PDF hosted at the Radboud Repository of the Radboud University Nijmegen

The following full text is a publisher's version.

For additional information about this publication click this link. http://hdl.handle.net/2066/145892

Please be advised that this information was generated on 2021-10-04 and may be subject to change. marjonne с.ш. creemers

Clinical and methodological studies in ankylosing spondylitis

Clinical and methodological studies in ankylosing spondylitis

Een wetenschappelijke proeve op het gebied van de Medische Wetenschappen

Proefschrift

ter verkrijging van de graad van doctor aan de Katholieke Universiteit Nijmegen volgens besluit van het College van Decanen in het openbaar te verdedigen op maandag 12 december 1994 des namiddags te 3.30 uur precies

door

Maria Christina Wilhelmina Creemers

geboren 17 september 1963 te Venlo

Drukkerij Benda bv, Nijmegen Promotores: Prof. dr. L.B.A. van de Putte Prof. dr. F.W.J. Gribnau

Co-promotores: Dr. P.L.C.M. van Riel Dr. M.A. van 't Hof

CIP-GEGEVENS KONINKLIJKE BIBLIOTHEEK, DEN HAAG

Clinical and methodological studies in ankylosing Spondylitis / Maria Christina Wilhelmina Creemers. - [S.l. : s.n.] - Ш. Proefschrift Nijmegen. - Met lit.opg. ISBN 90-9007674-3 Trefw.: Spondylitis ankylopoëtica

Cover-design: Fool Service - Bert Thomassen, Roermond

The studies described in this thesis were performed in the department of rheumato­ logy and gastroenterology of the University Hospital Nijmegen, in the department of rheumatology of the St. Maartenskliniek, Nijmegen, and in the department of pharmacology of the Catholic University Nijmegen, the Netherlands.

The investigations were financially supported by S.A. Nycomed Christiaens N.V., Brussels, Belgium and Chiesi SpA, Parma, Italy. Their financial support for the printing of this thesis is gratefully acknowledged. To the memory of my mother Annie Creemers-van de Wijer whose love and kindness never will be forgotten

Contents

Chapter I Introduction 9

Chapter II A 48-week long-term double-blind comparison of ß-cyclodextrin-piroxicam (Вгехіпе*) with naproxen in ankylosing spondylitis 17

Chapter ΠΙ Gastrointestinal toxicity during nonsteroidal anti­ inflammatory drugs in ankylosing spondylitis: increased faecal blood loss, and evidence for gastric adaptation in a 48-week clinical trial 33

Chapter Г Pseudoporphyria due to naproxen - a cluster of three cases 45

Chapter V Interaction under steady state conditions of ß-cyclo­ dextrin-piroxicam with the Hj-receptor antagonists Cimetidine and ranitidine S3

Chapter VI Second-line treatment in seronegative spondylar- thropathies 71

Chapter П Methotrexate in severe ankylosing spondylitis: an open study 91

Chapter Ш A Dutch version of the functional index for anky­ losing spondylitis: development and validation in a long-term study 99

Chapter EX Disease activity in ankylosing spondylitis: selection of a core set of variables and a first step in the development of a disease activity score 111

Chapter X A radiographic scoring system and identification of variables measuring structural damage in ankylo­ sing spondylitis 127

Chapter XI Summary and conclusions 139

Appendix Ion-pair solid-phase extraction of Cimetidine from plasma and subsequent analysis by high-perfor­ mance liquid chromatography 145

Samenvatting en conclusies 155

Dankwoord 161

Curriculum vitae List of abbreviations

AAU acute anterior uveitis ACTH adrenocorticotropic hormone AS ankylosing spondylitis AS-DAS ankylosing spondylitis - disease activity score AUC area under the curve AZA azathioprine ССР cyclophosphamide CD Crohn's disease CLK renal clearance с peak concentration CRP C-reactive protein OFl Dutch functional index EI enthesis index ESR erythrocyte sedementation rate FI functional index Gl gastrointestinal HPLC high performance liquid chromatography IBD inflammatory bowel disease MRT mean residence time MTP methylprednisolone MTX methotrexate NSAÏÏ) nonsteroidal antiinflammatory drug PsA psoriatic arthritis PCT porphyria cutanea tarda RA rheumatoid arthritis ReA reactive arthritis RS Reiter's syndrome

»И plasma half-live

ТПМ time to peak concentration SASP sulfasalazine SSpA seronegative spondylarthropathy UC ulcerative colitis yjF volume of distribution VJF steady state volume of distribution Chapter I

Introduction

Introduction Introduction Ankylosing spondylitis (AS) is a slowly progressive, chronic inflammatory disease, affecting predominantly the sacroiliac joints and the spine1"3. The most striking pathological finding is multiple focal microscopic inflammatory lesions confined to the ligamentous attachement (enthesophathy) with bony erosions and new bone for­ mations4. This may lead to fusion - ankylosis - of the sacoriliac joints and syndes­ mophyte formation in the spine, resulting in limited mobility of the back. Peripheral arthritis is present in 32% of cases5"7, but it is possible that this figure is an underestimate because arthritis of root joints - hips and shoulders - have frequently not been counted. This is remarkable since the latter joints are more frequently involved than other peripheral joints8. Commonly associated extra-articular features are acute anterior uveitis, aortitis, inflammatory bowel disease and oral ulcers1,9"11. Patients with AS carry in approximately 90% of cases the HLA-B27 antigen12. Although the precise etiology is still unclear, it has been suggested that next to this genetically determined host response (HLA-B27)13"21, environmental factors do play a role, such as the possible role for Klebsiella in triggering AS. However results of this study remain controversial22. In addition, a possible role for a gut-derived factor involved might be suggested because of the high frequency of (sub)clinical inflammatory bowel disease23,24, the association of AS with ulcerative colitis and Crohn's disease25, and a frequently raised IgA21·26"2*. To establish the diagnosis AS the Rome2', New York30, and modified New York criteria31 have been developed in the past As stated by the European Spon- dylarthropathy Study Group (ESSG) these criteria are too restricted and new criteria have been proposed32, that will encompass a wider spectrum of disease within the spondylarthropathies. As a result of the restrictions of the current criteria, the final diagnosis of AS only can be established if there is clear evidence of radiological sacroiliitis, resulting in a delay of diagnosis between 4 and 13 3 33 years • . The aims underlying the treatment of AS are the control of inflammation, and related to this, the modification of the course of the disease. At present drug treatment is mainly symptomatic, using non-steroidal anti-inflammatory drugs (NSAIDs)34;î5. Additionally, physiotherapy is applied often intermittently36, and eventually ergotherapy is often helpfull. NSAIDs reduce pain and stiffness and decrease inflammation. However, these drugs commonly cause gastro-intestinal (Gl) side effects with a wide variety in symptoms and severity37"41. Because of this, a large number of these drugs are now availabe, and still extensive efforts are being made to improve commonly used NSAIDs. ß-cyclodextrin-piroxicam (Brexine) is the complexed form of piroxicam with ß- cyclodextrin, a cyclic oligosaccharide of 7a (1-4) linked D-glucopyranose units42, ß-cyclodextrin-piroxicam is absorbed faster and peak concentration is higher without alteration of the area under the curve43. This might lead to potentially less mucosal damage, because exposure of the GI mucosa to the drug will be reduced. All literature available on ß-cyclodextrin-piroxicam has recently been reviewed44,

11 Chapter I including several studies in which improved GI toxicity of ß-cyclodextrin-piroxicam has been reported45-48. In this thesis ß-cyclodextrin-piroxicam was compared with naproxen over the long term in patients with ankylosing spondylitis. Efficacy of these two drugs is described in chapter 2. GI toxicity was evaluated by evaluation of the prevalence of GI symptoms, next to assessment of the degree of mucosal damage by measure­ ment of faecal blood loss (FBL), using the HemoQuant Assay49 (chapter 3). Additionally pseudoporphyria, a supposed rare side effect due to naproxen, occured in two patients of this clinical trial. Literature concerning pseudoporphyria due to naproxen was reviewed and the cases were described in chapter 4. Directly related to the previously mentioned GI toxicity of NSAIDs is the possible interaction with peptic ulcer treatments, such as H2-receptor antagonists. Cimetidine and ranitidine are two widely used Hj-receptor antagonists, which are commonly co-administered with NSAIDs in order to heal NSAID-induced peptic ulcers. H2-receptor antagonists might influence pharmacokinetics of other drugs, because H2-receptor antagonists are known to: 1. diminish gastric acid secretion, leading to an increase of gastric pH51, thus possibly influencing absorption; 2. inhibit the metabolic pathway of hydroxylation" by binding to cytochrome P^, leading to inhibition of metabolite formation of drugs; and 3. be excreted by glomerular filtration and tubular excretion, which might be influenced by other drugs such as NSAIDs. We studied the possible interaction between ß-cyclodextrin- piroxicam and two H2-receptor antagonists, Cimetidine and ranitidine; both the H2- receptor antagonists and ß-cyclodextrin-piroxicam were studied during steady state conditions (chapter 5). The method for solid-phase extraction of Cimetidine from plasma with subsequent analysis by high performance liquid chromatography was improved for this pharmacokinetic interaction study (appendix). Recent years have witnessed a growing interest in the use of second-line drugs as a treatment for seronegative spondylarthropathies, especially in AS. All the literature was reviewed (chapter 6). Methotrexate appeared promising, and in chapter 7 we report an open study with methotrexate in patients with severe AS. To assess functional status in chronic rheumatic diseases questionnaires are commonly used52"54. In this study we translated, modified and validated a question­ naire for the assessment of functional status in AS, in essence a French question­ naire55,56 (Chapter 8). This newly evaluated and validated Dutch Functional Index was applied to the clinical studies, presented in this thesis. One major problem in As is the difficulty in assessing disease activity in axial arthritis. In axial arthritis, in particular, most clinical variables are not very sensitive to change8,57,58. Laboratory variables are frequently not abnormal or only seem to reflect peripheral arthritis1,6,7. Besides this, disease activity can fluctuate largely: Goodacre et al. for example found almost individual patterns of disease activity in AS with exacerbations and remissions occurring at least once during a one-year period60. Data of the two clinical studies - ß-cyclodextrin-piroxicam versus naproxen in AS and methotrexate in severe AS - were used to select variables assessing disease activity and a first step was made for the development of a disease activity score (chapter 9).

12 Introduction

Apart from the assessment of disease activity, attention was paid to the assessment of outcome of this disease and a radiographic scoring system was developed and evaluated. A part of this radiographic scoring system was the Larsen hip score59 and a modification of the lumbar spine score as published by Taylor et al60. Radiographs of patients participating in the NSAID study were scored with this new radiographic scoring system. Attention was paid to differences in radiogra­ phic scores between ß-cyclodextrm-piroxicam and naproxen, and responders and non-responders. Regarding differences between men and women conflicting results have been published61'64. The relation of commonly used variables with this radiographic scoring system was studied additionally (chapter 10). The main conclusions of this thesis are summarized in chapter 11.

References

1. Moll JMH. Ankylosing spondylitis. Churchill Livingstone, Edingburgh. 1980 2. Carette S. The natural disease course of ankylosing spondylitis. Arthritis Rheum 1983,26*186-90 3. Mau W, Zeidler H, Mau R et al. Clinical features and prognosis of patients with possible ankylo­ sing spondylitis. Results of a 10-year follow-up. J Rheumatol 1988;15:1109-144. 4. Ball J. Enthesopathy of rheumatoid arthritis and ankylosing spondylitis. Ann Rheum Dis 1971;30* 213-23 5. Resnick D. Patterns of peripheral joint disease in ankylosing spondylitis. Radiology 1974; 110:523- 32 6. Gmsburg WW, Cohen MD. Peripheral arthritis in ankylosing spondylitis. Mayo Clin Proc 1983*58*593-6 7. Landewe RBM, Ooei The HS. Ankylosing spondylitis and peripheral joint disease. Clin Rheumatol 1989:8-887-90 8. Laurent R. Are there any antirheumatic drugs that modify the course of ankylosing spondylitis? In: N. Bellamy (ed.) Balliere's clinical rheumatology. Balliere Tindall volume 4, 1990, pp. 387-400 9. Calm A. Spondylarthropathies. 1984. Grane & Stratton, Inc. Orlando, Florida 10. Wilkinson M, Bywaters EGL. Cluneal features and course of ankylosing spondylitis. As seen m a follow-up of 222 hospital referred cases. Ann Rheum Dis 1958;17:209-28 11. Kinsella TD, MacDonald FR, Johnson LG. Ankylosing spondylitis: a late re-evaluation of 92 cases. Can Med Assoc 1966,95:1-9 12. Hcliwcll PS, Wright V. Ankylosing spondylitis. In: Bellamy N. Prognosis in the rheumatic diseases. Kluwer Academic Publishers 1991, pp 133-52 13. Brewerton DA, Hart FD, Nicholls A et al. Ankylosing spondylitis and HLA-B27. Lancet 1973,1: 904-7 14. Schlosstcin L, Terasaki PL В lust one R et al. High association of an HLA antigen, W 27, with ankylosing spondylitis. N Engl J Med 1973;288:704-6 15. Aheam JM, Hochberg MC. Epidemiology and genetics of ankylosing spondylitis. J Rheumatol 1988;15(suppl 16)·22-8 16. Aho K. Yersinia mfections and Yersinia arthritis m Europe. In* Espinoza L. Goldenberg D, Amett F, Alarcon G (eds): Infections m the rheumatic diseases. Orlando. Grune & Stratton. 1988, 267-72 17. Amor B. Suspected infectious agent and host environment interactions m spondyloarthropathies. Clin Exp Rheumatol 1987;5(suppl 1): 19-24 18. Khan MA, Yuc CC. Molecular basis for the HLA association of rheumatic diseases. Autoimmunity Forum Rheumatol 1989;1:2-4 19. Woodrow JC. Genetics of the spondyloarthropathies. Balherc's Clin Rheumatol 1988;2*602-22

13 Chapter I

20. Ebnnger A. The relationship between Klebsiella infections and ankylosing spondylitis Balliere's Qui Rheumatol 1989,3 312-38 21. Calgunen M, Swinburne L, Shmebaum R, Cokoke Em, Wright V. Secretory IgA- immune defence pattern m ankylosing spondylitis and klebsiella. Ann Rheum Dis 1981,40-600-4 22. Saag MS, Bennett JC. The infectious etiology of chronic rheumatic diseases. Seminars Arthr Rhcum 1987,17 1-23 23. Mielants H, Veys EM, Cuvclier C, Os de M. Subclinical involvement of the gut in undifferentiated spondyloarthropathies. Clin Exp Rheumatol 1989;7:499-304 24. Mielants H, Veys EM. The gut m the spondyloarthropathies. J Rheumatol 1990,17-7-10 25. Amett F. Seronegative spondylarthropathies. Bull Rheum Dis 1987,37-1-12 26 Franssen MJAM, Putte van de LBA, Onbnau FWJ. IgA serum levels and disease activity in ankylosing spondylitis a prospective study. Ann Rheum Dis 1983,44-766-71 27. Collado A, Sanmarti R, Brancós MA et al. Immunoglobulin A and С reactive protein levels m ankylosing spondylitis. Ann Rheum Dis 1987,46-719-20 28. Struthers OR, Lewin Г , Stanworth DR. Iga-ot, antitrypsin complexes in ankylosing spondylitis. Ann Rheum Dis 1989,48 30-4 29. Kellgren JH, Jeffrey MR, Ball J. The epidemiology of chronic rheumatism. Oxford- Blackwell, 1963,1-326-7 30. Moll JMH, Wnght V. New York clinical entena for ankylosing spondylitis. Ann Rheum Dis 1973,32 354-63 31. Lmden van der S, Valkenburg HA, Cats A. Evaluation of diagnostic entena for ankylosing spondylitis a proposal for modification of the New York entena. Arthritis Rheum 1987,5 167-71 32. Dougados M, Linden van der S, Juhlin R et al. The European Spondylarthropathy Study Group preliminary entena for the classification of spondylarthropathy. Arthntis Rheum 1991,34-1218-27 33. Gran JT, 0stcnsen M, Husby G. A clinical companion between males and females with ankylosing spondylitis. J Rheumatol 1985,12-126-9 34. Gran JT. Treatment of ankylosing spondylitis. Drugs 1992,44-586-603 35. Calm A, Elswood J. A prospective nationwide cross-sectional study of NSAID usage in 1331 patients with ankylosing spondylitis. J Rheumatol 1990,17:801-3 36. Hidding A, Van der Lmden S, de Witte L. Therapeutic effects of individual physical therapy m ankylosing spondylitis related to duration of disease. Qui Rheum 1993,12-334-40 37. Semble ES, Wu WC. Antiinflammatory drugs and gastnc mucosal damage. Seminars Arthr Rheum 1987,16 271-86 38. Soil AH, Weinstein WM, Kurata J, McCarthy D. Nonsteroidal antiinflammatory drugs and peptic ulcer disease. Ann Int Med 1991,114-307-19 39. Bjamasson L Williams G, So A et al. Intestinal permeability and inflammation m rheumatoid arhtntis, effects of nonsteroidal antiinflammatory drugs. Lancet 1984,2 1171-4 40. Hochberg MC. Association of nonsteroidal antiinflammatory drugs with upper gastrointestinal disease- epidemiologic and economic considerations. J Rheumatol 192,19 (suppl 36)-63-7 41. Gibson OR, Wnitacre EB, Ricotti CA. Cohtis-induccd by nonsteroidal anti-inflammatory drugs. Report of four cases and review of the literature Arch Intern Med 1992,152-625-32 42. Acerbi D, Bovis G, Carli F et al. Biopharmaceutical optimisation of B-cyclodextnn inclusion compounds. Drug Investigation 1990,2(suppl 4)-29-36 43. Acerbi D, Bonati C, Boscarmo G et al. Pharmacokinetic study on Piroxicam at the steady state m elderly subjects and younger adults after administration of piroxicam-ß-cyclodextnn. Int J Clin Pharm Res 1988,8-175-80 44. Riel van PLCM, Crecmers MCW. ß-cyclodextnn-piroxicam (Brcxine). Pharmac Weekblad 1993,128-1442-4 45. Patota L, Clausi G, Farroru F et al. Comparison of faecal blood loss, upper gastrointestinal mucosal integrity and symptoms after piroxicam-B-cyclodextnn, piroxicam and placebo administration. Eur J Clm Pharm 1989.36 599-604 46. Ambanelli U, Nervetti A, Colombo В et al. Piroxicam-ß-cyclodextnn m the treatment of rheumatic diseases: a prospective study. Curr Ther Res 1990-.48-58-68

