colour combination, yielding an orange hue.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 Papillomaviruses in malignant melanoma, S. Ambretti et al. 43 found in over 99% of cervical carcinomas16 and, in addition, In our study it was very important to detect HPV nucleic acids they have been implicated in the development of cancers at in cells that were without any doubt melanoma cells. So we other body sites, including nonmelanoma and melanoma skin preferred to use an antibody with a good specificity with the cancers. At present, high-risk mucosal HPVs (especially HPV drawback of a lower sensitivity. 16) are considered as potentially involved in skin carcinogene- As a detection system for the IHC procedure, the enzyme sis, while the significance of cutaneous HPVs in skin cell trans- HRP combined with a chemiluminescent substrate was used. formation has now been strongly reduced. In fact recent The use of chemiluminescence allows sensitive and sharp ana- works have stressed that cutaneous HPVs, which show a lyte localization on the target surface, thanks to its high detec- greater phylogenetic heterogeneity (groups A, B1, B2, E) with tability, specificity and low background.21 respect to mucosal ones, represent ubiquitous viruses, highly As regards the double detection of HPV and melanocytic mar- prevalent in the normal skin of healthy adults, causing asymp- ker (HMB-45), as the fluorescent precipitate obtained upon AP tomatic infections likely to have been acquired in early infancy cleavage was stable throughout the IHC procedure, both the and therefore suggesting a commensal nature.3,16 The involve- fluorescent and the chemiluminescent images could be acquired ment of high-risk mucosal HPVs in skin cancers is also sugges- sequentially in the same field at the end of the whole procedure, ted by the fact that E6 and E7 genes of HPV 16 in vitro then digitally combined. However, thanks to the availability of a introduced in melanocytes were able to generate a human computer-controlled motorized stage, images could also be melanocyte cell line.17 acquired separately at the end of the respective procedure. With In our study, we explored the concomitant presence of the developed procedure, consistent results were also obtained high-risk mucosal HPVs, nucleic acids and tumoural melano- when chemiluminescent IHC was performed before fluorescent cytic marker HMB-45 in eight biopsy specimens of skin mela- ISH and when chemiluminescence was used as a detection sys- nomas, previously found positive by PCR for the presence of tem both for AP and HRP detection. HPV 16 (seven samples) and HPV 18 (one sample).12 For this The presence of HPV nucleic acids and melanocytic marker purpose, a novel method based on the combination of an was found in all eight positive samples of primary melanoma. enzyme-amplified fluorescent ISH and a chemiluminescent The HPV staining in all positive samples of primary melanoma IHC was developed. Because the distribution of melanoma and in the positive controls (CaSki and HeLa cell lines) had a cells in the biopsy specimens was heterogeneous, the ISH nuclear and cytoplasmic localization, as the HPV16 and detection of HPV nucleic acids and HMB-45 marker in single HPV18 E6-probes detect both E6 DNA and E6 spliced- and sections proved necessary to show the presence of HPV in unspliced-transcripted RNAs. melanoma cells. The presence of mucosal HPV nucleic acids in the lower Whereas PCR has been considered the most sensitive layers of the eight melanomas emphasized the fact that HPV method for HPV detection, in the present study we used ISH was deeply associated with melanoma cells and excluded the to localize the virus at the level of individual cells. The useful- possibility of contamination of the surface of the tumour, as ness of conventional ISH has occasionally been limited by low suggested for cutaneous HPVs by Forslund et al.,4 who found detection sensitivity, therefore in our work we used a highly a high prevalence of cutaneous HPVs on the top of skin sensitive enzyme-amplified fluorescent ISH protocol. By using tumours but not in the ‘stripped’ biopsies of the same the enzyme-labelled fluorescence (ELF-97) signal amplification tumour. Moreover in the present study, the presence of HPV technology we were able to detect as few as 20–50 copies of nucleic acids was found in most of the melanoma area (mean HPV18 integrated genomes in HeLa cells. The fluorescent sub- value 70%), excluding the possibility that positivity for HPV strate for alkaline phosphatase ELF-97 yielded an intensely could be merely ascribed to the presence of HPV in just a few fluorescent, photostable yellow-green precipitate sharply cells of the tumour harbouring the virus. localized at the site of enzymatic activity. The precipitate ex- In the eight melanoma biopsies examined, although some hibited a large Stokes shift (over 180 nm), thus its fluorescent cells proved positive only for HPV or only for the melanocytic signal could be easily distinguished from tissue autofluores- marker HMB-45, the majority showed the concomitant pres- cence. In addition, ELF-97 produced a signal that is many ence of both (Table 1), excluding the fact that HPV can be times brighter (about 30-fold) than that achieved using fluo- present in normal keratinocytes. rescein labelled probes.18,19 The ISH protocol was therefore The areas of biopsy specimens with staining for HPV with- combined with a very sensitive chemiluminescent IHC method out HMB-45 (9Æ5%) could represent areas of melanoma where to detect the melanocytic marker HMB-45. the melanoma antigen HMB-45 is not expressed, but we can- The anti-melanoma monoclonal antibody HMB-45, used in not exclude the presence of other cell types (keratinocytes) the chemiluminescent IHC, is widely used in diagnostic path- infected with HPV. ology. A positive reaction with HMB-45 indicates active mel- The areas of biopsy specimens with staining for HMB-45 anosome formation and, therefore, melanocytic without HPV (20Æ5%) could be explained if the number differentiation. This reactivity is lost with melanosome matur- of HPV genomic copies/cell was under 20–100, as defined ation while many melanomas will express HMB-45 through- by the sensitivity experiments on HeLa and CaSki cell lines, out.20 However HMB-45 is not expressed by all melanoma or cells transformed with a ‘hit-and-run’ mechanism of cells and thus in some tumours the staining can be limited.14 carcinogenesis.22

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 44 Papillomaviruses in malignant melanoma, S. Ambretti et al.

The relationship between the role of HPV in melanoma, 5 Takamiyagi A, Asato T, Nakashima Y et al. Association of human however, remains still incompletely elucidated. Dreau et al.8 papillomavirus type 16 with malignant melanoma. Am J Dermatopa- suppose that a HPV-mediated block in one of the cell-cycle thol 1998; 20:69–73. 6 Rohwedder A, Philips B, Malfetano J et al. Vulvar malignant mela- checkpoint proteins would cause immortalization of the mel- 6 noma associated with human papillomavirus DNA: report of two anocytes. Alternatively other authors assume that neoplastic cases and review of literature. Am J Dermatopathol 2002; 24:230–40. melanocytes may have become more susceptible to HPV infec- 7 Dahlgreen L, Schedvins K, Kanter-Lewensohn L et al. Human papil- tion, as normal melanocytes are rarely susceptible to HPV in- loma virus (HPV) is rarely detected in malignant melanomas of fection. sun sheltered mucosal membranes. Acta Oncol 2005; 44:694–9. Apart from direct infection of melanocytes, HPV could act 8 Dreau D, Culberson C, Wyatt S et al. Human papilloma virus in as a cofactor in the development of malignant melanoma.7 It melanoma biopsy specimens and its relation to melanoma progression. Ann Surg 2000; 231:664–71. is possible that the combination of persistent inflammation, 9 Miracco C, Palummo N, Lavergne D et al. Malignant melanomas: scarring (sclerosis), and the presence of HPV creates an envi- search for human papillomaviruses. Arch Dermatol 2001; 137:826–7. ronment or field effect that promotes and initiates carcinogen- 10 Roussaki-Shulze AV, Kouskoukis C, Rammos C et al. Identification esis, mostly squamous cell carcinoma and, less frequently, of human papillomavirus DNA in melanoma biopsy specimens of malignant melanoma. Human papillomavirus could be a fac- Greek population. Int J Clin Pharmacol Res 2005; 25:145–50. tor, either as a primary agent (direct infection of melanocytes) 11 La Placa M, Ambretti S, Bonvicini F et al. Presence of high-risk mu- or as a cofactor (chronic inflammatory disorders) in malignant cosal human papillomavirus genotypes in primary melanoma and in acquired dysplastic melanocytic naevi. Br J Dermatol 2005; melanoma. Moreover some previous data suggest that HPV 152:909–14. might be important for tumour initiation and progression by 12 Meyer T, Arndt R, Nindl I et al. Association of human papillomavi- 22 means of a ‘hit-and-run’ mechanism of carcinogenesis as rus infections with cutaneous tumors in immunosuppressed proposed with regard to bovine papillomavirus type 4 infec- patients. Transpl Int 2003; 16:146–53. tions of the alimentary tract in cattle.23 13 Esclamado RM, Gown AM, Vogel AM. Unique proteins defined by Our combined enzyme-amplified fluorescent-ISH and monoclonal antibodies specific for human melanoma. Some poten- chemiluminescent-IHC represents a rapid, specific and sensi- tial clinical applications. Am J Surg 1986; 152:376–85. 14 Faulkner-Jones BE, Tabrizi SN, Borg AJ et al. Detection of human tive method for detecting the presence and the colocalization papillomavirus DNA and mRNA using synthetic, type-specific of mucosal high-risk HPV nucleic acids and a tumoural melan- oligonucleotide probes. J Virol Methods 1993; 41:277–96. ocytic marker in the same section of primary melanoma and 15 Walboomers JM, Jacobs MV, Manos MM et al. Human papillomavi- then demonstrate that viral nucleic acids are specifically pre- rus is a necessary cause of invasive cervical cancer worldwide. J sent in the melanoma cells. These combined methods should Pathol 1999; 189:12–9. be applied to perform an extensive study of the interesting 16 Le Poole IC, van den Berg FM, van den Wijngaard RM et al. Gen- and not yet quite clear relationship between high-risk mucosal eration of a human melanocyte cell line by introduction of HPV16 E6 and E7 genes. In Vitro Cell Dev Biol Anim 1997; 33:42–9. HPV infection and skin melanoma carcinogenesis. 17 Paragas VB, Zhang YZ, Haugland RP et al. The ELF-97 alkaline phosphatase substrate provides a bright, photostable, fluorescent Acknowledgments signal amplification method for FISH. J Histochem Cytochem 1997; 45:345–57. The skilful technical help of Davide Domenicali and Erica Bab- 18 Cox WG, Singer VL. A high-resolution, fluorescence-based method bore is gratefully acknowledged. for localization of endogenous alkaline phosphatase activity. J Histo- chem Cytochem 1999; 47:1443–55. 19 Harwood CA, Proby CM. Human papillomaviruses and non-mela- References noma skin cancer. Curr Opin Infect Dis 2002; 15:101–14. 20 Taatjes DJ, Arendash-Durand B, von Turkovich M et al. HMB-45 1 Lowy DR, Howley PM. Papillomaviruses. In: Fields Virology (Knipe antibody demonstrates melanosome specificity by immunoelectron DM, Howley PM, eds), 4th edn, Vol. 2 Philadelphia: Lippincott microscopy. Arch Pathol Lab Med 1993; 117:264–8. Williams & Wilkins Publications, 2001; 2231–64. 21 Roda A, Guardigli M, Pasini P et al. Bio- and chemiluminescence 2 Antonsson A, Forslund O, Ekberg H et al. The ubiquity and impres- imaging in analytical chemistry. Anal Chim Acta 2005; 541:25–36. sive genomic diversity of human skin papillomaviruses suggest a 22 Pfister H. Human papillomavirus and skin cancer. J Natl Cancer Inst commensalic nature of these viruses. J Virol 2000; 74:11636–41. Monogr 2003; 31:52–6. 3 Scheurlen W, Gissmann L, Gross G et al. Molecular cloning of two 23 Campo M, Moar M, Sartirana M et al. The presence of bovine papil- new HPV types (HPV 37 and HPV 38) from a keratoacanthoma lomavirus type 4 DNA is not required for the progression to, or and a malignant melanoma. Int J Cancer 1986; 37:505–10. maintenance of, the malignant state in cancers of the alimentary 4 Forslund O, Lindelof B, Hradil E et al. High prevalence of cutane- canal in cattle. EMBO J 1985; 4:1819–25. ous human papillomavirus DNA on the top of skin tumours but not in ‘‘stripped’’ biopsies from the same tumors. J Invest Dermatol 2004; 123:388–94.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp38–44 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07546.x Skin surveillance of a U.K. paediatric transplant population M.A. Thomson, N.R. Suggett,* P.G. Nightingale, D.V. Milford, U. Baumann,§ D.A. Kelly,§ C. Moss and V.A. Hill Departments of Dermatology, Nephrology and §Hepatology, Birmingham Children’s Hospital, Birmingham B4 6NH, U.K. *Department of Surgery, University Hospitals Birmingham NHS Trust, Birmingham B15 2TH, U.K. Wellcome Trust Clinical Research Facility, Birmingham B15 2TH, U.K.

Summary

Correspondence Background Solid organ transplant recipients are at increased risk of skin cancer. Michelle Thomson. Melanoma is less common than nonmelanoma skin cancer (NMSC) although the E-mail: [email protected] relative proportion of melanoma among skin cancers has been shown to be higher in paediatric than adult recipients. Multiple melanocytic naevi and/or Accepted for publication 3 July 2006 atypical naevi may be a risk factor for the development of melanoma. The relationship between naevus counts and phenotypic characteristics, disease-related Key words variables and sun exposure has not been explored in paediatric transplant atypical naevi, dysplastic naevi, patients. immunosuppression, melanoma, skin cancer, solid Objectives To determine the prevalence of premalignant and malignant skin lesions organ transplantation and to identify known risk factors associated with benign and atypical melanocy- Conflicts of interest tic naevi in a U.K. paediatric transplant population. None declared. Methods Paediatric (£ 19 years) renal and liver transplant patients, who were 5 or more years post-transplantation, were reviewed over 12 months. Lifetime history of sun exposure, episodes of sunburn, sunny holidays, sunscreen use, sun bed use, demographic and transplantation details were collected using interview, questionnaire and case note review. A skin examination was performed for regional counts of malignant lesions, benign and atypical naevi. Results Ninety-eight patients (82 liver, 13 renal, three multiorgan) with a median follow up of 9 years (range 5–16) were reviewed. Neither skin cancer nor premalignant lesions for NMSC were detected in this group. Eighty-five patients had benign naevi (median 6, range 1–57). Clinical risk factors for increased counts of benign naevi included increasing age (P ¼ 0Æ03), more episodes of sunburn (P ¼ 0Æ003) and prolonged treatment with ciclosporin (P ¼ 0Æ009). The presence of atypical naevi in six patients was significantly associated with more episodes of sunburn (P ¼ 0Æ006) and more transplants (P ¼ 0Æ04). Other variables including phenotype, skin type, sun exposure, holidays abroad, residence abroad and total duration of immunosuppression did not correlate with benign or atypical naevus counts. Conclusions Skin cancer was not observed in paediatric solid organ transplant recipients who were 5–16 years post-transplantation. Both benign and atypical naevus counts were higher in children with frequent episodes of sunburn. As both naevi and sunburn are risk factors for melanoma, we should target fair-skinned transplant recipients with naevi for intensive sun avoidance education. A prospective, longitudinal follow-up study should determine the onset of skin cancer post- transplantation and the significance of benign and atypical naevus counts in this cohort.

Solid organ transplant recipients are at increased risk of skin in adult recipients. The relative proportion of melanomas cancer. The incidence of skin cancer in paediatric recipients is among skin cancers in paediatric recipients is higher than in unknown although it is recognized to be different from that adult recipients (12% vs. 5%). As with the general population,

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp45–50 45 46 Skin surveillance of paediatric transplant patients, M.A. Thomson et al. large numbers of naevi and atypical naevi are the strongest chest, back and palms/soles. A benign melanocytic naevus was risk factor for the development of melanoma. This study was defined as a focal, uniformly brown lesion of at least 2 mm in undertaken to determine the prevalence of premalignant and diameter that was not obviously a freckle or simple lentigo. malignant skin lesions, and to explore the relationship An atypical naevus was defined as large (> 5 mm) and irregu- between benign and atypical melanocytic naevi and likely risk lar in pigmentation and outline. Actinic keratoses, intraepider- factors in this population. mal carcinoma and keratoacanthoma were considered premalignant lesions for nonmelanoma skin cancer (NMSC). Patients and methods Melanoma, squamous cell carcinoma, basal cell carcinoma and Kaposi’s sarcoma were considered malignant. Case notes were reviewed for details of transplantation and Patients immunosuppression. Recorded data included underlying dis- The paediatric liver and renal transplant programmes began at ease, date of first transplant, type and number of transplants, Birmingham Children’s Hospital in 1983 and 1995, respect- type and total duration of each immunosuppressive agent and ively. By December 2004, 456 liver transplants and 167 renal episodes of rejection requiring intravenous methylpredniso- transplants had been performed. Transplant recipients lone. (£ 19 years) with functioning renal or liver allografts, more than 5 years post-transplantation, were identified using trans- Immunosuppression plant database records. One hundred and four recipients were invited to participate when they attended transplant clinic A triple therapy regimen of prednisolone, azathioprine and between January and December 2004. Six patients with liver ciclosporin was used in 72 liver and 11 renal transplant transplants declined (two boys and four girls; mean age patients. Tacrolimus was used instead of ciclosporin in com- 15 years). Ninety-eight solid organ transplant recipients (82 bination with prednisolone and/or azathioprine in 13 liver liver, 13 renal, three multiorgan) were enrolled in this study. and two renal recipients. At 5 years post-transplantation, maintenance treatment with one immunosuppressant was achieved in 51% of liver transplant patients and oral predniso- Study parameters lone was added to single or double therapy regimens in 33% Patients were interviewed and examined by a consultant der- experiencing repeated episodes of rejection. All renal trans- matologist (V.A.H.: n ¼ 9) and a dermatology registrar plant patients generally received higher levels of immunosup- (M.A.T.: n ¼ 89) with 18 months of training in skin oncol- pression with dual or triple immunosuppressant therapy at ogy. Both dermatologists were blind to previous records at the 5 years post-transplantation. Acute rejection episodes were time of initial interview. Sensitivity to sun exposure was deter- treated with intravenous methylprednisolone (0Æ5g or1g) mined by assessing eye colour and skin type (Fitzpatrick clas- on three consecutive days in both renal and liver patients. sification). The patients were interviewed about residence in a Two patients received prednisolone and one patient received tropical climate (> 3 months), sunny holidays abroad (inclu- prednisolone and tacrolimus prior to transplantation (median, ded skiing trips), sun bed use, sunscreen use, episodes of sun- 1 month). One patient with hepatoblastoma had eight cycles burn (painful erythema > 48 h) and family history of skin of chemotherapy (cisplatin, doxorubicin and cardioxane) prior cancer. A previously validated scoring system was used to cal- to liver transplantation. culate cumulative sun exposure.1 Lifetime history of sun exposure was determined from data on average number of Statistical analysis hours spent outdoors per day (weekdays and weekends assessed separately) during the ages 0–5, 6–10, 11–15 and Statistical analysis was performed using the statistical soft- 16–19 years. In order to estimate sun exposure per year, we ware package SPSS (Chicago, IL, U.S.A.). Data on melanocy- assumed that 24 h/day exposure ¼ 1 year exposure/year tic naevus counts were summarized as totals, arithmetic recorded. Therefore x h/day exposure ¼ x/24 years exposure/ means and median values. Nonparametric tests were used year recorded. If the exposure was x h/day on weekdays and for the univariable risk factor analysis of phenotypic charac- y h/day on weekends, the exposure over 5 years would be teristics, transplant-related variables and sun exposure fac- (5/7 of 5x/24) + (2/7 of 5y/24). In addition, sunbathing tors. Differences between groups were considered habits were examined by determining frequency of sunbathing statistically significant when P £ 0Æ05. Benign naevus counts (never, rarely, occasionally, frequently) during these same and sun exposure score were analysed using Spearman cor- periods. These were scored (0, never; 1, rarely; 2, occasional- relation coefficients, the Jonckheere–Terpstra test (for ly; 3, frequently). The cumulative sun exposure score was cal- ordered categories), the Kruskal–Wallis test and the Mann– culated as the sum of lifetime sun exposure and sunbathing Whitney test. Atypical naevus counts and numbers of epi- scores expressed as years spent outdoors. sodes of sunburn were very low and therefore Kendall’s All areas of skin were examined with the exception of the tau-b statistic and Fisher’s exact test were used to analyse scalp and areas covered by underwear. Lesions were counted these. Logistic regression analysis was used to perform a in the following regions: head/neck, arms, legs, abdomen/ multivariable analysis for benign naevus counts.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp45–50 Skin surveillance of paediatric transplant patients, M.A. Thomson et al. 47

Results General dermatological ﬁndings

None of the 98 paediatric transplant patients in this study had General patient characteristics premalignant lesions for NMSC or malignant skin lesions. There were 82 liver transplant patients (43 male; mean age Total and regional counts were obtained for benign naevi and 11Æ8 years, range 5–17) with a median follow up of 9Æ3 years atypical naevi (Table 3). The most common neoplasm diag- (range 5–16) and 13 renal transplant patients (11 male; mean nosed in this study population was post-transplant lymphopro- age 13Æ2 years, range 8–17) with a median follow up of liferative disease (PTLD) (8%), but none of the children had 8Æ3 years (range 5Æ1–11Æ9) (Table 1). Three patients with cutaneous involvement. multiorgan transplants were grouped together with the liver transplant patients for analysis. Seventy-two liver patients were Benign melanocytic naevi transplanted once, 10 twice and 3 three times (Table 2). Twelve renal patients had one transplant and one child had Eighty-ﬁve patients had benign naevi (total 909, mean 11, two. Ethnicity was 83 white, eight Pakistani, four Indian, two median 6, range 1–57). Fifty-nine per cent of benign naevi Afro-Caribbean and one mixed race. Eighty-four of 98 recipi- occurred on sun-exposed sites (10Æ9% head/neck, 30Æ5% ents (86%) had skin types I–IV. The distribution of patients arms, 17Æ5% legs) with naevus counts on the arms contribu- according to gender, selected phenotypic characteristics and ting most to the total benign naevus count. The prevalence of sun exposure factors is given in Table 1. naevi in unusual locations such as the palms and soles was

Table 1 Demographics and sun exposure of allograft recipients Liver transplant Renal transplant Characteristic patients (n ¼ 85) patients (n ¼ 13) Mean age at transplantation, range (years) 3Æ0 (0–11Æ1) 4Æ9(2Æ5–9Æ1) Median follow up, range (years) 9Æ3 (5–16) 8Æ3(5Æ1–11Æ9) Gender (male) 43 11 Skin type, n (%) I 10 (12) 0 II 13 (15) 6 (46) III 33 (39) 4 (31) IV 16 (19) 2 (15) V 10 (12) 1 (8) VI 3 (4) 0 Sun bed use, n (%) Yes (occasionally) 1 (1) 0 No 84 (99) 13 (100) Residence abroad (> 3 months), n (%) Yes 3 (4) 0 No 82 (96) 13 (100) Episodes of sunburn, n (%) Never 46 (54) 7 (54) Rarely 30 (35) 5 (38) Occasionally 8 (9) 1 (8) Frequently 1 (1) 0 Use of sunscreen, n (%) Yes 69 (81) 12 (92) No 16 (19) 1 (8) Sun protection factor, n (%) None 16 (19) 1 (8) Didn’t know 7 (8) 1 (8) < 15 0 1 (8) 15–30 32 (38) 7 (54) 40–60 30 (35) 3 (23) Holidays abroad, n (%) Yes 44 (52) 4 (31) No 41 (48) 9 (69) Family history of skin cancer, n (%) Yes 1 (1) 0 No 84 (99) 13 (100)

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp45–50 48 Skin surveillance of paediatric transplant patients, M.A. Thomson et al.

Table 2 Transplantation, immunosuppression and sun exposure identified as significant risk factors for increased counts of scores of allograft recipients benign naevi. Patients using sunscreen (n ¼ 81) had higher total benign Liver transplant Renal transplant naevus counts (median 6, range 0–38) compared with naevus Characteristic patients (n ¼ 85) patients (n ¼ 13) counts (median 2, range 0–57) in the 17 patients not using Allografts, n (%) sunscreen (Mann–Whitney test, P ¼ 0Æ047) (Table 4). The 1 72 (85) 12 (92) sun protection factor (SPF) used was not significant (P ¼ 2 10 (12) 1 (8) 0Æ56). The significant association between type of transplant 3 3 (4) 0 and naevus counts (P ¼ 0Æ007) was due to the multiorgan Pulses methylprednisolone, n (%) group in which there were three patients, none of whom had 1 16 (19) 4 (31) any melanocytic lesions. Gender, eye colour, skin type, resi- 2 6 (7) 4 (31) dence abroad (> 3 months), holidays abroad, cumulative sun 3 2 (2) 1 (8) exposure score, age at transplantation, total duration of 400immunosuppression (each agent analysed separately) and use ‡ 5 2 (2) 0 of pulse methylprednisolone did not correlate with total or Pretransplant regional benign naevus counts. Although skin type did not immunosuppression, n (%) Yes 2 (2) 0 correlate with naevus counts, there was a tendency for fair- No 83 (98) 13 (100) skinned children to have more naevi. Sun exposure score, 2Æ77 (0–14Æ6) 1Æ74 (0–5Æ14) mean (range) Atypical melanocytic naevi

Six patients had atypical naevi (total 11, mean 2, median 1, range 1–5). The highest counts were documented on sites Table 3 Distribution of benign and atypical naevi according to site normally covered and exposed to less sunlight. There were (n,%) nine atypical naevi on the trunk (abdomen/chest, back) (Table 3). Atypical naevus counts were signiﬁcantly associated Benign naevi Atypical naevi with sunburn (P ¼ 0Æ006) and higher number of transplants (n ¼ 909) (n ¼ 11) (P ¼ 0Æ04) (Table 4). No signiﬁcant relationship was found Head/neck 99 (11) 0 between atypical naevus counts and the other phenotypic, sun Abdomen/chest 129 (14) 4 (36) exposure, transplantation or immunosuppression variables. Back 169 (19) 5 (45) Three patients with atypical naevus were previously referred Arms 277 (30) 1 (9) Legs 159 (17) 1 (9) for a dermatology review. An excision was performed in one Palms/soles 76 (8) 0 patient and the histology was benign. The other two patients had a family history of atypical mole syndrome and their parents had regular skin surveillance reviews.

8Æ4%. Greater numbers of benign naevi were signiﬁcantly Sun exposure associated with increasing age (P ¼ 0Æ03), more episodes of sunburn (P ¼ 0Æ003) and greater duration of ciclosporin ther- In view of the positive relationship between sunburn and both apy (P ¼ 0Æ009) (Table 4). Increasing age was also signiﬁ- benign and atypical naevi, further analysis was undertaken. cantly associated with higher regional counts of benign naevi Sunburn affected 10 of 98 (10%) recipients, with a positive on the abdomen/chest (P ¼ 0Æ002) and back (P ¼ 0Æ013). In correlation observed with fairer skin types assuming ordering the multivariable analysis more episodes of sunburn, greater of categories I through VI (Kendall’s tau-b )0Æ38, P <0Æ001) duration of ciclosporin therapy and use of sunscreen were and sun exposure score (P ¼ 0Æ007) (Table 5). Sunscreen, SPF

Table 4 The relation of signiﬁcant factors to Benign naevus counts Atypical naevus counts benign and atypical naevus counts

Spearman correlation P-value Kendall’s tau-bP-value Age 0Æ22 0Æ03* 0Æ11 0Æ20 Sunburn 0Æ29 0Æ003* 0Æ25 0Æ006* Number of transplants 0Æ07 0Æ48 0Æ16 0Æ04* Duration of ciclosporin 0Æ26 0Æ009* 0Æ02 0Æ82 Sunscreen use – 0Æ047* (MW) 0Æ10 0Æ59

MW, Mann–Whitney test; *statistically signiﬁcant.

Table 5 The relation of sun exposure variables to sun exposure scores and sunburn Sun exposure score Sunburn

Spearman correlation P-value Kendall’s tau-bP-value Sunburn 0Æ27 0Æ007* – – Sunscreen use – 0Æ41 (MW) 0Æ10 0Æ35 Sun protection factor )0Æ01 0Æ91 )0Æ01 0Æ93 Skin type )0Æ08 0Æ46 )0Æ38 <0Æ001* Holiday abroad 0Æ04 0Æ67 0Æ10 0Æ25

MW, Mann–Whitney test; *statistically signiﬁcant.

and holidays abroad had no significant effect on episodes of examiners were not assessed for concordance, most naevus sunburn. counts were performed by one investigator (M.A.T.: 89%), thus reducing this bias. Potential count bias was also reduced Discussion by ensuring that the examiner was blind to previous records at the time of examination. Solid organ transplantation is the definitive treatment for end- The interrelations between phenotypic characteristics, sun- organ failure. Liver and renal transplantation in recipients related factors and naevus counts are complex and no single younger than 18 years accounts for 8% of all transplants in factor has proved to be a good predictor of naevus den- the U.K.2 Allograft survival and function usually require long- sity.23–25 In our study, the most striking finding was that sun- term immunosuppression, and the resultant modification of burn was more common in patients with increased numbers the immune system is associated with an increased risk of skin of benign and atypical naevi. However, we recognize that the cancer. There are only a few reports in small series devoted to majority (86%) of our cohort was fair skinned, and naevus skin cancer in paediatric transplant patients. counts are higher and sunburn more common in such individ- This study showed no skin cancer in children with solid uals. Hence, the positive association between sunburn and organ transplants. Our results are consistent with other publi- naevi may not be directly related but rather two separate find- cations on post-transplant paediatric cancers.3–7 Other studies ings occurring in fair-skinned patients. Such individuals usu- report that skin cancers are the most common malignancy ally limit their sun exposure because they know they burn, following paediatric renal transplantation,8,9 and the second which may explain the lack of association between overall sun most common, after PTLD, in children with all types of exposure and naevus counts. Also, children using sunscreen transplants.10 In these studies, skin cancers developed had more naevi than those who did not use sunscreen (P ¼ 12–15 years after transplantation at an average age of 0Æ047), implying that fair children, who have more naevi and 26–28 years.7,8,10 This emphasizes the need for longer-term a tendency to sunburn, will use more sunscreen but possibly follow up of our cohort to determine the incidence of skin ineffectively. Some studies have suggested that use of sun- cancer. screens may be associated with increased naevus density.26,27 Several studies in the general population have identified As in our work, these studies may have suffered the difficulty excess benign melanocytic naevi as the most significant factor of controlling for the confounding effect of age, gender, phe- for the development of melanoma.11–13 Excess melanocytic notype and prior sun exposure. Data on lifestyle adjustments naevi have been noted after paediatric transplantation.14–17 and their effect on sunscreen use, sun exposure and sunburn Naevus counts in our transplant population (liver, mean 9; were not collected as we could not quantify the change in kidney, mean 13) were much lower than the excess melano- behaviour pre- and post-transplantation. cytic naevi (mean 71) reported in a U.K. study of children Our data suggested that longer immunosuppression with with renal allografts.16 The demographic profile and the cri- ciclosporin increased benign naevus counts although total dur- teria used for classifying naevi were similar so the discrepancy ation of immunosuppression had no effect. We cannot draw in naevus counts may suggest that liver transplant patients conclusions from this result as cumulative doses were not develop fewer naevi than renal transplant patients. The preva- known and other confounding variables would have had an lence of naevi in our population was actually more consistent impact. Although studies have shown that there is no signifi- with studies of melanocytic naevi in children without trans- cant difference in skin cancer between paediatric renal allo- plants in the U.K.18,19 graft recipients treated with ciclosporin and not treated with It was important to differentiate benign naevi from atypical ciclosporin,9 the role of ciclosporin in the development of naevi (‡ 5 mm) given the documented association between naevi remains unclear. Six patients received sirolimus, a new the presence20–22 and number13 of atypical naevi and the immunosuppressive agent, and our results suggest that siroli- development of melanoma. However, counts of naevi are sub- mus offered protection against the development of naevi jective and it is possible that benign and atypical melanocytic (P ¼ 0Æ03). However, there are not enough data to draw any lesions were incorrectly classified. Although counts by both conclusions from this finding.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp45–50 50 Skin surveillance of paediatric transplant patients, M.A. Thomson et al.

Both naevi and sunburn are risk factors for melanoma so 9 Gruber SA, Gillingham K, Sothern RB et al. Cancer development in we need to pay particular attention to fair-skinned children pediatric primary renal allograft recipients. Transplant Proc 1994; who tend to have more naevi and a greater risk of 26:3–4. 10 Penn I. De novo malignancy in pediatric organ transplant recipients. sunburn. Patient education regarding sun avoidance must be Pediatr Transplant 1998; 2:56–63. incorporated into the pre-transplantation assessment and 11 Holly EA, Kelly JW, Shpall SN et al. Number of melanocytic nevi as reinforced regularly post-transplantation. Pre-transplantation a major risk factor for malignant melanoma. J Am Acad Dermatol assessment is important but there are practical difficulties 1987; 17:459–68. because the timing of transplantation is unpredictable and 12 MacKie RM, Freudenberger T, Aitchison TC. Personal risk-factor resources for a dedicated skin surveillance clinic are limited. chart for cutaneous melanoma. Lancet 1989; ii:487–90. However, targeted surveillance of transplant patients in early 13 Grob JJ, Gouvernet J, Aymar D et al. Count of benign melanocytic nevi as a major indicator of risk for nonfamilial nodular and super- adulthood is sensible. ficial spreading melanoma. Cancer 1990; 66:387–95. In this study, skin cancer was not observed in paediatric 14 Barker JNW, Macdonald DM. Eruptive dysplastic naevi following solid organ transplant patients. Benign and atypical naevus renal transplantation. Clin Exp Dermatol 1988; 13:123–5. counts were higher in children with frequent episodes of sun- 15 McGregor JM, Barker JNW, Macdonald DM. The development of burn and as naevi and sunburn are risk factors for melanoma, excess numbers of melanocytic naevi in an immunosuppressed we should target fair-skinned transplant recipients with naevi identical twin. Clin Exp Dermatol 1991; 16:131–2. for intensive sun avoidance education. The association of more 16 Smith CH, McGregor JM, Barker JNW et al. Excess melanocytic nevi in children with renal allografts. J Am Acad Dermatol 1993; 28:51–5. naevi with increased ciclosporin use is different from previous 17 Euvrard S, Kanitakis J, Cochat P et al. Skin diseases in children with reports and may be important in subsequent findings. A pros- organ transplants. J Am Acad Dermatol 2001; 44:932–9. pective, longitudinal study of this paediatric transplant popula- 18 Sorahan T, Ball PM, Grimley RP et al. Benign pigmented nevi in tion should determine the incidence of skin cancer in this children from Kidderminster, England: prevalence and associated group and whether benign and/or atypical naevus counts rep- factors. J Am Acad Dermatol 1990; 22:747–50. resent significant risk factors for developing melanoma. 19 Pope DJ, Sorahan T, Marsden JR et al. Benign pigmented nevi in children. Prevalence and associated factors: The West Midlands, United Kingdom Mole Study. Arch Dermatol 1992; 128:1201–6. Acknowledgments 20 Grob JJ, Andrac L, Romano MH et al. Dysplastic naevus in non-familial melanoma. A clinicopathological study of 101 cases. Br J Der- The authors thank the Dermatology, Nephrology and Hepatol- matol 1988; 118:745–52. ogy Departments of the Birmingham Children’s Hospital for 21 Roush GC, Nordhead JJ, Forget B et al. Independence of dysplastic supporting this study. nevi from total nevi in determining risk for non familial melanoma. Prev Med 1988; 17:273–9. 22 Slade J, Marghoob AA, Salopek TG et al. Atypical mole syndrome: References risk factor for cutaneous malignant melanoma and implications for management. J Am Acad Dermatol 1995; 32:479–94. 1 Ramsay HM, Fryer AA, Reece S et al. Clinical risk factors associated 23 Kelly JW, Rivers JK, MacLennan R et al. Sunlight: a major risk fac- with nonmelanoma skin cancer in renal transplant recipients. Am J tor associated with the development of melanocytic nevi in Austra- Kidney Dis 2000; 36:167–76. lian schoolchildren. J Am Acad Dermatol 1994; 30:40–8. 2 UK Transplant. National Transplant Database and Organ Donor Register. Bris- 24 Gallagher RP, McLean DI, Yang CP et al. Suntan, sunburn, and tol, UK Transplant, 2005. pigmentation factors and the frequency of acquired melanocytic 3 Bucuvalas JC, Ryckman FC. Long term outcome after liver trans- nevi in children. Arch Dermatol 1990; 126:770–6. plantation in children. Paediatr Transplant 1998; 2:56–63. 25 Gallagher RP, McLean DI. The epidemiology of acquired melanocy- 4 Offner G, Latta K, Hoyer P et al. Kidney transplanted children come tic naevi. A brief review. Dermatol Clin 1995; 13:595–603. of age. Kidney Int 1999; 55:1509–17. 26 Autier P, Dore JF, Cattauzza MS et al. Sunscreen use, wearing 5 Nocera A, Ghio L, Dall’Amico R et al. De novo cancers in paediatric clothes, and number of nevi in 6- to 7-year-old European children: renal transplant recipients: a multicentre analysis within the North European Organisation for Research and Treatment of Cancer mela- Italy Transplant programme, Italy. Eur J Cancer 2000; 36:80–6. noma cooperative group. J Natl Cancer Inst 1998; 90:1873–80. 6 Wingen AM, Wiesel M, Mo¨hring K et al. Malignancies in children 27 Azizi E, Iscovich J, Pavlotsky F et al. Use of sunscreen is linked with with renal replacement. Transplant Proc 1994; 26:5–6. elevated nevi counts in Israeli school children and adolescents. Mel- 7 Euvrard S, Kanitakis J, Cochat P, Claudy A. Skin cancers following anoma Res 2000; 10:491–8. pediatric organ transplantation. Dermatol Surg 2004; 30:616–21. 8 Coutinho HM, Groothoff JW, Offringa M et al. De novo malignancy after paediatric renal replacement therapy. Arch Dis Child 2001; 85:478–83.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp45–50 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07556.x Altered innate and adaptive immune responses in patients with hidradenitis suppurativa E.J. Giamarellos-Bourboulis,* A. Antonopoulou,* C. Petropoulou,* M. Mouktaroudi,* E. Spyridaki,* F. Baziaka,* A. Pelekanou,* H. Giamarellou* and N.G. Stavrianeas *4th Department of Internal Medicine and 2nd Department of Dermatology and Venereology, University of Athens, Medical School, University General Hospital ‘Attikon’, 1 Rimini Str, 124 64 Athens, Greece

Summary

Correspondence Background The clinical improvement of hidradenitis suppurativa reported in a Evangelos J. Giamarellos-Bourboulis. small number of patients with antitumour necrosis factor (anti-TNF)-a therapies E-mail: [email protected] supports the hypothesis for an altered immune response in these patients. Objectives To evaluate the state of the innate and adaptive immune responses in Accepted for publication 26 June 2006 patients with hidradenitis suppurativa. Methods Fifty-three patients and six healthy controls were studied. Blood was sam- Key words pled and subpopulations of lymphocytes were analysed by ﬂow cytometry; cytokines, hidradenitis, lymphocytes, monocytes monocytes were isolated and their function was evaluated from the concentrations of TNF-a and interleukin (IL)-6 in supernatants of cell cultures after trig- Conﬂicts of interest gering with endotoxins (lipopolysaccharides). TNF-a and IL-6 were estimated by None declared. an enzyme immunoassay. Results CD3/CD8 lymphocytes were lower in patients with involvement of the perineum than in controls; patients with involvement of the breast had higher levels of natural killer (NK) cells than controls. A negative correlation was found between years lapsing since initial presentation of lesions of hidradenitis and the percentage of NK cells. Monocytes isolated from healthy volunteers were more active for the secretion of TNF-a and IL-6 than those of patients with hidradenitis suppurativa. Conclusions A reduction in the percentage of NK cells over time and a lower monocyte response to triggering by bacterial components is observed in patients with hidradenitis suppurativa. Further research is needed to clarify if these changes are connected to an autoimmune mechanism in the pathogenesis of hidradenitis suppurativa.

Hidradenitis suppurativa is a skin disorder of unknown aetiol- Based on the above probability for the existence of some ogy and pathogenesis. Its prevalence is reported to range from derangement of the activity of the host immune function, the one per 300 to four per 100 in the general population.1 Cur- present study aimed to evaluate characteristics of the innate rent theories for its pathogenesis implicate hyperkeratosis of and adaptive immune responses in patients with hidradenitis the follicular epithelium as the hallmark of the pathogenetic suppurativa. The study was designed to investigate (i) the sub- process leading to occlusion of the apocrine glands with subpopulations of lymphocytes of the adaptive immune response; sequent follicular rupture, inflammation and possible second- and (ii) the activity of monocytes to secrete proinflammatory ary infection.2,3 Clinical improvement with the application of cytokines after triggering by bacterial components. therapies targeted against tumour necrosis factor (TNF)-a may be compatible with the above theory of pathogenesis, as TNF- Patients and methods a is a major proinflammatory cytokine. In these studies, monoclonal anti-TNF-a infliximab antibodies4,5 or soluble A total of 53 patients with hidradenitis suppurativa were TNF-a etanercept receptors6 were administered in a small enrolled in the study. Diagnosis was made by clinical cri- number of patients. Positive responses with anti-TNF-a ther- teria2,3,7 comprising (i) onset after puberty; (ii) the presence apies have also addressed the question of whether any prob- of subcutaneous nodules in areas of skin rich in apocrine able autoimmune predilection might contribute to the glands; and (iii) a compatible history of recurrent drainage of pathogenesis of hidradentitis suppurativa.4 pus from the affected areas. Written informed consent was

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp51–56 51 52 Hidradenitis and immune function, E.J. Giamarellos-Bourboulis et al. provided by all patients. Exclusion criteria were the presence bated with RPMI 1640 (Biochrom) enriched with 10% fetal ) of (i) human immunodeficiency virus infection; (ii) neutro- bovine serum (FBS) and 2 mmol L 1 of glutamine in the pres- ) ) ) penia, i.e. < 500 neutrophils mm 3; (iii) any solid tumour or ence of 100 U mL 1 of penicillin G and 0Æ1mgmL 1 of any haematological malignancy; (iv) intake of corticosteroids; streptomycin (Sigma Co, St Louis, MO, U.S.A.) in 75-cm3 or (v) history of rheumatoid arthritis, systemic lupus erythe- flasks. After 1-h incubation at 37 Cin5%CO2, nonadherent matosus or inflammatory bowel disease. The study was cells were removed; adherent monocytes were washed thor- approved by the Ethics Committee of the Attikon University oughly with Hanks’ solution (Biochrom). Monocytes were Hospital. then harvested with a 0Æ25% trypsin/0Æ02% ethylenediamine- Blood (10 mL) was sampled from each patient after veni- tetraacetic acid solution (Biochrom) and counted in a Neuba- puncture of a peripheral vein under sterile conditions with a uer plate. Their purity was more than 95% as determined heparinized syringe; 3 mL was collected into a heparin-coated after staining with the anti-CD14 mAb with the fluorocolour tube (Becton Dickinson, Cockeysville, MD, U.S.A.) for flow FITC (emission 520 nm) and reading through the EPICS cytometric analysis of subpopulations of lymphocytes; the XL/MSL flow cytometer. Their viability was assessed by trypan remaining blood was used for the isolation of monocytes. blue staining. During sampling all enrolled patients had active disease, i.e. recurrent purulent discharges from several of the affected skin sites. Blood was also sampled from six healthy volunteers Perianal equally matched for age with the study population. 70 ) Red blood cells were lysed with 1Æ0 mmol L 1 of ammonium chloride. White blood cells were washed three times 60 with phosphate-buffered saline (PBS) (pH 7Æ2) (Merck, 95%CI)

Darmstadt, Germany) and subsequently incubated for 15 min ± 50 in the dark with the monoclonal antibodies (mAbs) anti-CD3 and anti-CD19 with the fluorocolour fluorescein isothiocya- 40 nate (FITC, emission 520 nm; Immunotech, Marseille, France) and with the mAbs anti-CD4, anti-CD8 and anti-CD(16+56) 30 a with the fluorocolour phycoerythrin (PE, emission 550 nm, Immunotech). The following combinations were applied: 20 anti-CD3/anti-CD4, anti-CD3/anti-CD8, anti-CD3/anti- CD3/CD4

CD(16+56); anti-CD19 was applied singly. Cells staining posi- (median % of mononuclears 10 CD3/CD8 tive for the above antibodies were analysed after running CD3/CD(16 +56) through the EPICS XL/MSL ﬂow cytometer (Beckman Coulter 0 Co, Miami, FL, U.S.A.) with gating for mononuclear cells. Controls No Ye s For the isolation of blood monocytes, the collected hepari- Breast nized venous blood was layered over Ficoll Hypaque (Bioch- 30 rom, Berlin, Germany) and centrifuged. Isolated mononuclear cells were washed three times with PBS (pH 7Æ2) and incu-

a 95%CI) ± Table 1 Clinical characteristics of 53 patients with hidradenitis 20 suppurativa enrolled in the study

Characteristics Details Patients (n)53 Age (years, mean ± SD) 35Æ59 ± 12Æ58 10 Male/female (n) 19/34 Years from initial diagnosis 10Æ00 ± 1Æ39 % of mononuclears (median % of mononuclears (median ± SE) CD19 Involved body areas n (%) of patients CD3(–)/CD(16 ± 56)(+) Axillae 29 (54Æ72) 0 Groins 25 (47Æ16) Controls No Yes Breasts 19 (35Æ84) Perianal area 8 (15Æ09%) Fig 1. Signiﬁcant immunophenotyping alterations of lymphocytes Back 2 (3Æ78) were seen in 53 patients with hidradenitis suppurativa compared with Thorax and back 1 (1Æ89) six healthy volunteers in relation to the affected skin site. Circles Other sites 3 (5Æ66) denote outliers and asterisks extremes. a, Statistically signiﬁcant decreases compared with controls.

) Isolated monocytes were distributed in two wells of a 12- lowest limits of detection were 0Æ5pgmL 1 for TNF-a and ) well plate; they were incubated with RPMI 1640 supplement- 6Æ25 pg mL 1 for IL-6. Concentrations of TNF-a and of IL-6 ) ) ed with 10% FBS and 2 mmol L 1 glutamine for 18 h at were expressed as pg 10 4 monocytes. The monocyte function )1 37 Cin5%CO2 in the absence or presence of 100 ng mL was determined after subtracting the concentrations of TNF-a of puriﬁed endotoxin (lipopolysaccharide, LPS) derived from and IL-6 yielded after incubation in the presence of LPS from Escherichia coli O144:H4 (Sigma). After incubation, cell superna- respective concentrations yielded after incubation in the tants were collected and kept refrigerated at )70 C until absence of LPS. assayed for TNF-a and IL-6. Patients were divided into subgroups depending on the Concentrations of TNF-a and IL-6 were estimated by an affected skin areas. Results were expressed as median ± 95% enzyme immunoabsorbent assay (Diaclone, Paris, France). The conﬁdence interval (CI). Comparisons were performed using

Axillae Groin 800 800

aa aa 95%CI) 95%CI) ± ± 600 600

400 400 (pg/10 000 cells, median (pg/10 000 cells, (pg/10 000 cells, median (pg/10 000 cells, 200 200 α α TNF TNF (–)LPS (–)LPS

(+)LPS (+)LPS 0 0 Controls No Ye s Controls No Ye s

Breast Perianal

800 800

a 95%CI) 95%CI) ±

± 600 600 a, b

400 a 400 (pg/10 000 cells, median (pg/10 000 cells,

(pg/10 000 cells, median (pg/10 000 cells, 200 200 α α TNF TNF (–)LPS (–)LPS

(+)LPS (+)LPS 0 0 Controls No Ye s Controls No Ye s

Fig 2. Comparative release of tumour necrosis factor (TNF)-a in supernatants of monocytes isolated from patients with hidradenitis suppurativa compared with six healthy volunteers in relation to the affected site of the skin; (–) lipopolysaccharide (LPS), incubation in the absence of endotoxins; (+) LPS, incubation in the presence of endotoxins. Circles denote outliers and asterisks extremes. a, monocyte response after triggering with LPS lower than controls; b, monocyte response after triggering with LPS lower than patients with involvement of the respective site.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp51–56 54 Hidradenitis and immune function, E.J. Giamarellos-Bourboulis et al. the Mann–Whitney U-test after Bonferroni correction. Patients disease and breast disease compared with those of healthy with disease of the back and chest were excluded from analy- controls (Fig. 1). More precisely, patients with perianal disease sis. Statistical correlations were assayed after assessment of the had lower CD3/CD8 lymphocytes than healthy controls (P ¼ nonparametric Spearman coefﬁcient (rs). P <0Æ05 was consid- 0Æ045) and patients with breast disease had higher percentages ered statistically signiﬁcant. of NK cells than healthy controls (P ¼ 0Æ049). Similar differences were not encountered for patients with involvement of Results the other skin areas. A negative correlation was found between years lapsing since initial presentation of hidradenitis lesions

The clinical characteristics of patients enrolled in the study are and the percentage of NK cells (rs ¼ )0Æ376, P ¼ 0Æ049). shown in Table 1. Signiﬁcant differences were seen in the Figure 2 shows the potency of monocytes for the release of subpopulations of mononuclear cells of patients with perianal TNF-a after triggering with LPS in relation to the affected skin

Axillae Groin

300 300 95%CI) 95%CI) ± ± 200 200

100 100 IL-6 (pg/10 000 cells, median IL-6 (pg/10 000 cells, IL-6 (pg/10 000 cells, median IL-6 (pg/10 000 cells, (–)LPS (–)LPS

(+)LPS (+)LPS 0 0 Controls No Ye s Controls No Ye s

Breast Perianal

300 300

a, b 95%CI) 95%CI) ± ± 200 200

100 100

IL-6 (pg/10 000 cells, median IL-6 (pg/10 000 cells, (–)LPS median IL-6 (pg/10 000 cells, (–)LPS

(+)LPS (+)LPS 0 0 Controls No Ye s Controls No Ye s

Fig 3. Comparative release of interleukin (IL)-6 in supernatants of monocytes isolated from patients with hidradenitis suppurativa compared with six healthy volunteers in relation to the affected site of skin; (–) lipopolysaccharide (LPS), incubation in the absence of endotoxins; (+) LPS, incubation in the presence of endotoxins. Circles denote outliers and asterisks extremes. a, monocyte response after triggering with LPS lower than controls; b, monocyte response after triggering with LPS lower than patients with involvement of the respective site.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp51–56 Hidradenitis and immune function, E.J. Giamarellos-Bourboulis et al. 55 site. Monocytes isolated from healthy volunteers were more treatment of this disorder. Their effect might be compatible active for the secretion of TNF-a than those isolated from with the hypothesis of an autoimmune predilection in hid- patients with involvement of the axillae (P ¼ 0Æ002), of the radenitis suppurativa.6 Furthermore, the disease may co-exist groin (P ¼ 0Æ002) and of the breast (P ¼ 0Æ004), but not of with Crohn disease, which is an autoimmune disorder,3 and the perianal area. pyoderma gangrenosum, in which derangements of the The potency of monocytes for the release of IL-6 after trig- immune function have been reported.14 gering with LPS in relation to the affected skin site is shown Therapeutic benefit with immunosuppressive and anti- in Figure 3. Monocytes isolated from healthy volunteers were TNF-a therapies would be compatible with the existence of more active for the secretion of IL-6 than patients without intense reactions of both the innate and adaptive immune involvement of the perianal area (P ¼ 0Æ036). systems upon antigenic triggering in hidradenitis suppurativa. Presented findings pointing towards defective immune Discussion responses give rise to much skepticism about the existence of an autoimmune predilection of the disease. However, Clinical improvement of patients with hidradenitis suppurativa these diminished immune responses may provide an ade- after administration of infliximab and etanercept directed quate explanation for the high recurrence rates even in ) against TNF-a4 6 has led to the assumption that significant patients undergoing extensive surgical excision of the affec- alterations of the immune responses may exist in these ted sites.15 patients. The present study was designed to evaluate the char- The presented results reveal the existence of alterations of acteristics of both the innate and adaptive immune responses the function of the innate immune system and to a lesser of patients with hidradenitis suppurativa. extent of the adaptive immune system in patients with hid- Alterations in the subpopulations of lymphocytes were radenitis suppurativa. These changes mainly involve a reduc- found only for patients with disease of the breast and peri- tion in the percentage of NK cells over time and a lower anum. These patients had elevated percentages of NK cells, i.e. monocyte response upon triggering by bacterial components. CD3())/CD(16+56)(+) cells and lower percentages of CD3/ Further research is mandatory to clarify if these changes are CD8(+) cells than their respective controls (Fig. 1). Further- connected to an autoimmune mechanism in the pathogenesis more, it was found in the entire study population that the of hidradenitis suppurativa. percentage of NK cells was decreased as the history of the disease was prolonged. Acknowledgments Another major finding of the present study was the decreased capacity of monocytes of patients with hidradenitis This study was co-funded by the European Social Fund & suppurativa to secrete proinflammatory cytokines upon trig- National Resources–EPEAEK II–PYTHAGORAS. gering by LPS compared with healthy controls. These findings involved mainly TNF-a and to a lesser extent IL-6 References (Figs 2, 3). These findings of decreased innate and adaptive immune 1 von der Werth JM, Jemec GBE. Morbidity in patients with hidrade- responses in hidradenitis suppurative should be considered nitis suppurativa. Br J Dermatol 2001; 144:809–13. in the light of the existing data in the literature. The latter 2 Slade DEM, Powell BW, Mortimer PS. Hidradenitis suppurativa: pathogenesis and management. Br J Plast Surg 2003; 56:451–61. are characterized by considerable discrepancies as two previ- 3 Wiseman MC. Hidradenitis suppurativa: a review. Dermatol Ther ous studies did not reveal any defects of the immune func- 2004; 17:50–4. 8,9 tions in patients with hidradenitis suppurativa, whereas 4 Sullivan TP, Welsh E, Kerdel FA et al. Infliximab for hidradenitis two other studies have identified some defects of the suppurativa. Br J Dermatol 2003; 149:1046–9. patients’ immune systems.10,11 More precisely, adequate 5 Rosi YL, Lowe L, Kang S. Treatment of hidradenitis suppurativa function of neutrophils has been reported in seven and 15 with infliximab in a patient with Crohn’s disease. J Dermatol Ther patients, respectively.8,9 A defective bactericidal effect associ- 2005; 16:58–61. 6 Cusack C, Buckley C. Etanercept: effective in the management of ated with low intracellular levels of cyclic guanosine mono- 10 hidradenitis suppurativa. Br J Dermatol 2006; 154:726–9. phosphate has been described in one patient, whereas 7 Shah N. Hidradenitis suppurativa: a treatment challenge. Am Fam decreased counts of T lymphocytes were reported in seven Physician 2005; 1554:1547–52. other patients.11 However, to our knowledge no study 8 Dvorak VC, Root RK, MacGregor RR. Host-defense mechanisms in exists such as the one performed herein, focusing on the hidradenitis suppurativa. Arch Dermatol 1977; 113:450–3. activity of blood monocytes in relation to hidradenitis sup- 9 Lapins J, Asman B, Gustafsson A et al. Neutrophil-related host purativa. response in hidradenitis suppurativa: a pilot study in patients with inactive disease. Acta Derm Venereol 2001; 81:96–9. The lack of a specific mechanism for the pathogenesis of 10 Ginder PA, Ousley M, Hinthorn D et al. Hidradenitis suppurativa: hidradenitis suppurativa has led to the application of a variety evidence for a bactericidal defect correctable by cholinergic agonist of therapeutic approaches such as antibiotics and immunosup- in vitro and in vivo. J Clin Immunol 1982; 2:237–41. pressive therapies; all were proved of limited or no bene- 11 O’Loughlin S, Woods R, Kirke PN et al. Hidradenitis suppurativa. fit.12,13 Anti-TNF-a strategies appear to be promising in the Glucose tolerance, clinical, microbiologic, and immunologic

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp51–56 56 Hidradenitis and immune function, E.J. Giamarellos-Bourboulis et al.

features and HLA frequencies in 27 patients. Arch Dermatol 1988; 14 Ah-Weng A, Langtry JAA, Velangi S et al. Pyoderma gangrenosum 124:1043–6. associated with hidradenitis suppurativa. Clin Exp Dermatol 2005; 12 Jemec GB. Medical treatment of hidradenitis suppurativa. Expert Opin 30:669–71. Pharmacother 2004; 5:1767–70. 15 Kagan RJ, Yakuboff KP, Warner P et al. Surgical treatment of hidrade- 13 Jemec GBE. Methotrexate is of limited value in the treatment of nitis suppurativa: a 10-year experience. Surgery 2005; 138:734–41. hidradenitis suppurativa. Clin Exp Dermatol 2002; 27:528–9.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp51–56 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07559.x Does adjuvant alpha-interferon improve outcome when combined with total skin irradiation for mycosis fungoides? D. Roberge, T. Muanza, G. Blake,* C. Shustik,* T. Vuong and C.R. Freeman Department of Radiation Oncology and *Division of Hematology, Department of Medicine, McGill University Health Centre, Montreal, Quebec H3G 1A4, Canada

Summary

Correspondence Background Patients with mycosis fungoides (MF) experience frequent disease David Roberge. recurrences following total skin electron irradiation (TSEI) and may benefit from E-mail: [email protected] adjuvant therapy. Objectives To review the McGill experience with adjuvant alpha-interferon (IFN) in Accepted for publication 19 June 2006 the treatment of MF. Methods From 1990 to 2000, 50 patients with MF were treated with TSEI: 31 with Key words TSEI alone and 19 with TSEI + IFN. Median TSEI dose was 35 Gy. In the interferon, mycosis fungoides, radiotherapy TSEI + IFN group, IFN was given subcutaneously at 3 · 106 units three times per week starting 2 weeks prior to start of TSEI, continued concurrently with the Conflicts of interest radiation and for an additional 12 months following TSEI. The TSEI alone group None declared. included 16 men and 15 women with a median age of 61 years (range 31–84). The TSEI + IFN group included 14 men and five women with a median age of 51 years (range 24–83). Clinical stage was IA, IB, IIA, IIB, III and IVA in 2, 9, 4, 8, 1 and 7 patients of the TSEI group and 0, 3, 3, 7, 4 and 2 patients of the TSEI + IFN group. Results Median follow up for living patients was 70 months. All patients responded to treatment. Complete response (CR) rate was 65% following TSEI and 58% following TSEI + IFN (P ¼ 0Æ6). Median overall survival (OS) was 61 months following TSEI and 38 months following TSEI + IFN (P ¼ 0Æ4). Acute grade II– III dermatitis was seen in all patients. Fever, chills or myalgia were seen in 32% of patients treated with TSEI + IFN. Conclusions Concurrent IFN and TSEI is feasible, with acceptable toxicity. Even when controlling for disease stage, the addition of IFN did not appear to increase CR rate, disease-free survival or OS.

Cutaneous T-cell lymphoma (CTCL) is a rare malignancy and long-term disease control are seen following total skin accounting for 2Æ2% of all lymphomas. It is characterized by electron irradiation (TSEI), although recurrences are the inﬁltrates of epidermotropic neoplastic T cells. Mycosis fun- norm.4,5 The role of the active therapeutic agents as adjuvants goides (MF) accounts for approximately half of CTCL cases.1 to TSEI remains unresolved. Although MF is often an indolent disease with skin lesions that Interferons (IFNs) are cytokines with antiviral, antiprolifera- can be present for months to years prior to diagnosis, progno- tive, antiangiogenic and immunomodulatory properties. They sis is highly dependent on disease stage. In contrast to patients are known to potentiate radiation effect by increasing blockage 6,7 with stage IA disease who can expect a survival similar to that at the G2-M phase of the cell cycle. They have been shown of matched controls in the general population,2 stage IV to be active in patients with MF, either as single agents8–10 or patients have a median survival of only 18 months.3 in combination with PUVA,8 retinoids11 or photopheresis.12 Therapeutic modalities for MF can be divided into skin- There is, however, no published experience with the combin- directed and systemic treatments. Common skin-directed treat- ation of single-agent alpha-IFN and TSEI. ments include psoralen with ultraviolet A radiation (PUVA), In an attempt to increase response rate and duration of topical chemotherapy, topical retinoids and external beam response, a cohort of patients was treated with a combination radiotherapy. Systemic therapies, generally reserved for more of IFN alfa-2b and TSEI. Our objective for this report is to advanced or refractory disease, include cytotoxic chemother- compare the outcome of these patients with that of a series of apy, photopheresis and biologic agents.1 High response rates patients treated at our institution with TSEI alone.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp57–61 57 58 TSEI with adjuvant interferon alfa-2b, D. Roberge et al.

Patients and methods Follow up

Between August 1990 and July 2000, 50 patients with bi- Patients were assessed weekly during TSEI for toxicity. Upon opsy-confirmed MF were treated at our institution. The cohort completion of TSEI they were followed in our clinic at 3–4- consisted of 31 patients treated with rotational TSEI alone and month intervals. As well, the patients were usually followed 19 patients treated with TSEI + IFN. The TSEI group included by their referring dermatologist or haematologist. 16 men and 15 women with a median age of 61 years (range 31–84). The TSEI + IFN group included 14 men and five Statistical analysis women with a median age of 51 years (range 24–83). Clinical stage was IA, IB, IIA, IIB, III and IVA in 2, 9, 4, 8, 1 and 7 Actuarial survival and disease-free survival (DFS) curves were patients of the TSEI group and 0, 3, 3, 7, 4 and 2 patients of calculated from the end of treatment and were plotted accord- the TSEI + IFN group (Table 1). ing to the Kaplan–Meier technique. Overall survival (OS) was calculated from the time of TSEI. In complete responders, DFS was the interval between the end of TSEI and either the date Total skin electron irradiation of pathological confirmation of the recurrence or initiation of TSEI was delivered using a previously described single beam salvage therapy. Univariate analyses of differences in actuarial rotational technique with 6 MeV electrons.13 TSEI was typic- curves were performed by using the log-rank test. The effect ally planned for a total of 35–36 Gy in 14–18 fractions. Actual of disease stage on DFS was controlled for by performing a delivered doses ranged from 20 to 36 Gy (median 35 Gy). multivariate Cox–Wilcoxon test. Because of the small number of stage IA patients, stages IA and IB were combined for the purpose of multivariate analysis. Differences in complete Interferon response (CR) rates by treatment group were calculated by Recombinant IFN alfa-2b was administered subcutaneously at Fisher’s exact test. All P-values correspond to two-sided signi- a dose of 3 · 106 units three times per week (one patient was ficance tests. Analyses were carried out using SPSS software treated with 5 · 106 units three times per week) starting (version 11; SPSS Inc., Chicago, IL, U.S.A.). 2 weeks prior to TSEI, continued concurrently with the radiation and for an additional 12 months following TSEI. Assessment of response

Response to treatment was determined primarily by physical Table 1 Patient characteristics examination. Pathological confirmation was obtained in some patients. CR was defined as a complete clinical disappearance TSEI TSEI + IFN of all disease. Partial response was defined as reduction by > 50% of the skin lesions. Patients (n)3119 Median age, years (range) 61 (31–84) 51 (24–83) Male 16 (52%) 14 (74%) Results Female 15 (48%) 5 (26%) Stage As of March 2006, 46% of patients were still alive. Living IA 2 (6%) 0 (0%) patients had been followed for a median of 70 months IB 9 (29%) 3 (16%) (38Æ3 months for the entire group). Median OS for the entire IIA 4 (13%) 3 (16%) group was 52 months. There was no significant difference in IIB 8 (26%) 7 (37%) III 1 (3%) 4 (21%) OS between the two groups (61 months in the TSEI group, IVA 7 (23%) 2 (10%) 38 months in the TSEI + IFN group, P ¼ 0Æ4). The actuarial Prior treatmenta survival curves are plotted in Figure 1. PUVA 5 (16%) 10 (53%) All patients responded to treatment. CR rate was 58% in the Topical chemotherapy 10 (32%) 7 (37%) TSEI + IFN group and 65% in the TSEI group (P ¼ 0Æ6). For Systemic chemotherapy 7 (23%) 8 (42%) the entire group, the median time to CR was 3Æ58 months Involved-field radiotherapy 7 (23%) 5 (26%) (corresponding to a scheduled 3-month follow-up appoint- Retinoids 1 (3%) 0 (0%) IFN 3 (10%) 0 (0%) ment). The median relapse-free survival following CR (Fig. 2) TSEI was 7Æ4 months for TSEI + IFN vs. 95Æ5 months for TSEI Dose (P ¼ 0Æ003). When factoring disease stage, age and prior < 35 Gy 9 (29%) 4 (21%) therapy in a multivariate analysis, the only predictor of pro- 35–36 Gy 22 (71%) 15 (79%) longed remission following CR was treatment with TSEI alone Median duration 35 days 44 days (P ¼ 0Æ03). TSEI, total skin electron irradiation; IFN, interferon; PUVA, psor- On multivariate analysis, stage (whether categorized into in- alen plus ultraviolet A radiation. aExcluding topical steroids. dividual stages or dichotomized into stages IA–IIA vs. IIB– IVA) was a strong predictor of OS (P ¼ 0Æ013) and DFS (P ¼

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp57–61 TSEI with adjuvant interferon alfa-2b, D. Roberge et al. 59

achieving CR, relapses are the norm (15–25% 10-year DFS for stage IA and IB patients).2,3 It has thus been logical to explore the use of agents known to be active in MF as adjuvants to TSEI. In a randomized trial from the National Cancer Institute,14 systemic chemotherapy (cyclophosphamide, doxorubicin, etoposide and vincristine) combined with TSEI significantly increased CR rate when compared with a conservative strategy of sequential topical treatments starting with topical nitrogen mustard. Despite the higher CR rate (38% vs. 18%) there was no impact on OS and combined therapy was associated with increased toxicity. In 1992, Jones et al.15 reported on the use of TSEI (3500 cGy in 12 fractions) combined with concurrent and ad- Fig 1. Overall survival. TSEI, total skin electron irradiation; IFN, juvant (6 months) oral etretinate. Twenty-three patients were interferon. treated with this regimen. In comparison with controls treated with TSEI alone, the intensity of acute TSEI toxicity, as assessed by patients, was not increased by the etretinate but was prolonged by 2 weeks. As expected, the use of etretinate was complicated by epistaxis, dry skin and elevation in trigly- ceride and cholesterol levels. At a median follow-up of 2 years, adjuvant etretinate did not appear to confer an advantage in DFS. In 1997, Quiro´s et al.4 reported the outcomes of 14 stage IA–IB patients treated with PUVA following TSEI. These patients were maintained on PUVA until clinical relapse. The 5-year DFS and OS for this cohort were 85% and 100%, respectively. DFS appeared superior (P <0Æ018) to that of a cohort of 100 stage IA–IB patients treated either with TSEI alone (10 patients) or with TSEI and another form of adjuvant therapy (90 patients). Toxicity to PUVA was considered acceptable although all patients experienced some degree of Fig 2. Freedom from recurrence in complete responders. CR, erythema/hyperpigmentation, dry skin and/or pruritus. complete response; TSEI, total skin electron irradiation; IFN, Many secondary cutaneous malignancies were seen in interferon. patients treated with adjuvant PUVA: malignant melanoma (two patients), basal cell carcinoma (five patients), squamous cell carcinoma (three patients) or the combination of both 0Æ04). Age and treatment group were not significant predictors basal cell carcinoma and squamous cell carcinoma (two of OS or DFS. There was a trend towards the use of prior sys- patients). temic chemotherapy predicting for a worse OS (P ¼ 0Æ057). In a 1999 retrospective review from Stanford University, Chinn et al.16 reported on the adjuvant use of topical nitrogen Toxicity mustard following TSEI. In patients with T2 disease having achieved CR to TSEI, topical nitrogen mustard was associated The primary treatment-related adverse effect in both groups with a longer freedom from relapse but did not improve OS. was acute dermatitis. Grade II–III dermatitis was observed in In a 2000 retrospective series of 44 patients with stage III– all patients. Fever, chills or myalgia were seen in 32% of IV erythrodermic MF, Wilson et al.17 have shown improve- TSEI + IFN patients and none of the TSEI patients. One 83- ments in progression-free and cause-specific survival when year-old patient had to discontinue IFN prior to TSEI because extracorporeal photopheresis was added to TSEI. In a series of anorexia, nausea, fever and depressed mood. Mild asymp- from the same author including early-stage patients, there was tomatic leucopenia and thrombocytopenia were each seen in a trend (P ¼ 0Æ06) towards improved OS for T3–T4 patients one patient treated with TSEI + IFN. treated, following TSEI, with either adjuvant doxorubicin/cyclophosphamide chemotherapy or photopheresis.18 This trend Discussion was not seen in the subset of patients with T1 or T2 disease. In 2003, Duvic et al.19 reported on the long-term results In the management of MF, TSEI is effective in producing high of a prospective trial of combined modality treatment of MF. response rates. Unfortunately, even in favourable stage patients Patients were treated with a 4-month induction regimen of

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp57–61 60 TSEI with adjuvant interferon alfa-2b, D. Roberge et al. oral isotretinoin and IFN alfa-2b. This induction phase was References followed by TSEI (for disease stages IA–IIA) or six cycles of 1 Siegel RS, Pandolfino T, Guitart J et al. Primary cutaneous T-cell multiagent cytotoxic chemotherapy (for disease stages IIB– lymphoma: review and current concepts. J Clin Oncol 2000; IVB). Patients with late-stage disease proceeded to TSEI fol- 18:2908–25. lowing chemotherapy. At the end of TSEI, patients in CR 2 Kim YH, Jensen RA, Watanabe GL et al. Clinical stage IA (limited were treated with maintenance IFN for 1 year and topical patch and plaque) mycosis fungoides. A long-term outcome analy- mechlorethamine for 2 years. Of 95 patients treated with this sis. Arch Dermatol 1996; 132:1309–13. regimen, 56 experienced a CR. The median duration of these 3 Kim YH, Liu HL, Mraz-Gernhard S et al. Long-term outcome of ´ CRs was 29 months (62 months for stages IA–IIA and 525 patients with mycosis fungoides and Sezary syndrome: clinical prognostic factors and risk for disease progression. Arch Dermatol 7 months for stages IIB–IV). Age and stage were both prog- 2003; 139:857–66. nostic factors for DFS. Actuarial 5-year OS was 66% for the 4 Quiro´s PA, Jones GW, Kacinski BM et al. Total skin electron beam entire group. therapy followed by adjuvant psoralen/ultraviolet-A light in the In our series, IFN failed to increase initial response to radi- management of patients with T1 and T2 cutaneous T-cell lymph- ation or to prevent recurrences following CR. Our series is oma (mycosis fungoides). Int J Radiat Oncol Biol Phys 1997; 38:1027– heterogeneous and the two groups of interest were not com- 35. parable for disease stage. Correcting for this, the TSEI + IFN 5 Freeman CR, Suissa S, Shenouda G et al. Clinical experience with a single field rotational total skin electron irradiation tech- group still fared worse than the TSEI group (although the dif- nique for cutaneous T-cell lymphoma. Radiother Oncol 1992; ference was reduced). Other than the small sample size, there 24:155–62. are several possible explanations for the lack of a positive 6 Chang AY, Keng PC. Potentiation of radiation cytotoxicity by re- effect of the addition of IFN to TSEI. The groups may have combinant interferons, a phenomenon associated with increased been imbalanced for other prognostic factors not controlled blockage at the G2-M phase of the cell cycle. Cancer Res 1987; for in our analysis. For example, pre-TSEI treatments were 47:4338–41. quite heterogeneous and differed in the two groups. It is also 7 Angioli R, Sevin BU, Perras JP et al. In vitro potentiation of radiation cytotoxicity by recombinant interferons in cervical cancer cell lines. possible that IFN was not given at sufficiently high doses or Cancer 1993; 71:3717–25. 8 for a sufficiently long period. In the series of Kuzel et al. 8 Kuzel TM, Roenigk HH Jr, Samuelson E et al. Effectiveness of inter- looking at adjuvant IFN added to PUVA, the target dose was feron alfa-2a combined with phototherapy for mycosis fungoides ) 12 · 106 IU m 2 and treatment was continued for and the Se´zary syndrome. J Clin Oncol 1995; 13:257–63. 24 months. In the series from Duvic et al.,19 starting doses of 9 Bunn PA Jr, Foon KA, Ihde DC et al. Recombinant leukocyte A 5 · 106 IU three times a week (in combination with isotretin- interferon: an active agent in advanced cutaneous T-cell lympho- oin) were found to be intolerable and the schedule had to be mas. Ann Intern Med 1984; 101:484–7. 10 Bunn PA Jr, Hoffman SJ, Norris D et al. Systemic therapy of cutane- changed to 3 · 106 IU three times a week. In our own series, ous T-cell lymphomas (mycosis fungoides and the Se´zary syn- added toxicity of the concurrent use of IFN may have led to a drome). Ann Intern Med 1994; 121:592–602. decrease in the dose-intensity of the radiation: the median 11 Knobler RM, Trautinger F, Radaszkiewicz T et al. Treatment of cuta- TSEI treatment time was 9 days longer in the TSEI + IFN neous T cell lymphoma with a combination of low-dose interferon group (P ¼ 0Æ005). However, although prolongation of over- alfa-2b and retinoids. J Am Acad Dermatol 1991; 24:247–52. all treatment time has clearly been shown to be a poor prog- 12 Suchin KR, Cucchiara AJ, Gottleib SL et al. Treatment of cutaneous nostic factor in the treatment of selected solid tumours,20,21 T-cell lymphoma with combined immunomodulatory therapy: a 14-year experience at a single institution. Arch Dermatol 2002; this has not been shown in CTCL. Moreover, at 14–18 daily 138:1054–60. fractions, the TSEI schedules used at McGill University are 13 Podgorsak EB, Pla C, Pla M et al. Physical aspects of a rotational shorter than that those commonly used at other institu- total skin electron irradiation. Med Phys 1983; 10:159–68. tions.4,16,19 14 Kaye FJ, Bunn PA Jr, Steinberg SM et al. A randomized trial com- In our opinion, our data do not suggest the addition of IFN paring combination electron-beam radiation and chemotherapy to TSEI as a promising avenue in increasing the therapeutic with topical therapy in the initial treatment of mycosis fungoides. ratio in the treatment of MF. We have dropped the routine N Engl J Med 1989; 321:1784–90. 15 Jones G, McLean J, Rosenthal D et al. Combined treatment with oral use of concurrent IFN in our clinical practice. No adjuvant etretinate and electron beam therapy in patients with cutaneous T- therapy has clearly been shown to affect OS in patients treated cell lymphoma (mycosis fungoides and Se´zary syndrome). JAm for MF with TSEI. Current adjuvant therapies appear to trade Acad Dermatol 1992; 26:960–7. possible prolongations in DFS for increased treatment toxicity, 16 Chinn DM, Chow S, Kim YH et al. Total skin electron beam therapy and the optimal balance in the therapeutic ratio remains to be with or without adjuvant topical nitrogen mustard or nitrogen defined. mustard alone as initial treatment of T2 and T3 mycosis fungoides. In conclusion, based on our retrospective series, concur- Int J Radiat Oncol Biol Phys 1999; 43:951–8. 17 Wilson LD, Jones GW, Kim D et al. Experience with total skin rent treatment of MF with IFN and TSEI is feasible, with in- electron beam therapy in combination with extracorporeal creased but acceptable acute toxicity. Even when controlling photopheresis in the management of patients with for disease stage and other prognostic factors, the addition erythrodermic (T4) mycosis fungoides. J Am Acad Dermatol 2000; of IFN did not appear to increase CR rate, duration of CR, 43:54–60. DFS or OS.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp57–61 TSEI with adjuvant interferon alfa-2b, D. Roberge et al. 61

18 Wilson LD, Licata AL, Braverman IM et al. Systemic chemotherapy 20 Langendijk JA, de Jong MA, Leemans CR et al. Postoperative radio- and extracorporeal photochemotherapy for T3 and T4 cutaneous therapy in squamous cell carcinoma of the oral cavity: the import- T-cell lymphoma patients who have achieved a complete response ance of the overall treatment time. Int J Radiat Oncol Biol Phys 2003; to total skin electron beam therapy. Int J Radiat Oncol Biol Phys 1995; 57:693–700. 32:987–95. 21 Souhami L, Corns R, Duclos M et al. Long-term results of high-dose 19 Duvic M, Apisarnthanarax N, Cohen DS et al. Analysis of long-term rate brachytherapy in cervix cancer using a small number of frac- outcomes of combined modality therapy for cutaneous T-cell tions. Gynecol Oncol 2005; 97:508–13. lymphoma. J Am Acad Dermatol 2003; 49:35–49.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp57–61 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07564.x Transcriptional proﬁles in melanocytes from clinically unaffected skin distinguish the neoplastic growth pattern in patients with melanoma C. Magnoni, E. Tenedini,* F. Ferrari,* L. Benassi, C. Bernardi, G. Gualdi, G. Bertazzoni, E. Roncaglia,* L. Fantoni,* R. Manfredini,* S. Bicciato, S. Ferrari,* A. Giannetti and E. Tagliaﬁco* Dipartimento di Medicine e Specialita` Mediche, Universita` di Modena e Reggio Emilia, Via del Pozzo 71, 41100 Modena, Italy *Sezione di Chimica Biologica, Dipartimento di Scienze Biomediche, Universita` di Modena e Reggio Emilia, Via Campi 287, 41100 Modena, Italy Dipartimento di Processi Chimici dell’Ingegneria, Universita` di Padova, Via Marzolo 9, 35131 Padua, Italy

Summary

Correspondence Background It is generally accepted that sunlight may contribute to the development Enrico Tagliafico. of melanoma. E-mail: tagliafi[email protected] Objectives To analyse gene expression of melanocytes obtained from clinically unaffected skin of patients with melanoma and healthy controls before and after ex- Accepted for publication 14 July 2006 posure to ultraviolet B radiation. Methods Using GeneChip array technology, the gene expression of melanocytes Key words obtained from the two donor groups was profiled, in order to identify transcrip- gene expression profiling, melanocytes, melanoma, tional differences affecting susceptibility to melanoma. microarrays, transcriptome Results The data collected did not show any difference between the expression Conflicts of interest profiles of melanocytes purified from normal donors and from patients with None declared. melanoma that was able to give a statistically significant class separation. How- ever, by means of unsupervised clustering our data could be divided into two main classes. The first class included the transcriptome profiles of melanocytes obtained from skin samples of patients with a vertical growth phase (VGP) melanoma, while the second class included the transcriptome profiles of melanocytes obtained from skin samples of patients with a radial growth phase (RGP) melanoma. Conclusions These data suggest that melanocytes in patients with VGP and RGP melanomas show significant differences in gene expression profiles, which allow us to classify patients with melanoma also from clinically unaffected skin.

Cutaneous malignant melanoma is the most aggressive among epidermis and the underlying dermis, shows high potential skin cancers and currently has become an important public for metastatic spread and is associated with a worse prognosis. health problem because of its increasing incidence worldwide. The presence of an ulceration that opens up to vascular access Melanoma progression is well defined in its clinical and histo- is a negative prognostic factor.1,2 However, melanomas may pathological aspects and in recent years many efforts have behave in an unexpected manner regardless of their size and been made to identify new immunological, serological and thickness. In some patients, early-detected small melanomas immunohistochemical parameters in order to formulate more display aggressive phenotype with a VGP, while in some oth- precise prognosis. Each parameter has led to promising results, ers melanoma shows an RGP for many years without convert- but has not provided a reliable prognostic factor in itself. Cur- ing to a VGP. Moreover, 5–22% of thin melanomas (including rently, prognoses for patients with melanoma are still based those <0Æ7 mm thickness) are characterized by metastatic dis- on tumour thickness and ulceration. Early stages of the disease and extremely rapid progression, while 25–50% of ease, such as melanoma in situ and microinvasive melanomas patients whose tumours are classified as ‘thick’ (>4 mm) confined to the epidermis and showing a radial growth phase inexplicably survive.3 Clearly, this prognostic characterization (RGP), are easily treated and are associated with a good prog- of melanoma based primarily on tumour thickness is inad- nosis. The invasive nodular melanoma characterized by a verti- equate and cannot explain the variability of clinical behaviour. cal growth phase (VGP), penetrating both the upper layer of Therefore, there is a strong need to understand the molecular

2006 The Authors 62 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 Gene expression of melanocytes in RGP vs. VGP melanoma, C. Magnoni et al. 63 mechanism regulating the progression of melanoma in order vival gene, amplified in metastatic malignant melanoma. At to provide criteria with predictive value for prognosis. the same time, other expression profile studies have been used To date, many authors have provided evidence that mela- to identify molecular predictors of treatment response. For noma, like many other cancers, has a genetic basis. It has long example, Wang et al.24 studied gene expression of melanoma been observed that melanomas can cluster in some families, metastases in patients before and after the administration of suggesting that defective genes can be passed through succes- immunotherapy. Only a few gene expression analysis studies sive generations of affected kindred.4 Again, evidence shows included normal human melanocytes and most of these studies that phenotypic traits genetically determined, such as high analysed normal melanocyte cell lines to determine normal naevus counts and a fair skin phenotype, represent important cell molecular markers or differences between normal and ma- risk factors for melanoma development.5 lignant melanocytes. These observations have been strengthened by experimental Based on current knowledge we can hypothesize that differ- data obtained in the last few years, which led to the identifica- ences in gene expression in melanocytes may influence predis- tion of some key genetic lesions associated with melanoma position, progression and prognosis of melanoma. The aim of onset and progression. These include losses of chromosomes our study was to identify any transcriptional behaviour that 6q, 8p and 10 and gains in copy number of chromosomes may discriminate normal melanocytes obtained from healthy 1q, 6p, 7 and 8 in primary melanomas,6,7 amplification of the control subjects from normal melanocytes obtained from 8,9 cyclin D1 locus in acral melanoma, frequent deletion of patients with melanoma, in order to detect any difference af- chromosome 13 and 17p in melanomas arising in chronically fecting personal susceptibility to develop this disease. To this sun-damaged skin, and frequent amplification of chromosome purpose we applied microarray technology to analyse global 12q in mucosal melanomas that can affect metastatic potential. gene expression of normal melanocytes obtained from two Moreover, mutations in the p16,10,11 p19ARF12 and PTEN/ donor groups: 12 patients with melanoma and nine healthy MMAC113,14 genes are involved in earliest events in mela- subjects. noma, whereas mutations in NRAS,15,16 BRAF17,18 and MYC19 It is generally accepted that sunlight may contribute to the genes are involved in the late phase of tumour progression development of melanoma and epidemiological evidence and metastatic dissemination. It is conceivable that the pres- showing the role of sunlight in causing melanoma onset is ence of some specific genetic alterations may influence mela- compelling. It seems that melanoma development is more as- noma prognosis and may predispose to development of early sociated with intense sunburn during childhood than with the metastatic or slowly progressive forms of the disease inde- lifetime level of total cumulative exposure to ultraviolet (UV) pendently from tumour thickness. Therefore, the identification radiation, but the exact relationship between malignant mela- of these genetic lesions may be a useful criterion to predict nomas and sun exposure has not yet been identified.25 There- the prognosis and to find new therapeutic approaches. fore, for 14 of the 21 available samples of melanocytes (from In the past few years, microarray technology has been seven patients with melanoma and seven healthy subjects) we applied to melanoma research in identifying interesting mole- performed gene expression profiling both before and after cules which may be of relevance for tumour development and UVB exposure. progression and may represent new potential therapeutic targets. Nevertheless, most researchers have focused on gene Materials and methods expression profiling of melanoma cell lines or malignant cells. In particular, Bittner et al.20 recently studied the expression Cell culture profile of different melanoma tumours and defined two puta- tive subsets with marker genes that were differentially Primary cultures of normal melanocytes were established from expressed. Many of the genes that distinguish the different skin biopsies obtained from patients with sporadic melanoma subsets are important for cell motility and invasive capacity, and from healthy control subjects with no personal or familial and in particular a dysregulation of the WNT5A pathway has history of melanoma. Skin biopsies from patients with mela- been identified. These results, validated by successive experi- noma were taken from healthy skin during the procedure of ments employing melanoma cells overexpressing WNT5A, selective sentinel lymphadenectomy. The characteristics of suggested that melanomas with decreased WNT5A expression patients with melanoma are listed in Table 1. Skin biopsies are less aggressive than tumours carrying an overexpression of from healthy control subjects were obtained both from plastic WNT5A, that show an increased potential for invasiveness.21 surgery procedures and from circumcision. The characteristics More recently, while analysing cell lines with different meta- of healthy control subjects are listed in Table 2. Informed con- static capacity, Clark et al.22 identified genes preferentially sent was obtained from patients according to the procedures expressed in cells with high invasive potential including genes approved by the Review Board of the University of Modena that regulate cytoskeletal organization, cellular motility and and Reggio Emilia. migration, such as fibronectin, thymosin b4 and RhoC. Finally, Skin biopsies were floated on a dispase solution (0Æ5%) at Garraway et al.23 combined gene expression data and single 4 C overnight. After separation of the epidermis, epidermal nucleotide polymorphism genotyping data, obtained using cells were plated at high density in Medium 154 (Cascade microarray technology, and identified MITF as a lineage sur- Biologics Inc., Portland, OR, U.S.A.) supplemented with

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 64 Gene expression of melanocytes in RGP vs. VGP melanoma, C. Magnoni et al.

