Frontal Fibrosing Alopecia: Possible Association with Leave&#X2010

Home , Hair gel, Hair mousse, Hair spray

BJD GENERAL DERMATOLOGY British Journal of Dermatology Frontal ﬁbrosing alopecia: possible association with leave-on facial skin care products and sunscreens; a questionnaire study* N. Aldoori,1 K. Dobson,1 C.R. Holden,1 A.J. McDonagh,1 M. Harries2 and A.G. Messenger1 1Department of Dermatology, Royal Hallamshire Hospital, Shefﬁeld S10 2JF, U.K. 2The Dermatology Centre, Manchester University, Salford Royal NHS Foundation Trust, Salford M6 8HD, U.K.

Linked Comment: Donati. Br J Dermatol 2016; 175:675–676.

Summary

Correspondence Background Since its first description in 1994, frontal fibrosing alopecia (FFA) has become Andrew G. Messenger. increasingly common, suggesting that environmental factors are involved in the aetiology. E-mail: a.g.messenger@sheffield.ac.uk Objectives To identify possible causative environmental factors in FFA. Methods A questionnaire enquiring about exposure to a wide range of lifestyle, Accepted for publication social and medical factors was completed by 105 women with FFA and 100 age- 10 May 2016 and sex-matched control subjects. A subcohort of women with FFA was patch Funding sources tested to an extended British standard series of allergens. The study was supported by a grant from the Results The use of sunscreens was significantly greater in the FFA group compared British Skin Foundation. with controls. Subjects with FFA also showed a trend towards more frequent use of facial moisturizers and foundations but, compared with controls, the differ- Conflicts of interest ence in frequencies just failed to reach statistical significance. The frequency of None declared. hair shampooing, oral contraceptive use, hair colouring and facial hair removal *Plain language summary available online were significantly lower in the FFA group than in controls. Thyroid disease was more common in subjects with FFA than controls and there was a high fre- DOI 10.1111/bjd.14535 quency of positive patch tests in women with FFA, mainly to fragrances. Conclusions Our findings suggest an association between FFA and the use of facial skin care products. The high frequency of sunscreen use in patients with FFA, and the fact that many facial skin care products now contain sunscreens, raises the possibility of a causative role for sunscreen chemicals. The high frequency of positive patch tests in women with FFA and the association with thyroid disease may indicate a predisposition to immune-mediated disease.

What’s already known about this topic?

• Frontal ﬁbrosing alopecia (FFA) is an inﬂammatory scarring alopecia that mainly involves the frontal and frontal/temporal hairline and eyebrows. The histopathology is identical to that of lichen planopilaris. • FFA mainly affects postmenopausal women although it also occurs in younger women and occasionally in men. • First reported in 1994, FFA remained rare during the 1990s but has become increasingly common in the last 10–15 years.

What does this study add?

• Leave-on facial skin care products are implicated in the aetiology of FFA. • It is suggested that sunscreen chemicals play a causative role. • It is suggested that there is an increased predisposition to immune-mediated disease in FFA.