14 Introduction

47. Santucci L, Fioriteci S, Patoia L et al. Gastric tolerance of piroxicam-B-cyclodextrin compared with placebo and with other NSAIDs: an endoscopic and functional study by evaluation of transmucosal potential difference. Drug Investigation 1990;2(suppl 4):56-60 48. Warrington S, Debbas N, Farthing M, Horton M, Umile A. Piroxicam-B-cyclodextrin: effects on gastrointestinal blood loss and gastric mucosal appearance in healthy men. Int J Tiss Reac 1991:13:243-8 49. Berg van den TWO, Edixhoven-Bosdijk A, Koole-Lesuis R, Wilson JHP. Faecal haem assay - some practical modifications of the haemoquant assay for haemoglobin in faeces. Clin Chim Acta 1987;169:319-22 50. Feldman M, Burton ME. Histamine^-receptor antagonists. Standard therapy for acid-peptic disease (first of two parts). N Engl J Med 1990:323:1672-80 51. Bast A, Smid J, Timmerman H. The effects of Cimetidine, ranitidine and famotidine on rat hepatic microsomal cytochrome P-450 activities. Agents Actions 1989;27:188-91 52. Meenan RF, Gertman PM, Mason JH. Measuring health status in arthritis: The Arthritis Impact Measurement Scales. Arthritis Rheum 1980;23:146-52 53. Fries JF, Spitz PW, Kraines RG. Measurement of patient outcome in arthritis. Arthritis Rheum 1980;23:137-45 54. Pincus T, Summey JA, Soraci SA, Walkton KA, Hummon NP. Assessment of patients satisfaction in activities of daily living using a modified Stanford Health Assessment Questionnaire. Arthritis Rheum 1985;26:1346-53 55. Dougados M, Gueguen A, Nakache J-P et al. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1988;15:302-7 56. Dougados M, Gueguen A, Nakache J-P et al. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1990;17:1254-5 57. Rigby AS, Silman AJ. Outcome assessment in clinical trials of ankylosing spondylitis. Br J Rheum 1991;30:321-5 58. Goodacre JA, Mander M, Carson Dick W. Patients with ankylosing spondylitis show individual patterns of variation in disease activity. Br J Rheum 1991;30:336-8 59. Larsen A, Dale K, Eck M. Radiographic evaluation of rheumatoid arthritis and related conditions by standard reference films. Acta Radiol 1977;18:481-91 60. Taylor HG, Wardle T, Beswick EJ, Dawes PT. The relationship of clinical and laboratory measurements to radiological change in ankylosing spondylitis. Br J Rheum 1991:30:330-5 61. Braunstein EM, Martel W, Moidel R. Ankylosing spondylitis in men and women: a clinical and radiographic comparison. Radiology 1982;144:91-4 62. Resnick D, Dwosh IL, Goergen TG et al. Clinical and radiographic abnormalities in ankylosing spondylitis: a comparison in men and women. Radiology 1976;119:293-7 63. Eustace S, Coghlan RJ, McCarthy C. Ankylosing spondylitis. A comparison of clinical and radiographic features in men and women. Irish Med J 1993;86:120-2 64. Jimenes-Balderas FJ, Mintz G. Ankylosing spondylitis: clinical course in women and men. J Rheumatol 1993;20:2069-72

15

Chapter Π

A 48-week long-term double-blind comparison of ß-cyclodextrin-piroxicam (Brexine®) with naproxen in ankylosing spondylitis

MCW Creemers, MA van 't Hof, MJAM Franssen, FWJ Gribnau, LB A van de Putte, PLCM van Riel

Submitted for publication

ß-cyclodextrin-piroxicam ν naproxen in AS

Summary

Nonsteroidal antiinflammatory drugs (NSAIDs) are used as a symptomatic treat­ ment in the majority of patients with ankylosing spondylitis (AS). Complexation of Piroxicam with ß-cyclodextrin (Brexine*) leads to improved absorptionkinetics, thus reducing gastrointestinal (GI) toxicity. In this study 48-week double-blind study efficacy and toxicity of ß-cyclodextrin-piroxicam (20 mg OD.) with naproxen (500 mg B.ID.) was compared in 59 AS patients, entering after a washout period. Patients were seen 12 times. Comparison over the whole study period showed no significant differences, although eight out of 109 tests favoured ß-cyclodextrin- piroxicam. The frequency and the profile of GÌ side effects were equal. 21 Patients did not complete the study of whom 11 because of inefficacy (7 naproxen, 4 ß- cyclodextrin-piroxicam), and overall efficacy (completion ν drop-out) was not statistically different. In conclusion, ß-cyclodextrin-piroxicam and naproxen had similar efficacy and toxicity over the long-term.

Introduction

Ankylosing spondylitis (AS) is a chronic inflammatory condition, in which sacroiliitis is the hallmark lesion, and pain and stiffness are the main symptoms. Involvement of the spine may lead to spinal ankylosis. The precise pathogenesis is still unclear and although involvement of gram negative bacterial infections strongly have been suggested1"3, no causal treatment is available. Therefore chronical intermittent nonsteroidal antiinflammatory drugs (NSAIDs) are used as a symptomatic treatment in the majority of patients. In the United Kingdom for example it was estimated that about 80% of AS patients are taking NSAIDs4. Because of a large interindividual variability in efficacy and toxicity, a number of different NSAIDs are needed for the treatment of AS. Gastrointestinal (GI) side effects are commonest, and to improve GI tolerance extensive efforts have been made, such as altering the route or the form of administration. Despite the chronic nature of AS, most clinical trials with NSAIDs published had a study duration of 2 to 12 weeks5, and it ought to be advocated that such studies should have a much longer study duration in order to pass judgement on overall efficacy*. Piroxicam is a widely used potent NSAID. Its half-life is on average 45 hours7, thus allowing once daily administration, ß-cyclodextrin is a soluble oligosaccharide obtained from the common starch by enzymic hydrolysis. It is able to complex with drugs by encapsulation and acts as an inert carrier8, ß-cyclodextrin-piroxicam (Brexine") is the complexed form of piroxicam with ß-cyclodextrin, and compared with the parent compound it has an about SO times increased solubility rate. As a consequence its peak concentration is higher and achieved faster910. Apart from the systemic inhibition of prostaglandin synthesis by NSAIDs, GI damage is related to

19 Chapter II the contact-time of the drug with the Gì mucosa. A reduction of this contact-time, as obtained by the complexation with ß-cyclodextrin, might result in less mucosal damage. This has been repotted in several studies with ß-cyclodextrin-piroxicam in which GI mucosal damage was assessed either by endoscopy or by measurement of faecal blood loss, although not all studies showed statistically significant differen­ ces11·13. The aim of the study we describe was to compare the efficacy and toxicity of ß-cyclodextrin-piroxicam with naproxen over the long-term in patients with AS.

Patients and methods

The study was conducted at the departments of rheumatology of the University Hospital Nijmegen St. Radboud and of the St. Maartenshospital Nijmegen, the Netherlands. Full approval of both ethical committees was obtained. Patients Patients with AS fullfilling the modified New York criteria15 of either sex, aged between 16 and 60 years, were selected. They were excluded from the study if they had any evidence of: 1. previous hypersensitivity to NSAIDs; 2. other major side effects on NSAIDs like GI hemorrhage or perforation; 3. active peptic ulcer disease in the previous years; 4. serious comorbidity or impaired organ function; 5. depres­ sion or other mental disorders; 6. alcohol or drug abuse; 7. pregnancy or breast feeding. In addition to the above criteria, only those patients with an active disease after a washout period for NSAIDs were admitted to the study. Active disease was defined as a clear need for analgesics or NSAIDs with at least two of the following features: (a) back pain during the night and/or the day; (b) morning stiffness of at least 15 minutes; (c) pain and stiffness in both buttocks; (d) pain and stiffness of the chest. After written informed consent 64 patients started a washout period. Of them 59 deteriorated and were allowed to enter the trial. Study design The trial was designed as a 48-week double-blind single-observer study in which ß- cyclodextrin-piroxicam at a dose of 20 mg O.D. (S.A. Nycomed Christiaens N.V., Belgium) was compared with naproxen at a dose of 500 mg B.I.D (S.A. Nycomed Christiaens N.V., Belgium). The double-dummy technique was used for blinding. Compliance was checked and stimulated at every visit by pill-counting. Patients were sequentially assigned to one of the treatment groups, balanced16 for age, sex, peripheral arthritis, and extent and duration of the disease. Concomitant treatment for other chronic and acute diseases was allowed, except for other NSAIDs, analgesics, anticoagulants, barbiturates, oral steroids and drugs interfering with absorption. Physical therapy was continued if started prior to the washout period.

20 ß-cydodextrin-piroxicam ν naproxen in AS

Participation in an AS-exercise group, exercises at home and swimming were allowed at any time of the study. Patients were interviewed and examined at a prestudy screening, during which also routine laboratory tests, an ECG and chest X-ray were made. After the washout period, of which the duration was depending on the patients deterioration, the trial medication was started. Assessments were made at start of the medication (week 0) and at weeks 2, 4, 8, 12, 18, 24, 30, 36, 42 and 48, generally at the same time of the day.

Measurements Every visit the following variables were assessed: a) Objective clinical variables: * occiput-wall-distance, measured in cm with the patient standing as erect as possible with heels and back against the wall; * chest expansion, assessed as the difference in cm between the circumference of the chest at nipple line on full inspiration and full expiration. The better one of two measurements was recorded; + Schober 10 cm test"; * fingertip-to-floor-distance, recorded as the distance in cm between the third fìnger and the floor with the patient bending forward maximally, without flexing the knees; * lumbar lateral flexion, measured as a percentage of the body height18; * the number of swollen joints (n=53); * enthesis index: pain graded as 0 = no pain; 1 = pain reported by the patient after request; 2 = spontaneous pain reported by the patient; 3 = severe pain leading to wincing or withdrawing. Enthesis examined are nuchal crests, manubriostemal joint, costochondral joints, greater tuberosity, medial and lateral epicondyl, medial and lateral condyles of the femur, calcaneal insertions of the plantar fascia and the achules tendons, the sacroiliac joints, the cervical, thoracic and lumbar spinous processes, the ischial tuberosities and the anterior posterior iliac spines. The maximal score of the enthesis index being 9019; * root joint index: pain of shoulders and hips by palpation and/or passive move­ ment was assessed with pain graded from 0 to 3 (see enthesis index). The sum of the score being the index; * mobility of the cervical spine, measured in degrees; rotation measured with a goniometer and lateralflexion, anteflexion and retroflexion with a hydrogoniome- ter. b) Subjective patients variables: * Spinal pain during the day, spinal pain during the night and general well-being were assessed by using Visual Analogue Scales (VAS). VAS of 100 mm were used in which 0 mm corresponds with no pain or very good and 100 mm with intolerable pain or very poor;

21 Chapter II

* Duration of morning stiffness, recorded in minutes after arising with a maximum of 360 minutes. c) Laboratory variables: * HLA-B27; * Erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), complete blood cell count, serum Creatinin, liverfunction tests, immunoglobin A, G and M, Creatinin Phosphokinase; * Urinanalysis. d) Functional assessment: A Dutch Functional Index for AS20 in essence a modification of the functional index of Dougados et al21,21 was completed at week 0 and every 12 weeks thereafter. e) Side effects: Side effects or other discomfort related to the study medication were obtained at every visit by direct questioning or as spontaneously reported complaints.

Statistical analysis

Patients' characteristics of both treatment groups being age, sex, duration and extent of the disease, duration of the washout period were compared using a t-test or chi-square test. Assessments at screening visit and at week 0 were used to test comparability of both groups. If necessary, logaritmic or square root transformation was applied, to obtain normality. The flare in disease activity, induced by the washout period, was evaluated in a within-group comparison, using paired t-tests between the prestudy screening visit and week 0. For a within-group comparison of improvement due to treatment paired t-tests were applied to weeks 0 and 2, thus analysing short-term efficacy of each treatment separately. These within-group comparisons between the prestudy screening visit and week 0, and between weeks 0 and 2 indicates the sensitivity to change of assessed variables. Response to treatment was analysed in two ways: 1. The patients situation after 2, 12, 24, 36 and 48 weeks of actual treatment between the two groups was described and analysed applying t-tests; 2. Out of the variables which changed significantly from week 0 to week 2 in the within-group comparison, subjective, clinical and laboratory variables were selected (fingertip-to-floor distance, chest expansion, Schober 10 cm test, enthesis index, spinal pain during the night, general well-being, morning stiffness, CRP and Dutch Functional Index). The size of the changes due to treatment, calculated as the differences between weeks 0 and 12, was evaluated by t-testing.

22 ß-cyclodextrin-piroxicam ν naproxen in AS

Table 1. Characteristics of patients

naproxen B-cyclodextrin- piroxicam

age (years) 37 ±9 4018 sex ratio male : female 23 : 7 22: 7 disease duration (years) 9±8 11 ±7

number of patients with: HLA-B27 26 29 peripheral arthritis 2 I presence of syndesmophytes on X-rays of lumbar and/or cervical spine 21 21

number of patients with different symptoms of disease activity at week 0: spinal pain 27 27 pain and stiffness in buttocks 7 14 chestpain 28 25 morning stiffness ¿IS minutes 16 8

*: mean ± standard deviation

Overall efficacy, i.e. patients who completed the study ν patients who were withdrawn due to either inefficacy or side effects, was analysed with a chi-square test

Results

Patients No patient had to be excluded because of non-compliance. The randomization procedure resulted in 29 patients in the ß-cyclodextrin-piroxicam-treated group and 30 patients in the naproxen-treated group. Patients' characteristics at the start of the study are shown in Table 1. The number of symptoms for active disease at study entry were not different (see Table 1). Between the two treatment groups there were no statistically significant differences at the prestudy screening visit, weeks 0 and 2 (see Table 2). The study was completed by 38 patients. For reasons of inefficacy and side effects 21 patients did not complete the study of whom 11 because of inefficacy (7 naproxen, 4 ß-cyclodextrin-piroxicam); one patient discontinued treatment because of the combination of inefficacy and side effects (naproxen).

23 Chapter II

с- о ·* ve ι> en en Ov 00 •* ρ о\ м сп r¡ VO о •ч °i VO VC* rj — —' ^ Ш* —« ЧГ Ч· — ri —* ^- •^ ~* en ci ^- ^- ^ -Η -Η •Н +1 -Η -Η -Η +1 +1 -H •Η -Η +1 -H* +1 Ή E •Ч- сп СП t- rj •* Ov vo я en en °і <1 ·"! *""ì Ρ Ч Ov oo 1 fi ел •^ СП ri -^ vo un' ел ^* О vo ^ ^ ri ri

0Ç г> ri ρ un ρ VC rj »С О СЧ СП Cj un r-* — Cj οό en Ö en en •ν* ri — ri -н —' en en r l - ~ *~ "^ "•*H ^-H Я 8 •Η -Η -Η -Η Ή •« Ή -Η +1 -H +) -Η +1 -Η СЧ VO VO СЧ 8 VC -Ч- Ov t Г; Г-; Ό t 00 f"- ri ri •* •<ƒ t°. СП Ή СП 00* v°ii un°i" ci ve' — r- ri ri

— r> r~ m un 00 О ON S •^ СП ***ì ve Г- η •» 00* rj p — 4* Ö un — r« §a ci en — Г) 1 R Я -Η -Η •Η Ή Ή -Η -Η -Η •H -H •H -H •H -H M -H — un t — η η t M 00 C-; Tt Ov — ρ en rj Η ci •ν* •* ^ СП 28 vi «n n г~ en Ov 00 о Iξ ri en

m ov »Λ vo VO S M en VI Tf СП 05 ri rj (-. ο\ Q -: Ó 't - ••"^ en ri Μ ^Ν t—1 ν* — *° en en s ^ ^н R R -Η -Η •Η -Η •Η +1 •H +1 44 -H -н +• •H -H +1 -H 1 VC 00 VO Π -: "l Cl СП ρ oo -Ч- en vo О VO vo — M 4* 't •Ч· СП "2 ¡¿ VO *rî 28 = я ri en

. : 4f V- vo *en 'θ Г? q rj ο\ oo *νθ ЧГ •4* vo ". я -- en $M r» •* ri *β _* ~- ~- СП СП ~* "^ 28' Ш* ^ .* RS +1 +1 •Η -Η •Μ -Η -H -H +4 -H -H -H +• +1 +1 -H g VC О ·* ο — Tt en -- Г-; СП o r~ Г- — en 00 СІ СП ·* τί •» СП vTÏ 00* IO wn vo Ov Ov Ov » 2 Я ri ci

# ve "—< *•* Ό rj — n un "1 *С Γ-; Ον U") rj VC UH Ο. Ον •ч* S ~ en' rj rj «» 1-1 *-* Г) Tt •" ^- О СП СП _* ^ СП ri Йй -Η -Η -Η -Η •Η -Η -H -H -Η -Η -H -H M -H •H -H 1 о о Ον ve "•Ч "t un eo «n ri r~ n r« n СП СП vo' vo' q- СП СП СП en «n «S ЬгІ Π en i 28 en en 5 »v о rj en Ov r~ Ov О ca VC Г) ve CI r¡ ^ Z* Zi ρ rj СП СП — ri 'S ¿ en m •H -H as f сч in •Н +і +1 -Η -Η -Η -н -и +1 -H •H -H -H +1 CN СП -* Ό — vo о un f". I> ^ Ov C- vTJ сп ri TÍ ^ ТГ СП wn wn «n Γ-* й Я SS 1 "* ¿ ri en

eu г cu Ζ cu Ζ ιχ Ζ a Ζ β. ζ eu Ζ ft. Ζ eu Ζ

*« 8 ¿3 ¿3 с И .S .Й .I3 e 'я я о * ε •e a О. 2 о 'л κ l о a. > * '1 .é- έ •a э и 1 1 о- ,—. οίί ε Ja § I? ε 8 1 л 1 •fi Л öl âl с а!