Table 1 Clinical parameters for patients with melanoma

Primary Growth Tumour Sentinel Regional Visceral Age Sample melanoma pattern thickness (mm) lymph node metastases metastases (years) Sex Biopsy site M3 Left shoulder Radial 0Æ43 Negative – – 68 M Axilla M4 Mastoid Vertical 6 Positive Neck Liver, spleen 69 M Laterocervical M5 Gluteus Vertical 3Æ5 Positive Leg – 72 F Groin M6 Right ear Radial 0Æ6 Negative – – 34 F Laterocervical M8 Axilla Vertical 8 Positive Liver, spleen 47 ? Axilla M17 Foot Vertical 3Æ5 Negative – – 59 F Groin M18 Left side Radial 0Æ75 Negative – – 59 M Axilla M35 Thorax Vertical 4Æ00 Positive Thorax – 63 M Axilla M45 Right foot Radial 9Æ00 Positive Right foot – 68 M Groin M46 Back Vertical In situ –––45F– M49 Back Radial In situ –––62M– M52 Back Vertical 0Æ69 Negative – – 54 M Groin

Table 2 Clinical parameters for healthy donors U.S.A.) were used to determine the concentration and purity/ integrity of RNA samples using the Agilent 2100 bioanalyser. Sample Age (years) Sex Biopsy site A24 9 M Foreskin Target synthesis, GeneChip hybridization and data A26 5 M Foreskin acquisition A28 2 M Foreskin A4 20 M Foreskin Biotin-labelled target synthesis was performed, starting from A6 52 F Abdomen 5 lg of total cellular RNA, according to the protocol supplied A11 44 F Abdomen by the manufacturer (Affymetrix, Santa Clara, CA, U.S.A.). A12 38 M Abdomen Labelled cRNA was purified using RNeasy spin columns (Qi- A31 24 F Abdomen A36 69 F Abdomen agen) and was fragmented (15 lg) as described in the Affyme- trix GeneChip protocol. Disposable RNA chips (Agilent RNA 6000 Nano LabChip kit) and Agilent 2100 bioanalyser were used to determine the concentration and quality of cRNAs as phorbol-12-myristate 13-acetate, transferrin, hydrocortisone, well as to optimize the fragmentation. The fragmented cRNAs insulin, bovine pituitary extract, basic fibroblast growth factor were then hybridized to Affymetrix HG-U133A GeneChip and fetal calf serum. After 24 h the cells were transferred into arrays for 16 h. GeneChips were washed and stained using the fresh medium. After three passages in culture, at near conflu- instrument’s standard Eukaryotic GE WS2 protocol, using anti- ence, the medium was removed and the cells were washed body-mediated signal amplification. GeneChips were finally twice with warm phosphate-buffered saline (PBS) with cal- scanned using the Affymetrix GeneChip scanner. cium and magnesium. Melanocytes from seven patients with melanoma and seven healthy subjects were treated with a sin- Normalization and intensity calculation gle UVB dose calibrated with a GoldiluxTM smart meter (Oriel, ) Stratford, CT, U.S.A.). The dose was 75 mJ cm 2. After UVB The detection call and amount of a transcript mRNA (signal) irradiation the PBS was replaced with fresh medium and cells was determined, using the MAS 5.0 absolute analysis algorithm were harvested after 4 h for RNA extraction. Control melano- as already described.26 All expression values for the genes in cytes both from patients with melanoma and from healthy the MAS 5.0 absolute analyses were determined using the glo- subjects were mock irradiated, the PBS was replaced with fresh bal scaling option that allows a number of experiments to be medium and cells were harvested after 4 h for RNA extraction. normalized to one target intensity. Alternatively, probe level data were converted to expression values using robust multiar- ray average with adjustment for probe sequence (GCRMA) pro- RNA extraction cedure.27 PM values were background adjusted, normalized Total cellular RNA was isolated from cultured cells using using invariant set normalization, and log transformed. RNeasy RNA isolation kit (Qiagen, Valencia, CA, U.S.A.) following the manufacturer’s recommendations. Subsequent RNA Statistical filtering and analysis extraction using phenol/chloroform/isoamyl alcohol (25 : 24 : 1, pH 6Æ6) was performed in order to remove The GCRMA-generated normalized data were uploaded on to residual contaminating melanin. Disposable RNA chips (Agi- GeneSpringTM software version 7.2 (Silicon Genetics, lent RNA 6000 Nano LabChip kit; Agilent, Palo Alto, CA, Redwood City, CA, U.S.A.) using the log2 transformation pro-

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 Gene expression of melanocytes in RGP vs. VGP melanoma, C. Magnoni et al. 65 cedure. A ‘per gene’ normalization was achieved by dividing and UV-irradiated melanocytes). Our results show that samples each signal by the median of its values in all samples. In order could be confidently separated into two main groups showing to remove genes not expressed or always expressed at low lev- two classes of differences. The first class of differences is rela- els, expression data were filtered using GeneSpringTM to select ted to the subtype of patients with melanoma: one group con- genes detected as ‘present’ according to the MAS 5.0 absolute tains the gene expression profiles of melanocytes isolated from analysis algorithm in at least 10% of samples. Then genes patients with VGP melanomas with the exception of patient whose normalized expression levels were always between 0Æ66 M17, while the other group contains profiles of melanocytes and 1Æ5 across all of the samples were filtered out. isolated from patients with RGP melanomas. The second class Welch’s t-test was performed with GeneSpringTM, using of differences is related to the biopsy location: one group con- multiple testing corrections, to find genes expressed at differ- tains the transcriptome profiles of melanocytes isolated from a ent levels with statistical significance between the compared trunk skin biopsy of normal donors, while the other group sample classes. Differentially expressed genes were also identi- contains profiles of melanocytes isolated from a foreskin biopsy fied using Significance Analysis of Microarray (SAM).28 of normal donors. Profiles of melanocytes isolated from a trunk Clustering analysis was performed using the analysis options skin biopsy of normal donors were clustered together with (gene trees and condition trees) included in the GeneSpringTM profiles of melanocytes isolated from patients with VGP mela- package, applying the standard and/or Pearson correlation nomas, while profiles of melanocytes isolated from a foreskin coefficient as a similarity measure. EASE29 software was used biopsy of normal donors were clustered together with profiles to examine selected lists of genes in order to identify over- of melanocytes isolated from patients with RGP melanomas. representation of functional classes according to the Gene Ontology (GO) classification. Identification of a signature associated with histological subtype in affected patients Results Our data suggest that there are differences in gene expression profiles of melanocytes, even if isolated from normal skin, that Unsupervised clustering identifies, among normal allow us to distinguish whether melanocytes were isolated melanocyte gene expression profiling, some groups from a patient with an RGP melanoma or with a VGP mela- corresponding to the melanoma histological subtype in noma. In order to identify these differences, we filtered out affected patients those genes showing lower than twofold difference between Clinical parameters for patients with melanoma, such as pri- the groups, obtaining a list of 2525 genes. On these genes we mary melanoma location, histological subtype etc., are sum- performed a Welch’s t-test (variances not assumed equal, marized in Table 1. P-value cutoff 0Æ01) using the Benjamini and Hockberg cor- Using an unsupervised hierarchical clustering approach, we rection for controlling false discovery rate: 279 genes passed tried to define natural subclasses of samples as determined by this test. Then, in order to improve consistency of our filter- gene expression profiles. We performed the unsupervised clus- ing procedure, we performed a SAM analysis on this subset of tering on a low-level filtered probe list (see Materials and 279 genes (number of permutations ¼ 100, minimum fold methods) using the Pearson correlation coefficient. The cluster- change ¼ 2; median false discovery rate ¼ 0). Thus we came ing results are shown in Figure 1. The dendrogram shows that up with a reliable signature of 233 transcripts (Fig. 2): 150 samples were cut by cluster algorithm into groups unexpect- showing a higher expression in normal melanocytes isolated edly neither associated with the melanocyte source (normal or from patients with a VGP melanoma and 83 transcripts show- melanoma-affected donors) nor with the treatment (untreated ing a higher expression in normal melanocytes isolated from

Fig 1. Unsupervised clustering. We performed unsupervised hierarchical clustering on a low-level ﬁltered probe list (see Materials and methods), using the Pearson correlation coefﬁcient as similarity measure. Samples divide into two main groups. Sample numbers are as given in Tables 1 and 2; C, untreated; UV, treated with ultraviolet radiation; VGP, vertical growth phase; RGP, radial growth phase.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 66 Gene expression of melanocytes in RGP vs. VGP melanoma, C. Magnoni et al.

Fig 2. Signature of normal melanocytes isolated from patients with vertical growth phase (VGP) and radial growth phase (RGP) melanomas (sample numbers are as given in Table 1). We identified 233 transcripts with statistically significant differences between expression level in melanocytes that were isolated from clinically unaffected skin of patients with RGP melanoma and with VGP melanoma. The signature was obtained using Welch’s t-test (P-value cutoff 0Æ01 and Benjamini and Hockberg false discovery rate as multiple testing correction) and Significance Analysis of Microarray analysis (number of permutations ¼ 100, minimum fold change ¼ 2; median false discovery rate ¼ 0).

patients with an RGP melanoma. The complete gene list is Identification of genes showing statistically significant available on our website (http://www.xlab.unimo.it/XLab/ differences between untreated and ultraviolet-treated ExpDataView.php). The signature gene list was uploaded on groups to EASE software to identify prevalent categories in the GO controlled vocabulary families ‘biological process’, ‘molecular Although an unsupervised approach testing global correlation function’ and ‘cellular component’. As shown in Table 3, between conditions failed to cut samples into untreated and ‘morphogenesis’, ‘cell adhesion’, ‘cell motility’, ‘cell prolifer- UV-treated groups, we attempted to identify, by means of a ation’ and ‘angiogenesis’ were prevalent GO categories in supervised approach, those genes that could be confidently ‘biological process’; among the ‘molecular function’ categories considered as modulated after UV treatment. In doing so, we we detected as statistically relevant ‘growth factor binding’; fi- analysed the gene expression data obtained from UV-treated nally, ‘extracellular space’ and ‘extracellular matrix’ were samples by comparing the transcriptome profiles of UV-treated among the most prevalent ‘cellular component’ categories. samples with those of the corresponding untreated samples.

Table 3 Prevalent Gene Ontology (GO) categories identiﬁed by EASE results we also performed a SAM analysis on this subset of software among the signature genes associated with histological 343 genes (number of permutations ¼ 100, fold change ¼ 2; subtype in affected patients median false discovery rate ¼ 0). Thus we obtained a gene list of 215 transcripts that constitute a consistent list of genes GO branch Gene category P-value modulated following UV treatment (Fig. 3): 164 genes GO biological process Morphogenesis 1Æ11E-07 downregulated and 51 genes upregulated by UV treatment GO biological process Growth 0Æ000202434 (complete data are available on our website: http:// GO biological process Cell adhesion 0Æ001002577 www.xlab.unimo.it/XLab/ExpDataView.php). These lists were GO biological process Regulation of cell cycle 0Æ006463384 uploaded on to EASE software to identify prevalent categories GO biological process Cell proliferation 0Æ007154648 GO biological process Cell migration 0Æ011120057 in the GO categories. Among the most statistically relevant cat- GO biological process Cell motility 0Æ013749415 egories we detected ‘nucleic acid binding’, ‘transcription regu- GO biological process Angiogenesis 0Æ03043848 latory activity’, ‘nucleobase, nucleoside, nucleotide and GO biological process Cell growth 0Æ034809746 nucleic acid metabolism’ and ‘nucleotide binding’. In order to GO biological process Blood vessel 0Æ034946228 test the hypothesis that melanocytes isolated from patients development with melanoma could exhibit a response to UV treatment dif- GO biological process Blood vessel 0Æ034946228 ferent from the one shown in normal donors, we performed a morphogenesis GO molecular function Protein binding 5Æ66E-07 two-way ANOVA analysis testing the statistical relevance of the GO molecular function Growth factor binding 2Æ65E-05 difference between gene expression proﬁles considering the GO cellular component Extracellular region 1Æ60E-06 ‘treatment’ and ‘donor’ parameters together. Unfortunately, GO cellular component Extracellular space 2Æ22E-05 we failed to obtain genes showing statistically relevant differ- GO cellular component Extracellular matrix 0Æ001504371 ences in the transcriptional response to UV damage.

Discussion

We filtered out genes showing lower than twofold difference We profiled gene expression of melanocytes obtained from between the groups, obtaining a list of 1352 genes. Then we clinically unaffected skin samples of patients with melanoma performed on this gene list a Welch’s t-test (variances not and healthy controls. This work was driven by the assumption assumed equal, P-value cutoff 0Æ01) using the Westfall and that the presence of some specific gene expression patterns, Young multiple testing correction: 343 genes passed this test. already characterizing normal melanocytes, may influence Finally, as described above, in order to obtain more reliable melanoma progression and prognosis. In other words, we were

Fig 3. Signature of untreated and ultraviolet (UV)-treated melanocytes. We identified 215 transcripts with statistically significant differences between expression level in untreated (C) and UV-treated melanocytes (sample numbers are as given in Tables 1 and 2). The signature was obtained using Welch’s t-test (P-value cutoff 0Æ01 and Westfall and Young permutation method in order to control the family-wise error rate) and Significance Analysis of Microarray analysis (number of permutations ¼ 100, minimum fold change ¼ 2; median false discovery rate ¼ 0).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 68 Gene expression of melanocytes in RGP vs. VGP melanoma, C. Magnoni et al. looking for any transcriptional behaviour that might predispose that is a member of the vascular endothelial growth factor normal melanocytes to develop a melanoma and/or determine (VEGF) family; NRP1,33,34 that is a VEGF coreceptor involved its biological features. The data collected showed some differ- in various tumours; CYR61,35–37 that is thought to act as an ences between gene expression profiles of melanocytes angiogenetic factor on melanoma cells; FBLN5;38 CTGF;39 and obtained from clinically unaffected skin samples of patients DDAH1.40 with melanoma and healthy controls, but such differences Secondly, regarding cell motility and matrix remodelling, were not sufficient to allow a statistically significant separation CKAP/p6341 and MRC242 regulate plasminogen activation and of the two classes of samples. However, unexpectedly, the are involved in matrix remodelling and cellular movement. unsupervised clustering showed that our data could be divided Thirdly, regarding cell motility and cell morphology, into two different main groups. The first group includes the EDG2,43,44 FBLN5,38,45 TWIST146 and fibronectin (FN1)47–49 transcriptome profiles of melanocytes obtained from skin sam- can stimulate cell migration and also morphology changes of ples of patients with a VGP melanoma as well as from trunk various cell types. skin biopsies of healthy donors, while the second group Lastly, regarding cell morphology and matrix remodelling, includes the transcriptome profiles of melanocytes obtained some other genes have a role in regulating either cell mor- from skin samples of patients with an RGP melanoma as well phology or interaction among cells and extracellular matrix: as from foreskin samples of healthy donors. These results show FSCN1;50 ADM;51 VCL,52 that is the major constituent of focal that there may be a trait in patients with VGP and RGP mela- adhesions; and COL5A1,53 that is a constituent of the extracel- nomas that allows for a possible classification of patients with lular matrix. melanoma also using clinically unaffected skin. This idea has Next, among genes with higher expression level in patients already been put forward and there are some proofs of the with VGP melanoma we found some cellular proliferation reg- 30 54 concept. Some evidence has come from Burczynski et al., ulators: Cyclin D2, involved in G1/S cell cycle transition; who provided an elegant demonstration that gene expression NOTCH2, belonging to the NOTCH genes family and involved profiling in peripheral blood, a clinically accessible surrogate in many tumours;55–58 CSPG2, that can play a role in stimula- tissue, can be used for identifying patterns of gene expression ting proliferation of melanoma cells;59 FGF5;60 CITED2;61 associated with higher likelihood of positive outcome in CRIP1;62 LOX;63 PRRX1;64,65 and PRRX2.64 Finally, we found patients with a solid tumour. In this study, the authors evalu- WNT5B,66 that seems to be involved in proliferation of some ated the association of expression profiles in peripheral blood cell types, and WNT5A, that seems to be involved both in cell mononuclear cells (PBMC) with clinical outcomes in patients proliferation66 and in regulation of cell adhesion and migra- with advanced renal cell cancer and demonstrated that unsu- tion.67,68 Some studies suggest a particular role for WNT5A in pervised hierarchical clustering of the PBMC profiles using all conferring invasive capacity to melanoma cells.21 expressed genes identified clusters of patients with significant Conversely, among the signature genes with lower expres- differences in survival. Furthermore, Cox proportional hazards sion in melanocytes from patients with VGP melanoma we modelling showed that the expression levels of many PBMC found some negative regulators of cell cycle progression: transcripts were predictors for patient outcomes in terms of ATM69 and LZTS1.70 Their lower expression could contribute progression and overall survival. Similarly, in our study, it to better conditions for cell proliferation. seems that some of the differences characterizing VGP and RGP Among the signature genes, we found some platelet-derived melanomas are probably driven by differences between tran- growth factor (PDGF) pathway genes with higher expression scriptional profiles also existing in normal cell populations. in melanocytes from patients with VGP melanoma: PDGFC In view of the above, we attempted to identify the distinc- (PDGF gamma chain), PDGF receptor alpha chain and PDGF tive genes for each group. The supervised Welch’s t-test and receptor beta chain. We consider this result particularly inter- SAM analyses allowed us to find a signature that is able to dis- esting because PDGFC is a ligand for PDGFRA and PDGFRB.71 tinguish significantly among the different groups. Within the In addition, there is evidence that PDGF can act as an auto- signature genes associated with melanoma histological sub- crine stimulus for melanoma cells.72 type, we found that melanocytes isolated from patients with This suggests that the PDGF pathway could be involved in VGP melanoma express at a higher level several genes that determining differences between these two groups of samples, play a role in regulating cell proliferation or in determining obtained from patients with different grade melanomas. the invasive capacity and hence metastatic potential of cells, as Indeed, we found that other PDGF pathway genes have a stat- suggested by the results of EASE software examination of istically significantly higher expression in patients with VGP, over-represented GO functional categories. Among the 150 even if these genes do not belong to the signature genes iden- signature genes with higher expression level in patients with tified and described above. Among these genes we found VGP melanoma and presumably involved in conferring upon PDGFA and PDGFD, that are other PDGF family members; and cells their invasive capacity and metastatic potential, we found SHC1, STAT1, PKR, JAK1 and STAT3, that are involved in genes controlling angiogenesis, cell motility, cell morphology PDGF signal transduction via the STAT pathway. This supports and extracellular matrix remodelling. the hypothesis of involvement of the PDGF pathway in deter- Firstly, we found various genes that are involved in stimula- mining significant differences among gene expression profiles ting normal or tumour-associated angiogenesis: VEGFC,31,32 of normal melanocytes retrieved from patients with VGP and

RGP melanomas, with different degrees of severity. We hypo- People with different genetic background such as the thesize that the second group of patients showing downregu- groups that we identified will probably react differently not to lation of some pathways, such as PDGF, under particular a single exposure to sunlight but to extreme conditions of ex- immunological conditions and interacting with specific envi- posure, such as intense exposure during childhood or to ronmental factors, can develop a less aggressive form of mela- chronic lifetime exposure. Subjects who were apparently not noma, prone to a long localized RGP. included in a high-risk category for melanoma development Surprisingly, as described before, we found that gene could be identified in the future as belonging to a high-risk expression profiles of melanocytes from foreskin of healthy group on the basis of a genetic analysis, and consequently donors cluster with profiles of melanocytes from patients with could be advised to avoid dangerous sun exposure behaviours. RGP melanoma. We have no explanation for these results; As a whole, the analysis of normal melanocytes through clustering of foreskin melanocytes with melanocytes from gene array could lead to the identification of a subset of sub- patients with RGP melanoma is possibly influenced by the jects with higher probability to develop more aggressive dis- physiology of this anatomical site (rare melanoma in this site) ease. Further research is needed to determine whether these or by the age of donors (rare melanoma in children). Foreskin new subsets discriminate patients with distinct natural history. is a generally accepted source of melanocytes and many In the future, gene expression profiling may be incorporated authors have utilized melanocytes from foreskin as cultured in clinical trials to screen subjects with a high risk to develop control cells compared with melanoma cells. Halaban et al. melanoma and to design tailored treatment regimens in studied foreskin melanocytes in culture and reported that the patients with melanoma. cells were highly differentiated as judged by melanin production, expression of melanocyte-specific proteins (tyrosinase, References gp75/tyrosinase related protein 1) and dendritic morphology.73,74 1 Barnhill RL, Mihm MC Jr. The histopathology of cutaneous malig- UV radiation from the sun has been implicated as the main nant melanoma. Semin Diagn Pathol 1993; 10:47–75. environmental factor responsible for the transformation of 2 Jhappan C, Noonan FP, Merlino G. Ultraviolet radiation and cutaneous malignant melanoma. Oncogene 2003; 22:3099–112. normal melanocytes into malignant cells.75 We chose to 3 Lomuto M, Calabrese P, Giuliani A. Prognostic signs in melanoma: expose cultured melanocytes to UVB, which is considered the state of the art. J Eur Acad Dermatol Venereol 2004; 18:291–300. most carcinogenic waveband, in order to verify if cells with 4 Goldstein AM, Tucker MA. Genetic epidemiology of familial mela- different genetic expression profiles could respond in different noma. Dermatol Clin 1995; 13:605–12. ways. We found no differences in the response to UVB, either 5 Bataille V, Bishop JA, Sasieni P et al. Risk of cutaneous melanoma between healthy donors and patients with melanoma, or in relation to the numbers, types and sites of naevi: a case–control between the groups associated with different histological sub- study. Br J Cancer 1996; 73:1605–11. 6 Chin L, Merlino G, DePinho RA. Malignant melanoma: modern types of melanoma. The basic UV response was identical in all black plague and genetic black box. Genes Dev 1998; 12:3467–81. samples, with transcriptional modulation of genes related to 7 Piepkorn M. Melanoma genetics: an update with focus on the cell cycle control, nucleus and DNA binding. We were expect- CDKN2A(p16)/ARF tumor suppressors. J Am Acad Dermatol 2000; ing this kind of response, because the environmental condi- 42:705–22. tions able to modify the genetic configuration are more 8 Sauter ER, Takemoto R, Litwin S et al. p53 alone or in combination complex than a single dose of UVB radiation. with antisense cyclin D1 induces apoptosis and reduces tumor size There are some data indicating that these tumours fre- in human melanoma. Cancer Gene Ther 2002; 9:807–12. 9 Sauter ER, Yeo UC, von Stemm A et al. Cyclin D1 is a candidate quently arise on areas of skin not exposed to sunlight, and oncogene in cutaneous melanoma. Cancer Res 2002; 62:3200–6. that indoor workers often have higher rates of the disease than 10 Hayward NK. Genetics of melanoma predisposition. Oncogene 2003; outdoor workers, suggesting that the relationship between UV 22:3053–62. radiation and melanoma development is complex and cannot 11 Talve L, Sauroja I, Collan Y et al. Loss of expression of the be simply explained on the basis of sunlight exposure. p16INK4/CDKN2 gene in cutaneous malignant melanoma corre- Moreover, exposure to UV radiation can probably lead to lates with tumor cell proliferation and invasive stage. Int J Cancer the development of melanoma when particular genetic charac- 1997; 74:255–9. 12 Kumar R, Sauroja I, Punnonen K et al. Selective deletion of exon 1 teristics of the host interact with different conditions of expos- beta of the p19ARF gene in metastatic melanoma cell lines. Genes ure. The ‘divergent pathway’ model is currently the most Chromosomes Cancer 1998; 23:273–7. reliable theory explaining the relationship between UV and 13 Tsao H, Goel V, Wu H et al. Genetic interaction between NRAS melanoma. Under this model, subjects with inherently low and BRAF mutations and PTEN/MMAC1 inactivation in melanoma. propensity for melanocyte proliferation require chronic sun J Invest Dermatol 2004; 122:337–41. exposure to develop clonal expansion of transformed epider- 14 Poetsch M, Dittberner T, Woenckhaus C. PTEN/MMAC1 in malig- mal melanocytes. In contrast, in subjects with inherently high nant melanoma and its importance for tumor progression. Cancer Genet Cytogenet 2001; 125:21–6. propensity for melanocyte proliferation (as characterized by 15 Kumar R, Angelini S, Hemminki K. Activating BRAF and N-Ras high naevus count) exposure to sunlight is crucial at the early mutations in sporadic primary melanomas: an inverse association stages of the carcinogenesis process, after which host factors with allelic loss on chromosome. Oncogene 2003; 22:9217–24. drive melanoma development independently.76

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 70 Gene expression of melanocytes in RGP vs. VGP melanoma, C. Magnoni et al.

16 Reifenberger J, Knobbe CB, Sterzinger AA et al. Frequent alterations taxis through integrin alpha(6)beta(1) and cell surface heparan of Ras signaling pathway genes in sporadic malignant melanomas. sulfate proteoglycans. Endocrinology 61 2002; 1441–50. Int J Cancer 2004; 109:377–84. 37 Mo FE, Muntean AG, Chen CC et al. CYR61 (CCN1) is essential for 17 Kumar R, Angelini S, Czene K et al. BRAF mutations in metastatic placental development and vascular integrity. Mol Cell Biol 2002; melanoma: a possible association with clinical outcome. Clin Cancer 22:8709–20. Res 2003; 9:3362–8. 38 Spencer JA, Hacker SL, Davis EC et al. Altered vascular remodeling 18 Omholt K, Platz A, Kanter L et al. NRAS and BRAF mutations arise in fibulin-5-deficient mice reveals a role of fibulin-5 in smooth early during melanoma pathogenesis and are preserved throughout muscle cell proliferation and migration. Proc Natl Acad Sci USA 2005; tumor progression. Clin Cancer Res 2003; 9:6483–8. 102:2946–51. 19 Kraehn GM, Utikal J, Udart M et al. Extra c-myc oncogene copies 39 Ivkovic S, Yoon BS, Popoff SN et al. Connective tissue growth factor in high risk cutaneous malignant melanoma and melanoma meta- coordinates chondrogenesis and angiogenesis during skeletal devel- stases. Br J Cancer 2001; 84:72–9. opment. Development 2003; 130:2779–91. 20 Bittner M, Meltzer P, Chen Y et al. Molecular classification of cuta- 40 Kostourou V, Robinson SP, Cartwright JE et al. Dimethylarginine neous malignant melanoma by gene expression profiling. Nature dimethylaminohydrolase I enhances tumour growth and angiogen- 2000; 406:536–40. esis. Br J Cancer 2002; 87:673–80. 21 Weeraratna AT, Jiang Y, Hostetter G et al. Wnt5a signaling directly 41 Razzaq TM, Bass R, Vines DJ et al. Functional regulation of tissue affects cell motility and invasion of metastatic melanoma. Cancer Cell plasminogen activator on the surface of vascular smooth muscle 2002; 1:279–88. cells by the type-II transmembrane protein p63 (CKAP4). J Biol 22 Clark EA, Golub TR, Lander ES et al. Genomic analysis of metastasis Chem 2003; 278:42679–85. reveals an essential role for RhoC. Nature 2000; 406:532–5. 42 Engelholm LH, List K, Netzel-Arnett S et al. uPARAP/Endo180 is 23 Garraway LA, Widlund HR, Rubin MA et al. Integrative genomic essential for cellular uptake of collagen and promotes fibroblast analyses identify MITF as a lineage survival oncogene amplified in collagen adhesion. J Cell Biol 2003; 160:1009–15. malignant melanoma. Nature 2005; 436:117–22. 43 van Leeuwen FN, Olivo C, Grivell S et al. Rac activation by lyso- 24 Wang E, Miller LD, Ohnmacht GA et al. Prospective molecular pro- phosphatidic acid LPA1 receptors through the guanine nucleotide filing of melanoma metastases suggests classifiers of immune re- exchange factor Tiam1. J Biol Chem 2003; 278:400–6. sponsiveness. Cancer Res 2002; 62:3581–6. 44 Weiner JA, Fukushima N, Contos JJ et al. Regulation of Schwann 25 Ortonne JP. Photobiology and genetics of malignant melanoma. Br cell morphology and adhesion by receptor-mediated lysophosphati- J Dermatol 2002; 146 (Suppl. 61):11–16. dic acid signaling. J Neurosci 2001; 21:7069–78. 26 Liu WM, Mei R, Di X et al. Analysis of high density expression 45 Schiemann WP, Blobe GC, Kalume DE et al. Context-specific effects microarrays with signed-rank call algorithms. Bioinformatics 2002; of fibulin-5 (DANCE/EVEC) on cell proliferation, motility, and inva- 18:1593–9. sion. Fibulin-5 is induced by transforming growth factor-beta and 27 Irizarry RA, Hobbs B, Collin F et al. Exploration, normalization, affects protein kinase cascades. J Biol Chem 2002; 277:27367–77. and summaries of high density oligonucleotide array probe level 46 Kwok WK, Ling MT, Lee TW et al. Up-regulation of TWIST in data. Biostatistics 2003; 4:249–64. prostate cancer and its implication as a therapeutic target. Cancer Res 28 Tusher VG, Tibshirani R, Chu G. Significance analysis of microar- 2005; 65:5153–62. rays applied to the ionizing radiation response. Proc Natl Acad Sci 47 Zagzag D, Shiff B, Jallo GI et al. Tenascin-C promotes microvascular USA 2001; 98:5116–21. cell migration and phosphorylation of focal adhesion kinase. Cancer 29 Hosack DA, Dennis G Jr, Sherman BT et al. Identifying biological Res 2002; 62:2660–8. themes within lists of genes with EASE. Genome Biol 2003; 48 Spessotto P, Rossi FM, Degan M et al. Hyaluronan–CD44 interaction 4:R70. hampers migration of osteoclast-like cells by down-regulating 30 Burczynski ME, Twine NC, Dukart G et al. Transcriptional profiles MMP-9. J Cell Biol 2002; 158:1133–44. in peripheral blood mononuclear cells prognostic of clinical out- 49 Podar K, Tai YT, Lin BK et al. Vascular endothelial growth factor- comes in patients with advanced renal cell carcinoma. Clin Cancer induced migration of multiple myeloma cells is associated with Res 2005; 11:1181–9. beta 1 integrin- and phosphatidylinositol 3-kinase-dependent PKC 31 Pietras RJ. Interactions between estrogen and growth factor recep- alpha activation. J Biol Chem 2002; 277:7875–81. tors in human breast cancers and the tumor-associated vasculature. 50 Jawhari AU, Buda A, Jenkins M et al. Fascin, an actin-bundling pro- Breast J 2003; 9:361–73. tein, modulates colonic epithelial cell invasiveness and differenti- 32 Gombos Z, Xu X, Chu CS et al. Peritumoral lymphatic vessel density ation in vitro. Am J Pathol 2003; 162:69–80. and vascular endothelial growth factor C expression in early-stage 51 Martinez A, Vos M, Guedez L et al. The effects of adrenomedullin squamous cell carcinoma of the uterine cervix. Clin Cancer Res 2005; overexpression in breast tumor cells. J Natl Cancer Inst 2002; 11:8364–71. 94:1226–37. 33 Soker S, Takashima S, Miao HQ et al. Neuropilin-1 is expressed by 52 Coll JL, Ben Ze’ev A, Ezzell RM et al. Targeted disruption of vincu- endothelial and tumor cells as an isoform-specific receptor for vas- lin genes in F9 and embryonic stem cells changes cell morphology, cular endothelial growth factor. Cell 1998; 92:735–45. adhesion, and locomotion. Proc Natl Acad Sci USA 1995; 92:9161–5. 34 Takashima S, Kitakaze M, Asakura M et al. Targeting of both mouse 53 Wenstrup RJ, Florer JB, Brunskill EW et al. Type V collagen controls neuropilin-1 and neuropilin-2 genes severely impairs developmen- the initiation of collagen fibril assembly. J Biol Chem 2004; tal yolk sac and embryonic angiogenesis. Proc Natl Acad Sci USA 279:53331–7. 2002; 99:3657–62. 54 Faussillon M, Monnier L, Junien C et al. Frequent overexpression of 35 Kunz M, Moeller S, Koczan D et al. Mechanisms of hypoxic gene cyclin D2/cyclin-dependent kinase 4 in Wilms’ tumor. Cancer Lett regulation of angiogenesis factor Cyr61 in melanoma cells. J Biol 2005; 221:67–75. Chem 2003; 278:45651–60. 55 Capobianco AJ, Zagouras P, Blaumueller CM et al. Neoplastic trans- 36 Grzeszkiewicz TM, Lindner V, Chen N et al. (CYR61, CCN1) sup- formation by truncated alleles of human NOTCH1/TAN1 and ports vascular smooth muscle cell adhesion and stimulates chemo- NOTCH2. Mol Cell Biol 1997; 17:6265–73.

56 Larsson C, Lardelli M, White I et al. The human NOTCH1, 2, and 3 lated with Prx2 in pulmonary vascular disease. Circ Res 2001; genes are located at chromosome positions 9q34, 1p13-p11, and 89:131–8. 19p13.2-p13.1 in regions of neoplasia-associated translocation. Ge- 66 Yang Y, Topol L, Lee H et al. Wnt5a and Wnt5b exhibit distinct ac- nomics 1994; 24:253–8. tivities in coordinating chondrocyte proliferation and differenti- 57 Hsieh JJ, Nofziger DE, Weinmaster G et al. Epstein–Barr virus ation. Development 2003; 130:1003–15. immortalization: Notch2 interacts with CBF1 and blocks differenti- 67 Jonsson M, Andersson T. Repression of Wnt-5a impairs DDR1 ation. J Virol 1997; 71:1938–45. phosphorylation and modifies adhesion and migration of mam- 58 Jundt F, Probsting KS, Anagnostopoulos I et al. Jagged1-induced mary cells. J Cell Sci 2001; 114:2043–53. Notch signaling drives proliferation of multiple myeloma cells. 68 Miller JR, Hocking AM, Brown JD et al. Mechanism and function of Blood 2004; 103:3511–5. signal transduction by the Wnt/beta-catenin and Wnt/Ca2+ path- 59 Touab M, Villena J, Barranco C et al. Versican is differentially ways. Oncogene 1999; 18:7860–72. expressed in human melanoma and may play a role in tumor de- 69 Niida H, Nakanishi M. DNA damage checkpoints in mammals. Mut- velopment. Am J Pathol 2002; 160:549–57. agenesis 2006; 21:3–9. 60 Zhan X, Bates B, Hu XG et al. The human FGF-5 oncogene encodes 70 Vecchione A, Ishii H, Baldassarre G et al. FEZ1/LZTS1 is down- a novel protein related to fibroblast growth factors. Mol Cell Biol regulated in high-grade bladder cancer, and its restoration suppres- 1988; 8:3487–95. ses tumorigenicity in transitional cell carcinoma cells. Am J Pathol 61 Tien ES, Davis JW, Vanden Heuvel JP. Identification of the CREB- 2002; 160:1345–52. binding protein/p300-interacting protein CITED2 as a peroxisome 71 Gilbertson DG, Duff ME, West JW et al. Platelet-derived growth fac- proliferator-activated receptor alpha coregulator. J Biol Chem 2004; tor C (PDGF-C), a novel growth factor that binds to PDGF alpha 279:24053–63. and beta receptor. J Biol Chem 2001; 276:27406–14. 62 Khoo C, Blanchard RK, Sullivan VK et al. Human cysteine-rich 72 Lazar-Molnar E, Hegyesi H, Toth S et al. Autocrine and paracrine intestinal protein: cDNA cloning and expression of recombinant regulation by cytokines and growth factors in melanoma. Cytokine protein and identification in human peripheral blood mononuclear 2000; 12:547–54. cells. Protein Expr Purif 1997; 9:379–87. 73 Halaban R, Cheng E, Smicun Y et al. Deregulated E2F transcrip- 63 Giampuzzi M, Botti G, Cilli M et al. Down-regulation of lysyl oxid- tional activity in autonomously growing melanoma cells. J Exp Med ase-induced tumorigenic transformation in NRK-49F cells charac- 2000; 191:1005–16. terized by constitutive activation of ras proto-oncogene. J Biol Chem 74 Hoek K, Rimm DL, Williams KR et al. Expression profiling reveals 2001; 276:29226–32. novel pathways in the transformation of melanocytes to mela- 64 ten Berge D, Brouwer A, Korving J et al. Prx1 and Prx2 are nomas. Cancer Res 2004; 64:5270–82. upstream regulators of sonic hedgehog and control cell prolifer- 75 Elwood JM. Melanoma and sun exposure. Semin Oncol 1996; ation during mandibular arch morphogenesis. Development 2001; 23:650–66. 128:2929–38. 76 Whiteman DC, Watt P, Purdie DM et al. Melanocytic nevi, solar 65 Jones FS, Meech R, Edelman DB et al. Prx1 controls vascular smooth keratoses, and divergent pathways to cutaneous melanoma. J Natl muscle cell proliferation and tenascin-C expression and is upregu- Cancer Inst 2003; 95:806–12.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp62–71 CLINICAL AND LABORATORY INVESTIGATIONS DOI 10.1111/j.1365-2133.2006.07570.x Melanomas arising from naevi and de novo melanomas — does origin matter? S.C. Weatherhead, M. Haniffa and C.M. Lawrence Department of Dermatology, Royal Victoria Inﬁrmary, Queen Victoria Road, Newcastle upon Tyne, Tyne and Wear NE1 4LP, U.K.

Summary

Correspondence Background It is widely accepted that some melanomas arise from pre-existing Sophie Weatherhead. naevi, while others appear de novo. The proportions involved and the effect of E-mail: [email protected] melanoma origin on prognosis is unclear. Objectives To determine whether melanomas reported by the patient to have devel- Accepted for publication 20 July 2006 oped from a pre-existing naevus are associated with a better or worse prognosis compared with those arising de novo when adjusted for confounding variables. Key words Methods All patients attending a dedicated melanoma screening clinic between melanoma, naevus, prognosis March 1997 and March 2002 were included. The distinction between melanoma arising without any pre-existing lesion (de novo) and those derived from a pre- Conflicts of interest existing lesion (naevus melanoma) was based on patient history. We categorized None declared. patients into three groups: those who gave a history of their lesion arising within a pre-existing naevus, those in whom the melanoma developed de novo and those in whom no conclusive history could be obtained. We compared prognostic indicators between the naevus and de novo melanoma groups. Results Of 8593 patients screened, 377 had a positive diagnosis of melanoma (in situ or invasive). Of these 42% had naevus melanomas, 34% new melanomas and 24% were uncertain. Patients presenting with a melanoma arising from a pre-existing naevus had a greater Breslow thickness despite presenting sooner than the de novo group, although no significant difference in thickness was found when other prognostic factors were controlled for. Conclusions This prospective study shows that naevi that undergo malignant change may result in melanomas that are thicker and thus potentially have a worse prognosis than de novo melanomas. Although our results were not statistically significant when other risk factors were also taken into account, it is possible that a larger study would identify a significant association.

Breslow thickness is the single most important prognostic We prospectively collected 5 years of data from all patients indicator of melanoma prognosis. Histological evidence of a referred to our centre with a possible diagnosis of melanoma. pre-existing naevus is also claimed to be associated with a better All patients were interviewed about the origin of their lesion prognosis.1 However, using histological evidence to show that a before an initial examination and diagnosis was made, thus melanoma has arisen from a pre-existing naevus is inevitably minimizing recall bias. going to identify only very early melanomas as more advanced disease will eventually obliterate any pre-existing Methods mole. An extensive search has identiﬁed only four studies which have used the patients’ history to determine whether or not a Patients were screened in a dedicated melanoma screening cli- melanoma has arisen within a pre-existing mole or de novo, i.e. nic between March 1997 and March 2002.5 At presentation a from apparently normal skin.1–4 These studies found that history was taken before the patients were examined and the 57–85% of melanomas developed within a pre-existing naevus. diagnosis made. Patients were asked whether their presenting However, this question was not the main focus of the studies lesion had arisen from a pre-existing mole, and if so how and details including the timing of the history in relation to long the preceding mole had been present and for how long diagnosis of melanoma were not provided. it had been changing. On the basis of the answers received

2006 The Authors 72 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp72–76 Origin of melanomas, S.C. Weatherhead et al. 73 patients subsequently diagnosed with melanoma were subdiv- thicker in the naevus melanoma group (P <0Æ004) (Fig. 1b). ided into three groups: (i) those whose lesion had developed The other values did not differ significantly between the two within an existing mole (naevus melanoma); (ii) those whose groups. Regression analysis showed a statistically significant lesion had developed where no previous mole existed (de novo association between Breslow thickness and whether there was melanoma); and (iii) those who could not be certain about a preceding naevus or de novo melanoma. However, the associ- the origin of the lesion. The demographic data, clinical history ation was no longer statistically significant when the analysis and examination findings were recorded for all patients. was adjusted for age, sex, past medical history of melanoma, Examination findings included presence of ulceration, descrip- family history of melanoma, diameter of lesion and ulcer- tion of the edge of the lesion, pigment pattern, whether the ation. lesion was palpable, its maximum diameter, and the site of the lesion. Data on the length of time for which the lesion Melanoma site had been changing were also recorded. We categorized melanomas into five categories for depth: in situ, < 1 mm, 1– The location of melanomas was divided into four categories: 1Æ9 mm, 2–3Æ9 mm and > 4 mm. head and neck, trunk, upper limbs and lower limbs. Signifi- cantly more naevus melanomas arose on the trunk (36% vs. 23%; P <0Æ015), whereas more de novo melanomas arose on Statistical analysis the head and neck (11% vs. 28%; P <0Æ001).There was no The naevus melanoma and de novo melanoma groups were significant difference between the number of melanomas compared using the v2 test for the discrete variables (sex, past found in the naevus melanoma and de novo melanoma groups medical history, family history and ulceration), Student’s t-test on the upper and lower limbs (17% vs. 18% for upper limbs for age, and with log-transformed values for maximum and 35% vs. 30% for lower limbs). tumour diameter, and the Mann–Whitney U-test for Breslow thickness. Simple logistic regression analysis was carried out Time course for association with Breslow thickness for each of the above variables. A stepwise multiple regression analysis was subse- In the naevus melanoma group four patients reported that quently performed for Breslow thickness taking into account their melanomas had arisen from a congenital naevus, 116 all of the variables together. patients (74%) reported the naevus had been present for as long as they could remember, and the remaining naevus mela- Results noma patients had a mean lesion duration of 7Æ8 years (Fig. 2a). In 107 out of 128 (84%) de novo melanoma patients Between 1997 and 2002, a total of 8593 patients were who answered this question the mean duration was 2Æ5 years screened and questioned before diagnosis. Invasive melanoma (Fig. 2a). There was no significant difference between the (269, 71%), in situ melanoma (65, 17%) or lentigo maligna reported length of time over which the naevus melanoma and (43, 11%) were histologically detected in 377 patients (222 de novo melanoma groups had noticed the lesion changing; this females, 155 males). Of these, 157 (42%) described their was 13 months in the naevus melanoma group and melanoma as originating within a pre-existing mole, 129 16 months in the de novo group (Fig. 2b). (34%) arose de novo and 91 (24%) patients were unsure and were therefore excluded from further analysis. Comparison of Discussion the first two groups is shown in Table 1. The average age of presentation showed the naevus melanoma group to be sig- This prospective study shows that, based on the patient’s his- nificantly younger than the de novo melanoma group tory, 42% of melanomas arise from a pre-existing naevus, (P <0Æ0012) (Fig. 1a). Breslow thickness was significantly 34% arise de novo and in almost a quarter the historical origin

Table 1 Comparison of melanomas between ‘naevus group’ and de novo group Naevus melanoma de novo melanoma (n ¼ 157) (n ¼ 129) P-value Average age (years) 54 60 0Æ0012 Female : male ratio 1Æ7:1 1Æ4:1 NS Past medical history of melanoma (%) 2 2 NS Family history of melanoma (%) 3 8 NS Maximum tumour diameter (mm) 11 10 NS Ulceration (%) 14 9 NS Median Breslow thickness (mm) 1Æ00Æ70Æ004

NS, not signiﬁcant.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp72–76 74 Origin of melanomas, S.C. Weatherhead et al.

(a) 40 (a) naevus melanoma de novo melanoma As long as 35 I can 30 remember 25 20 500 15 400 300 10 200 Percentage of patients 100 5 100 0 <30 30–39 40–49 50–59 60–69 70–79 80> Age group (years) 75 (b) 50 naevus melanoma de novo melanoma Time (months) 45 40 50 35 30 25 20 25 15 10 Percentage of patients 5 0 0 Naevus melanoma De novo melanoma insitu <1 1–1·9 2–3·9 >4 Breslow thickness (mm) (b) 200

Fig 1. Comparison of patient groups (a) adjusted for age group, (b) 150 adjusted for Breslow thickness. 100 100 cannot be assessed. Melanomas that arise in a pre-existing naevus have a greater Breslow thickness, and hence a potentially worse prognosis compared with melanomas that arise where no previous mole existed. This association, however, was not 75 statistically signiﬁcant in a multiple regression analysis after adjusting for age, sex, family history and past medical history of melanoma. Giant congenital naevi and multiple dysplastic naevi with fa- 6 50 milial melanoma are known risk factors for the progression Time (months) of naevi into melanoma and there have been numerous case series and reports of melanomas arising from previously documented naevi.7–14 The overall transformation rate of a single naevus into a melanoma has been estimated to range from 25 0Æ0005% to 0Æ003%, depending on the age of the patient. The overall lifetime transformation risk for any individual naevus in a 20-year-old has been estimated as 0Æ03%.15 Previous studies have examined the relationship between 0 melanomas and pre-existing naevi by studying clinical, histo- Naevus melanoma De novo melanoma logical or epidemiological evidence. Although it is a widely accepted maxim that many melanomas arise from pre-existing Fig 2. Duration of (a) lesion present before melanoma diagnosis, (b) change before melanoma diagnosis. moles, clinical studies of the frequency of this occurrence based on patient history and the relationship between melanoma origin and other prognostic indicators are poorly docu- Mackie et al. found a slightly poorer prognosis in melanomas mented. There are four clinical studies; Jones et al. reported from naevi that were reported to have developed before the that 57% of melanomas developed within pre-existing naevi,3 age of 2 years, but this was not conﬁrmed when multivariate whereas another smaller study found an incidence of 85%.2 analysis controlled for tumour thickness.4 The fourth study

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp72–76 Origin of melanomas, S.C. Weatherhead et al. 75 compared clinical and histological reports of melanomas are more common in younger patients,17,19 and a higher associated with moles1 but found that only 70% of patients proportion of naevus melanomas arise on the trunk compared with histological evidence of a naevus gave a history of a pre- with the site of de novo melanomas.17 This may be because ceding lesion, and 35% of patients without histological evi- younger patients have more moles, a larger proportion are dence recalled a preceding mole. Overall, three of the four truncal,20 and/or this group are more cosmetically aware of studies concluded that the presence of a preceding mole con- new lesions. ferred a better prognosis when the histological evidence was In summary, our study shows that common acquired naevi assumed, but did not analyse their data based on the clinical that undergo malignant change may result in melanomas that results. These studies are repeatedly quoted to justify the asser- are thicker and thus potentially have a poorer prognosis com- tion that mole-derived melanoma is associated with a good pared with de novo melanomas. Although our results were not prognosis. However, in each of the three studies, patients statistically significant when other risk factors were also taken were questioned retrospectively up to 30 days after the diag- into account, it is possible that a larger study would identify a nosis of melanoma had been made, which may lead to recall significant association. bias. Furthermore, the authors of the former two studies did not attempt to adjust for other confounding factors of progno- References sis. In our study there is an overlap of mole and melanoma duration suggesting that patients may mistake a de novo mela- 1 Friedman RJ, Rigel DS, Kopf AW et al. Favorable prognosis for ma- noma for a benign mole. Patients may not notice a pre-exist- lignant melanomas associated with acquired melanocytic nevi. Arch ing mole even when one was present. We have allowed for Dermatol 1983; 119:455–62. 2 Cameron JRJ. Melanoma of the skin. J R Coll Surg Edinb 1968; this by excluding from our study the quarter of patients who 13:233–54. were unable to give a clear history. Including just those 3 Jones WM, Williams WJ, Roberts MM et al. Malignant melanoma patients from whom a history could be obtained suggests that of the skin: prognostic value of clinical features and the role of almost 55% of melanomas arise from a pre-existing naevus. treatment in 111 cases. Br J Cancer 1968; 22:437–51. Based on histological reports approximately 25% of mela- 4 Mackie RM, Watt D, Doherty V et al. Malignant melanoma occur- nomas arise from pre-existing naevi.16,17 In these studies the ring in those aged under 30 in the west of Scotland 1979–1986: a diagnosis may be blurred by the coincidental coexistence of study of incidence, clinical features, pathological features and survival. Br J Dermatol 1991; 124:560–4. an adjacent naevus and subsequent invasion of the melanoma 5 Weatherhead SC, Lawrence CM. Melanoma screening clinics: are into this. Furthermore, if a melanoma arises in a mole this we detecting more melanomas or reassuring the worried well? will only be identified histologically in early lesions where the Br J Dermatol 2006; 154:539–41. tumour has not yet obliterated a pre-existing mole. It is im- 6 Greene MH, Clark WH Jr, Tucker MA et al. Acquired precursors of portant to acknowledge that when a naevus and melanoma cutaneous malignant melanoma. The familial dysplastic nevus syn- co-exist there is the potential for a spurious Breslow thickness drome. N Engl J Med 1985; 312:91–7. to be recorded. In such situations, the pathologist may have to 7 Benisch B, Peison B, Kannerstein M et al. Malignant melanoma originating from intradermal nevi. A clinicopathologic entity. Arch Der- measure Breslow thickness from the deepest melanocyte, even matol 1980; 116:696–8. though this may be part of a deeper mole and not the mela- 8 Rhodes AR, Sober AJ, Day CL et al. The malignant potential of small noma. Therefore our finding of a thicker Breslow thickness in congenital nevocellular nevi. An estimate of association based on a naevus melanoma cases may be influenced by this. Cohort histologic study of 234 primary cutaneous melanomas. J Am Acad studies whereby individuals are assessed before and after a Dermatol 1982; 6:230–41. melanoma has developed, are potentially the most reliable 9 Borrego L, Hernandez Santana J, Baez O et al. Naevus spilus as a way of answering the question. However these studies are precursor of cutaneous melanoma: report of a case and literature review. Clin Exp Dermatol 1994; 19:515–17. expensive, require observation of several million patients and 10 Okun MR, Bauman L. Malignant melanoma arising from an intra- consequently have not been done. Case–control studies dem- dermal nevus. Arch Dermatol 1965; 92:69–72. onstrate an increased number of new melanomas in patients 11 Little JH. Histology and prognosis in cutaneous malignant mela- with multiple moles,18 but have not demonstrated whether noma. In: Melanoma and Skin Cancer (McCarthy WH, ed.). Sydney: the melanomas that developed had arisen from a naevus or V.C.N. Blight, Govt Printer, 1972; 107–19. de novo. 12 Rhodes AR, Harrist TJ, Day CL et al. Dysplastic melanocytic nevi in Our study differs from previous attempts to identify any histologic association with 234 primary cutaneous melanomas. J Am Acad Dermatol 1983; 9:563–74. prognostic implications from melanomas arising from naevi 13 Schmoeckel C, Bockelbrink A, Bockelbrink H et al. Low- and high-risk by questioning patients before the lesion is examined and malignant melanoma. I. Evaluation of clinical and histological prog- diagnosis made. There may be errors resulting from the indi- nosticators in 585 cases. Eur J Cancer Clin Oncol 1983; 19:227–35. viduals’ ability to recognize or recall a pre-existing mole; 14 Huvos AG, Shah AP, Mike V. Prognostic factors in cutaneous ma- however, in our naevus melanoma group patients appear to lignant melanoma. A comparative study of long term and short have presented sooner following identification that the lesion term survivors. Hum Pathol 1974; 5:347–57. had been changing, yet still have a significantly worse progno- 15 Tsao H, Bevona C, Goggins W et al. The transformation rate of moles (melanocytic nevi) into cutaneous melanoma: a population- sis than those in the new melanoma group. Our results agree based estimate. Arch Dermatol 2003; 139:282–8. with other reports that melanomas arising within naevi

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp72–76 76 Origin of melanomas, S.C. Weatherhead et al.

16 Massi D, Carli P, Franchi A et al. Naevus-associated melanomas: 19 Rivers JK, Kelly MC, Kopf AW et al. Age and malignant melanoma: cause or chance? Melanoma Res 1999; 9:85–91. comparison of variables in different age-groups. J Am Acad Dermatol 17 Bevona C, Goggins W, Quinn T et al. Cutaneous melanomas associ- 1989; 21:717–22. ated with nevi. Arch Dermatol 2003; 139:1620–4 discussion 1624. 20 Rampen FH, van der Meeren HL, Boezeman JB. Frequency of 18 Swerdlow AJ, English J, MacKie RM et al. Benign melanocytic naevi as moles as a key to melanoma incidence? J Am Acad Dermatol 1986; a risk factor for malignant melanoma. Br Med J 1986; 292:1555–9. 15:1200–3.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp72–76 CONTACT DERMATITIS DOI 10.1111/j.1365-2133.2006.07565.x Severity of hand eczema assessed by patients and dermatologist using a photographic guide M. Hald,* N.K. Veien,* G. Laurberg and J.D. Johansen National Allergy Research Centre, Gentofte Hospital, University of Copenhagen, Copenhagen, Denmark *The Dermatology Clinic, Aalborg, Denmark

Summary

Correspondence Background The severity of hand eczema is of interest in epidemiological studies. Marianne Hald. Ideally, as no validated methods of self-assessment exist, a dermatologist should E-mail: [email protected] examine all subjects. However, this is very resource intensive. Objectives To examine if severity grading performed by patients with hand eczema Accepted for publication 7 July 2006 using a self-administered photographic guide was in agreement with the assessment performed by a trained dermatologist. Furthermore, to measure the correl- Key words ation between the severity of hand eczema expressed on a visual analogue scale hand eczema, photographic guide, severity, visual (VAS) and the clinical severity assessment, using the photographic guide. analogue scale Methods Fifty-three consecutive outpatients with hand eczema were included, a Conflicts of interest number based on a prestudy statistical calculation. The patients were asked to None declared. grade current severity of their hand eczema by choosing one of four groups of photographs representing differing severities of hand eczema. On the same day all patients were examined by an experienced dermatologist, who graded the severity using the same photographic guide. The photographic guide was a modified version of a validated guide for use by physicians. In addition, the patients rated the severity of their hand eczema on a VAS. Results Fifty-one of the respondents completed the full questionnaire. For 37 of the 51 patients (73%) the clinical severity assessments of patient and dermatologist were identical. The measure of agreement, Cohen’s kappa coefficient, was 0Æ61, indicating good inter-rater agreement. The correlation between the dermatologist-rated severity and the corresponding score by the patients on the VAS was only moderate (Spearman’s rank correlation coefficient rho ¼ 0Æ52). Conclusions The photographic guide for the self-assessment of hand eczema is an easy instrument to use, and for research purposes can be a reliable tool for patients with hand eczema to grade severity. A VAS can only be considered as a mediocre tool for estimation of the dermatologist-rated clinical severity, but should be validated as an independent instrument to assess severity of hand eczema.

Hand eczema is a common disease. The severity varies from patient can assess both the clinical severity and the impact of mild symptoms to severe illness and can result in decreased hand eczema. There have been several studies assessing the working capacity and impaired quality of life.1,2 The deter- validity of self-reporting hand eczema,3–5 but no method of mination of severity plays a central role in routine manage- severity grading has yet gained general acceptance. ment and in epidemiological studies as well as in clinical The aim of the present study was to validate a self-admin- trials. Epidemiological studies often require large study popu- istered photographic guide with photographs of hand eczema lations. Ideally, a dermatologist should examine all included of varying severity in relation to an assessment by a derma- subjects. However, this is very time consuming and might tologist using the same guide. Furthermore, we wanted to limit the number of participants. An option could be a ques- study the correlation between the severity of hand eczema tionnaire with self-rated assessment of severity. Whereas the expressed on a visual analogue scale (VAS) and the clinical dermatologist is limited to grading the clinical severity, the severity assessment, using the photographic guide.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp77–80 77 78 Severity of hand eczema assessed using a photographic guide, M. Hald et al.

Materials and methods clinical examination the patients were informed about the study and asked to complete the questionnaire. The derma- The study population consisted of 53 consecutive patients, tologists used the same photographic guide and graded the recruited among patients attending the Dermatology Clinic in hand eczema in one of four groups as it appeared at the Aalborg, Denmark. Enrolment took place from November examination on the more severely affected hand. They were 2005 to January 2006 (9 weeks). Patients meeting the follow- also offered the option: ‘There is no hand eczema right now’. ing criteria were enrolled: men or women aged 18 years or The patient and the dermatologist were not informed of each older who attended with hand eczema. All recruited patients other’s assessments. were white. No distinction was made between different subtypes or duration of hand eczema. Statistical methods

Construction of the self-administered photographic guide The concordance of assessment was measured by comparing the ability of the patients to classify severity of their hand In the generation of the self-administered photographic guide eczema in the same group as that determined by the derma- we employed a grading system developed by Coenraads et al., tologist. The agreement was expressed in two different ways: a photographic guide based on a five-point photographic gra- Cohen’s kappa coefficient and Spearman’s rank correlation ding system designed and validated for consistent assessment coefficient rho. Cohen’s kappa measures the proportion of 6 by dermatologists. We modified the original photographic agreement between ratings adjusted for agreement expected by guide by leaving out the group representing ‘clear’, which chance. It has a maximum of 1 when agreement is perfect, resulted in a guide constructed of 16 photographs distributed whereas a value of 0 indicates agreement no better than in four groups of severity of hand eczema: almost clear, mod- chance. To interpret values between 0 and 1 the following erate, severe and very severe. Each group was represented by a guideline should be applied: kappa < 0Æ20 may be taken to spectrum of pictures of male and female hands, dorsal and represent poor agreement, 0Æ21–0Æ40 represents fair agree- palmar views, different skin types and different types of ment, 0Æ41–0Æ60 represents moderate agreement, 0Æ61–0Æ80 chronic hand eczema. The groups were designated 1–4, with represents good agreement and 0Æ81–1Æ00 represents very good 1 being the mildest form. In addition to the photographic agreement.8 The kappa statistic does not take into account the guide, a self-administered questionnaire was composed. degree of disagreement as all deviations are treated equally A pretest of the questionnaire was performed on the basis without consideration of the magnitude of discrepancy. of structured interviews with 11 patients with hand eczema Spearman’s rho is based on a ranked analysis, relevant for attending a dermatology clinic, and was generally well accep- ordinal data of two variables. The correlation coefficient can ted and understood. have any value from )1 to +1, measuring the degree of asso- The patients were asked to grade their hand eczema, by ciation between the two variables. A correlation around 0 in- comparing their hands with the pictures. The following ques- dicates no relation of the variables. A coefficient below 0Æ5 tion was posed: please answer the following questions by represents weak correlation, whereas values above 0Æ9 repre- looking at the photographs: sent strong correlation. 1 Which group does your hand eczema match when your The necessary number of participants was calculated prior hands are most severely affected? to the study, expecting Cohen’s kappa to correspond to a fair 2 On average, which group did your hand eczema match agreement between the assessors and choosing a 5% signifi- over the last 12 months? cance level. On the assumption that most assessments would 3 If you look at your hands right now, which group does your be in groups 1, 2 and 3, the necessary sample size would be hand eczema match? – select the more severely affected hand. 50 subjects to obtain a power of 76%. A more even or skew They were also offered the option: I don’t have hand eczema distribution with few subjects in the two highest (or lowest) right now. groups would give power estimates of 94% or 86%, respect- The inclusion of question 1 was an attempt to offer the ively. The correlation between severity assessment performed patients the opportunity to express how severe their eczema using the photographic guide and the VAS was measured by could be and thereby avoid an unintended overestimation of calculating Spearman’s rho. the present severity. The patients were also asked to evaluate the average severity of their hand eczema on a global score for the last 12 months and Results at present. The instrument was designed as a VAS of 0–10, where A total of 53 patients, 47 women and six men (mean age 7 0 means no eczema and 10 means extremely severe eczema. 38 years, range 18–70) participated in the study. Forty-six of the patients had current hand eczema according to the derma- Collection of data tologist-rated assessment. On average, the patients reported an improvement in hand eczema, when comparing the current All patients were examined by one of two experienced derma- evaluation with the past 12 months, as assessed by the VAS tologists with a special interest in hand eczema. Following the and by using the photographic guide (Table 1).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp77–80 Severity of hand eczema assessed using a photographic guide, M. Hald et al. 79

Table 1 Result of severity ratings of hand eczema by the patients and Table 3 Relationship between clinical severity assessment by the the dermatologists patients and the dermatologists

Missing Identical rating by patients and dermatologists using the Mean ± SD score Range data photographic guide described as: Patient rateda Agreement (95% conﬁdence interval) 72Æ5% (58Æ3–84Æ1%) Æ Æ VAS 4Æ5±2Æ6 0–10 1 Inter-rater agreement (Cohen’s kappa) 0 61 (P <00001) VAS-12 5Æ5±2Æ30Æ5–9Æ52 SEM (Cohen’s kappa) 0Æ09 Æ Æ Photographic guide 1Æ84 ± 0Æ92 0–4 2 Correlation of agreement 0 82 (P <00001) Photographic guide-12 2Æ24 ± 0Æ80 1–4 3 (Spearman’s rho) b Correlation between visual analogue scale and: Dermatologist rated a Photographic guide 1Æ81 ± 0Æ98 0–4 0 Patient-rated current severity 0Æ68 (P <0Æ0001) Patient-rated severity for the last 0Æ45 (P <0Æ001) a aVAS, score on the visual analogue scale for current hand 12 months b eczema; VAS-12, score on the visual analogue scale for hand Dermatologist-rated current severity 0Æ52 (P <0Æ0001)

eczema over the last 12 months; Photographic guide, patient- a rated severity of current hand eczema using the photographic Correlation between patient-rated severity using the photo- guide; Photographic guide-12, patient-rated severity of hand graphic guide and the corresponding score on the visual analogue scale (VAS) of current hand eczema and hand eczema eczema over the last 12 months using the photographic guide. b bPhotographic guide, dermatologist-rated severity of current during the last 12 months, respectively. Correlation between hand eczema using the photographic guide. dermatologist-rated severity of current hand eczema using the photographic guide and the corresponding score on the VAS.

Of 51 patients grading their present hand eczema, identical grades were obtained for 37 (73%). The other 14 patients assessment using the photographic guide. The analysis was assigned a grade adjacent to the group assessed by the derma- carried out for the present time and for the last 12 months. tologist (Table 2). The inter-rater reliability analysed by Spear- The resulting correlation coefficient of 0Æ68 (P <0Æ0001) for man’s rho provided a coefficient of 0Æ82, representing good current hand eczema may be taken as good, whereas the cor- correlation. The inter-rater agreement evaluated by Cohen’s relation decreased to 0Æ45 (P <0Æ001) for the last 12 months. kappa, giving credit only for identical grades, was 0Æ61 and The score on the VAS for current hand eczema showed only represents good reliability (Table 3). In the group of more moderate concordance (Spearman’s rho ¼ 0Æ52, P <0Æ0001) severely affected cases, defined as groups 3 and 4, identical when the reference assessment was the physician-rated clinical grades were obtained for 91%, whereas agreement was seen severity (Table 3). for 67Æ5% of the milder cases defined as groups 0, 1 and 2. The difference was not statistically significant (Fisher’s exact Discussion test, P ¼ 0Æ47). To evaluate the correlation between the patient-rated clinical Self-reported severity of hand eczema has formerly been based severity and their assessment on a global scale, Spearman’s rho on information of frequency of relapses, and more indirect was calculated between the score on the VAS and the patient’s variables such as frequency of visiting a dermatologist and use of topical steroids.9,10 This study indicates that the photo- Table 2 Severity ratings of current hand eczema by the patients and graphic guide for self-assessment of hand eczema is an instru- the dermatologist using the same photographic guidea ment that contributes to a consistent grading and makes it possible to obtain direct information of the clinical severity in Dermatologist-rated severity a group of patients. The original photographic guide was developed as a stan- Patient-rated severity 0 1 2 3 4 Total dardized approach for physicians to assess the severity of hand 03 3eczema. The images were selected by a panel of five experi- 1 294 15 enced dermatologists identifying the most representative pho- 2 5 15 2 22 tographs and achieving a consensus rating of the images. The 3189 422photographic guide achieved a consistent grading by measur- Total 5 14 20 10 2 51 ing the inter- and intra-observer reliability in a session involving 11 dermatologists grading the hand eczema of 28 a The patient and the dermatologist used the same photographic patients.6 Use of the photographic guide requires the ability guide rating the current hand eczema in one of four groups (1– to assess the actual hand eczema by comparison with photo- 4) as it appeared on the more severely affected hand. They were graphs that represent the same stages. Not only the appearance also offered the option ‘There is no hand eczema right now’ (group 0). The table only illustrates assessments performed by of the eczema, but also the location and skin type can only be both the dermatologist and the patient. an approximation and clinical experience will be a factor increasing consistency. In the current study every case was

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp77–80 80 Severity of hand eczema assessed using a photographic guide, M. Hald et al. rated by the patient and one of the two dermatologists. The combination of photographs and the VAS in the same ques- dermatologists were both experienced with a presumed high tionnaire. As the physician-rated severity must be regarded as level of inter-rater reliability. As this study included the the reference assessment, the conclusion must be that the VAS patients’ assessment, we did not expect the same high inter- can only be considered to be a mediocre tool for estimation rater reliability as in the study performed by Coenraads et al.6 of the clinical severity. This does not mean that there is no However, the results indicate that the photographic guide can indication for application of the VAS, but that it should be serve as a reliable instrument for self-rated assessment of seen as an instrument in its own context. severity of hand eczema. In conclusion, the photographic guide for the self-assess- The study population was a random selection of new and ment of hand eczema is an easy instrument to use. For existing patients attending the clinic. A part of the population research purposes it is a reliable tool to grade severity of hand had previously consulted the clinic or alternatively another eczema. dermatologist. It cannot be ruled out that these patients might have had a preconceived opinion of the severity, but hand Acknowledgments eczema is a fluctuating disease and the clinical severity observed at one consultation might have changed at the sec- The authors thank Basilea Pharmaceutica Ltd for providing us ond. with the original photographic guide for this study. Aage There was no statistically significant evidence that the agree- Vølund is gratefully acknowledged for statistical assistance. ment was inferior in groups 0, 1 and 2 compared with groups The study was funded by the National Board of Health, Den- 3 and 4, which could have indicated difficulties of distinction mark. between minor changes. Still, the quality of the photographs plays a central role in the application of the photographic References guide. Minor clinical symptoms can be difficult to visualize on a photograph. To minimize the possibility that pictures repre- 1 Meding B, Wrangsjo K, Jarvholm B. Fifteen-year follow-up of hand senting ‘clear’ were misinterpreted as illustrating minimal eczema: persistence and consequences. Br J Dermatol 2005; eczema that was difficult to see, we decided to apply the con- 152:975–80. 2 Skoet R, Zachariae R, Agner T. Contact dermatitis and quality of cept of ‘no eczema’ by leaving out the pictures representing life: a structured review of the literature. Br J Dermatol 2003; group 0. Still two patients claimed to have current hand 149:452–6. eczema as opposed to the judgement of the dermatologist. 3 Meding B, Barregard L. Validity of self-reports of hand eczema. However, hand eczema presents a continuum and the bound- Contact Dermatitis 2001; 45:99–103. ary between minimal symptoms of active eczema and healed 4 Smit HA, Coenraads PJ, Lavrijsen APM et al. Evaluation of a self- skin may be subtle. administered questionnaire on hand dermatitis. Contact Dermatitis The photographic guide was originally designed and valid- 1992; 26:11–16. 5 Svensson A, Lindberg M, Meding B et al. Self-reported hand ated only in white individuals. Therefore we excluded non- eczema: symptom-based reports do not increase the validity of white patients. For ethnic groups with darkly pigmented skin diagnosis. Br J Dermatol 2002; 147:281–4. it is necessary to do further studies. 6 Coenraads PJ, van der Walle H, Thestrup-Pedersen K et al. Con- The VAS is a subjective instrument, in which many factors struction and validation of a photographic guide for assessing are summarized to a global score. Depending on the many severity of chronic hand dermatitis. Br J Dermatol 2005; 152:296– aspects of the patient’s life situation, the presence of hand 301. eczema may have a different impact on everyday life that is 7 Susitaival P, Flyvholm MA, Meding B et al. Nordic Occupational Skin Questionnaire (NOSQ-2002): a new tool for surveying occu- not necessarily strongly correlated with the clinical severity. pational skin diseases and exposure. Contact Dermatitis 2003; 49:70– Consistency between clinical symptoms and a global self- 6. 11 assessment is, however, to be expected. We found a signifi- 8 Altman DG. Practical Statistics for Medical Research. London: Chapman & cant good correlation between self-rated clinical assessment of Hall, 1991. current hand eczema and the corresponding score on the VAS, 9 Agner T, Flyvholm MA, Menne´ T. Formaldehyde allergy: a follow- a result indicating an interaction between how the patients up study. Am J Contact Dermat 1999; 10:12–17. perceive the impact of hand eczema and their visual percep- 10 Jungbauer FH, van der Vleuten P, Groothoff JW, Coenraads PJ. Irritant hand dermatitis: severity of disease, occupational exposure tion. The correlation decreased, as must be expected, when to skin irritants and preventive measures 5 years after initial diag- the time frame was extended to 12 months. Only a moderate nosis. Contact Dermatitis 2004; 50:245–51. correlation was found between dermatologist-rated severity 11 Holm EA, Wulf HC, Stegmann H et al. Life quality assessment and the score obtained on the VAS. This result is in line with among patients with atopic eczema. Br J Dermatol 2006; 154:719– a former study that described only a poor correlation between 25. the physician-rated clinical severity and the corresponding 12 Cvetkovski RS, Jensen H, Olsen J et al. Relation between patients’ score on the VAS.12 and physicians’ severity assessment of occupational hand eczema. Br J Dermatol 2005; 153:596–600. The closer connection between the self-rated clinical assessment and the score on the VAS might be explained by the

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp77–80 DERMATOLOGICAL SURGERY AND LASERS DOI 10.1111/j.1365-2133.2006.07574.x The use of confocal laser-scanning microscopy in microsurgery for invasive squamous cell carcinoma M. Horn, A. Gerger,* S. Koller, W. Weger, U. Langsenlehner,* P. Krippl,* H. Kerl, H. Samonigg* and J. Smolle Division of Analytical-Morphological Dermatology, Department of Dermatology, Medical University Graz, Auenbruggerplatz 8, A-8036 Graz, Austria *Division of Oncology, Department of Internal Medicine, Medical University Graz, Auenbruggerplatz 15, A-8036 Graz, Austria Institute for Medical Informatics, Statistics and Documentation, Medical University Graz, Auenbruggerplatz 2, A-8036 Graz, Austria

Summary

Correspondence Background Ex-vivo confocal laser-scanning microscopy offers rapid imaging of Armin Gerger. excised tissue specimens without conventional histotechnical procedures. As verti- E-mail: [email protected] cal sections are prepared, morphological features can be assessed according to standard criteria used in conventional histopathology. Accepted for publication 14 July 2006 Objectives To validate the diagnostic confocal examination of squamous cell carcinoma (SCC) in microscopy-guided surgery. Key words Methods Four independent observers received standardized instructions about diag- confocal microscopy, diagnostic validation, nostic confocal microscopy features of SCC. Subsequently, 120 confocal images squamous cell carcinoma, virtual histopathology of fresh excisions from SCC or normal skin, imaged using a commercially avail- Conflicts of interest able, near-infrared, reflectance confocal laser-scanning microscope, were evalu- None declared. ated by each observer. Results General morphology, such as location, size and shape of the cancer area M.H. and A.G. contributed equally to this work. could be visualized by the imaging system. Furthermore, densely packed and irregularly organized nuclei and nuclear atypia could be delineated. Overall, a sensitivity of 95% and a specificity of 96Æ25% were achieved by the four observers (positive predictive value 96Æ25%, negative predictive value 95Æ23%). Conclusions This study provides a set of well-described morphological criteria with obvious diagnostic impact which should be used in further investigations. In the future, confocal laser-scanning microscopy may guide microsurgery of any skin cancer.

Squamous cell carcinoma (SCC) is, apart from basal cell carci- have to be embedded in paraffin and stained with haematoxy- noma (BCC), the most common skin tumour in white-skinned lin and eosin. If margins are found which are positive for individuals. In moderate climate zones an estimated 100 cases tumour cells further excisions have to be performed until the per 100 000 occur in men and 50 cases per 100 000 occur in resected tissue is free of tumour. This is a slow procedure women every year. An increasing incidence of SCC corres- often requiring several days and the defect must be left open ponds with changing lifestyles and numerous environmental in the meantime. factors.1–3 Nonmelanoma skin cancer (NMSC) is not recorded In the Anglo-American areas Mohs micrographic surgery is equally in all countries, and therefore the number of unre- an established procedure that offers an intraoperative analysis corded SCCs is estimated to be high. to detect tumour complexes in the resected tumour mar- SCC occurs in high-risk areas, such as the mid-face or the gins.5,6 In many cases, the Mohs procedure requires several ear.3 A microsurgical excision must be performed with a small excisions in order to remove large complex lesions. The prep- security margin and minimal damage to surrounding healthy aration of frozen and haematoxylin and eosin-stained sections tissue. Whereas precise microsurgery in early stages of tumour typically requires 20–45 min per excision during which the development is almost always curative, imprecise excision puts patient has to wait with an open wound under local anaesthe- the patient at risk for destructive growth and death from dis- sia. This is slow and time-inefficient for Mohs surgeons and ease once the tumour has progressed to competence for meta- requires specialized equipment and trained medical and tech- stasis. nical personnel. In Central Europe, the three-dimensional histological exam- An alternative method for microscopy-guided surgery of ination of all tumour margins according to the guidelines of NMSC was recently published.7–9 Confocal laser-scanning mi- Breuninger has gained acceptance.4 Therefore, all tissue pieces croscopy (CLSM) was evaluated for classification of untreated

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp81–84 81 82 Confocal examination of squamous cell carcinoma, M. Horn et al. fresh specimens from SCC and BCC. To enhance contrast of axial dimension. This device allows visualization of cellular nuclei within the specimens, skin excisions were fixed with structures of the examined tissue. A field of view of approxi- 5% acetic acid to increase light back-scatter and make the mately 0Æ5 · 0Æ5 mm at a tissue level is provided by the nuclei bright and easily detectable. General morphology, such microscope. Adjacent fields can be imaged in sequence and as epidermis, dermis and location, shape and size of the can- then tiled together to produce a composite image of an area cer area could be clearly delineated by the method. Individual of tissue of approximately 2Æ5 · 2Æ5 mm. A glass slide con- nuclei were accurately visualized, allowing distinction between taining the tissue piece was fixed with the help of an adapter cancerous and normal cells; furthermore, distribution of indi- ring on to the objective lens. Overview images (maps) using vidual tumour cells within the tumour masses could be recog- the composite image procedure were captured in a two- nized. The corresponding histopathological analyses compared dimensional matrix. These maps were used to locate suspected well with the CLSM images. The procedure turned out to be cancerous sites. Within the map individual images could be fast, simple and inexpensive. While the diagnostic perform- selected and the system resumed imaging in that area of the ance of CLSM for evaluation of fresh specimens from BCC was tissue. At least 20 images comprising epidermal and dermal recently validated, possibly diagnostic SCC features have to areas and tumour tissue when present were recorded in each date only been described but not statistically assessed. specimen. All images were stored using BMP file format. To our knowledge, this study is the first to provide a systematic validation of confocal microscopy in diagnosing Diagnostic morphological features of squamous cell SCC in untreated and fresh skin excisions for application to carcinoma microscopy-guided surgery. Morphological CLSM features of SCC were assessed according 10 Materials and methods to standard criteria used in conventional histopathology. Tumour cell nuclei with variation in size and shape, irregular masses of epidermal cells and cell masses poorly demarcated Surgical procedures from surrounding tissue were taken into account for further Twenty patients with histologically verified SCC were enrolled analysis. in the study after informed consent was given. All patients were recruited prospectively from the Dermatological Surgery Training data and study setting Unit of the Department of Dermatology, Medical University of Graz, Austria. Only tumours with a clinically determined hori- Four independent observers (two residents and two dermato- zontal diameter exceeding 10 mm were included in the study. pathologists) all with experience in CLSM regarding the ex-vivo All surgical procedures were performed by the same surgeon. examination of BCC received short instructions about morpho- All institutional rules governing clinical investigation of logical CLSM features of SCC as an oral presentation. Further- human subjects were strictly followed. After local anaesthesia, more, 10 CLSM image examples of SCC including the a 3-mm punch biopsy was removed from the centre of the described features were demonstrated. All observers possessed tumour. Subsequently, the whole SCC was excised according knowledge about CLSM features of normal skin. For diagnostic to dermatosurgical guidelines and submitted to standard histo- assessment, three confocal SCC and three normal skin images logical examination. A primary closure of the wound was pos- were selected from each of the 20 patients. Each observer was sible in 14 cases, while in six cases a transpositional flap was faced with 120 CLSM images overall. Images were shown on performed. Moreover, normal-appearing skin from the site of the computer screen in random sequence using a macro pro- ‘dog-ears’ or from the contralateral ‘Burow’s triangle’ was col- cedure. Each observer evaluated each image as tumour tissue lected in each patient. Thus, the study did not interfere with or normal skin. Statistical analyses [sensitivity, specificity, the routine surgical and histological procedure and patient positive predictive value (PPV), negative predictive value care. Both fresh tissue pieces, one containing definitively (NPV)] were performed on the datasets with a personal com- tumour and the other definitively tumour-free tissue were pre- puter using the SPSS statistical software package for Windows pared immediately after the surgical procedure for CLSM version 12.0 (SPSS Inc., Chicago, IL, U.S.A.). All of the experi- evaluation: approximately 0Æ5 mm thick vertical sections were menters were blinded as to the clinical and histopathological cut and washed with 5% acetic acid for 30 s. diagnosis of the confocal images.

Confocal laser-scanning microscopy Results

CLSM was performed with a commercially available confocal Qualitative description of morphological confocal laser- microscope (Vivascope 1000; Lucid Inc., Rochester, NY, scanning microscopy features U.S.A.) using a diode laser with 830 nm wavelength and power < 35 mW at tissue level. Due to the low power of the General morphology, such as location, size and shape of the diode laser no tissue damage occurs. It images with a spatial cancer area could be visualized by the imaging system. Den- resolution of 0Æ5–1Æ0 lm in the lateral and 3–5 lm in the sely packed and irregularly organized nuclei were recognized

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp81–84 Confocal examination of squamous cell carcinoma, M. Horn et al. 83

(a)

Fig 1. Distribution of individual tumour cells: nuclei are densely (b) packed and irregularly organized.

(c)

Fig 2. Nuclear atypia can be delineated in confocal images. Acetic acid-induced compaction of chromatin makes nuclei bright.

(Fig. 1). Further details, such as nuclear atypia, could be delineated with the help of the field-of-view scan (Fig. 2). Moreover, normal skin structures as previously published could be clearly identified9 (Fig. 3). Overall, CLSM morphology compared well with the features seen in conventional histopathology. Within the study set six histologically verified SCCs in situ were found. In general, SCC in situ tumours were Fig 3. Normal skin architecture (a) and structures such as collagen not easily detected because of the bright pictured epidermis. fibres (b) and hairs (c) compare well with the morphological features Therefore, nuclear atypia was difficult to assess accurately; found in conventional histopathological images. however, cell distribution was a useful sign for classification.

sitivity values of 91Æ67% and 90% and specificity values of Sensitivity and specificity 95% and 96Æ67% (PPV 94Æ83% and 96Æ43%, NPV 91Æ94% and Best performance was achieved by the younger resident with- 90Æ63%), respectively. Overall, a sensitivity of 95% and a spe- out dermatopathological qualification and the senior dermato- cificity of 96Æ25% was achieved by the four observers (PPV pathologist, with sensitivity values of 98Æ33% and 100% and 96Æ25%, NPV 95Æ23%). The SCCs in situ were classified cor- specificity values of 100% and 93Æ33% (PPV 100% and rectly in 98Æ33%, 100%, 96Æ66% and 98Æ33% of cases (in 93Æ75%, NPV 98Æ36% and 100%), respectively, followed by order of the observers’ appearance) when statistically evaluated the senior resident and younger dermatopathologist, with sen- (overall performance 98Æ33%).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp81–84 84 Confocal examination of squamous cell carcinoma, M. Horn et al.