Frontal fibrosing alopecia (FFA) is an irreversible scarring Subjects alopecia that mainly affects postmenopausal women.1,2 The histopathology is identical to that of lichen planopilaris (LPP) Patients were recruited during 2012 at two sites in the U.K., but the clinical presentation is distinctive and differs from that the Royal Hallamshire Hospital, Sheffield, and Salford Royal of classical LPP. The disease is characterized by progressive NHS Foundation Trust, Salford, Greater Manchester. They regression of the frontal and temporal hairline, in most cases were recruited from general dermatology clinics and dedicated accompanied and sometimes preceded by loss of the eye- specialist hair clinics. The diagnosis of FFA was made by a brows. Follicular erythema and scale is commonly seen along consultant dermatologist with specialist expertise in hair dis- the receding hairline. The progression of the alopecia is vari- ease. The clinical diagnosis was based on typical scarring able and unpredictable, although it can spontaneously stabi- alopecia affecting the frontal hairline and causing recession. lize.3 Some women also report loss of hair from other body Additional features included complete or partial loss of eye- sites, such as the limbs and the axillae. Although predomi- brows, follicular erythema of the frontal hairline and involve- nantly a condition of postmenopausal women, FFA also occurs ment of the hairline over the sides of the scalp. – in premenopausal women and has been reported in men.4 6 Those patients in whom a dual diagnosis of FFA and a fur- Frontal fibrosing alopecia was first described by Kossard in ther form of scarring or nonscarring alopecia was made were 1994 in six postmenopausal women.1 One of the authors of excluded from the study. this paper (A.G.M.) had seen a single case in 19933 but, as far Control subjects were women aged over 40 years with no as is known, no other cases were observed before this time. history of hair loss. They were recruited through advertise- FFA remained rare during the 1990s. However, particularly ments to hospital staff on the hospitals’ intranet and from over the last 10 years, it has become increasingly common, a friends and family of the authors. All FFA and control subjects development that is occurring worldwide. Most published ser- were unrelated. ies have been in Caucasian women in westernized societies but cases have also been reported in women of African ethnic- Questionnaire ity from Africa and the U.S.A.,7,8 and from Japan.9,10 A review published in 2012 reported FFA as the most common form of We devised a questionnaire that enquired about exposure to a primary cicatricial alopecia seen by dermatologists in four of wide range of lifestyle, social and medical factors. Table 1 lists 11 regions of the U.K.11 The recent onset and rising incidence the subsections of the FFA questionnaire. Where relevant the of FFA strongly suggests that an environmental factor is subsections asked about the frequency of use, length of use involved in the aetiology. and the particular products used. The questionnaire sent to the A putative environmental cause for FFA needs to explain: control subjects excluded the section regarding hair loss but was otherwise identical. 1 the time course of the disease incidence: we hypothesize the first exposure probably occurred during the 1970s or 1980s (the onset of hair loss in Kossard’s cases occurred during the 1980s) and that there has been a large increase Table 1 Questionnaire categories in exposure in the last 10–15 years; 2 the predominant distribution of hair loss affecting eye- Demographic details brows and the frontal and temporal hairline; Profession History of hair loss 3 the predominant expression of the disease in post- Distribution of hair loss menopausal women and the occasional presentation in Previous medical history younger women and in men. Menopause history Hormonal medications, including hormone replacement therapy, In addition we expect that all women with FFA will have oral contraceptive pill, intrauterine device been exposed to a putative cause. Other prescribed medication In this study we used a questionnaire approach to investi- Over-the-counter medication gate exposure of subjects with FFA and an age- and sex- Smoking history matched control group to a range of potentially relevant Alcohol history environmental factors. Following the observation that some Facial skin care products (cleanser, toner, moisturizer, patients with FFA also complained of facial irritation we also foundation, sunscreen) Hair care products (shampoo, conditioner, hair spray, hair performed patch testing in a more limited cohort and com- mousse, hair gel) pared the results with those obtained in the overall patch- Hair care practices (perming, colouring, hair styling, heated tested population. styling aids) Hair removal (plucking, shaving, waxing, electrolysis, laser, intense pulsed light) Methods Facial resurfacing (chemical peels, laser resurfacing) The study was approved by the National Research Ethics Ser- Computer screen usage vice South Central – Berkshire Committee.

Patch testing 7%, P < 0Á05). There were no other significant disease associations. A group of 40 patients with FFA were patch tested to the Bri- tish Society for Cutaneous Allergy Standard Series (http:// www.cutaneousallergy.org/downloads/BSCA_baseline_series_ Medication 2014.pdf) plus hydroperoxides of limonene and linalool. Comparison between the FFA and control groups with regard Patch tests were applied to the back using FinnTM chambers on to contraception and hormonal medication showed a similar ScanporTM tape and read at 48 and 96 h. history for use of intrauterine device (37% vs. 38%, respec- tively) and hormone replacement therapy (21% vs. 23%). Statistical analysis A history of oral contraceptive usage was significantly higher in the control group (FFA 65Á7%, control 85%, P < 0Á01) There were no pre-existing data on which to base a power (Fig. 2). Thyroxine was taken more commonly by subjects calculation to determine the number of subjects required. with FFA than controls (16% vs. 6%, P < 0Á05). There were Therefore, a sample size of 100 cases of FFA and a similar no other associations in either group with other prescribed or number of control subjects was selected on an empirical basis. over-the-counter medications. Comparisons between FFA and control groups were analysed using Fisher’s exact test. A probability of less than 5% (P < 0Á05) that differences between the two groups occurred Alcohol and smoking by chance was taken as statistically significant. To correct for There was no difference between FFA and control subjects multiple testing the data were then analysed using the Ben- with regard to alcohol intake or cigarette smoking. jamini–Hochberg test with a false discovery rate set at 0Á05. Unanswered questions were excluded from analysis. Hair care