24 ß-cyclodextrin-piroxicam ν naproxen in AS

•4- ss αο Tf η — Ή -Η -H +1 •H -H •H +1 Ή -H -н -н •H -H О, οο о — <ч Г) σν ο M —ι ON «1 1 «л •л m 5 5? ve r~ η ¿Λ r- r- Я en ris

ι— o\ — NO 90 TI­ 1Ö OÑ — OÑ CS t^ 00 NO ON Г- NO ГЦ «л тг Г« — Г) — Π — •H -H •H •» •Η -Η Ή +1 +1 Ή •Η +1 •Η -Η NO Г- Ο Μ η η Ό fi 4"> S3 η Λ ri ?ì r- Γ-

f*l VC Γ~ TJ- NO Γ- Ο Γ- Г» — Г* -- Ή -Η -H +1 -н -Η •Η -Η •Η -Η +1 -Η -Η -Η Г4 00 3 * vi t^ ¡η 8 35 8$

NO тг VC 00 q οβ ON — τί oô Ο VC Ю ON г« — <ч CS — г) Г- rr +1 -H -Η +1 +1 -н •Η Ή •Η -Η -Η -Η •* ON Ρ-; (Ν 1 г» *t Й9 S3 SR Ρ"» (^ : .* ο — Г» «Ί •ч- NO ο! — ri rñ ON ON — r< ON V- r> οο η ri Г» — •Η +1 -H +1 •Η -Η •Η -Η I іл NO о о m νν ON οό ri -ί 8? 33 ri r-i

: 3 Ο ^ «η r¡ — NO »П Ν ГІ ri г« = s г» — _« ~* •Η -Η •Η -И •Η -Η •H -Н -H -н MI ve οο ТГ О Г) 93 η г^ η ¡я

< . Г- Г) о — fO ON TÍ οό 00 ^ — NO о о О 00 NO m ca |· ri η — г» η Π CS — •Η -Η •H -H -Η -Η +1 -Η -Η Ή •H +4 •H -H ri ri η η Г- ON ι ON NO 3 3 iss RS r*> Tf m тг

ft, Ζ ft. Ζ ft, Ζ ft, Ζ ft, Ζ ft, Ζ ft, Ζ ft, ζ

E -a i e •s р Χ· S о •S i о к о. — 's. I ε Ç5 ^ л 'I ё 13 13 a SP OÙ h Ее 'В-г. 'i-t з g g I* 8 ¿3

25 Chapter II

W II 45 ι^ 00 оо —; m *л .5 · •ч· Χ; 2 ο — ο' ο •Η Ή Ο•Η* -Η^ •Η -Η u- η Μ — 45 S S; fr Ρ i ι/ϊ 45 ο 6 о' о il 8 II % 00 ts 45 Tt Ό 45 45 й·.. ri Ον ~ 2 о о il Jí •Η Ή ^ «л rt •Η -Η •Η Ы 'S3 Г* 00 9 ^ en ο ο> 45 00 1 rt Ь I «i r-í ο ο о' о' Ì Я |, о · ζ I Tt ^ || о Я- °î *5 « ο г- г- 3 5 00 t~ Ö — о о * e < J! •H -H -Η -Η -н -н e - 3 8 00 ^ Οί Οι — •* 00 ON 45 45 ο © d о I 8 І' У С fi i i 00 V OO Ν οο ο\ 45 Г* »I oô "- ο' ο О О lil •Η -Η -H +1 -Η -Η •8 00 о* 00 Г> η 4t оо ОЧ 5 «Ξ «ri oô •β ч. ο' ο' о' о 2 J< ° ¡Ι Я U ι—I ¿s u С u : : Л * δ Г» *η *— li •й 2 <* •H -H •Η +1 }i gï ^! **Ί 45 ·* 00 9Î ó ο :1 ^ 8 •υ ι І 00 Г- V; V¡ η υ » о ri ri S3 S (Ч ^N ^И о о -Η -Η 2* •Η -Η Ή -Η 8 8 «η 45 — 00 οβ σ\ О •"• — ο' tl 2 ¿ ^*•* ^Μ Іlì з υ со II IS И< £^* 00 ~ Ρα ¿"g ri — <•": **! £ g "8-8 1 — ó +1 -Η •Η -Η s л "о u I! 1 г- о 45 — ε οό о! ο* ο •a ¿•ι >

ο- Ζ α. Ζ α. Ζ ^ в

9.g я •α J3 s 9 i-s. S JÍ α •ÜB· ο κ U κ M I'­ hl ! "ι и Ь ll- •H С * ci η .12 e S 8 * 'δ *~. g i I 8 'Я —s 11 3 If •s —« η Ι? £ S- S Ja * Η li ï!

26 ß-cyclodextrin-piroxicam ν naproxen in AS

Response Paired t-tesüng showed significant changes between the piestudy screening visit and week 0 in the majority of assessed variables in both treatment groups, reflec­ ting a flare of the disease (see Table 2). Sensitivity to change was tested by applying paired t-tests on variables of weeks 0 and 2. In both treatment groups significant improvement could not only be seen in subjective variables such as pain, morning stiffness and general well-being, but also in laboratory variables and variables measuring spinal mobility, except for measurements of the mobility of the cervical spine. The Dutch Functional Index, assessing functional status, improved significantly in both groups from week 0 to 12. Table 2 describes means and standard deviations of all assessed variables for each treatment group per measurement occasion. In only a small number of variables significant differences were found between the two treatments. Differen­ ces between weeks 0 and 12 were calculated of a selected group of variables and are shown in Table 3, to which t-tests were applied. C-reactive protein tended to favour ß-cyclodextrin, though not significantly (p < 0.06). Signs of synovitis (n=3) did not improve on study treatment. Side effects Similar frequencies of side effects were seen in both treatment groups (Table 4). The number of patients discontinuing treatment was not different for the two groups irrespective of the nature of the side effect The profile of GI symptoms

Table 3. Size of change due to treatment from week 0 to 12 (median, percentiles 10 and 90)

ß-cyclodextrin-piroxicam naproxen p-value'

plO p50 p90 plO p50 p90

fingertip-to-floordisL (cm) - 12.9 0.0 6.9 -14.0 -3.0 4.2 0.70 chestexpansion (cm) -0.6 1.0 2.0 -1.0 1.0 2.6 0.83 Schober (cm) -0.6 0.5 2.1 -0.6 0.0 1.0 0.14 enthesisindex (0-90) -22.2 -6.0 0.0 -23.2 -7.0 6.4 0.80 spinal pain night (VAS 100mm) -62.0 -23.0 5.0 -58.2 -14.0 10.2 0.42 general well-being (VAS 100 mm) -46.4 -18.0 6.6 -49.4 -16.0 11.4 0.83 morning stiffness (minutes) -324.0 -30.0 0.0 -240.0 -30.0 1.0 0.30 C-reactive protein (mg/1) -42.6 -13.0 3.0 -43.6 -12.0 14.6 0.06 Dutch functional index (0-3) -0.77 -0.15 0.25 -0.93 -0.14 0.33 0.69

*: t-test after transformation (to normality)

27 Chapter II

Table 4. Side effects

Naproxen* ß-cyclodextnn-piroxicam

discontinuation of treatment due to: GI-symptoms 5 3 other symptoms not related to the GI-tract 2 2

GI-symptoms* number of patients reporting QI-symptoms 12 12

pyrosis 5 5 ructus 1 0 abdominal discomfort 3 5 nauseousness 2 1 loss of apetite 0 1 flatulence 1 1 obstipation 2 0 diarrhoea 1 1

peptic ulcer 2 2 proctitis 1 0

Other symptoms not related to GI-tract* patients reporting symptoms 5 2

sleepiness/drowsiness 1 0 sleeplessness 1 0 headache 1 0

toxicodermia 0 1 psuedoporphyna 2 0

generalised edema and weight gain 0 1

- one patient experienced both headache and GI-sidc effects leading to discontinuation of treatment *· more than one symptom can occur in a patient was similar. The occurrence of pseudoporphyna in two naproxen-treated patients was remarkable as this is reported to be a rare side effect23. Central nervous system side effects only occurred in the naproxen-group

Overall efficacy 21 Patients were withdrawn prematurely from the study due to adverse drug reactions and lack of efficacy. The mean number of weeks patients continued in the study was 41 (standard deviation = sd = 14 weeks) and 36 weeks (sd = 16 weeks) for ß-cyclodextnn-piroxicam and naproxen respectively and revealed no statistical­ ly significant difference. Overall efficacy, i.e. patients who completed the study ν

28 ß-cyclodextrin-piroxicam ν naproxen in AS

Table 5. Distribution of completers and non-completeis

completing the study naproxen ß-cyclodextrin- total piroxicam

yes 17 21 38 no 13 8 21 total 30 29 59

chi-square test not significant (p = 0.32, continuity correction) patients who were withdrawn due to either inefficacy or side effects, revealed no differences between the two treatments as analysed using a chi-square test (Table 5).

Discussion

In this 48-week clinical trial a new formulation of piroxicam - ß-cyclodextrin- piroxicam - and the commonly prescribed naproxen were compared in AS. ß- cyclodextrin improves absorptionkinetics of piroxicam8·9 and it is suggested that this causes less mucosal damage, which was confirmed in studies with chronic treat­ ment of NSATDs by an improved GI tolerance10"13. Evaluation of treatment in AS is difficult because of the heterogeneity of the disease, resulting in heterogeneity in patients, like those with elevated acute phase proteins, clear enthesMs, peripheral arthritis and extensive spinal ankylosis. Apart from this, laboratory variables seem to have little relation to disease activity24"28, and therefore do not reflect treatment effects. Variables assessing spinal mobility appear to be a combination of momentary disease activity and irreversible damage caused by the disease, thus influencing sensitivity to change. When comparing treatments, groups of patients should therefore be balanced heterogeneity of the disease. In this study patients were allocated to either treatment, balanced for age, sex, peripheral arthritis, and duration and extent of the disease. As can be seen in Table 1, patients' characteristics and criteria for active disease were similar at the start of the study in both groups. In this study, with most patients having longstan­ ding disease, the majority of variables showed significant deterioration after a washout period. In addition, after two weeks of treatment a significant improve­ ment could be seen. It can be concluded that both treatment groups consisted of patients with active disease who were sensitive to change. Comparison of all variables at each measurement occasion showed no signifi­ cant difference between the treatment groups, although it was striking that eight out of 109 tests favoured ß-cyclodextrin-piroxicam. This number of significant tests might be attributed to coincidence. The size of the change due to treatment on the short-term have been analysed by calculation of the differences between weeks 0

29 Chapter II and 12, thus reducing the influence of the heterogeneity in disease duration, severity and extent of the disease of the studied patients. For this analysis selection of variables was based on significant changes found with paired t-tests between weeks 0 and 2, in order to test only sensitive variables. Overall efficacy and the number of weeks patients participated in the study were similar if drop out was chosen as the main criterium. The frequence and the profile of side effects did not differ between treatments and no differences in GI toxicity could be found in contrast with previous reports of piroxicam29, in which piroxicam compared with other NSATDs caused more GI toxicity. The concluded absence of difference in efficacy might be subject of a type Π error. After the inclusion period of this study, Bellamy et al30,31 published tables with sample sizes per treatment group for commonly used variables with correspon­ ding clinically important differences or deltas. According to these tables there was an optimal sample size for a maximal delta for Schober 10 cm test, chest expansi­ on, night pain and general well-being, and a minimal sample size for functional status in our study. Cervical anteflexion andretroflexion, lumba r lateral flexion, fingeitip-to-floor-distance, occiput-to-wall-distance and morning stiffness were other variables reported in these tables, and those variables required an optimal sample size varying from 91 to 291 patients per treatment group. Thus, in a part of the assessed variables sample size was sufficient to detect significanties. In conclusion, ß-cyclodextrin-piroxicam and naproxen had similar efficacy and toxicity in this study, in which sample sizes per treatment group for five important variables were sufficient.

References

1. Ebiinger RW, Cawdell DR, Dowling P, Ebringer A. Sequential studies in ankylosing spondylitis: association of Klebsiella pneumoniae with active disease. Ann Rheum Dis 1978;37:146-8 2. Shodjai-Moradi F, Ebringer A, Abuljadayel I. IgA antibody response to Klebsiella in ankylosing spondylitis measured by immunoblotting. Ann Rheum Dis 1992;51:233-7 3. O'Mahony S, Anderson N, Nuke G, Ferguson A. Systemic and mucosal antibodies to Klebsiella in patients with ankylosing spondyltis and Crohn's disease. Ann Rheum Dis 1992;51:1296-1300 4. Calin A, Elswood J. A prospective nationwide cross-sectional study of NSATO usage in 1331 patients with ankylosing spondylitis. J Rheumatol 1990;17:801-3 5. Laurent MR, Buchanan WW, Bellamy N. Methods of assessement used in ankylosing spondylitis clinical trials: a review. Br J Rheum 1991;30:326-9 6. Franssen MJAM, Gribnau FWJ, Putte van de LBA. A comparison of diflunisal and phe­ nylbutazone in the treatment of ankylosing spondylitis. Clin Rheumatol 1986;5:210-20 7. Day RO, Graham GG, Williams KM. Pharmacokinetics of nonsteroidal antiinflammatory drugs. Ballières Clin Rheumatol 1988;2:363-93 8. Orienti I, Cavallari С, Zecchi V. Availability of NSATOh beta-cyclodextrin inclusion complexes. Arch Pharm 1989;322:207-11

30 ß-cyclodextrin-piroxicam ν naproxen in AS

9. Acerbi D, Bovis G, Carli F et al. Biopharmaceutical optimisation of ß-cyclodextrin inclusion compounds. Drug Investigation 1990;2 (suppl 4):29-36 10. Acerbi D, Bonati C, Boscarino G et al. Pharmacokinetic study on piroxicam at the steady state in elderly subjects and younger adults after administration of piroxicam-ß- cyclodextrin. Int J Clin Pharm Res 1988;8:175-80 11. Ambanelli U, Nervetti A, Colombo В et al. Piroxicarn-ß-cyclodextrin in the treatment of rheumatic diseases: a prospective study. Curr Ther Res 1990;48:58-68 12. Bonardelli P, Oliani C, Preti PAM, Pellicano Ρ, Quattrocchi G. Efficacy and gastrointes­ tinal tolerability of ß-cyclodextrin-piroxicam and tenoxicam in the treatment of chronic osteoarthritis. Clin Therap 1990;12:547-55 13. Santucci L, Fiorucci S, Patoia L et al. Gastric tolerance of piroxicam-ß-cyclodextrin compared with placebo and with other NSAIDs: an endoscopic and functional study by evaluation of transmucosal potential difference. Drug Investig 1990;2 (suppl 4):56-60 14. Warrington S, Debbas N, Farthing M, Horton M, Umile Α. Piroxicam-ß-cyclodextrin: effects on gastrointestinal blood loss and gastric mucosal appearance in healthy men. Int JTissReac 1991;13:243-8 15. Van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic criteria for ankylosing spondylitis; a proposal for modification of the New York criteria. Arthritis Rheum 1984;27:361-8 16 Pocock SJ. Sequential treatment assignment with balancing for prognostic factors in the controlled clinical trial. Biometrics 1975;36:81-90 17. Schober P. Lendenwirbelsaule und Kreuzschmerzen. Muench Med Wschr 1937;84:336 18. Domján L, Nemes Τ, Bálint GP, Tóth Ζ, Gömör Β. A simple method for measuring lateral flexion of the dorsolumbar spine. J Rheumatol 1990;17:663-5 19. Mander MM, Simpson JM, McLellan A et al. Studies with an enthesis index as a method of clinical assessment in ankylosing spondylitis. Ann Rheum Dis 1987;46:197- 202 20. Creemers MCW, Hof van 't MA, Franssen MJAM et al. A Dutch version of the Functional Index for ankylosing spondylitis: development and validation in a long-term study. Br J Rheum 1994;33:842-6 21. Dougados M, Gueguen A, Nakache J-P et al. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1988;15:302-7 22. Dougados M, Gueguen A, Nakache J-P et al. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1990;17:1254-5 23. Suarez SM, Cohen PR, DeLeo VA. Bullous photosensitivity to naproxen: 'pseudopo- rphyria'. Arthritis Rheum 1990;33:903-8 24. Sheenan NJ, Slavin BM, Donovan MP, Mount JN, Mathews JA. Lack of correlation between clinical disease activity and erythrocyte sedimentation rate, acute phase proteins or protease inhibitors in ankylosing spondylitis. Br J Rheum 1986;25:171-4 25. Dixon JS, Bird HA, Wright V. A comparison of serum biochemistry in ankylosing spondylitis, seronegative and seropositive rheumatoid arthritis. Ann Rheum Dis 1981; 40:404-8 26. Scott DGL Ring EFJ, Bacon PA. Problems in the assessment of disease activity in ankylosing spondylitis. Rheumat Rehab 1981;20:74-80 27. Roberts WN, Larson MG, Liang MH et al. Sensitivity of anthropometric techniques for clinical trials in ankylosing spondylitis. Br J Rheum 1989;28:40-5 28. Laurent MR, Buchanan WW, Bellamy N. Methods of assessment used in ankylosing spondylitis clinical trials: a review. Br J Rheum 1991;30:326-9

31 Chapter II

29. Kaufman DW, Kelly JP, Sheehan JE et al. Nonsteroidal antiinflammatory drug use in relation to major upper gastrointestinal bleeding. Clin Pharmacol Ther 1993;53:485-94 30. Bellamy N, Buchanan WW, Esdaile JM et al. Ankylosing spondylitis antirheumatic drug trials. Π. Tables for calculating sample size for clinical trials. J Rheumatol 1991;18:1709-15 31. Bellamy N, Buchanan WW, Esdaile JM et al. Ankylosing spondylitis antirheumatic drug trials. Ш Setting the delta for clinical trials of antirheumatic drugs - results of a consensus development (delphi) exercise J Rheumatol 1991;18:1716-22

32 Chapter Ш

Gastrointestinal toxicity during nonsteroidal antiinflammatory drugs in ankylosing spondylitis: increased faecal blood loss, and evidence for gastric adaptation in a 48-week clinical trial

MCW Creemers, MA van 't Hof, A Tangerman, MJAM Franssen, FWJ Gribnau, LBA van de Putte, PLCM van Riel

Submitted for publication

Gl toxicity during NSAIDs in AS Summary

Aim of the study was to compare gastrointestinal (GI) toxicity of ß-cyclodextrin- piroxicam and naproxen during a 48-week study in patients with ankylosing spondylitis (AS) using clinical symptoms and faecal blood loss (FBL) as variables. FBL was measured at weeks 0, 4, 24 and 48 using a modified HemoQuant Assay, ß-cyclodextrin-piroxicam and naproxen revealed no differences in FBL and GI- symptoms over 48 weeks. GI-symptoms and FBL were not correlated. FBL increa­ sed from weeks 0 to 4 (p - 0.055) and decreased subsequently to weeks 24 and 48 (p < 0.001), indicating a mechanism of gastric adaptation. FBL was on average 1600 ml per year in AS patients, which is four times the FBL in healthy controls.