Discussion In several cases, there were difficulties with image quality, especially using the mosaic procedure. Mosaics often appeared In this study we demonstrate the application of CLSM to diag- bright in the borders of the examined tissue. Technical prob- nostic examination of untreated fresh specimens from SCC in lems could be resolved in all cases but were time consuming. microscopy-guided surgery. Acetic acid-washed sections of Finally, in this study each case was represented by preselected tumours and normal skin were imaged using CLSM and evalu- images. It might be possible that the evaluation of a larger ated in an observer-blinded manner. The analysis was sterile number of images per case taken from various areas of and artificial in that no clinical or histological diagnosis was the tumour might not add to the diagnostic accuracy, but taken into account. Vertical sections similar to those used for might, on the contrary, distract the observers from the correct conventional histology were prepared. Consequently, morpho- diagnosis. logical features could be assessed according to well-known cri- To evaluate the feasibility of the method further, a large- teria used in conventional histopathology. scale study with a higher number of cases and observers, jud- The morphological CLSM features are simple to learn and ging the whole set of images obtained in each case, would be use and this is reflected in the excellent and stable diagnostic helpful. At this point, CLSM ex vivo cannot replace standard his- performance of all observers. Moreover, confocal and histo- topathological procedures. Despite the described limitations of pathological morphology seems to correspond well, as con- the method, CLSM remains an exciting technology for rapid ventional microscopy features of SCC and normal skin were imaging and deserves further investigation. In the future the applied to confocal image examination. In contrast to the ob- method may be used as a prescreening tool for excised speci- servations of Chung et al.,8 SCCs in situ were correctly classified mes. The present study, however, provides a set of well- with an exceptional accuracy by all observers. This might be described morphological criteria with obvious diagnostic possible because each case was represented by preselected impact which should be used in future investigations. images, and this remains a bias in our study. The main advantage of CLSM is the unique opportunity to References image thin sections of fresh surgical excisions without processing the tissue as in standard histology. Tissue preparation 1 Miller DL, Weinstock MA. Nonmelanoma skin cancer in the United for confocal examination is simple and time-saving. Prepar- States: incidence. J Am Acad Dermatol 2000; 42:988–91. ation of a fresh excision takes approximately 1 min including 2 Diepgen TL, Mahler V. The epidemiology of skin cancer. Br J Der- matol 2002; 146:1–6. cutting a 0Æ5-mm thick section and washing with 5% acetic 3 Gallagher RP, Hill GB, Bajdik CD et al. Sunlight exposure, pigmen- acid for 30 s. Fixing the adapter ring on a glass slide and pla- tation factors, and risk of nonmelanocytic skin cancer. II. Squa- cing the objective lens on to the ring can be done in 1 min. mous cell carcinoma. Arch Dermatol 1995; 131:164–9. With the composite image function of the system, five to 10 4 Breuninger H. Histologic control of excised tissue edges in the maps including the entire section in most cases can be created operative treatment of basal-cell carcinomas. J Dermatol Surg Oncol in approximately 3 min. Overall, approximately 5 min is 1984; 10:724–8. required to prepare for confocal examination of surgical exci- 5 Mohs FE. Chemosurgery – a microscopically controlled method of cancer excision. Arch Surg 1941; 42:279–95. sions. Resolution of the confocal microscope is nearly equal to 6 Mikhail GR, ed. Mohs Micrographic Surgery. Philadelphia: W.B. Saun- that of conventional microscopes used to view histology ders, 1991. slides, and cellular and architectural details can be examined. 7 Rajadhyaksha M, Menaker G, Flotte T et al. Confocal examination Confocal image maps are rapidly made by the imaging system of nonmelanoma cancers in thick skin excisions to potentially to get an overview in large lesions and to detect suspected guide Mohs micrographic surgery without frozen histopathology. cancerous areas, followed by inspection of the cellular mor- J Invest Dermatol 2001; 117:1137–43. phology in the suspected regions. This is similar to the pro- 8 Chung VQ, Dwyer PJ, Nehal KS et al. Use of ex vivo confocal scanning laser microscopy during Mohs surgery for nonmelanoma skin cedure for examining histopathological sections. cancers. Dermatol Surg 2004; 30:1470–8. This particular study also shows several limitations. The 9 Gerger A, Horn M, Koller S et al. Confocal examination of untreated number of lesions is too small to guarantee statistical signifi- fresh specimens from basal cell carcinoma: implications for micro- cance. Furthermore, one cannot conclude from the results scopically guided surgery. Arch Dermatol 2005; 141:1269–74. of four independent observers that similar classification 10 Lever WF, Schaumburg-Lever G (eds). Histopathology of the Skin, 7th results would be achieved by the majority of dermatologists. edn. Philadelphia: Lippincott, 1990.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp81–84 DERMATOPATHOLOGY DOI 10.1111/j.1365-2133.2006.07575.x Effect of smoking on skin elastic ﬁbres: morphometric and immunohistochemical analysis M. Just, M. Ribera,* E. Monso´, J.C. Lorenzo and C. Ferra´ndiz* Department of Dermatology, Hospital Figueres, Figueres, Girona, Spain Departments of *Dermatology, Pneumology and Pathology, Hospital Universitari Germans Trias i Pujol, Badalona, Barcelona, Spain

Summary

Correspondence Background It has been established during recent years that smoking is an inde- Miquel Just. pendent risk factor for the development of premature facial wrinkling. The E-mail: [email protected] underlying mechanism is not well known, but elastic fibres of the dermis seem to be the major target of smoke components. Accepted for publication 14 July 2006 Objectives To determine quantitative and qualitative changes of the dermal elastic tissue of nonsun-exposed skin induced by smoking, as well as the possible mech- Key words anisms responsible for them. dermis, elastic fibres, immunohistochemistry, Methods Sixty-nine patients were recruited (20 nonsmokers, 19 former smokers morphometry, smoking and 30 smokers). Using static morphometry and immunohistochemistry and lec- Conflicts of interest tin staining we analysed elastic fibres of the dermis and their major components, None declared. elastin and microfibrillar component. ) Results Significantly higher values for the number of elastic fibres mm 2 and the percentage of the area filled by them in the reticular dermis were found in smokers. Cumulative tobacco dose showed statistically significant correlations with both morphological parameters (Spearman’s rank correlation coefficient). Immu- nohistochemistry demonstrated that the two main components of elastic fibres were altered in smokers. Plasma protease inhibitors and lectin staining were negative in all the samples. Conclusions Smoking is an independent risk factor for the increase of elastic fibres in the reticular dermis of nonexposed skin, and it acts on their two main structural components, elastin and microfibrillar component. This increase in the area of elastic fibres in smokers is not due to newly synthesized elastic material, but to their degradation, as occurs in solar elastosis and which acts in an additive manner.

Tobacco smoking is one of the major health problems in total area occupied by dermal elastic fibres in smokers com- Western countries and is causally related to many chronic and pared with nonsmokers. A more recent study, however, shows malignant diseases. Human skin is exposed to smoke directly contradictory findings which question the effects of smoking through its irritant components on the epidermis and indir- on elastic fibres of the skin.18 The hypothesis that elastin ectly on the dermis via the bloodstream. It is not surprising could also be the main target in smoke-induced wrinkles, as therefore that smoking has manifold effects on the skin and is occurs in emphysema, has recently been supported by the fact associated with significant morbidity of this organ.1,2 In the that lung function is related to the morphological characteris- past three decades, several studies have shown that smoking is tics of elastic fibres in the reticular dermis, and that both char- an independent risk factor for the development of premature acteristics depend on the amount of tobacco consumed.19 facial wrinkling and skin ageing.3–12 The goal of this study was to investigate quantitative and In order to establish the pathophysiological mechanisms qualitative changes of the dermal elastic tissue of nonsun- underlying this premature wrinkling, and based on the finding exposed skin induced by smoking, as well as the possible that elastin is the major target in smoke-induced emphys- mechanisms responsible for them. Morphometric analyses and ema,13,14 several reports have used morphometry to analyse immunohistochemical techniques were used to assess dermal the elastic fibres of the skin in smokers and nonsmokers.15–17 elastic fibres in a well-defined cohort of smokers and non- The results showed an increase in the number, thickness and smokers.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 85 86 Effect of smoking on skin elastic ﬁbres, M. Just et al.

Patients and methods

Subjects

Sixty-nine consecutive male outpatients were recruited in a general hospital by the dermatology and pneumology units: 20 nonsmokers, 19 former smokers and 30 current smokers. Only men were included in the study to avoid sex-related confounding factors. Those patients reporting a1-antitrypsin deficiency, connective tissue disease, body mass index )2 >40kgm , or ever treated with D-penicillamine, colchicine or methotrexate or artificial ultraviolet radiation, were not enrolled. Because skin ageing progresses slowly after the third decade and accelerates only in old age,20,21 only patients aged Fig 1. Binary image showing elastic fibres of the dermis as white. 40–70 years participated in the study to preserve the homogeneity of the sample. The study was approved by the Hospi- tal’s Ethics Committee and written informed consent was Immunohistochemical and lectin staining obtained from all participating subjects. Subjects were categorized as nonsmokers, former or cur- Tissue sections of 3 lm thickness from the punch biopsies rent smokers. Subjects who had never smoked daily and were stained by the immunoperoxidase method using antielas- had smoked fewer than 150 cigarettes during their lifetime tin monoclonal antibody (Sigma, St Louis, MO, U.S.A.) diluted were considered nonsmokers, daily smokers who had quit 1/50, and antisera to amyloid P component (APC) (Dakopatts, > 6 months before as former smokers, and subjects who Glostrup, Denmark) diluted 1/100, a1-antitrypsin (Dakopatts) had been smoking at least one cigarette per day during the diluted 1/100, a1-antichymotrypsin (Dakopatts) diluted 1/ 6 months prior to the examination as current smokers. Life- 600 and a2-macroglobulin (Biosystem, Barcelona, Spain) dilu- time cumulative tobacco dose was expressed as pack-years ted 1/20. Sections were also stained by the lectins concanav- (average number of packs smoked per day multiplied by alin A and Triticum vulgaris (Sigma) diluted 1/20. Histological years smoked). Pipe or cigar smokers were excluded. evaluations were made in a blinded fashion. We assessed the Patients who had used oral corticosteroids for > 3 months intensity of the staining using a visual scale ranging from 1 to in their lifetime were considered long-term oral corticoster- 2 (1, weakly positive; 2, strongly positive). oid users.

Statistics Morphometric analysis All data were introduced in a database and analysed using the Punch biopsies 4 mm in diameter were obtained from non- SPSS statistical software package version 11.5 (SPSS, Chicago, sun-exposed skin of the upper inner part of the arm, as in IL, U.S.A.). Results for categorical variables were expressed as previous studies.15,16,18 Specimens were fixed in formalin and absolute and relative frequencies and results for continuous embedded in paraffin. Sections of 3 lm thickness were stained variables as mean and SD, or as median and interquartile range with orcein and examined under an optical microscope when the distribution was not normal. (· 40) (Carl Zeiss, Jena, Germany) in a blinded fashion. Five Firstly, the relationship between skin morphometry, long- fields of papillary and reticular dermis were randomly selected term oral corticosteroid use and smoking habit was assessed from those that did not include large nonconnective tissue ele- by univariate analysis (Student’s t-test) to determine if long- ments (i.e. hair follicles and sebaceous glands) and the image term corticosteroid therapy and/or quitting smoking were as- of every field was transferred to a monitoring screen using a sociated with changes in the morphology of skin elastic fibres. semiautomatic video camera (Kontron Bildanalyse, Eching, Secondly, skin parameters in nonsmokers and smokers were Germany). From every section, an image of 256 grey levels compared (Student’s t-test), and the correlation between skin was obtained and converted to a binary image showing elastic morphology and cumulative tobacco dosage was examined to fibres as white (Fig. 1). This image was processed (Microm determine which skin parameters were affected by lifetime Image Processing; Microm, Barcelona, Spain) and the number amount of tobacco consumed (Spearman’s rank correlation of elastic fibres in every field and the area filled by them were coefficient, rs). ) determined, with the results expressed as elastic fibres mm 2 Finally, in order to analyse if changes in elastic fibres in the and the percentage of the field containing elastic fibres. reticular dermis related to smoking habit are due to changes Morphometric values obtained in the five examined fields of in the microfibrillar and/or in the amorphous component, im- papillary and reticular dermis were averaged to determine the munochemistry was performed and elastin and APC content values for the elastic fibre content for each subject. determined using a semiquantitative scale with two categories

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 Effect of smoking on skin elastic ﬁbres, M. Just et al. 87

(weak/strong), and the morphometry of both categories was smokers when compared with nonsmokers (Table 3). Cumula- compared (Student’s t-test). tive tobacco dose showed statistically significant correlations All statistical tests were two sided, and P £ 0Æ05 was repor- with the morphology of elastic fibres in the reticular dermis, )2 ted as statistically significant. both with their number mm (rs ¼ 0Æ26, Spearman’s rank correlation coefficient) and with the percentage of the area Results filled by them in the histological field (rs ¼ 0Æ46, Spearman’s rank correlation coefficient). Smoking, however, was not significantly correlated with the morphology of the elastic fibres Characteristics of the study subjects in the papillary dermis (Table 4). The characteristics of our study sample are summarized in Table 1. Participating subjects were white males, aged Immunohistochemical analysis between 43 and 70 years (mean ± SD 58Æ35 ± 7Æ2). The mean ± SD age of the smokers was 58 ± 5 years, of former When the relationship between immunostaining with elastin smokers 58 ± 7Æ9 years and of nonsmokers 59Æ1 ± 8 years. and APC and the morphology of skin elastic fibres in the reticu- These differences were not significant (P >0Æ1, one-way ANO- lar dermis was examined, subjects with higher elastin and APC VA). Cumulative tobacco dosage was over 30 pack-years for content were found to have higher values for area filled by most of those participating. The morphological characteristics elastic fibres (mean ± SD 15Æ2±4Æ4 and 16Æ3±3Æ5, respect- of dermal elastic fibres were not significantly correlated with ively) than the subgroup with lower elastin and APC content age in the studied sample (P >0Æ1, Pearson’s correlation coef- (mean ± SD 13Æ0±3Æ0 and 13Æ2±3Æ6, respectively) (P ¼ ficient), confirming the homogeneity of the sample according 0Æ037 and P ¼ 0Æ006, respectively, Student’s t-test). Moreover, to this parameter. Differences in the measured morphological subjects with higher APC content were also found to have a parameters according to long-term oral corticosteroid use higher number of elastic fibres (mean ± SD 745 ± 231) than were not statistically significant. Similarly, nonstatistically sig- the subgroup with lower APC content (mean ± SD 611 ± 160; nificant differences in these parameters were found between P ¼ 0Æ020, Student’s t-test; Table 5, Figs 4, 5). current and former smokers, and these subgroups were con- Immunostaining for the plasma protease inhibitors (a1- sidered as a whole for the following analyses (Table 2). antitrypsin, a1-antichymotrypsin and a2-macroglobulin) and staining for lectins (concanavalin A and T. vulgaris) were negative in all the samples. Morphometric analysis

The elastic fibres of the reticular dermis were more numerous Discussion and fragmented in smokers than in nonsmokers (Figs 2, 3). Significantly higher values in the number of elastic Our data support that smoking is an independent risk factor ) fibres mm 2 (P ¼ 0Æ041) and the percentage of the area filled for the increase of elastic fibres in the reticular dermis of non- by them (P <0Æ001) in the reticular dermis were found in exposed skin, but not in the papillary dermis. This association is in agreement with earlier studies,15,16 but not with results reported recently by Knuutinen et al.18 Our study sample is ex- Table 1 Population characteristics (n ¼ 69) tensive, confounding variables are controlled, and histological evaluations were made in a blinded fashion. Therefore, the Sociodemography dose-related effect of smoking on the morphology of the re- Age (years; mean ± SD) 58Æ35 ± 7Æ2 ticular dermis shown in this paper is in agreement with the Gender (male) [n (%)] 69 (100) Inhaled corticosteroids [n (%)] 17 (25) dose-related effect of smoking on facial wrinkling and on the 5,22,23 Long-term oral corticosteroids [n (%)] 15 (22) amount of skin elastosis reported previously, and lends Smoking support to the fact that smoking is a risk factor for the devel- Never [n (%)] 20 (29) opment of facial wrinkling. Former [n (%)] 30 (43) Smoking-induced changes in skin elastic fibres occur deeper Current [n (%)] 19 (28) in the dermis and are less intense than those reported as asso- Pack-years [median (interquartile range)]a 32 (0–60) ciated with sun exposure, and must be considered as being Morphometry of skin elastic fibres Reticular dermis related to cigarette smoke components that reach the reticular ) 24 No. fibres mm 2 (mean ± SD) 634Æ8 ± 204Æ1 dermis through the bloodstream. So, it is not surprising that Area (%; mean ± SD) 14Æ3±3Æ7 smoking by itself is unable to change the macroscopic appear- Papillary dermis ance of the sun-protected skin of smokers.25,26 However, )2 No. fibres mm (mean ± SD) 2451Æ9 ± 816Æ5 when exposed skin was analysed, a significant increase of the Area (%; mean ± SD) 5Æ0±1Æ6 elastic tissue was found in both the papillary and the reticular 17 aExpressed as median (interquartile range) due to non-normal dermis of smokers compared with nonsmokers. This allows distribution. us to explain, at least in part, the increase in facial wrinkling of smokers, which depends on the cumulative dose of tobacco

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 88 Effect of smoking on skin elastic ﬁbres, M. Just et al.

Table 2. Skin morphometry according to oral corticosteroid use, former and current smoking

No (n ¼ 54) Yes (n ¼ 15)

Long-term oral corticosteroids Mean (SD) 95% CI Mean (SD) 95% CI P-valuea Morphometry of skin elastic fibres Reticular dermis Area (%) 13.8 (3.0) 13.0–14.6 15.9 (5.3) 13.0–18.9 0.15 ) No. fibres mm 2 632 (218) 572–692 644 (145) 564–725 > 0.25 Papillary dermis Area (%) 5.0 (1.4) 4.6–5.4 5.2 (2.0) 4.1–6.3 > 0.25 ) No. fibres mm 2 2458 (863) 2115–2701 2432 (658) 2067–2796 > 0.25

Former (n ¼ 30) Current (n ¼ 19)

Smoking habitb Mean (SD) 95% CI Mean (SD) 95% CI P-valuea Morphometry of skin elastic fibres Reticular dermis Area (%) 14.9 (3.6) 13.5–16.26 16.0 (3.9) 14.2–17.9 > 0.25 ) No. fibres mm 2 638 (201) 563–714 710 (201) 614–808 > 0.20 Papillary dermis Area (%) 5.1 (1.8) 4.4–5.8 5.2 (1.4) 4.5–5.9 > 0.25 ) No. fibres mm 2 2503 (709) 2228–2778 2286 (602) 1987–2585 > 0.25

CI, conﬁdence interval. aStudent’s t-test; bonly in smokers (n ¼ 49).

consumed.5,22,24 This relationship is probably modulated by a genetic predisposition, because only one quarter of heavy smokers show abnormally severe wrinkling in their exposed skin.5 Ultrastructurally, elastic fibres from adult human skin consist of two main components: a central amorphous core of elastin surrounded by a peripheral mantle of tubular microfibrils, which are composed of APC in addition to other glycopro- teins.27–29 Our data show a significant relationship between the increase in the area occupied by elastic fibres in the reticular dermis in smokers and the increase in elastin and APC immunostaining, as well as a significant relationship between the increase in the number of these elastic fibres and the increase Fig 2. Elastic fibres of the reticular dermis in nonsmokers. in APC immunostaining. These findings suggest that smoke acts on the two main structural components of elastic fibres in the dermis, although it seems to be more pronounced on the microfibrillar component than on the amorphous material. These results are in line with an earlier study by Frances et al.,15 who found differences in immunostaining of elastin and microfibrils in smokers compared with nonsmokers, as well as that microfibrillar dense zones are enlarged two- to fourfold in smokers when examined under electron microscopy. The implications of these findings need to be determined. The underlying mechanisms of the morphological changes of the elastic fibres in the reticular dermis in smokers remain to be determined. The increase in number seems to be due to fibre fragmentation, as reported previously.15,16 The increase in the area occupied by elastic fibres in the reticular dermis could result from the synthesis of new normal or abnormal Fig 3. Elastic fibres of the reticular dermis in smokers. elastic tissue, from degradation of elastic fibres, from a

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 Effect of smoking on skin elastic ﬁbres, M. Just et al. 89

Table 3 Skin morphometry according to smoking habit

Nonsmokers (n ¼ 20) Former/current smokers (n ¼ 49)

Mean (SD) 95% CI Mean (SD) 95% CI P-valuea Reticular dermis Area (%) 11Æ7(2Æ1) 10Æ7–12Æ715Æ3(3Æ7) 14Æ3–16Æ4<0Æ001 ) No. ﬁbres mm 2 557 (191) 468–646 667 (203) 608–725 0Æ041 Papillary dermis Area (%) 4Æ7(1Æ4) 4Æ1–5Æ45Æ2(1Æ6) 4Æ7–5Æ7>0Æ20 ) No. ﬁbres mm 2 2529 (1098) 2015–3043 2418 (671) 2219–2617 > 0Æ20

CI, conﬁdence interval. aStudent’s t-test.

Table 4 Skin morphometry according to cumulative tobacco dose tors as well as the negative lectin reactivity in all the samples analysed suggest the hypothesis that the increase in the area of Cumulative tobacco dose elastic fibres in the reticular dermis in smokers is not due to (pack-years) newly synthesized elastic material, but to their degradation, as occurs in solar elastosis.31 ra P-value s Over recent years substantial progress has been made in Morphometry of skin elastic fibres understanding molecular mechanisms that bring about skin Reticular dermis ageing induced by smoking and sun exposure. Matrix metallo- Area (%) 0Æ46 < 0Æ001 ) No. fibres mm 2 0Æ26 0Æ035 proteinases (MMPs) are zinc-dependent proteases that degrade Papillary dermis dermal collagen and other extracellular matrix molecules, in- Area (%) )0Æ04 > 0Æ25 cluding elastin. The proteolytic activity of MMPs is inhibited ) No. fibres mm 2 )0Æ10 > 0Æ25 by tissue inhibitors of metalloproteinases (TIMPs). The expression of MMP-1 is increased in a dose-dependent manner in aSpearman’s rank correlation coefficient. the skin of smokers compared with nonsmokers, whereas the TIMP remains unchanged or even decreased, suggesting that decrease in dermal thickness or from external substance depos- smoking-induced MMP-1 might be important in the skin-age- ition. No difference in dermal thickness between smokers and ing effects of tobacco smoking.33–36 On the other hand, MMPs nonsmokers or external substance deposition in dermal elastic can also be induced by ultraviolet radiation, and maximum fibres of smokers was found in previous reports.15,18,30 On induction was observed when both were applied together, the other hand, plasma protease inhibitors (a1-antitrypsin, indicating that the two factors act in an additive manner.34,37 a1-antichymotrypsin and a2-macroglobulin) seem to play a The fact that tobacco smoke seems to have phototoxic proper- protective role during the formation of the elastic fibres, by ties supports this hypothesis.38 This emerging information attenuation of extracellular proteolytic enzymes. In the same reveals that skin ageing induced by smoking and photoageing way, lectin staining is usually associated with the presence of could share a fundamental molecular pathway. immature elastic fibres. Therefore, the presence of both lectins In conclusion, this study supports that smoking is an inde- and plama protease inhibitors in association with elastic fibres pendent risk factor for the increase of elastic fibres of the re- suggests their relatively recent formation.31,32 In the present ticular dermis. This increase is due to elastic tissue study the lack of immunostaining for plasma protease inhibi- degradation rather than new synthesis, as occurs in solar elas-

Table 5 Morphometry of skin elastic ﬁbres in the reticular dermis according to elastin and amyloid P component (APC) immunostaining

Weak Strong

Mean (SD) 95% CI Mean (SD) 95% CI P-valuea Elastin immunostaining Area (%) 13Æ0(3Æ0) 12Æ0–14Æ115Æ2(4Æ4) 13Æ4–17Æ00Æ037 ) No. ﬁbres mm 2 648 (189) 579–718 620 (210) 534–707 > 0Æ25 APC immunostaining Area (%) 13Æ2(3Æ6) 11Æ9–14Æ516Æ3(3Æ5) 14Æ5–18Æ10Æ006 ) No. ﬁbres mm 2 611 (160) 554–668 745 (231) 626–864 0Æ020

CI, conﬁdence interval. aStudent’s t-test.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 90 Effect of smoking on skin elastic ﬁbres, M. Just et al.

References

1 Just M, Ribera M. Efectos del consumo de tabaco sobre la piel. Piel 2000; 15:176–81. 2 Wolf R, Lo Schiavo A, Ruocco V. Smoking out the skin. J Appl 20 000 Cosmetol 1995; 13:1–14. 3 Daniell HW. Smoker’s wrinkles: a study in the epidemiology of ‘crow’s feet’. Ann Intern Med 1971; 75:873–80. 4 Model D. Smoker’s face: an underrated clinical sign? Br Med J 1985; 291:1760–2. 15 000 5 Kadunce DP, Burr R, Gress R et al. Cigarette smoking: risk factor for premature facial wrinkling. Ann Intern Med 1991; 114:840–4. 6 Schnohr P, Lange P, Nyboe J et al. Does smoking increase the degree of wrinkles on the face? The Copenhagen City Heart Study. Area of dermal elastic fibres (%) Ugeskr Laeger 1991; 153:660–2. 10 000 7 Joffe I. Cigarette smoking and facial wrinkling. Ann Intern Med 1991; 115:659–60. 8 Ernster VL, Grady D, Miike R et al. Facial wrinkling in men and women, by smoking status. Am J Public Health 1995; 85:78–82. 1 2 9 Koh JS, Kang H, Choi SW et al. Cigarette smoking associated with Elastin immunostaining premature facial wrinkling: image analysis of facial skin replicas. Int J Dermatol 2002; 41:21–7. Fig 4. Scatter plot and Student’s t-test of the percentage of the area of 10 Raitio A, Kontinen J, Rasi M et al. Comparison of clinical and com- the reticular dermis filled by elastic fibres and elastin immunostaining puterized image analyses in the assessment of skin ageing in smok- intensity in the histological field. ers and non-smokers. Acta Derm Venereol (Stockh) 2004; 84:422–7. 11 Lo´pez Hernańdez B, Tecedor J, Ro´denas JM et al. Envejecimiento cutańeo y tabaquismo. Rev Clin Esp 1995; 195:147–9. 12 Leung WC, Harvey I. Is skin ageing in the elderly caused by sun exposure or smoking? Br J Dermatol 2002; 147:1187–91. 13 Wright JL, Churg A. Smoke-induced emphysema in guinea pigs is associated with morphometric evidence of collagen breakdown and repair. Am J Physiol 1995; 268:17–20. 20000 14 Vlahovic G, Russell ML, Mercer RR, Crapo JD. Cellular and connective tissue changes in alveolar septal walls in emphysema. Am J Respir Crit Care Med 1999; 160:2086–92. 15 Frances C, Boisnic S, Hartmann DJ et al. Changes in the elastic tissue of the non-sun-exposed skin of cigarette smokers. Br J Dermatol 15000 1991; 125:43–7. 16 Lister RK, Barnes R, Khorshid M et al. Structural changes in nonsun-exposed skin of smokers: histological assessment of dermal

Area of dermal elastic fibres (%) thickness and elastin. Br J Dermatol 1999; 141 (Suppl. 55): 24 (Abstract). 10000 17 Boyd AS, Stasko T, King LE et al. Cigarette smoking-associated elas- totic changes in the skin. J Am Acad Dermatol 1999; 41:23–6. 18 Knuutinen A, Kallioinen M, Va¨ha¨kangas K, Oikarinen A. Smoking 1 2 and skin: a study of the physical qualities and histology of skin in Amyloid P component immunostaining smokers and non-smokers. Acta Derm Venereol (Stockh) 2002; 82:36– 40. Fig 5. Scatter plot and Student’s t-test of the percentage of the area of 19 Just M, Monso´ E, Ribera M et al. Relationships between lung func- the reticular dermis filled by elastic fibres and amyloid P component tion, smoking and morphology of dermal elastic fibres. Exp Dermatol immunostaining intensity in the histological field. 2005; 14:744–51. 20 Braveman IM, Fonferko E. Studies in cutaneous aging I: the elastic fibers network. J Invest Dermatol 1982; 78:434–43. tosis, and is correlated with changes in the microfibrillar com- 21 Hornebeck W, Robert L. Elastase-like enzymes in aorta and human ponent more than in the amorphous material. Furthermore, breast carcinomas: quantitative variations with age and pathology. sunlight and smoking have multiplicative effects on facial age- Adv Exp Med Biol 1979; 79:145–64. ing and could share, at least in part, a pathophysiological 22 Chung JH, Lee SH, Youn CS et al. Cutaneous photodamage in Kore- pathway. ans. Influence of sex, sun exposure, smoking, and skin color. Arch Dermatol 2001; 137:1043–51. 23 Kennedy C, Bastiaens MT, Bajdik CD et al. Effect of smoking and Acknowledgments sun on the aging skin. J Invest Dermatol 2003; 120:548–54. 24 Frances C. Smoking and the skin. Int J Dermatol 1992; 31:779–80. This work was supported in part by Fondo de Investigaciones 25 Allen HB, Johnson BL, Diamond SM. Smoker’s wrinkles? JAMA Sanitarias ISCiii-RTIC-03/11. 1973; 225:1067–9.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 Effect of smoking on skin elastic ﬁbres, M. Just et al. 91

26 O’Hare PM, Fleischer AB, D’Agostino RB et al. Tobacco smoking 32 Davies JD, Mera SL. Elastosis in breast carcinoma: II. Association of contributes little to facial wrinkling. J Eur Acad Dermatol Venereol 1999; protease inhibitors with immature elastic fibres. J Pathol 1987; 12:133–9. 153:317–24. 27 Uitto J. Biochemistry of the elastic fibers in normal connective tis- 33 Lahmann C, Bergemann J, Harrison G, Young AR. Matrix metallo- sues and its alterations in diseases. J Invest Dermatol 1979; 72:1–10. proteinase-1 and skin ageing in smokers. Lancet 2001; 357:935–6. 28 Breathnach SM, Melrose SM, Bhogal B et al. Amyloid P component 34 Yin L, Morita A, Tsuji T. Skin aging induced by ultraviolet expos- is located on elastic fibre microfibrils in normal human tissue. Nat- ure and tobacco smoking: evidence from epidemiological and mo- ure 1981; 293:652–4. lecular studies. Photodermatol Photoimmunol Photomed 2001; 17:178–83. 29 Dahlba¨ck K, Ljungquist A, Lo¨fberg H et al. Fibrillin immunoreactive 35 Knuutinen A, Kokkonen N, Risteli J et al. Smoking affects collagen fibers constitute a unique network in the human dermis: immu- synthesis and extracellular matrix turnover in human skin. Br J Der- nohistochemical comparison of the distributions of fibrillin, vitro- matol 2002; 146:588–94. nectin, amyloid P component, and orcein stainable structures in 36 Yin L, Morita A, Tsuji T. Alterations of extracellular matrix induced normal skin and elastosis. J Invest Dermatol 1990; 94:284–91. by tobacco smoke extract. Arch Dermatol Res 2000; 292:188–94. 30 Whitmore SE, Sago NJ. Caliper-measured skin thickness is similar 37 Fisher GJ, Datta SC, Talwar HS et al. Molecular basis of sun-induced in white and black women. J Am Acad Dermatol 2000; 42:76–9. premature skin ageing and retinoid antagonism. Nature 1996; 31 Mera SL, Lovell CR, Russel R, Davies JD. Elastic fibres in normal 379:335–9. and sun-damaged skin: an immunohistochemical study. Br J Derma- 38 Placzek M, Kerkmann U, Bell S et al. Tobacco smoke is phototoxic. tol 1987; 117:21–7. Br J Dermatol 2004; 150:991–3.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp85–91 DERMATOPATHOLOGY DOI 10.1111/j.1365-2133.2006.07603.x Characterization of the expression and activation of the epidermal growth factor receptor in squamous cell carcinoma of the skin G.B. Fogarty, N.M. Conus,* J. Chu and G. McArthur Division of Radiation Oncology, *Translational Research Laboratory, Division of Research and Division of Haematology and Medical Oncology, Peter MacCallum Cancer Centre, Melbourne, Vic., Australia Department of Medicine, St Vincent’s Hospital, University of Melbourne, Melbourne, Vic., Australia

Summary

Correspondence Background Locally advanced skin cancers including squamous cell carcinoma (SCC) Gerald Fogarty, Radiation Oncology Associates, PO of the skin are increasing in incidence. Patients are often elderly with significant Box 1003, Crows Nest 1585, NSW, Australia. comorbidities and therapy can be difficult. New targeted therapies, such as treat- E-mail: [email protected] ment directed at the epidermal growth factor receptor (EGFR), may be effective Accepted for publication and less toxic in these patients. However, before designing appropriate clinical 9 June 2006 trials it is necessary to characterize the expression and activation of targets such as the EGFR to evaluate the rationale of using EGFR inhibitors (EGFRIs) in the Key words treatment of this type of cancer. epidermal growth factor receptor, skin cancer, Objectives To characterize the expression and activation by phosphorylation of squamous cell carcinoma, Western blotting EGFR in SCC of the skin by quantitative Western blotting using the LiCor immu- Conflicts of interest nofluorescence detection system with validation by immunohistochemistry. Sec- None declared. ondary objectives were to evaluate downstream targets of EGFR expression and activation in SCC of the skin and to examine the associations between EGFR, pathological features and clinical behaviour of these tumours. Methods Twenty-one mainly locally advanced skin SCCs collected in our institution and stored in our tissue bank over a 4-year period were used for the study. Results Nine of 21 (43%) tumours expressed EGFR above background. Of those nine, five expressed phosphorylated EGFR. There was no correlation with downstream activation of canonical signalling pathways, pathological features or clinical behaviour. Conclusions EGFR is expressed in a minority of tumours and then is not always activated. These results show that, before designing a trial with a targeted agent such as an EGFRI in SCC of the skin, it is important to verify the presence of the appropriate target to maximize the best outcome.

Skin cancers are common and increasing in incidence in Austra- that have the target, these agents give fewer side-effects in tis- lia and other Western countries of lower latitudes. In Australia sues that lack the target, that is, in normal cells. in 1995 there were 270 000 persons having one or more non- Epidermal growth factor (EGF) receptor (EGFR) is a 170- melanoma skin cancers removed and this increased to 374 000 kDa transmembrane glycoprotein that plays a pivotal role in in 2002.1 One significant component of this case load is squa- normal epithelial biology.2 It is expressed in human epidermis mous cell carcinoma (SCC) of the skin, and many patients especially in the basal layers and in epidermal appendages,3 referred to tertiary referral centres have more advanced disease. and expression has been described in the proliferating areas of These patients can be elderly with significant comorbidities such skin tumours.4 EGFR can be overexpressed in primary and as immunosuppression, diabetes and other malignancies. Treat- perhaps even more so in metastatic deposits of SCC of the ments such as aggressive surgery with postoperative radical skin.5 In many epithelial tumours high expression of EGFR radiotherapy can be associated with significant morbidity. and/or its ligands is associated with more aggressive disease, Therefore, targeted therapies are worthy of evaluation in this poor prognosis and resistance to hormonal therapy, cytotoxic patient population because for the same effect in tumour cells drugs and radiotherapy.6–11

2006 The Authors 92 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 EGFR in skin SCC, G.B. Fogarty et al. 93

Mitogen activation of EGFR induces phosphorylation of radiotherapy within 6 months prior to operation were exclu- EGFR and downstream activation of signalling pathways in- ded. Tumour specimens were stored in liquid nitrogen until cluding the phosphatidyl inositol 3-phosphate kinase/AKT and the time of analysis. mitogen-activated protein kinase (MAPK) pathways.12 The level of EGFR phosphorylation rather than EGFR expression Quantitative Western blotting alone in tumours has been shown to be associated with decreased time to progression in nonsmall-cell lung cancer.13 Approximately 100 mg of frozen tumour sample was pow- Histopathological and molecular studies in the skin of can- dered on dry ice, lysed for 20 min on ice in 500 lL lysis buf- ) ) cer patients receiving the EGFR inhibitor (EGFRI) gefitinib fer [10 mmol L 1 Tris-HCl (pH 7Æ5), 100 mmol L 1 NaCl, ) revealed significant inhibition of EGFR phosphorylation and 2 mmol L 1 ethylenediamine tetraacetic acid, 0Æ5% deoxycho- )1 )1 downstream signalling in EGFR-expressing cells. Reduced kera- late, 1% Triton X-100, 0Æ1 mmol L NaVO3, 50 mmol L ) tinocyte proliferation with increased maturation markers, skin NaF, 1 mmol L 1 phenylmethylsulphonyl fluoride and one thinning and apoptosis were also observed.14,15 These effects protease inhibitor cocktail tablet (Roche, Basel, Switzerland) were observed at all dose levels, before reaching dose-limiting per 10 mL lysis buffer], and homogenized in a Dounce ho- toxicities. Similarly, Bonner et al.,16 in a phase III trial using mogenizer. Cellular debris was removed by 2 · 10 min cen- the monoclonal antibody to EGFR, cetuximab, in combination trifugation at 4 C. Protein concentration was determined in with radiation in head and neck mucosal SCC, showed clinical the supernatants using a Lowry protein assay kit (BioRad, Her- benefit but also skin toxicity outside of the treatment fields. In cules, CA, U.S.A.) and aliquots of 50 lg protein were separ- vitro studies have shown that small molecule inhibitors of ated in reducing denaturing conditions by standard sodium EGFR in cell lines derived from SCC of skin inhibit EGFR dodecyl sulphate–polyacrylamide gel electrophoresis on 10% phosphorylation and its downstream effects including inva- gels. A431 cell lysates using 10 lg of protein stimulated or siveness.17 Given the apparent oversensitivity of normal skin not with EGF were run on each gel as internal controls. The to EGFRIs compared with other epithelial tissues, we hypo- separated proteins were then transferred on to nitrocellulose thesized that SCC of the skin and other malignancies of skin membranes. The membranes were then sequentially blocked origin may express higher levels of EGFR and therefore be for 1 h in LiCor blocking buffer and incubated with primary more sensitive to EGFRIs than other malignancies. Also, given and secondary antibodies with 4 · 5-min phosphate-buffered that EGFR has been implicated in radioresistance,18 patients saline washes in between. Primary antibodies used were at with unresectable SCC of the skin may benefit from a combin- 1 : 1000 dilution and included rabbit polyclonal antiphosphor- ation of an EGFRI and radiation. ylated EGFR (phos-EGFR) (Tyr1068; #2234), mouse What is known about EGFR and its expression and phos- polyclonal anti-EGFR (Tyr1F4; #2239), anti-AKT (#9272), phorylation in skin cancer is limited and is often inferred anti-p44/42 MAPK (#9102), anti-C-myc (#9402), mouse from other cancers. The primary aim of this study was to polyclonal antiphos-AKT (Ser473; 587F11; #4051), anti- characterize the expression and activation by phosphorylation phos-p44/42 MAPK (Thr202/Tyr204; E10; #9106) and an- of EGFR in SCC of skin to assess if this cancer is a realistic tar- ticyclin D1 (DCS6; #2926) (all antibodies were from Cell get for EGFRI treatment. Quantitative Western blotting (WB) Signaling Technology, Danvers, MA, U.S.A.). with the LiCor immunofluorescence detection system with val- Secondary antibodies used at 1 : 5000 dilution were Alexa idation by immunohistochemical (IHC) analysis was used. The Fluor 680 goat antirabbit IgG (Molecular Probes, Eugene, OR, levels and activation of downstream targets of EGFR activation U.S.A.) and IRDye 800 antimouse IgG (Rockland Immuno- in SCC of the skin were also evaluated by quantitative WB. chemicals, Gilbertsville, PA, U.S.A.). Membranes were scanned These include AKT, phosphorylated AKT (phos-AKT), MAPK, and bands were quantified with the Odyssey infrared Imaging phosphorylated MARK (phos-MAPK), c-Myc and cyclin D1, all System LiCor. This system allows accurate quantification of of which have been shown to be important in proliferation proteins on WB across a wide linear range and probing the and tumorigenesis. Another aim was retrospectively to investi- same membrane with different antibodies raised in different gate the association (if any) between EGFR expression/activa- species (e.g. mouse and rabbit) at the same time, hence allow- tion, pathological features and clinical behaviour. ing determination and quantification of total protein vs. the phosphorylated form of the same protein. Background levels Materials and methods were between 0Æ5 and 1Æ2, and values above 1 (with 0Æ5 background) or 2 (with 1Æ2 background) were considered positive. Positive bands could also be visualized by eye. The values in- Selection of tumours dicate levels of fluorescence detected; they depend on the lev- From 1999 to 2004, patients with locally advanced histologic- els of protein present as well as on the specificity of the ally proven skin SCC gave consent for postoperative storage of antibody used. They are not absolute levels but allow quantita- their tumours in the Peter MacCallum Cancer Centre tissue tive comparison between tumours. Positivity quantified using bank and for the use of their tissue in research as per our LiCor was confirmed by enhanced chemiluminescence and institutional ethics committee guidelines. Patients receiving also by using different antibodies for EGFR: rabbit polyclon- systemic cytotoxic chemotherapy, systemic EGFRI or local al against total EGFR (1005; clone H11, Dako, Glostrup,

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 94 EGFR in skin SCC, G.B. Fogarty et al.

Denmark; Sc-03; Santa Cruz Biotechnology, Santa Cruz, CA, Table 1 Epidermal growth factor receptor (EGFR) expression and U.S.A.) and mouse polyclonal antiphos-EGFR (Tyr1068; activation (phosphorylation) from highest to lowest expressing 1H12; #2236, Cell Signaling Technology). Samples were run samples with corresponding clinical information three times to conﬁrm results. EGFR Phos-EGFR Tumour or LiCor LiCor Primary or node or Immunohistochemical analysis Tumour value value secondary dermal deposit

IHC analysis was performed to ensure that the EGFR and 1 233Æ523Æ2 Secondary Node 236Æ16Æ0 Secondary Dermal D phos-EGFR found by LiCor was within tumour cells and not 316Æ52Æ4 Primary T3 within stroma. IHC analysis was performed using the same 48Æ52Æ2 Secondary Node antibodies used in the WB using a standard method. Follow- 54Æ62Æ1 Secondary Node ing a deparafﬁnization and rehydration step, blocking with 62Æ40Æ9 Primary T2 3% (v/v) hydrogen peroxide in distilled water and antigen 72Æ41Æ0 Primary T2 retrieval in citrate buffer, slides were probed with the anti- 82Æ30Æ5 Secondary Dermal D bodies to EGFR (1005; Dako clone H11; Sc-03) and phos- 92Æ21Æ9 Secondary Node 10 1Æ50Æ7 Secondary Node EGFR (Tyr1068; 1H12; Cell Signaling antibody #2236) at the 11 1Æ40Æ6 Primary T4 concentration recommended. Incubation with biotinylated sec- 12 1Æ30Æ6 Primary T4 ondary antibody, streptavidin peroxidase and diaminobenzi- 13 1Æ30Æ5 Primary T1 dine substrate was done using the LSAB2+ kit and protocol 14 1Æ20Æ4 Secondary Node (Dako). All cases were stained simultaneously. Slides were 15 1Æ11Æ1 Primary T2 analysed qualitatively in a blinded fashion. Staining was scored 16 0Æ91Æ2 Primary T3 as 0, none; 1, weak; 2, moderate; 3, strong. 17 0Æ90Æ4 Secondary Node 18 0Æ60Æ1 Primary RT 19 0Æ5 0 Secondary Node Pathological features and clinical behaviour 20 0Æ50Æ6 Primary T2 21 0Æ50Æ5 Secondary Node We investigated whether EGFR expression correlated with clinical behaviour. The clinical presentation and postoperative For EGFR and phosphorylated EGFR (phos-EGFR) a value > 2 is histopathological description were reviewed by a clinician considered positive. Node, tumour within lymph nodes; Dermal D, dermal deposit; RT, recurrent tumour. blinded to the results of the laboratory analyses. Tumour stage,19 site, size, differentiation and presence of extracapsular disease were recorded and then compared with EGFR expres- correlation between the relative levels of total EGFR and phos- sion. The clinical outcome of the patients from whom the EGFR (Table 1). The LiCor values also correlated with visual tumours were harvested was also retrospectively assessed. This inspection of the blots, as shown in representative Western was done from medical records by a clinician blinded to the blots for both EGFR (Fig. 1a) and phos-EGFR (Fig. 1b). Pro- results of the laboratory analyses. Age, sex, duration of follow bing for actin showed that similar levels of proteins (50 lg) up and disease status at follow up were recorded. were loaded in each lane (Fig. 1a).

Results Immunohistochemical analysis IHC analysis to localize EGFR and phos-EGFR within tumour Western blotting analysis of epidermal growth factor cells was performed using the same antibodies used in the receptor (EGFR) and phosphorylated EGFR WB. IHC analysis for total EGFR confirmed that EGFR was pre- To quantify the expression and activation of the EGFR in sent within the tumour cells that were EGFR positive on WB. human SCC of the skin we utilized quantitative WB using Li- The intensity of IHC staining correlated with the LiCor value Cor infrared immunofluorescence technology. This technology for all the tumours investigated (Table 2). Representative IHC offers significant advantages over conventional chemilumines- staining for EGFR is shown in Figure 2a. IHC analysis was less cence techniques, having a linear relationship between input sensitive than LiCor WB in the case of phos-EGFR determin- and signal over a much wider range. Nine of 21 (43%) ation. Phos-EGFR IHC staining was positive in only one sam- tumours expressed EGFR above background. Of these nine, ple. This sample is shown in Figure 2b. five (56%) demonstrated phosphorylation of the EGFR above background. The 12 tumours that did not express EGFR did Western blotting/LiCor quantification of potential not have detectable phosphorylation of the EGFR. Table 1 downstream proteins of epidermal growth factor receptor shows EGFR and corresponding phos-EGFR immunofluores- signalling cence LiCor values from the strongest EGFR-expressing tumour to the weakest. For those five tumours expressing both EGFR There was no correlation between the expression of EGFR and and phos-EGFR above background, there was a strong positive the downstream proteins MAPK, phos-MAPK, c-Myc and

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 EGFR in skin SCC, G.B. Fogarty et al. 95

(a)

50 µm

(b)

Fig 1. (a) Representative Western blots of epidermal growth factor (EGF) receptor (EGFR) showing LiCor values from various skin squamous cell carcinomas numbered as in Table 1. A LiCor reading > 2 is considered positive and corresponds to visual detection. Positive control A431 cells have been treated with EGF. Actin blots show equal loadings of 50 lg protein lysates. (b) Western blots of phosphorylated EGFR (phos-EGFR) for the same tumours as in (a).

Table 2 Correlation of amount of epidermal growth factor receptor 50 µm (EGFR) as found on Western blotting by LiCor analysis with EGFR as found by visualization of immunohistochemical (IHC) analysis Fig 2. (a) Representative immunohistochemical (IHC) staining for epidermal growth factor receptor (EGFR) of a skin squamous cell Tumour LiCor value IHC value carcinoma (SCC) expressing EGFR (tumour 2 in Table 1). The LiCor 1 233Æ53value for EGFR in this tumour is 36Æ1. The IHC grade of this sample 236Æ13is 3, that is, a high intensity of staining. (b) IHC staining for 316Æ51phosphorylated EGFR (phos-EGFR) in the same tumour, the only skin 48Æ52SCC expressing phos-EGFR in IHC out of the 21 analysed. The LiCor 54Æ61value for phos-EGFR in this tumour is 6Æ0. Original magniﬁcation: 62Æ41(a), (b) · 40. 72Æ41 Æ 8231cyclin D (Table 3). AKT/phos-AKT WB values are not repor- 92Æ21 1 ted as they were not higher than background levels in all 10 1Æ50 11 1Æ41tumours despite showing strong expression in our control 12 1Æ30A431cells. Representative Western blots are shown in Figure 3 13 1Æ30(a–d). These results suggest that in SCC of the skin signalling 14 1Æ20pathways downstream of the EGFR can be modulated by other 15 1Æ11mechanisms. 16 0Æ90 17 0Æ90 18 0Æ60Clinical behaviour and pathological description 19 0Æ51 20 0Æ50No correlation was found between EGFR expression and clin- 21 0Æ50ical presentation and behaviour including age, sex, duration of follow up and disease status at follow up. There was no cor- IHC staining was scored as 0, none; 1, weak; 2, moderate; 3, relation at a median follow up postsurgery of 10Æ4 months. strong. IHC staining correlated with the LiCor value. LiCor is EGFR expression also did not vary with postoperative histo- more quantitative than IHC. For EGFR a LiCor value > 2 is considered positive. pathological descriptors including tumour stage, site, size, differentiation and presence of extracapsular disease.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 96 EGFR in skin SCC, G.B. Fogarty et al.

Table 3 LiCor ﬂuorescence values for epidermal growth factor receptor (EGFR), phosphorylated EGFR (phos-EGFR) and downstream proteins in locally advanced skin squamous cell carcinomas, arranged by tumours with highest to lowest EGFR expression

Tumour EGFR Phos-EGFR MAPK Phos-MAPK Cyclin D1 c-Myc 1 233Æ523Æ28Æ70Æ53Æ66Æ6 236Æ16Æ05Æ50Æ34Æ65Æ4 316Æ52Æ421Æ30Æ526Æ63Æ2 48Æ52Æ24Æ30Æ82Æ63Æ7 54Æ62Æ16Æ00Æ82Æ12Æ7 62Æ40Æ915Æ32Æ37Æ22Æ4 72Æ41Æ016Æ51Æ28Æ83Æ0 82Æ30Æ513Æ70Æ83Æ41Æ9 92Æ21Æ92Æ90Æ30Æ612 10 1Æ50Æ77Æ10Æ52Æ83Æ8 11 1Æ40Æ65Æ10Æ54Æ53Æ3 12 1Æ30Æ64Æ50Æ42Æ310 13 1Æ30Æ53Æ60Æ41Æ61Æ5 14 1Æ20Æ413Æ80Æ523Æ73Æ5 15 1Æ11Æ114Æ81Æ67Æ87Æ5 16 0Æ91Æ211Æ20Æ93Æ02Æ5 17 0Æ90Æ45Æ20Æ41Æ61Æ3 18 0Æ60Æ17Æ40Æ44Æ22Æ8 19 0Æ50 6Æ00Æ63Æ13Æ0 20 0Æ50Æ67Æ30Æ96Æ53Æ3 21 0Æ50Æ56Æ01Æ11Æ35Æ8

MAPK, mitogen-activated protein kinase; phos-MAPK, phosphorylated MAPK. For EGFR and phos-EGFR a value > 2 is considered positive.

Discussion

Our LiCor WB results show that EGFR and the corresponding Fig 3. Representative Western blots of (a) mitogen-activated protein phos-EGFR are respectively expressed in approximately 40% kinase (MAPK); (b) phosphorylated MAPK (phos-MAPK); (c) cyclin D ; (d) c-Myc from various skin squamous cell carcinomas as (nine of 21) and 25% (five of 21) of this small series of 1 numbered in Table 1. mainly locally advanced skin SCCs. For those five tumours that expressed both total EGFR and phos-EGFR above background, there was a correlation between the relative levels of total median of 10Æ4 months of follow up. This suggests that other EGFR and phos-EGFR. Tumours that did not express EGFR pathways are likely to be involved in the modulation of these above background did not express phos-EGFR above back- downstream proteins.20–22 ground. IHC analysis confirmed that the EGFR was indeed of To our knowledge, this is the first study looking at expres- tumour origin. The IHC intensity of total EGFR staining sion levels of EGFR and phos-EGFR in SCC skin tumour sam- correlated with the LiCor WB values. IHC analysis was less ples, using a quantitative WB technique. Most studies sensitive than WB in the case of phos-EGFR determination. examining EGFR expression in SCC skin tumour samples have Phos-EGFR could only be detected by IHC in one of the five been performed by IHC techniques. There is a lack of consist- tumours that had LiCor values above background for phos- ency in these results. Kikuchi et al.4 showed by IHC analysis EGFR. that EGFR was expressed in all of six skin SCCs examined and As phosphorylation of the EGFR is normally associated with was seen especially in the invasive parts of the tumour. Shi- activation of the kinase and downstream signalling pathways mizu et al.5 showed by IHC analysis that in four of five cases we examined the expression and activation of molecules previ- of metastatic skin SCC, EGFR expression was weakly positive ously associated with activation of the EGFR. However, our in the primary, and strongly positive in the corresponding study has shown that there was no correlation between EGFR lymph node lesions, suggesting that metastatic potential was expression/phosphorylation and the downstream targets AKT, linked to EGFR expression. EGFR expression was not detected 20 phos-AKT, MAPK, phos-MAPK, c-Myc and cyclin D1. Simi- in normal skin in their experiments. Nazmi et al. investigated larly, there was no correlation between EGFR expression/acti- by IHC analysis 66 specimens from skin that included normal, vation, pathological description and clinical behaviour at a benign, premalignant and malignant specimens, including five

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 EGFR in skin SCC, G.B. Fogarty et al. 97 skin SCCs. Interestingly, they found that the more malignant http://www.nhmrc.gov.au/publications/_files/cp87.pdf (cited the phenotype, the more significant was the absence of EGFR January 2005). expression. Lavrijsen et al.,21 using an IHC method on frozen 2 Jost M, Kari C, Rodeck U. The EGF receptor – an essential regulator of multiple epidermal functions. Eur J Dermatol 2000; 10:505–10. sections of punch biopsies, found that EGFR was expressed in 3 Nanney LB, Magid M, Stoscheck CM, King LE Jr. Comparison of all of 10 invasive skin SCCs examined, with tumour differenti- epidermal growth factor binding and receptor distribution in nor- ation making no difference to the intensity of the expression. mal human epidermis and epidermal appendages. J Invest Dermatol Because different antibodies and methods have been used and 1984; 83: 385–93. because IHC analysis is more a qualitative rather than a quanti- 4 Kikuchi A, Amagai M, Hayakawa K et al. Association of EGF recep- tative method, results reported are not always consistent. tor expression with proliferating cells and of ras p21 expression Techniques other than IHC analysis have been used. Ikeuchi with differentiating cells in various skin tumours. Br J Dermatol 1990; 123: 49–58. et al.,22 using light microscopic autoradiography with 125 5 Shimizu T, Izumi H, Oga A et al. Epidermal growth factor receptor [ I]EGF showed only one of 10 skin SCCs to have increased overexpression and genetic aberrations in metastatic squamous-cell 23 expression of EGFR compared with normal skin. Krahn et al. carcinoma of the skin. Dermatology 2001; 202: 203–6. in 2001 found by conventional and differential and quantita- 6 Mendelsohn J, Baselga J. Status of epidermal growth factor receptor tive reverse transcriptase–polymerase chain reaction that EGFR antagonists in the biology and treatment of cancer. J Clin Oncol was expressed in four of five skin SCCs, with two expressing 2003; 21: 2787–99. EGFR strongly. They also found that six of 16 samples of nor- 7 Umekita Y, Ohi Y, Sagara Y, Yoshida H. Co-expression of epidermal growth factor receptor and transforming growth factor-alpha mal skin expressed EGFR, of which four did so strongly. In predicts worse prognosis in breast-cancer patients. Int J Cancer our study, we have used a combination of traditional IHC an- 2000; 89: 484–7. alysis and quantitative WB using the LiCor fluorescence system 8 Grandis JR, Melhem MF, Gooding WE et al. Levels of TGF-alpha to measure both the total and activated levels of EGFR. The and EGFR protein in head and neck squamous cell carcinoma and superior correlation between the levels of total EGFR and patient survival. J Natl Cancer Inst 1998; 90: 824–32. phos-EGFR found with this new LiCor system (Table 1) com- 9 Fontanini G, De Laurentiis M, Vignati S et al. Evaluation of epider- pared with what was found on IHC analysis in the same mal growth factor-related growth factors and receptors and of neo- angiogenesis in completely resected stage I–IIIA non-small-cell tumour cell population (Table 2 and the finding of phos-EGFR lung cancer: amphiregulin and microvessel count are independent in one tumour only with IHC analysis alone) suggests that the prognostic indicators of survival. Clin Cancer Res 1998; 4: 241–9. LiCor system is superior. 10 Uhlman DL, Nguyen P, Manivel JC et al. Epidermal growth factor Targeted therapies only work in cancer if the tumour cells receptor and transforming growth factor alpha expression in papil- express the target and the target is activated and implicated in lary and nonpapillary renal cell carcinoma: correlation with meta- tumour growth. This is seen in the HER-2 trastuzumab story. static behavior and prognosis. Clin Cancer Res 1995; 1: 913–20. Overexpression of HER-2 in breast cancer is associated with 11 Tateishi M, Ishida T, Mitsudomi T et al. Immunohistochemical evidence of autocrine growth factors in adenocarcinoma of the aggressive disease and decreased overall survival time. Trast- human lung. Cancer Res 1990; 50: 7077–80. uzumab, a monoclonal antibody, targets an extracellular 12 Solomon B, Hagekyriakou J, Trivett MK et al. EGFR blockade with 24 domain of the HER-2 receptor. The combination of chemo- ZD1839 (‘Iressa’) potentiates the antitumor effects of single and therapy and trastuzumab resulted in significantly higher over- multiple fractions of ionizing radiation in human A431 squamous all response rates with a longer median time to disease cell carcinoma. Epidermal growth factor receptor. Int J Radiat Oncol progression and overall survival time than with chemotherapy Biol Phys 2003; 55: 713–23. alone25,26 in patients who had overexpression and activation 13 Kanematsu T, Yano S, Uehara H et al. Phosphorylation, but not overexpression, of epidermal growth factor receptor is associated of HER-2. Quality of life in 400 patients was also significantly with poor prognosis of non-small cell lung cancer patients. Oncol improved with patients having chemotherapy combined with Res 2003; 13: 289–98. 27 trastuzumab than without. 14 Baselga J, Rischin D, Ranson M et al. Phase I safety, pharmacokinetic, Our study, despite using a small sample size of 21 SCC skin and pharmacodynamic trial of ZD1839, a selective oral epidermal tumours, shows the great differences in the amounts of EGFR growth factor receptor tyrosine kinase inhibitor, in patients with and phos-EGFR between each sample, and highlights the vari- five selected solid tumor types. J Clin Oncol 2002; 20: 4292–302. ability in the phenotype of carcinomas despite their similar 15 Albanell J, Rojo F, Averbuch S et al. Pharmacodynamic studies of the epidermal growth factor receptor inhibitor ZD1839 in skin histologies. Our results suggest that it is vital to characterize from cancer patients: histopathologic and molecular consequences the target in tumours for which targeted agents (e.g. EGFRIs) of receptor inhibition. J Clin Oncol 2002; 20: 110–24. are to be trialled so as to avoid useless therapy and toxicity in 16 Bonner JA, Giralt J, Harari PM et al. Phase III evaluation of radi- patients whose tumours lack the target. ation with and without cetuximab for locoregionally advanced head and neck cancer. Int J Radiat Oncol Biol Phys 2004; 60 (Suppl. 1): S147–8. References 17 Barnes CJ, Bagheri-Yarmand R, Mandal M et al. Suppression of epi- 1 Australian Cancer Network Management of Non-Melanoma Skin dermal growth factor receptor, mitogen-activated protein kinase, Cancer Working Party. Epidemiology. In: Clinical Practice Guidelines. and Pak1 pathways and invasiveness of human cutaneous squa- Non-Melanoma Skin Cancer: Guidelines for Treatment and Management in Austra- mous cancer cells by the tyrosine kinase inhibitor ZD1839 (Iressa). lia. National Health and Medical Research Council, 2003; 9. Mol Cancer Ther 2003; 2: 345–51.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 98 EGFR in skin SCC, G.B. Fogarty et al.

18 Ke Liang MS, Ang KK, Milas L et al. The epidermal growth factor 23 Krahn G, Leiter U, Kaskel P et al. Coexpression patterns of EGFR, receptor mediates radioresistance. Int J Radiat Oncol Biol Phys 2003; HER2, HER3 and HER4 in non-melanoma skin cancer. Eur J Cancer 57: 246–54. 2001; 37: 251–9. 19 American Joint Committee on Cancer. Carcinoma of the skin 24 Pegram MD, Konecny GE, O’Callaghan C et al. Rational combin- (excluding eyelid, vulva, and penis). In: AJCC Cancer Staging Manual, ations of trastuzumab with chemotherapeutic drugs used in the 6th edn. New York: Springer-Verlag, 2002; 203–8. treatment of breast cancer. J Natl Cancer Inst 2004; 96: 739–49. 20 Nazmi MN, Dykes PJ, Marks R. Epidermal growth factor recep- 25 Slamon DJ, Leyland-Jones B, Shak S et al. Use of chemotherapy plus tors in human epidermal tumours. Br J Dermatol 1990; 123: 153– a monoclonal antibody against HER2 for metastatic breast cancer 61. that overexpresses HER2. N Engl J Med 2001; 344: 783–92. 21 Lavrijsen AP, Tieben LM, Ponec M et al. Expression of EGF 26 Marty M, Cognetti F, Maraninchi D et al. Randomized phase II trial receptor, involucrin, and cytokeratins in basal cell carcinomas and of the efﬁcacy and safety of trastuzumab combined with docetaxel squamous cell carcinomas of the skin. Arch Dermatol Res 1989; 281: in patients with human epidermal growth factor receptor 2-posi- 83–8. tive metastatic breast cancer administered as ﬁrst-line treatment: 22 Ikeuchi T, Urano Y, Fukuhara K et al. Light microscopic autoradio- the M77001 study group. J Clin Oncol 2005; 23: 4265–74. graphical analysis of [125I] epidermal growth factor binding in 27 Osoba D, Slamon DJ, Burchmore M et al. Effects on quality of life basal cell epithelioma and squamous cell carcinoma of the skin. J of combined trastuzumab and chemotherapy in women with meta- Dermatol 1993; 20: 219–25. static breast cancer. J Clin Oncol 2002; 20: 3106–33.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp92–98 EPIDEMIOLOGY AND HEALTH SERVICES RESEARCH DOI 10.1111/j.1365-2133.2006.07537.x The risk for cutaneous malignant melanoma, melanoma in situ and intraocular malignant melanoma in relation to tobacco use and body mass index A˚. Odenbro,* P. Gillgren, R. Bellocco,* P. Boffetta,*§ N. Ha˚kansson– and J. Adami*,** *Department of Medical Epidemiology and Biostatistics, PO Box 281, Karolinska Institutet, 171 77 Stockholm, Sweden Department of Surgery, Stockholm So¨der Hospital, 118 83 Stockholm, Sweden Department of Statistics, University of Milan-Bicocca, 20126 Milan, Italy §International Agency for Research on Cancer, Lyon, France –National Institute of Environmental Medicine, Karolinska Institutet, 171 77 Stockholm, Sweden **Department of Medicine, Clinical Epidemiology Unit, Karolinska University Hospital, 171 76 Stockholm, Sweden

Summary

Correspondence Background The incidence of cutaneous malignant melanoma (CMM) and mela- A˚sa Odenbro. noma in situ (MIS) has been increasing during the last 50 years. Malignant mela- E-mail: [email protected] noma (MM) is also the most common intraocular malignancy (IMM). Besides ultraviolet radiation, the cause of these tumours is largely unknown. Accepted for publication 14 June 2006 Objectives We designed a study to examine the effect of body mass index (BMI) and tobacco use on the risk for MM and MIS. Key words Methods Analyses were performed on a nationwide cohort of 339 802 Swedish body mass index, cohort study, epidemiology, construction workers. Exposure information was collected prospectively by ques- malignant melanoma, tobacco use tionnaires combined with personal interviews. Conﬂicts of interest Results Follow up yielded a total of 7 663 400 person-years during which 1639 None declared. workers developed MM ⁄MIS. The risk for MM ⁄MIS was reduced in current or previous smokers compared with those who had never smoked, both when analysing all smoking tobacco products combined and when analysing cigarette and pipe smokers separately. The risk was further diminished with longer duration of smoking and greater quantity of tobacco smoked. The effect was more evident in CMM/MIS than in IMM. Snuff taking conferred a decreased risk for CMM/MIS, and a BMI over normal weight range conferred an increased risk for CMM. Conclusions Tobacco smoking was found to be inversely associated with the risk for CMM and MIS. The mechanism of action is unknown but it has been suggested to be due to the immune suppressive effect that tobacco exerts which would be protective against deleterious immune reactions caused by, for example, the sun. Neither is the mechanism behind the higher risk for CMM due to being overweight known. One hypothesis is that it is an effect of a hormonal imbalance. Further studies are required to elucidate these mechanisms.

Cutaneous malignant melanoma (CMM) has been the most frequent intraocular tumour and it presents with a bad prog- rapidly increasing malignancy in developed countries with nosis.6 The only conﬁrmed risk factors for CMM and MIS are large white-skinned populations over the last 50 years.1–3 In phenotype, phototype and sunlight exposure, but other aetio- Sweden, the incidence rate has increased by more than 300% logical factors may also contribute.7,8 The relation between since 1970.4 Despite the rapidly growing number of malig- sunlight exposure and CMM/MIS is complex: intermittent nant melanomas (MMs) the survival rate has improved sub- intense exposure (recreational) and sunburn history have been stantially, a trend attributable mainly to earlier detection.1,5 shown to confer an elevated risk, whereas some studies show The lesions therefore tend to be thinner and are even classiﬁed that chronic sun exposure, e.g. occupational sun exposure, is as melanoma in situ (MIS) at diagnosis today. Intraocular MM inversely related to the risk for CMM/MIS.9 No major causa- (IMM) is an uncommon form of cancer, but it is still the most tive agent has so far been established for IMM. There has been

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 99 100 Malignant melanoma in relation to tobacco and BMI, A˚. Odenbro et al. some evidence for an association with racial characteristics and quality of the smoking data has been reviewed and when with sun exposure, but the results have been inconsistent.10,11 comparing answers from 2 to 3 years apart perfect concord- Tobacco constituents contain a wide range of carcinogenic ance was found in 89%. There were missing values in 1Æ3% of substances and exert multiple toxic effects on the respiratory, current and 1Æ4% of previous smokers and inconsistencies cardiovascular and immunological systems.12,13 There has were found in 2Æ6% of the smoking data.24 Inconsistencies in been conflicting evidence for an association between CMM/ the data on snuff taking were present in 7% of the workers. MIS or IMM and smoking in previous studies.7,14–21 Most of Information on BMI was missing in < 2% of the cohort. these studies have suffered from limited statistical power and Occupational exposure to sunlight has been assessed as pre- have often been of the case–control design. Different specific viously described.25 Briefly, an experienced industrial hygien- tobacco products, such as pipe tobacco, cigars and oral moist ist from the construction industry (N. Hallin) assessed the snuff have rarely been investigated before. amount of sunlight exposure of every worker, using the ex- The association between body mass index (BMI) and CMM tensive job task coding available in this cohort. Sunlight ex- has been explored previously and there has been some evi- posure was categorized into four exposure groups. Subjects dence for an increased risk with increasing BMI.8,14,18,22 As with multiple occupations were assigned an average score these results are based on very few studies with limitations in based on the exposure levels of the component job tasks. We their study designs, the relationship is still far from estab- only used information on job task from the first visit but it lished. has been shown that the workers seldom changed their duties With this very large, nationwide occupational cohort study, and therefore 96Æ3% had the same exposure level throughout with prospectively collected exposure information, we aim to follow up.25 elucidate further the role of tobacco use and BMI in the development of CMM, IMM and MIS. The cohort and follow-up

Subjects and methods First visit defined entry into the cohort. As no information on smoking history was collected from 1975 to 1977 we included only workers who were registered between 1971 and Study setting 1975 and between 1978 and 1992. Over 95% of the cohort The Construction Industry’s Organization for Working Envi- consisted of men; hence we restricted our study to male ronment, Safety and Health (Byggha¨lsan) provided outpatient workers. After excluding 1232 persons who had developed services to construction workers all over Sweden from 1969 cancer before entry into the cohort, 3258 individuals who to 1993. The organization was a joint venture launched by emigrated before entry and 570 individuals with erroneous trade unions and the Swedish Construction Employers’ Associ- personal identification numbers we were left with 339 802 ation. The basic units were stationary and mobile clinics typic- men for analyses. ally staffed by a physician and a few nurses. The main activity Each cohort member was followed from date of entry into was preventive health check-ups, offered to all blue- and the cohort until cancer diagnosis, death, emigration, or end of white-collar employees in the building industry through regu- follow up (31 December 2004), whichever occurred first. Fol- lar (every second year during the early years, every third year low up was made possible by the unique 10-digit personal thereafter) invitations and through visits or advertisements at identification number given to every Swedish citizen.4 Identifi- virtually all major building sites. Although the programme cation numbers were checked exhaustively with usual tech- was voluntary, 85–90% of eligible workers participated at least niques to ensure that they were complete and valid. The once.23 Beginning in 1971 the information collected was personal identification number was then used for record link- stored in a computerized register. age to the National Death Registry (date and cause of death), the Migration Registry (date of emigration) and the Swedish Cancer Registry (SCR). The SCR, founded in 1958 and func- Exposure information tional throughout this study, receives reports of all incident Exposure information was collected by letting each worker malignancies diagnosed in Sweden. By law, both treating complete a questionnaire before every visit. On average each physician and pathologist are required to report to the Regis- cohort member underwent three health check-ups. To avoid try, which ensures recording of > 98% of all malignant misunderstandings or inconsistencies, the answers were dou- tumours, with histological verification of 97%.4,26 There is no ble-checked by a nurse. During the period from 1971 to 1993 scientific study available on the quality and completeness of four different questionnaires were used. The questionnaires MIS registration in the SCR. The reporting of intraocular used during 1971–1974 are extensive with almost 200 items tumours has been studied previously.27 With time, eye-saving including a detailed history of smoking and snuff using, an- treatments have become more usual and therefore histological thropometric measures and occupational coding in 200 categ- verification of these tumours has varied, but it is not as high ories. This questionnaire was removed during the period as for tumours registered in the SCR generally. During the 1975–1977 but was resumed and expanded in 1978 to record period of this study the International Classification of Diseases, 7th an even more detailed smoking history from then on. The edition (ICD-7) was in use and the following codes were used

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 Malignant melanoma in relation to tobacco and BMI, A˚. Odenbro et al. 101 for identifying incident cases of CMM and IMM: ICD-7: Table 1 Basic characteristics of study population 190.1–190.9 (CMM; Pathological Anatomical Diagnosis, PAD 176 for CMM and 173/174 for MIS) and 192.0 (PAD 176 for No. cases in the cohort IMM). CMM 1309 MIS 267 IMM 63 Statistical analyses Total 1639 Person-years 7 663 400 The distribution of BMI was categorized in four groups Mean age at entry, years (range) 34Æ2 (14–82) ) according to World Health Organization criteria: underweight, Mean BMI, kg m 2 (range) 24Æ2 (10Æ7–55Æ0) BMI < 18Æ5; normal weight, BMI 18Æ5–25; overweight, BMI Tobacco use (all melanoma cases combined), 25–30; obese, BMI > 30. Because there were too few cases in no. cases (person-years) the extreme categories, data were analysed in two groups: Tobacco nonusers 599 (2 289 820) a underweight/normal and overweight/obese. Smoking tobacco combined Former smokers 236 (811 380) The distribution of age was categorized in eight groups, Current smokers 454 (2 468 390) each with a range of 5 years. Pure cigarette smokers 440 (2 254 730) When assessing the total amount of tobacco smoked by a Pure cigar smokers 15 (46 850) worker daily, cigarettes were assumed to contain 1 g of Pure pipe smokers 88 (399 970) tobacco and cigars 6 g of tobacco. Snuff users and pipe smok- Pure snuff users 96 (815 050) ers reported the amount of tobacco (in g) used every week. Mixed tobacco users 401 (1 856 970) Age-, BMI- and sun exposure-adjusted rate ratios (RRs) Occupational sun exposure (during working days) were estimated using the Cox proportional hazards regression Seldom or never 781 (3 616 800) model, which allows studying time to event data, accounting Intermittently 601 (2 633 280) for censoring and loss to follow up; the relevant exposure fac- Continually 68 (408 980) tors were included as independent variables and their effects Incessantly 26 (125 130) were estimated in separate models. Removing the variables birth cohort and sun exposure from the analyses did not alter CMM, cutaneous malignant melanoma; MIS, melanoma in situ; IMM, intraocular malignant melanoma; BMI, body mass index. the estimated incidence RRs or the 95% confidence intervals aSnuff users excluded. (CIs) significantly; however, they were kept in the models as they might be considered as potential confounders. Smoking status, duration of smoking, time since cessation of smoking and BMI were established on the date of entry. Parameter estimates and 95% CIs were obtained by maxim- users) and 25% used two or more tobacco products—most izing the partial likelihood, using the stset procedure in Stata frequently cigarettes and snuff (14%). The mean BMI in the ) 8.2 (StataCorp 2003: Stata Statistical Software, Release 8.2; cohort was 24Æ2kgm 2. Most workers had a low or medium Stata Corporation, College Station, TX, U.S.A.). Trend tests amount of occupational sun exposure. were calculated, when applicable, by taking the mean expos- Throughout follow up, a total of 1639 men developed ure in each class of the continuous variable, after categoriza- MM ⁄MIS. Of these, 1309, 267 and 63 men developed CMM, tion, and then including the new ‘scored’ variable as a MIS and IMM, respectively. Table 2 displays the adjusted inci- continuous variable in the model. Interaction effects between dent RRs and 95% CIs for MM ⁄MIS due to the combined biologically plausible exposures were also included in the effect of tobacco smoking, including cigarettes, cigars and model and the likelihood ratio test was used to assess their pipe smoking. Current smokers were at a 35–50% lower risk statistical significance. of developing all outcomes. Ex-smokers had a 25% decreased The main underlying assumptions of the Cox model were risk for CMM compared with those who had never smoked assessed by looking at the distribution of Schoenfeld residuals but there was no risk change observable for MIS or IMM. This and by the related test of proportionality; when these were effect of tobacco smoking is confirmed by the analyses of not fulfilled a stratified Cox regression was fitted instead. quantity smoked, accumulated quantity smoked and duration of smoking. With increasing duration of smoking, the risk for Results CMM and MIS decreased. Cessation of smoking within the last 10 years from entry into the cohort conferred approximately The baseline characteristics of the study population are shown the same risk for CMM and MIS as smoking cessation of more in Table 1. Mean age at entry was 34Æ2 years and the cohort than 10 years from entry into the cohort. The risk for IMM members were followed on average for 22Æ6 years (range increased markedly with longer time since cessation but with 0Æ01–33Æ53), yielding a total of 7 663 400 person-years of a very wide CI. The risk for CMM, MIS and IMM was also follow up. About 70% had ever used some kind of tobacco decreased in the analyses of quantity smoked. The lowest esti- product. Forty-seven per cent used one tobacco product only mate of CMM and MIS risk was detected in the category of (30% were pure cigarette smokers and 10% were pure snuff moderate quantity smoked. By calculating accumulated

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 102 Malignant melanoma in relation to tobacco and BMI, A˚. Odenbro et al.

Table 2 Age-, sunlight exposure-, birth cohort- and body mass index-adjusted incidence rate ratios (RRs) and 95% conﬁdence intervals (CIs) for developing melanoma due to smoking tobacco exposure

Cutaneous malignant Intraocular malignant All melanoma combined melanoma Melanoma in situ melanoma

RR 95% CI RR 95% CI RR 95% CI RR 95% CI Tobacco nonusers 1 1 1 1 Smoking status Previous 0Æ80 0Æ68–0Æ94 0Æ75 0Æ63–0Æ91 0Æ97 0Æ66–1Æ43 1Æ05 0Æ52–2Æ13 Current 0Æ63 0Æ55–0Æ72 0Æ66 0Æ58–0Æ77 0Æ47 0Æ33–0Æ67 0Æ57 0Æ28–1Æ13 Quantity of tobacco smoked per day (g)a 1–5 0Æ76 0Æ62–0Æ94 0Æ75 0Æ60–0Æ95 0Æ71 0Æ42–1Æ21 0Æ92 0Æ36–2Æ32 6–15 0Æ62 0Æ53–0Æ73 0Æ63 0Æ53–0Æ75 0Æ56 0Æ38–0Æ82 0Æ78 0Æ39–1Æ58 >15 0Æ69 0Æ59–0Æ81 0Æ71 0Æ60–0Æ85 0Æ63 0Æ42–0Æ95 0Æ50 0Æ20–1Æ25

Ptrend ¼ 0Æ18 Ptrend ¼ 0Æ32 Ptrend ¼ 0Æ23 Ptrend ¼ 1Æ00 Duration of smoking (years) 1–10 0Æ74 0Æ62–0Æ89 0Æ76 0Æ62–0Æ93 0Æ67 0Æ43–1Æ05 1Æ07 0Æ42–2Æ76 11–20 0Æ74 0Æ63–0Æ87 0Æ75 0Æ62–0Æ89 0Æ59 0Æ39–0Æ91 1Æ23 0Æ59–2Æ52 >20 0Æ59 0Æ49–0Æ70 0Æ60 0Æ50–0Æ73 0Æ58 0Æ37–0Æ90 0Æ39 0Æ17–0Æ90

Ptrend <0Æ001 Ptrend <0Æ001 Ptrend ¼ 0Æ006 Ptrend ¼ 0Æ03 Accumulated quantity smoked (duration · quantity, g) 1–499 0Æ82 0Æ71–0Æ93 0Æ82 0Æ71–0Æ96 0Æ69 0Æ49–0Æ97 1Æ19 0Æ64–2Æ18 500–999 0Æ55 0Æ46–0Æ66 0Æ56 0Æ46–0Æ68 0Æ61 0Æ40–0Æ94 0Æ17 0Æ04–0Æ72 > 999 0Æ38 0Æ28–0Æ52 0Æ41 0Æ29–0Æ58 0Æ28 0Æ11–0Æ68 0Æ22 0Æ03–1Æ62

Ptrend ¼ 0Æ15 Ptrend ¼ 0Æ25 Ptrend ¼ 0Æ34 Ptrend ¼ 0Æ84 Recency of smoking cessation (years) 1–10 0Æ80 0Æ66–0Æ96 0Æ75 0Æ61–0Æ94 1Æ03 0Æ66–1Æ59 0Æ72 0Æ27–1Æ93 >10 0Æ87 0Æ68–1Æ12 0Æ77 0Æ58–1Æ04 1Æ12 0Æ61–2Æ07 1Æ76 0Æ74–4Æ17

Ptrend ¼ 0Æ55 Ptrend ¼ 0Æ64 Ptrend ¼ 0Æ13 Ptrend ¼ 0Æ07

aCombined quantity from cigarette (one cigarette ¼ 1 g), cigar (one cigar ¼ 6 g) and pipe smoking.

quantity smoked (quantity smoked · duration of smoking) a not reach statistical significance. The risk for MIS was not decreasing risk trend for CMM, MIS and IMM was again altered by duration of using snuff. observed, although the RR estimates for IMM were based on The risk for CMM was significantly increased in the over- few cases. weight/obese group compared with the underweight/normal The effect on MM ⁄MIS incidence by the exclusive use of weight group (Table 4). There was no association between one type of tobacco product is shown in Table 3. Compared BMI and MIS, and the slightly increased risk for IMM was with those who had never used tobacco, pure cigarette smok- based on very few cases. ers had an RR of 0Æ7(0Æ6–0Æ8) of developing CMM, an RR of Biologically plausible interactions, mainly between age and 0Æ7(0Æ5–0Æ9) of developing MIS and an RR of 0Æ9(0Æ5–1Æ6) tobacco exposure category, were tested when possible and did of developing IMM. In the same setting pure pipe smokers not reveal any modification of the main effects. were at a 0Æ6(0Æ5–0Æ8) risk for CMM, a 0Æ4(0Æ2–0Æ7) risk for MIS, and a 0Æ6(0Æ2–1Æ7) risk for IMM. The analysis on pure Discussion cigar smoking was based on small numbers but cigar smoking did not appear to confer significant risk changes for CMM or With this large cohort of Swedish male construction workers MIS and the risk could not be estimated for IMM. Pure snuff we have shown consistent evidence of a decreased risk for using reduced the risk for CMM and MIS by 40% but had no CMM and MIS by tobacco smoking and snuff using. A BMI detectable effect on IMM. The results on amount of pure above the normal weight range conferred an increased risk for cigarette tobacco smoked per day were very similar to those CMM but not for MIS or IMM. on combined smoking tobacco. The impact of the amount of Ultraviolet (UV) radiation is the only established risk factor pipe tobacco smoked per week and number of cigars smoked for CMM/MIS. It is generally accepted that characteristics rela- per day was hard to evaluate due to small numbers of ted to ethnicity, such as skin type and eye colour, number of MM ⁄MIS in the exposure categories. Risk of CMM and MIS atypical naevi and sunlight exposure have a strong impact on decreased with increasing duration of snuff using. The RR for the risk of developing CMM and MIS.9 CMM was 0Æ7(0Æ5–0Æ9) for 1–29 years’ duration and 0Æ5 Tobacco smoke contains a number of carcinogenic com- (0Æ2–1Æ0) for ‡ 30 years’ duration, while the risk for MIS did pounds and has been shown to have an aetiological role in a

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 Malignant melanoma in relation to tobacco and BMI, A˚. Odenbro et al. 103

Table 3 Age-, body mass index-, birth cohort- and sunlight exposure-adjusted incidence rate ratios (RRs) and corresponding 95% conﬁdence intervals (CIs) for developing melanoma according to the use of different tobacco products separately

Cutaneous malignant Intraocular malignant All melanoma combined melanoma Melanoma in situ melanoma

RR 95% CI RR 95% CI RR 95% CI RR 95% CI Tobacco nonusers 1 1 1 1 Pure cigarette smokers 0Æ69 0Æ61–0Æ79 0Æ69 0Æ59–0Æ80 0Æ67 0Æ49–0Æ94 0Æ86 0Æ45–1Æ62 Quantity of cigarette tobacco smoked daily (g) 1–9 0Æ73 0Æ61–0Æ86 0Æ73 0Æ60–0Æ88 0Æ68 0Æ45–1Æ04 0Æ86 0Æ38–1Æ96 10–19 0Æ62 0Æ51–0Æ76 0Æ61 0Æ49–0Æ76 0Æ56 0Æ33–0Æ93 1Æ19 0Æ52–2Æ72 ‡ 20 0Æ73 0Æ57–0Æ92 0Æ72 0Æ55–0Æ94 0Æ92 0Æ51–1Æ66 0Æ46 0Æ11–2Æ02

Ptrend <0Æ001 Ptrend <0Æ001 Ptrend ¼ 0Æ10 Ptrend ¼ 0Æ79 Pure pipe smokers 0Æ58 0Æ46–0Æ74 0Æ62 0Æ48–0Æ81 0Æ35 0Æ17–0Æ73 0Æ64 0Æ24–1Æ72 Pure cigar smokers 0Æ79 0Æ46–1Æ38 1Æ00 0Æ57–1Æ73 - - Pure snuff users 0Æ65 0Æ52–0Æ82 0Æ63 0Æ48–0Æ81 0Æ64 0Æ36–1Æ14 1Æ14 0Æ43–3Æ07 Duration of snuff use (years) 1–29 0Æ71 0Æ55–0Æ90 0Æ70 0Æ53–0Æ92 0Æ67 0Æ37–1Æ23 1Æ17 0Æ33–4Æ10 ‡ 30 0Æ51 0Æ27–0Æ98 0Æ47 0Æ22–1Æ00 0Æ39 0Æ05–2Æ88 1Æ05 0Æ23–4Æ79

Ptrend <0Æ001 Ptrend <0Æ001 Ptrend ¼ 0Æ08 Ptrend ¼ 0Æ75 Mixed tobacco usea 0Æ71 0Æ62–0Æ81 0Æ71 0Æ61–0Æ82 0Æ75 0Æ55–1Æ05 0Æ57 0Æ28–1Æ16

aCigarette smoking and snuff taking was the most common combination.