Results Information on hair care is shown in Table 3 and Figure 3. All subjects in both groups reported using shampoos. One hundred and ﬁve subjects with FFA and 100 control subjects were recruited. FFA and control subjects were well matched with regard to their age (mean: FFA 63Á8 years, con- Frontal trols 59Á4 years) and had almost identical mean age at onset Eyebrows of menopause (FFA 48Á8 years, controls 48Á7 years) Side scalp (Table 2). Back scalp Within the FFA group the mean age of onset of hair loss Top scalp Á – was 54 8 years (range 30 75). Figure 1 shows the distribu- Arms tion of hair loss in the FFA group. The majority (90%) of Legs subjects with FFA had loss of eyebrows in addition to loss of Axillae frontal hairline, and almost half had noticed hair loss from Pubic their limbs, axillae or pubic area. 020406080100 % involved Medical history

Subjects with FFA gave a history of other medical ailments Fig 1. Distribution of hair loss in subjects with frontal fibrosing more commonly than control subjects (64% vs. 40%, alopecia. P < 0Á05). A history of thyroid disease was significantly more common in subjects with FFA than in controls (19% vs. HRT Control Table 2 Demographics FFA IUD Frontal fibrosing alopecia Controls OCP Number of subjects 105 100 Mean age (range, years) 63Á8 (42–79) 59Á4 (44–77) 0 20406080100 Postmenopausal 87% 96% % Mean age at menopause 48Á8 (30–58) 48Á7 (28–61) (range, years) Fig 2. Hormonal and contraceptive history. Subjects giving a past or Á – Onset of hair loss (range, 54 8 (30 75) N/A current history of use. FFA, frontal fibrosing alopecia; HRT, hormone years) replacement therapy; IUD, intrauterine device; OCP, oral contraceptive.

Table 3 History of skin care, hair care and contraceptive practices in FFA and control subjects – frequencies and statistical probabilities. A Shampoo history of facial hair removal and oral contraceptive use was Conditioner significantly less frequent in subjects with FFA than controls. The Hair spray frequencies of shampooing, hair dyeing and traction hairstyles were Hair gel Control also significantly lower in subjects with FFA than controls. The Hair mousse frequency of sunscreen use was significantly greater in subjects with Traction FFA FFA than controls. The frequency of facial moisturizer and foundation Styling aid use was also greater in subjects with FFA than controls but just failed Perm to reach statistical significance Hair dye Facial hair removal Used by (%) P-values Fisher’s exact % of subjects using FFA Controls test B–H test Facial hair removal 30 61 0Á0001 0Á0001 Fig 3. Hair care. Shampoo, hair spray, hair gel, hair mousse, styling Computer screen use 65 85 0Á0001 0Á0001 aid – twice a week or more frequently; conditioner – with every Shampooa 73 92 0Á0004 0Á0024 wash; traction hairstyle – at least once a week; perm, hair dye – at Sunscreena 48 24 0Á0005 0Á0024 least once a year. Hair dyeb 61 82 0Á0011 0Á0042 Oral