Introduction

Nonsteroidal antiinflammatory drugs (NSAIDs) are widely used because of their antiinflammatory and analgesic effects. However, one of their major disadvantages is the high frequency of gastrointestinal (GI) side effects, varying from pyrosis and nausea to peptic ulcer disease, gastric hemorrhage and ulcerations of the small and large intestine1"*. Two independent mechanisms apparently play a role in GI damage by NSAIDs2: first, a systemic effect due to inhibition of prostaglandin synthesis, and second, a direct damaging, local effect5. Recently it has been suggested that it might be possible to bypass the systemic effect of NSAIDs by selective inhibition of cyclooxygenase 2, found to be the predominant cyclooxyge- nase isoform throughout all stages of inflammation6. However, by now, most attention has been paid to influencing the local damaging effects of NSAIDs, for example altering the route or form of administration in order to decrease the local damaging effect of the drug itself. Brexine is the complexed form of piroxicam with ß-cyclodextrin, a cyclic oligosaccharide of 7a (1-4) linked D-glucopyranose units7. The pharmacokinetic profile of ß-cyclodextrin-piroxicam is identical to that of piroxicam except for absorption kinetics, caused by complexation with ß-cyclodextrin. ß-cyclodextrin- piroxicam is absorbed faster and peak concentration is higher without alteration of the area under the curve8. This might lead to less mucosal damage, because exposure of the GI mucosa to the drug will be reduced. This reduced GI toxicity has been reported in several studies with ß-cyclodextrin-piroxicam9"13. Next to the prevalence of GI symptoms, GI toxicity of NSAIDs can be evaluated as the degree of mucosal damage. The latter can be assessed directly by endoscopy1* or indirectly by measurement of faecal blood loss (FBL). Blood loss in stools can be measured quantitatively by measurement of 51Cr labeled erythrocytes or by measurement of fluorescing haem-derived porphyrins, the latter method known as the HemoQuant Assay. Advantages of measurement with the HemoQuant technique is that it is a non-invasive technique, and that the entire GI tract can be evaluated. In particular with respect to this last point it has to be stressed that

35 Chapter III

Table 1. Characteristics of patients at start of study

ß-cyclodextrin- naproxen Piroxicam

number of patients 23 28 male 16 21 HLA-B27 positive 23 24 age (years)* 41 ± 9 39 ± 9 disease duration (years)" 10 ± 8 9 ± 8

*: mean ± standard deviation

NSAID-enteiOpathy not only involves the upper GI tract, but also the lower GI tract4·15·16 However, NSAID-related GI toxicity has been studied in only small numbers of healthy volunteers and scarcely in patients, and the time of drug administration varied from 4 to 28 days17"21. Two studies form an exception by evaluating a larger number of patients (6523 and 249") and by administration of NSAIDs for 6 weeks and 1 year respectively. The gastric mucosa has shown evidence for a mechanism of adaptation to aspirin in studies in animals25,26 and man27,2", and to indomethacin in man29. This mechanism has not been reported in response to other NSAIDs. This adaptation occurred within 4 weeks but could take up to 9 weeks26, for which it is important to perform studies on the longterm. Since many patients with AS use NSAIDs for many years, and in addition, the prevalence of (sub)clinical inflamma­ tory bowel disease has been reported to be increased in spondylarthropamy30·31, it should be stressed that it is important to evaluate longterm effects of these drugs on GI mucosa. The aims were to study and compare GI toxicity on the long-term of two NSAIDs, ß-cyclodexrrin-piroxicam and naproxen, the relation between GI com­ plaints and the amount of FBL, and to study FBL in patients with AS. For these purposes patients with AS were studied in a 48-week clinical trial. GI toxicity was evaluated using clinical symptoms and faecal blood loss (FBL) as variables.

Patients and methods Patients This study was part of a 48-week double-blind randomized clinical trial comparing naproxen with ß-cyclodextrin-piroxicam (Brexine) in patients with AS. The following criteria were used: Inclusion criteria: 1. age 16-60 years; 2. patients with ankylosing spondylitis according to the modified New York criteria32; 3. clear evidence of activity of the disease when not treated with antiinflammatory drugs. Exclusion criteria: 1. previous hypersensitivity to NSAIDs; 2. previous major side

36 Gl toxicity during NSAIDs in AS effects on NSAIDs like GI hemorrhage or perforation; 3. active peptic ulcer disease in the previous years; 4. serious systemic disease or impaired organ function; 5. depression or other mental disorders; 6. alcohol or drug abuse; 7. pregnancy or breast feeding. Additionally for this GI toxicity study: 8. the presence of hemorr­ hoids. After written informed consent, 59 patients were enrolled in the clinical trial, and 51 of them fulfilled the inclusion criteria for study; patients' characteristics are shown in Table 1.

Study design The study was performed at the rheumatology departments of the University Hospital Nijmegen St. Radboud and St Maartenskliniek Nijmegen, the Netherlands with one observer (MC). All patients were seen twelve times during 48 weeks, starting after a prestudy screening visit and a washout period. Study-medication was naproxen 500 mg b.i.d. and ß-cyclodextrin-piroxicam 20 mg o.d. Medication was supplied double-blind using the double-dummy technique for the whole study- period. Between the two study-drugs there was no cross-over in case of discontinu­ ation. Compliance was checked and stimulated by pill-counting at every visit. Patients were allocated to one of the treatment groups balanced for sex, age, severity and extent of the disease". FBL was measured at week 0 (after a washout period serving as a baseline) and at weeks 4, 24 and 48. Patients who had to stop study-medication because of a lack of effect or an adverse drug reaction were treated with other NSAIDs and/or sulfasalazine, and continued in the GI toxicity study despite discontinuation of study treatment (intention to treat analysis). Side effects were reported spontaneously and asked for by the investigator.

Instruction, diet and faeces collection Instructions, which had to be followed three days prior to faeces collection, were given to all patients. In order to prevent self-induced gum-bleeding dental flossing was forbidden and only soft tooth brushes were allowed to use. To rninimize the possibility of gum-bleeding caused by food, hard fruits and fish with bones were forbidden. In case a patient had easily gum-bleeding it was forbidden to brush teeth for this period and the patient was supplied with a mouthwash. Female patients, who were still menstruating, were asked not to collect stools during their period. A diet, in which red meat was forbidden34·3', was prescribed to be held three days prior to faeces collection. Intake of alcohol was not restricted, but advised to use as sparse as possible36. Complete stools were obtained on two consecutive days for 48 hours on four occasions (week 0, 4, 24 and 48). Patients were given plastic containers of 1 litre. Faeces was collected seperate from urine and collected directly into plastic containers to avoid loss of hemoglobin transfer from stools to toilet water36. Containers with faeces were stored at room temperature till they were brought to the hospital. The visit at the hospital always took place within at maximum two days after the stool collection. At this visit blood was taken for determination of

37 Chapter III

Table 2. Median daily faecal blood loss, and percentiles 10 and 90 (ml/24 hours) of the two treatments

ß-cyclodextrin-piroxicam naproxen

week plO p50 p90 η plO p50 p90 η

0 3.2 5.2 11.4 21 1.9 6.2 11.9 25

4 3.9 6.3 13.6 19 2.3 6.0 16.5 28

24 2.3 4.0 9.6 21 2.3 4.4 9.0 22

48 1.8 5.2 9.9 20 1.4 3.9 8.4 17

hemoglobin. Principally all 51 patients collected stools at weeks 0, 4, 24 and 48, whether they were still using the study drug or not.

The HemoQuant Assay Quantitative hemoglobin loss in the faeces was determined using the modified HemoQuant Assay38. Analyses were performed blinded for treatment. In short, the haem was extracted with acidified isopropanol and reduced (removal of the iron ion out of the haem) into fluorescing protoporphyrines, which were measured by means of fluorescence. In the modified procedure (personal communication of the same authors38) reduction of haem was performed at 60° Celsius with 2 ml of a solution of 0.5 g FeS04.7H20 in 9 ml aqua and 6 ml HCl (37%), instead of 2 ml of the oxalic reagent at 100° C, as published previously38. Stools were stored in plastic containers at -20° Celsius until analyses. Before analysis stools were defrosted and whole stools of 48 hours weighted, diluted with two parts of water and homogenized using a domestic blender. One-gram samples were put into 10 ml polyethylene tubes for the determiniation of haem-derived porphyrins. The total amount of haem-derived porphyrins were measured, and not the amount of haem separately. A porphyrin molecule is similar to a haem molecule except that in haem an iron ion is present. The amount of porphyrins equals the amount of haem in a sample. The total blood loss in ml per 24 hours was calculat­ ed from the known dilution factor, the blood hemoglobin concentration of the patient, the total amount of faeces produced and the amount of porphyrin in the faeces. In the usual HemoQuant Assay FBL is calculated as mg Hb per g faeces. Fibers in the diet dilute faecal hemoglobin because of increased bulk and signifi­ cantly decrease the faecal hemoglobin concentration35. FBL in this study was, therefore, calculated in ml blood per 24 hours.

38 Gl toxicity during NSAIDs in AS

Figure 1. Boxwhisker plots of median faecal blood loss of all measurements occasions

ml/24 hr 30

IS

12

9

6 3 Τ τ τ 0 24 48 weeks

Statistical analysis

Median FBL and percentiles 10 and 90 were calculated for each treatment group for each measurement occasion. Due to the observed skewness in the FBL-data logaritmic transformation was applied. A t-test was calculated to compare the two treatments using FBL of weeks 0, 4, 24 and 48. For all measurements together differences of FBL at weeks 0, 4, 24 and 48 were calculated and a paired t-test on these mean differences was done. Correlations were calculated in order to deter­ mine a possible influence of the length of the washout period on the difference in FBL between weeks 0 and 4. To analyse a relation between FBL and occurence of GI complaints, a 3-way ANOVA was calculated with patient, week and occurence of GI complaints as independent variables and FBL as dependent variable.

Results

In total, 184 stools for FBL-measurements were available. Not of every patient FBL could be measured on all four occasions, and several times faeces collection was postponed because of menstruation. No patient experienced such severe gumbleeding that faeces collection had to be put off. Apart from this, two patients were lost-to-follow-up: one patient because of severe complications of ulcerative colitis (belonging to the naproxen-treated group) and one patient because of refusal at week 36 (piroxicam-treated group).

39 Chapter III

Table 3. Gastrointestinal symptoms

ß-cyclodextrin naproxen' Piroxicam*

number of patienta 23 28

reporting OI symptoms 10 (43%) 12 (42%)

symptoms* pyrosis 5 5 ructus 0 1 abdominal discomfort 4 3 nauseousness 1 2 loss of apetite 0 0 flatulence 0 1 obstipation 0 2 diarrhoea 0 1

number of patiens which had to stop treatment due to OI symptoms 3

endoscopy 3 peptic ulcer 2 proctitis 0 no abnormalities 1

X-ray stomach/colon 1 peptic ulcer 0 proctitis 0 no abnormalities 1

': percentage is given as a percentage of all patients per treatment group *: in a patient more than one symptom can occur

There were 81 FBL measurements in the ß-cyclodextrin-piroxicam group and 103 in the naproxen group of patients who did not discontinue study treatment, and these were used for evaluation of GI toxicity of the two treatments. Median FBL with its 10 and 90 percentiles for each treatment are shown in Table 2. A t-test did not reveal any statistically significant differences for the four measurement occasions between ß-cyclodextrin-piroxicam and naproxen. GI symptoms and complaints occurred in both treatment groups. The number of patients which reported GI symptoms were 10 and 12 for ß-cyclodextrin-piroxicam (η = 23) and naproxen (n = 28), respectively (see Table 3). Treatment was stopped due to GI symptoms in 3 (ß-cyclodextrin-piroxicam) patients and 5 (naproxen) respectively. Endoscopic evaluation was performed in six patients: peptic ulcers have been found in 4 patients (2 ß-cyclodextrin-piroxicam, 2 naproxen), proctitis in 1 patient both cases. A 3-way ANOVA did not show any relation between the

40 Gl toxicity during NSAIDs in AS occurence of GI symptoms and the amount of FBL (p > 0.05). All 184 stools for FBL were used to study FBL in patients with AS using chronically NSAIDs, and in addition to study a possible mechanism of gastric adaptation. FBL at week 4 (6.3 ml/24 hours, range 1.9 - 28.3) was increased compared to baseline, week 0 (5.4 ml/24 hours, range 0.3 - 11.6). This increase revealed no statistically significant difference applying paired t-test (p = 0.055). Subsequent FBLs showed a reduction to week 24 (4.1 ml/24 hours, range 0.9 - 10.2) and week 48 (4.9 ml/24 hours, range 1.0 - 12.4), both highly statistically significant applying paired t-tests (p < 0.001). Analyses of FBL at weeks 0, 24 and 48 were not statistically significantly different between the two groups. FBL of all four measurement occasions is shown in figure 1. The difference in FBL between weeks 0 and 4 was not related to the length of the washout period, which was on average 14 days with a range from 3 to 42 days. FBL in patients with AS using NSAIDs chronically was calculated to be on average 1600 ml per year.

Discussion

The prevalence of GI side effects and the quantity of FBL per 24 hours were used to evaluate NSAID-induced GI damage in this 48-week study in patients with AS. Comparison of ß-cyclodextrin-piroxicam with naproxen showed that both treatment groups were not statistically different regarding median daily FBL at all four measurement occasions. Comparison with other studies using these drugs is difficult, because on the one hand correlations between different methods of assessment mucosal damage are moderate and quite variable40"43. In addition, dosages of drugs, study period, time of administration and kind of populations (patients or healthy volunteers) differed largely. Because a relation was found between use of alcohol and FBL44, intake of alcohol was forbidden in most studies with healthy volunteers. In most longterm studies with patients as in this study use of alcohol was allowed, although it was advised to use alcohol as sparsely as possible with at maximum one unit per day. As has been shown in many studies23,43, no correlation was found between GI complaints and mucosal damage measured as daily FBL. The number of peptic ulcers is too small to analyse a possible relation between the amount of FBL and the development of peptic ulcers. Median daily FBL showed a trend to increase from baseline (week 0) to week 4 and a statistically highly significant reduction from week 4 to weeks 24 and 48. This increase at week 4 and subsequent reduction suggests a mechanism of gastric adaptation, which has been observed in animals and humans after aspirin intake24"27, and in indomethacine in man29, and could take up to nine weeks to occur26"28. The mechanism is thought to be based on an increased epithelial cell regeneration and mitoses44. In healthy volunteers FBL, measured using the modified HemoQuant Assay38, is on average 1.1 ml per 24 hours, with an upper limit of 2.7 ml per 24 hours38,

41 Chapter III which means that FBL is on average 400 ml per year. In this study FBL was found to be larger compared to FBL in healthy volunteers38 and estimated to vary on average between 1600 and 1900 ml per year, after gastric adaptation and at baseline (week 0), respectively. By measurement of FBL the entire GI tract can be evaluated, and the increased FBL, even after gastric adaptation, suggests that a larger part of the GI tract is involved than only the gastric and duodenal mucosa. FBL at baseline (week 0) did not show a significant difference with FBL at weeks 24 and 48, besides, no correlation has been found between the amount of FBL and the length of the washout period. This might be explained by the fact that low grade bowel inflammation has been found to be present in about 70% of patients using NSAIDs chronically, and this inflammation might persist for up to 18 months after discontinuation of NSAIDs3. Additionally, the prevalence of (subclinical inflammatory bowel disease has been reported to be increased in spondylarthro- pathy30,31, which might be worsened by use of NSAIDs16,17. Therefore, it seems worthwhile to study second-line drugs in AS, since these drugs might be less haimfull to the GI tract compared to NSAIDs45. Besides, second-line drugs might influence the course of (sub)clinical inflammatory bowel disease. In summary, there seems to be some evidence for gastric adaptation reflected in an increase and subsequent reduction of FBL. Further studies in patients have to be done regarding this mechanism of gastric adaptation. The lack of correlation between GI complaints and the amount of FBL was in accordance with other studies2,23, ß-cyclodextrin-piroxicam and naproxen did not differ in the amount of daily FBL and the frequency of Gl-complaints on the longterm. Finally, FBL in patients with AS is increased compared to FBL in healthy volunteers, probably related to chronic use of NSAIDs and the increased prevalence of inflammatory bowel disease. This once more stresses evaluation of second-line drugs in AS.