Table 4 Age-, body mass index (BMI)-, birth cohort-, sunlight exposure- and tobacco Cutaneous Intraocular product usage-adjusted incidence rate ratios All melanoma malignant Melanoma malignant (RRs) and corresponding 95% conﬁdence combined melanoma in situ melanoma intervals (CIs) for developing melanoma according to different levels of BMI RR 95% CI RR 95% CI RR 95% CI RR 95% CI BMI Underweight/normal 1 1 1 1 Overweight/obese 1Æ27 1Æ14–1Æ42 1Æ34 1Æ19–1Æ52 0Æ95 0Æ72–1Æ25 1Æ22 0Æ72–2Æ07

number of malignancies.28–30 Smoking also decreases cutane- ated for MIS or IMM as there were too few cases. Only three ous blood flow and has been demonstrated to depress case–control studies have previously investigated the effect of immune responses.31 However, the evidence for an association pure pipe and cigar smoking on the development of CMM/ with CMM/MIS is contradictory. Most studies have been of MIS/IMM.7,15,21 These studies showed no association with case–control design and therefore might have been prone to pipe or cigar smoking; they did, however, suffer from poor bias and were limited by small sample sizes. Freedman et al.18 statistical precision. found a decreased risk for melanoma in men who had ever Oral moist snuff is a form of smokeless tobacco primarily smoked that was consistent in measurements of quantity used in Sweden. It has been suggested that snuff may be smoked and duration of smoking. They also performed a involved in the aetiology of some cancers.28,29 However, we review of epidemiological studies and found an indicated pat- failed to find any previous studies on the relationship between tern of decreased risk for melanoma in tobacco smokers. A snuff and CMM/MIS or IMM. Our study provides novel evi- possible explanation to the apparently negative association dence of a decreased risk for CMM and MIS in snuff users, with smoking could be its immune depressive effect that which appears to be independent from the effect of tobacco hypothetically would protect the melanocytes from the inflam- smoking. ) matory reaction induced by UV radiation. An alternative ex- In this study, a BMI > 25 kg m 2 was associated with an planation of this finding is the fact that smokers are less increased risk for CMM but had no effect on MIS or IMM. physically active than nonsmokers and therefore spend more This is supported by a few other studies and hormonal factors time indoors.32,33 Pipe smoking conferred a decreased risk for have been suggested as one possible explanation.8,14,18,22 CMM/MIS/IMM in this study. There was no effect on CMM The main strengths of this study are the large size, the long risk of cigar smoking and this association could not be evalu- and meticulous follow up, and the prospective, detailed

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 104 Malignant melanoma in relation to tobacco and BMI, A˚. Odenbro et al. collection of exposure information. There are also a few limi- 4 The Swedish Cancer Registry. Cancer Incidence in Sweden. Stockholm: tations to this study. We did not have direct information on National Board of Health and Welfare, 2003 [updated 13 May the amount of occupational or recreational sunlight exposure. 2005]. Available at http://www.socialstyrelsen.se/NR/rdonlyres/ 86E6C7D0-4650-43DA-82DF-FAC11CDB54C5/3019/20044210.pdf Almost all construction workers during this period were, how- (accessed 16 September 2005). ever, in approximately the same socioeconomic class and the 5 Gillgren P, Brattstro¨m G, Frisell J et al. Body site of cutaneous ma- recreational tanning behaviour ought not to differ consider- lignant melanoma—a study on patients with hereditary and mul- ably among workers born close in time to each other. There- tiple sporadic tumours. Melanoma Res 2003; 13:279–86. fore, the recreational sunlight exposure can, to a large extent, 6 Lutz JM, Cree IA, Foss AJ. Risk factors for intraocular melanoma be dealt with by adjusting for birth cohort. The occupational and occupational exposure. Br J Ophthalmol 1999; 83:1190–3. sunlight exposure was adjusted for by creating a sun expos- 7 Westerdahl J, Olsson H, Ma˚sba¨ck A et al. Risk of malignant melanoma in relation to drug intake, alcohol, smoking and hormonal ure matrix. Any remaining misclassification of sunlight ex- factors. Br J Cancer 1996; 73:1126–31. posure is most likely not to differentiate between tobacco 8 Thune I, Olsen A, Albrektsen G, Tretli S. Cutaneous malig- users and nonusers and would therefore bias our results nant melanoma: association with height, weight and body-sur- towards the null. The lack of assessment of the quality and face area. A prospective study in Norway. Int J Cancer 1993; completeness of MIS reporting in the SCR as well as the lim- 55:555–61. ited histological verification of IMM are also limitations to 9 Gandini S, Sera F, Cattaruzza MS et al. Meta-analysis of risk factors for this study, and the results on MIS and IMM should therefore cutaneous melanoma. II. Sun exposure. Eur J Cancer 2005; 41:45–60. 10 Singh AD, Rennie IG, Seregard S et al. Sunlight exposure and patho- be interpreted with caution. There is, however, no reason to genesis of uveal melanoma. Surv Ophthalmol 2004; 49:419–28. believe that reporting depends on smoking status and even- 11 Shah CP, Weis E, Lajous M et al. Intermittent and chronic ultravi- tual bias would therefore also be nondifferential. We were olet light exposure and uveal melanoma. A meta-analysis. Ophthal- unable to perform analyses on women in this cohort. In two mology 2005; 112:1599–607. previous studies the relationship between the BMI and CMM 12 Wiencke JK. DNA adduct burden and tobacco carcinogenesis. Onco- found in men could not be detected in women.8,18 Neither gene 2002; 21:7376–91. could the male relationship to smoking be replicated in 13 National Center for Chronic Disease Prevention and Health Promo- tion. Tobacco Information and Prevention Source. The Health Conse- women in the article by Freedman et al.18 Lastly, this cohort quences of Smoking: A Report of the Surgeon General, 2004 [updated 31 had a mean age of 34Æ2 years at entry and was followed for January 2005]. Available at http://www.cdc.gov/tobacco/sgr/ an average of 22Æ6 years. The cohort is therefore relatively sgr_2004/index.htm (accessed 19 November 2005). young and some of the workers had not reached the mean 14 Shors AR, Solomon C, McTiernan A, White E. Melanoma risk in age for melanoma diagnosis. This will not affect the studied relation to height, weight and exercise. Cancer Causes Control 2001; association with tobacco but considerably more cases can be 12:599–606. expected to accrue with longer follow up and thereby add 15 Osterlind A, Tucker MA, Stone BJ, Jensen OM. The Danish case– control study of cutaneous malignant melanoma. IV. No associ- more power to future analysis. ation with nutritional factors, alcohol, smoking or hair dyes. Int J In conclusion, with this large cohort study we have presen- Cancer 1988; 42:825–8. ted evidence of a decreased risk for CMM and MIS in tobacco 16 De Hertog SAE, Wensveen CAH, Bastiaens MT et al. Relation users and an increased risk for CMM related to obesity. The between smoking and skin cancer. J Clin Oncol 2001; 19:231–8. biological mechanisms behind these findings are unclear. 17 Egan KM, Gragoudas ES, Seddon JM, Walsh SM. Smoking and the Future studies might therefore include experimental parts to risk of early metastasis from uveal melanoma. Ophthalmology 1992; increase our understanding. 99:537–41. 18 Freedman DM, Sigurdson A, Doody MM et al. Risk of melanoma in relation to smoking, alcohol intake, and other factors in a large Acknowledgments occupational cohort. Cancer Causes Control 2003; 14:847–57. 19 Siemiatycki J, Krewski D, Franco E, Kaiserman M. Associations We are grateful to the Swedish Cancer Society for funding this between cigarette smoking and each of 21 types of cancer: a project. multi-site case–control study. Int J Cancer 1995; 24:504–14. 20 Veirod MB, Thelle DS, Laake P. Diet and risk of cutaneous malignant melanoma: a prospective study of 50, 757 Norwegian men References and women. Int J Cancer 1997; 71:600–4. 21 Gallagher RP, Elwood JM, Hill GB. Risk factors for cutaneous ma- 1 Ries LAG, Eisner MP, Kosary CL et al. (eds). SEER 9 Registries 1973– lignant melanoma: the western Canada melanoma study. Recent 2002. SEER Cancer Statistics Review. Bethesda, MD: National Cancer Results Cancer Res 1986; 102:38–55. Institute, 2004. Available at http://seer.cancer.gov/cgi-bin/csr/ 22 Kirkpatrick CS, White E, Lee JA. Case–control study of malignant 1975_2002/search.pl (accessed 25 March 2006). melanoma in Washington State. II. Diet, alcohol, and obesity. Am J 2 World Health Organization. International Agency for Research on Epidemiol 1994; 139:869–80. Cancer and International Association of Cancer Registries. Cancer 23 Engholm G, Englund A. Morbidity and mortality patterns in Swe- Incidence in Five Continents, Vols I–VIII, 2006. Available at http:// den. Occup Med 1995; 10:261–8. www-dep.iarc.fr (accessed 1 April 2006). 24 Nyreń O, Bergstro¨m R, Nystro¨mLet al. Smoking and colorectal 3 Gandini S, Sera F, Cattaruzza MS et al. Meta-analysis of risk factors cancer: a 20-year follow-up study of Swedish construction work- for cutaneous melanoma: I. Common and atypical naevi. Eur J Can- ers. J Natl Cancer Inst 1996; 88:1302–7. cer 2005; 41:284–7.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 Malignant melanoma in relation to tobacco and BMI, A˚. Odenbro et al. 105

25 Ha˚kansson N, Floderus B, Gustavsson P et al. Occupational sunlight 30 Freiman A, Bird G, Metelitsa AI et al. Cutaneous effects of smoking. exposure and cancer incidence among Swedish construction work- J Cutan Med Surg 2004; 8:415–23. ers. Epidemiology 2001; 12:552–7. 31 Moszczynski P, Zabinski Z, Moszczynski P Jr et al. Immunological 26 Mattsson B, Wallgren A. Completeness of the Swedish cancer regis- ﬁndings in cigarette smokers. Toxicol Lett 2001; 18:121–7. ter. Acta Radiol Oncol 1984; 23:305–13. 32 Santmyire BR, Feldman SR, Fleisher AB. Lifestyle high-risk behav- 27 Bergman L, Seregard S, Nilsson B et al. Incidence of uveal mela- iors and demographics may predict the level of participation in noma in Sweden from 1960 to 1998. Invest Ophthalmol Vis Sci 2002; sun-protection behaviors and skin cancer primary prevention in 43:2579–82. the United States. Cancer 2001; 92:1315–24. 28 International Agency for Research on Cancer. Tobacco Habits Other than 33 Kimm SYS, Glynn NW, Kriska AM. Decline in physical activity in Smoking; Betel-quid and Areca-nut Chewing; and Some Related Nitrosamines. black girls and white girls during adolescence. N Engl J Med 2002; IARC Monograph on the Evaluation of the Carcinogenic Risk of Chemicals to 347:709–15. Humans, Vol. 37. Lyon: IARC, 1985. 29 International Agency for Research on Cancer. Tobacco Smoking. IARC Monograph on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 38. Lyon: IARC, 1986.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp99–105 EPIDEMIOLOGY AND HEALTH SERVICES RESEARCH DOI 10.1111/j.1365-2133.2006.07539.x Characteristics and completeness of clinical trial registrations in psoriasis and atopic dermatitis R.D. Quain and K.A. Katz University of Pennsylvania School of Medicine, Philadelphia, PA, U.S.A.

Summary

Correspondence Background Controversy over the failure to publish results of clinical trials linking Rhonda Quain. antidepressant treatment to suicidal behaviour in adolescents has increased inter- E-mail: [email protected] est in clinical trial registration. Objective To assess numbers, characteristics and completeness of registrations of tri- Accepted for publication 1 June 2006 als for psoriasis and atopic dermatitis registered at two web-based trial registries: ClinicalTrials.gov and isrctn.org. Key words Methods In this cross-sectional study we identified trials by searching ClinicalTri- atopic dermatitis, clinical trials, clinical trial als.gov and isrctn.org on 18 January 2006 for trials registered up to 31 Decem- registration, psoriasis ber 2005. We included only trials of therapeutic interventions for atopic Conflicts of interest dermatitis or psoriasis. We ascertained the date of submission of registration, the None declared. funding source of the trial, and whether a registration listed the specific name of the intervention studied, the specific outcome measure used (e.g. ‘Psoriasis Area and Severity Index’), the criterion used to gauge success on the outcome measure (e.g. ‡ 75% decrease), and the time at which the outcome would be assessed (e.g. at 12 weeks). Results There were 156 registered trials, including 128 (82%) at ClinicalTrials.gov [36 (23%) in atopic dermatitis, 92 (59%) in psoriasis] and 28 (18%) at isrctn.org [23 (15%) in atopic dermatitis, 5 (3%) in psoriasis]. Pharmaceutical companies funded 87 trials (56%), federal or governmental agencies 28 (18%), universities or organizations 21 (13%), and a combination of funders 20 (13%). Of atopic dermatitis trials (13 of 36) and (24 of 92) of psoriasis trials at Clinical- Trials.gov were registered in September 2005. The specific name of the intervention studied was listed in 150 registrations (96%), 89 (57%) listed the specific measure used, 69 (44%) listed the criteria to gauge success and 62 (40%) listed the time of assessment. Conclusions While trial registrations in atopic dermatitis and psoriasis are increasing, more complete information in these registrations may increase their value for dermatologists and their patients.

The idea that clinical trials should be prospectively registered adolescents were made public.4,5 In the wake of this debate, dates to 1974, when it was suggested in the context of Presi- the issue of prospective clinical trial registration gained dent Nixon’s ‘War Against Cancer.’1 The idea only began to momentum,6–8 receiving endorsements from the American gain real momentum in 1997, however, when the U.S. Con- Medical Association,9 the Association of American Medical gress passed the Food and Drug Administration Modernization Colleges,10 the Cochrane Collaboration,11 and—with some Act (FDAMA).2 The Act required prospective trial registration caveats—the U.S. and international pharmaceutical industry in serious or life-threatening diseases and mandated the estab- trade groups [Pharmaceutical Research and Manufacturers of lishment of a U.S. government-run clinical trial registry.2 Since America (PhRMA) and the International Federation of Pharma- named ClinicalTrials.gov and now administered by the ceutical Manufacturers and Associations (IFPMA), respective- National Library of Medicine,2,3 the ClinicalTrials.gov registry ly].12–14 Additionally, in September 2004 the International allows sponsors to register trials free of charge. Committee of Medical Journal Editors (ICMJE), a group that In 2004 controversy ensued after the results of unpublished comprises editors from 11 journals, including the New England clinical trials linking antidepressants to suicidal behaviour in Journal of Medicine, JAMA and the Lancet, adopted a policy stating

2006 The Authors 106 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp106–110 Trial registration—psoriasis and AD, R.D. Quain and K.A. Katz 107 that their journals would only consider publishing clinical tri- of terms considered not to denote specific intervention names als that had been prospectively registered in a public trial included a combination of probiotics and new dressing. Exam- registry.15 This policy applied to trials starting after 1 June ples of terms considered not to denote specific outcomes 2005, while ongoing trials required registration by 13 Sep- included safety, efficacy and feasibility. Disagreements were tember 2005.15,16 Currently five registries meet ICMJE specifi- resolved by consensus. cations, including ClinicalTrials.gov, isrctn.org, http:// www.actr.org.au, http://www.umin.ac.jp/ctr/index and Results http://www.trialregister.nl.17 The ICMJE policy has also been adopted by the Archives of Dermatology,18 the Journal of Investigative We identified 168 registrations, including 140 at ClinicalTri- Dermatology19 and the British Journal of Dermatology.20 als.gov and 28 at isrctn.org. Of these, 128 of the studies While not hosting a trial registry itself, the World Health (91%) registered on ClinicalTrials.gov were intervention trials, Organization (WHO) has also advocated universal trial regis- as were all 28 of those registered on isrctn.org, yielding 156 tration.21 In November 2005 the WHO’s International Clinical trials in total. The characteristics of the registrations of these Trials Registry Platform promulgated a list of the 20 elements trials are shown in Table 1. The sponsors of 71 of the 156 tri- it deemed essential for trial registrations;22 these have since als (46%) were based in the U.S.A., while 52 (33%) were been adopted by the ICMJE.16 The WHO Platform is currently based in Europe and 33 (21%) in other areas. setting up an international network of trial registries21 and is Over 80% of trials registered for both psoriasis and atopic examining ways to balance the need for transparency in clinic- dermatitis in both registries included specific intervention al trials with companies’ proprietary concerns regarding the names. Of trials registered on ClinicalTrials.gov, 51 of 92 release of plans and data.23 psoriasis trials (55%) and 18 of 36 atopic dermatitis trials In this study, we assessed the numbers, characteristics and (50%) listed specific outcome measures; of those on is- completeness of trial registrations in psoriasis and atopic rctn.org, two of five (40%) did so for psoriasis and 18 of dermatitis, two areas of recent intense clinical research in 23 (78%) did so for atopic dermatitis. Of trials registered dermatology, registered on ClinicalTrials.gov and isrctn.org. on ClinicalTrials.gov, 39 of 92 psoriasis trials (42%) and 17 Unlike ClinicalTrials.gov, isrctn.org is privately run, based in of 36 atopic dermatitis trials (47%) listed the specific change the U.K. and charges a fee of £120 per trial registration. in outcome measure to be assessed; of those on isrctn.org, no psoriasis trials and 13 of 23 atopic dermatitis trials Methods (56%) did so. Finally, of the ClinicalTrials.gov registrations, 35 of 92 psoriasis trials (38%) and 13 of 36 atopic derma- We identified trial registrations by entering the search terms titis trials (36%) listed the specific time of outcome assess- ‘psoriasis’ and ‘atopic dermatitis’, respectively, in the ‘Disease ment; of the isrctn.org registrations, one of five psoriasis or Condition’ field at http://ClinicalTrials.gov/ct/screen/Ad- trials (20%) did so and 13 of 23 atopic dermatitis trials vancedSearch and in the ‘Search For’ field at http://www.con- (57%) did so. trolled-trials.com/isrctn/ on 18 January 2006. We included Figures 1 and 2 show the monthly number of trial registra- only trials with registrations submitted by 31 December 2005. tions, by type of funding source, for the 92 psoriasis trials Each of us independently reviewed each registration to assess (80%) and 36 atopic dermatitis trials (100%) registered on whether the study was an intervention trial (i.e. a trial of a ClinicalTrials.gov from January 2004 through December 2005. medicine, device or psychosocial intervention) for psoriasis or The month with the greatest number of registrations for both atopic dermatitis. If so, we recorded the trial’s sponsor(s) and diseases was September 2005, when 24 psoriasis trials (26%), the date that the registration was submitted. including 17 industry-sponsored trials, and 13 atopic derma- Independently, we then recorded whether the registration titis trials (36%), including nine of university- or organiza- noted a specific intervention name, outcome measure, change tion-sponsored trials, were registered. There were five in outcome measure and time of outcome assessment. We psoriasis trials with registrations submitted to http:// based our assessment of these last four factors on methods www.clinicaltrials.org in November 1999; in no other month used in a National Library of Medicine study of registrations before January 2004 were there more than two psoriasis trials on ClinicalTrials.gov from May 2005 through October registered. In September 2005 and December 2005, respect- 2005.24 That study assessed whether the intervention name ively, there were four and six atopic dermatitis trials registered gave clinically meaningful insight into the specific treatment on isrctn.org, with no more than three trials registered in any that was being tested. For example, nonspecific terms such as other month. Only one psoriasis trial was registered in each of ‘investigational drug’ were occasionally used. Records were five separate months on isrctn.org. considered acceptable if they specified at least one drug name 24 or a company’s unique serial number. It assessed outcome Discussion measure and time of outcome assessment as follows: ‘It should include the measure used and the time of measurement Prospective clinical trial registration has been endorsed by relative to the start of the intervention, such as death by numerous groups of physician, journal editors and pharma- 180 days after the start of treatment.’24 In our study, examples ceutical company organizations since 2004 as a method to

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp106–110 108 Trial registration—psoriasis and AD, R.D. Quain and K.A. Katz

Table 1 Characteristics of trial registrations for atopic dermatitis and psoriasis at ClinicalTrials.gov and isrctn.org through 31 December 2005

ClinicalTrials.gov (n ¼ 128) isrctn.org (n ¼ 28)

Psoriasis Atopic dermatitis Psoriasis Atopic dermatitis Characteristic (n ¼ 92) [n (%)] (n ¼ 36) [n (%)] (n ¼ 5) [n (%)] (n ¼ 23) [n (%)] Funding source Government agency 10 (11) 0 4 (80) 14 (63) Pharmaceutical company 62 (67) 20 (56) 1 (20) 4 (17) University or organization 7 (8) 11 (31) 0 3 (13) Combination of funding types 13 (14) 5 (14) 0 2 (9) Sponsor locationa U.S. 57 (64Æ8%) 14 (38Æ9%) 0 (0Æ0%) 0 (0Æ0%) Europe 10 (11Æ4%) 14 (38Æ9%) 4 (80Æ0%) 20 (87Æ0%) Other 21 (23Æ9%) 8 (22Æ2%) 1 (20Æ0%) 3 (13Æ0%) Registration listed specific intervention name(s) 91 (99) 35 (97) 5 (100) 19 (83) Registration listed specific outcome measure(s) used 51 (55) 18 (50) 2 (40) 18 (78) Registration listed specific change in outcome 39 (42) 17 (47) 0 13 (56) measure(s) to be assessed Registration listed specific time(s) at which change in 35 (38) 13 (36) 1 (20) 13 (56) outcome measure to be assessed

aFour studies did not include location.

Fig 1. Number of psoriasis trials registered on ClinicalTrials.gov each month between 1 January 2004 and 31 December 2005, by trial funding source.

ensure full disclosure of the existence of clinical trials. This suggests that the impending ICMJE deadline did prompt the study empirically examined trial registration in two areas of registration of previously unregistered ongoing trials. This intense interest in clinical research in dermatology (psoriasis may be especially true for psoriasis and atopic dermatitis, and atopic dermatitis) in two prominent public trial registries depending on whether trial sponsors believe these diseases (ClinicalTrials.gov and isrctn.org) and yielded several import- meet the definition of ‘serious’ elaborated by the FDA in inter- ant findings. preting the FDAMA regulations on mandatory trial registration. Firstly, the absolute number of registrations in psoriasis and A similar trend of increased registration during this period atopic dermatitis has increased over time on ClinicalTrials.gov, was not noted on isrctn.org. This may reflect the fact that as shown in Figures 1 and 2. The increased frequency of reg- there were fewer registrations for these two diseases on is- istrations in the months prior to the 13 September 2005, the rctn.org compared with ClinicalTrials.gov, or that trial spon- ICMJE deadline for the registration of ongoing studies parallels sors in these disease areas favour ClinicalTrials.gov to an overall trend for registrations at ClinicalTrials.gov24 and isrctn.org.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp106–110 Trial registration—psoriasis and AD, R.D. Quain and K.A. Katz 109

15 Pharmaceutical company University or organization Combination of types

5 Number of trials registered

Fig 2. Number of atopic dermatitis trials 0 registered on ClinicalTrials.gov each month between 1 January 2004 and 31 December Jul 04 Jul 05 Jan 04Feb 04Mar 04Apr May04 Jun04 04 Aug Sep04 04Oct 04Nov Dec04 04Jan 05Feb 05Mar 05Apr May05 Jun05 05 Aug Sep05 05Oct 05Nov Dec05 05 2005, by trial funding source.

Secondly, the patterns of registration were different at the While the problems caused by unregistered trials were dem- two registries. A relatively higher percentage of industry onstrated in the case of antidepressants and adolescent sui- sponsored trial registrations were found on ClinicalTrials.gov, cides, deficiencies in the completeness of registration are also while a relatively higher percentage of government sponsored important.25 Two examples illustrate the importance of com- trial registrations were found on isrctn.org. Indeed, PhRMA plete and specific trial registration. Firstly, articles published has pledged that its member companies will register trials at describing clinical trials of intramuscular alefacept did not the U.S.-based ClinicalTrials.gov,12 while the IFPMA, while report results for the trial’s primary endpoint, as specified in noting that numerous public registries exist, also specifically the original protocol.26 Secondly, a report of the gastrointest- identified ClinicalTrials.gov as an appropriate registry.13 inal side effects of the Cox-2 inhibitor celecoxib presented While industry sponsored studies accounted for the majority results, favourable to the drug, from 6 months of follow up; of registrations in both psoriasis and atopic dermatitis on results less favourable to the drug, from 12 and 16 months of ClinicalTrials.gov, most trials registered on ClinicalTrials.gov follow up, as specified in the original study protocols and through October 2005 were government sponsored. This available at the time of submission, were not submitted or may reflect greater commercial interest in these diseases published in the initial journal report of the study.27–29 If compared with other diseases overall. Meanwhile, more trials complete and specific information about these trials’ designs with U.S.-based sponsors were registered on ClinicalTri- had been prospectively registered, these issues could have als.gov, while more trials with European-based sponsors were been addressed prior to publication. registered on the U.K.-based isrctn.org. The cost associated This study has limitations. Firstly, we examined trial regis- with registering a trial on isrctn.org (£120 per registration), trations as they appeared on the two websites. The National compared with free registration at ClinicalTrials.gov, may Library of Medicine study, by contrast, examined raw data also have influenced the numbers and characteristics of regis- entered by trial sponsors into the fields of the ClinicalTri- trations at each site. als.gov database.24 In the processing of these data by registry Thirdly, the information contained in many trial registra- staff to prepare a trial registration for publication on a web- tions was incomplete. The percentages of registrations listing a site, some specificity may have been lost. It is important to specific intervention name was generally high (range 83– bear in mind, however, that the users of registries, patients, 100%), in accord with the study of registrations on Clinical- physicians, journal editors and reviewers, and researchers per- Trials.gov overall.24 However, percentages of registrations list- forming systematic reviews, have access only to processed reg- ing specific information about other important aspects of trial istrations, as we did, and not to raw registry data. Secondly, design, including the outcome measure used (range 40–78%), registrations may be updated over time. Indeed, the National criteria used to gauge success (range 0–56%), and time of Library of Medicine study documented an improvement in the outcome assessment (range 16–62%), were lower. Similarly, specificity of registrations over time.24 Therefore, the registra- in the subset of industry sponsored registrations examined by tions that we examined in this study may not reflect their cur- the National Library of Medicine study of ClinicalTrials.gov, rent specificity. Thirdly, we did not assess other only 31% of registrations were found to include a specific dermatological diseases or other ICMJE-approved registries, to outcome measure and time frame.24 which our results may not necessarily generalize. Fourthly, we

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp106–110 110 Trial registration—psoriasis and AD, R.D. Quain and K.A. Katz did not examine non-ICMJE-approved registries, including one 12 Pharmaceutical Research and Manufacturers of America. PhRMA dermatology-specific registry.30 As of 15 May 2006, however, Clinical Trial Registry Proposal. Washington, DC: PhRMA, c2002–5. this registry included only three eczema and two psoriasis tri- Available at http://www.phrma.org/publications/policy/ 06.01.2005.1111.cfm (accessed 6 January 2005). als, suggesting that it is not widely used for registrations in 30 13 International Federation of Pharmaceutical Manufacturers and Asso- these diseases at present. Finally, we did not assess the com- ciations. Joint Position on the Disclosure of Clinical Trial Information via Clin- pleteness of other elements of trial registrations that are ical Trial Registries and Databases. Geneva, Switzerland: IFPMA, 2005. 22 16 deemed essential by the WHO and ICMJE. Available at http://www.ifpma.org/clinicaltrials.html (accessed 10 Prospective clinical trial registration has become an increas- February 2006). ingly high profile issue. This study demonstrates an increase 14 International Federation of Pharmaceutical Manufacturers and Asso- in trial registration on two prominent public trial registries, ciations. Joint Position on the Disclosure of Sensitive Information via Clinical Trial Registries. Geneva, Switzerland: IFPMA, 2005. Available at ClinicalTrials.gov and isrctn.org, for psoriasis and atopic http://www.ifpma.org/clinicaltrials.html (accessed 10 February dermatitis. While this increase in prospective clinical trial 2006). registration is encouraging, the specificity and completeness of 15 DeAngelis CD, Drazen JM, Frizelle FA et al. Clinical trial registra- information included in these registrations could be improved, tion: a statement from the International Committee of Medical and this would increase their usefulness to physicians and, Journal Editors. JAMA 2004; 292:1363–4. ultimately, to patients. 16 DeAngelis CD, Drazen JM, Frizelle FA et al. Is this clinical trial fully registered? A statement from the International Committee of Med- ical Journal Editors. JAMA 2005; 293:2927–9. Acknowledgments 17 International Committee of Medical Journal Editors. Frequently Asked Questions—Questions About Clinical Trials Registration. Philadelphia, PA: IC- Funding source: Dr Katz’s fellowship is funded by a Kirsch- MJE, 2005. Available at http://www.icmje.org/faq.pdf (accessed stein National Research Service Award from the National Insti- 10 February 2006). tutes of Health. 18 Callen JP, Robinson J. Clinical trial registration: a step forward in providing transparency for the positive and negative results of clinical trials. Arch Dermatol 2005; 141:75 (abstract). References 19 Williams HC, Stern RS. Prospective clinical trial registration. J Invest Dermatol 2005; 124:viii–x (abstract). 1 Dickersin K, Rennie D. Registering clinical trials. JAMA 2003; 20 Ormerod AD, Williams HC. Compulsory registration of clinical tri- 290:516–23. als. Br J Dermatol 2005; 152:859–60. 2 Food and Drug Administration Modernization Act of 1997, Pub. L. 21 World Health Organization. International Clinical Trials Registry Platform. No. 105–115, 113. Geneva: WHO, c2006. Available at http://www.who.int/ictrp/en/ 3 Guidance for Industry: Information Program on Clinical Trials for Serious or Life- (accessed 15 May 2006). threatening Diseases and Conditions. Washington, DC: U.S. Department of 22 World Health Organization. Registration Data Set (Version 1.0). Geneva: Health and Human Services, Food and Drug Administration, Center WHO, c2006. Available at http://www.who.int/ictrp/data_set/ for Drug Evaluation and Research (CDER), Center for Biologics en/index1.html (accessed 15 May 2006). Evaluation and Research (CBER), 2002. Available at http:// 23 World Health Organization. Registry Platform—Discussion Document on www.fda.gov/cder/guidance/4856FNL.PDF (accessed 5 January Disclosure Timing—20 April 2006. Geneva: WHO, c2006. Available at 2005). http://www.who.int/ictrp/Formal_Consult_Discuss_Document.pdf 4 Marshall E. Antidepressants and children. Buried data can be haz- (accessed 15 May 2006). ardous to a company’s health. Science 2004; 304:1576–7. 24 Zarin DA, Tse T, Ide NC. Trial registration at ClinicalTrials. 5 Hensley S, Abboud L. Medical research has ‘black hole’. Wall St J gov between May and October 2005. N Engl J Med 2005; 2004; June 4, Sect B:3. 353:2779–87. 6 Rennie D. Trial registration: a great idea switches from ignored to 25 Drazen JM, Wood AJ. Trial registration report card. N Engl J Med irresistible. JAMA 2004; 292:1359–62. 2005; 353:2809–11. 7 Steinbrook R. Registration of clinical trials—voluntary or manda- 26 Lott JP, Katz KA. Pharmaceutical companies’ policies and practices tory? N Engl J Med 2004; 351:1820–2. regarding prospective registration of dermatology-related clinical 8 Steinbrook R. Public registration of clinical trials. N Engl J Med trials. Br J Dermatol 2005; 155:635–638. 2004; 351:315–17. 27 Fontanarosa PB, Rennie D, DeAngelis CD. Postmarketing 9 American Medical Association. AMA Recommends that DHHS Establish a surveillance–lack of vigilance, lack of trust. JAMA 2004; 292:2647– Registry for all U.S. Clinical Trials (June 15, 2004). Chicago: American 50. Medical Association, 2004. Available at http://www.ama-ssn.org/ 28 Hrachovec JB, Mora M. Reporting of 6-month vs 12-month data in ama/pub/category/13934.html (accessed 13 February 2005). a clinical trial of celecoxib. JAMA 2001; 286:2398; author reply 9– 10 Association of American Medical Colleges. AAMC Expresses Support for 400. Clinical Trials Registry (June 30, 2004). Washington, DC: Association of 29 Wright JM, Perry TL, Bassett KL et al. Reporting of 6-month vs 12- American Medical Colleges. Available at http://www.aamc.org/ month data in a clinical trial of celecoxib. JAMA 2001; 286:2398– newsroom/pressrel/2004/040630.htm (accessed 13 February 400. 2005). 30 Cochrane Skin Group: Ongoing Skin Trials Register. Nottingham: Dermatol- 11 The Cochrane Collaboration. The Cochrane Collaboration Supports Prospect- ogy Department, Queen’s Medical Center, University Hospital, ive Registration of Clinical Trials. Oxford: Cochrane Collaboration, 2006. Available at http://www.nottingham.ac.uk/ongoingskintri- c2004. Available at http://www.cochrane.org/news/articles/ als/ (accessed 15 May 2006). 2004.07.26.htm (accessed 5 January 2005).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp106–110 EPIDEMIOLOGY AND HEALTH SERVICES RESEARCH DOI 10.1111/j.1365-2133.2006.07606.x Validation of the U.K. Working Party diagnostic criteria for atopic eczema in a Xhosa-speaking African population D.A. Chalmers, G. Todd, N. Saxe, J.T. Milne, S. Tolosana, P.N. Ngcelwane, B.N. Hlaba, L.N. Mngomeni, T.G. Nonxuba and H.C. Williams* Department of Dermatology, Faculty of Health Sciences, University of Cape Town, South Africa *Centre of Evidence-Based Dermatology, Queen’s Medical Centre, University of Nottingham, Nottingham, U.K.

Summary

Correspondence Background Reliable diagnostic criteria for eczema are important for epidemiologic- D. Chalmers. al comparisons. Although the U.K. diagnostic criteria for atopic eczema have per- E-mail: c/o [email protected] formed well in an English language setting, limited data are available from other countries where cultural and linguistic factors may affect their validity. Accepted for publication 27 June 2006 Objectives We sought to determine the validity of the U.K. criteria for eczema in relation to clinical assessment by a dermatologist in a Xhosa-speaking South Afri- Key words can population. Africa, diagnostic criteria, eczema, validation Methods A cross-sectional survey of 3067 children aged 3–11 years was conducted in rural, peri-urban and urban settings in South Africa. The prevalence of atopic Conflicts of interest eczema was determined using the U.K. diagnostic criteria and a clinical assess- None declared. ment by a dermatologist. Questions were translated into the local language (Xhosa). Trained researchers administered the questions to the children’s parents or carers. The validity of the U.K. criteria was then determined by calculating the sensitivity, specificity, positive and negative predictive values, and Youden’s Index in relation to the dermatologist’s examination. Results The point prevalence of atopic eczema according to a dermatologist was 1Æ0% [95% confidence interval (CI) 0Æ6–1Æ4], while the prevalence of visible flexural eczema according to the U.K. protocol was 1Æ8% (95% CI 1Æ3–2Æ2). The sensitivity and specificity of the U.K. criteria in this setting was 43Æ7% (95% CI 26Æ3–62Æ3) and 97Æ9% (97Æ3–98Æ4), respectively. The positive and negative predictive values of the U.K. criteria were 18Æ4% (95% CI 10Æ4–28Æ9) and 99Æ4% (95% CI 99Æ0–99Æ6), respectively. The presence of visible flexural eczema according to the U.K. photographic protocol was the best predictor of atopic eczema, with a sensitivity and specificity of 81Æ2% (95% CI 63Æ5–92Æ7) and 99Æ0% (95% CI 98Æ6–99Æ3), respectively, and a positive and negative predictive value of 48Æ1% (95% CI 34Æ3–62Æ1) and 99Æ8% (95% CI 99Æ5–99Æ9), respectively. Conclusions The validity of the full question-based version of the U.K. diagnostic criteria for atopic eczema in this South African setting is low, which may be due to a combination of translational and cultural issues. However, the one physical sign of visible flexural eczema performed well, suggesting that it alone might be a useful tool for future international comparative prevalence studies.

Atopic eczema (syn. atopic dermatitis) refers to the skin mani- a particular phenotype that may lead to better understanding festations of a hypersensitivity phenotype. In the past no clear of the underlying causes of this poorly understood condition. definition of atopic eczema existed so that diagnosis may have While the Hanifin and Rajka criteria were a major step for- varied between clinicians. In the early 1980s Hanifin and Raj- ward in suggesting the many clinical features that character- ka proposed a set of four major and 23 minor diagnostic fea- ized the phenotype of atopic eczema, their complexity, lack of tures for atopic eczema in an attempt to guide physicians in standardization and unknown validity and repeatability made their diagnosis.1 These criteria suggested the main features of them unsuitable for epidemiological studies. Therefore, in

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp111–116 111 112 Validation of U.K. diagnostic criteria for atopic eczema, D.A. Chalmers et al.

1994, a U.K. Working Party set about the task of developing a were also calculated for individual questions, and for combin- minimum set of valid and reliable diagnostic criteria for atopic ations of questions as alternatives to the combination stipula- eczema that could be used in epidemiological studies.2 The ted in the final formulation of the U.K. criteria. The data criteria have since been validated in hospital and community collected were double-entered into a Microsoft Access XP-Pro- settings in the U.K. and in other countries including Romania, fessional database and checked for errors and inconsistencies. Iran, Denmark, Spain, China, Germany, Australia, Italy and Statistical analysis was performed using Stata-8 (Statistics/Data Ethiopia.3–11 The demonstrated validity in these settings has Analysis; Stata Corporation, College Station, TX, U.S.A.). varied from excellent (e.g. China) to poor (Iran). Some of the Consent was obtained from all the schools participating in differences in validity might be due to methodological or the research as well as from local community leaders and the study population differences; others might be genuine differ- local authorities. Written or verbal consent was obtained from ences in instrument performance related to translational and the parent or guardian of the child of interest. The care givers cultural factors. then answered the questionnaire, with the understanding that As those who developed the U.K. diagnostic criteria have they could refuse to answer particular questions should they recommended, the criteria should ideally be validated in the wish. Disclosure of information was on a voluntary basis and setting in which they are to be used. We wished to undertake participants were informed that they could withdraw from the a study exploring whether a rural–urban gradient for atopic study at any stage if they were not satisfied with the process. eczema occurred among the Xhosa people of South Africa, the Ethical approval was granted for this study by the University results of which will be published elsewhere. In order to of Cape Town Research and Ethics Committee and by the choose the best instrument for comparing disease prevalence Queen’s Medical Centre (Nottingham) Ethics Committee. in different rural–urban settings across South Africa, we sought to determine the validity of the U.K. Working Party’s Results refinement of the Hanifin and Rajka criteria against clinical diagnosis by a dermatologist. Population characteristics and overall atopic eczema prevalence Materials and methods Of the 3144 eligible participants, 3069 were examined, repre- A community-based cross-sectional prevalence survey was senting a response rate of 97Æ6%. Two cases were excluded conducted. For the comparative study, three areas were because they did not meet the age inclusion criteria. A total of defined: urban Cape Town, as defined by the local municipal- 3067 individuals were included in the study. Roughly equal ity; peri-urban; two informal or unplanned settlements within numbers were examined in each of the three study areas 10 km of the local municipality of Cape Town and a rural (urban, n ¼ 1002; peri-urban, n ¼ 1017; rural n ¼ 1048). area, further than 10 km away from the local municipality.12 The mean age of the total study population was 6Æ6 years (SD A cluster sampling technique was employed. 2Æ5), with a slight female preponderance (52Æ4% girls, 47Æ5% All parents of children aged 3–11 years were invited to par- boys). All participants were black Xhosa-speaking children. ticipate in the survey. The questions used for the U.K. diag- There was no statistically significant association between the nostic criteria were identical to those recommended in the person interviewed and the likelihood of being diagnosed online manual.13 Other questions were included as a means of with atopic eczema, either according to the questionnaire or internal validation. Questionnaires were translated into Xhosa on clinical grounds. However, it was noted that while the as- and then verified by back-translation prior to the commence- sociation was not statistically significant (P ¼ 0Æ76), a high ment of the study. The translated questionnaire was also proportion of those deemed positive according to the ques- piloted with 38 parents, and changed slightly as a result of tionnaire had information provided by an unclassified carer translation problems. Individuals who took part in the pilot (n ¼ 24, 31Æ5%). study were not included in the main analysis. All question- In total, 76 cases were classified as having atopic eczema naires were administered by trained bilingual Xhosa–English according to the full U.K. Working Party diagnostic criteria, interviewers. corresponding to a 1-year period prevalence of 2Æ48% [95% In order to assess the validity of the questionnaire the defin- confidence interval (CI) 1Æ9–3Æ0]. A total of 32 children were itive test, or gold standard, was defined as the clinical examin- deemed to be suffering from atopic eczema at the time of the ation by a dermatologist.14,15 One of three qualified clinical examination by the examining dermatologists, corres- dermatologists examined all children who had been inter- ponding to a point prevalence of 1Æ04% (95% CI 0Æ6–1Æ4). viewed. The skin was examined for the presence of all skin diseases as well as for the presence of visible flexural eczema. Validity of the U.K. diagnostic criteria The validity of the questionnaire was then assessed by calculating sensitivity, specificity and positive and negative pre- The sensitivity, specificity and predictive values of each item dictive values. Youden’s Index (the sum of sensitivity plus in the U.K. Working Party diagnostic criteria along with com- specificity minus 100) was also calculated to give an overall posite definitions in relation to the reference standard of a measure of relative value.16 These measurements of validity clinical diagnosis of atopic eczema are shown in Table 1.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp111–116 ora Compilation Journal 06TeAuthors The 2006 06BiihAscaino Dermatologists of Association British 2006

Table 1 Validity of U.K. diagnostic criteria for atopic eczema when compared with diagnosis by a dermatologist

Sensitivity % Speciﬁcity % Positive predictive Negative predictive Youden’s (95% CI) (95% CI) value % (95% CI) value % (95% CI) Indexa Individual questions 1a. Itchy skin condition in the last year 71Æ8 (53Æ2–86Æ2) 49Æ2 (47Æ4–51Æ0) 1Æ4(0Æ9–2Æ2) 99Æ4 (98Æ8–99Æ7) 21 1b. Itchy skin condition in the last week 68Æ7 (49Æ9–83Æ8) 51Æ6 (49Æ8–53Æ4) 1Æ4(0Æ9–2Æ2) 99Æ3 (98Æ8–99Æ7) 20Æ3 2. Onset of this skin condition under 2 years 9Æ3(1Æ9–25Æ0) 97Æ5 (96Æ9–98Æ0) 3Æ9(0Æ8–10Æ9) 99Æ0 (98Æ6–99Æ3) 6Æ8 3. History of this skin condition ever 68Æ7 (49Æ9–83Æ8) 62Æ8 (61Æ1–64Æ5) 1Æ9(1Æ2–2Æ8) 99Æ4 (99Æ0–99Æ7) 31Æ5

affected the skin creases Chalmers D.A. eczema, atopic for criteria diagnostic U.K. of Validation 4. History of generally dry skin in the last year 62Æ5 (43Æ6–78Æ9) 81Æ5 (80Æ1–82Æ9) 3Æ4(2Æ1–5Æ2) 99Æ5 (99Æ1–99Æ7) 44 • 5a. Personal history of asthma or hayfever 6Æ2(0Æ7–20Æ8) 97Æ1 (96Æ4–97Æ6) 2Æ2(0Æ2–7Æ8) 98Æ9 (98Æ5–99Æ3) 3Æ3 rts ora fDermatology of Journal British 5b. Family history of asthma, hayfever or 0Æ0(0Æ0–10Æ8) 98Æ8 (98Æ4–99Æ2) 0Æ0(0Æ0–10Æ0) 98Æ9 (98Æ5–99Æ2) )1Æ2 eczema for children under 4 years One physical sign 6. Visible ﬂexural eczema 81Æ2 (63Æ5–92Æ7) 99Æ0 (98Æ6–99Æ3) 48Æ1 (34Æ3–62Æ1) 99Æ8 (99Æ5–99Æ9) 80Æ2 Combinations of criteria Positive according to U.K. criteria (Q1 plus 43Æ7 (26Æ3–62Æ3) 97Æ9 (97Æ3–98Æ4) 18Æ4 (10Æ4–28Æ9) 99Æ4 (99Æ0–99Æ6) 41Æ6 three or more of features of Q2–6)

2007 Q1 plus four or more features of Q2–6 12Æ5(3Æ5–28Æ9) 99Æ7 (99Æ4–99Æ8) 30Æ7(9Æ0–61Æ4) 99Æ0 (98Æ6–99Æ3) 12Æ2 Q1 plus one or more features of Q2–6 71Æ8 (53Æ2–86Æ2) 63Æ8 (62Æ1–65Æ5) 2Æ0(1Æ3–3Æ0) 99Æ5 (99Æ1–99Æ7) 35Æ6

156, Q1 plus two or more features of Q2–6 65Æ6 (46Æ8–81Æ4) 87Æ7 (86Æ5–88Æ8) 5Æ3(3Æ3–8Æ0) 99Æ5 (99Æ2–99Æ7) 53Æ3 Self-report of ‘eczema’ ever (not in criteria) 18Æ7(7Æ2–36Æ4) 94Æ7 (93Æ8–95Æ5) 3Æ6(1Æ3–7Æ7) 99Æ1 (98Æ6–99Æ4) 13Æ4 pp111–116

aYouden’s Index (sensitivity plus speciﬁcity minus 100). When the sensitivity plus the speciﬁcity are equal to 1, then the technique used is no better than random chance. In this instance the Youden’s Index is therefore zero.16 tal. et 113 114 Validation of U.K. diagnostic criteria for atopic eczema, D.A. Chalmers et al.

Only the presence of visible flexural eczema consistently ing hardships from other problems encountered on a daily predicted the clinical diagnosis of atopic eczema with a sensi- basis. Another clue that translation/cultural factors may have tivity of 81Æ3% and specificity of 99Æ1%. Questions designed affected the question performance is the observation that only to elicit a personal history of hay fever, asthma or eczema had 68Æ7% of parents of clinical cases reported a history of flexural low sensitivity, as did questions relating to age of onset and a eczema, yet a greater proportion had visible flexural eczema family history of atopy. when examined (81Æ2%). The response to the extra question on a personal history of ever having had ‘eczema’ is also revealing—over 80% of those with clinical eczema did not Analysis of false positive and negative results think they had it. Conversely, 5% of those who did not have Only 14 of the 76 individuals classified as having atopic clinical atopic eczema thought that they had eczema. Disease eczema according to the U.K. criteria had atopic eczema on severity played no role in the likelihood of being diagnosed examination. To better understand these 62 ‘false positives’ positive according to the U.K. criteria (P ¼ 0Æ466). detailed reviews of their records were undertaken. The distri- In a recent validation of the Xhosa translation of the EQ- bution was not equal between areas, with a larger proportion 5D, a health-related quality of life measure, researchers found being found in the urban area. There was no significant age that while translation of certain words into Xhosa may be or sex difference noted. There was also no statistically signifi- done easily, the concept may not translate as effectively so that cant difference according to the person conducting the inter- the desired interpretation of the word is lost.18 Back-transla- view (P ¼ 0Æ894). tion is an acceptable means of assessing accuracy of transla- Of the 32 individuals diagnosed as having atopic eczema by tion, but it is often possible to pick up only gross errors clinical examination, 18 were classified by the questionnaire rather than subtle changes in the concept being conveyed.19 as not affected, i.e. ‘false negatives’. Again, relatively more Thus the two versions may be comparable in language, but ‘false negatives’ were found in the urban district. The clinical not in meaning. In a recent review of translation methodo- features of the 18 false negative cases were not recorded as logy, comparisons were made between various methods of being unusual by the dermatologist. Disease severity did not translation, including back-translation with monoglot testing play a role in this instance. as was used in this study.20 They concluded that back-translation is the most appropriate method for transcultural research, Discussion and while testing on bilingual subjects is perhaps the most desirable, they conceded that resource limitations may pre- This study has shown that the full U.K. Working Party criteria clude this in most instances. In this study, although every for diagnosing atopic eczema do not work well when applied effort was taken to ensure comparability of language and in- to a Xhosa population. However, the single physical sign of vis- terpretation for the Xhosa translation of the questionnaire, it is ible flexural eczema, which is used as part of the U.K. criteria, still possible that systematic errors of translation occurred. performed well. Several factors may have affected the validity of As highlighted previously,3 low disease prevalence is a key the questions used in this survey. factor in adversely affecting positive predictive values, with Translation and cultural issues are likely to have been im- values dropping quickly when the prevalence of disease falls portant factors in influencing the validity of specific questions. below around 5%.13 Thus, even a criterion with very good Questions relating to time periods such as ‘in the last year’ sensitivity and specificity such as the sign of visible flexural were particularly unreliable (j ¼ 0Æ09), possibly reflecting eczema (81% and 99%, respectively) will have a poor positive cultural differences in concepts of time and seasons. Lack of predictive value of 48% when the overall prevalence is as low familiarity with ‘hay fever’, ‘eczema’ and ‘asthma’ may also as 1% as in this study, whereas the corresponding positive have contributed to the low validity of these particular ques- predictive value for this sign with the same sensitivity and tions in this population with poor access to health care.17 specificity in a setting where the overall prevalence was 10% The fact that 30% of those who were deemed to have active would be very high at 90%. atopic eczema by the dermatologist did not answer positively The prevailing clinical concept of what might constitute a to an itchy skin condition suggests a problem in understand- typical case of atopic eczema might be different in those used ing questions relating to itch or it may be considered to be to examining dark skins in Africa when compared with those commonplace and therefore unimportant. Further investiga- used to seeing mainly pale skins in Europe. It is possible that tion into the Xhosa cultural understanding of the phrases in rural South Africa, more follicular and papular forms with ‘itchy skin’ and ‘eczema’ are warranted before specific conclu- eczema on extensors rather than flexural areas might represent sions can be drawn. Another explanation for the loss in sensi- atopic eczema. Examination of a sample of clinical photo- tivity of the symptom of itching is that the questions were graphs of cases and noncases by one of the U.K. authors answered by mothers and carers rather than through a direct (H.C.W.) during a site visit suggests remarkable similarity response from the child. It is also possible that some of those rather than differences in what constitutes a typical case of with clinical atopic eczema answering negatively to the itching atopic eczema. question were simply not bothered enough by the symptom Recall of symptoms over a long period might have contri- either because it was not severe enough or because of compet- buted to an overall misclassification bias. Given the seasonal

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp111–116 Validation of U.K. diagnostic criteria for atopic eczema, D.A. Chalmers et al. 115 variation of atopic eczema, clinical examination is a reflection for large surveys where privacy cannot always be guaranteed. of disease at one point in time only, and may not be an accu- Fieldworkers can be trained in the sign using the U.K. diag- rate estimate as it may not identify disease which is in remis- nostic criteria online manual, where a set of 48 training pho- sion at the time of examination. This is less problematic in tographs are provided,13 as well as a quality control test darker skins as postinflammatory hyperpigmentary changes which is marked independently.21 Paradoxically, nonclinical would be visible. However, there was little difference in the fieldworkers might be more reliable at recording this sign validity of the criteria when symptoms over the last week only than doctors as they are more likely to follow the protocol were used as the qualifying criterion. exactly. The high response rate (over 97%) and the large sample size The visible flexural eczema sign is less prone to cultural fac- of around 3000 children enabled validity to be estimated with tors in determining its performance. Despite concerns that ery- good precision. All of those interviewing parents and carers thema may be missed in black skin,22 the photographic were trained beforehand. Care was taken to translate the original protocol for recording the sign of visible flexural eczema questionnaires by using suitable educated bilingual local people, includes examples of pigmented skins, with clear instructions with back-translation checks to ensure accuracy of that transla- on the importance of including surface change when erythema tion. Xhosa fieldworkers were deliberately chosen to conduct is not obvious. The clinicians who participated in this study this survey because of their sensitivity to cultural issues. The fact were also experienced in recognizing these skin lesions in a that only three dermatologists with similar views on what con- pigmented population. The fact that the visible flexural eczema stitutes atopic eczema were used in the study was also a sign performed well in this study reinforces the view that it is strength, as was the fact that they were blinded from the ques- suitable for use in black individuals. tionnaire responses. In the rural area, questionnaires and clinical It is tempting to suggest that criteria for atopic eczema examinations were performed on the same day. In the urban should be developed de novo for different countries, rather areas it was not always possible to schedule a clinician to be pre- than trying to adapt existing criteria. While there is some sent at the time of administering the questionnaire, and it is merit in this suggestion for within-country studies, such a possible that some children with intermittent disease might have suggestion is likely to turn the clock of atopic eczema been misclassified as their eczema was no longer active when research back 30 years, as we could end up with as many subsequently examined. However, this is unlikely given the criteria as there are countries—a move that would hamper postinflammatory changes which are often associated with atop- efforts at international standardization and comparisons. It is ic eczema in this population. also tempting on the basis of this study and another from Misclassification may have occurred as a result of informa- Iran to suggest that questions relating to atopic eczema tion provided by an unclassified carer and could have been simply cannot be translated.4 Yet, a number of other studies limited if the older children had been asked to complete the in diverse cultural, socioeconomic and language settings questionnaire themselves. Our fieldworkers faced other issues have demonstrated good validity after appropriate transla- such as fire, theft and hijacking during the survey, which led tions.3,5–8,10,11 No blanket rule exists, therefore, and further to some cases not being examined. studies in different study populations are needed to build The full U.K. criteria for atopic eczema cannot be recom- up a picture of the validity of the U.K. criteria in these mended as the primary outcome measure for estimating dis- diverse settings. ease prevalence in the Xhosa communities who participated in this study. Other measures such as clinical examination by a Acknowledgments dermatologist or objective measures such as recording the one sign of visible flexural eczema, which are less prone to transla- Special thanks must be given to the Wellcome Trust who pro- tion/cultural influences, should be used instead. vided funding for this study and to Professor Rodney Ehrlich Although clinical examination by a dermatologist might as supervisor for the Master of Public Health thesis that seem to be the most common sense method of estimating formed the basis of this paper. atopic eczema prevalence in all future studies, it does mean identifying a suitably qualified dermatologist with a standard- References ized concept of what constitutes the clinical phenotype of atopic eczema. The ability to compare different studies directly 1 Hanifin JM, Rajka G. Diagnostic features of atopic dermatitis. Acta will be affected. Derm Venereol (Stockh) 1980; 92 (Suppl.):44–7. On the other hand, visible flexural eczema is beginning to 2 Williams HC, Burney PG, Hay RJ et al. The UK Working Party’s Diagnostic Criteria for Atopic Dermatitis. I. Derivation of a mini- emerge as an international reference standard for atopic mum set of discriminators for atopic dermatitis. Br J Dermatol 1994; eczema, and has now been used in over 40 studies worldwide 131:383–96. including the International Study of Asthma and Allergies in 3 Popescu CM, Popescu R, Williams H et al. Community validation of Childhood Phase 2.21 In addition to the desirable goal of stan- the United Kingdom diagnostic criteria for atopic dermatitis in dardization, the sign is also easy to ascertain by nonclinical Romanian schoolchildren. Br J Dermatol 1998; 138:436–42. fieldworkers, taking less than 1 min to complete, and not 4 Firooz A, Davoudi SM, Farahmand AN et al. Validation of the diag- requiring children to remove their clothes—a big advantage nostic criteria for atopic dermatitis. Arch Dermatol 1999; 135:514–16.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp111–116 116 Validation of U.K. diagnostic criteria for atopic eczema, D.A. Chalmers et al.

5 Johnke H, Vach W, Norberg LA et al. A comparison between cri- www.nottingham.ac.uk/dermatology/eczema/index.html (accessed teria for diagnosing atopic eczema in infants. Br J Dermatol 2005; 29 August 2006). 153:352–8. 14 Weinstock MA. Validation of a diagnostic test. Epidemiological 6 Ortiz de Frutos FJ, Guerra Tapia A, Gomez de la Camara A et al. principles in dermatology. Arch Dermatol 1989; 125:1260–4. Validation of the Spanish version of the diagnostic questionnaire of 15 Williams HC, Burney PG, Hay RJ et al. The UK Working Party’s the Working Group on Atopic Dermatitis of the United Kingdom. Diagnostic Criteria for Atopic Dermatitis. I. Derivation of a mini- Rev Clin Esp 1998; 198:424–8 (in Spanish). mum set of discriminators for atopic dermatitis. Br J Dermatol 1994; 7 Gu H, Chen XS, Chen K et al. Evaluation of diagnostic criteria for 131:383–96. atopic dermatitis: validity of the criteria of Williams et al. in a hos- 16 Pekkanen J, Pearce N. Defining asthma in epidemiological studies. pital-based setting. Br J Dermatol 2001; 145:428–33. Eur J Respir 1999; 14:951–7. 8 Mohrenschlager M, Schafer T, Williams HC et al. Ubersetzung} und 17 Flohr C, Johansson SG, Wahlgren CF, Williams H. How atopic is Validierung der Britischen Diagnosektriterien in Augsberg. Allergo J atopic dermatitis? J Allergy Clin Immunol 2004; 114:150–8. 1998; 7:373–7. 18 Jelsma J, Mkoka S, Amosun L et al. The reliability and validity 9 Foley P, Zuo Y, Plunkett A, Marks R. The frequency of common of the Xhosa translation of the EQ-5D. Disabil Rehab 2004; 26: skin conditions in preschool-age children in Australia: atopic 103–8. dermatitis. Arch Dermatol 2001; 137:293–300. 19 Sperber AD. Translation and validation of study instruments for 10 Girolomoni G, Abeni D, Masini C et al. The epidemiology of atopic cross cultural research. Gastroenterol 2004; 126:S124–8. dermatitis in Italian schoolchildren. Allergy 2003; 58:420–5. 20 Maneersriwongol W, Dixon JK. Instrument translation process: a 11 Haileamlak A, Lewis SA, Britton J et al. Validation of the Interna- methods review. J Adv Nurs 2004; 48:175–86. tional Study of Asthma and Allergies in Children (ISAAC) and U.K. 21 International Study of Asthma and Allergies in Childhood Phase 2, criteria for atopic eczema in Ethiopian children. Br J Dermatol 2005; online reference: http://isaac.auckland.ac.nz/Phasetwo/Phs2Frame. 152:735–41. html (accessed 8 November 2006). 12 Byarugaba J. The impact of urbanization on the health of black 22 Ben-Gashir MA, Hay RJ. Reliance on erythema scores may mask pre-school children in the Umtata district, Transkei, 1990. S Afr severe atopic dermatitis in black children compared with their Med J 1991; 79:444–8. white counterparts. Br J Dermatol 2002; 147:920–5. 13 Williams HC, Flohr C. ‘So how do I define atopic eczema? A practical manual for researchers wishing to define atopic eczema’, online reference: http://

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp111–116 PHOTOBIOLOGY DOI 10.1111/j.1365-2133.2006.07569.x Photodegradation of folic acid during extracorporeal photopheresis M. Der-Petrossian, M. Fo¨dinger,* R. Knobler, H. Ho¨nigsmann and F. Trautinger Division of Special and Environmental Dermatology, Department of Dermatology, and *Department of Medical and Chemical Laboratory Diagnostics, Medical Uni- versity of Vienna, Waehringer Guertel 18-20, 1090 Vienna, Austria

Summary

Correspondence Background Photodegradation of folic acid (FA) by ultraviolet (UV) radiation is a Manon Der-Petrossian. well-documented photochemical reaction, and decreased serum levels of FA have E-mail: [email protected] been found in patients receiving photochemotherapy (psoralen plus UVA). Dur- ing extracorporeal photopheresis (ECP) leucocytes and plasma are subjected to 8- Accepted for publication 14 July 2006 methoxypsoralen (8-MOP) plus UVA. Objectives To investigate whether ECP leads to the photodegradation of FA in the Key words extracorporeal system. extracorporeal photopheresis, folic acid, Methods In 30 patients undergoing ECP on two consecutive days the FA levels photodegradation, ultraviolet A were measured in the extracorporeal collected plasma prior to and after UVA ex- Conflicts of interest posure. Healthy donor plasma was exposed to 8-MOP and increasing doses of None declared. UVA in vitro. In five patients serum folate levels were determined before and after ECP. Results We found a mean reduction of 44% and 46% on the first and second day of treatment, respectively. This effect could be reproduced in vitro: the irradiation of healthy donor plasma with UVA led to a dose-dependent reduction of FA of ) up to 54Æ75% at 16 J cm 2. This was independent of the presence of 8-MOP and the base concentration of 5-methyltetrahydrofolate; minimal changes were

observed for vitamin B12 and homocysteine, not undergoing photodegradation. Serum folate levels did not change signiﬁcantly before and after ECP. Conclusions We conclude that extracorporeal exposure of plasma to UVA during ECP leads to photodegradation of FA. Further investigations are required to determine the biological effects of folate photoproducts and whether clinically relevant loss of FA might be a consequence of ECP.

Photodegradation of folic acid (FA) by ultraviolet (UV) radi- Epidemiological data have shown that the prevalence of NTD ation is a well-documented photochemical reaction. FA is a is higher in light-skinned people, supporting the hypothesis photosensitive compound that undergoes photolysis by UV. that dark skin protects the body’s folate stores from destruc- Recent studies have investigated the photoproducts of this tion.13–15 In fair skin types photolysis of FA with following reaction, showing that the photodegradation of FA is divided folate deficiency is presumed to lead to increased risk of NTD. into three phases leading to the formation of specific photo- A correlation between NTD and sunbed exposures has been products such as p-aminobenzoyl-L-glutamic acid, 6-formyl described by Lapunzina. He reports three otherwise healthy pterin and pterin-6-carboxylic acid.1–4 women using sunbeds for tanning in the first weeks of preg- Folate, the conjugated form of FA, is an essential vitamin nancy and each giving birth to an infant with NTD.16 It has for the formation of purines and pyrimidines for DNA and been suggested that degradation of folate in vivo played a role RNA synthesis. Deficiency of folate leads to macrocytic in the evolution of mankind. Jablonski and Chaplin hypothes- anaemia, and is associated with an increased risk of cardiovas- ized that dark skin has been favoured in environments with cular diseases and carcinogenesis.5–9 Furthermore, folate significant UV radiation because pigmentation would, by pro- deficiency in pregnant women has clearly been shown to be tecting against folate photolysis, prevent NTD.17,18 Further- related to neural tube defects (NTD), such as spina bifida.10–12 more, a correlation between folate deficiency and impaired

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp117–121 117 118 Photodegradation of folic acid, M. Der-Petrossian et al. spermatogenesis has also been reported, supporting the hy- 800; Waldmann, Villingen-Schwenningen, Germany). Energy pothesis of Jablonski and Chaplin.19 output of the lamp was measured with a Waldmann UV- ) Photodegradation of FA has also been observed in photo- meter. The applied dose was 2 J cm 2. All experiments were therapy. In the late 1970s Branda and Eaton reported light- done in the dark and sham-treated samples with and without skinned patients with different skin disorders undergoing photo- 8-MOP/5-MTHF were used as controls. chemotherapy (psoralen plus UVA; PUVA) for 3 months 5-MTHF and vitamin B12 were determined simultaneously who had significantly lower serum folate levels than healthy by radioassay (Simultrac-SNB Radioassay Kit Vitamin B12 controls. They also showed a 30–50% loss of folate in human [57Co]/Folate [125I]; Becton-Dickinson, Franklin Lakes, NJ, plasma after in vitro exposure to simulated sunlight.20 How- U.S.A.). Homocysteine was measured with a fluorescence ever, a study on healthy volunteers receiving only UVA 2 days polarization immunoassay (IMx analyzer; Abbott Laboratories, per week for 3 weeks showed no difference in the serum fo- Abbott Park, IL, U.S.A.). Levels of vitamin B12 and homocyste- late levels before and after UVA treatment.21 Another form of ine, not undergoing photodegradation, were used as controls phototherapy, where plasma is directly exposed to UV, is for non-UV-specific effects. extracorporeal photopheresis (ECP). ECP is a well-established treatment modality for different disorders including Se´zary Statistics syndrome, systemic sclerosis, graft-versus-host disease (GvHD) and others. During ECP a fraction of leucocytes and plasma is The paired t-test was used to calculate statistical significance subjected extracorporeally to 8-methoxypsoralen (8-MOP) for the comparison of FA, vitamin B12 and homocysteine lev- plus UVA and given back to the patient. Research on the els before and after UVA. mechanisms of action of ECP is mainly focused on the effects of 8-MOP/UVA on the treated lymphocytes, whereas informa- Results tion on the photochemistry of irradiated plasma is lacking. Due to the known photochemistry of FA we assumed that ECP On the first day of ECP mean ± SD FA levels decreased ) might lead to photodegradation of this essential component of from 7Æ11 ± 5Æ8to3Æ9±2Æ6 nmol L 1 (P <0Æ005) after human plasma. To address this question we measured the UVA exposure. On the second day the reduction was from ) plasma levels of FA prior to and after extracorporeal UVA ex- 5Æ9±4Æ9to3Æ4±3Æ0 nmol L 1 (P <0Æ005). The mean posure during ECP. reduction after UVA was 44% and 46%, respectively. In

contrast, the mean ± SD vitamin B12 levels decreased slightly on the first day from 259Æ3 ± 405Æ1to Materials and methods ) 203Æ4 ± 293Æ9 pmol L 1 (P ¼ 0Æ14) and on the second day ) Thirty patients undergoing ECP on two consecutive days were from 210Æ2 ± 287Æ5 to 196Æ8 ± 259 pmol L 1 (P ¼ 0Æ025) enrolled into the study. Written informed consent was after UVA exposure. Here the mean reduction was 13% and obtained from all patients. The study protocol was approved 6%, respectively. The mean ± SD homocysteine levels by the Ethics Committee of the Medical University of Vienna. showed almost no change: from 5Æ17 ± 1Æ95 to ) Of these patients 11 had systemic sclerosis, eight GvHD, five 5Æ08 ± 1Æ95 lmol L 1 (P ¼ 0Æ44) on the first day and from ) cutaneous T-cell lymphoma, four were after lung transplant- 4Æ81 ± 1Æ6to4Æ65 ± 1Æ65 lmol L 1 (P ¼ 0Æ05) on the ation, one had erosive lichen planus and one nephrogenic fi- second day of ECP. The mean reduction was 1% and 3%, brosing dermopathy. Exclusion criteria were serum folate respectively (Fig. 1). deficiency and folate supplementation. Mean ± SD serum folate levels did not change significantly Heparinized plasma was taken from the extracorporeal sys- before and after ECP on two consecutive days. On the first day ) ) tem before and immediately after UVA radiation (2 J cm 2) levels were 11Æ16 ± 4Æ55 and 11Æ10 ± 4Æ34 nmol L 1, respect- on both treatment days. In addition, in five patients serum ively. On the second day folate levels were 12Æ06 ± 6Æ65 and ) samples were also taken before and after ECP. Samples 11Æ20 ± 5Æ54 nmol L 1, respectively (Fig. 2). Homocysteine were protected from light and kept at )70 C until further levels showed a slight but nonsignificant reduction on both days ) analysis. after ECP (from 10Æ02 ± 2Æ13 to 8Æ82 ± 2Æ15 lmol L 1 and ) For in vitro experiments healthy donor plasma was collected from 9Æ18 ± 1Æ89 to 8Æ80 ± 2Æ00 lmol L 1). from heparinized blood and was diluted 1 : 2Æ5 with NaCl The results of the in vitro experiments are shown in 0Æ9% corresponding to the dilution of the extracorporeal Table 1. After 8-MOP/UVA irradiation of plasma the FA plasma during ECP. The diluted samples were supplemented levels showed a UVA dose-dependent mean reduction to ) ) with 5-methyltetrahydrofolate (5-MTHF, 5-methyltetrahydro- 90Æ5% (2 J cm 2), 74Æ5% (8 J cm 2) and 54Æ75% ) folic acid disodium salt; Sigma-Aldrich, St Louis, MO, U.S.A.), (16 J cm 2), compared with control. This effect was inde- the most common form of FA in human plasma. 8-MOP pendent of the baseline concentration of 5-MTHF. The vita- )1 (Gerot, Vienna, Austria) 100 ng mL was added according to min B12 levels showed a mean reduction to 89Æ25% ) ) ) the ECP protocol. The samples were irradiated in Petri dishes (2 J cm 2), 89Æ5% (8 J cm 2) and 77Æ25% (16 J cm 2). with a Waldmann PUVA 800 irradiation unit equipped with These results were independent of the presence of 8-MOP Philips TKL 40 W/09N lamps emitting at 315–400 nm (UV during irradiation.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp117–121 Photodegradation of folic acid, M. Der-Petrossian et al. 119

greatly exceeded that of vitamin B12 and homocysteine 140% changes, indicating a UV-speciﬁc effect. Further to investigate 120% First day Second day UV dose dependency and the role of 8-MOP in FA photodeg- 100% radation we performed in vitro experiments with plasma from 80% healthy donors. Conﬁrming previous results, we found that 20 60% the effect is UV dose dependent in vitro. In addition, we

% of control showed that the initial concentration of 5-MTHF has no influ- 40% ence on the extent of the reaction. Earlier observations in 20% patients treated with PUVA and our results in ECP raise the 0% question of whether photoactivated 8-MOP might be involved Folic Vitamin Homocysteine Folic Vitamin Homocysteine acid B12 acid B12 in FA photochemistry; we show here that – at least at pharma- cological concentrations – this is not the case. Fig 1. Photodegradation of folic acid during extracorporeal Loss of FA leads to megaloblastic anaemia and is associated photopheresis (ECP). Heparinized plasma of 30 patients undergoing with NTD and other disorders. Our results might raise con- ECP on two consecutive days was taken from the extracorporeal cerns that these conditions could be a consequence of ECP and system prior to and after ultraviolet (UV) A exposure. Folic acid and possibly other forms of phototherapy. However, at least for vitamin B levels were determined simultaneously by radioassay. The 12 ECP clinically manifest FA deficiency is highly unlikely to mean ± SD reduction of FA after UVA was 44 ± 28% and 46 ± 25% occur. It has been estimated that during a typical treatment on the first and second day, respectively. Vitamin B12 showed a mean ± SD reduction of 13 ± 30% and 6 ± 21%, respectively. cycle of ECP 5–10% of circulating leucocytes are separated 22 Homocysteine was measured with a fluorescence polarization and treated extracorporeally. Assuming a similar percentage immunoassay. The mean ± SD reduction of homocysteine after UVA for plasma and a reduction of FA by about 46%, as observed exposure was 1 ± 16% and 3 ± 8%, respectively. in our study, a loss of only 4Æ6% of total plasma FA can be calculated. This reduction would not be detectable by an assay system that has an interassay variability of 4–5%, as used in 25 our study. Accordingly, in five of our patients we could not find any changes in the serum folate levels before and after

) 20

–1 ECP. However, we cannot exclude the development of folate deficiency in patients under long-term treatment with ECP. 15 This has to be confirmed in prospective, long-term, clinical trials. 10 In the present study we did not analyse red blood cell (RBC) folate levels that are claimed to represent a better meas- Folic acid (nmol L 5 ure of folate tissue stores. This is due to the fact that RBCs incorporate folate as they are formed. Thereafter, RBC folate 0 Day 1/before Day 1/after Day 2/before Day 2/after levels remain constant throughout the life span of the cell and ECP ECP ECP ECP are less sensitive to dietary and other exogenous effects than serum folate levels.23 In contrast, serum folate levels decrease Fig 2. Serum folate levels in five patients before and after within a few days upon dietary folate restriction although tis- extracorporeal photopheresis (ECP). Serum samples were taken before sue stores may be normal.24 Because we have measured folate and after ECP on both treatment days. Folic acid was determined by levels on two consecutive days before and after ECP, changes radioassay. Serum folate levels did not change significantly before and of RBC folate levels were very unlikely to occur and thus were after ECP. On the first day of ECP the mean ± SD folate levels were ) not analysed in our study. The assumption of no changes in 11Æ16 ± 4Æ55 and 11Æ10 ± 4Æ34 nmol L 1, and on the second day ) 12Æ06 ± 6Æ65 and 11Æ20 ± 5Æ54 nmol L 1, respectively. RBC folate concentrations is confirmed by our observation of no increase in serum total homocysteine level, a functional marker of (tissue) folate deficiency. In addition, as RBCs, a Discussion major reservoir of folate, are not irradiated in ECP it is highly unlikely that ECP will result in clinically relevant or even Our study shows that UVA irradiation of extracorporeal measurable (tissue) FA deficiency. plasma during ECP leads to photodegradation of FA. We Earlier studies have addressed the question of whether showed that on both days of ECP FA levels decreased signifi- phototherapy of the skin leads to FA reduction. In one clin- cantly by up to 46% in the extracorporeal system. To exclude ical observation 10 light-skinned patients under PUVA treat- that this reduction might be due to any effect independent of ment were found to have lower serum folate levels 20 UV, such as dilution or adhesion to plastic, vitamin B12 and following treatment compared with healthy controls. homocysteine were measured simultaneously. We found However, this study suffers from major methodological minor changes in these factors, indicating that nonspecific problems, e.g. folate levels were not measured before PUVA effects might occur. However, the magnitude of FA reduction therapy, and small sample size.25 Another study with 16

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp117–121 120 Photodegradation of folic acid, M. Der-Petrossian et al.

Table 1 Folic acid and vitamin B12 levels in healthy donor plasma irradiated with ultraviolet (UV) A, with or without 8-methoxypsoralen (8-MOP)

)2 UVA (J cm ), Folic acid Vitamin B12 Folic acid Vitamin B12 5-MTHFa with or without (% of control (% of control (% of control (% of control ) (ng mL 1) 8-MOP with 8-MOP) with 8-MOP) without 8-MOP) without 8-MOP) 0 Control 100 100 100 100 2 91938183 8 64795670 16 55 82 55 104 2 Control 100 100 100 100 2 87938584 8 67905775 16 50 85 55 88 4 Control 100 100 100 100 2 968886101 8 92 101 60 87 16 67 66 56 88 8 Control 100 100 100 100 2 888391109 8 758865103 16 47 76 52 123

5-MTHF, 5-methyltetrahydrofolate. a5-MTHF supplementation to healthy donor plasma (diluted 1 : 2Æ5 with NaCl 0Æ9%).

healthy volunteers receiving six exposures to UVA during a sitization in established in vitro and in vivo models of the period of 3 weeks showed no difference in folate status immunomodulatory effects of ECP.29,30 before and after treatment.21 It cannot be excluded that this study missed UV-dependent alterations in folate species and References effects of more intense exposure.26 The alerting observation of NTD in three newborns of mothers who used sunbeds 1 Akhtar MJ, Khan MA, Ahmad I. High performance liquid chroma- in early pregnancy might provide evidence that UVA – in tographic determination of folic acid and its photodegradation addition to causing photoageing and photocarcinogenesis – products in the presence of riboflavin. J Pharm Biomed Anal 1997; 16:95–9. is not safe with regard to FA deficiency.16 Taken together, 2 Akhtar MJ, Khan MA, Ahmad I. Photodegradation of folic acid in the pertinent information on the effect of tanning parlour aqueous solution. J Pharm Biomed Anal 1999; 19:269–75. use and phototherapy on FA stores is insufficient and war- 3 Akhtar MJ, Khan MA, Ahmad I. Identification of photoproducts of rants further investigations. folic acid and its degradation pathways in aqueous solution. J Pharm A consequence of FA photodegradation is not only FA loss Biomed Anal 2003; 31:579–88. but also the appearance of specific photolysis products such as 4 Off MK, Steindal AE, Porojnicu AC et al. Ultraviolet photodegradation of folic acid. J Photochem Photobiol B 2005; 80:47–55. p-aminobenzoyl-L-glutamic acid, 6-formyl pterin and pterin-6- 5 Moat SJ, Lang D, McDowell IF et al. Folate, homocysteine, endo- carboxylic acid. The photochemistry of FA has been thor- 1–4 thelial function and cardiovascular disease. J Nutr Biochem 2004; oughly investigated in vitro in aqueous solutions. It has been 15:64–79. described that the resulting pterins also act as photosensitizers 6 Mason JB, Choi SW. Folate and carcinogenesis: developing a unify- and can accelerate FA photodegradation and induce DNA ing hypothesis. Adv Enzyme Regul 2000; 40:127–41. photo-oxidation.4,27 Determination of low amounts of FA 7 Prinz-Langenohl R, Fohr I, Pietrzik K. Beneficial role for folate in photoproducts in biological samples is technically difficult and the prevention of colorectal and breast cancer. Eur J Nutr 2001; is not yet established for human plasma.28 Thus further 40:98–105. 8 Potter JD. Methyl supply, methyl metabolizing enzymes and colo- research will be necessary to develop sensitive methods for rectal neoplasia. J Nutr 2002; 132:S2410–12. the quantitative analysis of FA photoproducts in human body 9 Giovannucci E. Epidemiologic studies of folate and colorectal neo- fluids and tissue samples. In addition, these molecules have to plasia: a review. J Nutr 2002; 132:S2350–5. be investigated for their contribution to the clinical and bio- 10 Bower C, Stanley FJ. Dietary folate as a risk factor for neural-tube logical effects of UV radiation. The results described here pro- defects: evidence from a case–control study in Western Australia. vide for the first time evidence that photochemical changes Med J Aust 1989; 150:613–19. occur in human plasma to a significant extent during ECP. It 11 MRC Vitamin Study Research Group. Prevention of neural tube defects: results of the Medical Research Council Vitamin Study. Lan- is likely that besides FA other photosensitizers exist in human cet 1991; 338:131–7. plasma that might be biologically relevant. The aim of follow- 12 Mitchell LE. Epidemiology of neural tube defects. Am J Med Genet C ing studies will be to investigate the role of plasma photosen- Semin Med Genet 2005; 135:88–94.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp117–121 Photodegradation of folic acid, M. Der-Petrossian et al. 121

13 Grace HJ. Prenatal screening for neural tube defects in South 23 Snow CF. Laboratory diagnosis of vitamin B12 and folate defici- Africa: an assessment. S Afr Med J 1981; 60:324–9. ency. Arch Intern Med 1999; 159:1289–98. 14 Khoury MJ, Erickson JD, James LM. Etiologic heterogeneity of 24 Herbert V. Experimental nutritional folate deficiency in man. Trans neural tube defects: clues from epidemiology. Am J Epidemiol 1982; Assoc Am Physicians 1962; 75:307–20. 115:538–48. 25 Cohn BA. Sunlight, skin color, and folic acid. J Am Acad Dermatol 15 Stevenson RE, Allen WP, Pai GS et al. Decline in prevalence of 2002; 46:317–18. neural tube defects in a high-risk region of the United States. Pedi- 26 Lucock M, Yates Z, Glanville T et al. A critical role for B-vitamin atrics 2000; 106:677–83. nutrition in human developmental and evolutionary biology. Nutr 16 Lapunzina P. Ultraviolet light-related neural tube defects? Am J Med Res 2003; 23:1463–75. Genet 1996; 67:106. 27 Hirakawa K, Suzuki H, Oikawa S, Kawanishi S. Sequence-specific 17 Jablonski NG, Chaplin G. The evolution of human skin coloration. DNA damage induced by ultraviolet A-irradiated folic acid via its J Hum Evol 2000; 39:57–106. photolysis product. Arch Biochem Biophys 2003; 410:261–8. 18 Jablonski NG, Chaplin G. Skin deep. Sci Am 2002; 287:74–81. 28 Han F, Huynh BH, Shi H et al. Pteridine analysis in urine by capil- 19 Wong WY, Merkus HM, Thomas CM et al. Effects of folic acid and lary electrophoresis using laser-induced fluorescence detection. Anal zinc sulfate on male factor subfertility: a double-blind, random- Chem 1999; 71:1265–9. ized, placebo-controlled trial. Fertil Steril 2002; 77:491–8. 29 Klosner G, Trautinger F, Knobler R et al. Treatment of peripheral 20 Branda RF, Eaton JW. Skin color and nutrient photolysis: an evolu- blood mononuclear cells with 8-methoxypsoralen plus ultraviolet tionary hypothesis. Science 1978; 201:625–6. A radiation induces a shift in cytokine expression from a Th1 to 21 Gambichler T, Bader A, Sauermann K et al. Serum folate levels after Th2 response. J Invest Dermatol 2001; 116:459–62. UVA exposure: a two-group parallel randomised controlled trial. 30 Maeda A, Schwarz A, Kernbeck K et al. Intravenous infusion of syn- BMC Dermatol 2001; 1:8. geneic apoptotic cells by photopheresis induces antigen-specific 22 Gasparro FP. Extracorporeal Photochemotherapy: Clinical Aspects and the regulatory T cells. J Immunol 2005; 174:5968–76. Molecular Basis for Efficacy. Georgetown, TX: R.G. Landes Company, 1994.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp117–121 PHOTOBIOLOGY DOI 10.1111/j.1365-2133.2006.07584.x Effects of psoralen plus ultraviolet A irradiation on cultured epidermal cells in vitro and patients with vitiligo in vivo C-S. Wu, C-C.E. Lan,* L-F. Wang, G-S. Chen,* C-S. Wu* and H-S. Yu* Faculty of Biomedical Laboratory Science, College of Health Science and *Department of Dermatology, College of Medicine, Kaohsiung Medical University, Kao- hsiung, Taiwan Department of Dermatology, National Taiwan University Hospital and National Taiwan University College of Medicine, No. 7 Chung-San South Road, Taipei, Taiwan Summary

Correspondence Background Both psoralen plus ultraviolet (UV) A (PUVA) and narrowband UVB Hsin-Su Yu. (NB-UVB) irradiation are effective treatments for vitiligo vulgaris. However, the E-mail: [email protected] mechanisms of PUVA and NB-UVB in repigmentation are not thoroughly clarified. Our previous results showed that NB-UVB irradiation directly promotes Accepted for publication 23 July 2006 melanocyte (MC) migration and stimulates MC proliferation via keratinocytes (KCs). Key words Objectives In the present study, we used NB-UVB as a reference for comparison to keratinocyte, melanocyte, psoralen plus ultraviolet A, investigate the immediate effects of PUVA on MC proliferation and migration. repigmentation, vitiligo Methods Cultured MCs and KCs were treated with PUVA or irradiated with NB- Conflicts of interest UVB. The direct impact of PUVA treatment on MCs was assessed in terms of its None declared. effect on MC proliferation and migration. The indirect effect of PUVA treatment and NB-UVB irradiation on MC proliferation via KCs was also investigated. The activities of matrix metalloproteinase (MMP)-2 and MMP-9, known for their influence on cell migration, were evaluated in the PUVA-treated MC and KC supernatants. The concentrations of MC mitogens/growth factors in the PUVA-treated KC supernatants were also determined. In addition, the serum levels of MC mitogens/growth factors in healthy controls, in patients with active vitiligo and in patients with repigmenting vitiligo after PUVA treatment were determined to elucidate the mechanisms of how PUVA induces vitiligo repigmentation in vivo. Results Our results demonstrated that PUVA treatment did not significantly stimulate the release of MC mitogens/growth factors from KCs. The migration of MCs was also not enhanced after PUVA treatment. The expression of MMP-2 activity in supernatants derived from PUVA-treated MCs was significantly increased as compared with the control group. However, neither MMP-2 nor MMP-9 activity in KC supernatants was stimulated by PUVA treatment. In contrast to NB-UVB, immediate effects of PUVA on MC proliferation and migration were not observed in this study. Sera from patients with repigmenting vitiligo after PUVA treatment contained higher levels of basic fibroblast growth factor, stem cell factor and hepatocyte growth factor as compared with healthy controls and patients with active vitiligo. Conclusions Our results indicate that in addition to immune suppression, PUVA treatment creates a favourable milieu for promoting the growth of MCs in patients with vitiligo instead of directly stimulating the regrowth of MCs. Based on our results, we propose that in the active stage of vitiligo, PUVA treatment is the therapy of choice to slow down the destruction of MCs and to create a favourable environment for MCs to survive. In the stable stage of vitiligo, NB- UVB irradiation should be used to stimulate the proliferation and migration of MCs directly.