contraceptivec 66 85 0Á0020 0Á0063 Traction hairstyled 210 0Á0166 0Á0451 Cleanser Foundationa 62 49 0Á0689 NS Control Toner Facial moisturizera 93 85 0Á0706 NS FFA Facial cleansera 60 51 0Á2083 NS Foundation Facial tonera 15 20 0Á4631 NS Moisturizer Facial resurfacing 2 0 0Á4980 NS Hair spraya 34 30 0Á5516 NS Sunscreen Conditionere 54 59 0Á5731 NS Permanent waveb 10 8 0Á6322 NS 0 20406080100 Hair moussea 23 20 0Á7339 NS % of subjects using at least twice a week Intrauterine devicec 21 23 0Á7388 NS a Hair gel 67 0Á7794 NS Fig 4. Facial skin care. Hair styling aida 52 55 0Á7796 NS e Á HRT 37 38 1 0000 NS Facial skin care B–H test, Benjamini–Hochberg test; FFA, frontal fibrosing alope- Information on facial skin care is shown in Table 3 and Fig- cia; HRT, hormone replacement therapy. aTwice a week or more ure 4. The regular use of facial skin care products was defined frequently; bat least once a year; cpast history of use; dat least once a week; ewith every wash. as a frequency between daily to twice weekly for at least 5 years. Facial moisturizers were used more commonly by subjects with FFA (93%) than controls (85%) although the difference just failed to reach statistical significance However, the proportion of subjects with FFA that shampooed (P = 0Á07). Foundations were also used more frequently by their hair at least twice a week was significantly lower than in subjects with FFA (62%) than controls (49%), although again control subjects (77% vs. 92%, P < 0Á001). Fewer subjects the difference was not statistically significant (P = 0Á07). with FFA also dyed their hair compared with controls (64% There was no difference between the two groups for the regu- vs. 82%, P < 0Á01). The use of conditioners (defined as used lar use of cleansers (FFA 60%; controls 51%) and toners (FFA with each wash) was similar in the two groups (FFA 54%, 15%; controls 20%). control 59%). The use of other hair care products (mousse, The use of dedicated sunscreens in the FFA group was dou- gel, spray) was much lower and similar in the two groups. ble that of the controls, with 48% of the FFA group using A history of permanent waving was low in both groups sunscreens at least twice a week compared with 24% in the (FFA 10%, controls 8%). control group (P < 0Á001). The frequency of potentially damaging hair styling aids (e.g. use of hair straighteners) was similar in the two groups. Other considerations Subjects with FFA reported tight traction-causing hair styles less frequently (2%) than controls (10%) but the frequency Subjects with FFA were significantly more likely to have was low in both groups. retired from work than controls (59% vs. 24%; P < 0Á001). The removal of unwanted facial hair was significantly more A history of using a computer screen was significantly more common in the control group than in subjects with FFA (61% common in control subjects than in subjects with FFA (85% vs. 30%, P < 0Á001). vs. 65%; P < 0Á001).