Acknowledgements

We thank С Bessems for performing all HemoQuant Assays, A. van Schaijck for analytical assistance and E. Brummelkamp for computational assistance and advice. We are indebted the department of internal medicine Π Rotterdam (the Nether­ lands)38 for giving us the description of the modified HemoQuant Assay.

References

1. Semble ES, Wu WC. Antiinflammatory drugs and gastric mucosal damage. Seminars Arthr Rheum 1987;16:271-86 2. Soll AH, Weinstein WM, Kurata J, McCarthy D. Nonsteroidal antiinflammatory drugs and peptic ulcer disease. Ann Intern Med 1991;114:307-19 3. Bjamasson I, Williams G, So A et al. Intestinal permeability and inflammation in rheumatoid arthritis; effects of nonsteroidal antiinflammatory drugs. Lancet 1984;2:1171-4 4. Gibson GR, Whitacrc EB, Ricotti CA. Colitis-induced by nonsteroidal anti-inflammatory drugs. Report of four cases and review of the literature. Arch Intern Med 1992;152:625-32

42 Gl toxicity during NSAIDs in AS

5. Levine RA, Petokas S, Nandi J, Enthoven D. Effects of nonsteroidal antiinflammatory drugs on gastro-intestinal injury and prostanoid generation in healthy volunteers. Dig Dis Sci 1988;33:660-6 6. Appleton L Tomlinson A, Willoughby DA. Inducible cyclooxygenase (Cox-2): a safer therapeutic target? Br J Rheum 1994;33:410-2 7. Acerbi D, Bovis G, Carli F, Pasini M, Pavesi L, Peveri T. Biopharmaceutìcal optimisation of ß- cyclodextrin inclusion compounds. Drug Investigation 1990;2(suppl 4):29-36 8. Acerbi D, Bonati C, Boscarino G et al. Pharmacokinetic study on piroxicam at the steady state in elderly subjects and younger adults after administration of piroxicam-o-cyclodextrin. Int J Clin Pharm Res 1988;8:175-80 9. Patoia L, Clausi G, Farroni F et al. Comparison of faecal blood loss, upper gastro-intestinal mucosal integrity and symptoms after piroxicam beta-cyclodextrin, piroxicam and placebo administration. Eur J Clin Pharm 1989;36:599-604 10. Ambanelli U, Nervetti A, Colombo В et al. Piroxicam-B-cyclodextrin in the treatment of rheumatic diseases: a prospective study. Curr Ther Res 1990;48:58-68 11. Bonardelli P, Oliani С Preti PAM, Pellicano Ρ, Quattrocchi G. Efficacy and gastro-intestinal tolcrability of beta-cyclodextrin-piroxicam and tenoxicam in the treatment of chronic osteoarthritis. ClinTherap 1990;12:547-55 12. Santucci L, Fiorucci S, Patoia L et al. Gastric tolerance of piroxicam-B-cyclodextrin compared with placebo and with other NSAIDs: an endoscopic and functional study by evaluation of transmucosal potential difference. Drug Investigation 1990;2(suppl 4):56-60 13. Warrington S, Debbas N, Farthing M, Horton M, Umile A. Piroxicam-B-cyclodextrin: effects on gastro-intestinal blood loss and gastric mucosal appearance in healthy men. Int J Tiss Reac 1991;13:243-8 14. Lanza F, Rouer G, Nelson R. An endoscopic evaluation of the effects of nonsteroidal antiinflam­ matory drugs of the gastric mucosa. Gastrointest Endose 1975;21:103-5 15. Schwartz HA. Lower gastro-intestinal side effects of nonsteroidal antiinflammatory drugs. J Rheumatol 1981;8:952-4 16. Ramptom DS. Nonsteroidal antiinflammatory drugs and the lower gastro-intestinal tract Scand J Gastroenterol 1987;22:1-4 17. Hooper JW, Anslow JA, Martin WS, Araujo P, Darke A. Fecal blood loss during isoxicam and piroxicam adminstration for 28 days. Clin Pharmacol Ther 1985;38:533-7 IB. Jallad NS, Sanda M, Salom IL et al. Gastro-intestinal blood loss in arthritic patients receiving chronic dosing with etodolac and piroxicam. Am J Med Sci 1986;292:272-6 19. Bianchine JR, Procter RR, Thomas FB. Piroxicam, aspirin, and gastro-intestinal blood loss. Clin Pharmacol Ther 1982;32:247-52 20. Arnold JD, Salom IL, Berger AE et al. Comparison of gastro-intestinal microbleeding associated with use of etodolac, ibuprofen, indomethacin, and naproxen in normal subjects. Curr Ther Res 1985;37:730-8 21. Loebl DH, Craig RM, Cullic DD et al. Gastro-intestinal blood loss: effect of aspirin, fenoprofen and acetaminophen in rheumatoid arthritis as determined by sequential gastroscopy and radioactive fecal markers. JAMA 1977;237:976-81 22. Bjamasson I, Smethurst P, Hayllar J, Levi AT. NSAJD enteropathy; the main site of chronic blood loss in patients on NSAJD. Gut 1990;31A:1203 23. Caruso I, Bianchi Porro G. Gastroscopie evaluation of antiinflammatory agents. Br Med J 1980; 280:75-8 24. St John DJB, Yeomans ND, McDermott GT, de Boer WGRM. Adaptation of the gastric mucosa to repeated administration of aspirin in the rat. Dig Dis 1973;18:881-6 25. Hurley JW, Crandall LA. The effect of salicylates upon the stomachs of dogs. Gastroenterology 1964;46:36-43 26. Graham DY, Smith JL, Dobbs SM. Gastric adaptation occurs with aspirin administration in man. Dig Dis Sci 1983;28:1-6 27. Smith HL, Spjut HJ, Torres E, Jones RD, Graham DY. Mechanism of gastric mucosal adaptation to aspirin-induced injury in man. Gastroenterology 1984;86:1257

43 Chapter III

28. Graham DY, Smith JL, Spjut Ш, Torres E. Gastric adaptation* studies in humans during continuous aspirin administration. Gastroenterology 1988,95 327-33 29. Shorrock CJ, Rees WDW. Effect of lndomethacin on human gastroduodenal 'mucus-bicarbonate' barrier. Gut 1987,28A1411 30 Grillet В, de Clerck L, Dequeker J, Rutgeerts P, Gebœs K. Systematic lleocolonoscopy and bowel biopsy study in spondylarthropathy. Br J Rheum 1987;26-338-40 31. Mielants H, Veys EM. HLA-B27 related arthritis and bowel inflammation. Part 1. Sulfasalazine (salazopynn) m HLA-B27 reactive arthritis. J Rheumatol 198S;12'287-93 32. Van der Linden S, Valkenburg HA, Cats A Evaluation of diagnostic entena for ankylosing spondylitis. A proposal for modification of the New York entena. Arthntis Rheum 1984,27-361-8 33. Pocock SJ. Allocation of patients to treatment in clinical tnals Biometrics 1979,35*183-97 34 Fcmberg EJ, Sternberg WM, Banks BL, Henry JP. How long to abstain from eating red meat before fecal occult blood tests. Ann Intern Med 1990;113*403-4 35. Schwartz S, Ellefson M. Quantitative fecal recovery of ingested hemoglobin-heme in blood- comparisons by HemoQuant assay with ingested meat and fish. Gastroenterology 1985,89-19-26 36. Fleming JL, Ahlquist DA, McGiIl DB. Aspirin, ethanol and hemoquant Mayo Clin Proc 1987; 62*159-63 37. Ahlquist DA, Schwartz S, Isaacson J, Ellefson M. A stool collection device* the first step m occult blood testing. Ann Intern Med 1988,108*609-12 38. Berg van den JWO , Edixhoven-Bosdijk A, Koole-Lesuis R, Wilson JHP. Faecal haem assay -some practical modifications of the haemoquant assay for haemoglobin in faeces. Clin Chun Acta 1987.169*319-22 39. Slavin JL, Melchor EA, Sundeen M, Schwartz S. Effects of high-fiber diet on fecal blood content (HemoQuant Assay) in healthy subjects. Dig Dis Sci 1991-36*929-32 40. Katz LB, Fernandez JA, Schwartz A, Shnver DA. Rioprostil prevents aspinn-induced fecal blood loss m dogs correlation of Hemoquant (Hb) with endoscopic findings. Digest 1989,42:217-23 41 Lange S, Hench V, Ahlquist D, Malagelada J.-R. The relationship between drug-induced gastroduo­ denal injury and fecal blood loss. Gastroenterology 1986,92-1511 42. Lynch NM, Deacon M, McHutchison JG et al. Gastro-intestinal blood loss from a new buffered aspirin (Ostopnn): measurement by radiochromium and hemoquant techniques. Aust NZ J Med 1989,19*89-96 43. Leahy MBG, Pippard MJ, Salzmann MB et al. Quantitative measurement of feacal blood loss: comparison of radioisotopic and chemical analyses. J Clin Pathol 1991:44*391-4 44. DeSchepper PJ, Tjandramaga ТВ, De Roo M et al. Gastro-inteslinal blood loss after diflunisal and after aspirin· effect of ethanol. Clin Pharmacol Ther 1978:23-669-76 45. Hayllar J, Smith T, MacPherson A et al. Nonstenodal antiinflammatory drug-induced small intestinal inflammation and blood loss. Effect of sulfasalazine and other disease-modifying antirheumatic drugs. Arthntis Reum 1994,37* 1146-50

44 Chapter IV

Pseudoporphyria due to naproxen A cluster of three cases

MCW Creemers, A Chang, M JA M Franssen, TJW Fiselier, PLCM van Riel

Submitted for publication

Pseudoporphyria due to naproxen

Summary

Pseudoporphyria is a photo-induced blistering disorder with increased skin fragility, caused among others by nonsteroidal antiinflammatory drugs (NSAIDs). Lesions heal with scarring and milia. Porphyrin screen studies are normal in this disease. Histology and immunofluorescence resemble porphyria cutanea tarda. In this report we describe a cluster of three cases of naproxen-induced pseudoporphy­ ria, and review briefly previously reported cases induced by naproxen. The majority of reported cases involve children. Physicians should be aware of this reversible skin disorder.

Introduction

Pseudopoiphyria is a blistering disorder with increased skin fragility. It resembles porphyria cutanea tarda (РСГ) in its reaction to the intake of certain drugs and sunlight exposure, but without disorders in the porphyrin metabolism. Only sunlight exposed areas of the skin are involved; face, dorsal surfaces of hands and feet and less often the neck sparing the area underneath the chin and behind the ears. Mostly blisters heal with scarring and milia. Blisters are subepidermal and histolo­ gy is identical to abnormalities found in PCT. PCT patients lack uroporphyrinogen decarboxylase, an enzyme of the haem biosynthesis, by contrast in pseudoporphyria porphyrin screen in erythrocytes, urine and faeces is normal. Pseudoporphyria due to naproxen was believed to be a rare adverse reaction to the drug, although by now a hundred cases have been reported. We report here three cases of pseudoporphyria due to naproxen and review briefly the literature, the establishment of diagnosis and hypothetical mechanisms.

Cases

Case 1 During holidays at the Canary Islands a 42-year-old white woman experienced blistering and skin fragility of the dorsa of both hands, fingers and feet for eight weeks. Her medical history included a moderately active ankylosing spondylitis since three years and in the past a pityriasis rosea. She had been taking an oral contraceptive for several years and naproxen 500 mg b.i.d. for eight months. Before she was treated with other nonsteroidal antiinflammatory drugs (NSAIDs), of which some caused gastrointestinal side effects. There was no history of hepatitis, porphyria or other photosensitivity disorders. Examination of the skin revealed small blisters with slight scarring from old lesions. No hyperpigmentation, hirsutism, milia or atrophic scarring were present. Laboratory examination, repeated

47 Chapter IV twice, revealed normal liverfunction tests and a normal porphyrin screen in erythrocytes, urine and faeces was demonstrated twice. Perilesional skin biopsies showed subepidermal bullae formation. Direct immunofluorescence of the perilesio­ nal skin showed aspecific condensation of IgM and Ç, at the edge of the bulla. IgG, fibrinogen, IgA and Clq were diffusely deposited in the dermal layer. Indirect immunofluorescence revealed no abnormalities. No signs of bullous pemphigoid were present. Naproxen was discontinued, but new lesions occurred for another eight weeks, thereafter blistering ceased with minimal scarring.

Case 2 A 58-year-old white man, who was taking sunbeds regularly for several years, had a six-week history of blistering and skin fragility of the dorsa of both hands. His medical history comprised a longstanding moderately active ankylosing spondylitis and prostatism, without hepatitis, porphyria or photosensitivity disorders. The patient reported that be had experienced the same lesions about five to six years ago but to a lesser extent and degree. In the past he had been treated with other NSAIDs, all discontinued because of inefficacy. On both occasions he was using naproxen 500 mg b.i.d. Examination of the skin showed small blisters and some old lesions with slight scarring. No other abnormalities of the skin were seen. Liverfunction tests, porphyrin screen in erythrocytes and urine revealed normal. Skin biopsies were not conclusive because of a sampling error. He decided to continue naproxen because of good efficacy, despite advise to stop, and started to use a sunfilter when taking sunbeds. Skin lesions healed with scarring within a few weeks. Almost 1.5 years later, no new skin lesions occurred despite naproxen use, but slight skin fragility remained.

Case 3 A nine-year-old white girl with polyarticular systemic-onset juvenile chronic arthritis revealed blistering and skin fragility of the face and dorsa of both hands, aggravated by sunlight exposure. She had no history of hepatitis, porphyria or photosensitivity disorders. She was taking methotrexate 10 mg weekly and na­ proxen 250 mg b.i.d. (20 mg/kg/day). She had not been treated with other NSAIDs. Skin examination showed blistering with crustae and scarring of old lesions. Liverfunction tests were normal. A skin biopsy was not felt to be justifiable in view of the age of the patient and the site of the blistering. After naproxen was changed into diclofenac, no new blistering occurred and lesions healed within several weeks.

Discussion

Though other NSAIDs are known to cause pseudoporphyria as well, the majority of cases published were caused by naproxen1"9. A total of 100 cases of pseudopor-

48 Pseudoporphyria due to naproxen phyria due to naproxen are known, remarkably 74 of these were children. The mean age of children was seven years (range 1-18 years) and of adults age ranged from 21 to 84 years. In adults it appeared that mostly women were involved (14 women versus seven men, five not reported), in most cases involving children sex was not reported. Most cases manifested itself during summer, not always clearly related to sunlight exposure, three cases occurred after visiting a tanning salon. Dose of naproxen in adults was 250 to 1500 mg/day and in children the mean dose was 13 mg/kg/day and the use of naproxen ranged from a few weeks to several years before onset of pseudoporphyria. Most frequently lesions occurred on the dorsal surfaces of hands and feet, in less cases in the face, in a few cases on the neck and in one case also on both knees. Scarring usually remained after healing but showed a tendency to fade. Porphyrin screening was done in the majority of cases and revealed normal. Histology was reported in 16 cases and in 11 of these 19 biopsies immunofluorescence was done, all biopsies were conclusive for PCT- like abnormalities. Naproxen was stopped in all patients after which skin lesions healed. In 14 cases the time needed for healing was described and was on average five weeks (range 1 week to 2 months). All three cases described here manifested itself in the summer of 1992, which was an ordinary Dutch summer, though one case manifested itself at the Canary Islands. Pseudoporphyria due to naproxen was not seen ever before at our clinics. Our two adults took part in a 48-week double-blind study comparing ß-cyclodex- trin-piroxicam (n=30) with naproxen (n=29) in ankylosing spondylitis. One of our cases (case 2) differs from other cases in the literature as skin lesions healed while naproxen was still being administered but this patient had started to use a sunfilter. One case of pseudoporphyria due to nalidix acid behaved similarly10. No porphyrin screen was done in case 3, but the causal relation with discontinuation of naproxen and subsiding of characteristic skin lesions confirm the diagnosis. Although phototoxicity is widely accepted, several hypotheses have been developed. Naproxen is known to stabilize the liposomal membrane in vitro and to be a potent inhibitor of cyclo-oxygenase. In most hypotheses sunlight plays a central role. The inflammation of the sun exposed skin may be due to: a. the chemical structure of NSAJDs itself which may induce photosensitization of the skin 4,n; b. the release of free oxygen radicals by naproxen which may contribute to the phototoxic reaction*·', probably as well aggravated by complement compounds7. Other hypotheses suggest that pseudoporphyria patients have minor disturbances of haem-biosynthesis leading to symptoms in case of excessive sunlight expos­ ure5,12, perhaps sometimes associated with intake of alcohol5, estrogens5 or eventual­ ly minor skin trauma6. This relation with sunlight exposure is confirmed especially in three patients with vitiligo in whom bullae were only present on areas of vitiligo12"14. Not only exposure to a high intensity of sunlight may induce pseudo­ porphyria, but also multiple prolonged low-grad exposure may cause skin lesion resulting mostly in decreased adhesion of dermis and epidermis9,15. It is not easy to explain the larger frequency of pseudoporphyria due to naproxen in children (Allen et al6 even reported that 6% of all children using

49 Chapter IV naproxen experienced pseudoporphyria), especially as far less children than adults use naproxen. Possibly children experience more frequently trauma of the skin during sunlight exposure and are more exposed. Pseudoporphyria might be seen as a 'diagnosis per exclusionem'. РСГ, epidermolysis bullosa acquisita (EBA) and hydroa vacciniforme must first be ruled out as a diagnosis. Clinically and histologically pseudoporphyria are similar, except that there is a negative porphyrin screen in pseudoporphyria. In EBA sites prone to minor trauma, i.e. dorsae of hands, elbows and knees, are most frequently affected, sites which can be involved in pseudoporphyria too. However, histologically in EBA deposition of antibodies in the sublamina densa can be seen on indirect immunofluorescence analysis16. In addition, treatment of pseudoporphyria contains discontinuation of the drug9 in contrast to treatment of EBA, which requires mostly administration of corticosteroids and is often unsatisfactory. Especially in children hydroa vacciniforme must be considered in the differential diagnosis, related to recent vaccination. For establishment of the diagnosis the following diagnostic criteria for pseudo­ porphyria have been published by Suarez et al9: a. clinical lesions consisting of bullae and erosions on sun-exposed skin; b. granular IgG and Ç, on direct immu­ nofluorescence and negative findings on indirect immunofluorescence analysis; these histologic findings resemble those of PCT; and с negative findings on porphyrin studies. It may be concluded that the pathogenesis of pseudoporphyria due to naproxen is still not clear, but there seems to be a relation to sunlight exposure and perhaps minor skin trauma. Discontinuation of naproxen is the best treatment. One should be aware of this adverse drug reaction, especially in fair skinned patients and even more in children.