2006 The Authors 122 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp122–129 Effects of PUVA on epidermal cells and vitiligo, C-S. Wu et al. 123

) ) Functional melanocytes (MCs) in patients with vitiligo disap- extract (BPE), 1Æ5 lgmL 1 bovine insulin, 10 lgmL 1 ) ) ) pear from the involved skin by a mechanism(s) that has yet to bovine transferrin, 10 7 mol L 1 hydrocortisone, 3 lgmL 1 ) be identified. This depigmentary disease affects all races and heparin and 10 ng mL 1 phorbol 12-myristate 13-acetate occurs in 0Æ3–1Æ0% of the world population.1,2 The underly- (PMA). The BPE, bovine insulin, bovine transferrin, hydrocor- ing mechanism of how functional MCs disappear from the tisone, heparin and PMA were purchased from Sigma. The involved skin is still unclear. As the pathogenesis of this dis- epidermal cell suspension was plated on to 10-cm2 Petri ease remains obscure, treatment of vitiligo has generally been dishes, and incubated at 37 C in a humidified atmosphere unsatisfactory and often disappointing. Regardless of thera- containing 5% CO2. For KC culture, the epidermal cells were peutic modality employed, recovery from vitiligo is initiated pelletted by centrifugation (500 g, 10 min) and dispersed into by the activation3 and proliferation4 of immature pigment individual cells by repeated aspiration with a pipette. The cells cells, followed by the upward migration to the nearby epider- were gently resuspended in 5 mL of keratinocyte-SFM med- ) ) mis to form perifollicular pigmentation islands and the down- ium (Gibco) containing 25 lgmL 1 BPE and 5 ng mL 1 re- ward migration to the hair matrices to produce melanin.5 combinant human epidermal growth factor. Twenty-four Other mechanisms proposed to participate in the vitiligo hours after primary seeding, the medium was changed, and recovery process include migration of normal MCs from the then changed regularly every 2 days. About 7–10 days after adjacent normal skin to the affected site.6 primary seeding, the semiconfluent cells were incubated with Both psoralen plus ultraviolet (UV) A (PUVA) and narrow- 0Æ25% trypsin in 0Æ01% ethylenediamine tetraacetic acid solu- band UVB (NB-UVB) irradiation are well known for their ef- tion (Gibco) at 37 C for 3–5 min, harvested with soybean fectiveness in treating patients with vitiligo.7–11 Despite their solution (Gibco), centrifuged (500 g, 10 min), and the cell clinical efficacy, the underlying mechanisms of how PUVA pellet was resuspended in culture medium and reincubated. and NB-UVB irradiation induce repigmentation in vitiligo have The second- or third-passage cells were used in the following not been thoroughly elucidated. In clinical settings, although experiments. both PUVA and NB-UVB are effective for vitiligo, there are intrinsic differences between the two modalities. More specif- Treatment of melanocytes and keratinocytes with ically, the numbers and durations of treatments required psoralen plus ultraviolet A or narrowband ultraviolet B before repigmentation occurred were lower with NB-UVB than with PUVA.7,8,10,12,13 We recently clarified the underly- Cultured human MCs and KCs were either treated with PUVA ing mechanisms of how NB-UVB induces repigmentation in or irradiated with NB-UVB. In clinical phototherapy, the min- vitiligo. Our previous study demonstrated that NB-UVB imal erythema dose is used to determine the initial dose of directly promotes the proliferation and migration of MCs.14 UVA/UVB applied. To apply this principle in vitro, we used the Despite our previous results from NB-UVB irradiation, there sublethal dose as the treatment dose in our experiments. For are few reports investigating the mechanisms of how PUVA PUVA treatment,16 MCs and KCs were incubated with 8-meth- ) ) induces repigmentation in patients with vitiligo. In the present oxypsoralen (8-MOP; Sigma; 5 · 10 7 mol L 1 in medium) study, by using NB-UVB irradiation as a reference for com- at 37 C for 2 h in the dark. Then the dishes were washed parison, we set out to investigate the immediate effects of and rinsed with phosphate-buffered saline (PBS); the covers of PUVA on MC proliferation and migration in vitro. In addition, the dishes were opened; and the cells were exposed to various ) to elucidate the mechanisms of how PUVA induces vitiligo re- doses of UVA (1 · 15 W, intensity 0Æ75 mW cm 2) irradi- ) pigmentation in vivo, serum samples from patients with vitiligo ation (0, 30, 60, 120 and 240 mJ cm 2 for MCs; 0, 100, ) were analysed. 200, 400 and 800 mJ cm 2 for KCs). MCs and KCs were ) exposed to NB-UVB 25 mJ cm 2 (1 · 15 W, intensity )2 Materials and methods 1Æ28 mW cm ) irradiation as a control (NB-UVB control) for comparison. The doses of irradiation were measured by a UVX radiometer (UVP, San Gabriel, CA, U.S.A.). Melanocyte and keratinocyte culture

The MCs and keratinocytes (KCs) used in our studies were XTT assay for cell viability obtained from normal human foreskins. Three samples from different adults donors undergoing routine circumcisions were A colorimetric method for determining the activity of mitoch- cleaned of excess subcutaneous tissue, cut into small pieces ondrial enzymes was used to detect viability of MCs and KCs (5 · 5 mm), and incubated with 0Æ25% trypsin (Gibco, Grand after PUVA treatment. The CellTiter cell proliferation assay kit Island, NY, U.S.A.) at 4 C overnight. The epidermal sheets (XTT assay) was purchased from Promega (Madison, WI, were separated from dermis using ﬁne forceps, and the epi- U.S.A.). MCs and KCs were seeded on 96-well plates at a den- dermal cell suspension was prepared as previously described.15 sity of 3 · 104 cells/well and incubated overnight. Then the For MC culture, the cells were pelleted by centrifugation cells were treated with PUVA or irradiated with NB-UVB and (500 g, 10 min) and resuspended in serum-free medium con- incubated for 24 h. Ten microlitres of CellTiter reagent containing MC basal medium MCDB 153 (Sigma, St Louis, MO, taining tetrazolium compound and electron coupling reagents ) U.S.A.) supplemented with 30 lgmL 1 bovine pituitary was added into each well to be catalysed by mitochondrial

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp122–129 124 Effects of PUVA on epidermal cells and vitiligo, C-S. Wu et al. dehydrogenase enzymes in metabolically active cells. The were soaked for 2 h in 2Æ5% Triton-X 100 solution with four plates were incubated for 4 h and the colorimetric absorbance washings to remove the SDS. The gels were incubated for ) was recorded at 490 nm by a microplate reader (Molecular 18 h at 37 C in buffer containing 50 mmol L 1 Tris/HCl, 17 )1 )1 Devices, Sunnyvale, CA, U.S.A.). pH 7Æ6, 150 mmol L NaCl, 10 mmol L CaCl2 and 0Æ02%

NaN3. Afterwards, the gels were stained with 0Æ2% Coomassie blue, destained in destaining solution (acetic acid/methanol/ Cell proliferation assay distilled water 1 : 4 : 5), and dried. The gels were recorded MCs were seeded on 24-well plates at a density of and analysed on a digital imaging system (Alpha Imager 4 · 104 cells/well and incubated overnight. Then MC growth 2000; Alpha Innotech Corp., San Leandro, CA, U.S.A.). supplements were deprived of culture media for 24 h. Then MCs were treated with PUVA/NB-UVB or incubated with Selection of patients 10% KC-conditioned media collected from PUVA-treated or NB-UVB-irradiated KCs and incubated for 6 days. Then the Thirty-three patients with active vitiligo were enrolled in our ) cells were washed with PBS and harvested by gentle trypsini- study. Each patient was treated with oral 8-MOP 0Æ6mgkg 1 zation, followed by mixing with an equal volume of trypan twice weekly followed 2 h later by UVA exposure. Only 15 blue. The number of living cells was counted with a haemo- patients had clinical signs of repigmentation of some lesions cytometer. at the end of the study period. The mean ± SD number of PUVA treatments required to induce repigmentation was 38Æ6±20Æ1. Serum samples were obtained from the 15 Melanocyte migration assay patients with repigmenting vitiligo (six females and nine CELLocate slips (Eppendorf, Hamburg, Germany) were pre- males; age range 10–48 years, mean ± SD 20Æ9±12Æ3), from ) coated with 10 lgmL 1 type I collagen, and incubated at 12 age- and sex-matched healthy volunteers selected for com- 37 C for 2 h. MCs were seeded on CELLocate coverslips in a parison and from 15 patients with vitiligo in the active stage 24-well culture plate at a density of 1 · 104 cells/well and in- (lesions displaying progression in the previous 3 months).20 cubated for 2 h. The cells were treated with PUVA or NB- All the studied subjects were cleared of the possibility of asso- UVB. For measurement of migration distance, we first selected ciated systemic diseases such as thyroid disease (Hashimoto’s and photographed a field on the CELLocate slip. Then photo- thyroiditis and Graves’ disease), Addison’s disease, pernicious graphs of the same field were taken at 0, 4, 8, 12, 16, 20 and anaemia, insulin-dependent diabetes mellitus and alopecia ar- 24 h after stimulation. Linear measurements of migration dis- eata. None of the participants had received any medical treat- tance were made using cell nuclei as the reference points. Ten ment in the preceding 3 months. Serum samples were stored to twelve cells for each group were randomly selected. The at )70 C. The study was approved by the Ethics Committee migration distances for every selected MC in each indicated of the Kaohsiung Medical University Hospital. time interval (i.e. 0–4, 4–8, 8–12, 12–16, 16–20 and 20– 24 h) were measured with a ruler. The summation of distan- Measurements of melanocyte mitogens/growth factors in ces for each indicated point represented the random migration sera and keratinocyte supernatants by enzyme-linked distance of MCs in 24 h. As the photographs taken under the immunosorbent assay microscope were magnified, conversion from ‘measured’ distance to ‘actual’ distance was required. According to the man- The concentrations of MC mitogens/growth factors in sera col- ual, every side of the square on the CELLocate slip measures lected from healthy individuals and from patients with active 175 lm. Therefore, a conversion ratio was established by and repigmenting vitiligo and in the culture supernatants ) measuring the side of the square (175 lm) on the photograph derived from 1 · 106 mL 1 KCs at 24 h after treatment and equating this distance to 175 lm, the actual distance on with PUVA or irradiation with NB-UVB were measured by com- the CELLocate slip.18 mercially available enzyme-linked immunosorbent assay kits for basic fibroblast growth factor (bFGF), endothelin 1 (ET-1), stem cell factor (SCF) and hepatocyte growth factor (HGF) Zymography for matrix metalloproteinase (MMP)-2 and (Quantikine; R&D Systems, Minneapolis, MN, U.S.A.). The MMP-9 results are presented as mean ± SD. The detectable concentra- ) ) Zymography was performed in sodium dodecyl sulphate tion range was 0–640 pg mL 1 for bFGF, 0–120 pg mL 1 for ) ) (SDS)-polyacrylamide gel electrophoresis (10%) containing ET-1, 31Æ2–2000 pg mL 1 for SCF and 125–8000 pg mL 1 0Æ1% gelatin as previously described.19 Twelve to fifteen mi- for HGF. crolitres of supernatants derived from MCs treated with PUVA ) ) (8-MOP, 5 · 10 7 mol L 1; UVA dose 0, 30, 60 and ) Statistical analysis 120 mJ cm 2) or KCs treated with PUVA (8-MOP, ) ) ) 5 · 10 7 mol L 1; UVA dose 0, 100, 200 and 400 mJ cm 2) SPSS system for Windows version 10.0 (SPSS Inc., Chicago, IL, was mixed with nonreducing sample buffer and subjected to U.S.A.) was used for statistical analysis. The results are electrophoresis without boiling. After electrophoresis, the gels expressed as mean ± SD. Student’s t-test was used for statistical

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp122–129 Effects of PUVA on epidermal cells and vitiligo, C-S. Wu et al. 125 evaluation between control and experimental groups in the (a) study. P <0Æ05 was considered to be statistically signiﬁcant. 2·5

) 2 -1 ml

Results 5 1·5 Melanocytes are more sensitive to psoralen plus 1 ultraviolet A-induced cellular damage than are

keratinocytes Cell No· (x 10 0·5

In comparison with the viability of nontreated MCs and KCs 0 at 24 h (100%), the viability of MCs treated with PUVA 30, 0 UVA 30 mJ cm-2 UVA 60 mJ cm-2 UVA 120 mJ cm-2 )2 60, 120 and 240 mJ cm for 24 h was 94Æ5±9Æ7%, (8-MOP: 5 x 10-7 M) 98Æ6±6Æ5%, 101Æ2±10Æ1% and 75Æ2±8Æ9%, respectively. (b) The viability of KCs treated with PUVA 100, 200, 400 and ) 2·5 800 mJ cm 2 for 24 h was 99Æ3±4Æ8%, 99Æ4±5Æ7%, *

98Æ9±6Æ1% and 80Æ1±7Æ3%, respectively. These results ) -1 2 )2 show that PUVA 240 and 800 mJ cm produces signiﬁcant ml 5 damage in MCs and KCs, respectively (P <0Æ05). Therefore, 1·5 ) irradiation doses higher than PUVA 120 mJ cm 2 in MC treat- ) 1 ments and 400 mJ cm 2 in KC treatments were excluded from

our following experiments. Cell No· (x 10 0·5

0 The proliferation of melanocytes was not stimulated by NBUVB 25 0 UVA 100 UVA 200 UVA 400 -2 -2 -2 -2 psoralen plus ultraviolet A (PUVA) treatment or by mJ cm mJ cm mJ cm mJ cm supernatants from PUVA-treated keratinocytes Treated keratinocyte supernatants (8-MOP: 5 x 10-7 M)

To examine the effect of PUVA or of supernatants derived Fig 1. Effects of melanocyte proliferation stimulated with psoralen from PUVA/NB-UVB-treated KCs on MC growth, the trypan plus ultraviolet (UV) A (PUVA) or with supernatants derived from blue exclusion test was used to determine cell number of keratinocytes treated with PUVA or narrowband UVB (NB-UVB). After ) treating with (a) PUVA 0, 30, 60 or 120 mJ cm 2 (8- MCs. As demonstrated in Fig. 1a, the cell number of MCs trea- ) ) ) 7 1 ted with PUVA 0, 30, 60 and 120 mJ cm 2 for 6 days was methoxypsoralen, 8-MOP: 5 · 10 mol L ) or (b) supernatants 5 5 5 derived from keratinocytes treated with PUVA 0, 100, 200 and 1Æ06 ± 0Æ30 · 10 ,1Æ01 ± 0Æ50 · 10 ,0Æ98 ± 0Æ40 · 10 )2 )7 )1 )2 ) 400 mJ cm (8-MOP: 5 · 10 mol L ) or NB-UVB 25 mJ cm and 0Æ91 ± 0Æ26 · 105 cells mL 1, respectively. The cell for 6 days, the cell numbers of melanocytes were enumerated by number of MCs treated with 10% supernatants derived from trypan blue exclusion test. Results are representative of three )2 KCs treated with PUVA 0, 100, 200 and 400 mJ cm independent experiments and are shown as mean ± SD. *P <0Æ05 5 5 for 6 days was 1Æ05 ± 0Æ50 · 10 ,1Æ01 ± 0Æ30 · 10 , compared with control group. ) 1Æ29 ± 0Æ40 · 105 and 1Æ39 ± 0Æ50 · 105 cells mL 1, respectively. The number of MCs treated with 10% supernatants ) Psoralen plus ultraviolet A treatment significantly derived from KCs irradiated with NB-UVB 25 mJ cm 2 for ) stimulated the matrix metalloproteinase-2 activity in 6 days was 1Æ75 ± 0Æ25 · 105 cells mL 1 (Fig. 1b). The melanocytes results show that MC proliferation was not stimulated directly or indirectly via KCs by PUVA treatment. On the other hand, In comparison with the activity of matrix metalloproteinase NB-UVB irradiation indirectly stimulated the proliferation of (MMP)-2 in supernatant derived from untreated MCs (100%), MCs via KCs (P <0Æ05). the MMP-2 activity in supernatants obtained from MCs treated ) with PUVA 30, 60 and 120 mJ cm 2 was 166Æ3±4Æ5%, 188Æ7±8Æ4% and 141Æ5±3Æ8%, respectively. Although the Melanocyte migration was not enhanced by psoralen MMP-2 activity in supernatant derived from MCs treated with plus ultraviolet A treatment ) PUVA 120 mJ cm 2 was slightly lower than in supernatant from ) The migration distances of MCs at 24 h after PUVA 0, 30, 60 MCs treated with PUVA 30 and 60 mJ cm 2, the supernatants ) ) and 120 mJ cm 2 treatment were 78Æ64 ± 50Æ44, 87Æ06 ± derived from MCs treated with PUVA 30, 60 and 120 mJ cm 2 38Æ48, 84Æ33 ± 36Æ25 and 78Æ10 ± 34Æ78 lm, respectively. significantly stimulated MMP-2 activity, as shown by zymogra- The migration distance of MCs at 24 h after NB-UVB phy (Fig. 3a). However, the supernatants derived from KCs ) ) 25 mJ cm 2 was 157Æ05 ± 26Æ05 lm(Fig. 2). PUVA treat- treated with PUVA 100, 200 and 400 mJ cm 2 showed no ment produced no significant changes on MC migration. significant effect on MMP-9 activity (Fig. 3b). The activity of However, NB-UVB irradiation significantly enhanced the MC MMP-2 and MMP-9 in PUVA-treated MC supernatants and KC migration (P <0Æ05). supernatants was analysed by densitometric scanning (Fig. 3c).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp122–129 126 Effects of PUVA on epidermal cells and vitiligo, C-S. Wu et al.

250 Table 2. The results demonstrate that the sera from patients m)

μ * 200 with repigmenting vitiligo contained signiﬁcantly higher levels of bFGF, SCF and HGF as compared with healthy controls and 150 patients with active vitiligo. 100

Migration distance ( 0 NBUVB 0UVA 30UVA 60 UVA 120 25 mJ cm-2 mJ cm-2 mJ cm-2 mJ cm-2 (a) (8-MOP: 5 x 10-7 M) 1 2345

Fig 2. Effects of psoralen plus ultraviolet (UV) A (PUVA) treatment )2 (0, 30, 60 and 120 mJ cm ; 8-methoxypsoralen, 8-MOP: 120 92 KDa ) ) 5 · 10 7 mol L 1) and narrowband UVB (NB-UVB) irradiation 84 ) 72 KDa (25 mJ cm 2) on migration of cultured human melanocytes. Cultured human melanocytes were seeded on type I collagen-precoated CELLocate coverslips at a density of 1 · 104 cells/well. Time-lapse photographs were taken at selected time points and the migration distance (lm) in 24 h was measured. Results are representative of two independent experiments. Error bars represent mean ± SD of 10–12 cells. *P <0Æ05 compared with control group.

Release of melanocyte mitogens/growth factors from (b) keratinocytes was not upregulated by psoralen plus 12 3 4 5 ultraviolet A treatment

As shown in Table 1, PUVA treatment gave no significant stimulatory effect on release of bFGF, ET-1, SCF or HGF by KCs. 120 92 KDa The serum levels of basic fibroblast growth factor, stem 84 cell factor and hepatocyte growth factor in patients with repigmenting vitiligo after psoralen plus pultraviolet A 72 KDa treatment were significantly higher than those in healthy controls and in patients with active vitiligo. (c) The concentrations of bFGF, SCF, HGF and ET-1 in sera 250 MMP-2 obtained from healthy controls and patients are shown in MMP-9 200 * * 150 *

100 Fig 3. Gelatin zymogram of conditioned media from melanocytes and keratinocytes treated with psoralen plus ultraviolet (UV) A (PUVA). 50 (a) Supernatants derived from PUVA-treated melanocytes were 0 subjected to gelatin zymography analysis. Lane 1, marker; lane 2, (% of control medium) Activity Control UVA 30 UVA 60 UVA 120 ) medium mJcm-2 mJ cm-2 mJ cm-2 control medium; lane 3, PUVA (UVA 30 mJ cm 2;8- ) ) Treated melanocyte supernatants (8-MOP: 5 x 10-7 M) methoxypsoralen, 8-MOP: 5 · 10 7 mol L 1); lane 4, PUVA (UVA ) ) ) 60 mJ cm 2; 8-MOP: 5 · 10 7 mol L 1); lane 5, PUVA (UVA ) ) ) 120 mJ cm 2; 8-MOP: 5 · 10 7 mol L 1). (b) Supernatants derived 250 MMP-2 MMP-9 from PUVA-treated keratinocytes were subjected to gelatin 200 zymography analysis. Lane 1, marker; lane 2, control medium; lane 3, ) ) ) PUVA (UVA 100 mJ cm 2; 8-MOP: 5 · 10 7 mol L 1); lane 4, PUVA 150 ) ) ) (UVA 200 mJ cm 2; 8-MOP: 5 · 10 7 mol L 1); lane 5, PUVA (UVA ) ) ) 100 400 mJ cm 2; 8-MOP: 5 · 10 7 mol L 1). (c) The activity of matrix metalloproteinase (MMP)-2 and MMP-9 in PUVA-treated melanocyte 50 supernatants (top) and keratinocyte supernatants (bottom) was 0 analysed by densitometric scanning of the zymograms after adjusting (% of control medium) Activity Control UVA 30 UVA 60 UVA 120 for control medium (100%). *P <0Æ05 as compared with the control medium mJcm-2 mJ cm-2 mJ cm-2 group. Similar results were observed in two separate experiments. Treated melanocyte supernatants (8-MOP: 5 x 10-7 M)

Table 1 Mean ± SD concentrations of melanocyte mitogens/growth factors in keratinocyte supernatants after psoralen plus ultraviolet (UV) A (PUVA) treatment

) Mitogens/growth factors (pg mL 1) PUVAa ) UVA (mJ cm 2) SCF HGF bFGF ET-1 015Æ20 ± 2Æ10 58Æ98 ± 5Æ78 4Æ35 ± 1Æ18 9Æ74 ± 5Æ22 100 15Æ24 ± 2Æ85 60Æ37 ± 6Æ94 4Æ00 ± 1Æ22 9Æ77 ± 6Æ18 200 14Æ09 ± 3Æ47 59Æ36 ± 4Æ81 5Æ40 ± 1Æ26 10Æ87 ± 7Æ04 400 15Æ47 ± 4Æ04 61Æ47 ± 8Æ60 4Æ46 ± 1Æ53 10Æ16 ± 7Æ52

SCF, stem cell factor; HGF, hepatocyte growth factor; bFGF, basic ﬁbroblast growth factor; ET-1, endothelin 1; a8-methoxypsoralen: ) ) 5 · 10 7 mol L 1.

Table 2 Mean ± SD serum concentrations of melanocyte mitogens/growth factors in healthy controls, in patients with active vitiligo and in patients with repigmenting vitiligo

) Mitogens/growth factors (pg mL 1)

bFGF SCF HGF ET-1 Healthy controls (n ¼ 12) 3Æ04 ± 0Æ06 315Æ26 ± 216Æ09 562Æ75 ± 222Æ82 4Æ52 ± 1Æ57 Active vitiligo (n ¼ 15) 2Æ58 ± 1Æ11 414Æ34 ± 82Æ61 453Æ66 ± 64Æ63 3Æ67 ± 1Æ66 Repigmenting vitiligo (n ¼ 15) 7Æ65 ± 3Æ95* 746Æ22 ± 155Æ93* 1175Æ46 ± 422Æ85* 3Æ76 ± 2Æ81

SCF, stem cell factor; HGF, hepatocyte growth factor; bFGF, basic ﬁbroblast growth factor; ET-1, endothelin 1; *P <0Æ05 compared with healthy controls and active vitiligo.

MMP-2 activity in supernatants derived from PUVA-treated Discussion MCs was significantly increased as compared with the control Previous studies have shown that recovery of vitiligo is initi- group. However, neither MMP-2 nor MMP-9 activity in KC ated by activation, proliferation, migration and melanogenesis supernatants was stimulated by PUVA treatment. Studies from 29 of MCs.3–5,21 However, the molecular mechanisms of these Lei et al. demonstrated that MMP-2 activity is required for processes have remained unravelled. Although several thera- the migration of melanoblasts. On the other hand, after NB- peutic modalities have been used clinically to treat vitiligo, UVB irradiation, the expressions of MMP-2 activity in MCs PUVA and NB-UVB are still the most common options. Our and MMP-9 activity in KCs were significantly increased. These previous studies demonstrated that NB-UVB directly or indir- zymography results revealed that both PUVA treatment and ectly (via KCs) regulated the physiological functions of MCs NB-UVB irradiation could promote a favourable environment that could facilitate the repigmentation in vitiligo lesions.14 In for MC migration; however, NB-UVB is more effective as it the current study, we used NB-UVB irradiation as a reference also promotes MMP-9 activity in KC supernatant. for comparison to explore in vitro the mechanisms involved in In our in vitro experimental system, PUVA seemed to have the processes of PUVA-induced repigmentation in patients limited immediate effects on cultured epidermal cells. How with vitiligo. then, in vitiligo, can PUVA stimulate the repigmentation Our results showed that PUVA treatment did not significant- in vivo? In vitiliginous lesions, the autoimmune mechanisms ly stimulate the release of MC mitogens/growth factors from appear to involve the progressive destruction of MCs by auto- 30 KCs. In contrast, an increase in bFGF and ET-1 secretion from antibody-dependent cellular cytotoxicity or by cytotoxic T 31 KCs was observed after NB-UVB irradiation.14 Both bFGF and lymphocytes. It has been reported that PUVA treatment is 16 ET-1 are recognized as mitogens for MCs, which could able: (i) to deplete vitiligo-associated MC antigens; (ii) to 32 enhance the growth and survival of MCs.22,23 Therefore, in deplete Langerhans cells in vitiliginous epidermis; and (iii) 16 our experimental system, PUVA treatment did not promote significantly to increase the activity of tyrosinase in MCs. In MC proliferation via KCs. addition, 8-MOP stimulates melanin production by melano- 33 The MMPs are a family of enzymes involved in degradation blasts. These results indicated that PUVA therapy (i) blocks of extracellular matrix components such as collagen, gelatin the binding of anti-MC autoantibodies to MCs; (ii) inhibits and fibronectin.24 Expression of MMPs is markedly increased the progression of antibody-dependent cytotoxicity to MCs in in situations involving active tissue remodelling and cell vitiligo; and (iii) stimulates melanogenesis. Our previous migration.25–28 As our results indicated, the expression of study revealed that PUVA treatment significantly increased the

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp122–129 128 Effects of PUVA on epidermal cells and vitiligo, C-S. Wu et al. number of split-DOPA-positive MCs but decreased Langerhans impact of NB-UVB on epidermal cells was more prompt as cells in repigmented areas of epidermal sheets as compared compared with PUVA. with those of depigmented lesions in stable vitiligo.16 These In summary, we have shown that NB-UVB and PUVA have in vivo pieces of evidence provide reasonable explanations for different effects on cultured epidermal cells; more specifically, the clinical efficacy of PUVA treatment on vitiligo lesions. NB-UVB imparted immediate effects in cultured MCs and KCs To explore further the effects of PUVA in patients with viti- while rapid responses were not observed in cultured epider- ligo, serum samples from patients with vitiligo were collected mal cells after PUVA treatment except for induction of MMP-2 and analysed. Our present data revealed that the sera from from MCs. PUVA treatment creates a favourable milieu for PUVA-treated patients with repigmenting vitiligo contained promoting the growth of MCs through inducing immune sup- higher concentrations of bFGF, SCF and HGF as compared pression and elevating MC mitogens. Based on our results, we with those of healthy controls and patients with active vitiligo. propose a novel therapeutic scheme for treating vitiligo: in the As bFGF, SCF and HGF are the documented factors that stimu- active stage of vitiligo, PUVA treatment is the therapy of late the growth of MCs,34,35 our result corroborated the previ- choice to slow down the destruction of MCs and create a ous study from Abdel-Naser et al.,36 showing that sera from favourable environment for MCs to survive. In the stable stage patients with vitiligo could promote MC growth. Although of vitiligo, NB-UVB irradiation should be used to stimulate PUVA did not induce release of MC mitogens from cultured MC proliferation and migration directly. We believe that this KCs, the possible explanations for the increase seen in our therapeutic regimen with a solid theoretical basis will bring in vivo results include (i) fibroblasts, another natural cellular more prompt and satisfactory clinical responses to patients neighbour of MCs in vivo, that may release these mitogens after with vitiligo. phototherapy;37–39 and (ii) because at least 20–40 PUVA treatments are generally required before repigmentation occurs in Acknowledgments patients,7 short-term PUVA treatment may not be adequate to stimulate the release of growth factors by KCs or fibroblasts This study was supported by the National Science Council in vitro. In our study, PUVA-treated patients with vitiligo who research grants NSC 93-2314-B-002-070 and NHRI-EX94- showed repigmentation had elevated serum concentrations of 9231SI. MC mitogens/growth factors in vivo. Therefore, it is reasonable to suggest that the increase of bFGF, SCF and HGF after PUVA References treatment contributes to the therapeutic efficacy in treating vitiligo. In our present studies, the levels of MC mitogens/ 1 Howitz J, Brodthagen H, Schwartz M, Thomsen K. Prevalence of growth factors in patients with repigmenting vitiligo follow- vitiligo: epidemiological survey on the Isle of Bornholm, Denmark. ing NB-UVB treatment were not determined. We cannot Arch Dermatol 1977; 113:47–52. 2 Kovac SO. Vitiligo. J Am Acad Dermatol 1998; 38:647–66. exclude the possibility of increased serum levels of those mit- 3 Staricco RG. Activation of amelanotic melanocytes in the outer root ogens/growth factors in patients with vitiligo following NB- sheath of the hair follicle following ultraviolet exposure. J Invest Der- UVB irradiation. However, the main point of our studies sug- matol 1962; 39:163–4. gested that PUVA therapy lacked immediate effects on cultured 4 Ortonne JP, Schmitt D, Thivolet J. PUVA-induced repigmentation epidermal cells, as seen with NB-UVB. Therefore, the changes of vitiligo: scanning electron microscopy of hair follicles. J Invest seen in serum levels of MC mitogens/growth factors following Dermatol 1980; 74:40–2. PUVA therapy provides an explanation for the therapeutic effi- 5 Cui J, Shen LY, Wang GC. Role of hair follicles in the repigmentation of vitiligo. J Invest Dermatol 1991; 97:410–16. ciency of PUVA. Further studies are required to compare the 6 Morelli JG, Kincannon J, Yohn JJ et al. Leukotriene C4 and TGF- changes in serum levels of MC mitogens/growth factors after alpha are stimulators of human melanocyte migration in vitro. J Invest PUVA or NB-UVB therapy. Dermatol 1992; 98:290–5. Westerhof and Nieuweboer-Krobotova8 reported that 7 Plott RT, Wagner FR. Modern treatment approaches to vitiligo. Cu- patients receiving PUVA and NB-UVB treatments experienced tis 1990; 45:311–16. a repigmentation of 46% and 67%, respectively, after 8 Westerhof W, Nieuweboer-Krobotova L. Treatment of vitiligo with 4 months. The study from Scherschun et al.10 showed that UV-B radiation vs topical psoralen plus UV-A. Arch Dermatol 1997; 133:1525–8. among NB-UVB-treated patients, 70% achieved more than 9 Njoo MD, Bos JD, Westerhof W. Treatment of generalized vitiligo 75% repigmentation with a mean treatment course of 19 ses- in children with narrow-band (TL-01) UVB radiation therapy. JAm sions (about 6 months). However, 9–20 months of PUVA Acad Dermatol 2000; 42:245–53. therapy was required to achieve slight to moderate repigmen- 10 Scherschun L, Kim JJ, Lim HW. Narrow-band ultraviolet B is a use- tation.13,40,41 Therefore, clinical evidence seemed to indicate ful and well-tolerated treatment for vitiligo. J Am Acad Dermatol that the responses obtained from NB-UVB treatment were 2001; 44:999–1003. more prompt as compared with PUVA therapy. However, in 11 Roelandts R. Photo(chemo)therapy for vitiligo. Photodermatol Photoim- munol Photomed 2003; 19:1–4. many of the previous studies, the stage of vitiligo, either ac- 8,9,42–44 12 Lindelof B, Sigurgeirsson B, Tegner E et al. Comparison of the car- tive or stable, is often not mentioned. It may be pos- cinogenic potential of trioxsalen bath PUVA and oral methoxsalen sible that many of the patients enrolled in the previous studies PUVA. A preliminary report. Arch Dermatol 1992; 128:1341–4. were in the stable stage of vitiligo, and therefore the direct

13 Sahin S, Hindioglu U, Karaduman A. PUVA treatment of vitiligo: 30 Cui J, Arita Y, Bystryn JC. Cytolytic antibodies to melanocytes in a retrospective study of Turkish patients. Int J Dermatol 1999; 38: vitiligo. J Invest Dermatol 1993; 100:812–15. 542–5. 31 al Badri AM, Todd PM, Garioch JJ et al. An immunohistological 14 Wu CS, Yu CL, Wu CS et al. Narrow-band ultraviolet-B stimulates study of cutaneous lymphocytes in vitiligo. J Pathol 1993; proliferation and migration of cultured melanocytes. Exp Dermatol 170:149–55. 2004; 13:755–63. 32 Kao CH, Yu HS. Depletion and repopulation of Langerhans cells in 15 Kao CH, Yu HS. Effects of TPA on melanocytes in vitiligo. Arch Der- nonsegmental type vitiligo. J Dermatol 1990; 17:287–96. matol Res 1991; 283:119–24. 33 Lei TC, Virador V, Yasumoto K et al. Stimulation of melanoblast 16 Kao CH, Yu HS. Effects of PUVA on melanocytes in vitiligo and in pigmentation by 8-methoxypsoralen: the involvement of microph- patients with vitiligo vulgaris. J Invest Dermatol 1992; 98:734–40. thalmia-associated transcription factor, the protein kinase A signal 17 Scudiero DA, Shoemaker RH, Paull KD. Evaluation of a soluble tetr- pathway, and proteasome-mediated degradation. J Invest Dermatol azolium/formazan assay for cell growth and drug sensitivity in cul- 2002; 119:1341–9. ture using human and other tumor cell lines. Cancer Res 1988; 34 Yarr M, Gilchrest BA. Human melanocyte growth and differenti- 48:4827–33. ation: a decade of new data. J Invest Dermatol 1991; 97:611–17. 18 Horikawa T, Norris DA, Yohn JJ et al. Melanocyte mitogens induce 35 Jimbow K, Quevedo WC, Prota G, Fitzpatrick TB. Biology of mel- both melanocyte chemokinesis and chemotaxis. J Invest Dermatol anocytes. In: Dermatology in General Medicine (Freedberg IM, Eisen AZ, 1995; 104:256–9. Wolff K, Austen KF, Goldsmith LA, Katz S, eds), 5th edn, New 19 Woessner JF. Quantification of matrix metalloproteinases in tissue York: McGraw-Hill, 1999; 192–219. samples. Methods Enzymol 1995; 248:510–28. 36 Abdel-Naser MB, Hann SK, Bystryn JC. Oral psoralen with UV-A 20 Moellmann G, Klein-Angerer S, Scollay DA et al. Extracellular therapy releases circulation growth factor(s) that stimulates cell granular material and degeneration of keratinocytes in the normally proliferation. Arch Dermatol 1997; 133:1530–3. pigmented epidermis of patients with vitiligo. J Invest Dermatol 1982; 37 Puri N, Mojamdar M, Ramaiah A. Growth defects of melanocytes 79:321–30. in culture from vitiligo subjects are spontaneously corrected in vivo 21 Norris DA, Horikawa T, Morelli JG. Melanocyte destruction and rein repigmenting subjects and can be partially corrected by the add- population in vitiligo. Pigment Cell Res 1994; 7:193–203. ition of fibroblast-derived growth factors in vitro. Arch Dermatol Res 22 Halaban R, Langdon R, Birchall N et al. Basic fibroblast growth fac- 1989; 281:178–84. tor from human keratinocytes is a natural mitogen for melano- 38 Yu HS, Wu CS, Yu CL et al. Helium-neon laser irradiation stimu- cytes. J Cell Biol 1988; 107:1611–19. lates migration and proliferation in melanocytes and induces re- 23 Yada Y, Higuchi K, Imokawa G. Effects of endothelins on signal pigmentation in segmental-type vitiligo. J Invest Dermatol 2003; transduction and proliferation in human melanocytes. J Biol Chem 120:56–64. 1991; 266:18352–7. 39 Poon VK, Huang L, Burd A. Biostimulation of dermal fibroblast by 24 Werb Z. ECM and cell surface proteolysis: regulating cellular ecol- sublethal Q-switched Nd:YAG 532 nm laser: collagen remodeling ogy. Cell 1997; 91:439–42. and pigmentation. J Photochem Photobiol B 2005; 81:1–8. 25 Uitto VJ, Airola K, Vaalamo M et al. Collagenase-3 (matrix metallo- 40 Tran D, Kwok YK, Goh CL. A retrospective review of PUVA ther- proteinase-13) expression is induced in oral mucosal epithelium apy at the National Skin Centre of Singapore. Photodermatol Photoim- during chronic inflammation. Am J Pathol 1989; 152:1489–98. munol Photomed 2001; 17:164–7. 26 Salo T, Ma¨kela¨ M, Kylma¨niemi M et al. Expression of matrix metal- 41 Parsad D, Kanwar AJ, Kumar B.. Psoralen–ultraviolet vs. narrow- loproteinase-2 and -9 during early human wound healing. Lab Invest band ultraviolet B phototherapy for the treatment of vitiligo. J Eur 1994; 70:176–82. Acad Dermatol Venereol 2006; 20:175–7. 27 Parks WC. The production, role, and regulation of matrix metallo- 42 Tjioe M, Gerritsen MJ, Juhlin L, van de Kerkhof PC. Treatment of proteinases in the healing epidermis. A23–37. Wounds 1995; 7: vitiligo vulgaris with narrow band UVB (311 nm) for one year A23–37. and the effect of addition of folic acid and vitamin B12. Acta Derm 28 Vaalamo M, Mattila L, Johansson N et al. Distinct populations of Venereol (Stockh) 2002; 82:369–72. stromal cells express collagenase-3 (MMP-13) and collagenase-1 43 Chen GY, Hsu MM, Tai HK et al. Narrow-band UVB treatment of (MMP-1) in chronic ulcers but not in normally healing wounds. vitiligo in Chinese. J Dermatol 2005; 32:793–800. J Invest Dermatol 1997; 109:96–101. 44 Hong SB, Park HH, Lee MH. Short-term effects of 308-nm xenon- 29 Lei TC, Vieira WD, Hearing VJ. In vitro migration of melanoblasts chloride excimer laser and narrow-band ultraviolet B in the treat- requires matrix metalloproteinase-2: implications to vitiligo ther- ment of vitiligo: a comparative study. J Korean Med Sci 2005; apy by photochemotherapy. Pigment Cell Res 2002; 15:426–32. 20:273–8.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp122–129 THERAPEUTICS DOI 10.1111/j.1365-2133.2006.07561.x A double-blind, randomized quantitative comparison of calcitriol ointment and calcipotriol ointment on epidermal cell populations, proliferation and differentiation J.E.M. Ko¨rver, W.H.P.M. Vissers, D.W.A. Van Rens, M.C. Pasch, P.E.J. Van Erp, J.B.M. Boezeman and P.C.M. Van De Kerkhof Department of Dermatology, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands

Summary

Correspondence Background Calcitriol and calcipotriol are widely used in the topical treatment of J.E.M. Ko¨rver. psoriasis. However, studies comparing both treatment modalities are scarce. Espe- E-mail: [email protected] cially, there are almost no studies comparing the effects on epidermal cell populations in a quantitative manner. Accepted for publication 11 June 2006 Objectives The aim of this study was to quantitatively compare the effects of topical calcitriol and topical calcipotriol on clinical scores and epidermal subpopulations. Key words Patients and methods From five patients with stable plaque psoriasis, skin biopsies calcipotriol, calcitriol, immunohistology, keratin were taken from two symmetrical regions on the trunk or extremities before and 15, psoriasis, transit-amplifying cells after treatment with either calcitriol or calcipotriol. Frozen sections were labelled b Conflicts of interest immunofluorescently using direct immunofluorescence for -1 integrin and the P.C.M. van de Kerkhof has been active in lecturing Zenon labelling technique for keratin (K) 6, K10 and K15. The digital photo- at symposia, conducting clinical trials and graphs of the stained sections were quantitatively analysed and the results of both providing consultancies to various companies having treatments were compared. vitamin D3 products in their portfolio. There are Results The clinical SUM-score improved significantly for both the calcitriol- and no other conflicts of interests to declare. the calcipotriol-treated lesions. In the calcipotriol-treated group the expression of K10 and K15 increased and the expression of K6 decreased significantly. No changes were seen for the marker b-1 integrin. In the calcitriol-treated group none of the markers changed significantly. A tendency towards significance was seen for the changes in the expression of K6 and K15 in favour of calcipotriol. Conclusions Both calcitriol and calcipotriol gave a significant improvement in clinical scores. However, treatment with calcipotriol resulted in a normalization of K6, K10 and K15, whereas treatment with calcitriol did not. Comparison of both treatments showed a tendency towards significance for the above-mentioned markers for calcipotriol only.

Vitamin D3 derivatives are used over many years as a safe and Whether calcitriol has as strong antipsoriatic properties as effective topical treatment for psoriasis. When used with a calcipotriol is unclear. Comparative clinical studies between ) maximum of 100 g week 1 for calcipotriol ointment calcitriol and calcipotriol are scarce and individual study ) ) (50 lgmL 1) and 210 g week 1 for calcitriol ointment results are difficult to compare because of differences in study ) (3 lgmL 1) they also have little effect on calcium metabo- design. Most importantly, various scoring systems have been lism or bone density.1,2 The clinical efficacy of one of the used to compare the clinical improvements after treatment most used vitamin D analogues, calcipotriol, in clearing psor- with either calcitriol or calcipotriol. Bourke et al.1 found an iasis is comparable or better than topical steroids or dithra- improvement in PASI of 32% in the group of patients treated nol.3,4 Whereas calcipotriol is effective, it also has some with calcitriol and 68% in the group of patients treated with dermatological side-effects such as perilesional erythema and calcipotriol. The difference between these therapies was statis- stinging or burning at the applied areas. Calcitriol has these tically significant. In another trial, Franssen et al.7 found an side-effects at a lower frequency5 and therefore the use of cal- improvement of 65% and 74% in the SUM scores of selected citriol has been advocated in sensitive areas of the skin, such lesions on nonflexural areas for calcitriol and calcipotriol, re- as the flexural, retro-auricular and facial areas.6 spectively. However, the difference between the improvements

2006 The Authors 130 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp130–137 Quantitative comparison of calcitriol and calcipotriol ointment, J.E.M. Ko¨rver et al. 131 of these two therapies was not significant.7 Using a global Exposure to sun or ultraviolet radiation was prohibited during nonquantitative scoring system Ortonne et al.6 showed that cal- the course of the study. citriol gave better results in flexural areas compared with calci- Patients were double-blind randomized to treat all psoriatic potriol. lesions on either the left or the right side of their body with Psoriatic lesions are characterized by inflammation, an in- calcitriol ointment or calcipotriol ointment. Patients were creased epidermal proliferation, and a disturbed keratinization instructed to wash their hands between applications and after with a decreased compartment of normally differentiated kera- the applications. During the study no other topical or systemic tinocytes. Vitamin D derivatives are known to inhibit keratino- antipsoriatic treatments were allowed. cyte proliferation, induce keratinization and cornified envelope Clinical efficacy of both treatments was evaluated by select- formation in psoriasis.8–10 The inhibition of the epidermal ing a target lesion on each of the body parts treated with hyperproliferation seems to be mainly due to the direct inhi- either calcitriol or calcipotriol. Of this target lesion a SUM- bition of the hyperproliferative basal keratinocytes.11 Accord- score was performed. This SUM-score was calculated by using ing to Gniadecki this is ascribed to the induction of the a five-point scale ranging from 0 to 4 for erythema, indura- differentiative pathway in the hyperproliferative epidermal tion and desquamation, resulting in a SUM score with a max- progenitor cells residing in the basal cell layer.12 imum of 12 points. Care was taken to ensure that both target Only three cell biological studies have compared the effects lesions on each of the body parts had the same SUM-score at of calcitriol and calcipotriol on the various components of the baseline. The target lesions were located symmetrically on the human skin. One study focusing on the effects of both topical trunk or extremities. vitamin D derivatives on cultured keratinocytes showed that At baseline and after 6 weeks of treatment 4-mm punch both calcitriol and calcipotriol inhibited the keratinocyte pro- biopsies were taken from both target lesions on the trunk or liferation, induced the cornified envelope formation and had extremities after local anaesthesia with lidocaine 2%/adrenal- no effect on keratin (K) 5 and K1 expression.9 Another study ine 1 : 10 0000 (AstraZeneca BV, Zoetermeer, the Nether- reported a significant reduction in factor XIII expressing der- lands). All specimens were embedded in Tissue-Tek (Sakura mal dendritic cells after treatment with calcitriol, but not after Finetek, Zoeterwoude, the Netherlands), snap frozen in liquid treatment with calcipotriol.13 Finally, a flow cytometric study nitrogen and stored at ) 80 C. on keratinocytes by Franssen et al.7 discovered a significant increase in K10+/K6) subpopulations, K6 and K10 double- Fluorescent immunohistochemistry and Zenon labelling negative subpopulations and in overall K10 expression after technique treatment with calcipotriol. A similar tendency was found for calcitriol, but this was not significant.7 Overall, the data com- This method has been described previously.15,16 In brief, paring the clinical and cell biological effects of calcitriol and monoclonal antibodies (mAbs) were labelled and/or diluted. calcipotriol are not only scarce, but also incongruous with Frozen sections (7-lm) of the biopsies were incubated with each other. the labelled antibodies. A mAb against K10 (RKSE60; Mono- Because the affinity for the vitamin D receptor and the san, Uden, the Netherlands) was labelled using the Zenon la- pharmacokinetics of calcitriol and calcipotriol differ, a dispar- belling technique and was used for the assessment of ity in clinical efficacy and cell biological effects might be epidermal differentiation. For examination of epidermal expected.14 Therefore, the aim of our study was to quantita- hyperproliferation, K6 antibody (LHK6B; Novocastra Labora- tively compare the effects of topical calcipotriol and topical tories, Newcastle upon Tyne, U.K.) was labelled using the Ze- calcitriol on clinical scores, epidermal subpopulations and ker- non labelling technique. mAbs against b-1 integrin (CD29 atinocyte proliferation and differentiation in patients with sta- FITC, clone K20; DakoCytomation, Copenhagen, Denmark) ble plaque psoriasis. and K15 (LHK15; Neomarkers, Fremont, CA, U.S.A.) were used for analysis of the basal layer of the epidermis. K15 was Patients and methods labelled using the Zenon labelling technique whereas b-1 integrin was covalently labelled with fluorescein–isothiocyanate (FITC). Patients and biopsies For the double staining, equal volumes of two antibody Five patients with stable plaque psoriasis were included in this dilutions were mixed and incubated on the slides. K10 was study, after giving informed consent. Patients were diagnosed combined with b-1 integrin and with K6, while b-1 integrin with stable and mild to moderate plaque psoriasis. They all was also mixed with K15. had symmetrical lesions on their trunk and/or extremities. The sections and mAbs were examined using an immuno- Patients had not received any topical treatment for 2 weeks fluorescence microscope (Axioskop 2 MOT; Zeiss, Oberko- and no systemic treatment for 4 weeks. They had not used chen, Germany) with a magnification of 200· and a halogen medication that could interfere with psoriasis, e.g. systemic lamp of 100 W. The mAbs were detected using the follow- corticosteroids, b-blockers, lithium carbonate, antimalaria ing filters: for Alexa fluor 594 excitation BP 546/12, FT medication or immunosuppressants. Patients with known or 580, emission LP 590 (red), for FITC and Alexa Fluor 488 suspected abnormality of calcium metabolism were excluded. excitation BP 485/20, FT 510, emission LP 515 (green) and

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp130–137 132 Quantitative comparison of calcitriol and calcipotriol ointment, J.E.M. Ko¨rver et al. for DAPI excitation BP 365/12, FT 395, emission LP 397 Results (blue). Images were photographed using a digital camera (Axiocam The quantification of the various markers and scores have been MRc5; Zeiss) and AxioVision software (Zeiss). For each anti- summarized in Figure 1. body five different parts of the epidermis were photographed. A statistically significant reduction of the SUM scores was The photos made with the different filters were later merged seen in both the calcitriol- and calcipotriol-treated lesions with the AxioVision software to get the triple stained pictures. (from 7Æ0to3Æ8; P <0Æ05 for calcipotriol and from 7Æ0to 5Æ0; P <0Æ05 for calcitriol). However, the decline in the calcipotriol-treated lesions was more pronounced ()46%) than in Image analysis the calcitriol treated lesions ()29%). The images were analysed with IP-lab software (Scanalytics, The K6 expression in the calcipotriol-treated lesions Rockville, MD, U.S.A.) in a standardized manner.15,16 For decreased significantly (from 54Æ0to9Æ98; P <0Æ001). The expression of b-1 integrin, K6, K10 and K15 a region of K6 expression in the calcitriol-treated lesions also decreased, interest (ROI) was drawn in a frame. This ROI contained the but this decrease was not significant (from 46Æ6to35Æ72; epidermis without the stratum corneum. In this ROI the inten- P ¼ 0Æ46). sity and percentage of positive signal in the epidermis was In the calcipotriol-treated lesions the K10 expression in- measured. For b-1 integrin expression this was done by set- creased significantly (from 49Æ3to62Æ6; P <0Æ05). The K10 ting the reverse intensity and green signal on the median expression in the calcitriol-treated lesions did not change sig- value +20, after which the percentage of ROI and mean inten- nificantly (from 48Æ1to50Æ1; P ¼ 0Æ78). sity signal were measured. Before treatment no K15 expression could be detected in For K6, K10 and K15 the reverse intensity was set on the either treatment group (Fig. 2). After 6 weeks of treatment median value +15, after which the percentage expression in the K15 expression in the lesions treated with calcipotriol in- the ROI and the mean intensity signal were measured. creased significantly (from 0 to 1Æ19; P <0Æ05). In contrast, the K15 expression in the lesions treated with calcitriol did not increase significantly (from 0 to 0Æ13; P ¼ 0Æ88). Statistical analysis In both the calcipotriol- and the calcitriol-treated lesions the Statistical analysis was performed using simple statistics b-1 integrin expression did not change. (mean, SEM). A t-test for dependent samples was done for the When the effects of treatment with calcipotriol and calcitri- separate analysis of the effects of both ointments. For statistical ol were compared using a repeated measurement ANOVA, for analysis of the differences between calcitriol and calcipotriol a the SUM score, K10 expression and b-1 integrin expression, repeated measurements analysis of variance (ANOVA) was per- no statistically significant differences between both treatments formed. This was all done using Statistica 6 (StatSoft Inc, Tul- were found. However, for K6 and K15 expression, there was sa, OK, U.S.A.). P-values < 0Æ05 were considered statistically a tendency towards significance (P ¼ 0Æ06 for both markers) significant. in favour of the calcipotriol-treated lesions.

(a) (b) (c)

(d) (e)

Fig 1. Quantitative results of the diverse markers for both calcipotriol and calcitriol at baseline and after 6 weeks of treatment. Results are expressed as mean ± SEM, *P <0Æ05. (a) Clinical score of the target lesions, expressed as SUM score. (b) Relative surface area of keratin (K) 6- expressing cells. (c) Relative surface area of K10-expressing cells. (d) Relative surface area of K15-expressing cells. (e) Relative surface area of b-1 integrin-expressing cells.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp130–137 Quantitative comparison of calcitriol and calcipotriol ointment, J.E.M. Ko¨rver et al. 133

(a) (b)

Fig 2. The expression of keratin (K) 15 (red) before and after treatment with calcipotriol (a,b) and calcitriol (c,d). Note the reappearance of basal K15 (visible as a patchy, basal red ﬂuorescent staining) after treatment with calcipotriol.

When comparing the double labelling of K6 and K10 cipotriol or calcitriol this cell population disappeared or was (Fig. 3) we found three different subpopulations of cells in reduced to one cell layer only (Fig. 4b,d). the epidermis. Before treatment the majority of the epidermal cells in the lesions in both treatment groups co-expressed K6 Discussion and K10. A small percentage of the cells showed an expression of only K6, whereas single expression of K10 within one cell Vitamin D3 derivatives are a cornerstone in the topical treat- was hardly seen. After treatment with calcipotriol there was a ment of psoriasis. Both calcipotriol and calcitriol are approved shift in cell populations in the lesions. In these lesions all cells and have been shown to be effective in several studies.4,17–19 showed a single expression of K10 and there were no cells Comparative studies between calcipotriol and calcitriol are co-expressing K6 and K10 (Fig. 3b). In the lesions treated scarce but calcipotriol appears to be the more powerful antip- with calcitriol this shift in the cell populations expressing K6 soriatic compound in chronic plaque psoriasis on nonﬂexural, and K10 was not seen. nonfacial sites.1,4,7 Although these studies used a variety of With respect to the double labelling of b-1 integrin and clinical scoring systems, our clinical data conﬁrms that calci- K10 three different cell populations were also found (Fig. 4). potriol appears to be the most potent vitamin D3 derivative. The cells in the suprabasal compartment below the stratum After 6 weeks of treatment with calcipotriol the clinical SUM corneum showed single expression of K10. The basal cell scores in our study were reduced by 46%, whereas treatment layers showed single expression of b-1 integrin. Nevertheless, with calcitriol reduced the SUM scores by 29%. before treatment a population of two to three cell layers co- Vitamin D3 derivatives are known to inhibit hyperprolifera- expressing b-1 integrin dim cells and K10 was seen just above tion and to promote keratinocyte differentiation in the psoriat- the basal cell layer (Fig. 4a,c). After treatment with either cal- ic epidermis.9,11,20,21 In this study we investigated the effects

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp130–137 134 Quantitative comparison of calcitriol and calcipotriol ointment, J.E.M. Ko¨rver et al.

(a) (b)

Fig 3. Double labelling of keratin (K) 6 (green) and K10 (red), before and after treatment with calcipotriol (a,b) and calcitriol (c,d). Co-expressing cells are visible as yellow or orange cells. Clearly visible is the disappearance of K6 expression after treatment with calcipotriol (b). of topical calcipotriol and topical calcitriol on epidermal tively visible as a diminished expression of K6, did not reach subpopulations and keratinocyte proliferation and differenti- statistical significance. This was even more visible in a qualit- ation in psoriasis patients. We used the marker K6, a keratin ative way in our pictures showing the almost complete disap- upregulated in hyperproliferative epidermal keratinocytes, to pearance of K6 expression after calcipotriol treatment, whereas assess the effects of both vitamin D3 derivatives on hyperpro- after calcitriol treatment only a slight shift in the cell popula- liferation.22,23 tions expressing K6 and K10 was seen, with a slight increase Only one study has been published investigating the effect in cells showing single K10 expression. of vitamin D derivatives on K6 expression.7 In this compara- The different results of the study from Franssen et al.7 can tive, flow cytometric study on K6/K10 subpopulations, no perhaps be explained by the difference in methodology used. change in overall K6 expression was seen. However, studies Franssen et al. used flow cytometry to quantitatively measure on K16, a keratin co-expressed with K6, which also has an in- cell populations co-expressing K6 and K10. This was done by creased expression in hyperproliferative skin such as in psoria- measuring co-expression in single cells, out of their histo- sis, show a decreased expression after treatment with logical context. In our experiments we also qualitatively visu- calcipotriol or calcitriol.7,24–28 This is in accordance with our alized these cell populations, but they were still in their quantitative experiments that show a definite inhibition of histological context. After treatment with calcipotriol, a signifi- hyperproliferation as was measured by a decreased K6 expres- cant increase in the expression of the differentiation-related sion. This inhibition of hyperproliferation after treatment with marker K10 was seen. The expression of K10 in psoriatic le- calcipotriol was statistically significant. After treatment with sions treated with calcitriol did not change at all, and was not calcitriol the inhibition of hyperproliferation, although qualita- even visible in a qualitative way. This difference between

(a) (b)

Fig 4. Double labelling of b-1 integrin (green) and keratin (K) 10 (red), before and after treatment with calcipotriol (a,b) and calcitriol (c,d). Just above the basal cell layer a population of two to three cell layers co- expressing b-1 integrin and K10 is seen, visible as cells with a red cytoplasm surrounded by a green cell membrane. These cell layers disappear after treatment with either calcipotriol or calcitriol. The cells co- expressing b-1 integrin and K10 can be seen even more clearly in the magnified inlays. Note the dermal immunofluorescent staining of b-1 integrin on fibroblasts and the endothelium of the vasculature. calcipotriol and calcitriol was even more obvious in our dou- changes in both the calcitriol- and calcipotriol-treated lesions. ble-labelling experiments with K6 and K10. After treatment This is also in agreement with the aforementioned study in with calcipotriol there was a decrease of the cell population which the epidermal cell population not co-expressing K6 and co-expressing K6 and K10 and a concomitant increase of the K10 represents for the most part the b-1 integrin-positive keratinocyte population expressing only K10, thus showing a population.7 However, in our double-labelling experiments of shift from the hyperproliferative cell compartment towards the b-1 integrin with K10 we observed a definite reduction in the differentiating cell compartment. This shift was not seen in number of basal cell layers expressing b-1 integrin. Interest- the group of psoriatic lesions treated with calcitriol. Although ingly this reduction was mainly due to a reduction in the there is scant literature on the changes in K10 expression after number b-1 integrin dim cells. These cells also co-expressed treatment with calcitriol, all the literature on the K10 changes K10 before treatment and are thought to represent the transit- after treatment with calcipotriol shows a marked increase in amplifying cells that are the direct descendants of the epider- K10 expression, confirming our results.21,29,30 Franssen et al.7 mal stem cell and have a very high proliferative activity.31,32 also found that calcipotriol significantly increased the K10 Furthermore, the number of cell layers containing the b-1 in- expression whereas calcitriol did not. This is in accordance tegrin dim cells is increased in the psoriatic lesion, giving rise with literature showing that treatment with vitamin D deriva- to the thickening of the epidermis.16,31 It is conceivable that tives also tends to normalize other differentiation-related vitamin D derivatives inhibit hyperproliferation in the psoriatic markers as involucrin.9,20 epidermis by controlling the epidermal progenitor cell into a The relative surface area of b-1 integrin compared with the differentiating pathway rather than a hyperproliferative path- total epidermal surface area did not show any statistical way.12

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp130–137 136 Quantitative comparison of calcitriol and calcipotriol ointment, J.E.M. Ko¨rver et al.

The reappearance of K15 in the basal cell layers after treat- predictor for the disease remission period.35 The pharmacoki- ment with calcipotriol further strengthens this hypothesis netic differences between calcitriol and calcipotriol might also because K15 is known to be downregulated in activated basal explain why calcipotriol has a higher incidence of side-effects epidermal cells.33 Treatment with calcipotriol switches the such as skin irritation and erythema than calcitriol. basal progenitor cell towards a differentiating pathway, result- Both calcipotriol and calcitriol are vitamin D derivatives ing in differentiation instead of a pathway resulting in the which are able to inhibit proliferation and induce differenti- typical hyperproliferation seen in psoriasis. In this differentiation in the psoriatic plaque in vitro; in vivo, calcipotriol seems ating pathway the basal keratinocytes become less activated to be more potent in doing so than calcitriol. After 6 weeks of and basal K15 expression returns. The diminishment of the treatment calcipotriol is able to switch the activated basal cells transit-amplifying cell compartment and increase of the differ- from a hyperproliferative state into a differentiating state, lead- entiative cell compartment results in normalization of the ing to a decrease in the transit-amplifying cell compartment structure of the psoriatic epidermis. and finally giving rise to an almost normally differentiated su- Calcitriol showed the same tendency to do this; a slight prabasal cell population and a definite reduction in epidermal increase in K15 was visible with a diminishment of the tran- thickness. Calcitriol probably is also able to do this, but needs sit-amplifying cell compartment which was qualitatively vis- more time to do so. ible. However, the tendency to increase the differentiated cell compartment was not even qualitatively visible (as our dou- Acknowledgments ble-labelling experiments show). This indicates that after treatment with calcitriol there is an induction of the progenitor The authors would like to thank Gijs de Jongh for his assist- cells in the basal layer from a hyperproliferative state into a ance in analysing the data and Tim Smits for his contribution differentiative state, leading to a qualitatively visible decrease in the collection of the skin biopsies. in the transit-amplifying cell compartment. However, this effect is not strong enough to influence the suprabasal cells; References therefore, this does not lead to an increase in suprabasal epidermal differentiation as measured by an increase in K10 1 Bourke JF, Iqbal SJ, Hutchinson PE. A randomized double-blind expression. comparison of the effects on systemic calcium homeostasis of top- Both calcitriol and calcipotriol are vitamin D derivatives ical calcitriol (3 micrograms/g) and calcipotriol (50 micrograms/ g) in the treatment of chronic plaque psoriasis vulgaris. Acta Derm with comparable affinity for the vitamin D receptor. Perhaps Venereol 1997; 77:228–30. an even lower affinity for the vitamin D receptor is found for 2 van de Kerkhof PC, Green C, Hamberg KJ et al. Safety and efficacy 14 calcipotriol compared with calcitriol. Nevertheless, the con- of combined high-dose treatment with calcipotriol ointment and centration of the active compound in calcitriol ointment is not solution in patients with psoriasis. Dermatology 2002; 204:214–21. only more than 10 times lower than that of calcipotriol, it 3 Scott LJ, Dunn CJ, Goa KL. Calcipotriol ointment. A review of its also has a 30 times higher affinity for vitamin D-binding pro- use in the management of psoriasis. Am J Clin Dermatol 2001; 2:95– tein in the blood.34 In vitro experiments show that due to this 120. 4 Ashcroft DM, Po AL, Williams HC et al. Systematic review of com- higher affinity of calcitriol for the vitamin D-binding protein, parative efficacy and tolerability of calcipotriol in treating chronic combined with the higher concentration of the active com- plaque psoriasis. BMJ 2000; 320:963–7. pound in calcipotriol ointment, the local biological activity of 5 Gerritsen MJ, van de Kerkhof PC, Langner A. Long-term safety of calcitriol in the psoriatic epidermis is probably markedly topical calcitriol 3 microg g(–1) ointment. Br J Dermatol 2001; 144 decreased compared with calcipotriol.14 Although calcitriol is (Suppl. 58):17–9. as capable as calcipotriol in inducing differentiation in hyper- 6 Ortonne JP, Humbert P, Nicolas JF et al. Intra-individual comparison proliferative progenitor epidermal cells in vitro, the above-men- of the cutaneous safety and efficacy of calcitriol 3 microg g(–1) ointment and calcipotriol 50 microg g(–1) ointment on chronic tioned mechanisms can be the reason why calcipotriol is the 12 plaque psoriasis localized in facial, hairline, retroauricular or flexu- clinically more powerful compound. Our experiments show ral areas. Br J Dermatol 2003; 148:326–33. that calcitriol is probably capable of inducing the same route 7 Franssen ME, de Jongh GJ, van Erp PE et al. A left/right comparison into differentiation in psoriatic epidermal cells, but does this of twice-daily calcipotriol ointment and calcitriol ointment in less insistently. For calcitriol to be able to achieve similar clin- patients with psoriasis: the effect on keratinocyte subpopulations. ical improvement, it requires more time for the psoriatic le- Acta Derm Venereol 2004; 84:195–200. sion to return to a nonactivated, normally differentiated state. 8 Gerritsen MJ, Rulo HF, Van Vlijmen-Willems I et al. Topical treatment of psoriatic plaques with 1,25-dihydroxyvitamin D3: a cell This may also explain the difference in the potential of calci- biological study. Br J Dermatol 1993; 128:666–73. potriol and calcitriol to re-induce the marker for nonactivated 9 Takahashi H, Ibe M, Kinouchi M et al. Similarly potent action of basal keratinocytes, K15. This has been demonstrated in 1,25-dihydroxyvitamin D3 and its analogues, tacalcitol, calcipotri- another study from our group, which found the reappearance ol, and maxacalcitol on normal human keratinocyte proliferation of K15 to be a relatively late phenomenon in the normaliza- and differentiation. J Dermatol Sci 2003; 31:21–8. tion of the psoriatic epidermis, occurring even after normal- 10 van der Vleuten C, de Jong EM, van de Kerkhof PC. Epidermal dif- ization of differentiation-related markers had taken place. ferentiation characteristics of the psoriatic plaque during treatment with calcipotriol. Arch Dermatol Res 1996; 288:366–72. Interestingly, K15 reappearance also seemed to be a possible

11 Jensen AM, Llado MB, Skov L et al. Calcipotriol inhibits the prolif- proliferating cells in microwave-processed formalin-fixed paraffin eration of hyperproliferative CD29 positive keratinocytes in psoriat- sections. J Pathol 1992; 168:357–63. ic epidermis in the absence of an effect on the function and 23 Weiss RA, Eichner R, Sun TT. Monoclonal antibody analysis of number of antigen-presenting cells. Br J Dermatol 1998; 139:984– keratin expression in epidermal diseases: a 48- and 56-kdalton 91. keratin as molecular markers for hyperproliferative keratinocytes. 12 Gniadecki R. Effects of 1,25-dihydroxyvitamin D3 and its 20-epi J Cell Biol 1984; 98:1397–406. analogues (MC 1288, MC 1301, KH 1060), on clonal keratinocyte 24 Jiang CK, Magnaldo T, Ohtsuki M et al. Epidermal growth factor growth: evidence for differentiation of keratinocyte stem cells and and transforming growth factor alpha specifically induce the activa- analysis of the modulatory effects of cytokines. Br J Pharmacol 1997; tion- and hyperproliferation-associated keratins 6 and 16. Proc Natl 120:1119–27. Acad Sci U S A 1993; 90:6786–90. 13 Uhoda I, Quatresooz P, Hermanns-Le T et al. Histometric assess- 25 Reichrath J, Perez A, Muller SM et al. Topical calcitriol (1,25-dihyd- ment of psoriatic plaques treated by vitamin D3 derivatives. Derma- roxyvitamin D3) treatment of psoriasis: an immunohistological tology 2003; 206:366–9. evaluation. Acta Derm Venereol 1997; 77:268–72. 14 Bouillon R, Allewaert K, Xiang DZ et al. Vitamin D analogs with 26 Berth-Jones J, Fletcher A, Hutchinson PE. Epidermal cytokeratin low affinity for the vitamin D binding protein: enhanced in vitro and immunocyte responses during treatment of psoriasis with cal- and decreased in vivo activity. J Bone Miner Res 1991; 6:1051–7. cipotriol and betamethasone valerate. Br J Dermatol 1992; 126:356– 15 van Duijnhoven MW, van de Kerkhof PC, Pasch MC et al. The com- 61. bination of the Zenon labeling technique and microscopic image 27 Castelijns FA, Gerritsen MJ, van Erp PE et al. Efficacy of calcipotriol analysis to study cell populations in normal and psoriatic epider- ointment applied under hydrocolloid occlusion in psoriasis. Derma- mis. J Cutan Pathol 2005; 32:212–19. tology 2000; 200:25–30. 16 Korver JE, van Duijnhoven MW, Pasch MC et al. Assessment of epi- 28 Lu I, Gilleaudeau P, McLane JA et al. Modulation of epidermal dif- dermal subpopulations and proliferation in healthy skin, symptom- ferentiation, tissue inflammation, and T-lymphocyte infiltration in less and lesional skin of spreading psoriasis. Br J Dermatol 2006; psoriatic plaques by topical calcitriol. J Cutan Pathol 1996; 23:419– 155:688–94. 30. 17 Harrington CI, Goldin D, Lovell CR et al. Comparative effects of 29 van de Kerkhof PC. Reduction of epidermal abnormalities and in- two different calcipotriol (MC903) formulations versus placebo in flammatory changes in psoriatic plaques during treatment with psoriasis vulgaris. A randomised, double-blind, placebo-controlled, vitamin D3 analogs. J Investig Dermatol Symp Proc 1996; 1:78–81. parallel group multi-centre study. J Eur Acad Dermatol Venereol 1996; 30 Vissers WH, Berends M, Muys L et al. The effect of the combin- 11:152–8. ation of calcipotriol and betamethasone dipropionate versus both 18 Kowalzick L. Clinical experience with topical calcitriol (1,25-di- monotherapies on epidermal proliferation, keratinization and T-cell hydroxyvitamin D3) in psoriasis. Br J Dermatol 2001; 144 (Suppl. subsets in chronic plaque psoriasis. Exp Dermatol 2004; 13:106–12. 58):21–5. 31 Franssen ME, Zeeuwen PL, Vierwinden G et al. Phenotypical and 19 Langner A, Ashton P, van de Kerkhof PC et al. A long-term multi- functional differences in germinative subpopulations derived from centre assessment of the safety and tolerability of calcitriol oint- normal and psoriatic epidermis. J Invest Dermatol 2005; 124:373–83. ment in the treatment of chronic plaque psoriasis. Br J Dermatol 32 Watt FM. Epidermal stem cells: markers, patterning and the control 1996; 135:385–9. of stem cell fate. Philos Trans R Soc Lond B Biol Sci 1998; 353:831–7. 20 Sebag M, Gulliver W, Kremer R. Effect of 1,25 dihydroxyvitamin 33 Waseem A, Dogan B, Tidman N et al. Keratin 15 expression in stra- D3 and calcium on growth and differentiation and on c-fos and tified epithelia: downregulation in activated keratinocytes. J Invest p53 gene expression in normal human keratinocytes. J Invest Derma- Dermatol 1999; 112:362–9. tol 1994; 103:323–9. 34 Kissmeyer AM, Binderup L. Calcipotriol (MC 903): pharmacokinet- 21 Glade CP, van Erp PE, Boezeman JB et al. Multiparameter flow cy- ics in rats and biological activities of metabolites. A comparative tometry as a tool to evaluate antipsoriatic therapy. Br J Dermatol study with 1,25(OH)2D3. Biochem Pharmacol 1991; 41:1601–6. 1997; 137:367–75. 35 van Duijnhoven MW, Korver JE, Vissers WH et al. Effect of calcipo- 22 Cattoretti G, Becker MH, Key G et al. Monoclonal antibodies against triol on epidermal cell populations alefacept-treated psoriatic recombinant parts of the Ki-67 antigen (MIB 1 and MIB 3) detect lesions. J Eur Acad Dermatol Venereol 2006; 20:27–33.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp130–137 THERAPEUTICS DOI 10.1111/j.1365-2133.2006.07585.x Comparison of clinical and pharmacokinetic proﬁles of etanercept 25 mg twice weekly and 50 mg once weekly in patients with psoriasis B. Elewski, C. Leonardi,* A.B. Gottlieb, B.E. Strober, M.A. Simiens,§ M. Dunn§ and A. Jahreis§ University of Alabama, Birmingham, Ste 414, 700 18th St S, Birmingham, AL 35233-3805, U.S.A. *St Louis University, 1034 South Brentwood Boulevard, Suite 600, St Louis, MO 63117-1206, U.S.A. Department of Dermatology, Tufts-New England Medical Center, 750 Washington St, Box 114, Boston, MA 02111, U.S.A. Ronald O. Perelman Department of Dermatology, New York University School of Medicine, Ste 7R, 560 1st Avenue, New York, NY 10016-6402, U.S.A. §Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320-1799, U.S.A.

Summary

Correspondence Background Etanercept is a tumour necrosis factor antagonist that is approved in the Boni Elewski. U.S.A., Canada and Europe for treating adult patients with chronic moderate to E-mail: [email protected] severe plaque psoriasis. Objectives To assess whether clinical efficacy, safety and pharmacokinetic (PK) Accepted for publication 31 July 2006 profiles of etanercept 50 mg once weekly are comparable to etanercept 25 mg twice weekly. Key words Methods Patients from a U.S. phase 3 study and a global phase 3 study were subse- clinical trials, efficacy, etanercept, quently enrolled in an open-label extension study (extension study) where they pharmacokinetics, psoriasis, safety all received etanercept at a dose of 50 mg once weekly for an initial 12 weeks. Conflicts of interest Patients who had received at least 24 weeks of etanercept 25 mg twice weekly in B.E. is a clinical investigator and/or consultant the global phase 3 study and were enrolled in the extension study (n ¼ 265) for Amgen Inc., Biogen, Centocor Inc. and were assessed for efficacy and safety at extension study baseline and after Genentech. 12 weeks of etanercept 50 mg once weekly. Efficacy endpoints included the C.L. is a consultant for Abbott, Amgen Inc., Psoriasis Area and Severity Index (PASI), the Dermatology Life Quality Index and Bristol-Myers Squibb, Centocor Inc., Genentech the Physician’s Global Assessment of psoriasis. In addition, PK profiles from and Serono. He has been an investigator for studies sponsored by 3M Pharmaceuticals, Abbott, patients in the U.S. phase 3 study were compared with PK profiles from another Allergan, Amgen Inc., Astellis, Biogen, Bristol- set of patients in the extension study. Comparison was made between a subset of Myers Squibb, CombinatoRx, Centocor Inc., patients receiving etanercept 25 mg twice weekly dosing in the U.S. phase 3 Fujisawa, Galderma, Genentech, MedImmune, study (n ¼ 13) and those receiving etanercept 50 mg once weekly in the exten- RTL, Schering-Plough Corp. and Serono. He has sion study (n ¼ 84). received educational grants from Amgen Inc. and Results The mean PASI score was 5Æ77 at extension study baseline after treatment Genentech. He is on the speakers’ bureau for Abbott, Amgen Inc., Centocor Inc., Genentech and with etanercept 25 mg twice weekly, which was sustained at 5Æ82 after 12 weeks Warner-Chilcott. of etanercept 50 mg once weekly. Similar results were observed in other efficacy A.B.G. is a consultant for several companies endpoints. Etanercept 50 mg once weekly was generally well tolerated. No new (Amgen Inc., BiogenIdec Inc., Centocor Inc., safety findings were reported. PK profiles overlapped extensively between the Wyeth Pharmaceuticals, Schering-Plough Corp., two dosing regimens. Eisai, Celgene Corp., Bristol Myers Squibb Co., Beiersdorf, Inc., Warner Chilcott, Abbott Labs, Conclusions In this report, we demonstrate that efficacy, safety and PK profiles were Roche, Sankyo, Medarex, Kemia, Celera, TEVA, comparable between etanercept 25 mg twice weekly and 50 mg once weekly in Actelion, UCB, Novo Nordisk, Almirall, Immune patients who had received at least 24 weeks of etanercept 25 mg twice weekly Control) and is on the speaker’s bureau for Amgen prior to receiving etanercept 50 mg once weekly in the extension study. Inc. and Wyeth Pharmaceuticals. B.E.S. has been a speaker, consultant and investigator for Amgen Inc. M.A.S., M.D. and A.J. are employees of Amgen Inc.

138 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp138–142 Comparison of etanercept 25 mg twice weekly vs. 50 mg once weekly, B. Elewski et al. 139

Psoriasis is a chronic inflammatory disease associated with Study design substantial physical and psychological disabilities.1,2 Psoriasis manifests as erythematous plaques that can cause considerable The institutional review boards of participating medical cen- pain and discomfort. The visible skin lesions have serious tres approved the protocol. Study design, disposition and social implications that often leave patients stigmatized, lead- results from the blinded portions of two phase 3 studies have ing to a markedly reduced health-related quality of life. been reported previously.6,7 The first phase 3 trial was a glo- Long-term use of traditional psoriasis therapies, including bal study that consisted of a 12-week, double-blind period phototherapy and systemic therapies, has been limited either where three treatment groups were compared (etanercept because of ineffectiveness, drug-induced toxicities, or time- 50 mg twice weekly, etanercept 25 mg twice weekly or pla- consuming treatment regimens. Consequently, many patients cebo), followed by up to 36 weeks of open-label period have not been satisfied with these treatments.3,4 The tumour where all patients received etanercept 25 mg twice weekly.7 necrosis factor (TNF) antagonist etanercept offers patients with Patients described in this report received at least 24 weeks of psoriasis an effective therapy with a favourable risk-benefit etanercept 25 mg twice weekly prior to enrolling in the ratio. Etanercept is a soluble TNF receptor that binds and neut- extension study (Fig. 1). In the second phase 3 study (U.S. ralizes endogenous TNF molecules,5 and has been shown to phase 3 study), four treatment groups were compared (etaner- be safe and effective in patients with psoriasis in two random- cept 50 mg twice weekly, etanercept 25 mg twice weekly, et- ized, controlled, phase 3 studies, referred to as the U.S. phase anercept 25 mg once weekly and placebo) during a 24-week, 3 study and the global phase 3 study.6,7 Patients from these double-blind period, followed by continuous or intermittent two studies were subsequently enrolled in an open-label dosing of etanercept during open-label treatment for up to extension study for up to 72 weeks where they received 72 weeks of total treatment.6 Details of the initial 24 weeks of 50 mg once weekly dosing for the first 12 weeks, with the the global phase 3 study and U.S. phase 3 study are described option of increasing their dose to 50 mg twice weekly there- elsewhere.6,7 Both studies were terminated early to allow after. In this report, we assessed the clinical efficacy and safety patients to enrol in a separate, long-term, open-label extension of the first 265 enrolled patients who received etanercept study where they received etanercept 50 mg once weekly 25 mg twice weekly in the global phase 3 study7 and subse- (2 injections of 25 mg/mL at the same time) for 12 weeks. quently converted to etanercept 50 mg once weekly in the After the first 12 weeks of the extension study, patients had extension study. the option of increasing their dose to 50 mg twice weekly or We also assessed the pharmacokinetic (PK) profile of a maintaining their dose at 50 mg once weekly. small cohort of patients while receiving etanercept 25 mg twice weekly in the U.S. phase 3 study6 and the PK profile of Endpoints another cohort of patients while receiving etanercept 50 mg once weekly in the extension study. In this analysis, clinical efficacy and safety data for the initial 12 weeks of the extension study were collected for the first Patients and methods 265 patients who enrolled within 15 days of completing the global phase 3 trial. All 265 patients had received at least 24 weeks of etanercept 25 mg twice weekly prior to entering Patients the extension study. Efficacy and safety of 25 mg twice weekly Adult patients were eligible if they had active, stable plaque and 50 mg once weekly dosing were analysed by comparing psoriasis [‡ 10% of the body surface area (BSA) and Psoriasis data from these 265 patients at extension baseline and at Area and Severity Index (PASI) score ‡ 10 at screening of ori- 12 weeks. The primary endpoint was the incidence of adverse ginal studies], and were candidates to receive phototherapy or events, serious adverse events (SAEs), infectious events and systemic therapy in the investigator’s opinion. Patients were serious infectious events (SIEs). Other endpoints included not to have received systemic psoriasis therapy or psoralen mean change in the PASI, PASI 50/75/90 (representing 50%, plus ultraviolet (UV) A phototherapy for 4 weeks before the 75% or 90% improvement of disease severity from global study; topical corticosteroids, vitamin A or D analogue prepa- phase 3 study baseline), Dermatology Life Quality Index rations, dithranol or UVB phototherapy for 2 weeks before (DLQI; scale of 0–30, with higher scores indicative of worse the study; or etanercept or an anti-TNF antibody at any time. outcome), the Physician’s Global Assessment (PGA; scale of 0– Patients were allowed to use topical corticosteroids of moder- 5; 0 ¼ clear, 5 ¼ severe) of Clear or Almost Clear psoriasis, ate strength on the scalp, axilla and groin, or tar compound and the Patient’s Global Assessment (PtGA; scale of 0–5, with or steroid-free topical emollients. higher scores indicative of worse outcome) score of 0 or 1. PK data for patients receiving etanercept 50 mg once weekly in the extension study (n ¼ 84) were compared with Treatment PK data for a subset of patients receiving etanercept 25 mg In both phase 3 studies and the extension study, etanercept was pre- twice weekly dosing in the U.S. phase 3 study (n ¼ 13). sented as a sterile lyophilized powder (25mg/mL). All study drugs These 13 patients had extensive PK sampling at steady state, were self-administered by the patient by subcutaneous injections. permitting direct comparisons with 50 mg once weekly

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp138–142 140 Comparison of etanercept 25 mg twice weekly vs. 50 mg once weekly, B. Elewski et al.

Fig 1. Study designs. BIW, twice weekly.

dosing. Only 13 patients were available because PK sampling 20Æ5 years (Table 1). Mean affected BSA was 26Æ6% and 28% and analysis was not planned but was added to an amended of the patients had psoriatic arthritis. study protocol, and these were the only patients who had extensive PK sampling as patients were rolled over to the exten- Efficacy and safety sion study. Patients described in this report received at least 12 weeks of etanercept 25 mg twice weekly prior to their PK The mean PASI score improved from 18Æ43 at global phase 3 profile. Blood samples were obtained for up to 168 h post- study baseline to 5Æ77 at extension study baseline after treat- dose. Etanercept serum concentrations were determined by an ment with etanercept 25 mg twice weekly for at least enzyme-linked immunosorbent assay. PK parameters were cal- 24 weeks, representing a mean improvement of 67%. After culated by noncompartmental analysis (NCA) of the serum 12 weeks of etanercept 50 mg once weekly, the mean PASI concentration-time data (WinNonlin Professional Version 3Æ3, score was sustained at 5Æ82, a mean difference of )0Æ06 (95% Pharsight Corp., Mountain View, CA, U.S.A.) using the Steady CI )0Æ40 to 0Æ29, P ¼ 0Æ76) between the two dosing regi- State Option of the NCA analysis tool. For the calculation of mens (Fig. 2a). At entry into the extension study, after at least the NCA parameters, the actual sampling times were used. 24 weeks on 25 mg twice weekly dosing, patients had the following PASI 50/75/90 responses: 118 patients (45%) had a PASI-75 response, 87 (33%) had between a PASI-50 and PASI- Statistical analysis 75 response, and 60 (23%) had a less than PASI-50 response. All analyses were based on observed cases. Comparisons of ef- Eighty-four percent of PASI-75 responders at extension study ficacy between etanercept 25 mg twice weekly and 50 mg once weekly dosing were performed using the t-test. Table 1 Baseline demographics

Results Total (n ¼ 265) Age, mean years (SD) 45Æ6 (12Æ1) Patients Men, % 63 White, % 89 Two hundred and sixty-ﬁve patients from the global study Duration of psoriasis, mean years (SD) 20Æ5 (12Æ0) who enrolled in the extension study were assessed for clinical Affected body surface area, mean % (SD) 26Æ6 (17Æ1) efﬁcacy and safety after 12 weeks of open-label treatment. The Reported history of psoriatic arthritis, n (%) 74 (28) group was predominantly male (63%) and white (89%), with Patients with prior psoriasis therapy, n (%) 225 (85) a mean age of 45Æ6 years and a mean duration of psoriasis of

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp138–142 Comparison of etanercept 25 mg twice weekly vs. 50 mg once weekly, B. Elewski et al. 141

(a) tious events. These rates were less than those previously reported for patients receiving etanercept 25 mg twice weekly during the blinded global phase 3 study, in which 58% and 30% of patients receiving etanercept 25 mg twice weekly reported adverse events and infectious events, respectively, after 12 weeks of therapy. No SIEs were reported during the first 12 weeks of the extension study. One percent of patients reported SAEs after converting to 50 mg once weekly dosing, compared with 3% of patients receiving etanercept 25 mg twice weekly during the blinded study. No new safety findings were observed and there were no reports of tuberculosis (b) or opportunistic infections. In conclusion, conversion from 25 mg twice weekly to 50 mg once weekly for 12 weeks resulted in a comparable safety profile of both dosing regimens.