contraceptive use was significantly greater in the control group Patch testing than among subjects with FFA, suggesting a protective role. Of the 40 patients tested, 21 (52Á5%) had at least one positive The use of shampoos and conditioners was high in both patch test and nine had more than one positive reaction. The subjects with FFA and controls and a component in these majority of positive reactions were to fragrances [Fragrance products cannot be excluded as having a causative role. How- mix 1, n = 4; linalool hydroperoxide, n = 9; Myroxylon pereirae ever, the use of products applied all over the scalp does not (Balsam of Peru), n = 5] and to methylisothiazolinone easily explain the predominant distribution of FFA. Also, the (n = 7). Reactions to Fragrance mix 1 and methylisothiazoli- frequency of shampooing was significantly lower in the sub- none occurred at a similar frequency to that in the overall jects with FFA compared with controls. This may reflect the patch-tested population, all of whom were being investigated common response among patients with hair loss to minimize for dermatitis. Reactions to linalool hydroperoxide and Balsam hair washing due to concern that it might increase hair shed- of Peru were significantly more common in patients with FFA ding. An alternative explanation is that frequent shampooing (Table 4). reduces the risk of developing FFA. Hair colouring was quite common in both groups but the proportion of women using Discussion hair dyes was significantly lower in the subjects with FFA than in controls. Again, this may reflect a concern that hair dye This study would not have been possible 15 years ago. The contributes to hair damage and loss. fact that we and others6,12 are now able to recruit large num- Significantly fewer subjects with FFA reported using a com- bers of patients with FFA with relative ease is illustrative of puter screen compared with controls. This is possibly the rising incidence of this disease, making it difficult to explained by the fact that a higher proportion of controls than escape the conclusion that an external cause is responsible. subjects with FFA were still in work (76% vs. 41%). The latter The results of our questionnaire study are not conclusive is probably due to the high number of control subjects but do provide pointers to possible culprits and rule out recruited from hospital staff. others. In choosing the questions we selected environmental Hair loss in FFA predominantly involves the frontal and factors that would most likely fulfil the requirements outlined frontotemporal hairline extending to in front of the ears and, in the introduction. Key requirements were that all women in most cases, also affects the eyebrows. The significantly with FFA should have been exposed to the putative cause, that lower frequency of facial hair removal by subjects with FFA it should have entered the environment within the last compared with control subjects also suggests involvement of 30 years or so, and that it could explain the predominant dis- facial hair follicles by the disease, as has also been shown by tribution of hair loss. the presence in some patients of facial papules that show Hormonal factors have been suggested as a possible cause lichenoid inflammation involving vellus follicles.3,13 This pat- of FFA. Although the predominance of the disease in post- tern of hair loss accords most closely to the site of application menopausal women may indicate a role of hormonal factors of ‘leave-on’ facial skin care products such as moisturizers and in susceptibility to FFA, there must be something more to foundations. Almost all the subjects with FFA reported using explain its recent emergence and rising incidence – the meno- facial moisturizers regularly and over 60% used foundations. pause and its associated hormonal changes have been around The frequency of moisturizer and foundation use was greater for a very long time! We did consider a possible role for hor- among subjects with FFA compared with controls but the dif- mone-modifying drugs but the results of the questionnaire ference just failed to reach statistical significance. argue against this explanation. Interestingly, a history of oral Could an ingredient or ingredients in leave-on facial products play a role in FFA? The high frequency of positive patch tests could implicate fragrances but fragrances have been pre- Table 4 Patch-test results. The data in the ‘All patch tests’ columns are sent in facial cosmetics for many years and this would not the results from all patients patch tested by the Sheffield Contact explain the apparent recent surge in incidence of FFA. More- Dermatitis Service between 2010 and 2013. The frequencies of over, we do not know whether conventional patch testing positive reactions in the subjects with FFA and this general patch-test aimed at an epidermal reaction is able to identify substances population were compared using Fisher’s exact test that cause an immune reaction in the hair follicle. The patch- test results may reflect a greater use of fragrance-containing FFA products by women with FFA rather than indicating a direct (n = 40) All patch tests causative role. Alternatively, it may point to an increased sus- Allergen +ve % +ve n % +ve P ceptibility to immune-mediated disease, an idea that is sup- Fragrance mix 1 4 10 3541 5Á5NS ported by the increased frequency of thyroid disease in Myroxylon pereirae 512Á5 3502 3Á70Á017 subjects with FFA. An association between LPP and thyroid Linalool hydroperoxide 9 22Á5 1292 9Á80Á016 disease has been reported previously.14 Á Á Methylisothiazolinone 7 17 5 279 14 3NS A clue may reside in the result of the question concerning NS, not significant. sunscreen use. Twice as many women in the FFA group regularly used a sunscreen compared with controls, a difference