References

1. Tucker LB, Miller LC, Szet IS, Schaller JG. Pseudoporphyria in children treated with nonsteroidal antiinflammatory agents. Pediatric Research 1992;31:155A 2. Shelley WC, Elpcm DJ, Shelley ED. Naproxen photosensitization demonstrated by challenge. Cutis 1986;39:169-70 3. Rivers JK, Bametson RStC. Naproxen-induced bullous Photodermatitis. Med J Aust 1989;151:167-8 4. Levy ML, Barron KS, Eichenfield A, Honig PJ. Naproxen-induced pseudoporphyria: a distinctive Photodermatitis. J Pediatr 1990;117:660-4 5. Stenberg A. Pseudoporphyria and sunbeds. Acta Derm Venereol 1990;70: 354-6 6. Allen R, Rogers M, Humphrey I. Naproxen induced pseudoporphyria in juvenile chronic arthritis. J Rheumatol 1991;18:893-6 7. Cox NH, Wilkinson DS. Dermatitis artefacta as the presenting feature of auto-erythrocyte sensitization syndrome and naproxen-induced pseudoporphyria in a single patient. Br J Dermatol 1992;126:86-9 8. Southwood TR, Ryder CAJ. The frequency of pseudoporphyria in childeren treated with non­ steroidal anti-inflammatory drugs. In: congress report of 55th annual meeting American College of Rheumatology, abstract cll9, 1991

50 Pseudoporphyria due to naproxen

9. Suarez SM, Cohen PR, DeLeo VA. Bullous photosensitivity to naproxen: 'pseudoporphyria'. Arthritis Rheum 1990;33:903-8 10. Ramsay CA, Obreshkova Б. Photosensitivity from nalidixic acid. Br J Dermatol 1974;91:523-8 11. Kochevar Ш. Phototoxicity of nonsteroidal inflammatory drugs. Arch Dermatol 1989;125:824-6 12. Farr PM, Diffey BL. Pseudoporphyria due to naproxen. Lancet 1985;i; 1166-7 13. Bums DA. Naproxen pseudoporphyria in a patient with vitiligo. Clin Exp Dermatol 1987;12:296-7 14. Pierach A. Pseudoporphyria due to naproxen. Lancet 1985;i:l 166-7 15. Poh-Fitzpatrick MB, Ellis DL. Porphyria like bullous dermatosis after chronic intense tanning bed and/or sunlight exposure. Arch Dermatol 1989;125:1236-9 16. Judd LA, Henderson DW, Hill DC. Naproxen-induced pseudoporphyria. Arch Dermatol 1986; 122: 451-4

51

Chapter V

Interaction under steady state conditions of ß-cyclodextrin-piroxicam with the H2-receptor antagonists Cimetidine and ranitidine

MCW Creemers, FGM Rüssel, PLCM van Riel, FWJ Gribnau

Submitted for publication

Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

Summary

The pharmacokinetic interaction between β-cyclodextrín-piroxicam and Cimetidine and ranitidine has been studied under steady state conditions in 16 healthy male volunteers, aged 18 - 35 years, at random divided into two groups, ß-cyclodextrin- piroxicam was administered once daily (20 mg), Cimetidine (400 mg) and ranitidine (150 mg) both twice daily, one hour prior to ß-cyclodextrin-piroxicam ingestion, β- cyclodextrin-piroxicam was studied at 4 different periods, the first and last period ended with a washout period, in the 3rd period the H2-receptor antagonist was administered concomitantly. Plasma and urine samples were collected regularly at each seventh day of a period. Basic pharmacokinetic parameters were calculated for piroxicam, Cimetidine and ranitidine, and of each drug the percentage of the dose excreted unchanged in urine were determined, including the 5-hydroxy metabolite of piroxicam. Data of all volunteers of the different periods have been analyzed using paired t-tests, and revealed no statistically significant differences. The differences between Cimetidine and ranitidine have been analyzed applying t- tests on increments of a subset of parameters of the four different periods. No statistically significant differences have been found between the two H2-receptor antagonists. Pharmacokinetics of Cimetidine and ranitidine were consistent with literature. In conclusion, there is no interaction between ß-cyclodextrin-piroxicam and the H2-receptor antagonists Cimetidine and ranitidine, under steady state conditions in healthy volunteers, absorption of ß-cyclodextrin-piroxicam is unchan­ ged during administration of Cimetidine and ranitidine.

Introduction

Nonsteroidal antiinflammatory drags (NSAIDs) are widely prescribed because of their antiinflammatory, analgesic and antipyretic effects1. However, they frequently cause gastrointestinal (GI) side effects, varying from mild abdominal discomfort to peptic ulceration and occasionally hemorrhage2,3. Extensive efforts have been made to reduce the prevalence of GI side effects by altering the route or the form of administration. However, NSAIDs are known to inhibit prostaglandin synthesis sys- temically, resulting in vulnerability of the GI mucosa4. Peptic ulcers and NSAID 6 gastropathy are treated with H2-receptor antagonists' and proton pump antagonists , without reduced efficacy during concomitant administration of NSAIDs7. By now, 8,9 no preventive effects have been found with H2-receptor antagonist administration . Various interactions between NSAIDs and H2-receptor antagonists have been reported, recently reviewed by Dixon et al.10. Piroxicam is a widely used NSAID in rheumatologic practice. It has a long half-life of on average 45 hours (range 30 to 60 hours1112), thus, allowing once daily administration. Usually two or more peaks are observed in the concentration- time profile at two to 12 hours after dosing13. Piroxicam is largely eliminated by

55 Chapter V biotransformation, namely hydroxylation followed by glucuronidatìon and about 2 to 8% of a dose is excreted unchanged in urine14,15. Its major, inactive, metabolite 5-OH-piroxicam is mainly excreted in urine unchanged or as glucuronide16. ß-cyclodextrins are able to form inclusion compounds and modify the behavior of many drugs in aqueous solution, increasing solubilization, stabilization and pharmaceutical availability17, ß-cyclodextrin-piroxicam is the complexed form of ß- cyclodextrin and piroxicam. This complexation has led to an improved absorption in the first two hours1', and a reduced prevalence of GI side effects is suggested and found in several studies with ß-cyclodextrin-piroxicam1*'23. Changes in gastric pH scarcely influence the dissolution of the ß-cyclodextrin complex. The equilibri­ um of free and ionized drug will shift at larger pH and the amount of ionized, non­ absorbable, drug will increase. Besides the altered absorption kinetics, the pharma­ cokinetic profile of ß-cyclodextrin-piroxicam is similar to that of piroxicam18. Anti-ulcer treatments still have to be prescribed and thus possible interactions with H2-receptor antagonists might occur. Hj-receptor antagonists reduce gastric acid secretion, leading to an increase in gastric pH up to 4м, thus possibly influenc­ ing absorption of other drugs. Cimetidine and ranitidine are two Hrreceptor an­ tagonists, differing from each other in their structural formula; Cimetidine having an imidazole ring and ranitidine having a furane ring25. Cimetidine is weaker than ranitidine in inhibiting fT-secretion25'28, and binds more avidly to the haem portion of the cytochrome P^ system of mixed function oxidases, leading to a more powerful inhibition of the metabolic pathway of hydroxylation29. This influence on the metabolic pathway probably is the basis for the large number of interactions of mainly Cimetidine with other drags30,31. Cimetidine and ranitidine are excreted by glomerular filtration and tubular secretion, and the percentage of the dose excreted unchanged in urine of Cimetidine is on average 7S%32 and of ranitidine 70% to 78%33,34. Renal excretion might be influenced directly or indirectly by other drags, such as NSAIDs. The interaction of piroxicam with ranitidine and Cimetidine has been studied in rats and humans. In general a clear influence of Cimetidine on piroxicam pharmaco­ kinetics has been observed35"37, and a lack of this interaction with ranitidine38. In rats both piroxicam and ranitidine were administered intravenously at high doses, and plasma piroxicam concentrations and plasma half-life were significantly increased during Cimetidine administration38. Almost similar results were found in two studies with healthy male volunteers: in both studies piroxicam was given as a single oral dose of 20 mg and Cimetidine 300 mg four times a day35 or 200 mg three times a day38. In a 2-way cross-over study in 11 (évaluable) arthritic patients both Cimetidine and nizatidine caused a small but statistically significant change in the ratio of the areas under the curve (AUC) of 5-OH-piroxicam and of piroxicam in plasma37. The authors considered this change clinically not relevant, however, an influence on the half-life of piroxicam cannot be ruled out as this was not inves­ tigated. In an intrasubject comparison of 10 healthy male volunteers a small but significant increase in the area under the curve (AUC) of 15% was found35. Said and Foda38 studied 12 male healthy volunteers, divided into two equal groups, and

56 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

Table 1. Characteristics of healthy male volunteers (mean ± standard deviation)

Cimetidine group ranitidine group (n-8) (n=8)

age (years) 25 ± 4 24 ± 3 body mass index (kg τη"1) 22.4 ± 1.9 22.9 ± 2.5 weight (kg) 74 ± 9 77 ± 10

Cimetidine increased the AUC and plasma half-life significantly. Two placebo- controlled cross-over studies in 18 healthy volunteers were performed by Dixon et al36: in the first study piroxicam (20 mg) was taken twice daily for 10 days with a single dose of ranitidine (150 mg), in the second study ranitidine (150 mg) was administered twice daily with piroxicam (20 mg) as a single dose. No interaction between ranitidine and piroxicam was found. Up until recently only one study has addressed the interaction under steady conditions of both piroxicam and H^-receptor antagonists37. In our study clinical practice was mimicked by assessing the interaction under steady state conditions of piroxicam as well as of the H2-receptor antagonists, Cimetidine and ranitidine.

Furthermore, special reference was paid to the influence of the H2-receptor antagonists on the absorption kinetics of ß-cyclodextrin-piroxicam. The phar­ macokinetics of both (ß-cyclodextrin-)piroxicam and the H2-receptor antagonists were analysed in plasma as well as in urine. Measurements were paired because of the known large interindividual variability in piroxicam pharmacokinetics.

Materials and methods

Selection of volunteers Healthy male volunteers, aged between 18 and 35 years, were selected after complete interview, physical examination, routine laboratory tests and an ECG. They were excluded if they had evidence of: a. previously experienced side effects to NSAIDs or Hj-receptor antagonists; b. history of blood dyscrasias or coagulation disorders; с depression or other mental disorders; d. alcohol, nicotine or drug abuse; e. use of medication other than study medication one month before the start of the study period; f. smoking39,40; g. use of alcohol two weeks previously to enrollment in the study. Smoking, consumption of alcohol, and use of medication other than study medication was disallowed during the whole study period.

Study design The study was conducted at the department of pharmacology of the Catholic University Nijmegen, the Netherlands, after approval of the ethical committee was obtained. Volunteers (n=16), enrolled into the study after written informed consent,

57 Chapter V

Figure 1. Schedule of the study

Д H2-nc«ptorl mtagonirt |

ρ -CD-piroxicam Piroxicam I

period t period 2 period 3 period 4

0 7 12 18 25 33 40 time (d*y·)

The blocks indicate the time a drug is being administered. The arrows indicate every seventh day of a period, during which blood and urine samples are taken frequently

were allocated to one of the H2-receptor antagonist groups. The allocation proce­ dure41 was guided by the body mass index (weight/height2, kg/m2) and age. Characteristics of volunteers are shown in Table 1. ß-cyclodextrin-piroxicam (20 mg tablet Brexine* S.A. Nycomed Christiaens B.V., Belgium) was given once daily in the morning, and ranitidine (ISO mg tablet Zantac*, Glaxo B.V., the Netherlands) and Cimetidine (tablet 400 mg Tagamet*, SmithKline French B.V., the Netherlands) were both adminstered twice daily. The H2-receptor antagonist was administered one hour prior to ß-cyclodextrin-piroxicam. During the first two days of drug ingestion of the three drugs a double dose was administered. Tablets were ingested with a 150 ml tap of water. Volunteers were supplied with tablets for eight days every period and compliance was checked and stimulated by tablet counting. Volunteers participated in the study for 40 days, divided into four periods during which ß-cyclodextrin-piroxicam was administered (see Figure 1). Every seventh day of a period blood samples, starting with a predose sampling, were taken at times 0, 0.25, 0.5, 0.75, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 24 hours after ß- cyclodextrin-piroxicam ingestion. Urine was collected over the following time intervals 0-1, 1-2, 2-3, 3-6, 6-9, 9-12, 12-24 hours, starting with a predose collecti­ on. Period 1 and 4 both contained a washout period for ß-cyclodextrin-piroxicam starting at the seventh day (see figure 1) and additonial blood samples were taken after the last dosing at times 36, 48 and 96 hours and at times 36, 48, 96 and 168 hours for period 1 and 4, respectively. After the last dosing, urine portions were collected in period 1 over the time intervals 24-48 and 48-% hours and in period 4 over the time intervals 24-48, 48-96 and 96-168 hours for period 4. Additionally, a predose blood sample was taken at day 5 of period 2, 3, and 4. In total, of each volunteer 70 blood and 39 urine samples were obtained.

58 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

The H2-receptor antagonists, Cimetidine and ranitidine, were administered concomi­ tantly during period 3 and at day 7 blood and urine samples were taken, starting with a predose sampling, at times 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 11, 13, 25 hours after the morning dose. Urine samples were collected over time intervals: 0-1, 1-2, 2-3, 3-4, 4-7, 7-10, 10-13, 13-25 hours. Prior to predose sampling volunteers had to stay fasted overnight Every seventh day of a period volunteers were not allowed to leave the clinic until the last samples were taken in the evening. After ß-cyclodextrin-piroxicam ingestion meals and Quid intake were standardized for 12 hours. Meals were given after 2.5, 5 and 10 hours after the ß-cyclodextrin-piroxicam ingestion. Fluids supplied were tree of caffeine. Side effects were asked for each visit as well as consumption of alcohol and tobacco.

Laboratory procedures During the frequent blood sampling over 12 hours, an intravenous catheter was used, flushed with heparin solution after sampling. Blood was stored on ice until centrifugation at 3000 rpm for 10 minutes. Plasma was collected from each specimen and stored at -20° С pending analysis. Urine was collected over the defined time intervals, the volume measured and samples stored at -20° С pending analysis. Analysis of piroxicam The development, validation and the analyses of piroxicam in plasma and urine and of 5-OH-piroxicam in urine have been performed at the Department for Bioanalysis of Pharma Bio-Research Laboratories B.V. (Assen, the Netherlands)42. Procedures were in accordance with the guidelines for Good Laboratory Practice as published by the US Food and Drug Administration and the Organization for Economical Cooperation and Development. 5-OH-piroxicam has been determined after de- glucuronidation with ß-glucuronidase.

Analysis of ranitidine and Cimetidine Concentrations of Cimetidine and ranitidine were measured by high performance liquid chromatography (HPLC) after solid-phase extraction (according to a method which will be published in detail elsewhere)43. Briefly, 5 μg of either ranitidine or Cimetidine were added as internal standard to 500 ul of plasma or 1000 ul urine. Following protein precipitation in plasma with 5% (w/v) metaphosphoric acid, the supernatant was brought to pH 2.6 with a phosphate solution containing 0.02 M of the ion-pairing agent octanesulfonate and loaded onto a Bond Elut PH Column (Analytichem Int., Haba City, CA). After washing the column four times, Cimetidi­ ne and ranitidine were eluted with 0.75 ml of a solution consisting of acetonitrile: 0.01 M phosphate buffer/0.005 M octanesulfonate (30:70). Urine samples were extracted on the same columns, but under alkaline conditions without octanesulfo­ nate, essentially based on a method described by Chiou et al44. Separation of 10 ul

59 Chapter V eluate was achieved on a Hewlett Packard 5 μιη LiChrosorb RP-18 column (200 χ 4.6 mm) at 40° С The mobile phase, consisting of a 22:78 mixture of the elution fluid mentioned before, was delivered at a rate of 1 ml/min and column effluent was monitored at a wavelength of 228 nm. Retention times of Cimetidine and ranitidine were 5.9 and 7.6 min, respectively. Peak height ratios were measured and linear calibration curves were obtained for concentrations ranging form 0.025 - 5 pg/ml (r2 > 0.9994). The absolute recovery of Cimetidine and ranitidine was 90% with a coefficient of variation less than 2.3%. The mean precision of analysis in spiked samples ranged from 4.0 - 0.7% over a concentration range of 0.063 - 2.5 pg/ml. The lower limit of detection (four times baseline noise) was 10 ng/ml.