Pharmacokinetics

PK data were similarly comparable for the two dosing regimens. The steady-state concentration–time proﬁles resulting from the two etanercept dosing regimens overlapped extensively over the 1-week sampling interval (Fig. 3). Overall exposure, as measured by Area Under the Curve (AUC) and

average concentration (Cave), was also comparable between Fig 2. Sustained improvement in (a) Psoriasis Area and Severity Index the two dosing regimens (data not shown). (PASI) (± SE) and (b) Dermatology Life Quality Index (DLQI) responses (± SE) after converting from etanercept 25 mg twice weekly to etanercept 50 mg once weekly. BIW, twice weekly. Discussion

Etanercept has been shown in previous clinical trials to be safe baseline maintained this high response after 12 weeks of etan- and effective in treating patients with psoriasis.6,7 Most of ercept 50 mg once weekly. Among those achieving between these patients were treated with etanercept 25 mg, adminis- PASI-50 and PASI-75 improvement, 59% maintained the tered as a subcutaneous injection twice weekly. Based on the response after 12 weeks of open-label 50 mg once weekly results of these trials, etanercept was approved for the treat- dosing, with 20% achieving a PASI-75 or greater response and ment of adult patients with chronic moderate to severe plaque 21% achieving a less than PASI-50 response. In addition, psoriasis who are candidates for systemic therapy or photo- among patients who did not achieve a PASI-50 response with therapy. etanercept 25 mg twice weekly, 34% achieved PASI-50 or Herein, we report for the first time that efficacy, safety greater response after treatment with etanercept 50 mg once and PK profiles achieved with a dosing of etanercept 25 mg weekly. The percentages of patients who achieved a PGA Clear or Almost Clear status were also sustained 12 weeks after switching from 25 mg twice weekly (43%) to 50 mg once weekly (38%). In addition, improvements in patient-reported outcomes were similarly sustained after dose conversion, with mean DLQI scores improving from 10Æ77 at global phase 3 study baseline to 3Æ13 at extension study baseline to 3Æ03 at week 12 of the extension study, a mean difference of 0Æ10 (95% CI )0Æ18 to 0Æ39, P ¼ 0Æ48) (Fig. 2b). The proportion of patients who achieved a PtGA score of 0 or 1 was also maintained 12 weeks after switching from 25 mg twice weekly (47%) to 50 mg once weekly (52%). During the extension study, 42% of patients reported at least one adverse event and 23% of patients reported infectious events after 12 weeks of etanercept 50 mg once weekly, Fig 3. Mean (+ SD) profiles of etanercept at steady state following with nasopharyngitis (3Æ4%) and upper respiratory tract infec- 25 mg twice weekly and 50 mg once weekly administration. BIW, tion (3Æ4%) representing the most frequently reported infec- twice weekly.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp138–142 142 Comparison of etanercept 25 mg twice weekly vs. 50 mg once weekly, B. Elewski et al. twice weekly were substantially sustained with a dosing of References 50 mg once weekly in patients with moderate to severe 1 Krueger JG. The immunologic basis for the treatment of psoriasis plaque psoriasis. It should be noted that the overall exposure with new biologic agents. J Am Acad Dermatol 2002; 46:1–23; quiz to the 50 mg once weekly dosing is relatively short at 23–6. 12 weeks compared with up to 48 weeks on etanercept 2 Nickoloff BJ, Nestle FO. Recent insights into the immunopathogen- 25 mg twice weekly. Therefore, additional time on the 50 mg esis of psoriasis provide new therapeutic opportunities. J Clin Invest once weekly dosing may be required to truly discern 2004; 113:1664–75. differences in efficacy and safety between the two dosing 3 Krueger G, Koo J, Lebwohl M et al. The impact of psoriasis on qual- regimens. Nevertheless, the results described are consistent ity of life: results of a 1998 National Psoriasis Foundation patient- membership survey. Arch Dermatol 2001; 137:280–4. with a previous study of patients with rheumatoid arthritis, 4 Stern RS, Nijsten T, Feldman SR et al. Psoriasis is common, carries a whereby comparable efficacy, safety and PK outcomes were substantial burden even when not extensive, and is associated with demonstrated when patients were administered with either of widespread treatment dissatisfaction. J Investig Dermatol Symp Proc these dosing regimens.8 2004; 9:136–9. In clinical practice, some patients may prefer twice weekly 5 Enbrel (etanercept). Package Insert. Thousand Oaks, CA: Amgen Inc. injections of 25 mg etanercept administered 3 or 4 days apart. and Wyeth-Ayerst Pharmaceuticals, 2004. However, this report may be of interest to patients and der- 6 Leonardi CL, Powers JL, Matheson RT et al. Etanercept as monotherapy in patients with psoriasis. N Engl J Med 2003; 349:2014–22. matologists as once-weekly dosing of etanercept may improve 7 Papp KA, Tyring S, Lahfa M et al. A global phase III randomized the convenience of use for patients with psoriasis. The avail- controlled trial of etanercept in psoriasis: safety, efficacy, and effect ability of etanercept in a 50 mg liquid prefilled syringe can of dose reduction. Br J Dermatol 2005; 152:1304–12. reduce the injections needed by half. In addition, the ability 8 Keystone EC, Schiff MH, Kremer JM et al. Once-weekly administra- to administer a single injection of etanercept 50 mg once tion of 50 mg etanercept in patients with active rheumatoid weekly could potentially increase patient compliance with the arthritis: results of a multicenter, randomized, double-blind, treatment regimen. placebo-controlled trial. Arthritis Rheum 2004; 50:353–63.

Acknowledgments

We thank Ting Chang for editorial support. This research was funded by Amgen Inc. and Wyeth Research.

2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp138–142 CASE REPORT DOI 10.1111/j.1365-2133.2006.07516.x Nephro-urological complications of epidermolysis bullosa in paediatric patients S.M.H. Chan, M.J. Dillon,* P.G. Duffy and D.J. Atherton Departments of Dermatology, *Nephrology and Urology, Great Ormond Street Hospital for Children, London WC1N 3JH, U.K.

Summary

Correspondence A small but important proportion of patients with epidermolysis bullosa (EB) E-mail: [email protected] may develop significant renal and urological complications which can have a major impact on their morbidity and mortality. During the last 10 years, five of Accepted for publication a large group of children with EB under our care, with either dystrophic or junc- 2 April 2006 tional types of disease, experienced major nephro-urological complications. Two Key words patients with recessive dystrophic EB (REDB) developed macroscopic haematuria epidermolysis bullosa, kidney, paediatric, surgery, - one had renal failure and underwent a renal biopsy showing IgA nephropathy. urogenital tract, urology A third patient with RDEB also developed renal failure and his biopsy demonstrated postinfectious glomerulonephritis ⁄type III membranoproliferative (mesan- Conflicts of interest giocapillary) glomerulonephritis. Both patients with renal failure underwent None declared. peritoneal dialysis. Two patients with junctional EB developed obstructive uropathies, which required bladder reconstruction and the fashioning of a Mitrofanoff channel in one.

Epidermolysis bullosa (EB) is the name given to a group of Recessive dystrophic epidermolysis bullosa genetic disorders characterized by trauma-induced separation of the skin and mucosae from the underlying tissues. In the Patient 1: Renal failure—postinfectious dystrophic forms (DEB), this separation occurs immediately glomerulonephritis/type III membranoproliferative below the lamina densa of the basement membrane zone as a (mesangiocapillary) glomerulonephritis result of mutations of the COL7A1 gene on chromosome 3 (3p21.1 locus). The clinical spectrum is variable, severity Patient 1 was diagnosed soon after birth with RDEB. He lived being generally greater when DEB is inherited as an autosomal in Italy and first presented to our dermatology service at the recessive trait. Junctional EB (JEB) reflects separation of the age of 4 years, initially with digital fusion and skin infections. epithelium within the basement membrane zone through the At 14Æ5 years, he presented to St Thomas’s Hospital with a lamina lucida. All forms are inherited as autosomal recessive 1-year history of malaise and fatigue with evidence of renal traits and three broad categories exist. The Herlitz type is the and secondary cardiac failure. Laboratory parameters revealed ) commonest and frequently results in death in the first 2 years the following: haemoglobin 6 g dL 1 (normal 11Æ5– ) ) of life. Non-Herlitz JEB is compatible with survival to adult- 15Æ5gdL 1), serum potassium 6Æ6 mmol L 1 (3Æ3– ) ) ) hood. A third subgroup is associated with pyloric atresia (JEB/ 4Æ5 mmol L 1), urea 46 mmol L 1 (2Æ5–6Æ7 mmol L 1), cre- ) ) PA) and often causes death in infancy. atinine 890 lmol L 1 (39–70 lmol L 1), complement C3 ) ) ) Great Ormond Street Hospital for Children is a tertiary 0Æ78 g L 1 (0Æ75–1Æ65 g L 1), complement C4 0Æ34 g L 1 ) referral unit for paediatric patients with EB. Here we present (0Æ14–0Æ54 g L 1), with negative antineutrophil antibodies five patients with renal disease occurring in the last 10 years. (ANCA) (< 1 : 40 titre), antistreptolysin-O antibody titre Two of the three patients with recessive dystrophic EB (RDEB) (ASOT) (< 200), and an antideoxyribonuclease B (anti DNase had renal biopsies demonstrating one case of postinfectious B) level of 480 units (< 300). Staphylococci and streptococci were glomerulonephritis/type III membranoproliferative glomerulo- isolated from skin swabs. Cardiac failure with poor left nephritis and one of IgA nephropathy. Both of these patients ventricular contraction was evident on chest X-ray and echo- are also the first such reported cases to be managed by perito- cardiography. An ultrasound demonstrated a horseshoe kidney. neal dialysis. Two patients with JEB and obstructive urological His renal biopsy showed that 15 of 35 glomeruli were lesions are also described, and a novel method for surgical obsolete, and a great majority of the remainder showed very alleviation is described for the first time in the literature. A severe chronic damage, although three showed cellular cres- summary of our five patients with nephro-urological involve- cents. One glomerulus was almost normal, but all glom- ment, with type of EB and outcomes is given in Table 1. eruli showed increased cellularity, mesangial expansion and

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp143–147 143 144 Nephro-urological complications of EB in paediatric patients, S.M.H. Chan et al.

) ) Table 1 Summary of patients with nephro-urological involvement, 649 IU mL 1 and > 3440 U mL 1. She had received repeated with type of EB and outcomes courses of antibiotics including flucloxacillin and amoxicillin. Complement levels during the acute phase and an autoim- Type mune screen were normal, while IgG was markedly raised at ) ) Patient no. of EB Diagnosis Outcome 38 g L 1 (normal 5Æ3–16Æ5gL 1) and IgA was significantly ) ) 1 RDEB Postinfectious Renal failure increased at 4Æ72 g L 1 (0Æ8–4Æ0gL 1). She had a normal glomerulonephritis renal ultrasound and plain abdominal X-ray. Patient 2 and her Type III family declined further investigation in the way of a formal membranoproliferative glomerular filtration rate (GFR) examination or renal biopsy. (mesangiocapillary) glomerulonephritis Patient 2 had clinical features that were in keeping with 2 RDEB Undiagnosed Resolution poststreptococcal glomerulonephritis, although these were not 3 RDEB IgA nephropathy Renal failure entirely typical. Without histological confirmation, an IgA-type 4 JEB/PA Obstruction Surgery nephropathy associated with skin infections and other renal 5 JEB Hypospadias, meatal stenosis Surgery conditions unrelated to her skin disease could not be excluded. EB, epidermolysis bullosa; RDEB, recessive dystrophic EB; JEB, junctional EB; PA, pyloric atresia. She was commenced on penicillin prophylaxis in an attempt to reduce cutaneous streptococcal infections. The microscopic haematuria resolved. Four years later, as only MRSA (methicil- segmental double contouring of capillary walls. There was lin-resistant Staphylococcus aureus) was detected on skin swabs severe widespread chronic atrophy of the tubules, severe scar- and her biochemical parameters of renal function remained ring and a patchy mononuclear cell infiltrate in the intersti- unimpaired, penicillin was discontinued. On her last review in tium, and severe hyaline change in the arterioles with normal July 2003, at the age of 17 years, she had had no further interlobular arteries. There was coarse granular localization of problems, normal biochemical renal function and no evidence moderate amounts of C3 alone on capillary walls with im- of macroscopic or microscopic haematuria. munoperoxidase. These findings could be in keeping with end-stage renal disease with proliferative changes and comple- Patient 3: Haematuria ⁄ renal failure - IgA nephropathy ment staining consistent with postinfectious glomerulonephritis, although the history and histological findings of segmental Patient 3 presented at birth in Syria with RDEB. She developed double contouring and the immunostaining data would also gastro-oesophageal reflux, oesophageal strictures, failure to be consistent with the so-called type III membranoproliferative thrive, digital fusion, lower limb contractures and impaired (mesangiocapillary) glomerulonephritis. mobility. She had recurrent skin infections with S. aureus and A peritoneal dialysis catheter was initially inserted and ultra- later MRSA. At the age of 13 years, a squamous cell carcinoma filtration performed. A Tenckhoff catheter was later inserted to on the left knee was successfully excised. She developed permit peritoneal dialysis which continued for 3 years after he splenomegaly, required multiple blood transfusions and ery- returned home to Italy. Peritoneal infections and adhesions led thropoietin for anaemia. to a change to haemodialysis for a year, which was not well At the age of 11 years, she developed macroscopic haema- tolerated. Malnutrition and anaemia complicated his course turia without proteinuria. Her abdominal X-ray was normal, and he died at the age of 20 years as a consequence of fluid while a renal ultrasound demonstrated enlarged kidneys. Her overload and cardiac failure. family declined a renal biopsy. By the age of 13, she had developed slight proteinuria and microscopic haematuria and evidence of increasingly enlarged kidneys with increased echo- Patient 2: Haematuria—poststreptococcal ) genicity on ultrasound. An impaired GFR of 34 mL min 1 glomerulonephritis/IgA nephropathy ) ) ) 1Æ73 m 2 (normally c. 100 mL min 1 1Æ73 m 2) preceded a Patient 2 presented in the early neonatal period with blisters, decline in her biochemical parameters as her urea rose to ) ) and a diagnosis of autosomal RDEB was made. Significant fea- 11 mmol L 1 and her creatinine to 228 lmol L 1. Comple- tures included oesophageal strictures, constipation and digital ment C3 and C4, IgA, ASOT anti-DNase B levels and autoanti- fusion. bodies were within normal limits, and investigations for She developed marked macroscopic haematuria at the age amyloid were negative. She subsequently underwent a renal of 11 years. This feature became intermittent, although micro- biopsy which demonstrated IgA nephropathy. scopic haematuria persisted in the absence of significant pro- A Tenchkoff catheter was inserted for peritoneal dialysis. teinuria, hypertension or abnormality in serum renal Renal transplantation was discussed with the family, but biochemistry. She had a history of recurrent skin infections deemed hazardous in view of the increased risk of infection with groups A and G b-haemolytic streptococci, Pseudomonas aerugi- and cutaneous malignancy associated with immunosuppres- nosa and staphylococci, with raised ASOT and anti-DNase B levels sion. She died from sepsis 4 months after the insertion of the ) ) ) of 804 IU mL 1 (normal < 300 IU mL 1) and 3090 U mL 1 peritoneal catheter, aged 15 years. Postmortem examination ) (< 200 U mL 1) respectively. Levels 6 weeks later were was not performed for religious reasons.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp143–147 Nephro-urological complications of EB in paediatric patients, S.M.H. Chan et al. 145

In the course of a renal consultation, patient 3’s elder sister, He was noted to have a coronal hypospadias and underwent who also had RDEB, revealed that she too had experienced a two-stage Thiersch–Duplay repair with good results at the haematuria that she had never disclosed. She was anaemic but age of 4 years. Two years later, he developed dysuria associ- had normal renal biochemistry. In the absence of any inter- ated with a reluctance to micturate, and a poor urinary stream. vention, she had been discharged at the age of 17 to the adult Flow rates showed a poor flow with small volumes and high team, without apparent sequelae. residuals. A urethral dilatation was performed and a suprapubic catheter was inserted, although this became dislodged 3 days later. He continued to have severe dysuria. It was pos- Junctional epidermolysis bullosa tulated that a true urethral stricture or a stricture at the site of the hypospadias repair was the cause of his symptoms. An Patient 4 ultrasound demonstrated normal kidneys and a smooth, 4-mm Patient 4 was born at 35 weeks’ gestation following a preg- thick bladder wall, and a dimercaptosuccinic acid (DMSA) nancy complicated by polyhydramnios. He developed nonbil- scan was normal. A cystourethroscopy suggested a urethral ious vomiting and underwent a gastroduodenostomy for PA. stenosis that was dilated. Continuing symptoms despite a Skin biopsy demonstrated separation at the dermo-epidermal further dilatation and simple meatotomy led to the placement junction, with immunohistochemistry showing normal immu- of another suprapubic catheter. Since then, he has developed noreactivity for laminin 5 (GB3), collagen type VII (LH7Æ2) recurrent urinary tract infections, bladder soreness and pain and plectin (7A8), findings consistent with the known muta- on clamping the catheter, and continues to pose a manage- tions in the b4 integrin gene in patients with JEB with PA. ment dilemma. Other problems included an inflammatory arthritis presenting at the age of 8 years, and failure to thrive. Discussion At the age of 2Æ5 years, he developed severe dysuria with haematuria and loss of epithelial tissue via the urethra. Cystos- There have been a limited number of case reports of the renal copies demonstrated evidence of urethral and ureteral obstruc- and urological manifestations of EB. Some suggest that neph- tion as a result of oedema, bullae and mucosal thickening. A ro-urological pathology may be present from early neonatal suprapubic catheter was inserted, which was well tolerated, life, sometimes before development of overt nephro-urological but over a period of time he developed urinary colonization manifestations.1,2 and occasional symptomatic urinary infections with coliforms Renal amyloidosis has been described in the literature in a and Pseudomonas aeruginosa. Bilateral vesicoureteric reflux was number of patients, with the majority of reports in patients demonstrated on micturating cystourethrogram (MCUG) and with RDEB presenting in the older teenage/young adult age serial isotope-labelled technetium MAG 3 renogram (MAG 3) group.3–9 All were managed symptomatically, and autopsy scans demonstrated a gradual reduction in function of the findings, where noted, invariably reported systemic and renal right kidney to 9% at the age of 9 years. Later, ureteric calculi deposition of amyloid A, the form associated with chronic were noted and were attributed to chronic infection. Other inflammation. The same type of renal problem has been complications included severe bladder spasms and pain with reported in patients with severe dominant DEB10 and JEB.11 bladder instability, and hypertension. Biochemical parameters Specific case reports have highlighted obstructive lesions of renal function remained relatively normal throughout, with including vesicoureteric junction obstruction, urethral stenosis, ) ) a GFR measured at 70 mL min 1 1Æ73 m 2 (normally 89– meatal and labial stenosis. Vesicoureteric junction obstruction ) ) 165 mL min 1 1Æ73 m 2). has been reported in patients with JEB/PA,12–14 some of Hypertension was well controlled on relatively low doses of whom have undergone ureterostomies15–18 or ureteral reim- atenolol and nifedipine (Adalat Retard; Bayer plc, Newbury, plantation15,16 with varying degrees of success. Two patients Berks, U.K.). A percutaneous nephrolithotomy and cystoscopy with JEB11 (including one case associated with PA19) demon- enabled soft calcium and magnesium ammonium phosphate strated evidence of obstruction mainly at the level of the pos- stones to be removed. Bladder pain and spasm were initially terior urethral valve, and were treated with repeated urethral managed medically. Eventually, at the age of 12 years, an ileo- dilatations,11 electroresection, urethral reconstruction, tempor- cystoplasty to reconstruct the bladder, appendix Mitrofanoff ary vesicostomy, ureterosigmoideostomy and nephrostomy19 channel and right nephrectomy were performed. Clinically, he to alleviate symptoms. Meatal stenosis in patients with RDEB20 has coped well with this and inserts a catheter through the and JEB21 has also been documented in the literature, as has abdominal wall into the reconstructed bladder to empty the its management by meatotomy,20,22 ureterosigmoidostomy, contents four times per day. intermittent catheterization and periurethral urethrostomy.21 A case of labial fusion in a patient with DEB has also been described.23 Patient 5 Other reports have included patients with RDEB with clin- Patient 5 developed denudement of the scalp following a vent- ical (in one case)24 and histological (one case)25 evidence of ouse delivery, and blisters affecting his hands and back. Biopsy IgA nephropathy, one case of post-infectious glomerulone- indicated a diagnosis of non-Herlitz JEB. phritis3 secondary to staphyloccal colonization and, in a

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp143–147 146 Nephro-urological complications of EB in paediatric patients, S.M.H. Chan et al. patient with JEB/PA, a case of focal segmental glomeruloscle- ureters. If this is well tolerated, complex urological reconstruc- rosis.26 Other individual case reports have been in patients tion may subsequently be performed without major problems. with JEB, mainly those with JEB/PA. Problems have included Surgical correction of minor abnormalities at the tip of the ure- renal tubular acidosis,13 renal calculi in obstructed systems19 thra should not be attempted. (described in one patient with JEB/PA and another with Her- Our experience would suggest that nephro-urological com- litz JEB11), urinary tract infections in two siblings with unob- plications can cause significant morbidity in a small propor- structed systems,13 bladder spasms and irritability.17 External tion of paediatric patients with EB. In keeping with the genital abnormalities have included hypospadias1 in one findings of Fine’s group,28 this tends to occur more frequently patient, and bladder extrophy, epispadias, bilateral inguinal in patients with the severe subtypes of EB, notably JEB and hernias with an anteriorized anus27 in another. RDEB. However, our patients are a tertiary population drawn A report of 3280 patients followed up in the National Epi- from various secondary referral centres around the world, and dermolysis Bullosa Registry in the U.S.A.28 demonstrated that interpretation of epidemiological data should be viewed in urinary tract complications occurred mainly in those patients this context. with more severe forms of EB. Urethral meatus stenosis, the most common complication, occurred in 11Æ6% and 8Æ0% of References patients with Herlitz JEB and RDEB, respectively. Urinary retention, hydronephrosis and bladder hypertrophy occurred 1 Chang CH, Perrin EV, Bove KE. Pyloric atresia associated with epi- in 9Æ3%, 7Æ0% and 4Æ6% of Herlitz JEB cases, respectively. Py- dermolysis bullosa: special reference to pathogenesis. Pediatr Pathol elonephritis and cystitis were most often seen in the setting of 1983; 1:449–57. 2 Schachner L, Lazarus GS, Dembitzer H. Epidermolysis bullosa he- generalized EB simplex (Koebner variant) and inversa RDEB. reditaria letalis: pathology, natural history and therapy. Br J Dermatol Drawing together all these reports, there appear to be three 1977; 96:51–8. major causes of nephro-urological involvement in patients 3 Mann JF, Zeier M, Zilow E et al. The spectrum of renal involvement with EB.29 Firstly, there are the acute or chronic glomerulopa- in epidermolysis bullosa dystrophica hereditaria: report of two thies, probably triggered by chronic skin infections, secondly, cases. Am J Kidney Dis 1988; 11:437–41. amyloidosis, probably provoked by chronic antigen stimula- 4 Gunduz K, Vatansever S, Turel A et al. Recessive dystrophic epi- tion from the skin, and, thirdly, obstructive uropathies. dermolysis bullosa complicated with nephrotic syndrome due to secondary amyloidosis. Int J Dermatol 2000; 39:151–3. We have presented our experience of two patients with 5 Bourke JF, Browne G, Gaffney EF et al. Fatal systemic amyloidosis RDEB with renal failure, secondary to postinfectious glome- (AA type) in two sisters with dystrophic epidermolysis bullosa. J rulonephritis/type III membranoproliferative (mesangiocapil- Am Acad Dermatol 1995; 33:370–2. lary) glomerulonephritis and IgA nephropathy, and their 6 Yi S, Naito M, Takahashi K et al. Complicating systemic amyloidosis management with peritoneal dialysis. A third patient with in dystrophic epidermolysis bullosa, recessive type. Pathology 1988; RDEB and undiagnosed haematuria was managed conserva- 20:184–7. tively and her symptoms resolved spontaneously. There seems 7 Brownstein MH, Helwig EB. Systemic amyloidosis complicating dermatoses. Arch Dermatol 1970; 102:1–7. to be reasonably good evidence that infective episodes can be 30 8 Csikos M, Orosz Z, Bottlik G et al. Dystrophic epidermolysis bullosa linked to the onset of IgA nephropathy and to recrudes- complicated by cutaneous squamous cell carcinoma and pulmonary cences of disease activity. Where the infective agent has been and renal amyloidosis. Clin Exp Dermatol 2003; 28:163–6. identified there could be a case for prophylaxis. Anecdotal evi- 9 Kaneko K, Kakuta M, Ohtomo Y et al. Renal amyloidosis in reces- dence suggests that this may reduce, for example, episodes of sive dystrophic epidermolysis bullosa. Dermatology 2000; 200:209– macroscopic haematuria, although there are no controlled tri- 12. als to support this approach. Poststreptococcal glomerulone- 10 Dunnill MG, Mallett RB, Hawkins PN et al. Severe dominant dystrophic epidermolysis bullosa complicated by systemic amyloidosis. phritis has also been reported to occur as a superimposed 31 Br J Dermatol 1993; 128:708–9. phenomenon on top of established IgA nephropathy. 11 Ridley CM, Levy IS. Epidermolysis bullosa and amyloidosis: a case We have also discussed our experience with the surgical report. Trans St Johns Hosp Dermatol Soc 1968; 54:75–82. management of obstructive uropathies. In patient 4, with JEB, a 12 El Shafie M, Stidham GL, Klippel CH et al. Pyloric atresia and epi- definitive procedure was delayed because of the uncertainty of dermolysis bullosa letalis: a lethal combination in two premature the eventual outcome. In practice, the operation was successful newborn siblings. J Pediatr Surg 1979; 14:446–9. in alleviating his symptoms and managing his urinary tract 13 Hayashi AH, Galliani CA, Gillis DA. Congenital pyloric atresia and junctional epidermolysis bullosa: a report of long-term sur- under difficult circumstances. In retrospect, performing the vival and a review of the literature. J Pediatr Surg 1991; procedure earlier might have avoided the complications of 26:1341–5. chronic urinary sepsis caused by the indwelling suprapubic 14 Price AP, Hanna M, Katz DS. Epidermolysis bullosa of the bladder. catheter. In patient 5, also with JEB, surgical correction of the AJR Am J Roentgenol 2001; 177:1486–7. hypospadias led to a series of events that may have contributed 15 Bull MJ, Norins AL, Weaver DD et al. Epidermolysis bullosa—pylor- to urethral stenosis and a disproportionate degree of dysuria, ic atresia. An autosomal recessive syndrome. Am J Dis Child 1983; resulting in the placement of a suprapubic catheter. From our 137:449–51. 16 Reitelman C, Burbige KA, Mitchell ME et al. The urological mani- experience, suprapubic catheterization should be the initial step festations of epidermolysis bullosa. J Urol 1986; 136:1320–2. where there is evidence of involvement of the bladder base and

17 Berger TG, Detlefs RL, Donatucci CF. Junctional epidermolysis bul- 26 Kambham N, Tanji N, Seigle RL et al. Congenital focal segmental losa, pyloric atresia, and genitourinary disease. Pediatr Dermatol glomerulosclerosis associated with beta4 integrin mutation and 1986; 3:130–4. epidermolysis bullosa. Am J Kidney Dis 2000; 36:190–6. 18 Rosenberg D, Dodat H, Cottin X et al. Involvement of the urinary 27 Moretti G, Mazzaglia E, D’Anieri A et al. Epidermolysis bullosa tract in a syndrome of congenital epidermolysis bullosa and atresia junctionalis associated with urinary bladder exstrophy: a case of the pylorus. Arch Fr Pediatr 1987; 44:867–70. report. Pediatr Dermatol 1995; 12:239–41. 19 Eklof O, Parkkulainen K. Epidermolysis bullosa dystrophica with 28 Fine JD, Johnson LB, Weiner M et al. Genitourinary complications urinary tract involvement. J Pediatr Surg 1984; 19:215–17. of inherited epidermolysis bullosa: experience of the National Epi- 20 Kretkowsk RC. Urinary tract involvement in epidermolysis bullosa. dermylosis Bullosa Registry and review of the literature. J Urol Pediatrics 1973; 51:938–41. 2004; 172:2040–4. 21 Glazier DB, Zaontz MR. Epidermolysis bullosa: a review of the 29 Atherton DJA, Martinez A, Mellerio J. Clinical management of chil- associated urological complications. J Urol 1998; 159:2122–5. dren and adults with epidermolysis bullosa. Proceedings of a multidisci- 22 Cohen EL. Genital complications of epidermolysis bullosa. Clin Pedi- plinary international symposium, 23 October 2003. DebRA, 2004. atr 1983; 22:443. 30 Pola E, Logroscino G, De Santis V et al. Onset of Berger disease 23 Shackelford GD, Bauer EA, Graviss ER et al. Upper airway and after Staphylococcus aureus infection: septic arthritis after anterior cruci- external genital involvement in epidermolysis bullosa dystrophica. ate ligament reconstruction. Arthroscopy 2003; 19:E29. Radiology 1982; 143:429–32. 31 Horita Y, Tadokoro M, Taura K et al. Histologically conﬁrmed 24 Cuesta-Estelles G, Escobedo-Rumoroso JM, Garces-Lopez L et al. superimposition of post-streptococcal acute glomerulonephritis Epidermolysis bullosa and chronic renal failure. Nephrol Dial Transplant during IgA nephropathy. Clin Exp Nephrol 2004; 8:351–5. 1998; 13:2133–4. 25 Boulanger E, Catteau B, Pagniez D et al. Recessive dystrophic epidermolysis bullosa and IgA glomerulonephritis. Clin Nephrol 1998; 49:68.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp143–147 CASE REPORT DOI 10.1111/j.1365-2133.2006.07566.x CD30+ large cell transformation of mycosis fungoides after psoralen plus ultraviolet A photochemotherapy J. Ogino, K. Saga, M. Kagaya, A. Kamada, K. Hirosaki, R. Kaneko and K. Jimbow Department of Dermatology, Sapporo Medical University School of Medicine, Minami 1 Nishi 16, Chyuo-ku, Sapporo, Japan

Summary

Correspondence Psoralen plus ultraviolet A (PUVA) photochemotherapy is widely used for the Kenji Saga. therapy of mycosis fungoides (MF). Clinical progression of MF is often associated E-mail: [email protected] with an increase in the size of tumour cells known as transformation. We report two patients with CD30+ large cell transformation that appeared after low-dose Accepted for publication 14 July 2006 PUVA therapy for MF. Clinical data, histopathology, immunohistopathology and T-cell receptor gene rearrangement were studied. Nodules consisted of atypical Key words large cells that expressed CD30. Monoclonal rearrangement of T-cell receptors anaplastic large cell lymphoma, cutaneous T-cell was observed in one case. Low-dose PUVA therapy may be associated with lymphoma, lymphomatoid papulosis, psoralen, CD30+ large cell transformation in patients with MF. ultraviolet A

Conﬂicts of interest None declared.

Mycosis fungoides (MF) is the most common type of cutane- On admission, scaly erythematous brown patches and ous lymphoma that has a clonal proliferation of epidermotrop- plaques were scattered on poikilodermatous skin over the ic CD4+ T cells.1 Clinical progression of MF is often whole body. There was neither superficial lymphadenopathy associated with an increase in the size of tumour cells which nor hepatosplenomegaly. Laboratory examination revealed the ) is known as transformation, a loss of epidermotropism sec- following abnormal findings: total serum protein 6Æ3gdL 1 ) ondary to mutations in skin-homing receptors,2 and an escape (normal 6Æ5–8Æ0), lactate dehydrogenase 450 mg dL 1 (nor- ) from apoptosis. Psoralen plus ultraviolet (UV) A (PUVA) thermal 190–440) and total cholesterol 261 mg dL 1 (normal apy-induced transformation or PUVA-accelerated dissemin- 130–230). Other laboratory tests, including antihuman T-cell ation of transformed CD30+ large cells in MF has not been lymphotropic virus type 1 (HTLV-1) antibody, were within well recognized, although acute myeloid leukaemia following normal range or negative. Chest X-ray and thoracoabodominal PUVA therapy for MF3 and PUVA-induced lymphomatoid computed tomography showed neither lymphadenopathy nor papulosis in patients with MF have been reported.4,5 We organomegaly. A bone marrow biopsy showed no abnormal report two patients with classical MF who developed general- findings. ized nodules consisting of CD30+ large cells that appeared We started PUVA therapy. Two hours after oral administra- ) after low-dose PUVA therapy. We speculate that low-dose tion of 8-methoxypsoralen 0Æ33 mg kg 1 the whole body was PUVA therapy induced CD30+ large cell transformation and irradiated with UVA. The initial dose of UVA was low ) accelerated the progression of the disease in these patients. (1Æ0Jcm 2) because she had severe skin atrophy. One month later, when the dose of UVA had gradually been increased to )2 )2 Case reports 2Æ8Jcm and the total dose had reached 20Æ8Jcm , the extremities started to reveal small shiny solid nodules. These solid nodules showed neither spontaneous regression nor ne- Patient 1 crosis. Skin biopsies were taken from erythematous nodules Plaque-stage MF was diagnosed for the first time in a 73-year- on her left forearm and her left knee (Fig. 1a). The histopa- old woman in September 2001. CD30+ cells were not detec- thology showed nodular infiltration of large atypical cells ted in the plaque lesion by retrospective immunohistochemical accompanied by sparse medium-sized lymphoid cells in the studies. Before this diagnosis, she had had recurrent ‘eczema- upper dermis (Fig. 1b). About 60% of the infiltrating cells tous’ eruptions for 20 years. After the diagnosis of MF, her were large atypical cells (Fig. 1c). On immunohistochemistry, family doctor in her home town had treated her with topical atypical large cells were CD30+ (Fig. 1d), CD68+, CD3), corticosteroids. She was referred to us in June 2004, because CD4), CD8) and CD20). Anaplastic lymphoma kinase was her skin lesions had become impossible to control with topical negative. Southern blot analysis revealed no clonal T-cell corticosteroids. receptor Jb1 gene rearrangement.

2006 The Authors 148 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp148–151 CD30+ large cell transformation of MF after PUVA therapy, J. Ogino et al. 149

Fig 1. Clinical and histopathological photographs in patient 1. (a) A shiny solid nodule on the left knee. (b) Haematoxylin and eosin staining of the nodule (original magnification ·4). Nodular dense infiltration of lymphocytic cells is present in the papillary and reticular dermis. (c) Haematoxylin and eosin staining of the nodule (original magnification ·20). Atypical large cells comprise 60% of the cellular infiltration. (d) Immunostaining shows large atypical cells that are positive for CD30 (original magnification ·40).

) The dose of UVA was increased to 5Æ5Jcm 2. The nodules and two other skin biopsies were taken from plaques. Histopa- ) started flattening when the total dosage reached 67Æ8Jcm 2. thology of the plaques showed typical features of plaque-stage ) The dose of UVA was further increased to 8 J cm 2 and PUVA MF. No CD30+ cells were found in the infiltrating cells. On therapy was continued at this dose. When the total dosage the other hand, 50% of cells in the ulcerative tumour on the ) reached 109Æ8Jcm 2, all nodules and papules had disap- abdomen were CD30+ large atypical cells. peared. Then her family doctor in her home town continued He had been treated with topical steroids and PUVA. PUVA ) only topical steroid therapy. Two months after the remission, therapy was started at a low dose of UVA (1Æ0Jcm 2). Two multiple nodules reappeared over the whole body. These nod- hours after oral administration of 8-methoxypsoralen ) ules responded poorly to PUVA therapy. Although complete 0Æ36 mg kg 1 the whole body was irradiated with UVA. The ) remission was not achieved, intramuscular injection of inter- dose of UVA was gradually increased to 3Æ0Jcm 2. Two feron-c suppressed the development of nodules. weeks later, when the total dosage of UVA had reached ) 8Æ5Jcm 2, multiple solid erythematous nodules suddenly started to appear on the plaques over the whole body. These Patient 2 nodules showed neither spontaneous regression nor necrosis. A 65-year-old man presented in May 2004 with skin erup- Two biopsies were taken from these new skin nodules. Histo- tions that had lasted for about 20 years. He had erythema- pathology showed a nodular infiltration of large atypical cells tous brown scaly patches and plaques on the whole body with sparse inflammatory cells involving the full dermis and along with an ulcerative tumour that had appeared on the upper subcutaneous tissues. About 90% of the infiltrating cells abdomen 1 month before the visit. The tumour had been were large atypical cells. Southern blot analysis revealed a clo- rapidly growing and was 3 cm in diameter at presentation. nal T-cell receptor Jb1 gene rearrangement. Immunohisto- There was neither superficial lymphadenopathy nor hepato- chemical studies revealed that atypical large cells were CD30+, splenomegaly. CD3), CD4+, CD8), CD20) and CD68). Anaplastic lymph- Laboratory examination revealed the following abnormal oma kinase was negative. ) findings: zinc turbidity test 1Æ6 KU (normal 2–10) and alka- The dose of UVA was increased up to 6Æ5Jcm 2. When ) ) line phosphatase 602 mg dL 1 (normal 110–370). Other la- the total dosage reached 65Æ1Jcm 2 almost all the nodules boratory tests including anti-HTLV-1 antibody were negative became flat. The PUVA therapy was continued at an outpatient or within normal range. Chest X-ray and thoracoabodominal clinic of another hospital. Three months after the remission, a computed tomography showed neither lymphadenopathy nor new skin tumour appeared on his lower right leg. This organomegaly. A bone marrow biopsy showed no abnormal tumour was surgically removed. This tumour showed nodular findings. A skin biopsy was taken from an ulcerative tumour infiltration of CD30+ large cells.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp148–151 150 CD30+ large cell transformation of MF after PUVA therapy, J. Ogino et al.

PUVA induced mutation in the HPRT gene of diploid human Discussion 19 fibroblasts. These data support the thesis that low-dose The incidence of transformation is highly variable, ranging PUVA may induce transformation and accelerate the progres- from 8%4 to 55%5 of patients with MF depending on var- sion of the disease. In both of the present cases, complete ious diagnostic criteria for transformation. The transformation regression was achieved by increasing the dose of UVA. This of MF is now defined as at least one skin biopsy showing indicates that high-dose PUVA therapy can suppress prolifer- large cells (‡ 4 times the size of a small lymphocyte) exceed- ation of transformed large cells through direct damage to neo- ing 25% of the infiltrate throughout or forming microscopic plastic cells, even though it may decrease immune surveillance nodules.6 Transformation of MF often occurs in advanced sta- by normal lymphocytes. ges.7,8 Diamandidou et al.7 reported that patients with MF Transformation should be differentiated from primary and Se´zary syndrome in stage IIB–IV disease, and in particu- CD30+ lymphoproliferative disorders, including lymphoma- lar those with tumours, have a high incidence of large cell toid papulosis, primary cutaneous anaplastic large cell lymph- transformation. In patient 1, low-dose PUVA therapy was oma and cutaneous infiltration of systemic or nodal CD30+ started at stage IB (T2N0M0). In patient 2, low-dose PUVA lymphoma. Transformed cases with tumours have poor prog- therapy was started at stage IIB (T3N0M0). Patients with nosis with median survival of < 1 year, while patients with transformation have relatively poor survival, especially if MF associated with lymphomatoid papulosis have a similar or transformation occurs early (within 2 years) in the course of better prognosis compared with those without lymphomatoid 7 the disease. Prognosis is poor if the stage is IIB or higher.7 papulosis. CD30+ large cell transformation can easily be dif- Transformation to a clinically more aggressive form of ferentiated from secondary infiltration of systemic and nodal lymphoma or leukaemia is not unique to MF. Approximately CD30+ lymphoma. The absence of anaplastic lymphoma kin- 25% of follicular B-cell lymphomas undergo transformation ase protein in the CD30+ primary cutaneous lymphoprolifera- to diffuse large cell lymphoma or occasionally to acute lym- tive disorders is useful for distinguishing these lesions from a phoblastic leukaemia.9 secondary skin dissemination of systemic CD30+ anaplastic 20 CD30 is often expressed in transformed large cells in MF. large cell lymphoma. In the present two cases, negative im- Salhany et al.6 reported that 10 of 17 cases expressed CD30 in munohistochemical staining for anaplastic lymphoma kinase large cell transformation that occurred in cutaneous T-cell indicated that these cases were not skin infiltration of systemic lymphoma. The molecular mechanisms responsible for CD30+ or nodal CD30+ anaplastic large cell lymphoma. Lymphoma- large cell transformation in MF are not yet fully understood. toid papulosis presents itself as recurrent crops of reddish pa- Wood et al.10 demonstrated that MF and large cell lymphoma pulonodular lesions that regress spontaneously within a few arose from a single T-cell clone. They speculated the sequen- weeks leaving only small scars and pigmentation. Cellular tial progression from a premalignant T-cell clone to MF, and infiltrates of lymphomatoid papulosis consist of CD30+ atyp- from MF to large cell lymphoma. Transformation would occur ical cells, which may be scattered or in small clusters (two to when a second somatic mutation, or series of events, confers five atypical cells), admixed with a reactive cellular back- 20 a proliferative advantage to a subclone over the original neo- ground including neutrophils, eosinophils and histiocytes. plastic clone.10 Large cell transformation in MF is associated Atypical CD30+ cells in lymphomatoid papulosis usually have with expression of Th2-associated receptors such as CCR4.2 a CD4+ T-helper phenotype, namely CD3+, CD4+ and 18 Therapy-associated large cell transformation in MF has CD8). In the present two cases, histopathological findings rarely been reported. Pielop et al.11 reported a case in which were different from lymphomatoid papulosis. In the present ciclosporin induced CD30+ large cell transformation in axil- cases, CD30+ large cells formed nodules without infiltration lary lymph nodes of undiagnosed MF. The lymphadenopathy of neutrophils and eosinophils. Primary cutaneous CD30+ regressed after cessation of ciclosporin.11 Abd-el-Baki et al.12 large cell lymphomas present themselves as single or multiple demonstrated a significant association between methotrexate medium-sized to large reddish nodules or tumours that may 20 exposure and transformation to large cell lymphoma in initially show temporary regression, but often persist. The patients with MF. However, PUVA-induced or PUVA-acceler- large cell lymphomas show a grossly nodular or diffuse pat- ated large cell transformation in MF has not been widely tern, or both patterns, of dermal involvement with frequent 20 recognized, although acute myeloid leukaemia following extension to the subcutis. CD30+ large cell lymphomas are PUVA therapy to MF3 and PUVA-induced lymphomatoid diagnosed by detecting more than 75% of CD30+ large cells 8 papulosis in patients with MF have been reported.13,14 in the lymphoma cells. 21 The lymphotoxicity in PUVA therapy has mainly two Vergier et al. reported that the diagnosis of CD30+ large effects, namely a decrease in immune surveillance and direct cell transformation in MF is certain if: (i) clinical transform- damage to T cells.15–17 Johnson et al.18 demonstrated that low ation occurs on a lesion of MF (such as a patch); and (ii) cel- doses of PUVA were highly cytotoxic for phytohaemaggluti- lular pleomorphism is observed with cerebriform T nin-activated normal lymphocytes and transformed T lympho- lymphocytes mixed with fewer than 75% of CD30+ large cytes assessed with two viability assays and with flow cells. When more than 75% of large cells are CD30+, cytofluorometry. Chiou and Yang reported that low-dose only patients’ evolution may reveal the difference between

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp148–151 CD30+ large cell transformation of MF after PUVA therapy, J. Ogino et al. 151 transformed MF and primary CD30+ lymphoproliferative dis- 8 Beljaards RC, Kaudewitz P, Berti E et al. Primary cutaneous CD30- orders.21 In both of the present cases, CD30+ large atypical positive large cell lymphoma: definition of a new type of cutane- cells were admixed with cerebriform lymphocytes, although ous lymphoma with a favorable prognosis. A European multicenter study on 47 patients. Cancer 1993; 71:2097–104. nodules had about 90% of CD30+ large atypical cells in 9 Sheehan-Dare RA, Cotterill JA, Barnard DL. Transformation of mye- patient 2. In addition, the nodules appeared on MF plaques lodysplasia to acute myeloid leukaemia during psoralen photo- after low-dose PUVA therapy. In patient 1, no CD30+ large chemotherapy (PUVA) treatment of psoriasis. Acta Derm Venereol cells had been observed in a biopsy from a plaque lesion that (Stockh) 1989; 69:262–4. had been taken 33 months before the nodules had appeared. 10 Wood GS, Bahler DW, Hoppe RT et al. Transformation of mycosis Thus, we speculate that low-dose PUVA therapy induced fungoides: T-cell receptor beta gene analysis demonstrates a com- CD30+ large cell transformation in patient 1. In patient 2, mon clonal origin for plaque-type mycosis fungoides and CD30+ large-cell lymphoma. J Invest Dermatol 1993; 101:296–300. only the abdominal ulcerative tumour had had more than 11 Pielop JA, Jones D, Duvic M. Transient CD30+ nodal transform- 75% of CD30+ large cells, which fulfilled the criteria of trans- ation of cutaneous T-cell lymphoma associated with cyclosporine 6 formation in MF. However, two other plaques had no treatment. Int J Dermatol 2001; 40:505–11. CD30+ large cells when a biopsy had been taken 20 days 12 Abd-el-Baki J, Demierre MF, Li N et al. Transformation in mycosis before the PUVA therapy. Therefore, we speculate that large fungoides: the role of methotrexate. J Cutan Med Surg 2002; 6:109– cell transformation had occurred spontaneously before PUVA 16. therapy in patient 2, and then proliferation and dissemination 13 Wolf P, Cerroni L, Smolle J et al. PUVA-induced lymphomatoid papulosis in a patient with mycosis fungoides. J Am Acad Dermatol of CD30+ large cells had been accelerated by the low-dose 1991; 25:422–6. PUVA therapy. 14 Zackheim HS, Jones C, Leboit PE et al. Lymphomatoid papulosis associated with mycosis fungoides: a study of 21 patients including References analyses for clonality. J Am Acad Dermatol 2003; 49:620–3. 15 Edelson R, Berger C, Gasparro F et al. Treatment of cutaneous T-cell 1 Kim YH, Hoppe RT. Mycosis fungoides and the Se´zary syndrome. lymphoma by extracorporeal photochemotherapy. N Engl J Med Semin Oncol 1999; 26:276–89. 1987; 316:297–303. 2 Jones D, O’Hara C, Kraus MD et al. Expression pattern of T-cell-as- 16 Okamoto H, Horio T, Maeda M. Alternation of lymphocyte func- sociated chemokine receptors and their chemokines correlates with tion by 8-methoxypsoralen and long-wave ultraviolet radiation. II. specific subtypes of T-cell non-Hodgkin lymphoma. Blood 2000; The effect of in vivo PUVA on IL-2 production. J Invest Dermatol 96:685–90. 1987; 89:24–6. 3 Kwong YL, Au WY, Ng MH et al. Acute myeloid leukemia follow- 17 Ananthaswamy HN. Neoplastic transformation of C3H mouse ing psoralen with ultraviolet A therapy: a fluorescence in situ embryo 10T1/2 cells by 8-methoxypsoralen plus UVA radiation. hybridization study. Cancer Genet Cytogenet 1997; 99:11–13. J Invest Dermatol 1985; 85:102–6. 4 Dmitrovsky E, Matthews MJ, Bunn PA et al. Cytologic transform- 18 Johnson R, Staiano-Coico L, Austin L et al. PUVA treatment selec- ation in cutaneous T cell lymphoma: a clinicopathologic entity as- tively induces a cell cycle block and subsequent apoptosis in sociated with poor prognosis. J Clin Oncol 1987; 5:208–15. human T-lymphocytes. Photochem Photobiol 1996; 63:566–71. 5 Cerroni L, Rieger E, Hodl S et al. Clinicopathologic and immuno- 19 Chiou CC, Yang JL. Mutagenicity and specific mutation spectrum logic features associated with transformation of mycosis fungoides induced by 8-methoxypsoralen plus a low dose of UVA in the hprt to large-cell lymphoma. Am J Surg Pathol 1992; 16:543–52. gene in diploid human fibroblasts. Carcinogenesis 1995; 16:1357–62. 6 Salhany KE, Cousar JB, Greer JP et al. Transformation of cutaneous 20 Paulli M, Berti E. Cutaneous T-cell lymphomas (including rare sub- T-cell lymphoma to large cell lymphoma: a clinicopathologic and types). Current concepts. II. Haematologica 2004; 89:1372–88. immunologic study. Am J Pathol 1988; 132:265–77. 21 Vergier B, de Muret A, Beylot-Barry M et al. Transformation of my- 7 Diamandidou E, Colome-Grimmer M, Fayad L et al. Transformation cosis fungoides: clinicopathological and prognostic features of 45 of mycosis fungoides/Se´zary syndrome: clinical characteristics and cases. Blood 2000; 95:2212–18. prognosis. Blood 1998; 92:1150–9.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp148–151 CASE REPORT DOI 10.1111/j.1365-2133.2006.07583.x A failure of mucocutaneous lymphangiogenesis may underlie the clinical features of lipoid proteinosis T. Uchida, H. Hayashi, M. Inaoki, T. Miyamoto* and W. Fujimoto Department of Dermatology, Kawasaki Medical School, 577 Matsushima, Kurashiki, Okayama 701-0192, Japan *Department of Dermatology, Tsuyama Central Hospital, 1756 Kawasaki, Tsuyama, Okayama 708-0841, Japan

Summary

Correspondence Lipoid proteinosis (LiP) (OMIM 247100) is a rare autosomal recessive disease Wataru Fujimoto. caused by loss of function mutations in the extracellular matrix protein 1 gene, E-mail: [email protected] ECM1, on chromosome 1q21. LiP is characterized clinically by hoarseness in early infancy, followed by waxy papules and plaques on the face and body along with Accepted for publication 23 July 2006 pox-like and acneiform scars. We studied a 20-year-old Japanese woman with LiP. She was born of consanguineous parents. Biopsy specimens obtained from a Key words nodule on the elbow were used for histopathology, immunohistology and elec- calcinosis cutis, extracellular matrix protein 1, tron microscopy. Exons 6 and 7 of ECM1 were amplified by polymerase chain lipoid proteinosis, lymphangiogenesis, Weibel– reaction (PCR) from genomic DNA from the proband, her parents, her brother Palade body and an unrelated person. PCR products were sequenced to detect the mutation. Conflicts of interest Histopathological examination revealed an irregular mass of calcium beneath None declared. deposits of a hyaline material in the dermis. Immunofluorescence double staining showed that the CD31-positive microvascular density was increased but that staining for the lymphatic-specific hyaluronan receptor LYVE-1 was drastically diminished in lesional compared with nonlesional skin of the patient and with normal skin. Electron microscopy revealed marked concentric reduplication of basal laminae not only around blood vessels but also around solitary dermal cells positive for Weibel–Palade bodies scattered in the hyaline material. Sequencing of the PCR products revealed a homozygous frameshift mutation, 507delT, in exon 6. This led to a premature stop codon 23 bp downstream. The results of immunopathological and ultrastructural characterization suggest that a failure of mucocutaneous lymphangiogenesis may underlie the clinical features of LiP. Identification of mutation 507delT in a Japanese patient with LiP further supports the thesis that this mutation represents a recurrent mutation in ECM1 in patients with LiP. To our knowledge, this case represents the first report of calcinosis cutis occurring in LiP.

Lipoid proteinosis (LiP) (OMIM 247100) is a rare autosomal gene, ECM1.2,4 Thus far, mutations in ECM1 have been detec- recessive genodermatosis characterized predominantly by ted in more than 20 different families from all around the hoarseness, and flesh-coloured papules and nodules involving world.4 It has been shown that ECM1 appears to have an im- the skin and mucosa.1,2 LiP was first reported by Urbach and portant role in regulating blood vessel homeostasis in human Wiethe3 in 1929, and was originally named ‘lipoidosis cutis skin. In this paper, we labelled biopsy specimens with anti- et mucosae’. The disorder typically presents in early infancy bodies to the panvascular marker CD31 and lymphatic-specific with hoarseness, followed by blisters and varicelliform scar- hyaluronan receptor LYVE-1, and detected a failure of cutane- ring of the skin during childhood. Gradually diffuse infiltra- ous lymphangiogenesis in lesional skin of LiP. Furthermore, tion and thickening of the skin give a yellow and waxy we report the identification of a recurrent mutation of ECM1 appearance, resulting in warty plaques on the elbows and bea- in a Japanese patient with LiP. Our results of immunofluores- ded papules on the eyelid margins.1 Pathological mutations cence double staining provide a novel hypothesis of the patho- were recently identified in the extracellular matrix protein 1 mechanism underlying this unique genodermatosis.

2006 The Authors 152 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp152–157 Failure of lymphangiogenesis in lipoid proteinosis, T. Uchida et al. 153

Case report hair, eyebrows and nails were normal. Neurological examination, the patient’s general level of intellect, and language A 20-year-old Japanese woman was referred for evaluation of skills were normal. Blood evaluation was unrevealing and a scar-like lesions in March 2004. The patient was the second computerized brain scan and magnetic resonance imaging child born to a Japanese couple who were known to be first scans revealed bilateral and symmetrical paracellar calcifica- cousins. Her parents and brother were not affected. She had tions. Videolaryngoscopy revealed thickened epiglottis, swol- had hoarseness since infancy. At the age of 18 months, the len arytenoids and aryepiglottic fold. patient developed bullae and haemorrhagic blisters on her occiput and then on her wrists and fingers (Fig. 1a). A diag- Materials and methods nosis of epidermolysis bullosa was suspected. Throughout the first few years of life her gingiva gradually became hyper- Histopathology, immunofluorescence microscopy and trophic and her tongue and lips became stiff. From the age of electron microscopy 6 years, the patient noticed her skin gradually becoming dry with numerous waxy, yellowish papules over her face, neck After obtaining informed consent, biopsy specimens were and elbows. She also started to experience recurrent ulcera- taken from the nodule at the patient’s elbow and from nonle- tions on her axillae and antecubital fossa and then round or sional skin of the patient’s left leg. Normal skin used as a con- striae-like scarring gradually increasing on her trunk and ex- trol was obtained from surgical specimens. Skin was fixed in tremities. Examination revealed yellowish papular infiltrations 10% formalin and processed for routine light microscopy with on her forehead, perioral region and upper eyelids paraffin embedding. Sections were stained with haematoxylin (Fig. 1b,d). Her lips were thick and her gingiva was irregu- and eosin and with periodic acid–Schiff (PAS). Immunofluo- larly hypertrophic (Fig. 1c). The tongue was enlarged and its rescence staining was performed on 4-lm cryostat sections movement was reduced because it was tethered with a thick- using mouse monoclonal antibodies to human CD31 (Dako, ened frenulum. Yellowish waxy papules and nodules were also Carpinteria, CA, U.S.A.), rabbit antihuman collagen type IV noted on the elbows (Fig. 1e), knees, legs and backs of the (Monosan, Uden, Netherlands), rabbit antihuman hyalunonan hands. Numerous, round atrophic scars were present on her receptor LYVE-1 (Abcam Ltd, Cambridge, U.K.) and corres- trunk and striae-like scars were intermingled within violaceous ponding secondary antibodies labelled with Cy3 or fluorescein or yellowish plaques on the antecubital fossae (Fig. 1f). The isothiocyanate (Jackson ImmunoResearch Laboratories, Inc.,

Fig 1. Clinical features of lipoid proteinosis. (a) A retracted haemorrhagic blister on the tip of a finger at the age of 18 months; (b) yellowish papular infiltration on the forehead, perioral region and upper eyelids; (c) the lips, gingiva and tongue were irregularly hypertrophic and movement of the tongue was reduced; (d) beaded or cobblestone-like papules along the eyelids and inner canthus; (e) yellowish waxy infiltrated plaques and nodules on the elbow; (f) striae-like scars in (b) (f) yellowish plaques on the antecubital fossae.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp152–157 154 Failure of lymphangiogenesis in lipoid proteinosis, T. Uchida et al.

Westgrove, PA, U.S.A.). Antibody staining was photographed with an Olympus FV300 confocal laser scanning microscope. A small skin sample was ﬁxed in 2Æ5% glutaraldehyde and routinely embedded in epoxy resin, cut and stained with ura- nyl acetate and lead citrate, and examined in a JEM-2000EX (II) electron microscope (Jeol Ltd, Tokyo, Japan).

Mutation analysis Fig 2. Histological findings of lipoid proteinosis. (a) Deposits of After obtaining informed consent, peripheral blood samples homogeneous hyaline-like material throughout the dermis with were taken from the affected individual and her clinically irregular masses of calcium (haematoxylin and eosin); (b) hyaline-like normal parents, and DNA was extracted using Dnaquick material stained with periodic acid–Schiff (PAS). PAS-positive material (Dainippon Seiyaku Co., Osaka, Japan). Primers were designed is prominent surrounding blood vessels but not at the dermal– for amplification of all exons of the ECM1 gene as described epidermal junction. Note nuclei scattered in the homogeneous in detail elsewhere.4 Specifically, primers for amplification hyaline-like material. of exon 6 were: forward primer 5¢-AGCCTTGAGAAG CAGGAGGA-3¢ and reverse primer 5¢-AGTGAACGGGACC contained capillaries and a few scattered elongated cells. An TGAGGTT-3¢. For polymerase chain reaction (PCR) ampli- irregular mass of calcium was present in the degenerated der- fication, 250 ng of genomic DNA was used as the template in mis (Fig. 2a). PAS-positive basement membrane (BM) thicken- an amplification buffer containing 5 pmol of each primer, ing was observed around blood vessels but not at the dermal–

2Æ5 mmol MgCl2,0Æ5 mmol of each nucleoside triphosphate epidermal junction (DEJ) (Fig. 2b). Immunofluorescence dou- and 1Æ25 U of AmpliTaq Gold polymerase (Applied Biosys- ble staining with antibodies to the panvascular marker CD31 tems, Foster City, CA, U.S.A.) in a total volume of 50 lLina and collagen type IV showed an increase of microvascular GeneAmp PCR System 9700 thermal cycler (Applied Bio- structures in the dermis and confirmed that type IV collagen systems). Samples were denatured for 9 min before 35 ampli- was strongly expressed as bright, thick bands of staining fication cycles (95 C for 1 min, followed by 60 C for 1 min around CD31-positive vessels (Fig. 3). The staining for type and 72 C for 1 min), followed by a final extension at 72 C IV collagen was not increased at the DEJ. To address whether for 5 min. PCR products were then subcloned into the increased microvascular density is due solely to an increase of pCR2.1-TOPO TA-cloning vector (Invitrogen, Carlsbad, CA, capillaries and blood vessels, immunofluorescence double U.S.A.) and sequenced in an ABI 310 Genetic Analyzer staining with antibodies to the lymphatic-specific hyaluronan (Applied Biosystems). For the genotyping, restriction fragment receptor LYVE-1 and to CD31 was performed in both lesional length polymorphism was performed. The PCR product and nonlesional skin of the patient. Interestingly, the staining was purified using the QIAquick PCR purification kit for LYVE-1 was drastically diminished in lesional skin com- (Qiagen, Valencia, CA, U.S.A.), digested with the appropriate pared with nonlesional skin of the patient and with normal restriction enzymes overnight, and analysed by 3% agarose gel skin (Fig. 4b). The staining for both CD31 and LYVE-1 in electrophoresis. nonlesional skin of the patient was comparable with that in normal skin (Fig. 4d,e,g,h). Ultrastructural examination at the Results papillary dermis revealed fine collagen fibrils embedded in an amorphous, fine-granular matrix under the basal lamina Histologically, the lesional skin revealed marked deposition of (Fig. 5a). The basal lamina at the DEJ was intact, but there a homogeneous, hyaline material in the upper dermis that was marked reduplication of the basal lamina around capillary

Fig 3. Immunoﬂuorescence double staining for CD31 and collagen type IV. Double staining with antibodies to the panvascular marker CD31 (green) and collagen type IV (a) (b) (c) (red) in lesional skin of the patient (a–c) and in normal skin (d–f) shows an increase of microvascular structures in the dermis of lesional skin (a). Type IV collagen was strongly expressed as bright, thick bands of staining around CD31-positive vessels (c). The staining for type IV collagen was not (d) (e) (f) increased at the dermal–epidermal junction in lesional skin (b,c).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp152–157 Failure of lymphangiogenesis in lipoid proteinosis, T. Uchida et al. 155

(a) (b) (c)

Fig 4. Immunoﬂuorescence double staining for CD31 and the lymphatic-speciﬁc hyaluronan receptor LYVE-1. Double staining with antibodies to CD31 (green) and LYVE-1 (red) in lesional (a–c) and nonlesional (d–f) skin of the patient, and in normal skin (g–i). (d) (e) (f) Interestingly, the staining for LYVE-1 (red) was drastically diminished in lesional skin (b) compared with nonlesional (e) and with normal skin (h). The staining for CD31 (green) and LYVE-1 (red) in nonlesional skin of the patient (d,e) was comparable with that in normal skin (g,h). The white lines indicate (g) (h) (i) the dermal–epidermal junction.

venules consisting of multiple layers arranged in a concentric The characteristic cutaneous manifestations of LiP such as fashion (Fig. 5b). Reduplication of basal lamina was also waxy, yellowish papules over the face, neck and elbows observed around a number of cells scattered in hyaline-like appear at around 5 or 6 years of age. This could be the period material that had Weibel–Palade bodies in their cytoplasm necessary for a sufficient amount of ‘hyaline-like’ material to (Fig. 5c, inset). form in the dermis to be identified as papules or nodules. Amplified DNA from the affected individual disclosed a Hoarseness is usually present during infancy and is due to a homozygous, 1-bp deletion of T at nucleotide 507 (507delT) hyaline-like material deposited in the mucous membranes of within exon 6 of the ECM1 gene (Fig. 6b). This mutation vocal cords as observed in the present case.6 Next, the gingiva, leads to a frameshift and premature stop codon 23 bp down- tongue and lips gradually become hypertrophic in the first stream. The 507delT mutation results in a new cut site for few years of life, before the cutaneous manifestations. From restriction endonuclease HpaII (New England Biolabs, Beverly, these features it is speculated that the lymphatic system plays a MA, U.S.A.), and digestion of exon 6 PCR products for all crucial role in these mucosal tissues that require high dynam- family members with HpaII verified the presence of a mutation ics and plasticity for eating and speaking from an early age. on one allele in the parents and the brother (i.e. heterozygous This concept could also be applied to the cutaneous lesions, carriers) (Fig. 6c). because they appear characteristically on skin exposed to extension and flexion such as the face, knees, elbows and ante- Discussion cubital fossae. On the other hand, cutaneous manifestations do not seem Although immunohistological and ultrastructural features in to progress further in adult patients with LiP. Also, lymph- LiP are well established, the composition of the ‘hyaline-like’ oedema has never been a chief complaint in patients with LiP. material and its mechanism of accumulation in the dermis are A comparable staining for both CD31 and LYVE-1 in nonle- poorly understood. We determined if the cutaneous lymphatic sional skin of the patient with that in normal skin suggests system is preserved in the lesional skin of a patient with LiP. that both cutaneous angiogenesis and lymphangiogenesis are In contrast to an increase of blood vessels with marked expres- preserved overall in the patient but are affected only in lesion- sion of type IV collagen around them, selective reduction of al skin. In the lesional skin of LiP, tortuous, dilated blood ves- cutaneous lymphatic vessels was demonstrated in lesional skin sels are prominent (Fig. 3a–c). The presence of solitary cells of LiP. The lymphatic system controls tissue fluid homeostasis containing Weibel–Palade bodies, the endothelial cell-specific by draining protein-rich lymph from tissues. In the skin, fluid storage granule, in their cytoplasm might indicate that endo- and macromolecules enter the vessels through the gaps thelial cell migration is accelerated but endothelial cell lumen between lymphatic endothelial cells when interstitial pressure formation is immature.7,8 Our findings might correspond to increases.5 Therefore, it is easily assumed that the inefficient the presence of a number of collapsed vessels in the lesional draining of fluids and macromolecules due to the lack of cuta- skin of LiP, implicating partial loss of endothelial polarity, as neous lymphatic vessels leads to the accumulation of such pro- reported recently.9 These clinical and immunopathological fea- tein-rich material in the extracellular matrix. tures of LiP suggest that the blood vascular system may play a

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp152–157 156 Failure of lymphangiogenesis in lipoid proteinosis, T. Uchida et al.

(a) (b)

Fig 5. Transmission electron microscopy. (a) Deposit of an amorphous, fine-granular matrix under the basal lamina. (b) The basement membrane of a small blood vessel shows marked concentric reduplication. The lamina densa is not reduplicated at the dermal–epidermal junction. (c) Marked reduplication around a single cell embedded in amorphous hyaline-like material. These cells have rod-shaped, tubulated Weibel– (c) Palade bodies (WP) characteristic of vascular endothelial cells (c, inset). compensatory role for the lymphatic system in perfusion and studies that a failure of cutaneous lymphangiogenesis reabsorption in the skin and mucosa. accounts, at least in part, for the clinicopathological features Many pro- and antiangiogenic factors are now known to of LiP. However, the current data cannot really separate out promote and inhibit angiogenesis.7,8 ECM1 has been reported cause and effect. In other words, reduction of cutaneous lym- to promote endothelial cell proliferation and angiogenesis.10,11 phatic vessels could just be a sequela followed by the depos- ECM1 has recently been shown to bind important members of ition of hyaline-like material in the lesional skin of LiP. the BM molecules, perlecan and fibulin-1,12,13 and to inhibit Therefore, it is intriguing to know whether interactions the activity of matrix metalloproteinase 9 (MMP-9) through between altered ECM1 and other extracellular matrix proteins protein/protein interaction.14 Both perlecan and fibulin-1 are are involved not only in vascular anomalies but also in select- thought to affect endothelial growth, matrix synthesis and vas- ive reduction of cutaneous lymphatic vessels in the lesional cular maturation through interactions with other BM mole- skin of LiP. cules.9 A role for MMP-9 in tubular morphogenesis In this study, we have elucidated the molecular basis of is reported, although it is still controversial.7 Based on these in LiP in one Japanese individual. Since Urbach and Wiethe’s vitro studies it is likely that interactions between altered ECM1 publication in 1929, over 250 cases of LiP have been de- and other extracellular matrix proteins such as perlecan, fibu- scribed. To the best of our knowledge, only 24 patients with lin-1 and MMP-9 are involved in vascular anomalies in LiP, LiP have been reported in the dermatological literature in although their precise role remains to be established. Japan. Among them, a mutation of the ECM1 gene was iden- Much emphasis has previously been placed on the role of tified in only two patients. One had a homozygous 892delC fibroblasts in the production and accumulation of ‘hyaline- mutation in exon 7 and the other had a homozygous R53X like’ material in LiP. This was once regarded as a lysosomal mutation in exon 3.4 In contrast, we have identified the storage disease, suggested by the cytoplasmic vacuolization in 507delT mutation in exon 6 of the ECM1 gene for the first dermal fibroblasts.15,16 On the other hand, it has been repor- time in a Japanese patient with LiP. The 507delT mutation is ted that ‘hyaline-like’ materials originate from the overproduc- possibly caused by slipped mispairing during DNA replica- tion of noncollagenous proteins, most of which are normal tion because this part of the gene is composed of palin- constituents of human skin.17 Although our understanding of dromic runs of multiple C or G nucleotide repeats. This the function of the lymphatic system and its role in disease is particular mutation has been documented previously, in two still rudimentary,5,18 it is conceivable from the results of our Thai brothers, an Iranian family4 and an Indian child with

References (a) 1 Hofer PA. Urbach–Wiethe disease (lipoglycoproteinosis; lipoid proteinosis; hyalinosis cutis et mucosae): a review. Acta Derm Venereol (Stockh) 1973; 53 (Suppl.):1–52. 2 Hamada T, McLean WHI, Ramsay M et al. Lipoid proteinosis maps to 1q21 and is caused by mutations in the extracellular matrix protein 1 gene (ECM1). Hum Mol Genet 2002; 11:833–40. 3 Urbach E, Wiethe C. Lipoidosis cutis et mucosae. Virchows Arch A Pathol Anat 1929; 273:285–319. 4 Hamada T, Wessagowit V, South AP et al. Extracellular matrix protein 1 gene (ECM1) mutations in lipoid proteinosis and genotype– phenotype correlation. J Invest Dermatol 2003; 120:345–50. 5 Hong Y-K, Shin JW, Detmar M. Development of the lymphatic vas- (b) (c) cular system: a mystery unravels. Dev Dyn 2004; 231:462–73. 6 van Hougenhouck-Tulleken W, Chan I, Hamada T et al. Clinical Fig 6. Pedigree and identification of a recurrent ECM1 mutation in and molecular characterization of lipoid proteinosis in Namaqua- lipoid proteinosis (LiP), 507delT, in exon 6 in a family with LiP. (a) land, South Africa. Br J Dermatol 2004; 151:413–23. Pedigree of the family affected with LiP. The mutant allele is present 7 Davis GE, Senger DR. Endothelial extracellular matrix. Biosynthesis, in three generations which is consistent with an autosomal recessive remodeling, and functions during vascular morphogenesis and trait. (b) Sequencing of exon 6 of ECM1 in the proband (P) reveals a neovessel stabilization. Circ Res 2005; 97:1093–107. 1-bp deletion of T at nucleotide 507 (507delT). N indicates the 8 Bhushan M, Young HS, Brenchley PEC, Griffiths CEM. Recent normal sequence. (c) Restriction endonuclease digestion of advances in cutaneous angiogenesis. Br J Dermatol 2002; 147:418– polymerase chain reaction (PCR) products spanning exon 6 of ECM1 25. with HpaII. In the normal individual (C), the original PCR product of 9 Mirancea N, Hausser I, Beck R et al. Vascular anomalies in lipoid 671 bp is digested into fragments of 313, 243, 91 and 24 bp (the proteinosis (hyalinosis cutis et mucosae): basement membrane components and ultrastructure. J Dermatol Sci 2006; 42:231–9. smallest band is not visible). In the proband, the PCR product is 10 Han Z, Ni J, Smits J et al. Extracellular matrix protein 1 (ECM1) further digested into fragments of 313, 224, 91, 24 and 19 bp (the has angiogenic properties and is expressed by breast tumor cells. smallest two bands are not visible), indicating the presence of FASEB J 2001; 15:988–94. 507delT on both ECM1 alleles. In the parents and the brother, the PCR 11 Chan I. The role of extracellular matrix protein 1 in human skin. product is digested into fragments of 313, 243, 224, 91, 24 and Clin Exp Dermatol 2004; 29:52–6. 19 bp (the smallest two bands are not visible), indicating 12 Mongiat M, Fu J, Oldershaw R et al. Perlecan protein core interacts heterozygosity for 507delT. with extracellular matrix protein 1 (ECM1), a glycoprotein involved in bone formation and angiogenesis. J Biol Chem 2003; 278:17491–9. LiP.19 Therefore, identification of the 507delT mutation in a 13 Fujimoto N, Terlizzi J, Brittingham R et al. Extracellular matrix pro- Japanese patient with LiP adds further molecular data to sup- tein 1 interacts with the domain III of fibulin-1C and 1D variants port the earlier reports that this particular mutation repre- through its central tandem repeat 2. Biochem Biophys Res Commun sents the only recurrent mutation described so far in ECM1 2005; 333:1327–33. in patients with LiP. 14 Fujimoto N, Terlizzi J, Aho S et al. Extracellular matrix protein 1 Calcium deposits were seen in the dermis. Patients with LiP inhibits the activity of matrix metalloproteinase 9 through high- are known to have calcification of the temporal lobes or hip- affinity protein/protein interactions. Exp Dermatol 2006; 15:300–7. 15 Bauer EA, Santa-Cruz DJ, Eisen AZ. Lipoid proteinosis: in vivo and pocampi in the brain,1,20 but cutaneous calcification has not in vitro evidence for a lysosomal storage disease. J Invest Dermatol previously been reported. Because the patient was normocalc- 1981; 76:119–25. aemic and normophosphataemic, and the calcium deposits 16 Moy LS, Moy RL, Matsuoka LY et al. Lipoid proteinosis: ultrastruc- were juxtaposed with hyaline material in the dermis, this tural and biochemical studies. J Am Acad Dermatol 1987; 16:1193– should be categorized as dystrophic calcification. Although 201. various genetic disorders are accompanied by dystrophic calci- 17 Fleischmajer R, Krieg T, Dziadek M et al. Ultrastructure and compo- nosis cutis, the exact mechanism of cutaneous calcification is sition of connective tissue in hyalinosis cutis et mucosae skin. J Invest Dermatol 1984; 82:252–8. not known.21 The hyaline material in the dermis may serve as 18 Hirakawa S, Detmar M. New insights into the biology and path- a nidus for calcification. However, further investigation of this ology of the cutaneous lymphatic system. J Dermatol Sci 2004; patient and other future LiP patients with calcinosis cutis is 35:1–8. required to determine its aetiology. 19 Chan I, Bingewar G, Patil K et al. An Indian child with lipoid proteinosis resulting from a recurrent frameshift mutation (507delT) in the extracellular matrix protein 1 gene. Br J Dermatol Acknowledgments 2004; 151:726–7. 20 Hamada T. Lipoid proteinosis. Clin Exp Dermatol 2002; 27:624–9. We thank Yoko Yoshida, Department of Dermatology, and 21 Fairley JA. Calcifying and ossifying disorders of the skin. In: Derma- Taiji Suda, Electron Microscopy Research Center, Kawasaki tology (Bolognia JL, Jorizzo JL, Rapini RP, eds), London: Mosby, Medical School, for their assistance in immunofluorescence 2003; 691–8. and electron microscopy.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp152–157 Gene Corner

MSH6 mutation in Muir–Torre syndrome: could this be a rare ﬁnding?

DOI: 10.1111/j.1365-2133.2006.07607.x evaluation identified the lesion as a cystic sebaceous tumour (Fig. 1a). Cystic sebaceous tumours were earlier described as Clinical criteria for diagnosis of Muir–Torre syndrome (MTS, specific cutaneous markers for MTS.8 Due to this finding the MIM 158320) include synchronous or metachronous occur- dermatohistopathologist suspected MTS/Lynch syndrome. rence of at least one sebaceous gland neoplasia (adenoma, epi- Analysis of the tumour histories of the patient and his fam- thelioma or carcinoma) and at least one internal neoplasm in ily revealed that he had been diagnosed with a seminoma at a patient.1,2 A subset of patients with MTS was found to be 39 years of age, and with ascending colon cancer at 59 years carriers of germline mutations in DNA mismatch repair of age. The patient’s younger son was diagnosed with ‘benign (MMR) genes mainly in MSH2.3,4 This finding uncovered cutaneous tumours’ at age 40 years. His mother died at the many clinically diagnosed MTS cases as phenotypic variants of age of 74 years from a ‘cancer of the lower part of the abdo- the autosomal dominant cancer susceptibility syndrome men’ (no detailed information on the affected organ was known as Lynch syndrome or ‘hereditary nonpolyposis colo- available) and his maternal grandmother died at the age of rectal cancer’ (HNPCC). HNPCC is diagnosed clinically when 59 years from either gastric or colorectal cancer. The patient’s the patient’s family meets the Amsterdam criteria (three family father died at the age of 54 years of unknown reasons members affected with colorectal cancer or an HNPCC-associ- (Table 1). ated tumour over at least two consecutive generations, one of them being a first-degree relative of the other two, and one of Patients 2–10 them diagnosed before the age of 50 years).5 HNPCC should also be suspected when the patient’s individual history or fam- Tumour histories and family tumour histories of nine add- ily history meet one of the less stringent Bethesda criteria, e.g. itional patients identified with MMR gene mutations in MSH2 early-onset colorectal cancer (before age 50 years), synchron- or MLH1 are summarized in Table 1. Seven of these patients ous or metachronous HNPCC-associated tumours, or a certain were clinically diagnosed with MTS. In two patients (2 and less stringent familial clustering of tumours irrespective of age 10), MTS was suspected due only to the diagnosis of a cystic at onset.6 Diagnosing Lynch syndrome through identification sebaceous tumour. of the underlying MMR gene mutation (in MSH2, MLH1, MSH6 All patients gave their written informed consent authorizing or PMS2) is of utmost importance not only for the patient but data documentation, analysis of tumour tissue for HNPCC also for the patient’s family, as patients with Lynch syndrome characteristics and mutation analysis in MMR genes. The study and their families are known to profit from specific surveil- was approved by the Ethics Committees of the Medical Facul- lance recommendations (for review see Lynch et al.7). ties of the Bonn and Duesseldorf Universities. To date, patients with MTS with mutations in MSH2 and For immunohistochemical analyses mouse antihuman MLH1 have been reported in the literature, with the majority monoclonal antibodies were used: G219-1129 for MSH2, of mutations located in MSH2, a finding also recently presented G168-15 for MLH1, 610918, BD for MSH6 and 556415, BD in our large sample of patients with MTS.4 Here, we present for PMS2 (PharMingen, San Diego, CA, U.S.A.). Staining pro- the first MSH6 mutation in MTS. In addition, we present nine cedures were performed as previously described.9 Analysis for patients with mutations in either MSH2 or MLH1 and provide microsatellite instability (MSI) was performed on matched an update on proportions of MSH2, MLH1 and MSH6 mutations pairs of tumour DNA and normal DNA using the Bethesda in our mutation-positive patients with MTS. panel of microsatellite markers as previously described.10 A second panel of markers, which was used in the case of Case and methods unclear MSI results after testing with the first set, comprised the markers BAT40, D3S1298, D3S1611, D6S470 and D18S69. Patient 1 Peripheral blood was drawn from all index patients to The patient who was later identified carrying an MSH6 muta- extract genomic DNA by standard salting out procedure. Muta- tion was diagnosed at the age of 61 years with a cutaneous le- tion analysis in MSH6 was performed by direct sequencing of sion located at his front upper trunk. Histopathological the entire coding region (primers available on request). Search

2006 The Authors 158 Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp158–162 MSH6 mutation in MTS, E. Mangold et al. 159

(a) (b) (c)

Fig 1. Histopathological findings and molecular findings in patient 1. (a) Histology of the cystic sebaceous tumour of the skin showing a circumscribed proliferation of enlarged sebaceous lobules. Note the focal prominence of the basal cell population in the periphery of the lobules. Cystic sebaceous tumours represent benign sebaceous adenomas with a unique cystic growth pattern that is different from conventional superficial sebaceous adenomas (haematoxylin and eosin; original magnification · 25). (b) Immunohistochemistry of the cystic sebaceous tumour demonstrating a positive staining for MSH2 and loss of MSH6 staining, a finding that pinpoints an MSH6 mutation (original magnification · 200). (c) Sequencing pattern of the heterozygous germline mutation c.3436C fi T; p.Gln1146X in exon 5 of the MSH6 gene. for small germline mutations in MSH2 and MLH1 was per- of these in MSH2 and two in MLH1. In all patients, prior formed as previously described.10 For detection of large microsatellite analysis and/or immunohistochemical analyses genomic deletions in MSH2, MLH1 and MSH6 multiplex liga- indicated MMR deficiency (Table 1). tion-dependent probe amplification (MLPA) was applied by Now, after the first MSH6 mutation in MTS has been identi- use of SALSA MLPA Kit P003 for MLH1/MSH2 and P008 for fied, the question arises: what is the frequency of MSH6 muta- MSH6/PMS2 according to the manufacturer’s protocol (MRC- tions in MTS? Our sample of mutation-positive patients with Holland, Amsterdam, The Netherlands). MTS might provide an answer to this question: previously we reported 27 patients with MTS who were found to be carriers Results and discussion of mutations, more than 90% of them (25 of 27) located in MSH2 and a small number (two of 27) in MLH1.4 Adding the 10 patients from this study, in summary, we have identified Patient 1 37 mutation carriers in our MTS patient sample to date: 32 Microsatellite analysis performed in the cystic sebaceous (86%) are carriers of MSH2 mutations, four (11%) patients tumour of patient 1 revealed high MSI: four of five markers have MLH1 mutations and one patient has a mutation in MSH6. (BAT25, BAT40, D2S123, D18S69) showed additional alleles. From these data one might conclude that MSH6 mutations are The immunohistochemical staining showed a loss of expres- a rare finding in MTS. sion of the MMR protein MSH6 and normal expression of Most of the mutations reported in patients with Lynch syn- MSH2, MLH1 and PMS2, pinpointing an MSH6 mutation drome are located in MSH2 and MLH1, while the contribution (Fig. 1b). This suspicion was subsequently confirmed by dis- of the two other MMR genes – MSH6 and PMS2 – to the classic covery of an MSH6 germline mutation (MSH6, c.3436C fi T; Lynch syndrome appears to be low. Several recent studies, p.Gln1146X) (Fig. 1c). The mutation is predicted to lead to a however, have provided growing evidence that MSH6 and premature stop at codon 1146 and was therefore classified as PMS2 lead to a different and possibly milder phenotype that pathogenic. Segregation analysis revealed that the patient has might have been missed in previous investigations due to inherited the mutation from his mother. The patient is now strict patient selection for a ‘typical’ Lynch syndrome his- undergoing annual colonoscopies and further recommended tory.11–13 Interestingly, non-HNPCC-associated tumours were Lynch syndrome surveillance examinations and his family reported to be statistically more frequent in MSH6 mutation- members have been referred on for genetic counselling, pre- positive families compared with families with MSH2 and MLH1 dictive testing and screening measures. mutations.12 Patient 1 and his family history reflect these findings – neither does the family history meet the Amsterdam criteria, nor Patients 2–10 does the patient or one of his family members meet the Beth- In nine other patients with clinically diagnosed or suspected esda criteria as they were initially defined in 1997. It was not MTS, germline mutations in MMR genes were detected: seven until the patient developed his third neoplasia, the cystic seba-

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp158–162 160 Table 1 Tumour histories, family tumour histories and mismatch repair gene mutations of 10 new patients with Muir–Torre syndrome MSH

Patient’s tumour history: Family’s tumour history: 6 malignancy (age at relative: malignancy Exon MSI IHC for IHC for IHC for IHC for Mangold E. MTS, in mutation Patient diagnosis) (age at diagnosis) Gene (intron) Mutation Consequence status MSH2 MSH6 MLH1 PMS2 1 Seminoma (39 years), Son: benign skin tumours MSH6 5 c.3436C ﬁ⁄T; p.Gln1146X Nonsense MSI-H + ) ++ colorectal cancer (40 years); mother: ‘cancer of (59 years), cystic seba- the lower part of the abdo- ora Compilation Journal ceous tumour men’c (74 years); maternal (61 years) grandmother: gastric or colorectal cancer (59 years) 2 Cystic sebaceous tumour Brother: gastric or colorectal can- MSH2 1–6 del exon 1–6 Large deletion MSI-H ))++

(48 years) cer (age unknown); mother: al. et gastric or colorectal cancer (age

unknown) a a )) 06BiihAscaino Dermatologists of Association British 2006 3 Colorectal cancer Maternal sister: colorectal cancer MSH2 3 c.303T fi G; p.Gly101 Silent MSI-H ++ (39 years), sebaceous (54 years) tumour (39 years) 4 Colorectal cancer (43 years), Paternal aunt: uterine cancer MSH2 (intron 5) c.942 + 3A fi Tb Splice defectb MSI-Hd ND ND ND ND squamous cell cancer of (70 years) the palatine tonsil (58 years), rectal cancer (60 years), prostate cancer (66 years), sebaceous carcinoma (70 years) 5 Sigmoid cancer Mother: colorectal cancer MSH2 (intron 5) c.942 + 3A fi Tb Splice defectb MSI-H ))++ (29 years), sebaceous (35 years); maternal grand- adenoma (47 years), mother: ‘cancer of the lower carcinoma of the right part of the abdomen’c • rts ora fDermatology of Journal British ureter (48 years), blad- (63 years) der cancer (49 years) 6 Colorectal cancer Sister: colorectal cancer (41 and MSH2 (intron 5) c.942 + 3A fi Tb Splice defectb MSI-He )e )e +e +e (40 years), colorectal 54 years), ‘cancer of the lower cancer (53 years), duo- part of the abdomen’c denal cancer (71 years), (43 years); sister: uterine cancer neuroendocrine tumour (33 years), her son: colorectal of the pancreas cancer (35 years); brother: liver 2007 (71 years), sebaceous cancer (40 years), his daughter:

06TeAuthors The 2006 adenoma (72 years) cervical cancer (age unknown); 156, sister: colorectal cancer

pp158–162 (35 years); father: colorectal cancer (42 years) ora Compilation Journal 06TeAuthors The 2006 Table 1 (Continued) 06BiihAscaino Dermatologists of Association British 2006

Patient’s tumour history: Family’s tumour history: malignancy (age at relative: malignancy Exon MSI IHC for IHC for IHC for IHC for Patient diagnosis) (age at diagnosis) Gene (intron) Mutation Consequence status MSH2 MSH6 MLH1 PMS2 7 Colorectal cancer Brother: colorectal cancer (age MSH2 7 del exon 7 Large deletion MSI-H ) NE + ND (44 years), basal cell unknown); mother and seven carcinomas of her siblings: colorectal can- (> 57 years), bladder cer or other malignancies (ages cancer (70 years), seba- unknown) ceous carcinoma (70 years), cystic sebaceous tumour

• (71 years) rts ora fDermatology of Journal British 8 Colorectal cancer Brother: colorectal cancer MSH2 (intron 13) c.2211-10T ﬁ Aa Unknowna ND ))++ (47 years), prostate (54 years); father: colorectal cancer (66 years), cancer (age unknown); melanoma (68 years), mother: oesophageal cancer sebaceous adenoma (age unknown) (68 years) 9 Two colorectal cancers Sister: unknown malignancy MLH1 13 c.1489dupC; p.Arg497ProfsX6 Frameshift MSI-Hf +f +f )f )f (44 years), sebaceous (24 years); brother: colorectal 2007 adenoma (45 years) cancer (44 years); mother: MSH

156, colorectal cancer (72 years)

10 Cystic sebaceous tumour Two sisters: colorectal cancer MLH1 16 c.1852_1854delAAG; p.Lys615del In-frame deletion MSI-H ND ND ND ND 6 pp158–162 (51 years) (age unknown) Mangold E. MTS, in mutation