British Journal of Dermatology (2016) 175, pp762–767 © 2016 British Association of Dermatologists Frontal fibrosing alopecia, N. Aldoori et al. 767 that was highly significant. Although the use of dedicated sun- References screens by subjects with FFA did not approach universality it is noteworthy that many facial moisturizers now contain sun- 1 Kossard S. Postmenopausal frontal fibrosing alopecia. Scarring alopecia in a pattern distribution. Arch Dermatol 1994; 130:770–4. screen chemicals, a trend that has accelerated in recent years. 2 Kossard S, Lee MS, Wilkinson B. Postmenopausal frontal fibrosing In most cases we were unable to assess whether products used alopecia: a frontal variant of lichen planopilaris. J Am Acad Dermatol by our subjects contained a sunscreen as, in reply to the ques- 1997; 36:59–66. tion about the products used, they often gave a brand name 3 Tan KT, Messenger AG. Frontal fibrosing alopecia: clinical presen- rather than a specific product, and brands may encompass tations and prognosis. Br J Dermatol 2009; 160:75–9. many products with different formulations. Nevertheless, it is 4 Chen W, Kigitsidou E, Prucha H et al. Male frontal fibrosing alope- 55 – likely that a high proportion of women who use facial mois- cia with generalised hair loss. Australas J Dermatol 2014; :e37 9. 5 Salido-Vallejo R, Garnacho-Saucedo G, Moreno-Gimenez JC, turizers are regularly exposed to sunscreens. Allergy to sun- 15 Camacho-Martinez FM. Beard involvement in a man with frontal screen chemicals is well recognized although uncommon. fibrosing alopecia. Indian J Dermatol Venereol Leprol 2014; 80:542–4. Their introduction is relatively recent and the widespread regu- 6Van~o-Galv an S, Molina-Ruiz AM, Serrano-Falcon C et al. Frontal lar exposure of the female population in the form of various fibrosing alopecia: a multicenter review of 355 patients. J Am Acad nondedicated sunscreen products, such as moisturizers, is a Dermatol 2014; 70:670–8. feature of the last 10–15 years, a trend that parallels the rise in 7 Miteva M, Whiting D, Harries M et al. Frontal fibrosing alopecia in 167 – incidence of FFA. It is also worth noting that the first cases of black patients. Br J Dermatol 2012; :208 10. 8 Dlova NC, Jordaan HF, Skenjane A et al. Frontal fibrosing alopecia: FFA were reported from Australia where the ‘Slip-Slop-Slap’ a clinical review of 20 black patients from South Africa. Br J Derma- campaign promoting the use of sunscreens to prevent skin can- tol 2013; 169:939–41. cer was launched in 1981. However, we cannot exclude the 9 Inui S, Nakajima T, Shono F, Itami S. Dermoscopic findings in possibility that the higher use of sunscreens in the FFA group frontal fibrosing alopecia: report of four cases. Int J Dermatol 2008; reflects a tendency by these women to ‘look after’ their skin. 47:796–9. The preponderance of FFA in postmenopausal women is 10 Nakamura M, Tokura Y. Expression of Snail1 in the fibrotic dermis not explained by our results. However, it is known that sub- of postmenopausal frontal fibrosing alopecia: possible involvement of an epithelial–mesenchymal transition and a review of the Japa- stances applied to the skin, including ingredients in skin care 162 – 16,17 nese patients. Br J Dermatol 2010; :1152 4. products, can enter the follicular infundibulum. We spec- 11 Griffin LL, Michaelides C, Griffiths CE et al. Primary cicatricial ulate that the flow of sebum contributes to the removal of alopecias: a U.K. survey. Br J Dermatol 2012; 167:694–7. exogenous substances from the infundibulum and that the 12 Holmes S, MacDonald A. Frontal fibrosing alopecia. J Am Acad Der- low sebum production in older women18 leads to their more matol 2014; 71:593–4. prolonged retention in this site. 13 Donati A, Molina L, Doche I et al. Facial papules in frontal fibros- In conclusion, the results of this study in terms of the univer- ing alopecia: evidence of vellus follicle involvement. Arch Dermatol 2011; 147:1424–7. sality of use, the time course and the distribution of hair loss, 14 Atanaskova Mesinkovska N, Brankov N, Piliang M et al. Association raise the possibility of a causative role for leave-on facial skin of lichen planopilaris with thyroid disease: a retrospective case– care products containing sunscreen chemicals in this increas- control study. J Am Acad Dermatol 2014; 70:889–92. ingly common disease. We should, of course, be wary about the 15 Heurung AR, Raju SI, Warshaw EM. Adverse reactions to sunscreen conclusions of association studies of this type and the low level agents: epidemiology, responsible irritants and allergens, clinical of discrimination between subjects with FFA and control groups characteristics, and management. Dermatitis 2014; 25:289–326. with regard to the use of facial products is a significant limita- 16 Rancan F, Vogt A. Getting under the skin: what is the potential of the transfollicular route in drug delivery? Ther Deliv 2014; 5:875– tion. Also, direct evidence is difficult if not impossible to obtain 7. as there is no method for testing that can prove any particular 17 Schwartz JR, Shah R, Krigbaum H et al. New insights on dandruff/ substance causes immune-mediated follicular lichenoid reac- seborrhoeic dermatitis: the role of the scalp follicular infundibu- tions of the type seen in FFA. We anticipate that the regular use lum in effective treatment strategies. Br J Dermatol 2011; 165(Suppl. of leave-on facial skin care products will be much lower in men 2):18–23. than in women and a similar study conducted in men with FFA 18 Pochi PE, Strauss JS, Downing DT. Age-related changes in seba- 73 – may provide a more conclusive result. ceous gland activity. J Invest Dermatol 1979; :108 11.

Acknowledgments

We are also grateful to Dr Ruth Sabroe who provided some of the patch-test data and to all the subjects who participated in the study.