Pharmacokinetic analysis

For each subject, plasma concentration curves of piroxicam and Cimetidine and ranitidine were analysed according to model independent methods. The following pharmacokinetic parameters were calculated: elimination rate constant (K.,), calculated by log-linear regression analysis of the terminal phase of the curve (of periods 1 and 4); half-lives (tM); areas under the plasma concentration time curves (AUC), determined by the linear trapezoidal rule; volumes of distribution (Vjj/F = Dose/k.|*AUC; F = bioavailability). Time to peak concentrations (t„,„) and peak concentrations (Cm„) were calculated by using the first peak for Cmmi, and !„,„_, and the second peak for Сш„ 2 and t,^ 2. Oral clearance of piroxicam was cal­ culated as CLR = Dose/AUC. Renal clearance (CLR/F) of the H2-receptor an­ tagonists was calculated from the average value of the renal excretion rate divided by the midpoint plasma concentration of three different urine collection intervals, in the descending phase of the curve. Furthermore, the mean residence time in steady state (MRT = (AUC*t + T*

(Cm)ss/kel)/AUC; T=dosing interval; (Cm)„=concentration of the last sample of the dosing interval)45 and the steady state volume of distribution (V„/F) of piroxicam for each period were calculated, applying statistical moments theory44. Cumulative urinary excretion and the percentage of the dose excreted were calculated for piroxicam and 5-OH-piroxicam. Creatinine was measured in all 24 hours urine portions to check completeness of urine collection.

Statistical analysis

A number of pharmacokinetic parameters calculated for piroxicam in plasma and urine and for 5-OH-piroxicam in urine were mutually dependent. Therefore, a subset of parameters was selected for statistical analysis in order to avoid multiple testing, and thus reduce the probability of false significancies.

60 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

Figure 2a. Plasma concentration time course Piroxicam (mg/l) 1(h ί a. . Ν . •,

0.1 -1— 10 20 30 40

time (days) The plasma concentration time course of one volunteer is shown in this figure. The dotted lines, calculated by log-linear regression, are used for determination of k^ for the plasma half life

Figure 2b. Pharmacokinetic profile of piroxicam Piroxicam (mg/L) 8

—ι 10 12 time (hours)

For evaluation of the interaction between piroxicam and the two H2-receptor antagonists paired t-testing was applied to the following parameters of all 16 volunteers: periods 1 and 2: AUC, the percentage of the dose in urine of piroxicam and 5-OH-piroxicam; periods 2 and 3, periods 3 and 4, and periods 2 and 4: AUC,

61 Chapter V

„ _, f; q ir, »! Ч «1 » » PI m —ι ·* о м _* _* О О — О' Γ­ "8 -Η -Η -Η -Η +1 -Η -Η Η •» -н -н -н 'С ιπ « 00 η fl\ β - ·* M ON Tt — К. » « a - ^ à » Г^ — •* Ö m

(Ч о ^. Г~; « сч f4 f*1 о -* о ci о •8 •H ' • -н · ' -н -*•н Ή •н Ή •с о ÍS г·; 06 о ГЛ о' IL г»

о О 00 чо сч о — - о ri Ö -н • ι -Η •н •Η -Η -Η г» ON •* «Ч ts •* ~ t- ι- ** •w О а

Γ<ΙΟΝ(4©*-<4-- — о -* о ~* "8 -Η-Η+{-Η·Η+Ι-Η-Η-Η-Η-ΗΉ о\<лаог-ідч:г~«->оо<чтгоо I in іл о ·* о

ί Χ Χ Χ ι Χ Ι Ι Ι Ι Ι Ι Ι

χ χ χ χ χ χ χ Ν Ν Π

ι ι ι ι ι ι ι ι ι χ χ

e? .а ώ> ^- ώ ώ •a £ 5j^ollJJJJ I Q

62 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

Ρ Se q r« о _: <4 8 см Ö (4 —. ci 06 S t r- m" Si

.S

o 0s m во io" ^ f^ о o Л оо « с) —< η g •ä О — -ч CS \¿ o ¿· о сГ ρ" νο to — во" _¿ υ 4 ob m \o г- — 3 # 00 (S X ¿ ¿ ¿ ¿ 9 Tf «?T 83 sa ss κ= ss; aa ss Ρ? do' dó 90 00' 09 rid do' Ыо ε

o 00 00 о 00 сч o S s f4 ι -ч Г« -^ •^ »С О •ч· α «η о" оо" ri 8 ο m ? г 8 9 η ri m «? σ\ ι ^ -* ©m fn*n —ÍN tso —м mm 00 —σ* £ ς>©' ÖÖ 90 09 -¡о 99 ©ri и °9 8 α

8 o г e o о ri o % ¡Ô 8 m •о 9 — — сч — νο 9 ili ι Τ 1 -*r* ^ —* ΟΝΟ orí oom vom о ІІ ττη CJ-« r»r1 i^rj оо — 28 09 С4СЧ •a 99 90 SO 99 9*0' 9» υ ОЙ и α; оси ucci uc¿ uo< им и as им ueü им им

¡S £ 5 •ч Ö г > υ υ J J 8

63 Chapter V

C„,„,i, Cma2, t^^, ^¿, and the percentage of the dose in urine of piroxicam and 5-OH-piroxicam; and of periods 1 and 4: t^, MRT, and Vjj/F. To evaluate the differences between the influence of the two H2-receptor antagonists on piroxicam pharmacokinetics, the increments of the same subgroup of parameters were analysed using t-tests. Calculation of these increments reduces the influence of intersubject variability.

Results

Figure 2a shows a representative pharmacokinetic profile of piroxicam in plasma of a volunteer over the whole study period. The profiles of the 24 hour curves of piroxicam - in total 64 curves - showed two to four peaks (Figure 2b). For all 16 volunteers, the pharmacokinetic parameters of the four periods are shown in Table 2. Paired t-testing revealed only statistical significant changes in the AUC from period 1 to 2, and of t,,,,, 2 from period 3 to 4 (marked in Table 2). The comparison of the influence of the two H2-receptor antagonists on the piroxicam pharmacokine­ tics, analysed by using the increments between the different periods in a subset of parameters, For analysis of the influence of the two H^receptor antagonists on piroxicam pharmacokinetics, the increments between the different periods were calculated and showed large ranges of most parameters (see Table 3). None of the parameters, selected for statistical analysis, were significantly different, and p-values found varied from 0.11 to 0.98. Pharmacokinetic parameters of Cimetidine and ranitidine are shown in Table 4. Half of the volunteers showed pharmacokinetic plasma profiles with two peaks. Pharmacokinetic parameters calculated were consistent with literature33'34'47'52.

Discussion

The interaction between ß-cyclodextrin-piroxicam and two H2-receptor antagonists, Cimetidine and ranitidine, has been studied under steady state conditions of both ß- cyclodextrin-piroxicam and the H2-receptor antagonists in healthy volunteers, thus mimicking clinical practice. The study design allowed intrasubject comparison of parameters of piroxicam, eliminating the effects of the known large variability in piroxicam pharmacokinetics between subjects. Changes in the pattern of metabolite formation were determined by measuring 5-OH-piroxicam in urine over regular time intervals before, during and after administration of the H2-receptor antagonist. Pharmacokinetics of piroxicam in plasma were consistent with those reported in other studies with healthy volunteers. The percentage of the dose of piroxicam excreted unchanged has been reported previously and estimated to be 2 to 5%15. In

64 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

Table 4. Phannacokinetíc parameters Cimetidine and ranitidine

parameter Cimetidine ranitidine

AUC (mg.h.r" 8.41 ± 1.40 2.80 ± 0.57 2.4 ± 0.7 2.9 ± 0.6 234 ±75 VF(1) 160 ±35 9 ±14 29 ±6 CU (l.h') 2.8 ± 0.5 0.46 ± 0.07 C„ (mg.11) bl 13 ± 0.2 2.0 ± 0.7 t^., (h) 1.7 ± 0.9 (n-4) 0.51 ±0.09(n=5) C„j (mg.r') 2.7 ± 0.6 4.0 ± 0.6 «™л (h) % of dose excreted in urine 64±8 45 ± 10

our study this percentage was even less than 1% over a period of 24 hours after dosing. 5-OH-piroxicam, measured after de-glucuronidation with ß-glucuronidase, was excreted in urine for about 20 to 25% and this was in line with the percentages published recently by Milligan et al.37. Pharmacokinetic parameters are mutually dependent, although a subset of primary parameters can be identified. In this way multiple testing can be avoided, and the probability of false significance reduced. To eliminate intersubject variab­ ility in pharmacokinetics of piroxicam, influences of the Hj-receptor antagonists were evaluated by applying t-tests to increments between the different periods. The AUC showed a significant difference between periods 1 and 2 (see Table 2), although volunteers were loaded with a double dose of ß-cyclodextrin-piroxicam during the first two days. Probably steady state has not been reached at day 7 in period 1, however as calculated from the volume of distribution, steady state concentrations should have been approximated closely by the loading dose. There is no reason to doubt that all volunteers were less or non-compliant in the period 1 than later in the study. The statistical difference found, however, does not interfere with the analyses and conclusions of the study. The comparison between periods 2 and 3, and between periods 3 and 4 are the most important. Absorption kinetics of ß-cyclodextrin-piroxicam was not altered significantly due to Cimetidine or ranitidine administration. The time of the second peak (t^ 2) of piroxicam follows about one hour after volunteers had their breakfast, probably related to emptying of the gallbladder and the supposed enterohepatic re-circulation of piroxicam. The drop in t^ 2 in period 3 might be attributed to an interaction as a result of a changed gastric motility due to an increased gastric pH. It might be suggested that the subsequent increase in !„,„, 2 after discontinuation of the H2- receptor antagonists points out a possible rebound phenomenon in gastric acid production, which could accelerate gastric motility. Glucuronidation of a drug can take place in the liver and in the kidney. To study the interaction of the I^-receptor antagonists on glucuronidation in the liver as well as in the kidney, concentrations of 5-OH-piroxicam in urine were measured

65 Chapter V after de-glucuronidation. Measurement of the concentration of 5-OH-piroxicam in plasma separately would give no additional information concerning the route, i.e. liver of kidney, of glucuronidation. In this study the percentage of the dose of 5- OH-piroxicam excreted in urine was not influenced by the H2-receptor antagonists. It can be concluded that both Cimetidine and ranitidine do not give a clinically relevant interaction with piroxicam. The 95% confidence intervals of the increments calculated between the different periods of pharmacokinetic parameters are large and, therefore, differences between the two Hj-receptor antagonists have to be quite large to reach significance. Previously an increase in the half-life35 3B and of the AUC38 has been found due to Cimetidine administration. The increase in the AUC has not been confirmed by Milligan et al37, although in this study the ratio of the AUCs of 5-OH-piroxicam and of piroxicam in plasma changed significantly for both Cimetidine and nizatidine37. Nizatidine has an inhibitory effect on the enzyme 31 cytochrome ¥4X about ten times less potent compared to Cimetidine . Therefore, it is surprising that both drugs showed significance as only influence of Cimetidine might be expected. The lack of an interaction might be explained by the fact that cytochrome P^, is known to comprise multiple isoenzymes53,54. The isoenzymes inhibited by Cimetidine53 differ from the isoenzyme responsible for the oxidation of NSAIDs55,56. Therefore, piroxicam might be oxidated by an isoenzyme for which Cimetidine has a low affinity, resulting in a lack of interactioa In this study healthy volunteers have been studied, and results apply for patients with a normal liver and kidney functioa The only study available in patients included arthritic patients which apparently not had a very active disease reflected in albumin concentration of on average 40 g.l"1 37. Additional information about liver and kidney function was not provided in this study37. In summary, our study satisfactory ruled out the possibility of an effect in man of the two Hj-receptor antagonists, Cimetidine and ranitidine, on the kinetics of piroxicam and the ab­ sorption kinetics of ß-cyclodextrin-piroxicam.

Acknowledgements

We are indebted to Dr. M.A. van 't Hof for statistical assistance.

References

1. Langman MJS. Epidemiologic evidence on the association between peptic ulceration and anti- inflammmatory drug use. Gastroenterology 1989;96:640-6 2. Semble ES, Wu WC. Antiinflammatory drugs and gastric mucosal damage. Seminars Arthr Rheum 1987:16:271-86 3. Soil AH, Weinstein WM, Kurata J, McCarthy D. Nonsteroidal antiinflammatory drugs and peptic ulcer disease. Ann Int Med 1991;114:307-19

66 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

4. Cryer В, Feldman M. Effects of nonsteroidal anti-inflammatory drugs on endogenous gastroin­ testinal prostaglandins and therapeutic strategies for prevention and treatment of nonsteroidal anti­ inflammatory drug-induced damage. Arch Intern Med 1992; 132:1145-55 5. Feldman M, Burton ME. Histamincj-receptor antagonists. Standard therapy for acid-peptic diseases, (second of two parts) N Engl J Med 1990;323:1749-55 6. Scheiman JM. Pathogenesis of gastroduodcnal injury due to nonsteroidal antiinflammatory drugs: implications for prevention and therapy. Seminars Arthr Rheum 1992;21:201-10 7. Croker JR, Cotton PB, Boyle AC, Kinsella P. Cimetidine for peptic ulcer in patients with arthritis. Ann Rheum Dis 1980;39:275-8 8. Roth SH, Bennett RWE, Mitchell CS, Hartman RJ. Cimetidine therapy in nonsteroidal antiinflamm­ atory drug gastropathy. Arch Intern Med 1987;147:1798-1801 9. Ehsanullah RSB, Page MC, Tildesley G, Wood JR. Prevention of gastroduodcnal damage induced by non-steroidal anti-inflammatory drugs: controlled trial of ranitidine. Br Med J 1988;297:1017-21 10. Dixon JS, Page MC. Interactions between NSAIDs and Hj-rcceptor antagonists or prostaglandin analogues. Rheumatol Int 1991;11:13-18 11. Hobbs DC, Twomey TM. Piroxicam pharmacokinetics in man: aspirin and antacid interaction studies. J Clin Pharmacol 1979;19:270-81 12. Brogden RN. Heel RC, Speight TM, Avery OS. Piroxicam, a reappraisal of its pharmacology and therapeutic efficacy. Drug 1984;28:292-323 13. Verbeeck RK, Richardson CK, Blocke KLN. Clinical pharmacokinetics of piroxicam. J Rheumatol 1986;13:789-96 14. Hawkey CJ, Rampton DA. Prostaglandins and the gastrointestinal mucosa: are they important in its function, disease, or treatment? Gastroenterology 1985;30:1162-88 15. Wiseman EH, Hobbs DC. Review of pharmacokinetic studies with piroxicam. Am J Med 1982; 72:9-17 16. bombardino JG. Synthesis and antiinflammatory activity of metabolites of piroxicam. J Med Chem 1981;24:39-42 17. Orienti I, Cavallari С, Zecchi V. Availability of NSAJDH* B-cyclodextrin inclusion complexes. Arch Pharm 1989;322:207-11 18. Acerbi D, Bonari С, Boscarine G, Bufalo L. Pharmacokinetic study on piroxicam at the steady state in elderly subjects and younger adults after administration of piroxicam-B-cyclodextrins. Int J Clin Pharm Res 1988:8:175-80 19. Patoia L, Clausi G, Farroni F et al. Comparison of faecal blood loss, upper gastrointestinal mucosal integrity and symptoms after piroxicam B-cyclodextrin, piroxicam and placebo administration. Eur J Clin Pharmacol 1989;36:599-604 20. Ambanelli U, Nervetti A, Colombo В et al. Piroxicam-ß-cyclodextrin in the treatment of rheumatic diseases: a prospective study. Curr Thcr Res 1990;48:58-68 21. Bonardelli P, Oliani C, Preti PAM, Pellicano Ρ, Quattrocchi G. Efficacy and gastrointestinal tolerability of ß-cyclodextrin-piroxicam and tenoxicam in the treatment of chronic osteoarthritis. Clin Therap 1990;12:547-55 22. Santucci L, Fiorucci S, Patoia L et al. Gastric tolerance of piroxicam-ß-cyclodextrin compared with placebo and with other NSAIDs: an endoscopic and functional study by evaluation of transmucosal potential difference. Drug Investig 1990;2 (suppl 4):56-60 23. Warrington S, Debbas N, Farthing M, Horton M, Umile Α. Piroxicam-ß-cyclodextrin: effects on gastrointestinal blood loss and gastric mucosal appearance in healthy men. Int J Tiss Reac 1991; 13:243-8 24. Feldman M, Burton ME. Histamin^-receptor antagonists, first of two parts N Eng J Med 1990;323: 1672-80

25. Linn JH. Pharmacokinetic and pharmacodynamic properties of histamine H2-receptor antagonists. Clin Pharmacokinet 1991;20:218-36 26. Brater DC, Peters MN, Eshelman FN, Richardson CT. Clinical comparison of Cimetidine and ranitidine. Clin Pharmacol Ther 1982;31:484-9