MSI, microsatellite instability; MSI-H, high MSI; IHC, immunohistochemistry; ND, not done; NE, not evaluable. aIt is not known whether this mutation is in fact deleterious/disease causing; bleads to an in-frame deletion of exon 5, MSH2; cprecise data on the affected organ are not known: only rough information on localization of the tumour (lower abdomen) is available; dmicrosatellite analysis was performed in tumour tissue of a rectal carcinoma; emicrosatellite analysis and immunohistochemical analysis were performed in tumour tissue of a colorectal carcinoma; fmicrosatellite analysis and immunohistochemical analysis were performed in tumour tissue of a duodenal carcinoma. tal. et 161 162 MSH6 mutation in MTS, E. Mangold et al. ceous tumour, that he met the revised Bethesda criteria,6 References which include the typical cutaneous lesions of MTS. A screen- 1 Cohen PR, Kohn SR, Davis DA, Kurzrock R. Muir–Torre syndrome. ing of the patient’s colorectal cancer for HNPCC characteristics Dermatol Clin 1995; 13:79–89. 13 – as recently proposed by Hampel et al. for all colorectal 2 Schwartz RA, Torre DP. The Muir–Torre syndrome: a 25-year retro- cancers regardless of any criteria – might have led to the spect. J Am Acad Dermatol 1995; 33:90–104. diagnosis of an MSH6 mutation in this family 2 years earlier. 3 Kruse R, Ru¨tten A, Lamberti C et al. Muir–Torre phenotype has a The seminoma in patient 1 is a non-HNPCC-associated frequency of DNA mismatch-repair-gene mutations similar to that tumour. Unfortunately, as the seminoma tissue is no longer in hereditary nonpolyposis colorectal cancer families defined by the available for immunohistochemical analysis, there is no possi- Amsterdam criteria. Am J Hum Genet 1998; 63:63–70. 4 Mangold E, Pagenstecher C, Leister M et al. A genotype–phenotype bility to prove or disprove that it evolved due to the MSH6 correlation in HNPCC: strong predominance of msh2 mutations in mutation. 41 patients with Muir–Torre syndrome. J Med Genet 2004; 41:567–72. In order to screen patients with MTS for HNPCC charac- 5 Vasen HF, Watson P, Mecklin JP, Lynch HT. New clinical criteria teristics and to preselect them for mutation analysis in for hereditary nonpolyposis colorectal cancer (HNPCC, Lynch syn- MMR genes, microsatellite analysis and/or immunohisto- drome) proposed by the International Collaborative group on chemical staining for MMR proteins in cutaneous or internal HNPCC. Gastroenterology 1999; 116:1453–6. tumours of patients with MTS can be performed – the same 6 Umar A, Boland CR, Terdiman JP et al. Revised Bethesda guidelines for hereditary nonpolyposis colorectal cancer (Lynch syndrome) preselection strategy as recommended in patients with and microsatellite instability. J Natl Cancer Inst 2004; 96:261–8. 9,14 suspected Lynch syndrome. Immunohistochemical analy- 7 Lynch HT, Boland CR, Gong G et al. Phenotypic and genotypic het- sis, as long as it is performed for all MMR proteins erogeneity in the Lynch syndrome: diagnostic, surveillance and involved in Lynch syndrome – MSH2, MLH1, MSH6 and management implications. Eur J Med Genet 2006; 14:390–402. PMS2 – can focus molecular genetic workup on the affected 8 Rutten A, Burgdorf W, Hugel H et al. Cystic sebaceous tumors as gene and helps to reduce time and costs spent for mutation marker lesions for the Muir–Torre syndrome: a histopathologic and analysis. molecular genetic study. Am J Dermatopathol 1999; 21:405–13. 9 Mathiak M, Ru¨tten A, Mangold E et al. Loss of DNA mismatch As we have previously reported and as our present data repair proteins in skin tumors from patients with Muir–Torre syn- demonstrate, there is a strong genotype–phenotype correlation drome and MSH2 or MLH1 germline mutations: establishment of in MTS: most of the MMR gene mutations in patients with immunohistochemical analysis as a screening test. Am J Surg Pathol MTS were identified in MSH2. MLH1 and MSH6 mutations pres- 2002; 26:338–43. ently appear to be rare in patients with MTS. 10 Mangold E, Pagenstecher C, Friedl W et al. Spectrum and frequencies of mutations in MSH2 and MLH1 identified in 1721 German families suspected of hereditary nonpolyposis colorectal cancer. Int J Acknowledgments Cancer 2005; 116:692–702. 11 Kariola R, Hampel H, Frankel WL et al. MSH6 missense mutations The authors thank the patients for their participation in the are often associated with no or low cancer susceptibility. Br J Cancer study and W. Weyers and M. Winzer for contributing cases. 2004; 91:1287–92. The project was financed by grants of the German Cancer Aid 12 Plaschke J, Engel C, Kruger S et al. Lower incidence of colorectal and BONFOR to E.M. and by a grant of the Deutsche For- cancer and later age of disease onset in 27 families with pathogenic schungsgemeinschaft to R.K. MSH6 germline mutations compared with families with MLH1 or MSH2 mutations: the German Hereditary Nonpolyposis Colorectal Cancer Consortium. J Clin Oncol 2004; 22:4486–94. Institute of Human Genetics, E. MANGOLD 13 Hampel H, Frankel W, Martin E et al. Screening for the Lynch syn- University of Bonn, Wilhelmstrasse 31, N. RAHNER drome (hereditary nonpolyposis colorectal cancer). N Engl J Med 53111 Bonn, Germany N. FRIEDRICHS* 2005; 352:1851–60. *Institute of Pathology, R. BUETTNER* 14 Kruse R, Ru¨tten A, Schweiger N et al. Frequency of microsatellite University of Bonn, Germany C. PAGENSTECHER instability in unselected sebaceous gland neoplasias and hyperpla- Department of Dermatology, S. ARETZ sias. J Invest Dermatol 2003; 120:858–64. University Hospital Duesseldorf, Germany W. FRIEDL § Accepted for publication: 25 July 2006 Department of Dermatology, T. RUZICKA§ University Hospital Munich, Germany P. PROPPING Key words: hereditary nonpolyposis colorectal cancer, Lynch syndrome, MLH1, àLaboratory of Dermatohistopathology, A. RU¨ TTENà MSH2, MSH6, Muir–Torre syndrome Friedrichshafen, Germany R. KRUSE Conflicts of interest: none declared. E-mail: [email protected]

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp158–162 Correspondence

Painless periungual pyogenic granulomata Treatment with silver nitrate was unsuccessful. The lamivu- associated with reverse transcriptase inhibitor dine/zidovudine was discontinued. One month later, the therapy in a patient with human immuno- lesions had nearly resolved (Fig. 1d–f). deﬁciency virus infection The differential diagnosis of pyogenic granuloma in the setting of HIV infection includes Kaposi sarcoma and bacillary DOI: 10.1111/j.1365-2133.2006.07512.x angiomatosis. Amelanotic melanoma, angiosarcoma, metastatic

SIR, We report a patient with human immunodeficiency virus (HIV) who developed painless periungual pyogenic granulomata after starting antiretroviral therapy with lamivudine, zidovudine and efavirenz. There are multiple published cases of painful paronychia, pyogenic-granuloma-like lesions or granulation tissue of the nails associated with protease inhibitors, (a) (b) particularly indinavir, but only two reports of paronychia associated with nonprotease-inhibitor antiretrovirals (lamivudine and zidovudine). Most pyogenic-granuloma-like lesions of the nails associated with antiretroviral therapy have been painful. Painless pyogenic granulomata of the nails may be a distinct entity associated with reverse transcriptase inhibitors. A 53-year-old man with HIV presented with painless, bleeding growths of the nails of the toes and left thumb of 1 month duration. Past medical history included non-Hodgkin lymphoma, pneumocystis pneumonia and cryptococcal menin- (c) gitis. His first exposure to antiretroviral medications was 3 months previously, when lamivudine/zidovudine and efavirenz were started. His other medications were co-trimoxazole, fluconazole and sertraline, which were longstanding treatments. On examination, he appeared well with normal vital signs. Friable, shiny, bright red papules arose between the lateral nail folds and nail plates of the left thumb, right fourth and fifth (d) toes and left first and second toes. The right first toenail appeared to have been shed, and a new nail plate extended approximately 4 mm from the proximal nail fold. Distally, this nail bed was covered with a shiny red plaque similar to the growths on the other affected digits (Fig. 1a–c). Sensation to pinprick and light touch was intact. The rest of the skin examination was unremarkable, and there was no adenopathy. ) The CD4 count was 293 cells mm 3, and viral load was (e) (f) ) 113 copies mL 1. Viral, fungal and bacterial cultures of the Fig 1. Friable, shiny, bright red papules arising between the lateral lesions were negative. A shave biopsy from the left hallux pa- nail folds and nail plates of the left thumb (a), left first and second pule revealed granulation tissue and a lobular arrangement of toes (b) and right fourth and fifth toes (c). The right first toenail capillaries within a background of necrosis and inflammation, appeared to have been shed, and a new nail plate extended consistent with pyogenic granuloma (Fig. 2). Warthin–Starry approximately 4 mm from the proximal nail fold. Distally, this nail stain and immunohistochemical staining for human herpes bed was covered with a shiny red plaque similar to the growths on virus (HHV)-8 with latent nuclear antigen-1 open reading the other affected digits (c). One month after the lamivudine/ frame (ORF) 73 were negative. zidovudine was discontinued, the lesions had nearly resolved (d–f).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 163 164 Correspondence

azole, co-trimoxazole and sertraline have not been associated with the formation of paronychia, pyogenic granulomata, granulation tissue or ingrown nails. We report a case of painless periungual pyogenic granulomata associated with reverse transcriptase inhibitor therapy. (b) This phenomenon may represent a distinct entity that is unrelated to the painful ingrown toenails and paronychia seen with the use of protease inhibitors. This report adds to the small number of published cases of antiretroviral-related ‘paronychia’ (a) not associated with protease inhibitors. In future reports, it would be helpful to distinguish (c) between painful and painless lesions of the nails and to differentiate paronychia, ingrown toenails and pyogenic-gran- Fig 2. Shave biopsy from the left hallux papule revealed granulation uloma-like lesions of the nails. With the increasing use of tissue and a lobular arrangement of capillaries within a background of antiretroviral regimens not based on protease inhibitors, fur- necrosis and inflammation, consistent with pyogenic granuloma (10·, ther cases of reverse-transcriptase-inhibitor-associated pyogenic a; 20·, b; and 40·, c). granulomata of the nails may become apparent. and other skin cancers, granulation tissue and haemangioma Acknowledgments could also be considered.1 Periungual pyogenic granulomata could be confused with the We thank Marcia Usui for her assistance with figures. We are also excess granulation tissue frequently seen in ingrown toenails.2 grateful for support from the Odland Endowed Research Fund. Acute paronychia due to bacterial, fungal or atypical herpes infection could be considered in the differential diagnosis, although Division of Dermatology, L.H. WILLIAMS acute infectious paronychia is not usually associated with granu- Department of Medicine, P. FLECKMAN lation tissue.3 Infectious paronychia and ingrown toenails are Box 356524, heralded by redness, swelling and pain of the nail fold.2,3 How- University of Washington, ever, isolated pyogenic granulomata are usually painless.1 Seattle, WA 98195, U.S.A Pyogenic-granuloma-like lesions of the nails have been de- Correspondence L.H. Williams. scribed in other conditions. Nine patients suffered from E-mail: [email protected] biopsy-proven painful pyogenic granulomata and onychomadesis of the fingernails following the application of a plas- References ter cast to an upper extremity after a fracture.4 Friction and penetrating trauma to the nail plate have induced painful 1 Lin RL, Janniger CK. Pyogenic granuloma. Cutis 2004; 74:229–33. pyogenic-granuloma-like lesions of toenail beds.5 Beau’s lines 2 Ikard RW. Onychocryptosis. J Am Coll Surg 1998; 187:96–102. and pyogenic-granuloma-like lesions of multiple fingernails 3 Tosti A, Piraccini BM. Nail disorders. In: Dermatology (Bolognia JL, Jorizzo JL, Rapini RP, Horn TD, Mancini AJ, Mascaro JM et al., followed hand trauma in two cases6 and Guillain–Barre´ syn- 7 eds), 1st edn, Vol. 1. New York: Mosby, 2003; 1072. drome in one case; it is not clear whether they were painful. 4 Tosti A, Piraccini BM, Camacho-Martinez F. Onychomadesis and Painful paronychia and pyogenic granulomata have been pyogenic granuloma following cast immobilization. Arch Dermatol reported many times in association with protease inhibitors, 2001; 137:231–2. particularly indinavir. Usually, these patients were taking other 5 Richert B. Frictional pyogenic granuloma of the nail bed. Dermatol- antiretrovirals, including lamivudine in several cases.8 ogy 2001; 202:80–1. The use of reverse transcriptase inhibitors has also been 6 Price MA, Bruce S, Waidhofer W, Weaver SM. Beau’s lines and pyogenic granulomas following hand trauma. Cutis 1994; 54: associated with paronychia. In one report, 12 HIV-positive 246–9. patients treated with lamivudine developed paronychia. Whe- 7 Mazereeuw–Hautier J, Bonafe JL. Bilateral Beau’s lines and pyogen- ther the paronychia was painful or featured excessive granula- ic granulomas following Guillain–Barre syndrome. Dermatology tion tissue is not known. Concurrent medications were not 2004; 209:237–8. listed, although lamivudine was said to be the only medication 8 Garcia-Silva J, Almagro M, Pena-Penabad C, Fonseca E. Indinavir- common to all patients. The potential mechanism of action is induced retinoid-like effects: incidence, clinical features and man- unknown.9 Another reverse transcriptase inhibitor, zidovu- agement. Drug Saf 2002; 25:993–1003. 9 Zerboni R, Angius AG, Cusini M et al. Lamivudine-induced parony- dine, was associated with Candida paronychia in a neutropenic 10 chia. Lancet 1998; 351:1256. neonate with mucocutaneous candidiasis. 10 Russo F, Collantes C, Guerrero J. Severe paronychia due to zidovu- Of the medications taken by our patient, only lamivudine dine-induced neutropenia in a neonate. J Am Acad Dermatol 1999; has previously been associated with noninfectious paronychia. 40:322–4. It is not clear whether pyogenic-granuloma-like lesions were present in those cases.9 To our knowledge, efavirenz, flucon- Conflicts of interest: None declared.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 165

Melanomas in renal transplant recipients: the London experience, and invitation to participate in a European study*

DOI: 10.1111/j.1365-2133.2006.07567.x

1 SIR, In response to the report of Le Mire et al. on the incidence of malignant melanoma in Oxford renal transplant recipients (RTRs), we detail our experience of melanoma in a lone; SCC, squamous cell carcinoma; series of RTRs at Bart’s and the London NHS Trust. Although incidence rates in our population were similar to those docu- Duration of follow up (months) Outcome mented in the Oxford study, metastatic disease was more common, and we suspect that RTRs with melanoma may have a worse prognosis than their immunocompetent counterparts,

although larger, multicentre studies are now required to con- Origin in naevus (on histology)

firm this. (Bowen’s disease); IHD, ischaemic heart disease; IS, immuno- A retrospective clinical and histological review of mela- in situ nomas diagnosed in patients attending a dedicated RTR dermatology clinic between 1990 and 2005 was performed. Inflammatory infiltrate All specimens were resectioned and were assessed by an experienced dermatopathologist (R.C.). In a total population AJCC stage of 861 RTRs with a follow-up period of 8557 patient-years, there were seven cases of de novo melanoma (five invasive and two in situ) in three men and four women. All diagnoses were made in the RTR dermatology clinic. Clinicopatho- logical data are presented in Table 1 and are illustrated in /I No 0 None No 24 No recurrence; died of IHD Figure 1. All patients were on prednisolone and azathio- /I No 0 Minimal No 29 No recurrence 25/II No IA None No 144 No recurrence 4/II4/II No No IA IA Minimal Minimal Yes No 69 54 No recurrence; died of NHL Died of metastatic melanoma 7/IV2/IV No Yes IIB Minimal IIB Minimal No No 25 23 Died of metastatic melanoma Died of metastatic melanoma Æ Æ Æ prine, with or without ciclosporin. The mean age at diag- Æ Æ In situ Breslow thickness (mm)/Clark level Ulceration In situ nosis was older than in Oxford [57Æ7 years (range 44–68) vs. 41] and the mean time from transplantation was shorter [102 months (range 10–247) vs. 132]. All patients had skin types I–III (although 20% of our RTR cohort have skin type V or VI). None had a family history of melanoma or fulfilled criteria for atypical mole syndrome, although two Histological type Site patients had several atypical naevi. A high proportion of patients had a history of other invasive skin tumours (five of seven vs. four of 10 in Oxford); in all but one patient (patient 1) these tumours arose prior to the diagnosis of Skin phototype/ atypical naevi III/no NM Back 4 melanoma. Melanomas were more commonly located on the III/yes LM Chest head and neck in our series, with four of seven tumours occurring at this site, in contrast to the Oxford tumours 2 I/no LM Cheek 1 N/A SSM Leg 0 7 2 4 4 13 II/no SSM Cheek 0 which all occurred at extracephalic sites. Ten of 12 7 · · · · · · · tumours in the Oxford series were superficial spreading · BCC Other invasive skin cancers CIS melanomas, whereas this subtype accounted for two of our BCC seven cases. As in the Oxford series, Breslow thickness in

the majority of cases (five of seven) was < 1 mm, and an IS drugs inflammatory response was either minimal or absent in all melanomas, as has previously been documented.2 A contigu- ous pre-existing naevus was present in only one case, and, similarly, was detected in only a minority (three of 12) of Time since transplant (months) the Oxford cases, in contrast to a previous proposal that over half of all transplant melanomas arise in pre-existing Clinicopathological data naevi.2 All melanomas in our series were treated by com- 3. M/60 180 P, A SCC 2. F/68 10 P, A, C BCC 1. F/44 94 P, A BCC Patient no., sex/age at diagnosis (years) 6. F/527. F/57 23 247 P, A, C P, A, C No SCC I/no LMM Nose 0 4. M/585. M/65 114 48 P, A P, A, No C SCC III/yes NM Jaw 2 A, azathioprine; AJCC, American Jointsuppressive; Committee LM, on Cancer; lentigo BCC,SSM, maligna; basal superficial LMM, cell spreading carcinoma; lentigo malignant C, melanoma. maligna ciclosporin; CIS, melanoma; squamous N/A, cell carcinoma not available; NHL, Non-Hodgkin lymphoma; NM, nodular melanoma; P, predniso

plete surgical excision; sentinel lymph node biopsy was not Table 1

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 166 Correspondence

a b

c d

Fig 1. Clinical and histological examples of melanoma in renal transplant recipients. (a) Patient 6: lentigo maligna melanoma on right side of nose, Breslow thickness 0Æ4 mm. (b) Patient 2: lentigo maligna on cheek. (c) Photomicrograph of histology, patient 3: nodular melanoma in vertical growth phase, Breslow thickness 4Æ7 mm. (d) Photomicrograph of histology, patient 4: ulcerated nodular melanoma in vertical growth phase, Breslow thickness 2Æ2 mm. available at the time these tumours were diagnosed. In most increase in males with melanoma in an Irish transplant cases immunosuppression was reduced, but not stopped. population. These more recent studies report rates higher than This series represents an elevated incidence of melanoma in previous estimates.4–6 our London RTR cohort: the rate in 2003 for the general Three patients in our series, but only one patient in population in North East Thames was 6Æ5 and 8Æ1 per Oxford, died from metastatic melanoma. Two deaths 100 000 population for men and women, respectively, and occurred in RTRs with melanomas exceeding 2 mm Breslow we would therefore have expected approximately 0Æ9 cases of thickness. In accordance with American Joint Committee on melanoma to have arisen in our cohort (in which 64% are Cancer staging, the predicted 5-year survival for patient 3 male). If in situ tumours are included, these 7 cases represent was 67% and that for patient 4 was 63%,7 yet both died an approximately 7Æ8-fold increased incidence compared with within approximately 2 years. Even more unexpectedly, the general population. This is remarkably similar to the eight- patient 6 developed metastases from a lentigo maligna mela- fold increased rate found in Oxford, and is of the same order noma of Breslow thickness 0Æ4 mm, for which the expected as that reported by Moloney et al.3 who reported a 6Æ6-fold 5-year survival is 95% or greater. Our series of patients is

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 167 too small to give statistical confirmation of a poorer prog- 6 Sheil AG, Flavel S, Disney AP et al. Cancer development in patients nosis in RTRs and previous studies have also not had suffi- progressing to dialysis and renal transplantation. Transplant Proc cient power to examine this, although a compromised 1985; 17:1685–8. 7 Balch CM, Buzaid AC, Soong SJ et al. Final version of the American outcome from melanoma has similarly been observed in Joint Committee on Cancer staging system for cutaneous mela- individuals with human immunodeficiency virus infection or noma. J Clin Oncol 2001; 19:3635–48. 8 AIDS. 8 Rodrigues LK, Klencke BJ, Vin-Christian K et al. Altered clinical The incidence of melanoma is rising faster than for any course of malignant melanoma in HIV-positive patients. Arch Derma- other major cancer and will continue to do so over at least the tol 2002; 138:765–70. next 30 years.9 In conjunction with the steadily improving 9 Diffey BL. The future incidence of cutaneous melanoma within the long-term survival from organ transplantation, it is likely that U.K. Br J Dermatol 2004; 151:868–72. 10 Otley CC, Berg D, Ulrich C et al. for the Reduction of Immunosup- post-transplant melanoma will emerge as an increasing clinical pression Task Force of the International Transplant Skin Cancer problem in coming years. Many important questions have yet Collaborative and the Skin Care in Organ Transplant Patients Eur- to be answered in this respect: whether prognosis is, indeed, ope. Reduction of immunosuppression for transplant-associated worse for post-transplant melanoma, whether more aggressive skin cancer: expert consensus survey. Br J Dermatol 2006; 154:395– management strategies are therefore required, and how reduc- 400. tion in immunosuppression should be approached. Although consensus statements such as those recently reported in this Conflicts of interest: none declared. Journal are a useful guideline to management of post-transplant skin cancer,10 a firmer evidence base is now required. *Part of these data were first presented at the British Society for Der- matopathology Annual Meeting, July 2002. V.L.B. and R.N.M. have This is a particular priority for management of post-transplant contributed equally to this work. melanoma, and sufficient power to reach meaningful conclusions is only likely to be achieved in the context of a multicentre study. Such a study is currently being coordinated within the SCOPE network (Skin Care in Organ transplant Patients, Europe: http://www.scopnetwork.org) by our centre, and we invite clinicians who are interested in participating to contact us. Melanomas in renal transplant recipients: the London experience, and invitation to partici- *Centre for Cutaneous Research, V.L. BROWN* pate in a European study: reply from authors Institute of Cell and Molecular Science and R.N. MATIN* Department of Pathology, R. CERIO* DOI: 10.1111/j.1365-2133.2006.07568.x Bart’s and the London Queen Mary’s School of M.E. LEEDHAM-GREEN* Medicine and Dentistry, London E1 2AT, U.K. C.M. PROBY* SIR, We are pleased that our previous results showing an Correspondence: Catherine Harwood. C.A. HARWOOD* eightfold increased risk of melanoma in renal transplant E-mail: [email protected] recipients (RTRs)1 is supported and confirmed by the study from London. However, we have re-examined the incidence Acknowledgments in our cohort because these two reports included melanoma in situ, which has a different prognosis and management from C.A.H. and C.M.P. are supported by Cancer Research UK. invasive melanoma and could give a higher incidence ratio C.A.H. is supported by the Research Advisory Board of compared with other centres that excluded in situ melanoma. St Bartholomew’s and The Royal London Charitable Founda- Moreover, since 2002 we have had further new cases of mela- tion (grant RAC404). noma in RTRs in Oxford, including recently a case of invasive melanoma in an Asian patient, indicating that immunosup- References pression may be an independent pathogenetic mechanism for melanoma. 1 Le Mire L, Hollowood K, Gray D et al. Melanomas in renal transplant recipients. Br J Dermatol 2006; 154:472–7. All new cases of invasive melanoma that occurred after 2 Greene M, Young T, Clark WH Jr. Malignant melanoma in renal 2002 in organ transplant recipients at the Oxford Transplant transplant patients. Lancet 1981; 1:1196–9. Centre were added to our previous data and cases of in situ 3 Moloney FJ, Comber H, O’Lorcain P et al. A population-based study melanoma were excluded. Data were taken from the Oxford of skin cancer incidence and prevalence in renal transplant recipi- Transplant Centre Clinical Database, on which all informa- ents. Br J Dermatol 2006; 154:498–504. tion on organ transplant patients and graft outcome are 4 Bouwes-Bavinck JN, Hardie DR, Green A et al. The risk of skin can- entered prospectively. We estimated the incidence of mela- cer in renal transplant recipients in Queensland, Australia. A follow-up study. Transplantation 1996; 61:715–21. noma among patients who underwent organ transplantation 5 Leveque L, Dalac S, Dompmartin A et al. Melanoma in organ trans- between May 1964 and March 2006. Standardized incidence plant patients. Ann Dermatol Venereol 2000; 127:160–5. ratios (SIRs) were calculated by dividing the observed

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Table 1 Clinicopathological data 168 Correspondence Patient no. Features of lesion Breslow Sex/age at Age (years) Time since at diagnosis: thickness Skin burns Duration of diagnosis at ﬁrst transplant size/border/ Histological (mm)/Clark IS Skin phototype/ in childhood/ follow-up (years) transplant (months) Site colour/irregularity type level drugs atypical naevi sun exposure NMSC (months) Outcome 1. M/36: 31 51: 2 MM; Back: 2 MM; 1: 7 · 5 mm/slightly SSM · 31:0Æ6/III; A, C I/yes Yes/+++ No 48 from No recurrence 2MM 88: 1 MM calf: 1 MM blurred/dark 2: 0Æ6/III; 1st MM

ora Compilation Journal (+ 1 in situ); brown/irr; 2: 5 mm/ 3: 0Æ4/III 39: 1 MM slightly blurred/ brown/irr; 3: 5 · 3 mm/blurred/ heterogeneous in colour/irr 2. M/48 41 81 Shoulder 3 cm/blurred/black/ NM 4Æ5/IV P, A I/yes NR/+ BD · 1 9 Died – metastatic irr, exophytic melanoma 06BiihAscaino Dermatologists of Association British 2006 3. F/45 37 96 Forearm 5 mm/slightly blurred/ SSM 0Æ3/II P, A II/yes NR/++ No 219 No recurrence dark brown/ slightly irr 4. M/73 64 109 Calf NR/NR/dark SSM 0Æ48/II P, C II/yes Yes/+++ BCC · 7, 83 Died – cardiac brown/irr SCC · 28, BD · 8 infarct 5. F/45 28 200 Arm 5 mm/well demarcated/ SSM 0Æ37/II C II/no No/+ BCC · 4, BD · 1 45 No recurrence red/irr, plaque 6. M/44 25 230 Back NR/blurred/dark SSM 0Æ49/III P, C II/yes Yes/++ No 69 No recurrence brown/irr (naevoid) 7. M/65 46 237 Abdomen 10 · 8 mm/blurred/ SSM 0Æ6/III P, A II/no No/++ BCC · 1, 61 No recurrence dark brown/irr SCC · 5, BD · 1 8. M/65 64 15 Abdomen 7 mm/NR/dark SSM 0Æ4/III P, A, C II/yes NR BCC · 4, 92 No recurrence brown/irr SCC · 3, BD · 7 •

rts ora fDermatology of Journal British 9. M/60 50 123 Arm 6 mm/blurred/ SSM 1Æ5/IV A, C I/yes Yes/+++ BCC · 1, 16 No recurrence heterogeneous in SCC · 2 colour/irr, pale halo 10. M/57 45 146 Back 6 mm/slightly blurred/ SSM 0Æ3/II P, A, C I/yes Yes/+ No 24 No recurrence dark brown/regular 11. M/41 20 248 Calf 9 · 6 mm/well SSM 0Æ7/III P, A V/no NR/++ No 4 No recurrence demarcated/black/ slightly irr

2007 12. M/46 44 23 Arm 4 mm/slightly blurred/ SSM 0Æ3/II MMF, S II/yes NR ? 4 No recurrence heterogeneous in 06TeAuthors The 2006 156, colour/slightly irr pp163–201 A, azathioprine; BCC, basal cell carcinoma; BD, Bowen’s disease; C, ciclosporin; irr, irregular; IS, immunosuppressive; MM, malignant melanoma; MMF, mycophenolate mofetil; NM, nodular melanoma; NMSC, nonmelanoma skin cancer; NR, not recorded; P, prednisolone; S, sirolimus; SCC, squamous cell carcinoma; SSM, superficial spreading malignant melanoma; +, mild; ++, moderate; +++, excessive. Correspondence 169 number in the Oxford transplant cohort by the expected dermatological clinic for organ transplant patients and follow number in the Oxford region (Berkshire, Buckinghamshire, patients regularly. Northamptonshire and Oxfordshire). Expected numbers of We agree with Brown et al. that, in the future, post- invasive melanoma were calculated by multiplying the age, transplant melanoma may emerge as an important clinical gender and 5-year calendar period (1963–2003) specific problem and agree that a multicentre study will permit us to incidence rates of the region by the corresponding person- optimize management and determine prognosis of melanoma years at risk in the Oxford cohort. Person-years were in organ transplant recipients. We are already involved in the derived from the date of transplantation up to the date of SCOPE (Skin Care in Organ transplant Patients, Europe) study melanoma, the date of failure of last transplantation, the and urge all our colleagues to collaborate in gathering statistic- date of death or the last follow-up, whichever came first. ally significant data and refining management. Assuming a Poisson distribution of the observed numbers, the 95% confidence intervals (CIs) and two-sided P-values Dermatology Department, B. IMKO-WALCZUK were derived. Collegium Medicum in Bydgoszcz and A. LALLY* Between 1964 and 2006, there were 1958 organ recipients Clinical Department of Plastic Surgery and Burns L. LE M IRE* of 2342 organ transplants (kidney or recently kidney and pan- Treatment, Medical University of Gdansk, Poland D. CASABONNE creas), giving a total of 16 676Æ19 total transplant years. Four- *Oxford Radcliffe Hospital, Oxford, U.K. K. HOLLOWOOD* teen invasive melanomas occurred de novo in 12 patients (one Cancer Epidemiology Unit, C. BORDEA* patient had three invasive melanomas), giving a rate of 8Æ4 University of Oxford, U.K. F. WOJNAROWSKA* per 10 000 person-years at risk. The expected number of Correspondence: Louise Le Mire. invasive melanomas in our cohort was 2Æ8, giving a SIR of 5Æ0 E-mail: mfl[email protected] (95% CI 2Æ7–8Æ4; P ¼ 0Æ00004). The patients with melanoma comprised 10 men and two References women. The mean age at first transplant was 40 years (range 20–64) and at diagnosis of melanoma was 50 years (range 1 Le Mire L, Hollowood K, Gray D et al. Melanomas in renal transplant 36–73), similar to our previous study. Patients from Oxford recipients. Br J Dermatol 2006; 154:472–7. were younger than those in London (mean 50 vs. 58 years). 2 Bouwes-Bavinck JN, Hardie DR, Green A et al. The risk of skin cancer in renal transplant recipients in Queensland, Australia. A follow- The mean interval between transplant and development of up study. Transplantation 1996; 61:715–21. melanoma was 121 months (range 15–248) compared with 3 Leveque L, Dalac S, Dompmartin A et al. Melanoma in organ trans- 132 months previously and 102 months in the London study. plant patients. Ann Dermatol Venereol 2000; 127:160–5. Clinical data are shown in Table 1. All patients had skin type 4 Sheil AG, Flavel S, Disney AP et al. Cancer development in patients I–III, with the exception of one Asian patient with skin type progressing to dialysis and renal transplantation. Transplant Proc 1985; V. Five patients had immunosuppression before transplant- 17:1685–8. ation. Most patients had a history of sunburn in childhood and/or Conflicts of interest: none declared. excess sun exposure as a child or adult, nine had multiple atypical naevi, and one patient had a mother with two melanomas at the same age. The most common site was the trunk and upper limb in men, and the upper limb in women. Half of the patients had other skin malignancies (Table 1). Thirteen of 14 cases were superficial spreading melanoma, and one was nodular mela- Clinical utility of an interferon-c-based assay noma. As in our previous study and the London study, Bre- for mycobacterial detection in papulonecrotic slow thickness in the majority of cases (12 of 14) was tuberculid < 1 mm, one patient had a 1Æ5 mm thick melanoma and one a4Æ5 mm thick melanoma. The latter died 9 months after DOI: 10.1111/j.1365-2133.2006.07563.x diagnosis. The rest of the patients have had no recurrences. Immunosuppression was reduced in some cases. SIR, Papulonecrotic tuberculid (PNT) is an uncommon cutane- The data demonstrate an increased incidence of invasive ous hypersensitivity reaction to Mycobacterium tuberculosis, charac- melanoma in the Oxford transplant population. The previously terized by a symmetrical eruption of necrotizing papules with reported increased risk of approximately eight times from atrophic scar formation. We describe a patient with PNT in Oxford and London included both invasive and in situ mela- whom mycobacterial DNA was not detected by polymerase nomas. The SIR in our group for invasive melanoma is 5Æ0 chain reaction (PCR) but the diagnosis of latent tuberculosis and is similar to the results from other centres.2–4 Better out- infection was supported by a whole-blood interferon (IFN)-c- come in our population (one melanoma death) may be the based assay. result of regular surveillance and early detection of melanoma. A 37-year-old Japanese man was referred to us in Septem- We suggest that all transplant centres should have a dedicated ber 2005 for the evaluation of papulonecrotic eruptions in the

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 170 Correspondence

(a) (b)

100µm

Fig 1. (a) Extensor aspect of forearm. A dusky-red, crusted papular lesion with multiple varioliform, atrophic scars. (b) An area of coagulation necrosis. Blood vessels show lymphocytic vasculitis with nuclear dust, ﬁbrinoid degeneration, thrombotic occlusion and complete necrosis (haematoxylin and eosin).

skin of 17 years’ duration. Physical examination revealed sym- sis of PNT. Although the patient was started on a daily metrically distributed, multiple, round, atrophic scars and regimen of rifampicin 450 mg and isoniazid 200 mg, new inflammatory, dusky-red, crusted papules on the extensor papulonecrotic lesions developed on his elbows after 5 months aspects of the elbows, knees and buttocks (Fig. 1a). Crusted of therapy. Therefore, a triple regimen of rifampicin 450 mg, papules were also observed on the ear helices, nape, and the isoniazid 300 mg and pyrazinamide 1Æ5 g was instituted. lateral aspect of the right foot. The patient denied trauma, PNT is one of the tuberculides, a group of disorders that insect bites, or topical application of medication or moxa cau- are thought to be hypersensitivity reactions to M. tuberculosis or tery. The results of laboratory studies were normal. An intra- its products in patients with significant immunity. A definitive dermal test with 10 tuberculin units of purified protein diagnosis of PNT is difficult and is supported mainly by a derivative produced an area of erythema of 75 · 45 mm with strongly positive Mantoux reaction, the tuberculin skin test a20· 20 mm induration at 48 h. Biopsy of a papulonecrotic (TST) and characteristic clinical and histopathological findings. lesion from the left elbow revealed a large central zone of The TST has several disadvantages, including false-positive coagulation necrosis with lymphocytic infiltration in the re- responses due to reactivity caused either by infection with ticular dermis. The necrotic area demonstrated lymphocytic nontuberculous mycobacteria (NTM) or by bacillus Camette– vasculitis with nuclear dust, fibrinoid degeneration, throm- Gue´rin (BCG) vaccination.1 The poor specificity of TST has botic occlusion of individual vessels and complete coagulation previously been reported in Japan where BCG vaccination was necrosis (Fig. 1b). Acid-fast bacilli were not found in the skin performed extensively.2 Although amplification of mycobacte- biopsy specimen. PCR of the snap-frozen tissue from the bi- rial DNA using PCR has become a valuable tool in the rapid opsy specimen was negative for M. tuberculosis and M. avium identification of slow-growing organisms, its reliability as a complex DNA. Neither chest X ray and a computed tomo- diagnostic method in clinical samples of tuberculides is still graphic scan of the neck and chest, nor gallium-67 scintigra- controversial.3–6 In the present case, the mycobacterial DNA phy revealed a mass suggesting a focus of M. tuberculosis was not detected by PCR analysis even of a snap-frozen skin infection. A whole-blood IFN-c assay [QuantiFERON-TB 2nd sample, which would be expected to have a higher sensitivity generation (QFT-2G); Japan BCG Supply, Tokyo, Japan] for than PCR analysis of paraffin-embedded tissue. The absence of ) the detection of M. tuberculosis infection revealed 1Æ688 IU mL 1 mycobacterial DNA could be explained by sampling error, of IFN-c (cut-off 0Æ35) for early secretory antigenic target-6 local rapid destruction of antigen by the inflammatory process, ) (ESAT-6) and 1Æ235 IU mL 1 of IFN-c (cut-off 0Æ35) for cul- the small number of organisms delivered to tissues and the ture filtrate protein-10 (CFP-10), thus establishing the diagno- possible presence of inhibitors.3,6

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 171

QFT-2G, the Japanese equivalent of QuantiFERON-TB Gold (Cellestis Ltd, Carnegie, Australia), has recently been developed as an accurate tool for detecting tuberculosis infection T cell/keratinocyte interactions in psoriasis: regardless of past history of BCG vaccination.1,2,7 This where is the trigger? enzyme-linked immunosorbent assay test detects the release of IFN-c in fresh heparinized whole blood from sensitized DOI: 10.1111/j.1365-2133.2006.07573.x persons when it is incubated with mixtures of synthetic peptides representing two proteins present in M. tuberculosis, SIR, It is now commonly assumed that psoriasis is an ESAT-6 and CFP-10. QFT-2G is expected to be more specific immune-mediated disorder and that skin-infiltrating activated for M. tuberculosis than the TST, because both ESAT-6 and T cells and the local overexpression of proinflammatory cy- 1 CFP-10 are absent from all BCG vaccine strains. However, tokines play a key role in the development of the lesions. because the genes encoding both ESAT-6 and CFT-10 are This has led to the development of various anti-inflammatory present in M. kansasii, M. marinum and M. szulgai, it is of the agents for the treatment of psoriasis, such as antitumour ne- 1 utmost importance to rule out these NTM infections from crosis factor-a reagents. In a recent issue of this Journal, the clinical and histopathological features and acid-fast bacilli Kru¨ger-Krasagakis et al. showed that infliximab treatment of cultures. patients with psoriasis causes a rapid increase in the number 2 We experienced a patient with PNT in whom mycobacterial of apoptosis-like keratinocytes in lesional epidermis, sug- DNA was not detected by PCR but the diagnosis of latent gesting that induction of keratinocyte apoptosis by infliximab tuberculosis infection was supported by QFT-2G. Our results may be complementary to the so-far reported anti-inflamma- indicate that the whole-blood IFN-c-based assay, QFT-2G, tory effects. Methotrexate and psoralen plus ultraviolet A could be an additional tool in diagnosing tuberculides. It is therapy may act as antipsoriatic agents through similar effects 3 of interest to determine the further utility of this assay as on keratinocytes. These observations are in line with recent additional patients with PNT and other tuberculides such as findings, based on abrogation of JunB/activator protein 1 erythema induratum Bazin and lichen scrofulosorum are (AP-1) or on activation of mitogen-activated protein kinase examined. (MAPK) or of cell-signalling molecules such as Stat3 in keratinocytes,4,5 indicating that alterations in epidermal keratino-

Departments of Dermatology and *Internal Medicine, R. TANAKA cyte-based signalling pathways may induce psoriatic lesions. Division of Respiratory Medicine, H. MATSUURA However, whether the primary pathogenic disturbance Kawasaki Medical School, 577 Matsushima, Y. KOBASHI* resides in the skin or in the immune system remains contro- Kurashiki-shi, Okayama 701-0192, Japan W. FUJIMOTO versial. Correspondence: W. Fujimoto. We and others have shown that, from a kinetic point of E-mail: [email protected] view, the architecture of clinically stable psoriatic epidermal lesions may be considered to be a steady-state condition achieved after epidermal cell proliferation has continued for References a sustained period.6,7 However, the link between the in- 1 Mazurek GH, LoBue PA, Daley CL et al. Comparison of a whole- flammatory infiltrate present in psoriatic lesions and this blood interferon c assay with tuberculin skin testing for detecting abrupt change in epidermal homeostasis leading to a latent Mycobacterium tuberculosis infection. JAMA 2001; 286:1740–7. new steady state of the germinative layer is not clear. More 2 Mori T, Sakatani M, Yamagishi F et al. Specific detection of tubercu- particularly, one should expect that if immune activation losis infection. An interferon-c-based assay using new antigens. Am J persists, epidermal hyperplasia should be progressive. Anal- Respir Crit Care Med 2004; 170:59–64. 3 Victor T, Jordaan HF, van Niekerk DJT et al. Papulonecrotic tubercu- ogously, the exact mechanisms leading to psoriasis remis- lid. Am J Dermatopathol 1992; 14:491–5. sion, i.e. return of epidermal architecture to its original 4 Quiros E, Bettinardi A, Quiros A et al. Detection of mycobacterial shape, need further clarification. The finding that effective DNA in papulonecrotic tuberculid lesions by polymerase chain reac- repair of psoriatic skin involves keratinocyte apoptosis in tion. J Clin Lab Anal 2000; 14:133–5. addition to anti-inflammatory effects suggests that both 5 Tan SH, Tan HH, Sun YJ, Goh CL. Clinical utility of polymerase hypotheses are not mutually exclusive. Rather, T cells and chain reaction in the detection of Mycobacterium tuberculosis in different keratinocytes are likely to pair up to generate psoriasis so types of cutaneous tuberculosis and tuberculids. Ann Acad Med Singapore 2001; 30:3–10. that nonlinear mathematics (chaos theory) may be more 6 Senturk N, Sahin S, Kocagoz T. Polymerase chain reaction in cutane- appropriate than linear mathematics to explain psoriasis. ous tuberculosis: is it a reliable diagnostic method in paraffin- Catastrophe theory may be of some relevance for this pur- embedded tissues? Int J Dermatol 2002; 41:863–6. pose. Catastrophe theory typically describes the progressive 7 Mazurek GH, Jereb J, Lobue P et al. Guidelines for using change in the forces leading to eventual breakdown in the QuantiFERON-TB Gold test for detecting Mycobacterium tuberculo- homeostasis, which may be subject to predictable delay sis infection, United States. MMWR Recomm Rep 2005; 54:49–55. (hysteresis) after the change in force occurs.8,9 Catastrophe theory has been used in a wide range of disciplines, such Conflicts of interest: none declared. as geology (e.g. to model fault formation), economics

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 172 Correspondence

(e.g. to model crashes in stock markets) or biology (e.g. to model tumour growth or tumour remission).10 In several aspects, psoriasis behaves more as a catastrophic Dapsone in the management of ‘insect event than as a linear process. Firstly, psoriasis involves bite-like reaction’ in a patient with chronic homeostasis and jumps. Secondly, there is an underlying lymphocytic leukaemia disposition toward maintaining homeostasis (cell loss by desquamation is compensated by an equivalent cell produc- DOI: 10.1111/j.1365-2133.2006.07576.x tion by the germinative compartment in normal skin as well as in established psoriatic lesions).6 Thirdly, the result- SIR, A 49-year-old woman presented with a 9-month history ing distribution of epidermal kinetics is bimodal: the under- of recurrent skin lesions. Treatment with several broad-spec- lying conditions can produce psoriatic lesions or normal trum antibiotics had been ineffective. There was no history of skin, depending on the host and existence of a trigger. insect bites. The patient had new lesions also during the win- Fourthly, under certain conditions, a slight change in an ter and when living at an altitude of > 2000 m. She had underlying condition can cause a jump from one equilib- chronic lymphocytic B-cell leukaemia which was diagnosed in rium to another, and thereby a radical change in epidermal March 2003 and was treated with ﬂudarabine and cyclo- kinetics. Thus, the ﬁnding that some effective antipsoriatic phosphamide until November 2004. At presentation, thera- therapies target both T cells and keratinocytes2 may be not peutic intervention for leukaemia was not indicated. just a coincidence but a necessity to display sustained thera- On admission, the patient had skin lesions at different sta- peutic effects. ges of evolution on her left cheek (Fig. 1), upper arm and neck. Each lesion showed a typical course of 10 days’ dur-

Department of Dermatology, Erasme University T. SIMONART ation: red papules progressed quickly to subcutaneous purple Hospital, 808 Route de Lennik, M. HEENEN nodules with surrounding erythematous swelling. The nodules B-1070 Brussels, Belgium showed central fluctuation and vesicle formation, broke down, E-mail: [email protected] discharged a great amount of serous material and resolved spontaneously without scar formation. The patient had splenomegaly and 3% leukaemic cells in her blood. References Histopathological examination revealed massive subepider- 1 Schon MP, Boehncke WH. Psoriasis. N Engl J Med 2005; 352:1899– mal oedema with extravasated erythrocytes, fibrin exudates 912. and eosinophils washing away the epidermis (Fig. 2). The 2 Kru¨ger-Krasagakis S, Galanopoulos VK, Giannikaki L et al. Pro- entire dermis was occupied by a dense perivascular, interstitial grammed cell death of keratinocytes in infliximab-treated plaque- and perifollicular lymphocytic infiltrate with many eosinophils type psoriasis. Br J Dermatol 2006; 154:460–6. that reached the subcutaneous fat. Lymphoctyes and eosinoph- 3 Laporte M, Galand P, Fokan D et al. Apoptosis in established and healing psoriasis. Dermatology 2000; 200:314–16. ils also infiltrated the follicular epithelium. More than 95% of 4 Hobbs RM, Silva-Vargas V, Groves R, Watt FM. Expression of acti- the lymphocytes were CD3+ and CD5+ T cells. Only a minor- vated MEK1 in differentiating epidermal cells is sufficient to gener- ity was CD20+ B lymphocytes. ate hyperproliferative and inflammatory skin lesions. J Invest Dermatol Direct and indirect immunfluorescence did not show find- 2004; 123:503–15. ings of bullous pemphigoid or other immunobullous diseases. 5 Sano S, Chan KS, Carbajal S et al. Stat3 links activated Circulating IgG autoantibodies against bullous pemphigoid keratinocytes and immunocytes required for development of antigens were not present. Molecular genetic findings revealed psoriasis in a novel transgenic mouse model. Nat Med 2005; 11:43–9. no T cell receptor gene rearrangements. However, clonal im- 6 Heenen M. On the morphogenesis of a psoriatic lesion. J Invest Der- munoglobulin gene rearrangements were detected. Skin swabs matol 1998; 111:174. were all negative for bacteria and fungi. Neither herpes sim- 7 Iizuka H, Takahashi H, Ishida-Yamamoto A. Pathophysiology plex DNA nor varicella-zoster virus DNA was detected by of generalized pustular psoriasis. Arch Dermatol Res 2003; 295:S55– polymerase chain reaction (PCR). 9. Due to the clinical and histopathological findings the diag- 8 Zeeman EC. Applications of Catastrophe Theory. Tokyo: Tokyo University nosis of insect bite-like reaction was made. Treatment with Press, 1973. 9 Thom R. Structural Stability and Morphogenesis: An Outline of a General Theory prednisolone at a maximum dose of 20 mg daily was not ef- of Models. New York: Addison-Wesley, 1989. fective. Three weeks after initiation of treatment with dapsone 10 Adam JA. Mathematical models of prevascular spheroid 100 mg daily, the skin lesions markedly improved and only development and catastrophe-theoretic description of rapid meta- small papules were noted. Tapering off dapsone after several static growth/tumor remission. Invasion Metastasis 1996; 16:247– weeks of therapy resulted in development of new lesions and 67. led to reintroduction of dapsone, which was again well tolerated. Methaemoglobinaemia was 3Æ9% and haemoglobin was Conflicts of interest: none declared. ) 11Æ4gdL 1.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 173

Fig 1. A subcutaneous nodule with surrounding erythematous swelling on the left cheek (a) resolved without scar formation while a new red papule evolved at the nasal tip (b).

The exact pathogenesis of insect bite-like reactions is largely unknown. Similar lesions have been reported in patients with human immunodeficiency virus infection6 and congenital agammaglobulinaemia,7 indicating that an altered immune reactivity most probably plays a causative role. Minor numbers of neoplastic cells that were detected by highly sensitive PCR- based methods in our case most probably represent part of an exaggerated immunological response to an unknown stimulus. Alternatively, the gene rearrangements may reflect contamination of the specimen with leukaemic peripheral lymphocyte DNA.1 Treatment of insect bite-like reactions is difficult. Antibiotics are ineffective. Corticosteroids at a dose of prednisolone 40 mg daily or more may suppress the eruptions but lesions recur when the dose is reduced.2,4 Treatment with dapsone Fig 2. Histopathology demonstrates massive subepidermal oedema, 100 mg daily led to marked improvement in our patient. We displacement of the epidermis and a dense lymphocytic infiltrate with are thus able to confirm a recently published case report of · eosinophils (haematoxylin and eosin; original magnification 40). Blum et al.4 who successfully used dapsone in a patient with an insect bite-like reaction and leukaemia. Dapsone is known Skin eruptions in patients with chronic lymphocytic leukae- positively to influence disorders with abnormal neutrophil and mia can be due to cutaneous involvement by malignant cells or eosinophil accumulation.8,9 to a nonmetastatic phenomenon. Among the latter group skin Our case highlights the importance of recognizing the dis- lesions in association with insect bites or simulating insect bites turbing nonspecific phenomenon of insect bite-like reaction in 1–5 have been infrequently described. As most patients do not patients with leukaemia. As in rare cases these reactions may 2 recall arthropod bites the term ‘insect bite-like reaction’ seems even precede the diagnosis of haematoproliferative disease a most appropriate. The morphology of the lesions varies widely: targeted evaluation is warranted.1,2 Although controlled stud- 1 1,2 1 3 macules, papules, nodules, some with central puncta, ies are missing, dapsone may be a therapeutic option. oedematous plaques2 as well as vesicles,1 bullae,1,3,4 vesiculo- 5 1,2 bullous eruptions and ulcers have been described. Universita¨ts-Hautklinik, Liebermeisterstrasse 25, A. ULMER The most important differential diagnosis includes specific D-72076 Tu¨bingen, Germany G. METZLER infiltrates of leukaemia. Characteristic histopathological find- E-mail: [email protected] S. SCHANZ ings of an insect bite-like reaction are a T cell-dominant G. FIERLBECK lymphocytic infiltrate accompanied by eosinophils in the superficial and deep dermis in association with massive subepidermal oedema. Autoimmune blistering disorders such as bul- References lous pemphigoid may be excluded by direct and indirect 1 Davis MD, Perniciaro C, Dahl PR et al. Exaggerated arthropod-bite immunofluorescence. lesions in patients with chronic lymphocytic leukemia: a clinical,

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 174 Correspondence

histopathologic, and immunopathologic study of eight patients. JAm were smooth-surfaced brownish infiltrated plaques and nod- Acad Dermatol 1998; 39:27–35. ules, 0Æ5–5 cm in diameter, on her chest, abdomen and back 2 Barzilai A, Shpiro D, Goldberg I et al. Insect bite-like reaction in (Fig. 1a). In addition, we found enlarged cervical, axillary and patients with hematologic malignant neoplasms. Arch Dermatol 1999; inguinal lymph nodes. Laboratory examination of her periph- 135:1503–7. 3 Weed RI. Exaggerated delayed hypersensitivity to mosquito bites in eral blood revealed normal values for red blood cells and chronic lymphocytic leukemia. Blood 1965; 26:257–68. white blood cells with a normal differential count and an ele- 9 )1 4 Blum RR, Phelps RG, Wei H. Arthropod bites manifesting as recur- vated platelet count of 676 · 10 L . Serum protein electro- rent bullae in a patient with chronic lymphocytic leukemia. J Cutan phoresis showed a polyclonal increase in c-globulins with ) Med Surg 2001; 5:312–14. elevated immunoglobulin levels: IgG, 5230 mg dL 1 (normal ) 5 Rosen LB, Frank BL, Rywlin AM. A characteristic vesiculobullous 748–1694); IgA, 610 mg dL 1 (normal 91–391); IgM, eruption in patients with chronic lymphocytic leukemia. J Am Acad ) 1130 mg dL 1 (normal 33–254). Her serum IL-6 level was Dermatol 1986; 15:943–50. )1 6 Diven DG, Newton RC, Ramsey KM. Heightened cutaneous reac- 38Æ9pgmL (normal £ 4Æ0). Urinalysis revealed a trace of tions to mosquito bites in patients with acquired immunodefi- protein but no Bence-Jones protein. Computed tomographic ciency syndrome receiving zidovudine. Arch Intern Med 1988; (CT) scans disclosed the presence of multiple enlarged paratra- 148:2296. cheal, pararenal, cervical and axillary lymph nodes. Skin bi- 7 Shibasaki M, Sumazaki R, Takita H. Hypersensitive reactions to mos- opsy from one of the cutaneous plaques on the back revealed quito bites in congenital agammaglobulinemia. Ann Allergy 1986; a focal perivascular infiltration of plasma cells mixed with 56:81–4. lymphocytes throughout the dermis (Fig. 1b, c). However, 8 Suda T, Suzuki Y, Matsui T et al. Dapsone suppresses human neutrophil superoxide production and elastase release in a calcium- the formation of lymphoid follicles was scarce in the dermis. dependent manner. Br J Dermatol 2005; 152:887–95. Histological sections from a cervical lymph node showed mod- 9 Pfeiffer C, Wozel G. Dapsone and sulfones in dermatology: overview erate hyperplasia of follicles without any clear atypical struc- and update. J Am Acad Dermatol 2003; 48:308–9. ture. The interfollicular space was heavily infiltrated by plasma cells together with lymphocytes. There was no proliferation of Conflicts of interest: none declared. small vessels with hyaline walls (Fig. 1d). Bone marrow biopsies demonstrated the features of normocellular marrow. Im- munostaining of both skin and lymph node showed no j or k light chain restriction. Because there was hardly any improvement with prednisolone 50 mg daily, we added melphalan 10 mg daily for 4 days every 4 weeks. She showed a little improvement with this treatment, but chronic myelomonocytic Indurated nodules and plaques showing a leukaemia (CMML) occurred 1 year later. While melphalan dense plasma cell infiltrate as a cutaneous might induce the onset of CMML, the possibility that the manifestation of Castleman’s disease CMML developed from CD could not be excluded. Patient 2. A 36-year-old Japanese woman had a 2-year his- DOI: 10.1111/j.1365-2133.2006.07577.x tory of multiple asymptomatic cutaneous nodules that had gradually developed on her trunk. More recently, she began to SIR, Castleman’s disease (CD) is an unusual lymphoid hyper- develop fatigue and slight fever up to 37Æ5 C. The cutaneous plasia that is thought to be induced by an immunological lesions were brownish, indurated nodules and plaques, 0Æ5– reaction to a virus or another infectious organism or to 4 cm in size (Fig. 2a). In addition, her axillary and inguinal drugs.1,2 The hyperplastic lymphoid cells in CD overproduce lymph nodes were enlarged. Laboratory studies revealed a red ) interleukin (IL)-6, which induces hypergammaglobulinaemia blood cell count of 3Æ39 · 1012 L 1 and a platelet count of ) and vascular proliferation.3 Although CD commonly occurs 741 · 109 L 1, but a normal white blood cell count and dif- in the mediastinal lymph nodes, there are scattered reports ferential. Serum protein electrophoresis showed a polyclonal on a variety of nonspecific skin lesions including maculopap- increase in c-globulins and immunoglobulin levels as follows: ) ) ) ular eruptions, pemphigus vulgaris and Kaposi sarcoma.4,5 IgG, 5765 mg dL 1; IgA, 467 mg dL 1; IgM, 531 mg dL 1. ) However, cutaneous involvement similar to that noted in the Her serum IL-6 level was 22Æ5pgmL 1. Urinalysis revealed lymph nodes is extremely rare in CD.6–8 Such skin involve- no Bence-Jones protein. CT scans disclosed the presence of ment in CD might have been overlooked or unreported multiple enlargements in the paratracheal, paraortal, cervical because CD is not well recognized in the dermatological and axillary lymph nodes. Skin biopsy from a cutaneous nod- field. We present two Japanese patients with CD who ule showed prominent lymphoid cell infiltrates and the forma- showed numerous cutaneous lesions consisting of indurated tion of lymphoid follicles with atrophic germinal centres nodules and plaques. throughout the dermis and subcutis (Fig. 2b, c). The histopa- Patient 1. A 45-year-old Japanese woman had a 3-year his- thology of the cervical lymph node showed a heavy infiltra- tory of gradually developing multiple asymptomatic cutaneous tion of plasma cells intermixed with small lymphocytes in patches and nodules on her trunk. She had lost 15 kg in the interfollicular space (Fig. 2d). In addition, some follicular weight in the previous 1 year. On physical examination, there centres were radially traversed by hyalinized capillary-sized

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 175

(a)

Fig 1. (a) Numerous brownish nodules and patches on the upper back of patient 1. (b, c) Histopathological features of a skin lesion show a perivascular inﬁltration of plasma cells together with lymphocytes. (d) Histopathological section from a cervical lymph node shows proliferation of plasma cells in the interfollicular space (haematoxylin and eosin; original magniﬁcation: b, · 20; c, d, · 200).

vessels. Immunostaining for j and k light chains confirmed be found hyalinized sclerotic collagen arranged radially around the polyclonal nature of the plasma cell infiltrate. Treatment the vessels. In the plasma cell type, the interfollicular areas with prednisolone 20 mg daily moderately improved her fever contain an infiltrate of predominantly plasma cells. The plasma and fatigue. cell type is often multicentric, and such patients often have CD is a benign lymphoproliferative disorder that is charac- systemic symptoms, such as fever and polyclonal hypergam- terized by marked enlargement of the lymph nodes,1,2 usually maglobulinaemia. This type may be associated with POEMS those of the mediastinum, while other locations might also syndrome, a rare plasma cell dyscrasia characterized by the be involved but less commonly. Recently, the importance of combination of polyneuropathy, organomegaly, endocrinopa- infection with human herpesvirus 8 (HHV-8) has been thies, M protein and skin changes. POEMS syndrome is associ- emphasized in the aetiology and management of CD.9 HHV-8 ated with HHV-8 infection and an increased level of serum was detected in lymph nodes excised from patients with CD, IL-6, similar to CD.12 In our two patients, the histopathologi- suggesting that HHV-8 infection may mediate cytokine abnor- cal features of both skin and lymph node lesions revealed lym- malities and plasma cell hyperproliferation. phoid follicules with a dense plasma cell infiltrate, which Histologically, CD can be classified into a hyaline-vascular suggested specific cutaneous involvement of the plasma cell type (80–90%), plasma cell type (10–20%) and intermediate type CD. The cutaneous plasma cell infiltration found in our types.10,11 The hyaline-vascular type is characterized by the patients is distinct from the skin changes reported in POEMS presence of small germinal centres with concentric layering of syndrome including hyperpigmentation, hypertrichosis, hyper- the mantle lymphocytes. Between the lymphoid follicles can hidrosis and angiomas.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 176 Correspondence

(a)

Fig 2. (a) Nodular lesions on the abdomen of patient 2. (b, c) Histopathological features of a skin lesion show a nodular lymphoproliferative inﬁltration extending from the mid-dermis into the deep dermis. There is lymphoid follicle formation. (d) Cervical lymph node shows similar histopathological changes (haematoxylin and eosin; original magniﬁcation: b, · 40; c, d, · 100).

For differential diagnosis of CD, we need to consider the aberrant population of mantle zone B lymphocytes in paraffin- possibility of low-grade B-cell lymphomas, although monoclo- embedded lymph node biopsies. Am J Clin Pathol 1996; 105: nality was excluded by immunostaining for j and k light 268–76. 3 Casper C. The aetiology and management of Castleman disease at chains. In fact, there has been a report of B-cell lymphoma 50 years: translating pathophysiology to patient care. Br J Haematol developing in some cases of CD which initially showed poly- 2005; 129:3–17. 2 clonal populations. In addition, cutaneous plasmacytoma 4 Frizzera G, Peterson BA, Bayrd ED et al. A systemic lymphoprolifer- should also be considered due to the abundant plasma cell ative disorder with morphologic features of Castleman’s disease: infiltrate in the skin lesions. However, the presence of mul- clinical findings and clinicopathologic correlations in 15 patients. tiple lymph node hyperplasia in the mediastinum and the J Clin Oncol 1985; 3:1202–16. increase of serum IL-6 definitely eliminated such a possibility 5 Kingsmore SF, Silva OE, Hall BD et al. Presentation of multicentric Castleman’s disease with sicca syndrome, cardiomyopathy, palmar in the present cases. and plantar rash. J Rheumatol 1993; 20:1588–91. CD has been considered rarely to involve the skin. How- 6 Kubota Y, Noto S, Takakuwa T et al. Skin involvement in giant ever, from our recent experience in these two cases the possi- lymph node hyperplasia (Castleman’s disease). J Am Acad Dermatol bility that cutaneous involvement in CD might have been 1993; 29:778–80. overlooked in the past cannot be excluded. This possibility 7 Skelton HG, Smith KJ. Extranodal multicentric Castleman’s disease should always be considered when patients present with skin with cutaneous involvement. Mod Pathol 1998; 11:93–8. lesions together with lymph node enlargement. 8 Sleater J, Mullins D. Subcutaneous Castleman’s disease of the wrist. Am J Dermatopathol 1995; 17:174–8. 9 Soulier J, Grollet L, Oksenhendler E et al. Kaposi’s sarcoma-associ- Departments of Dermatology, *Rheumatology and R. OKUYAMA ated herpesvirus-like DNA sequences in multicentric Castleman’s Hematology, Anatomic Pathology and H. HARIGAE* disease. Blood 1995; 86:1276–80. Hematologic Pathology, Tohoku University T. MORIYA 10 Frizzera G, Banks PM, Massarelli G et al. A systemic lymphoprolifer- Graduate School of Medicine, 1-1 Seiryo-machi, S. KAGATANI ative disorder with morphologic features of Castleman’s disease. Aoba-ku, Sendai 980-8574, Japan H. TAGAMI Pathological findings in 15 patients. Am J Surg Pathol 1983; 7:211– 31. E-mail: [email protected] R. ICHINOHASAMA 11 Keller AR, Hochholzer L, Castleman B. Hyaline-vascular and S. AIBA plasma-cell types of giant lymph node hyperplasia of the mediastinum and other locations. Cancer 1972; 29:670–83. References 12 Belec L, Mohamed AS, Authier FJ et al. Human herpesvirus 8 infection in patients with POEMS syndrome-associated multicentric Cas- 1 Castleman B, Iverson L, Menendez VP. Localized mediastinal lymph- tleman’s disease. Blood 1999; 93:3643–53. node hyperplasia resembling thymoma. Cancer 1956; 9:822–30. 2 Menke DM, Tiemann M, Camoriano JK et al. Diagnosis of Castle- Conflicts of interest: none declared. man’s disease by identification of an immunophenotypically

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 177

Retrospective enquiries revealed that the selenium dosage taken was by accident approximately 10 times higher than in- Selenium intoxication: undesirable effect of a dicated according to the treatment protocol. fasting cure Serum analysis revealed a selenium concentration of ) ) 4Æ4 lmol L 1 for the woman, and 4Æ9 lmol L 1 for the man DOI: 10.1111/j.1365-2133.2006.07578.x (normal range 0Æ75–1Æ8). Blood count, liver and kidney function tests, electrolytes, iron, ferritin, thyroxin and biotin in SIR, Selenium is considered an essential nutrient for humans serum, blood analysis for lead and mercury levels, as well as 1 and animals and is a cofactor of several important enzymes. urine analysis for arsenic, were within normal range. Based on Selenium deficiency syndrome is well known from the litera- the clinical symptoms and these findings, the diagnosis of ture as ‘Keshan disease’ and was first described in China. acute selenium intoxication was made. However, acute and chronic selenium poisoning via inhalation The gastrointestinal symptoms and headaches decreased or oral ingestion is rarely observed. Selenium poisoning is directly after discontinuation of the treatment. Four to five predominantly caused by occupational exposure, with inhala- weeks after admission to our clinic, both patients showed 2 tion usually having a fatal outcome. complete onycholysis of all nails of the hands and feet. Almost A married couple, the man aged 51 years and the woman 2 months afterwards the growth of nails and hair restarted. aged 44 years, was admitted to our outpatient clinic with an Serum selenium levels in both patients returned to within the acute massive loss of hair on the head that had started only a normal range after 2 months. No special therapy for the few days previously (Fig. 1). Until the week before admission, intoxication was given. they had undergone a fasting treatment (Mayr therapeutic fast- Selenium is an essential trace element in the human body ing cure) in a special clinic for 14 days. and plays a role in detoxification processes. It is a cofactor of The fasting treatment included an extremely hypocalorific several enzymes such as glutathione peroxidase, which con- diet of three meals a day, each consisting of a potato or a verts free radicals to harmless chemical species. In addition, bread roll with a glass of milk. On a daily basis, they selenium has a cancer-inhibiting effect. The World Health received a systemic supplement of selenium powder as well Organization specifies daily requirements of 50–100 lgtobe as magnesium powder, zinc drops and a teaspoon of laxative taken via oral ingestion. Main sources of selenium include salts. meat, egg yolk, cereals, legumes and nuts. During the first days of the treatment, severe diarrhoea was Acute selenium intoxication is usually due to excessive oral recorded. A few days later, they suffered from nausea and a or respiratory intake of hydrogen selenite. Overdose leads clin- worsening general condition. The man also developed a severe ically to gastrointestinal disturbances such as nausea, vomiting headache. During their treatment both patients lost 5–6 kg in or diarrhoea, headaches and hair loss. This is in addition to weight. One to two days after having finished the treatment, a nail discoloration and fragility, eventually leading to nail loss. considerable diffuse hair loss occurred resulting in a very sig- Following inhalation of hydrogen selenite the upper respira- nificant thinning of head hair with only a few short hairs tory tract is irritated which may lead to pulmonary oedema. remaining. In parallel, white-brownish discolorations of the Chronic intoxication can result in liver damage and peripheral nails appeared (Fig. 2). neuropathy.3–5

(a) (b)

Fig 1. (a, b) Diffuse efﬂuvium in the man only a few days after discontinuation of the Mayr fasting cure. The woman showed identical symptoms.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 178 Correspondence

References

1 Muth OH, Oldﬁeld JE, Weswis PH (eds). Selenium in Biomedicine. West- port, CT: Avi, 1967. 2Ko¨ppel C, Baudisch H, Beyer KH et al. Fatal poisoning with selenium dioxide. J Clin Toxicol 1986; 24:21–35. 3 Srivastava AK, Gupta BN, Bihari V, Gaur JS. Generalized hair loss and selenium exposure. Vet Hum Toxicol 1995; 37:468–9. 4 Vinceti M, Wie ET, Malagoli C et al. Adverse health effects of selenium in humans. Rev Environ Health 2001; 16:223–51. 5 Yang GQ, Wang SZ, Zhou RH, Sun SZ. Endemic selenium intoxication of humans in China. Am J Clin Nutr 1983; 37:872–81. 6 Schellmann B, Raithel HJ, Schaller KH. Acute fatal selenium poisoning. Toxicological and occupational medical aspects. Arch Toxicol 1986; 59:61–3. 7 Sioris LJ, Guthrie K, Peutel PR. Acute selenium poisoning. Vet Hum Toxicol 1980; 22:364.

Conﬂicts of interest: none declared.

Fig 2. Clinical appearance of the man’s thumb, showing white- brownish discoloration of the nail. Differential clinical response to alefacept in Our two patients accidentally ingested selenium orally at a combination with methotrexate in two patients concentration 10 times higher than that intended. As a result, with refractory palmar psoriasis clear symptoms of acute selenium intoxication were noticed a few days later. The determination of the selenium level in DOI: 10.1111/j.1365-2133.2006.07571.x serum revealed for both patients concentrations that were three times above the normal range. SIR, We here report on the use of alefacept in combination Only a few cases of acute selenium intoxication in humans with methotrexate in two patients with palmar psoriasis. These have been reported previously. In these cases, the most com- patients have significantly more physical disability compared mon symptoms described, among others, were nausea and with patients with chronic plaque psoriasis. Thus, treatment of diarrhoea. However, according to the literature death resulting palmar psoriasis is often challenging and at times unsatisfac- from haemolysis and cardiovascular failure occurred after poi- tory. Specifically, two patients with extensive and recalcitrant soning with exceptionally high doses.2,4,6,7 palmoplantar psoriasis were studied. In commercially available selenium preparations, such as A 61-year-old female presented with an 8-year history of sodium selenite, daily dosages of 100 lg are recommended, psoriatic plaques of the palms and psoriatic arthritis of the which reach 300 lg on a short-term basis. It is known that finger joints. The patient failed to respond to several treat- orally or parenterally administered selenium has a limited ments including fumaric acid, ciclosporin, psoralen plus therapeutic range.4 No agreement has been reached in the lit- ultraviolet A (PUVA), systemic retinoids and systemic cortico- ) erature regarding the safe upper limit of selenium intake. steroids. On treatment with methotrexate (20 mg week 1) Studies on selenium in humans revealed that toxic effects over 1 year, the patient still experienced erythematous hyper- occur following a daily intake of at least 300 lg. keratotic plaques of the palms with distinct pustules ) Noncritical intake of selenium without a physician’s recom- (Fig. 1a). A therapeutic trial with alefacept (15 mg week 1), mendation has to be urgently dissuaded. Our two cases show which was administered intramuscularly over a period of that during the administration of selenium, for example as 12 weeks, was initiated in addition to treatment with metho- part of a fasting treatment, it is necessary to consider its lim- trexate. The patient achieved significant improvement after ited therapeutic range in order to avoid an overdose and rela- receiving six doses of alefacept and skin lesions were almost ted toxicity. cleared at week 12 without any adverse effect and with a stable CD4+ T cell count (Fig. 1b). The patient received a Department of Dermatology and Venereology, B. SCHUH second treatment cycle consisting of 12 additional doses of University of Heidelberg, Voßstrasse 2, U. JAPPE alefacept leading to an almost complete and stable remission 69115 Heidelberg, Germany of palmar psoriasis. The significantly improved clinical status Correspondence: Uta Jappe. was maintained by continued treatment with methotrexate ) E-mail: [email protected] (20 mg week 1) and topical emollients with urea. The

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 179

(a) (b)

Week 0 Week 12

Fig 1. Psoriatic plaques of the palms of two patients at baseline and after 12 weeks of treatment with alefacept: patient 1 (a, b) and patient 2 (c, d). patient reported substantial improvement in her quality of patients with palmoplantar psoriasis make up a small percent- life. She has had no adverse events and CD4+ T cell counts age within the cohort of patients with psoriasis, they are sig- have remained within normal limits. nificantly physically disabled and more prone to being The second patient is a 57-year-old man with severe, ther- resistant against otherwise effective antipsoriatic treatment apy-resistant palmar psoriasis of 12 years’ duration leading to regimens. professional disability caused by significant impairment of Alefacept is a fully human dimeric fusion protein of soluble finger motility. Psoriatic plaques were extensive but limited lymphocyte function antigen-3 (LFA-3) with the Fc fragment to hands with localized fissuring. Several conventional treat- of human IgG1. It binds to CD2 on memory T cells and ment modalities including fumaric acid, ciclosporin, metho- blocks the interaction between LFA-3 on antigen-presenting trexate, systemic retinoids, systemic corticosteroids and PUVA cells and CD2 on activated T cells leading to inhibition of T- were applied, but induced no or only short remissions. In cell activation and proliferation.3 It also binds to natural killer light of persisting erythematous plaques with fissuring and cells via its Fc region and induces apoptosis of memory T scaling of the palms (Fig. 1c), alefacept treatment was initi- cells, limiting the inflammatory and uncontrolled keratinocyte ated at 15 mg intramuscularly once weekly for 12 weeks in proliferation in psoriatic skin.3,4 Alefacept has been approved addition to ongoing treatment with methotrexate by the Food and Drug Administration in the United States as ) (20 mg week 1). During this time the patient received top- monotherapy for the treatment of moderate to severe chronic ical treatment with 1% prednisone ointment. At week 12, plaque psoriasis. It is also marketed for the same indication in the patient did not show any significant improvement of the Australia, Canada, Argentina, Switzerland, Kuwait and Israel.4–6 psoriatic lesions (Fig. 1d). Circulating CD4+ T-cell counts Since its approval in 2003 over 12 000 patients with psoriasis remained stable throughout treatment and no adverse events have been treated with alefacept. At present, alefacept can be occurred. regarded as a moderately effective antipsoriatic drug with a Palmoplantar psoriasis is a disabling disease characterized by delayed onset of action and a favourable safety profile with- recurrent sterile pustules associated with erythema, fissuring out an increased risk for secondary malignancies or serious and scaling. The disorder typically has a chronic relapsing infections.7–11 course which last decades. Genetic and environmental factors The cases presented here show differential responsiveness of have been implicated in its aetiology.1,2 Topical treatment is palmar psoriasis to treatment with alefacept in combination frequently not as effective as in psoriasis vulgaris. Treatment with methotrexate. Although monotherapy with alefacept or options include systemic retinoids, ciclosporin, methotrexate methotrexate may be an effective treatment for moderate to and PUVA. Among these therapeutic options, systemic retin- severe psoriasis vulgaris, as both drugs affect the T-cell com- oids, topical glucocorticoids and PUVA are reasonably effective ponent of psoriasis pathogenesis. Alefacept inhibits T-cell acti- but may be associated with an increased risk of skin cancer vation and leads to apoptosis of memory T cells. Methotrexate (PUVA), skin atrophy (steroids), hyperlipidaemia and hepat- is a dihydrofolate reductase inhibitor and has antiproliferative opathy (retinoids) among other side-effects.1,2 Even though and immunosuppressive effects on T cells. As psoriasis is

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 180 Correspondence currently considered to be a T-cell-triggered immune disorder, 12 Mease PJ, Gladman DD, Keystone EC. Alefacept in combination the combined application of two drugs that affect different with methotrexate for the treatment of psoriatic arthritis: results of molecular functions of T cells may act synergistically. The dif- a randomized, double-blind, placebo-controlled study. Arthritis Rheum 2006; 54:1638–45. ferential response of recalcitrant palmar psoriasis suggests the potentially heterogenous aetiology of this disorder. One Conflicts of interest: none declared. patient was highly satisfied with this combination therapy of alefacept and methotrexate because of substantial improvement of functional disability associated with palmar psoriasis. In the other patient, a substantial therapeutic response was not observed. Of note, alefacept treatment was well tolerated in both patients. Alefacept may thus be a therapeutic option in therapy-resistant palmoplantar psoriasis, particularly in patients Cutaneous horn can be a clinical manifestation who have not shown a satisfactory clinical response to the of underlying sebaceous carcinoma standard therapeutic regimens. Hence, our present observation in palmar psoriasis is in line with results of a recently DOI: 10.1111/j.1365-2133.2006.07572.x published phase II trial which indicate that alefacept in combination with methotrexate may be an effective and SIR, We experienced a case of periocular sebaceous carcinoma well-tolerated therapeutic option for patients with psoriatic (SC) manifesting as a cutaneous horn (CH). Histopathological arthritis.12 findings showed well-differentiated SC with hyperkeratosis and preserved lobule structures. Electron microscopic study Department of Dermatology, A. JACOBI has shown nonmembrane-bound intracytoplasmic lipid inclu- University Hospital Erlangen, G. SCHULER sions in the tumour. CH can be a clinical manifestation of Hartmannstrasse 14, 91052 Erlangen, Germany M. HERTL* underlying sebaceous carcinoma. *Department of Dermatology and SC is divided into two entities: periocular SC (PSC) and Allergology, Philipps University Marburg, extraocular SC according to the affected sites.1 The most com- Deutschhausstrasse 9, 35037 Marburg, Germany mon presentation of SC is a small, enlarging, firm, deep-seated E-mail: [email protected] nodule.2 SC manifesting as CH is very rare. To our knowledge, only two cases of SC with CH have been reported.3,4 A 73-year-old female presented with an asymptomatic nod- References ule on the left periocular area of 1-year duration. She was 1 Pettey AA, Balkrishnan R, Rapp SR et al. Patients with palmoplantar referred in January 2005. Physical examination revealed a sub- psoriasis have more physical disability and discomfort than patients cutaneous, deep-seated nodule with cutaneous horn with other forms of psoriasis: implications for clinical practice. J 7 · 7 mm in diameter on the left periocular area (Fig. 1). Am Acad Dermatol 2003; 49:271–5. Her medical history was significant for ulcerative colitis treat- 2 Myers W, Christiansen L, Gottlieb AB. Treatment of palmoplantar psoriasis with intramuscular alefacept. J Am Acad Dermatol 2005; ed with prednisolone and she had a thyroid tumour. The sub- 53:127–9. cutaneous tumour and cutaneous horn were excised with a 3 Ortonne JP, Prinz JC. Alefacept: a novel and selective biologic agent for 1-mm margin. the treatment of chronic plaque psoriasis. Eur J Dermatol 2004; 14:41–5. Histopathological findings showed exogenous hypercornifi- 4 Gottlieb AB. Alefacept for psoriasis and psoriatic arthritis. Ann Rheum cation and accumulated neutrophils in the horny layers. Dis 2005; 64:58–60. 5 Vitez L, Heese E, Wozel G. Long-term remission after 1 course of 12 weeks of alefacept therapy—a case report. Dermatology 2005; 211:165–6. 6 Schmitt J, Stoller E, Wozel G. Alefacept. Successful long-term therapy of a severe psoriasis. Hautarzt 2005; 56:360–2. 7 Thaci D, Pa¨tzold S, Kaufmann R, Boehncke WH. Treatment of psoriasis with alefacept in patients with hepatitis C infection: a report of two cases. Br J Dermatol 2005; 152:1048–50. 8 Sterry W, Barker J, Boehncke WH et al. Biological therapies in the systemic management of psoriasis. International Consensus Confer- ence. Br J Dermatol 2004; 151:3–17. 9 Mrowietz U. Psoriasis therapy with biologicals. Hautarzt 2003; 54:224–9. 10 Jacobi A, Manger B, Schuler G, Hertl M. Therapeutic application of the TNF-alpha inhibitors infliximab and etanercept in inflammatory skin disorders. J Dtsch Dermatol Ges 2003; 1:259–72. 11 Jacobi A, Mahler V, Schuler G, Hertl M. Treatment of inflammatory dermatoses by tumor necrosis factor antagonists. J Eur Acad Dermatol Venereol (in press). Fig 1. A subcutaneous deep-seated nodule with cutaneous horn 7 · 7 mm in diameter on the left periocular area.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 181

Fig 2. Immunohistochemical stainings were performed using epithelial membrane antigen, periodic acid-Schiff, S-100 and carcinoembryonic antigen. (a) Continuous with the epidermis, irregularly lobulated tumour nests were found. (b) Tumour cells are composed of atypical cells with great variation in size and shape with (a) (c) hyperchromatic nuclei and scant cytoplasm and mitotic cells. (c) The foamy cells were positive for epithelial membrane antigen. (d) Tumour cells revealed nonmembrane- bound intracytoplasmic lipid inclusions. The tumour cells were adhered and coalescent with immature desmosomes. Original magniﬁcation: (a) ·40; (b) ·400; (c) ·200; (b) (d) (d) ·10 000.

Continuous with the epidermis, irregularly lobulated tumour berant mass of keratin.6 CH may complicate seborrhoeic nests were found (Fig. 2a). The structure of the lobule was keratosis, actinic keratosis, viral wart, basal cell carcinoma, preserved. The tumour nests consisted of sebaceous, cell-like keratoacanthoma and squamous cell carcinoma.7 CH is gener- foamy cells, irregular, proliferated basaloid cells and duct-like ally the manifestation of underlying epidermal dysplasia or structure cells. Tumour cells are composed of atypical cells squamous cell carcinoma. Brauninger et al.3 reported SC of the with great variation in size and shape with hyperchromatic lid margin masquerading as CH, and emphasized that CH may nuclei and scant cytoplasm and mitotic cells (Fig. 2b). The be the sign of an underlying SC. Boniuk and Zimmerman4 lobules showed small, duct-like structures with internal kerati- also reported SC with CH. To our knowledge, only two cases nization. According to histopathological findings, our case was of SC with CH have been reported until now. However, it is diagnosed as well-differentiated sebaceous carcinoma. important to be aware that CH can be a clinical manifestation For immunohistochemistry, epithelial membrane antigen of underlying SC. (EMA), periodic acid-Schiff (PAS), S-100, carcinoembryonic In our case, CK expression was similar to that seen in nor- antigen (CEA) stainings were performed. The foamy cells were mal sebaceous glands, as previously reported.8 The structure positive for EMA (Fig. 2c), and were negative for PAS, S-100 of the lobule in the present case was well preserved, suggest- and CEA. The basaloid cells and duct-like structure cells were ing that it is well-differentiated SC with similar CK expression. negative for EMA, PAS, S-100 and CEA. Ansai et al. reported that CK expression in SC was similar to The immunohistochemical study using antikeratin anti- that in normal sebaceous glands, but in some cases there were bodies were performed. The antikeratin antibodies used were simple epithelial keratins.8 We believe that the difference of as follows: 34bB4 (CK1), LHP1 (CK10), LL002 (CK14), CK expression may be attributable to the stage of differenti- LHK15 (CK15), LL025 (CK16), E3 (CK17), 5D3 (CK18) and ation of SC. b170 (CK19).5 The immunohistochemical study used the Electron microscopic study showed prominent nonmem- labelled streptoavidin–biotin method, which was performed as brane-bound lipid vacuoles, tonofilaments and immature des- previously described.5 mosomes in tumour cells as previously described.9 CK1 and CK10 were not expressed in tumour nests, but In the periocular area, the epidermis is thin, and the seba- CK14 was detectable in tumour nests. CK17 was detectable ceous glands are abundantly distributed. Therefore, we specu- only in duct-like structures. The other CKs were not found in late that germinative sebocytes are influenced by ultraviolet tumour nests. exposure, resulting in sebaceous carcinoma, and giving rise to In electron microscopic study, tumour cells revealed non- CH. membrane-bound intracytoplasmic lipid inclusions (Fig. 2d). It is important to direct primary treatment for SC before These tumour cells were adhered and coalescent with imma- orbital invasion or metastasis. Therefore, it is necessary to ture desmosomes. make an accurate initial diagnosis of SC. As clinical manifestations of SC are diverse, it is difficult to make an accurate and prompt diagnosis clinically as SC. It is Department of Dermatology, Yamada Red Cross H. KITAGAWA often misdiagnosed as pyogenic granuloma chalazion or unilat- Hospital, Ise, Mie, Japan M. MIZUNO 1 eral blepharitis or meibomitis, and squamous cell carcinoma. *Department of Dermatology, Mie University Y. NAKAMURA The definite diagnosis of SC is made by histological findings. Graduate School of Medicine, 2-174, Edobashi, I. KUROKAWA* The clinical manifestation of SC usually shows a yellow Tsu, Mie 514-8507, Japan H. MIZUTANI* nodule, erosion and ulceration. Interestingly, our case mani- Correspondence: Ichiro Kurokawa. fested CH. CH is a clinical diagnosis based on the large protu- E-mail: [email protected]

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 182 Correspondence

References

1 Lazar AJF, McKee PH. Tumors and related lesions of the sebaceous glands. In: Pathology of the Skin (McKee PH, Calonje E, Granter SR, eds), Vol. 2. Philadelphia: Elsevier Mosby, 2005; 1579–84. 2 Nelson MR, Hamlet KR, Gillard M et al. Sebaceous carcinoma. JAm Acad Dermatol 1995; 33:1–15. 3 Brauninger GE, Hood CI, Worthen DM. Sebaceous carcinoma of lid margin masquerading as cutaneous horn. Arch Opthalmol 1973; 90:380–1. 4 Boniuk M, Zimmerman LE. Sebaceous carcinoma of the eyelid, eye- brow, caruncle, and orbit. Trans Am Acad Opathalmol Otolaryngol 1968; 72:619–42. 5 Kurokawa I, Nishijima S, Kusumoto K et al. Trichilemmoma: an immunohistochemical study of cytokeratins. Br J Dermatol 2003; 149:99–104. 6 Brenn T, McKee PH. Tumors of the surface epithelium. In: Pathology Fig. 1. Thickening of the skin of the hands. of the Skin (McKee PH, Calonje E, Granter SR, eds), Vol. 2. Philadel- phia: Elsevier Mosby, 2005; 1153–1240. 7 Petzelbauer P, Hammer I, Konrad K. Cornu cutaneoum. Hautarzt 1989; 40:556–8. 8 Ansai S, Katagata Y, Yoshikawa K et al. An immunohistochemical study of sebaceous carcinoma with anti-keratin monoclonal antibodies: comparison with other skin cancers. J Dermatol 1994; 21:553–9. 9 Wolfe JT III, Yeatts RP, Wick MR et al. Sebaceous carcinoma of the eyelid. Errors in clinical and pathologic diagnosis. Am J Surg Pathol 1984; 8:597–606.

Conﬂicts of interest: none declared.

Pseudosystemic sclerosis as a complication of Fig. 2. Thickening of the skin of the calves. recombinant human interleukin 2 (aldesleukin) therapy A 49-year-old woman was diagnosed as having renal-cell DOI: 10.1111/j.1365-2133.2006.07579.x carcinoma involving her right kidney in March 1999; no metastatic lesions were seen and the patient underwent rad- SIR, Recombinant human interleukin 2 (rIL-2) or aldesleukin ical nephrectomy. In October 2004, the patient exhibited a is an effective immunotherapeutic agent in patients with meta- relapse of renal-cell carcinoma with biopsy-proven lung static renal-cell carcinoma and melanoma.1 Adverse cutaneous metastases, and she received subcutaneous aldesleukin at a ) effects are increasingly recognized as complications of aldes- dose of 11 · 106 Um 2, three times weekly for 1 month. leukin therapy. Previous authors have found cutaneous effects Three weeks after the institution of subcutaneous aldesleukin in up to 72% of patients receiving aldesleukin, mainly includ- therapy, the patient was admitted for a 1-week history of ing macular erythema with pruritus, injection site reaction, rapidly progressive thickening of the skin involving both urticaria, aphthous ulcers or bullous eruptions.1–4 Other aldes- her upper and lower limbs. Physical examination revealed leukin-related skin reactions have been described more rarely, thickening of: (i) the skin of the hands, wrists and fore- e.g. lobular panniculitis at the aldesleukin injection site, arms; and (ii) the skin of the feet, calves and thighs (Figs 1, immediate facial angio-oedema and erythema, reactions 2). General examination was otherwise normal. Laboratory resembling toxic epidermal necrolysis and multifocal ﬁxed ﬁndings were as follows: erythrocyte sedimentation rate ) drug eruption to other agents.1–6 We recently observed a case 38 mm in 1 h, haemoglobin 9Æ5gdL 1, white cell count ) ) ) of particular interest as the patient with metastatic renal-cell 13Æ9 · 109 L 1, total protein 68 g L 1, albumin 37 g L 1. carcinoma developed a diffuse pseudosystemic sclerosis pattern Autoantibody screening was negative for: rheumatoid fac- related to aldesleukin therapy. tors, antinuclear antibodies, anticentromere, antitopoiso-

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 183 merase, anticardiolipin and antiphospholipid antibodies. in patients with scleroderma).10 Puett and Fuchs11 have, Nailfold capillaroscopy was normal. A skin biopsy from the indeed, mentioned a rapid exacerbation of systemic sclerosis right thigh was performed. Histological analysis of skin (progression of truncal skin thickening) in a patient receiving biopsy specimens showed: (i) an intact epidermis; (ii) peri- aldesleukin therapy. We therefore suggest that it is question- vascular inflammatory infiltrates (mainly composed of able to initiate aldesleukin in patients with scleroderma and lymphocytes and plasma cells) and interstitial oedema in- malignancy. Finally, because aldesleukin is widely used for volving the dermis; (iii) fibrosis involving both the dermis therapy of carcinoma and melanoma, our findings highlight and hypodermis; and (iv) direct immunofluorescence was the importance of recognition of pseudosclerodermatous negative. The diagnosis of diffuse pseudosystemic sclerosis lesions as a complication of aldesleukin therapy, leading to related to aldesleukin therapy was made, and the therapy appropriate diagnosis and thereby avoiding unnecessary with aldesleukin was discontinued. Steroid therapy was investigations. ) instituted (at an initial dose of 1 mg kg 1) concomitantly, which resulted in improvement of scleroderma manifesta- Departments of Internal Medicine, I. MARIE tions, with a reduction of skin thickening of both the *Dermatology, and Cytology and Pathology, P. JOLY* upper and lower limbs. Therapy for the renal-cell carcinoma Rouen University Hospital, P. COURVILLE was initiated with interferon-alpha and 5-fluorouracil. 76031 Rouen Cedex, France H. LEVESQSUE Pseudoscleroderma lesions have been described in patients E-mail: [email protected] receiving various systemic drugs, such as bleomycin, L-tryp- 7 tophan, carbidopa, gemcitabine or taxanes. However, we References report, to our knowledge, the first case of aldesleukin-related pseudoscleroderma. In this instance, the diagnosis of aldes- 1 Schmidinger M, Hejna M, Zielinski CC. Aldesleukin in advanced leukin-related pseudoscleroderma could reasonably be made, renal cell carcinoma. Expert Rev Anticancer Ther 2004; 4:957– as: (i) there was a temporal relation between aldesleukin 80. 2 O’Reilly F, Feldman E, Yang J et al. Recurring cutaneous erup- administration and the rapid onset of sclerodermatous cuta- tion in a patient with metastatic renal cell carcinoma being trea- neous manifestations; (ii) our patient’s clinical presentation ted with high-dose interleukin 2. J Am Acad Dermatol 2003; was similar to previous reports of pseudoscleroderma lesions 48:602–4. induced by other drugs, i.e. she exhibited no Raynaud’s 3 Hofmann M, Audring H, Sterry W, Trefzer U. Interleukin-2- phenomenon, autoantibodies were undetectable and nailfold associated bullous drug dermatosis. Dermatology 2005; 210:74–5. capillaroscopy was normal; and (iii) improvement of the 4 Wolkenstein P, Chosidow O, Wechsler J et al. Cutaneous side pseudosclerodermatous features took place after stopping effects associated with interleukin 2 administration for metastatic melanoma. J Am Acad Dermatol 1993; 28:66–70. aldesleukin treatment. This latter data indicates that pseudo- 5 Abraham D, McGrath KG. Hypersensitivity to aldesleukin (interleu- scleroderma related to a paraneoplastic syndrome could be kin-2 and proleukin) presenting as facial and angioedema and ery- excluded in our patient; such reports of pseudoscleroderma thema. Allergy Asthma Proc 2003; 24:291–4. associated with cancer have been postulated to be due to the 6 Baars JW, Coenen JL, Wagstaff J et al. Lobular panniculitis after release of growth factors (e.g. basic fibroblast growth factor) subcutaneous administration of interleukin-2 (IL-2), and its exacer- by malignant cells resulting in fibroblast proliferation.8 bation during intravenous therapy with IL-2. Br J Cancer 1992; Moreover, previous authors have underscored the lack of 661:698–9. 7 Bessis D, Guillot B, Legouffe E, Guilhou JJ. Gemcitabine-associated inflammatory infiltrates in the dermis of patients with can- 8 scleroderma-like changes of the lower limbs. J Am Acad Dermatol cer-associated pseudoscleroderma; in our patient, histological 2004; 51:S73–6. analysis showed perivascular inflammatory infiltrates in the 8 Querfeld C, Sollberg S, Huerkamp C et al. Pseudoscleroderma asso- dermis, which also supports the diagnosis of aldesleukin- ciated with lung cancer: correlation of collagen type I and connect- related pseudoscleroderma. The exact pathological mecha- ive tissue growth factor gene expression. Br J Dermatol 2000; nisms of pseudoscleroderma cutaneous lesions due to 142:1228–33. aldesleukin are unknown, although they may be related 9 Schlegel PJ, Samlowski WE, Ward JH. Autoimmune hemolytic ane- mia in a patient with chronic lymphocytic leukemia and renal cell to an immunological process activating fibroblasts rather than carcinoma after treatment with high-dose intravenous interleukin-2. to a direct toxic effect from cutaneous accumulation of the J Immunother 2000; 23:507–8. drug. Aldesleukin therapy is indeed an immunotherapeutic 10 Clements PJ, Peter JB, Agopian MS et al. Elevated serum levels of agent leading to:1,9 (i) proliferation of T-cell subsets soluble interleukin 2 receptor, interleukin 2 and neopterin in dif- (activated natural killer cells and cytotoxic cells), and of acti- fuse and limited scleroderma: effects of chlorambucil. J Rheumatol vated T cells resulting in increased serum levels of IL-2 and 1990; 17:908–10. soluble IL-2 receptor; and (ii) proliferation of B cells, auto- 11 Puett DW, Fuchs HA. Rapid exacerbation of scleroderma in a patient treated with interleukin 2 and lymphokine activated kil- antibodies and Ig production. In light of these data, we sug- ler cells for renal cell carcinoma. J Rheumatol 1994; 21: gest that pseudoscleroderma may not be an unexpected 752–3. adverse effect of aldesleukin therapy, as IL-2 is considered to be important in the genesis of scleroderma (as shown by Conflicts of interest: none declared. elevated serum levels of both IL-2 and soluble IL-2 receptor

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 184 Correspondence

Clinically, the entire parietal scalp showed moderate to severe erythema, crusty scaling and diffuse areas with scarring A case of sporadic Bazex–Dupre´–Christol syn- alopecia. Culture of swabs from pustular scalp lesions includ- drome presenting with scarring folliculitis of ing hair bulbs revealed Staphyloccocus aureus, whereas cultures on the scalp conventional Sabouraud medium were negative for fungi. His- tology (haematoxylin and eosin staining) of a scalp lesion is DOI: 10.1111/j.1365-2133.2006.07580.x shown in Fig. 2. No evidence for microbes was found in PAS staining. Direct immunofluorescence was unspecific. Mutations SIR, Since its first description in 1964, the Bazex–Dupre´–Chri- in the PTCH gene (chromosome 9q22) could be excluded by stol syndrome (BDCS) has been reported in fewer than 90 gene analysis studies. Complete work-up including routine la- 1–3 patients. The three main features of BDCS are early onset of boratory tests and skull and skeletal radiography was otherwise multiple basal cell carcinomas (BCCS), congenital hypotricho- unremarkable. sis and follicular atrophoderma. BDCS is an X-linked dominant The expression of the main features of BDCS may vary.1–3 disease (gene location: Xp24–q27) that is thought to be a dis- Follicular atrophoderma and hypotrichosis is reported in about 2 order of the hair follicle. However, sporadic occurrence of 85% of patients. However, these symptoms appear to affect 3 this condition has been described only once in the literature. male subjects more severely. Early-onset BCCs are a very com- We describe a 31-year-old nulliparous woman with a his- mon finding in BDCS. In infancy and childhood, multiple tory of multiple BCCs (n ¼ 10) on her face, neck and dećol- milia are present (up to 70% of patients), whereas in adults lete´. Since early childhood she had complained of multiple only a few milia are usually observed. Hypohidrosis is found milia on her face, recurrent soreness of the scalp and sparse in less than 30% of patients. Other reported features of BDCS hair growth affecting the scalp, eyebrows and eyelashes. She include pinched nose with hypoplastic nasal alae and promin- was otherwise healthy. Her family history was unremarkable. ent columella, keratosis pilaris, and associated hair shaft ab- On examination, a few milia were observed on the cheeks. normalities such as pili torti and trichorrhexis nodosa.1–3 As A slight facial dysmorphism was noted including hypertelo- also described in previous reports, our patient showed normal rism and low, broadened nasal bridge. There was hypotricho- hairs intermingled with abnormal curly hairs.2 The latter sis predominantly affecting the scalp, eyebrows and eyelashes showed similar features to those usually found in pseudomo- (Fig. 1). The hairs of the scalp showed an admixture of irre- nilethrix, pili annulati and the uncombable hair syndrome. gularly curly hairs intermingled with normal hairs. There was Light microscopic hair anomalies in BDCS are variable, fre- no increased hair fragility and no tufting of hairs. Light micro- quently showing pili bifurcati, trichorrhexis nodosa and flat- scopy of curly hairs revealed fusiform beading along the hair tened, irregularly curly hairs.2 shaft without evidence of cracking on the thinner parts. Cross- We consider the clinical features of this patient to be con- sectional shapes were variable and irregular showing flat-ellip- sistent with the diagnosis of BDCS. The differential diagnoses tic and angular to reniform shapes.

Fig 1. Patient with sporadic Bazex–Dupre´– Christol syndrome showing slight hypertelorism and low, broadened nasal bridge, hypotrichosis of the eyebrows and eyelashes (a), admixture of irregularly curly hairs intermingled with normal hairs, and moderate erythema, crusty scaling and diffuse areas with scarring alopecia (b).

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 185

(b)

Fig 2. Histopathology of a biopsy specimen taken from the parietal scalp showing a dense inflammatory infiltrate in the dermis mainly consisting of neutrophils and lymphocytes. In the deep dermis an atrophic follicle is visible (a). Higher magnification reveals the follicular and perifollicular inflammation with neutrophils; condensed collagen bundles are present (b). Haematoxylin and eosin, original magnification (a) · 100, (b) · 200.

such as X-linked dominant chondrodysplasia punctata and reported in BDCS.1–3 However, it may be a feature of related Gorlin–Goltz syndrome (GGS) could widely be excluded.4,5 hereditary conditions such as Rapp–Hodgkin syndrome, ectro- X-linked dominant chondrodysplasia punctata may be charac- dactyly–ectodermal dysplasia–clefting syndrome and congen- terized mainly by skeletal abnormalities, cutaneous anomalies ital hypotrichosis of Marie Unna.7–9 and cataracts.4 Apart from multiple BCCs, jaw cysts, skeletal Our patient had suffered from a diffusely scarring scalp anomalies, ectopic calcification and pitting of the hands and dermatitis since early childhood. Histopathologically, a fibro- feet are the stigmatizing major features of GGS.5 Hypertelo- sing folliculitis and perifolliculitis was seen. Direct immuno- rism belongs to the minor criteria of GGS, and to the best of fluorescence was negative. These features widely excluded our knowledge, has not been reported previously in BDCS.1–3 distinct scalp conditions including folliculitis decalvans, lupus GGS is caused by mutations in the PTCH gene on chromo- erythematodes and lichen planus.10 some 9q22.5 We performed mutation analysis of this gene in Glaessl et al.3 reported the first patient with sporadic BDCS our patient, which did not reveal a pathogenic mutation, ren- who also did not show features of follicular atrophoderma. dering the differential diagnosis of GGS rather unlikely. However, it remains unclear whether our case, along with the Indeed, a more difficult differential diagnostic problem is the patient reported by Glaessl et al.,3 represents a genotype that is Rombo syndrome, which is clinically characterized by mul- different from BDCS, despite the phenotype, which is more or tiple facial milia, vermiculate atrophoderma, hypotrichosis, less identical to the classical X-linked dominant BDCS. As dem- multiple BCCs, trichoepitheliomas and peripheral vasodilation onstrated in other ectodermal dysplasia syndromes,7–9 BDCS with cyanosis.6 Nevertheless, the Rombo syndrome is different seems to include a variable clinical expression with a wide from BDCS in that follicular atrophoderma is absent and the spectrum of phenotypes showing clinical overlap between abnormalities become visible only in late childhood. To date, related disorders such as Gorlin–Goltz syndrome and Rombo scalp dermatitis and/or scarring folliculitis have not been syndrome.2,5,6

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 186 Correspondence

Departments of Dermatology and T. GAMBICHLER degree of risk, especially when managing patients with com- *Human Genetics, S. HOFFJAN* plicated concurrent medical problems and lifestyles. When Ruhr-University Bochum, P. ALTMEYER inpatient facilities are available, repeated admission to induce Gudrunstr. 56, 44791 Bochum, Germany F.G. BECHARA remission has significant cost implications. We wish to high- E-mail: [email protected] light the challenges and dilemmas that can be faced when dealing with some challenging psoriasis patients. References A 54-year-old man has had chronic plaque psoriasis since the age of 30. He also has psoriatic arthritis and suffers from hyper- 1 Vabres P, Lacombe D, Rabinowitz LG et al. The gene for Bazex–Du- tension, asthma, type 2 diabetes mellitus and atrial fibrillation. pre´–Christol syndrome maps to chromosome Xq. J Invest Dermatol He currently has severe psoriasis1 with at least 60% body surface 1995; 105:87–91. area affected, a psoriasis area and severity index (PASI) score of 2 Goeteyn M, Geerts ML, Kint A, De Weert J. The Bazex–Dupre´–Chr- 26Æ6 and a Dermatology Life Quality Index (DLQI) score of 30. istol syndrome. Arch Dermatol 1994; 130:337–42. 3 Glaessl A, Hohenleutner U, Landthaler M, Vogt T. Sporadic Bazex– He has tried various topical therapies but they were never Dupre´–Christol-like syndrome: early onset basal cell carcinoma, enough to provide adequate symptom relief and he always hypohidrosis, hypotrichosis, and prominent milia. Dermatol Surg denied not using prescribed therapy. He has exceeded his max- 2000; 26:152–4. imum lifetime cumulative number of phototherapy courses, i.e. 4 Lindenthal B, Repgen R, Emons D et al. Serum lipid analysis 150 treatments. Methotrexate was used previously but this had confirms the diagnosis of X-linked dominant chondrodysplasia to be stopped due to his excessive alcohol intake. Acitretin also punctata – Conradi–Hunermann–Happle syndrome. Klin Padiatr had to be stopped as he could not tolerate the side-effects. In 2004; 216:67–9. 5 Kimonis VE, Goldstein AM, Pastakia B et al. Clinical manifestations 1998, micronodular liver cirrhosis was demonstrated by liver in 105 persons with nevoid basal cell carcinoma syndrome. Am J biopsy. Despite this, he continues to drink alcohol excessively Med Genet 1997; 69:299–308. and has refused offers of professional support or help from 6 Ashinoff R, Jacobson M, Belsito DV. Rombo syndrome: a second Alcoholics Anonymous. He is convinced that alcohol helps with case report and review. J Am Acad Dermatol 1993; 28:1011–14. reducing the itching that he gets from psoriasis. He was seen in 7 Fosko SW, Stenn KS, Bolognia JL. Ectodermal dysplasias associated with clefting: significance of scalp dermatitis. J Am Acad Dermatol 1992; 27:249–56. 8 Sahin MT, Tu¨rel-Ermerctan A, Chan I et al. Ectodermal dysplasia showing clinical overlap between AEC, Rapp–Hodgkin and CHAND syndromes. Clin Exp Dermatol 2004; 29:486–8. 9 Truëb RM, Tsambaos D, Spycher MA et al. Scarring folliculitis in the ectrodactyly–ectodermal dysplasia–clefting syndrome. Histo- logic, scanning electron-microscopic and biophysical studies of hair. Dermatology 1997; 194:191–4. 10 Powell JJ, Dawber RPR, Gatter K. Folliculitis decalvans including tufted folliculitis: clinical, histological and therapeutic findings. Br J Dermatol 1999; 140:328–33.

Conﬂicts of interest: none declared.

Moral and cost dilemma of a psoriasis patient

DOI: 10.1111/j.1365-2133.2006.07582.x

SIR, The aim in psoriasis management is to provide patients with symptomatic relief and long-term control of the disease with as little risk as possible. The choice of the most appropriate treatment, either topical, phototherapy, systemic agents or more recently biologics, obviously depends on the severity of the disease in each individual patient and also whether there are contraindications or limitations. In patients with severe psoriasis, it is difﬁcult to achieve disease control without some Fig 1. On the day of admission.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 187

needed.3 We found it difﬁcult to refuse our patient admission as he is mentally competent and at the time of decis- ion to admit, his psoriasis was more severe than most other patients admitted. Hospital admissions have provided symptomatic relief and safe long-term control but as the average cost of his hospital admissions each year is at least £26 800 (at £298Æ78 per day for University Hospital Wales), is this an appropriate use of funds? This is the dilemma faced. Inpatient management of skin diseases is effective in inducing remission4 and improving patients’ quality of life.5 The many speciﬁc needs of a patient with very severe psoriasis can be met only by specialist nursing care as an inpatient and are almost impossible to be dealt with in an outpatient setting such as day care treatment or at home.6 Our patient’s history is an example of inpatient therapy working and illustrates why hospital beds continue to be needed in dermatology. Possibly inpatient therapy should be this patient’s primary or only form of management as systemic agents are not suitable. We need to manage most patients on an outpatient basis to reduce costs and, for many patients, outpatient care is essential as this also allows maintenance of their independence and normality of life. Even though day care treatment may initially cost less compared with admission, the slower rate of clearance of a patient’s psoriasis means that the overall cost may in the end be similar.7 For our patient, is it not wrong to deny him his preferred option, at least until safer therapy is available, despite the cost? We will probably continue to debate about Fig 2. At discharge. how to manage him.

2005 by a clinical psychologist who felt that he would not The Welsh Institute of Dermatology, S. ABDUL G HAFFAR benefit from any psychological intervention. University Hospital Wales, Heath Park, A.Y. FINLAY In view of his liver cirrhosis, drinking habits and medical Cardiff CF14 4XN, U.K. history, we felt that it was difficult to offer him any more sys- E-mail: [email protected] temic agents. The possibility of using the biologics has been discussed with him. Even if he met the inclusion/exclusion References criteria, he was not keen on any further systemic agents any- way because of their side-effects. Hospital admissions for in- 1 Finlay AY. Current severe psoriasis and the rule of tens. Br J Dermatol tensive topical treatment have therefore had to be offered. 2005; 152:861–7. Unfortunately, over the years, he has become reliant on fre- 2 Khilji FA, Gonzalez M, Finlay AY. Clinical meaning of change in Dermatology Life Quality Index scores. Br J Dermatol 2002; 147 quent admissions. (Suppl. 62): 50. He has refused alternative strategies to control his psoriasis, 3 Sohn S, Schoeffski O, Prinz J et al. Cost of moderate to severe plaque such as day care treatment, as he now has a fixed belief that psoriasis in Germany: a multicenter cost-of-illness study. Dermatology the only treatment option that works for him is hospital 2006; 212:137–44. admission. And he is not wrong. Before his most recent 4 Munro CS, Lowe JG, McLoone P et al. The value of in-patient derma- admission, he had a PASI score of 26Æ6 and a DLQI score of tology: a survey of in-patients in Scotland and Northern England. 30. At discharge, his PASI score improved to 10Æ6 and his Br J Dermatol 1999; 140:474–9. 5 Ayyalaraju RS, Finlay AY, Dykes PJ et al. Hospitalization for severe DLQI score, although still extremely high, had reduced by five 2 skin disease improves quality of life in the United Kingdom and the points, considered clinically significant (Figs 1,2). Since United States: a comparative study. J Am Acad Dermatol 2003; 49:249– 1995, he has had on average three admissions per year, five 54. in 2005 alone. The total number of days he has had in hospi- 6 Coleman DA, Bennett ML, Tilus D, Young D. A worst-case guide for tal altogether between January 1995 and April 2006 is any case of psoriasis. RN 1988; 51:39–43. 988 days, i.e. 7 days per month. The average length of stay is 7 Cockayne SE, Cork MJ, Gawkrodger DJ. Treatment of psoriasis: day 29Æ9 days, range 20–49 days. care vs. inpatient therapy. Br J Dermatol 1999; 140:375–6. Moderate to severe psoriasis has huge cost implications, Conflicts of interest: none declared. especially where repeated admissions for disease control are

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 188 Correspondence

Reiter syndrome triggered by adalimumab (HumiraÒ) and leﬂunomide (AravaÒ)ina patient with ankylosing spondylarthropathy and Crohn disease

DOI: 10.1111/j.1365-2133.2006.07586.x

SIR, The new class of biological agents targeting tumour necrosis factor (TNF)-a has been increasingly associated with the triggering and/or the aggravation of psoriasiform or pustular eruptions.1–4 We have recently evaluated a 26-year-old HLAB27-positive patient who has been treated with adalimumab (Humira; Abbott AG, Baar, Switzerland) and leflunomide (Arava; Sanofi-Aventis AG, Meyrin, Switzerland) to control Crohn disease and ankylosing spondylarthropathy with gastrointestinal and articular manifestations refractory to conventional therapies including methotrexate, infliximab and nonsteroidal anti-inflammatory drugs. Four days after the third Fig 2. Acute asymptomatic circinate balanitis. adalimumab injection (40 mg) the patient developed a diffuse rash consisting of erythematous, papular and squamous lesions involving the groin, axillar folds, periumbilical region (Fig. 1) and the trunk. He also had genital lesions with sharply delim- ited erosions of the glans that were consistent with a circinate balanitis (Fig. 2) and two erythematosquamous papules of the feet consistent with plantar keratoderma (Fig. 3). No oral involvement was observed. Light microscopy studies of a biopsy specimen obtained from the trunk disclosed parakeratosis, acanthosis, elongation of the epidermal ridges and a diffuse inflammatory infiltrate of the superficial dermis that was consistent with a psoriasiform pattern (Fig. 4). Of note, the patient had no new systemic complaints, including diges- tive, urinary or ocular symptoms. An aerobe urine culture detected an infectious urethritis caused by Ureaplasma urealyticum ) (104 mL 1). The skin lesions regressed completely after doxy- cycline treatment (200 mg daily for 14 days) and class 2 topical steroid therapy (hydrocortisone butyrate 0.1% once daily).

Fig 3. Erythematosquamous papules of the feet consistent with plantar keratoderma.

The clinical and histopathological ﬁndings in our patient raised the possibility of a psoriasiform eruption induced by TNF-a antagonism as previously described in single observations.1–4 Nevertheless, the peculiar eruption of the genital area consistent with a circinate balanitis with pustular plantar involvement, together with the detection of U. urealyticum in the urine were very suggestive for acute cutaneous manifestations Fig 1. Psoriasiform rash of the periumbilical region. in the context of Reiter syndrome. This idea was further suppor-

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 189

As paradoxical psoriasiform eruptions have been observed under TNF-a antagonists,1–4 this report illustrates the need to determine in the future if these cutaneous manifestations correspond to a reactive process that is consecutive to a latent infection and that may have been triggered by concomitant immunosuppression. Consequently, further investigations to exclude infection should be recommended in these patients.

Department of Dermatology, A.M. THIELEN University Hospital Geneva, Geneva, C. BARDE Switzerland V. JANER E-mail: [email protected] L. BORRADORI J.H. SAURAT

References Fig 4. Light microscope findings: psoriasiform pattern with parakeratosis, regression of the granular layer, elongation of the 1 Thielen AM, Kuenzli S, Saurat JH. Cutaneous adverse events of bio- epidermal ridges and dermal inflammatory infiltrate. Haematoxylin logical therapy for psoriasis: review of the literature. Dermatology and eosin, original magnification · 40. 2005; 211:209–17. 2 Verea MM, Del Pozo J, Yebra-Pimentel MT et al. Psoriasiform eruption induced by infliximab. Ann Pharmacother 2004; 38:54–7. ted by the response to the antibacterial therapy and the absence 3 Thurber M, Feasel A, Stroehlein J, Hymes SR. Pustular psoriasis of relapses after subsequent reintroduction of adalimumab. induced by infliximab. J Drugs Dermatol 2004; 3:439–40. Reiter syndrome is a rare systemic reactive process to uro- 4 Kary S, Worm M, Audring H et al. New onset or exacerbation of genital (including Chlamydia trachomatis and U. urealyticum)or psoriatic skin lesions in patients with definite rheumatoid arthritis intestinal (such as Yersinia and Shigella) bacterial agents, mostly receiving tumor necrosis factor alpha antagonists. Ann Rheum Dis in HLAB27-positive patients. Clinical manifestations include 2006; 65:405–7. 5 Thami GP, Garg G. Leflunomide in psoriasis and psoriatic arthritis: a ocular involvement (conjunctivitis, uveitis), articular manifes- preliminary study. Arch Dermatol 2004; 140:1288–9. tations (reactive arthritis consistent with an asymmetrical 6 Tlacuilo-Parra JA, Guevara-Gutierrez E, Rodriguez-Castellanos MA mono- or polyarticular involvement, enthesopathies, spondyli- et al. Leflunomide in the treatment of psoriasis: results of a phase II tis and sacroileitis) and urogenital manifestations (urethritis, open trial. Br J Dermatol 2004; 150:970–6. salpingitis) consecutive to a reactive inflammatory process. 7 Kaltwasser JP, Nash P, Gladman D et al. Treatment of Psoriatic Arth- Cutaneous manifestations consist of a psoriasiform rash, ritis Study Group. Efficacy and safety of leflunomide in the treatment endobuccal erosions, palmoplantar keratoderma and circinate of psoriatic arthritis and psoriasis: a multinational, double-blind, randomized, placebo-controlled clinical trial. Arthritis Rheum 2004; balanitis or vulvitis. There is no clear consensus with well- 50:1939–50. recognized diagnostic criteria. Atypical clinical manifestations, 8 Patel T, Gordon KB. Adalimumab: efficacy and safety in psoriasis as in our case, are usually the rule. and rheumatoid arthritis. Dermatol Ther 2004; 17:427–31. Although we cannot formally exclude that our patient developed a psoriasiform eruption directly related to an anti- Conflicts of interest: none declared. TNF-a therapy, our case may represent the first observation of acute cutaneous manifestations related to Reiter syndrome that developed under adalimumab and leflunomide therapy. As the pathogenic mechanism of the cutaneous manifestations seems to be similar to the pathogenesis of psoriasis, the emergence of this syndrome in a patient treated by two im- munomodulating agents is paradoxical considering that these molecules are effective for the treatment of psoriasis.5–8 We Efalizumab-induced aseptic meningitis suppose that concomitant immunosuppression and immuno- modulation by adalimumab and leflunomide may have sup- DOI: 10.1111/j.1365-2133.2006.07594.x ported the triggering role of U. urealyticum in this patient who has a genetic predisposition to develop Reiter syndrome. The SIR, Efalizumab, a recombinant humanized monoclonal anti- lack of recurrence after the initiation of antibiotic treatment body directed against the CD11a molecule on the T-cell sur- and topical steroid therapy and after the reintroduction of face, is considered as a safe and effective treatment for adalimumab and leflunomide supports this hypothesis. This psoriasis.1 The data from clinical trials indicate that the most evolution does not support the hypothesis of a psoriasiform common side-effect is an influenza-like syndrome that may eruption induced by TNF-a antagonism considering that the include headaches, chills, fever, nausea and myalgia.1 These lesions would probably persist or recur during the treatment. acute adverse events occur within 48 h following the first two

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 190 Correspondence injections, are dose dependent and are usually well tolerated To our knowledge, this is the first report of drug- ) at the dose of 0Æ7mgkg 1.1,2 Paracetamol or ibuprofen are induced aseptic meningitis (DIAM) associated with ef- suggested in case of such events or as premedication.1 Head- alizumab since its approval for the treatment of moderate- aches are the most common of these side-effects, affecting to-severe psoriasis. The arguments supporting a correlation 32% of patients in clinical trials.1,2 Serious adverse effects are between meningitis and efalizumab injection include: (i) rare and include thrombocythaemia, haemolytic anaemia and rapid development of symptoms within 48 h after the onset infections.2 We report the first detailed case of aseptic menin- of treatment; (ii) negative findings from extensive investiga- gitis induced by efalizumab. tions for infectious agents; (iii) absence of other causes of A 32-year-old woman was hospitalized for suspected acute meningitis; and (iv) reversibility after discontinuation of the meningitis. Her medical history was notable for a 14-year his- drug. In addition, two cases of efalizumab-induced menin- tory of severe psoriasis treated without benefit with photother- gitis were mentioned in a recent review of events reported apy, ciclosporin and, more recently, etanercept. Other medical during clinical trials.2 data included Graves’ disease treated with carbimazole for DIAM is an uncommon adverse reaction mainly caused by 3 years. One month after etanercept withdrawal, persistent nonsteroidal anti-inflammatory drugs, antibiotics, intravenous and diffuse psoriasis prompted the initiation of efalizumab. immunoglobulins, monoclonal antibodies (OKT3 antibodies, Because of an unexpected pregnancy, she medically aborted, infliximab3), intrathecal agents and vaccines.4 Of note, it taking mifepristone on 1 March followed by misoprostol appears that patients with autoimmune diseases are more sus- 2 days later. An oral contraceptive treatment (levonorgestrel) ceptible to develop DIAM.4 Our patient presented autoimmune was initiated on 8 March. The first subcutaneous injection of hyperthyroidism, although this association might have been efalizumab was administered the next day at a dose of fortuitous. ) 0Æ7mgkg 1. Forty-eight hours later she experienced severe The pathogenic mechanisms remain unknown, although an headaches of increasing intensity, and this was complicated immunological type III or IV hypersensitivity is suspected 24 h later by photophobia, phonophobia and neck stiffness. when the drug has not been directly introduced into the CSF.4 Neurological examination revealed nuchal rigidity without Efalizumab is a monoclonal antibody against the CD11a mol- Kernig’s or Brudzinksi’s signs. Focal neurological signs, fever, ecule. CD11a and CD18 are subunits of leucocyte function- chills, myalgia and arthralgia were absent. Except for wide- associated antigen-1 (LFA-1), an important T-cell surface spread psoriasis lesions, no other skin eruption or reaction at molecule for T-cell activation and migration. In the skin, ef- the site of efalizumab injection was observed. The cerebrospi- alizumab blocks the binding of LFA-1 to intercellular adhesion ) nal fluid (CSF) was clear, disclosing 100 cells mm 3 with molecule (ICAM)-1. In the central nervous system, ICAM-5 92% neutrophils; glucose, protein and interferon levels were (telencephalin), expressed by neurones of the mammalian normal; and a Gram stain was negative. Laboratory blood tests brain, is of major importance for leucocyte binding to neu- showed a slightly elevated white blood cell count rones. ICAM-5 binds specifically to the leucocyte integrin ) (10Æ7 · 109 L 1), with normal rates of polynuclear cells and CD11a/CD18 and antibodies against CD11a/CD18 can disrupt ) lymphocytes, and raised C-reactive protein (11Æ7mgL 1). this interaction.5 Thus, the aseptic meningitis could be a con- Erythrocyte sedimentation rate, serum protein electrophoresis, sequence of the blocking of LFA-1 and ICAM-5 interaction by kidney function, liver enzymes and antinuclear antibodies efalizumab. Another mechanism might be that efalizumab were either normal or negative. After admission, efalizumab binds to a cross-reacting neural antigen, thereby inciting local and levonorgestrel were withdrawn and empirical administra- inflammation, as suspected by some authors for OKT3 anti- tion of intravenous amoxicillin and ceftriaxone was started. bodies.4 The following CSF microbiological studies were negative: Aseptic meningitis should be added to the list of the seri- Ziehl–Neelsen stain; aerobic and anaerobic cultures; polymer- ous adverse side-effects of efalizumab. The physician should ase chain reaction tests for cytomegalovirus, herpesvirus, vari- be alerted by unusually severe headaches. Efalizumab-induced cella-zoster virus, Epstein–Barr virus, enterovirus and JC virus; meningitis may not be rare but simply under-recognized, Lyme borreliosis, brucellosis and listeriosis serologies; and given that headaches commonly occur during this cryptococcal antigen. Blood cultures, serum cryptococcal anti- treatment.1 gen, and Lyme borreliosis and human immunodeficiency virus serologies were also negative. Neuroimaging ruled out cere- Universite´ Montpellier I, Service de Dermatologie, N. KLUGER bral venous thrombosis and intraparenchymal lesions. Clinical Hoˆpital Saint-Eloi, CHU Montpellier, C. GIRARD improvement was noted 48 h later and the CSF profile at 72 h 80 avenue Augustin Fliche, 34295 Montpellier cedex 5, V. GONZALEZ* )3 showed 15 cells mm (80% lymphocytes) and normal glu- France B. GUILLOT cose and protein. The patient was discharged home 9 days *Universite´ Montpellier I, Service de Neurologie, D. BESSIS after admission, with no antibiotics. A diagnosis of ef- Hoˆpital Gui-de-Chauliac, CHU Montpellier, alizumab-induced aseptic meningitis was established. Two 80 avenue Augustin Fliche, 34295 Montpellier cedex 5, weeks later, the patient complained of intermittent minor France headaches. E-mail: [email protected]

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 191

References moderate; 3, severe), scored in 0Æ5-point increments. Telan- giectasia was also graded by the same scoring system. Addi- 1 Leonardi CL. Efalizumab in the treatment of psoriasis. Dermatol Ther tionally, side-effects, if any, were documented. After each 2004; 17:393–400. 2 Scheinfeld N. Efalizumab: a review of events reported during clinical assessment, one pass of diamond-wand microdermabrasion trials and side effects. Expert Opin Drug Saf 2006; 5:197–209. with a particle size of 120 lm was performed under a pres- 3 Marotte H, Charrin JE, Miossec P. Infliximab-induced aseptic menin- sure of 127 mmHg during the active treatment phase. All stat- gitis. Lancet 2001; 358:1784. istical analyses were performed using SAS software version 8.0 4 Moris G, Garcia-Monco JC. The challenge of drug-induced aseptic for Windows (SAS Institute, Cary, NC, U.S.A.). The differences meningitis. Arch Intern Med 1999; 159:1185–94. in the therapeutic efficiency and severity of telangiectasia 5 Tian L, Kilgannon P, Yoshihara Y et al. Binding of T lymphocytes to between the tacrolimus- and the clobetasol-treated side at the hippocampal neurons through ICAM-5 (telencephalin) and characterization of its interaction with the leukocyte integrin CD11a/ baseline and across five time points were compared using CD18. Eur J Immunol 2000; 30:810–18. Wilcoxon’s rank sum test with the Bonferroni’s correction. Of 20 enrolled patients, 11 women and seven men (13 Conflicts of interest: none declared. with malar rash of SLE, four with DLE and one with SCLE; mean ± SD age 33Æ0±9Æ8 years) completed the study. Both tacrolimus and clobetasol were effective in treating CLE, with reduction in erythema, desquamation and induration. There was no significant difference between tacrolimus and clobetasol (Table 1). With a (type I error) set at 0Æ05, presumed difference of each severity score at 1, SD of 0Æ9 and a sample Tacrolimus vs. clobetasol propionate in the size of 18, the power comparing two group means was calcu- treatment of facial cutaneous lupus erythema- lated as 0Æ993. All lesions worsened at week 8, although they tosus: a randomized, double-blind, bilateral were still better than at baseline. However, compared with comparison study tacrolimus, 11 patients (61%) developed telangiectasia on the clobetasol side as early as week 3 (P <0Æ05). DOI: 10.1111/j.1365-2133.2006.07595.x A few uncontrolled open-label case reports have shown 0Æ1% tacrolimus to be effective in treating various forms of CLE.3–7 SIR, Cutaneous lupus erythematosus (CLE), an autoimmune in- However, patients with hyperkeratotic DLE usually did not flammatory disorder, includes a variety of diseases such as sys- respond well to the treatment.3,5,6 A possible explanation is that temic lupus erythematousus (SLE) with malar rash, subacute cutaneous lupus erythematosus (SCLE) and discoid lupus erythematousus (DLE). Skin-infiltrating T lymphocytes are consid- Table 1. Scoring characteristics of clobetasol- and tacrolimus-treated ered to play a major pathological role in CLE. Potent topical lesions corticosteroids remain the mainstay of treatment for CLE.1 Their long-term use influences collagen synthesis and induces Duration d skin atrophy and telangiectasia. Tacrolimus, a calcineurin an- Parameter (weeks) Clobetasol Tacrolimus P-value tagonist, inhibits T-cell activation and suppresses proinflam- Erythema Baseline 2Æ36 ± 0Æ48 2Æ22 ± 0Æ60 0Æ281 matory cytokine production.2 It has been shown to be 21Æ06 ± 0Æ45a 1Æ06 ± 0Æ78b 1Æ0 a a effective in treating CLE in a few case reports.3–7 40Æ78 ± 0Æ49 0Æ83 ± 0Æ73 0Æ941 c c To compare the efficacy and safety of 0Æ1% tacrolimus oint- 81Æ47 ± 0Æ87 1Æ17 ± 0Æ69 0Æ133 Desquamation Baseline 1Æ61 ± 0Æ85 1Æ42 ± 0Æ88 0Æ289 ment and 0Æ05% clobetasol propionate ointment in the treat- 20Æ50 ± 0Æ34a 0Æ44 ± 0Æ73b 0Æ398 ment of facial CLE, we conducted a randomized, double-blind 40Æ25 ± 0Æ35a 0Æ39 ± 0Æ74a 0Æ656 study, which was approved by the local medical ethics com- 80Æ69 ± 0Æ69c 0Æ50 ± 0Æ57a 0Æ336 mittee. After written informed consents were obtained, Induration Baseline 2Æ06 ± 0Æ75 1Æ81 ± 0Æ91 0Æ125 patients with facial CLE were instructed to apply twice daily 20Æ54 ± 0Æ59a 0Æ50 ± 0Æ59a 1Æ0 a a 0Æ1% tacrolimus ointment to the affected areas on one side of 40Æ28 ± 0Æ43 0Æ17 ± 0Æ38 0Æ438 a b the face and 0Æ05% clobetasol propionate ointment to the 80Æ92 ± 0Æ84 0Æ67 ± 0Æ62 0Æ188 Telangiectasia Baseline 0Æ00 ± 0 0Æ00 ± 0Æ00 – other side of the face, randomly assigned for each patient. 10Æ06 ± 0Æ24 0Æ00 ± 0Æ00 1Æ0 Before initiation of therapy, any topical treatment was discon- 20Æ25 ± 0Æ52 0Æ00 ± 0Æ00 0Æ063 tinued for at least 8 weeks. For patients with SLE, only those 30Æ47 ± 0Æ50c 0Æ00 ± 0Æ00 0Æ004 who had a stable course and whose facial lesions did not 40Æ58 ± 0Æ49b 0Æ00 ± 0Æ00 0Æ001 respond to systemic treatment for at least 6 months were 80Æ64 ± 0Æ68c 0Æ00 ± 0Æ00 0Æ002 recruited. The severity of paired selected lesions was assessed aP <0Æ001, bP <0Æ01, cP <0Æ05, compared with baseline score; at each visit (weeks 0, 1, 2, 3, 4 and post-treatment week 4), dcomparison between clobetasol- and tacrolimus-treated lesions; based on the clinical signs of erythema, desquamation and in- values are given as mean ± SD. duration using a 7-point rating scale (0, none; 1, mild; 2,

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 192 Correspondence the tacrolimus molecule is relatively large (822Æ05 Da) and 10 Fang JY, Lee WR, Shen SC et al. Enhancement of topical 5-amino- might have difficulty in penetrating hyperkeratotic skin. Indeed, laevulinic acid delivery by erbium:YAG laser and microdermabra- descaling overnight with 2% salicylic acid increases the topical sion: a comparison with iontophoresis and electroporation. Br J Dermatol 2004; 151:132–40. antipsoriatic effect of tacrolimus.8 Microdermabrasion is known to increase skin permeation of topical medication by altering the Conflicts of interest: none declared. epidermal barrier.9,10 In this study, we therefore added microdermabrasion once weekly to accelerate the delivery of tacrolimus to CLE lesions. Despite microdermabrasion, total clearance of the lesions was rarely achieved. Walker et al.7 described the beneficial effect of 0Æ3% tacrolimus in 0Æ05% clobetasol propionate applied twice daily for recalcitrant DLE. In the present study, we found that tacrolimus was as efficient as clobetasol ‘Mechanic’s hands’: a misleading cutaneous propionate in treating CLE. However, an obvious advantage of sign of the antisynthetase syndrome tacrolimus over clobetasol was its safety in terms of skin telangiectasia, which was commonly observed in the clobetasol-trea- DOI: 10.1111/j.1365-2133.2006.07593.x ted side. In conclusion, our study suggests that tacrolimus is a safer alternative in treating patients with facial CLE. SIR, Cutaneous lesions may be present in idiopathic inflammatory myopathies and are essential for the diagnosis, as in der- ZUNG 1,2 *Department of Dermatology, T-Y. T * matomyositis. ‘Mechanic’s hands’ (MH) is a misleading Kaohsiung Veterans General Hospital, Y-S. LIU* sign consisting of lesions looking like occupational dermatosis HANG Kaohsiung, Taiwan H-W. C of the hands in some patients with idiopathic inflammatory School of Medicine, myopathy.1,3–7 This sign appears to be correlated with the National Yang-Ming University, presence of the circulating myositis-specific autoantibodies Taipei, Taiwan that defines the antisynthetase syndrome, particularly anti-Jo-1 Department of Biological Sciences, antibodies that are often associated with interstitial lung dis- National Sun-Yat-Sen University, Kaohsiung, Taiwan ease.1 We undertook a retrospective study of cutaneous lesions E-mail: [email protected] in patients with antisynthetase syndrome to evaluate their incidence and evolution. References Data from all patients seen in our institution between 1998 and 2001 were retrospectively analysed. Patients were 1 Goodfields MD, Johns SK, Veale DJ. The ‘connective tissue disease’. selected when the diagnosis of antisynthetase syndrome was In: Rook’s Textbook of Dermatology (Burns T, Breathnach SM, Cox NH, Griffiths CEM, eds), 7th edn, Vol. 3. Oxford: Blackwell Publishing, established on the presence of interstitial lung disease 2004; 56.21–56.22. and/or inflammatory muscle disease and/or polyarthritis 2 Nghiem P, Pearson G, Langley RG. Tacrolimus and pimecrolimus: along with the presence of anti-Jo-1 antibodies. Clinical and from clever prokaryotes to inhibiting calcineurin and treating atop- biological parameters were analysed with special attention to ic dermatitis. J Am Acad Dermatol 2002; 46:228–41. cutaneous signs. The presence of other autoantibodies was 3 Yoshimasu T, Ohtani T, Sakamoto T et al. Topical FK506 (tacroli- also noted. mus) therapy for facial erythematous lesions of cutaneous Seven patients were included in the study (Table 1). All lupus erythematosus and dermatomyositis. Eur J Dermatol 2002; 12:50–2. were women, aged 37–79 years (mean 58). Initial signs con- 4 Kanekura T, Yoshii N, Terasaki K et al. Efficacy of topical tacrolimus sisted of dyspnoea (4/7), cough (3/7), arthralgia (2/7), MH for treating the malar rash of systemic lupus erythematosus. Br J (2/7), fatigue, fever and Raynaud’s phenomenon (1/7 each). Dermatol 2003; 148:353–6. Mean delay between the first sign and diagnosis was 4 months 5 Lampropoulos CE, Sangle S, Harrison P et al. Topical tacrolimus (range 2–12). Physical examination disclosed the constant therapy of resistant cutaneous lesions in lupus erythematosus: a presence of inspiratory crackles, MH in two of seven cases possible alternative. Rheumatology (Oxford) 2004; 43:1383–5. presenting as asymptomatic well-demarcated hyperkeratosis 6 Heffernan MP, Nelson MM, Smith DI et al. 0Æ1% Tacrolimus ointment in the treatment of discoid lupus erythematosus. Arch Dermatol and erythema on the lateral and palmar aspects of the hands 2005; 141:1170–1. and fingers (Fig. 1a, b), and ‘V sign’ rash and erythematous 7 Walker SL, Kirby B, Chalmers RJ. The effect of topical tacrolimus papules on the knees and elbows in one case. All patients had on severe recalcitrant chronic discoid lupus erythematosus. Br J Der- detectable anti-Jo-1 antibodies, but no other autoantibodies matol 2002; 147:405–6. were detected. Median follow-up was 26 months (range 1– ) 8 Remitz A, Reitamo S, Erkko P et al. Tacrolimus ointment im- 52). All patients were treated with prednisone 1 or 2 mg kg 1 proves psoriasis in a microplaque assay. Br J Dermatol 1999; 141: daily except an old woman treated with low-dose prednisone 103–7. 9 Rajan P, Grimes PE. Skin barrier changes induced by aluminum and azathioprine. In two other cases, cyclophosphamide was oxide and sodium chloride microdermabrasion. Dermatol Surg 2002; given with steroids because of the severity of the lung 28:390–3. involvement.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 193

Table 1 Characteristics of patients at diagnosis

Patient 1 Patient 2 Patient 3 Patient 4 Patient 5 Patient 6 Patient 7 Sex/age (years) F/44 F/37 F/73 F/49 F/56 F/79 F/67 Fatigue No No No Yes No Yes Yes Temperature 38.3 C38C Normal 38C Normal Normal Normal Weight loss No 5 kg No No No 12 kg No Dry cough/dyspnoea Yes/NYHA III Yes/NYHA II No/NYHA II Yes/NYHA IV Yes/NYHA II No/NYHA IV Yes/NYHA IV (NYHA) Myalgia/weakness No/no Yes/no No/no No/no No/no No/no No/no Arthralgia/arthritis No/no Yes/no Yes/no Yes/no Yes/no No/no No/no Lung crackles Yes Yes Yes Yes Yes Yes Yes Raynaud’s phenomenon No Yes Yes No No No No Mechanic’s hands No Yes Yes Yes No No No Other cutaneous signs No No Papules of knees No Palmoplantar No No and elbows, erythema ‘V sign’ rash

NYHA, New York Heart Association classiﬁcation.

Fig 1. (a, b) Hyperkeratotic erythema on the lateral aspects of the ﬁngers typical of ‘mechanic’s hands’.

MH was present at diagnosis in two cases, but always in as- with idiopathic inflammatory myopathy, but in up to 60–70% sociation with involvement of at least one other deep tissue. of those with idiopathic inflammatory myopathy and intersti- In two other cases, MH was absent at diagnosis but developed tial pulmonary fibrosis.8 during a subsequent flare of the disease consisting of the same Our series shows that four of seven (57%) of patients previously described lesions. In these two patients, the erup- with antisynthetase syndrome had MH. This result is close to tion was not the only sign of flare: one had worsening of the 71% found by Love et al.1 Among the four patients who respiratory disease and the other developed muscular impair- had MH in our series, this sign was the only cutaneous le- ment. In the four cases, MH improved with treatment of the sion in three. The fourth had also a ‘V sign’ rash and ery- antisynthetase syndrome, along with systemic signs. thematosquamous lesions on the knees and elbows, as in The new classification of idiopathic inflammatory myopathy dermatomyositis. MH was either present at the initial diagno- is based upon the presence of the myositis-specific autoanti- sis or developed during relapse. MH consists of well-demar- bodies.1,2 These autoantibodies consist of several types of cated hyperkeratotic and nonpruritic erythema on the lateral autoantibodies directed towards aminoacyl-tRNA synthetases and palmar aspects of the hands and fingers.1 Fissuring and (mainly Jo-1), rarely towards signal recognition peptide and scaling are also associated. Skin biopsy shows parakeratosis, the helicase enzyme.2,8 Although the specificity of myositis- acanthosis and a scattered lymphohistiocytic infiltrate in the specific autoantibodies for myositis is high, the sensitivity is papillary dermis encroaching into the epidermis. The basal low and anti-Jo-1 antibodies are detected in 20–30% patients layer of epithelium shows focal vacuolization.4,7 These

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 194 Correspondence features have been termed MH and are close to the calloused 5 Bergoin C, Bure M, Tavernier JY et al. The anti-synthetase syn- hands of manual workers. Discrimination between occupa- drome. Rev Mal Respir 2002; 19:371–4. tional dermatitis and MH is difficult; however, lesions appear 6 Shibuya H, Arakawa S, Kai Y et al. Three cases of ‘mechanic’s hands’ associated with interstitial pneumonia: possible involvement well demarcated with interface dermatitis in the latter. There- with foot lesions. J Dermatol 2003; 30:892–7. fore, the diagnosis of MH should be considered in patients 7 White JM, Salisbury JR, Creamer D. Eczematous changes on the especially when they present with general signs, interstitial hands – quiz case. Diagnosis: ‘mechanic’s hands’ as part of the lung disease and/or inflammatory muscle disease and/or antisynthetase syndrome. Arch Dermatol 2005; 141:779–84. polyarthritis requiring a search for anti-Jo-1 antibodies. A 8 Imbert-Masseau A, Hamidou M, Agard C et al. Antisynthetase syn- close association has been shown between MH and anti-Jo-1 drome. Joint Bone Spine 2003; 70:161–8. antibodies, but MH has also been reported in polymyositis, 9 Garcia-Patos V, Bartralot R, Fonollosa V et al. Childhood sclerodermatomyositis: report of a case with the anti-PM/Scl antibody and dermatomyositis, childhood sclerodermatomyositis and mixed 1,9,10 mechanic’s hands. Br J Dermatol 1996; 135:613–16. connective disease. 10 Torok L, Danko K, Cserni G, Szucs G. PM-SCL autoantibody posi- In two of seven patients, MH occurred in the course of tive scleroderma with polymyositis (mechanic’s hand: clinical aid known idiopathic inflammatory myopathy in whom the skin in the diagnosis). J Eur Acad Dermatol Venereol 2004; 18:356–9. was not involved initially. In these cases, systemic involvement developed concomitantly or soon after. A skin flare was there- Conflicts of interest: none declared. fore always associated with deep tissue flare; such an event in patients with antisynthetase syndrome implies deep organ assessment. The patients with MH became clear of their lesions together with clearance of the deep tissue disease. When MH was associated with idiopathic inflammatory myopathy in our series, it never appeared as an isolated skin sign. This sug- A successful therapeutic trial of rituximab in gests that when MH is present in a patient free of systemic the treatment of a patient with recalcitrant, signs, the diagnosis cannot be antisynthetase syndrome; the high-titre epidermolysis bullosa acquisita hand lesions belong rather to a classical occupational dermatitis and a search for anti-Jo-1 antibodies need not be done in DOI: 10.1111/j.1365-2133.2006.07596.x this situation. Prognosis of the antisynthetase syndrome is poor as its mor- SIR, Rituximab is a humanized monoclonal antibody, which tality rate is 40%, with 62% in patients with interstitial lung binds to the CD20 molecule expressed on cells of B-lympho- disease, but could be improved by an early diagnosis.8 There- cyte lineage, causing cell depletion. It is licensed for the treat- fore, in view of the frequent occurrence of this sign in ment of advanced follicular lymphoma and large B-cell antisynthetase syndrome, and the importance of early manage- non-Hodgkin’s lymphoma. More recently, it has been used ment, careful examination of the hands is mandatory in successfully in the treatment of autoimmune blistering dis- patients with systemic signs such as interstitial lung disease, eases, predominantly pemphigus vulgaris.1,2 We report the myositis and arthritis. use of rituximab to treat a patient with severe epidermolysis bullosa acquisita (EBA). Service de Dermatologie and C. BACHMEYER A 58-year-old woman presented in November 2002 with *Service de Pneumologie, I. TILLIE-LEBLOND* oral ulceration and blisters and erosions on the upper torso CHU Tenon (AP-HP), 4 rue de la Chine, A. LACERT and flexures. Skin biopsy showed a subepidermal bulla with a 75020 Paris, France J. CADRANEL* mixed cellular infiltrate including eosinophils. Direct immuno- E-mail: [email protected] S. ARACTINGI fluorescence showed deposition of IgG and C3 along the basement membrane zone. Indirect immunofluorescence (IIF) References showed IgG antibasement membrane zone antibodies staining the dermal side of salt-split normal human skin at a titre of 1 Love LA, Leff RL, Fraser DD et al. A new approach to the classification of idiopathic inflammatory myopathy: myositis-specific auto- 1 : 400. Immunoblotting using dermal extract showed bind- antibodies define useful homogeneous patient groups. Medicine ing of IgG to a 290-kDa molecule, which comigrated with (Baltimore) 1991; 70:360–74. type VII collagen, confirming the diagnosis of EBA. 2 Troyanov Y, Targoff IN, Tremblay JL et al. Novel classification of Over time, the clinical features evolved and became typical idiopathic inflammatory myopathies based on overlap syndrome of mechanobullous EBA with spontaneous blisters and extreme features and autoantibodies: analysis of 100 French Canadian skin fragility (particularly on trauma-prone sites including patients. Medicine (Baltimore) 2005; 84:231–49. hands, feet, elbows and knees), scarring, milia formation and 3 Stahl NI, Klippel JH, Decker JL. A cutaneous lesion associated with myositis. Ann Intern Med 1979; 91:577–9. nail loss (Fig. 1). 4 Mitra D, Lovell CL, Macleod TIF et al. Clinical and histological fea- Response to treatment was at best partial and had been tures of ‘mechanic’s hands’ in a patient with antibodies to Jo-1 – a complicated by side-effects. Treatment included combinations case report. Clin Exp Dermatol 1994; 19:146–8. of prednisolone at doses up to 60 mg daily, azathioprine

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 195

(a) (b)

(c) (d) Fig 1. Patient with typical signs of mechanobullous epidermolysis bullosa acquisita: (a) ulceration of the tongue and lower labial mucosa, (b) erosions and erythematous patches, where erosions have healed, on the elbows, (c) blisters, erosions, milia, cutaneous atrophy and steroid purpura on the dorsum of the right hand and (d) disabling trauma-induced erosions on the palms of the hands.

) 1Æ5mgkg 1 daily (withdrawn due to hepatotoxicity), myco- thoracic oesophagus suggestive of a small submucosal lesion. phenolate mofetil (MMF) 1 g twice daily (constipation and She declined endoscopy but a thoracic computed tomographic faecal impaction at 1Æ5 g twice daily), and ciclosporin scan was normal and the abnormalities were attributed to ) 3mgkg 1 daily (withdrawn due to nephrotoxicity and hyper- oesophageal blistering. She had significant treatment side- tension). Monthly intravenous immunoglobulin infusions effects (cushingoid habitus, hypertension, diabetes, bilateral ) (2 g kg 1 per month) were combined with prednisolone and cataracts and skin atrophy) but poor disease control. Methyl- MMF, but were abandoned after four cycles due to clinical de- prednisolone infusions (500 mg) were given as a short-term terioration and rising IIF titres (Fig. 2) which peaked at measure on three consecutive days in May 2005 while fund- 1 : 3200. In May 2005, the patient was hospitalized with ex- ing for rituximab was secured. ) tensive cutaneous and oral ulceration, cellulitis and a deep Four rituximab infusions (375 mg m 2) were given without vein thrombosis secondary to immobility. Dressing changes, complication at weekly intervals in June 2005 in combination because of the large areas involved and significant pain, took with MMF 1 g twice daily and prednisolone 30 mg once daily. several hours. She had dysphagia for solids, was limited to a There has been slow but progressive clinical improvement, ini- soft diet and had been unable to swallow MMF tablets. A bar- tially observed within a week of the first rituximab infusion, ium swallow showed a filling defect in the upper third of the with complete healing by November 2005. One year after

CiA Aza MMF Pred IVIG 3200 2800 Me-Pred 2400 Rituximab 2000 IIF titre 1600 Fig 2. Changes in indirect 1200 immunoﬂuorescence (IIF) titres over time. Titres are expressed as the maximum serum 800 dilution at which basement membrane 400 antibodies were detected. CiA, ciclosporin; 0 Aza, azathioprine; MMF, mycophenolate 3 3 3 3 4 4 4 4 5 5 5 5 6 6 mofetil; Pred, prednisolone; IVIG, intravenous -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 -0 b y g v b y g v b y g v b y immunoglobulin infusions; Me-Pred, e u o e a u o e a u o e a F Ma A N F M A N F M A N F M methylprednisolone.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 196 Correspondence rituximab, she is in partial remission, suffering occasional and 3 Schmidt E, Benoit S, Brocker EB et al. Successful adjuvant treatment small trauma-induced blisters only, such that she has manage- of recalcitrant epidermolysis bullosa acquisita with anti-CD20 anti- able disease with a relatively normal quality of life. The dose body rituximab. Arch Dermatol 2006; 142:147–50. of prednisolone has been reduced to 5 mg daily, MMF contin- Conflicts of interest: none declared. ued at 1 g twice daily and there has been a fall in IIF titres to 1 : 10 (Fig. 2). Predictably, the peripheral B-cell count fell ) from normal pretreatment levels of 0Æ12 · 109 L 1 to ) <0Æ01 · 109 L 1 immediately post-treatment (normal 0Æ06– 0Æ47 · 109). One year later, B-cell levels remain low at 9 )1 )1 <0Æ01 · 10 L . Total IgG was mildly reduced at 5Æ7gL Spotted and rippled reticulate hypermelanosis: (normal 6–16) prior to rituximab, presumably due to immu- ) a possible variant of Dowling–Degos disease nosuppressive treatment. Levels fell to 4Æ9gL 1 following )1 treatment and remain low at 4Æ7gL at 1 year later. DOI: 10.1111/j.1365-2133.2006.07592.x EBA is a rare acquired, subepidermal bullous disease associated with autoimmunity to type VII collagen. It is notoriously SIR, Pigmentation disorders derive from melanoblast migration, difficult to treat and due to its rarity, evidence is anecdotal biogenesis of melanosomes, and secretion of melanosomes into and largely limited to case reports. We embarked on this keratinocytes.1 Reticulate hyperpigmentary disorders have been therapeutic trial because of the lack of response to more con- classified into dyschromatosis symmetrica hereditaria (DSH) of ventional immunosuppression in a patient who had aggres- Toyama (MIM 127400), dyschromatosis universalis hereditaria sive, uncontrolled disease. At the time, there were no (DUH) (MIM 127500), Dowling–Degos disease (DDD) (MIM published data on rituximab for EBA but the rationale for its 179850) and reticulate acropigmentation of Kitamura (RAK). use was the growing body of publications reporting successful We encountered a 3-year-old girl showing asymptomatic and treatment of other antibody-mediated diseases including pem- asymmetrical, spotted and rippled reticulate hypermelanosis 1 phigus vulgaris. Although her initial improvement could be without hypomelanosis. Here, we discuss the similarities to partly attributed to methylprednisolone, which was pulsed and differences from previously defined disorders. shortly before rituximab, the long-term disease control is A 3-year-old Japanese girl presented with spotted and rip- almost certainly the effect of rituximab. Schmidt et al. have re- pled reticulate hypermelanosis without hypopigmented spots. cently published another patient with EBA in whom rituximab The pigmentation showed an asymmetrical distribution. The 3 resulted in sustained clinical remission. Although rituximab is pigmentation gradually developed from the age of 1 year. Rip- expensive, in this case the cost of dressings alone during her pled reticulate hypermelanosis was observed around the left 4-week hospital admission was two-thirds that of the drug. axilla, and faint hypermelanosis around the right axilla. Spot- In summary, rituximab used as an adjuvant drug has ted and rippled reticulate hypermelanosis was present on the achieved a clinically useful partial remission in a patient with flexures of the extremities, especially on the right thigh severe, recalcitrant EBA, resulting in considerable improvement (Fig. 1a). Spotted reticulate hypermelanosis was present on in life quality and allowing a reduction in concomitant im- the dorsa of the hands and feet (Fig. 1b). The dorsum of the munosuppression. Rituximab should be considered as a thera- left foot showed more pronounced hypermelanosis (Fig. 1b). peutic option in such patients. V-neck-shaped asymmetrical hyperpigmentation was observed. The patient had normal palms and soles, and no other cutane- Acknowledgments ous abnormalities. Her relatives did not have reticulate hyperpigmentation. Skin biopsies were taken from sun-protected We thank Dr Nori Oyama for immunoblotting and Mr B.S. areas of the upper back: a hyperpigmented lesion on the left Bhogal for indirect immunofluorescence. side and a normal skin-coloured lesion on the right side. His- topathology of both biopsy specimens showed the same fea- Department of Dermatology, S.M. CRICHLOW tures. Haematoxylin and eosin staining showed no papillary University Hospitals Leicester, N.J. MORTIMER downgrowth of the epidermis. Fontana–Masson staining indi- Leicester Royal Infirmary, K.E. HARMAN cated asymmetrical distribution of melanin granules and per- Leicester LE1 5WW, U.K. sistence of melanin granules through the epidermis (Fig. 2a). E-mail: [email protected] Electron microscopy showed a small number of stage II melanosomes, about 45% of melanosomes being stage III, and the References remainder stage IV (Fig. 2b). The features of the patient were as follows: (i) the onset of 1 Schmidt E, Hunzelman N, Zillikens D et al. Rituximab in refractory hypermelanosis from age 1 year; (ii) clinical appearance: (a) autoimmune bullous diseases. Clin Exp Dermatol 2006; 31:503–8. spotted and rippled reticulate hypermelanosis that developed 2 Arin MJ, Engert A, Krieg T et al. Anti-CD20 monoclonal antibody (rituximab) in the treatment of pemphigus. Br J Dermatol 2005; asymmetrically on the flexures and acral regions, and (b) pro- 153:620–5. nounced asymmetrical hyperpigmentation after sun exposure;

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 197

(a)

(b) (b)

Fig 1. Clinical appearance of (a) the right thigh and (b) the dorsum Fig 2. Fontana–Masson stain and electron microscopy from the of the left foot. patient. (a) Fontana–Masson staining demonstrated a scattered distribution of melanin granules in the epidermis, and persistence of (iii) histopathology: (a) scattered distribution of melanin melanin granules through the epidermis. Original magnification granules in the epidermis, (b) persistence of melanin granules · 400. (b) Electron microscopy showed a small number of stage II melanosomes, about 45% of melanosomes being stage III, and the through the epidermis, and (c) no club-like elongation of the remainder stage IV. Bar ¼ 500 nm. rete ridges; and (iv) electron microscopy: the presence of about 45% stage III melanosomes and the remainder stage IV In cases of RAK, palmar and planter pits are observed. The his- melanosomes. topathological and electron microscopic features in RAK are DSH, DUH, DDD and RAK are genodermatoses with auto- similar to those in DDD.5 Occasional cases of overlapping dis- somal dominant inheritance. Most cases of DSH, DUH and tribution of DDD/RAK,6 DDD/RAK/DSH7 and DDD/DUH3 RAK have been reported from Japan. DSH and DUH are char- have been reported. acterized by hyper- and hypopigmented macules. The absence The onset at age 1 year in our patient was consistent with of hypomelanosis in our patient is clearly different from DSH RAK, but was earlier than in DDD. The appearance of the le- and DUH. DDD develops at the postpubertal period, and is sions was of a DDD/RAK overlap. Asymmetrical hypermelano- characterized by asymptomatic, symmetrical and progressive sis of the flexures and acral regions has not been described. reticulate hyperpigmentation, follicular plugging and pitted Skin biopsy showed different features, i.e. no papillary down- scars on the flexures.2 Sun exposure makes the pigmentation growth of the epidermis. Fontana–Masson staining showed more pronounced in some cases.3 Histopathological examin- manifestations similar to DDD: scattered distribution of mel- ation in DDD demonstrates club-like elongation of the rete anin granules in the epidermis and persistence of melanin ridges and an excess of melanin at the tip of the basal layer.2 granules through the epidermis. Electron microscopy indicated With Fontana–Masson stain, DDD shows scattered distribution stage III and IV melanosomes, similar to DDD and RAK. of melanosomes without nuclear cap-like structures, and per- Together, the clinicohistopathological examination showed sistence of melanosomes throughout the epidermis.2 Electron features similar to DDD and RAK, although distinct differences microscopy reveals melanin pigments in melanosomes of stage were present. DDD appears to be genetically heterogeneous. III and IV organelles.2 RAK develops within the first two dec- Very recently, two loci, keratin 5 on chromosome 12q and ades, and is characterized by reticulate hyperpigmented ma- another on chromosome 17p13.3, were shown to be linked cules without hypopigmented macules on the acral regions.4 to DDD.2,8 Keratin 5 is a component of the intermediate fila-

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 198 Correspondence ment (IF) cytoskeleton in the basal layer of the keratinocytes. virus (EBV) in immunocompromised patients and those Dysfunction of the IF cytoskeleton causes aberrant distribution induced by Borrelia burgdorferi.1,2 We report here a case occur- of melanosomes in keratinocytes,2 suggesting a delayed deg- ring on an area previously irradiated for breast cancer and dis- radation of melanin granules.2 cuss the potential pathogenetic role of radiotherapy. In summary, we have shown an uncommon case of reticu- A 74-year-old woman presented in October 2005 with a 3- late hypermelanosis, that is, spotted and rippled reticulate month history of a progressively growing lesion on her right hypermelanosis developing asymmetrically. We speculate that breast. She had been diagnosed in 1987 with an invasive lo- the clinicopathological manifestation may be a possible variant bular cancer of the right breast in the upper outer quadrant of DDD and that it may have been derived from aberrant deg- (T1N0M0), which was treated with lumpectomy, axillary radation of melanin granules in keratinocytes showing persist- lymph node dissection, interstitial brachytherapy (15 Gy) and ence of melanin granules through the epidermis. external-beam breast irradiation (49.5 Gy). On examination, the patient was asymptomatic. A 1.5 · 3.5 cm erythematous Department of Dermatology, N. OISO nodule was found on the skin of the left superior quadrant of Kinki University School of Medicine, D. TSURUTA* the right breast, close to the scar of lumpectomy (Fig. 1). No 377-2 Ohno-Higashi, Osaka-Sayama, T. OTA lymphadenopathy was palpable. Routine laboratory studies Osaka 589-8511, Japan H. KOBAYASHI* were normal. Borrelia burgdorferi serology was negative. Mammo- *Department of Dermatology, A. KAWADA graphy was normal and ultrasound imaging of the right breast Osaka City University Graduate School of Medicine, disclosed the skin nodule and two small right axillary lymph Osaka, Japan nodes. Positron emission tomography scan showed 18-fluo- E-mail: [email protected] rodeoxyglucose uptake in the skin nodule and in the lymph nodes related to a benign follicular hyperplasia. Biopsy of the References nodule disclosed a massive dermal and hypodermal lymphocytic infiltrate with a diffuse and follicular pattern. The diffuse 1 Bleehen SS, Anstey AV. Disorders of skin colour. In: Rook’s Textbook of infiltrate consisted of small lymphocytes with regular nuclei Dermatology (Burns T, Breathnach SM, Cox NH, Griffiths CEM, eds), associated with nodular clusters of middle-sized lymphocytes 7th edn, Vol. 2. Oxford: Blackwell Publishing, 2004; 39.13–39.15. displaying abundant cytoplasm and irregular nuclei (Fig. 2). 2 Betz RC, Planko L, Eigelshoven S et al. Loss-of-function mutations in Both types of cell strongly expressed CD20 and CD10, but not the keratin 5 gene lead to Dowling–Degos disease. Am J Hum Genet 2006; 78:510–5. Bcl-2. Immunoglobulin receptor rearrangement DNA was pre- 3 Sandhu K, Saraswat A, Kanwar AJ. Dowling–Degos disease with dys- sent. Computed tomography scan of the abdomen and thorax, chromatosis universalis hereditaria-like pigmentation in a family. J and bone marrow biopsy were normal. Therefore, diagnosis Eur Acad Dermatol Venereol 2004; 18:702–4. of primary cutaneous follicular centre cell lymphoma (FCCL) 4 Danesen P, Zanca A, Bertazzoni MG. Familial reticulate acropigmen- was established. Topical treatment with clobetasol propionate tation of Dohi. J Am Acad Dermatol 1997; 37:884–6. 0.05% cream and mechlorethamine 0.02% aqueous solution 5 Erel A, Gurer ML, Edali N. Reticulate acropigmentation of Kitamura: once daily was begun. Remission was obtained after 2 months two case reports. Int J Dermatol 1993; 32:726–7. 6 Al-Hawsawi K, Al-Aboud K, Alfadley A, Al-Aboud D. Reticulate ac- leaving a slight pigmented macule. ropigmentation of Kitamura–Dowling Degos disease overlap: a case FCCLs are rare low-grade lymphomas, presenting as asymp- report. Int J Dermatol 2002; 41:518–20. tomatic solitary or grouped skin lesions localized on the head 7 Thami GP, Jaswal R, Kanwar AJ et al. Overlap of reticulate acro- or trunk.3,4 Histologically, the condition is featured by both pigmentation of Kitamura, acropigmentation of Dohi and Dowling– Degos disease in four generations. Dermatology 1998; 196:350–1. 8 Li CR, Xing QH, Li M et al. A gene locus responsible for reticulate pigmented anomaly of the flexures maps to chromosome 17p13.3. J Invest Dermatol 2006; 126:1297–301.

Conﬂicts of interest: none declared.

Primary cutaneous follicular B-cell lymphoma arising at the site of radiotherapy for breast cancer

DOI: 10.1111/j.1365-2133.2006.07600.x

SIR, Primary cutaneous B-cell lymphomas are rare entities of Fig 1. Erythematous nodule of primary cutaneous follicular B-cell unknown cause, except for those induced by Epstein–Barr lymphoma on the scar of previous lumpectomy for breast cancer.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 Correspondence 199

Service de Me´decine Interne, C. BACHMEYER *Service de Dermatologie and K. KHOSROTEHRANI* Service d’Anatomo-Pathologie, P. MOGUELET CHU Tenon (AP-HP), 4 rue de la Chine, S. ARACTINGI* 75020 Paris, France E-mail: [email protected]

References

1 Wilkins K, Turner R, Dolev JC et al. Cutaneous malignancy and human immunodeficiency virus disease. J Am Acad Dermatol 2006; 54:189–206. 2 Bogle MA, Riddle CC, Triana EM et al. Primary cutaneous B-cell lymphoma. J Am Acad Dermatol 2005; 53:479–84. 3 Willemze R, Jaffe ES, Burg G et al. WHO-EORTC classification for cutaneous lymphomas. Blood 2005; 105:3768–85. 4 Bergman R, Kurtin PJ, Gibson LE et al. Clinicopathologic, immunophenotypic, and molecular characterization of primary cutaneous follicular B-cell lymphoma. Arch Dermatol 2001; 137:432–9. 5 Brenn T, Fletcher CDM. Radiation-associated cutaneous atypical vascular lesions and angiosarcoma. Clinicopathologic analysis of 42 Fig 2. Nodular infiltrate of follicular B-cell lymphoma involving the cases. Am J Surg Pathol 2005; 29:983–96. entire dermis down to the subcutaneous fat (haematoxylin and eosin, 6 Stein M, Haim N, Kuten A et al. Primary brain lymphoma after original magnification · 25). X-ray irradiation to the scalp for tinea capitis in childhood. J Surg Oncol 1992; 50:270–3. 7 Yukiiri K, Mizushige K, Ueda T, Kohno M. Second primary cardiac small and middle-sized neoplastic B cells with a follicular pat- B-cell lymphoma after radiation therapy and chemotherapy. Angiology tern. They rarely express Bcl-2 and are not associated with the 2001; 52:563–5. 8 Mao JH, Wu D, Perez-Losada J et al. Genetic interactions between t(14;18) translocation, in contrast to nodal follicular lympho- Pten and p53 in radiation-induced lymphoma development. Oncogene mas. 2003; 22:8379–85. Risk factors for the development of B-cell lymphomas are poorly understood. Briefly, there are reports of B-cell lympho- Conflicts of interest: none declared. proliferative diseases with the presence of EBV, mainly in patients treated with immunosuppressive drugs such as methotrexate, and in patients with human immunodeficiency virus.1 However, in these cases, the pathology was different as these were not follicular proliferations, but rather diffuse infiltration of large B cells. Rarely, B. burgdorferi has been associated with the development of FCCL.2 Additional dermoscopic presentation of We report a unique case of low-grade B-cell lymphoma that haemosiderotic dermatofibroma developed at the precise site of brachytherapy plus external- beam radiotherapy for breast cancer. Oncogenic transform- DOI: 10.1111/j.1365-2133.2006.07611.x ations due to the carcinogenetic effects of ionizing radiation 1 are known complications of irradiated skin, but usually lead to SIR, We read with interest the paper by Zaballos et al., who carcinoma and sarcoma.5 Although a fortuitous association reported the dermoscopic features of haemosiderotic and cannot be ruled out, the development of a rare condi- aneurysmal dermatofibroma. In this context we would like to tion—namely FCCL—on the precise site of an event known to report a recent additional observation concerning a 50-year- be able to promote oncogenesis of other cell types suggests old woman with an asymptomatic, pigmented nodular lesion that radiation may be a risk factor for cutaneous lymphoma. on the leg. Interestingly, a small number of single case reports of primary The lesion had appeared 3 years before, unrelated to lymphoma involving the brain or the heart after radiation trauma, and had increased in size and changed in colour dur- therapy have been described.6,7 The role of genetic alterations ing the last year. On examination, there was a firm, central in Pten and p53 pathways in radiation-induced lymphoma de- hyperpigmented, erythematous–violaceous nodular lesion, velopment is suggested by some authors.8 located on the posterior surface of the left leg, 9 mm in diam- In conclusion, a prolonged clinical surveillance of irradiated eter (Fig. 1a). Adjacent to the lesion a brown, symmetrical, skin especially in patients with breast cancer is mandatory in macule, 1 mm in diameter, was observed (Fig. 1a). Upon order to establish an early diagnosis of carcinoma and sarcoma dermoscopy, the lesion was asymmetrical in shape and colour but perhaps also primary cutaneous lymphoma. (Fig. 1b). In the centre a brown–grey pigmentation with a

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 200 Correspondence

histiocytic proliferation antler- to staghorn-like vascular spaces lined by endothelial cells revealing a haemangiopericytoma-like pattern were observed. In addition, at the specimen edge, but without connection to the main lesion, there was a proliferation of melanocytes arranged mainly in small nests at the dermoepidermal junction in accordance with a stereotypical junctional type of dysplastic naevus, thus explaining the brown macule at the right side of the lesion, which dermoscopically revealed a brown to black prominent reticulated pattern. Zaballos et al.1 reported as features of aneurysmal dermatofibroma the presence of a homogeneous blue–yellow to red- brownish area in the centre of the lesion, white linear and homogeneous structures, a delicate pigment network and vascular structures (comma vessels, dotted vessels) at the periphery, a scaly surface and a yellowish homogeneous area at the Fig 1. Clinical image: central hyperpigmented nodule surrounded by periphery. Our case differs from the cases reported by Zaballos an erythematous–violaceous halo on the posterior surface of the left et al. by the presence of a blue-whitish veil in the centre of leg. (b) Dermoscopic image: central blue-whitish veil structure, black the lesion, black dots and pigmented irregular streaks at the dots, irregular streaks and a scar-like area. upper part of the lesion and the absence of vascular structures and white linear homogeneous structures. Dermoscopic features of our case were highly suspicious for melanoma and the definitive diagnosis could be made only upon histopathology.2,3 Our case differs also from a case published by Blum et al.,4 who described globules in the centre of the lesion, streaks and a pigment network at the periphery and asymmetrical blue-greyish areas.4,5 In conclusion, we report an additional case of haemosiderotic dermatofibroma with dermoscopic features different from the previous observations reported in the literature, thus underlining the polymorphic dermoscopic appearances of this rare type of dermatofibroma.

*Department of Dermatology, R. CARDOSO* University Hospital, Coimbra, Portugal C. MASSONE Department of Dermatology, Medical H.P. SOYER OFMANN ELLENHOF Fig 2. Histological image: (a) dense fibrohistiocytic proliferation of University of Graz, Auenbruggerplatz 8, R. H -W fibrocytes and macrophages with haemosiderin depositions A-8036 Graz, Austria throughout the reticular dermis; (b) haemosiderin depositions Correspondence: Rainer Hofmann-Wellenhof. throughout the reticular dermis (Pearl’s stain). Haematoxylin and E-mail: [email protected] eosin; original magnification (a) · 20; (b) · 400. very delicate blue-whitish veil was seen, along with a few tiny References black dots and dark-brown irregular streaks at the upper pole 1 Zaballos P, Liambrich A, Ara M et al. Dermoscopic findings of hae- of the central brown–grey pigmentation. To the right of this mosiderotic and aneurysmal dermatofibroma: report of six patients. structure there was a semicircular scar-like area. At the periph- Br J Dermatol 2006; 154:244–50. ery a fine and regular pigmented network was observed. At 4 2 Zelger BW, Zelger BG, Steiner H et al. Aneurysmal and haemangio- o’clock a small area displaying a prominent brown to black pericytoma-like fibrous histiocytoma. J Clin Pathol 1996; 49:313–18. pigment network with a tiny spot of central hyperpigmenta- 3 Kaddu S, McMenamin ME, Fletcher CD. Atypical fibrous histio- tion could be observed (Fig. 1b). cytoma of the skin: clinicopathologic analysis of 59 cases with evidence of infrequent metastasis. Am J Surg Pathol 2002; 26:35– The lesion was subsequently excised and the histopathologi- 46. cal examination disclosed epidermal hyperplasia characterized 4 Blum A, Jaworsky S, Metzler G et al. Lessons on dermoscopy: der- by thickened rete ridges with an increased basal hyperpigmen- matoscopic pattern of hemosiderotic dermatofibroma. Dermatol Surg tation. However, the main histopathological substrate was a 2004; 30:1354–5. dense infiltrate of fibrocytes and macrophages with haemosid- 5 Johr RH, Soyer P, Argenziano G et al. Dermoscopy: the Essentials. London: erin deposition throughout the reticular dermis extending to Mosby, Elsevier Science Ltd, 2004. the subcutaneous fat (Fig. 2a,b). With scanning magnification in the upper and particularly in the lower part of the fibro- Conflicts of interest: none declared.

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201 News and Notices 201

News and Notices can be accessed via the BAD website http://www.bad.org.uk/ healthcare/annual_meeting/ DOI: 10.1111/j.1365-2133.2006.07694.x The closing date for the receipt of abstracts is Monday 8th January 2007 and the deadline will be adhered to strictly. Any abstracts received after this date will not be considered. PERI-OCULAR TUMOUR COURSE/2007 BSDS ADVANCED The deadline for abstract submissions to any of the special SURGERY WORKSHOP Wednesday interest group meetings will be Monday 5th February 2007. 28th February 2007 Freeman Hospital, Newcastle Poster submissions on the following topics are strongly This intensive one day course – taught by a faculty of oculo- encouraged: audit, medical education, and service delivery. plastic and dermatological surgeons - is aimed primarily at Conference & Event Services, British Association of Derma- senior trainees and consultants in ophthalmology and dermatologists, 4 Fitzroy Square, London, W1T 5HQ, UK or email tology, although delegates from other disciplines are most [email protected] welcome to attend. The fee is £150. For further details including bookings, please visit http:// Congress: 2nd International Congress on Psoriasis www.bsds.org.uk. Dates: JUNE 21–24, 2007 Venue: Paris, Palais des congre`s, FRANCE BSDS 2007 Annual surgical workshop Web site: http://www.pso2007.com Monday 16th – Thursday 19th April 2007 Contact: Ninewells Hospital & Medical School, Dundee PSO 2007 c/o MCI This course is aimed primarily at Specialist Registrars and Clin- 24 rue Chauchat ical Assistants in dermatology, and uses lectures and practical 75009 Paris sessions to teach dermatological surgical skills and techniques. FRANCE This course is always fully subscribed and early booking is Phone: +33 (1) 53 85 82 59 advised. The workshop fee is £625 for Specialist Registrars or Fax: +33 (1) 53 85 82 83 £895 for others. For further details including bookings, Email: [email protected] please visit www.bsds.org.uk. Main topics: 87th Annual Meeting of the British Association of – Scoring and monitoring the severity of the disease Dermatologists – Management of the severe clinical manifestations 10th–13th July 2007, Birmingham – Psoriasis in children and pregnancy The 87th Annual Meeting of the British Association of Derma- – Difﬁcult to treat localisations tologists will be held at the Brimingham International Confer- – Topical Treatment: what to choose and how to use ence Centre (ICC), Birmingham, 10th–13th July 2007, – Risk management and treatment optimisation: combination coorganised by Dr Colin Holden, BAD President Elect. and rational strategies Abstracts of papers and posters should be submitted for con- – Phototherapies: what to choose and how to use sideration by the Scientiﬁc Committee. Original communica- – Alternative treatment tions will be allotted 15 minutes, which must include time for – Biologics discussion. Clinicopathological cases will be allotted 7 minutes – Extracutaneous manifestations in Psoriasis for presentation, again allowing time for discussion. Online submission will be the only method of abstract submission available. Full instructions and the submission form

2006 The Authors Journal Compilation 2006 British Association of Dermatologists • British Journal of Dermatology 2007 156, pp163–201