67 Chapter V

27. Lcbcrt PA, MacLeod SM, Mahon WA, Soldin SJ, Vandcnberghe HM. Ranitidine kinetics and dynamics. I. Oral dose studies. Qui Pharmacol Tber 1981-.30-539-40 28. Lebert PA, Mahon WA, MacLeod SM et al. Ranitidine kinetics and dynamics Π. Intravenous dose studies and comparison with Cimetidine. Clin Pharmacol Ther 1981;30·545-50 29. Bast A, Smid J, Timmerman Η. The effects of Cimetidine, ranitidine and famotidine on rat hepatic microsomal cytochrome P-450 activities. Agents Actions 1989,27:188-91 30. Powell JR, Dorm КН. Histamine Hj-antagonist drug interactions in perspective: mechanistic concepts and clinical implications Am J Med 1984,77 (suppl 5b)-57-84 31. Klotz U, Kroemer HK. The drug interaction potential of ranitidine' an update. Pharmac Ther 1991,50233-44 32. Grahnen A, Bahr von C, Lindström В, Rosen A. Bioavailability and pharmacokinetics of Cimetidi­ ne. Eur J Clin Pharm 1979,16-335-40 33. Berardi RR, Tankanow RM, Nostrani ТТ. Comparison of famotidine with Cimetidine and ranitidine. Qmical Pharmacy 1988,7-271-84 34. Hecken van AM, Tjandramaga ТВ, Mullie A, Verbesselt R, Schepper de PJ Ranitidine single dose pharmacokinetics and absolute bioavailability m man. Br J Clin Pharmac 1982,14-195-200 35. Mailhot C, Dahl SL, Ward JR. The effect of Cimetidine on serum concentrations of piroxicam. Pharmacotherapy 1986;6112-7 36. Dixon JS, Lacey LF, Pickup ME, Langley SJ, Page MC. A lack of pharmacokinetic interaction between ranitidine and piroxicam. Eur J Clin Pharm 1990;39-583-6 37. Milhgan PA, McGill PE, Howden CW, Kelman AW, Whiting B. The consequences of H,-receptor antagonist - piroxicam coadministration in patients with joint disorders. Eur J Clin Pharm 1993,45-507-12 38. Said SA, Foda AM. Influence of Cimetidine on the pharmacokinetics of piroxicam m rat and man. Drug Res 1989-.39-790-2 39. Schuerer-Maly CC, Varga L, Koelz HR, Halter F. Smoking and pH response to Hj-receptor antagonists. Scand J Gastroenterol 1989,24 1172-8 40. Bauerfeind P, Cilluffo T, Fimmel CJ et al. Does smoking interfere with the effect of histamine ba­ roceptor antagonists on intragastric acidity in man? Gut 1987,28-549-56 41. Pocock SJ. Sequential treatment assignment with balancing for prognostic factors in the controlled clinical trial. Biometrics 1975,36 81-90 42. Mensink CK, Somhorst YAM, Jonkman JHG. Determination of piroxicam in human plasma and urine and of 5-OH-piroxicam in human unne by high performance liquid chromatography and UV- detection. Assay Method Reports PBRL-RD-129, PBRL-RD-134, PBRL-RD-138, 1993. Pharma Βίο-Research International В V, Assen, the Netherlands 43 Rüssel PGM, Creemers MCW, Tan Y, Riel van PLCM, Gribnau FWJ. Ion-pair solid-phase extraction of Cimetidine from plasma and subsequent analysis by high-performance liquid chromatography. J Chromatogr В (accepted for publication, 1994) 44. Chiou R, Stubbs RJ, В ay ne WF. Determination of Cimetidine in plasma and unne by high perfor­ mance liquid chromatography. J Chromatogr 1986,377-441-6 45. Pfeffer M. Estimation of mean residence tune from data obtained when mulbplc-dosing steady state has been reached. J Pharmac Sci 1984,73-854-6 46. Cheng H, Jusko WJ. Noncompartmental determination of the mean residence tune and steady state volume of distribution during multiple dosing. J Pharm Sci 1991,80 202-4 47. Gladwiza U, Klotz U. Pharmacokinetics and pharmacodynamics of Hj-rcceptor antagonists in patients with renal insufficiency. Clin Pharmacokinet 1993,24-319-32 48 Hu YO, Poa L-H, Chung P-Η, Tang Η-S. Pharmacokinetic properties of ranitidine m Chinese people. Acta Gastrc-Enterol Belgica 1991,54-328-35 49. Somogyi A, Rohner Η-G, Gugler R. Pharmacokinetics and bioavailability of Cimetidine m gastric and duodenal ulcer patients Clin Pharmacokinet 1980,5-84-94 50. Walkenstein SS, Dubb JW, Randolph WC et al. Bioavailability of Cimetidine in man. Gastroenterol­ ogy 1978;74 360-5

68 Interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

51. Bauer LA, McDonnell Ν, Hom Ж et al. Single and multiple doses of oral Cimetidine do not change liver blood flow in humans. Clin Pharmacol Ther 1990;48:195-200 52. Griffiths R, Lee RM, Taylor DC. Kinetics of Cimetidine in man and experimental animals. In; Exc Mod Proc 2nd symposium on Hj-receptor antagonists 1977, eds. Burland WL, Simkins MA. Amsterdam, the Netherlands 53. Knodcll RG, Brwne DG, Gwozdz GP, Bnan WR, Guengerich FP. Differential inhibition of

individual liver cytochromes Рш by Cimetidine. Gastroenterology 1991,101:1680-91 54. Lu AYH. Multiplicity of liver drug metabolism enzymes. Drug Metab Rev 1979;10:187-208 55. Lemann T, Transon C, Dayer P. Cytochrome P^TB (CYP2Q: a major monooxygenase catalyzing diclofenac 4'-hydroxylation in human liver. Life Sci 1992;52:29-34 56. Zhao J, Leemann T, Dayer P. In vitro oxidation of oxicam NSATOs by a human liver cytochrome P«,. Life Sci 1992;51:575-81

69

Chapter VI

Second-line treatment in seronegative spondylarthropathies

MCW Creemers, PLCM van Riel, MJAM Franssen, LBA van de Putte, FWJ Gribnau

Accepted for publication: Seminars of Arthritis and Rheumatism: April 1994

2nd-line treatment in SSpA

Summary

The literature concerning second-line treatment of seronegative spondylarth- ropathies (SSpA) from 1940 to August 1993 has been reviewed. Sulfasalazine (SASP) appeared to be effective in the treatment of ankylosing spondylitis, and promising in reactive arthritis (ReA) and Reiters' syndrome (RS). Methotrexate and azathioprine were associated with a remarkable improvement in some cases of AS and RS. Methylprednisolone and levamisole were both efficacious in AS, but levamisole was associated with occasional severe side effects. Radiation therapy led to short-term improvement in AS, but was abandoned because of severe long- term side effects. Only sulfasalazine has been studied in sufficient detail to allow definitive conclusions, but methotrexate and azathioprine may be promising drugs.

Introduction

The seronegative spondylarthropathies (SSpA) include a broad spectrum of diseases, such as ankylosing spondylitis (AS), Reiters' syndrome (RS), reactive arthritis (ReA), psoriatic arthritis (PsA), arthritis related to inflammatory bowel disease (IBD), Whipples' disease and Behçets' syndrome1. Overlap syndromes also occur, and some cases cannot be classified. As present classification criteria are limited, new criteria are currently being developed1. In general, patients with SSpA can be divided into two major groups; a. patients with inflammatory spinal disease (sacroiliitis and spondylitis) and b. patients with primarily peripheral, asymmetric synovitis1,2. A high prevalence of HLA-B27 is seen in both groups1,3·4. Psoriasis, acute anterior uveitis (AAU), enthesopathies3'4, balanitis, urethritis and oral ulcers are common extra-articular features. SSpA is generally not associated with systemic malaise, fatigue and disability as seen in rheumatoid arthritis (RA). Hence, the administration of second-line drugs are used less common in SSpA. Nonetheless, second-line drugs have been adminis­ tered to some patients with a severe SSpA over the last three decades. It is most likely that this drug regimen partly reflects the natural course of this group of diseases; in a minority of cases the disease will have a continuous and unremittent course of SSpA, leading to a need for almost continuous drug administration3. In this article we summarize the literature concerning use of second-line treatments, including second-line drugs and radiation therapy in AS, RS, ReA and EBD-related arthritis. Because most studies in PsA6"17 include only patients with active peripheral arthritis, in whom only 20% have sacroiliitis1''21, PsA is not included in detail. Each treatment will be reviewed separately, starting with the results of double-blind studies, followed by open studies and case reports. When­ ever possible the type of arthritis, disease duration and the variables in which

73 Chapter VI improvement was seen are indicated. Finally, we conclude with suggestions for further studies of SSpA.

Materials and methods

A literature search was conducted to identify all reports concerning use of second- line drugs - hydroxychloroquine, gold, d-penicillamine, sulfasalazine, methotrexate, cyclophosphamide, 6-mercaptopurine, azathioprine, cyclosporine and levamisole as well as radiotherapy, methylprednisolone and coiticotrophin - in seronegative spondylarthropathies. Our search included AS, RS, EBD-related arthritis, ReA, unclassified SSpA and not PsA. All general articles on SSpA in English, French and German were reviewed, and relevant references, dating from 1940 up to August 1993, identified. Further search was undertaken using Medline on CD-ROM from 1966 to July 1993. We excluded editorials and papers based only on retrospective data collection. If the same set of data was published by the same authors in different journals, we chose only the article which appeared in the journal having the greatest impact factor22. Papers were included only if patients, methods and results were described, with sufficiently detailed data on patient characteristics and on the method of data collection; i.e. whether the paper concerned case reports, open uncontrolled studies or clinical trials, and the duration of follow-up. In the case of an open study or a clinical trial, we required enumeration of inclusion and exclusion criteria as well as the statistical methods used for data analysis.

Results Hydroxychloroquine This drug has not been studied in relation to SSpA. Only Giordano23 described two AS patients with frequently recurring acute anterior uveitis (AAU) in whom attacks of AAU did not recur for at least 10 months and whose AS symptoms improved after starting hydroxychloroquine 400 mg daily. Side effects were not reported. D-penicillamine Conflicting reports have been published concerning the efficacy of d-penicillamine in AS. In a double-blind placebo-controlled trial, carried out over a six-month period on 17 AS patients, 13 of whom had peripheral AS, d-penicillamine was reported to be ineffective24. The drug was begun at a dose of 125 mg per day and increased monthly up to 750 mg per day. Objective and subjective variables of disease activity were assessed monthly. Despite randomization, the placebo group had less severe disease upon entering the trial, reflected in the number of inflamed joints and morning stiffness. The mean duration of the disease in the group treated with d-penicillamine was 16 years (range 8-28 years) versus nine years in the pla-

74 2nd-l¡ne treatment in SSpA

•β -а § 1 i η" Ο« CL.-C i J СО S < ь 1г І ε α. α. îl в, 9 S ¿s 1I I ε ¡s м a .§ 8>й .S ft, •si "3 — ft, .SOT η S to .В Л чсо spi ε g-O •о « 8 8 о — о η о Oh. 11 li а 8 J?. «τ, * S_ § О, .-ч ft, ^ 00 ν ft. * com И" ^Нэ gS ft ^δ 3 d. i soft, < toft, со ft, „.ft, и ft. ил о δ II η S Г- >β СЛО\ VO ¡о -"-" or« Í4 — - fi s s 1 J — <ч •Si l£ !£ SE S «e s* S Я 2 2 00 'cco = 8 w •si а ·· α, o se •о 2 8 lì ft, to Ι ι І i

1 ¡a г ? & я Я ч il о ja -й ja Ja ja ja J3 ja I« ε 8 1 8 Ì s ε s Ή £J 2 S 1* .3 3 ой •о со 1 0 α 00 ca Q Q Q О о Κ Κ g. g- К К К І К £

№, f*, fe |ь Iй« .gV ЭЬ js fe st. ¡i, 5 oo Sao .goo goo Кос > oo }• ifc яa [2a aa Sa да áa 5a ga <3а За л

75 Chapter VI cebo group (range 1-26 years). In a nine-month-long open study of 49 AS patients (17 with peripheral AS) Bemacka et al found a significant improvement in spinal mobility and the number of swollen joints, but not in the ESR. The results are in marked contrast to the double-blind study, although interpretation is complicated by hospitalization and physiotherapy2"6. All 17 patients with peripheral arthritis improved. In these two studies three patients experienced side effects, consisting of an allergic reaction, a blistering rash and diarrhea with abdominal pain. Similar contradictory results have been found in many case reports of patients with axial as well as peripheral AS27"31. In a case report d-penicillamine proved to be an effective treatment in two patients with IBD-related sacroiliitis, both of whom unable to tolerate treatment with NSAIDs28.

Gold No placebo-controlled studies with intramuscular gold salts in SSpA are available. In a clinical report of 128 AS patients, six out of 18 patients with peripheral arthritis, who were treated with intramuscular gold salts, experienced major benefit, and two others minor benefit32. One RS patient was successfully treated with intramuscular gold up to a cumulative dose of 2.835 mg33. No side effects were reported. The oral gold compound auranofin was studied a placebo-controlled multi- center trial, double-blind for six months and single-blind for six additional months. In the final six months the auranofin dosage could be increased from 6 mg to 9 mg daily. Of the 60 patients, 55 could be evaluated. No difference could be found between treatment with auranofin or placebo in clinical or laboratory variables, although an improvement could be seen in all ten AS patients with peripheral arthritis, termination of treatment because of side effects was seen in 11.6% of patients34; the most frequently reported side effects were diarrhea and soft stools.

Sulfasalazine Sulfasalazine (SASP) has been shown be effective in placebo-controlled double- blind studies in both axial and peripheral AS (see Table la). Dougados et al performed a trial in 60 patients, nine of whom had peripheral AS, using a daily dose of 2 g of SASP. The two treatment groups showed distinct differences from the third month of treatment. At the end of the study improvement in a functional index, daily NSAID dose and immunoglobulin levels was seen in 15 of the 25 SASP-treated patients who were available for evaluation. SASP treatment was discontinued in eight patients as a result of side effects39. Another double-blind placebo-controlled 12-week study of 37 patients revealed improvements in the SASP-treated group after eight weeks36. Five almost similar trials were published using average doses of 2 g37"41. All studies lasted at least 12 weeks, and all but one41 study indicated efficacy, although the mean disease duration in this study was

76 2nd-line treatment in SSpA

Table lb. Nature of side effects during S ASP treatment*

symptoms/number of patients withdrawal of SASP continuation of SASP

rash 3 0 rash + fever 3 0 oral ulcers 1 0 gastrointestinal complaints 11 23 liver function disturbances 2 0 diarrhea 1 0 anemia 1 0 leucopenia 3 0 headache 8 5 dizziness 2 1 headache + dizziness 0 2 depression 1 1 pruritus 1 0 total number of patients 37 32

* references 35 to 42 total number of patients: 430 relatively long (20 years)41. Only Nissila included a large number of patients with peripheral AS (55 out of 85); no significant difference in efficacy was found between axial and peripheral AS3*. A 24-week interventional study of 20 patients with peripheral AS indicated relapse after withdrawal of SASP in the majority of patients (60%)42. Improvement in a reduction of pain, stiffness, sleep disturbance, fingertip-to-floor distance and chest expansion was observed in almost all studies. Quite different measurements were used for spinal mobility. Peripheral arthritis and laboratory variables also were improved. The mean time interval between the initiation of SASP treatment and significant improvement was not specified in most of the studies. The prevalence of side effects was relatively low: 10% of patients had to terminate treatment and another 10% of patients experienced dose-dependent and/or transient side effects (see Table lb). A meta-analysis of five randomized, double-blind, placebo-controlled trials33'39 demonstrated that SASP can be a safe and effective drug in the short-term treat­ ment of AS and they found no evidence of a different response between axial AS and peripheral AS43. In two open studies44,45 and 10 reported cases46,47 the majority of patients experienced benificial effect to SASP treatment Two placebo-controlled studies, one open study and several case reports were available on use of SASP in ReA. Both placebo-controlled studies reported beneficial effects due to SASP treatment*8,49; joint pain, number of swollen joints, spinal pain, ESR, CRP and immunoglobulins improved in the SASP-treated group compared to the placebo-treated group, although differences were statistically significant in only one study49.

77 Chapter VI

Two open studies44"50 and the three case reports47 indicated efficacy and toxicity of SASP in ReA similar to those in the placebo-controlled studies. A group of 15 HLA-B27 positive ReA patients (12 men, 3 women, mean age 13 years; range 10-25 years), were treated for an average of 13 months (range 3-36 months). After 3-6 months, all patients showed significant improvement in objecti­ ve and subjective variables. Within 3-12 months 11 patients went into remission and all the remaining patients improved. No significant adverse reactions were encountered51. In an open uncontrolled study of 16 patients with RS greater improvement was seen than in a group of AS patients studied simultaneously. None of these RS patients experienced side effects47. One 6-month placebo-controlled double-blind study (n=351)52 and one open study with a follow-up of 28 months (n=37)53 were performed of SASP, 2 to 3 g daily, in SSpA. Greater efficacy was seen in the group treated with SASP, although the changes were not statistically significant in the double-blind study. Side effects were observed in 28 and five patients, respectively,i n the two studies.

Methotrexate A one-year open study of 10 AS patients (mean age 32 years, mean disease duration 11 years) with active axial arthritis treated with MTX 12.5 mg i.m. weekly indicated favorable results54. Nine patients completed the study and one dropped out due to a lack of effect. It was possible to either reduce or terminate NSAID treatment in most patients. Dose dependent side effects were observed in eight patients, including elevated liver function tests (n=4), dizziness (n=4) and dyspeptic syndrome (n=5). Prior to this study, four other patients (two axial AS, two periphe­ ral AS) were reported to have been treated succesfully with weekly doses of MTX (7.5-15 mg)55Jí. Nausea and headache were observed as side effects (n=l) which resolved after treatment with leucovorin. During the period covered by this review 33 patients with RS who were treated with MTX were reported57'73. Most of these patients had experienced failure of treatment with NSAIDs or corticosteroids. The MTX dose varied from 2.5 to 25 mg daily during six to seven days in total or from 7.5 to 50 mg weekly for at least three months. Administration was oral, i.v. or i.m. Treatment with NSAIDs, etrenitate (Tigason*: a synthetic analogue of retinoid acid, of which acitretine (Neotigason*) is its free acid and main metabolite) or steroids was usually continu­ ed. Significant improvement in arthritis was observed in 25 (78%) patients (n=32: one patient could not be evaluated). A good or even dramatic response of mucocu­ taneous lesions to MTX treatment was seen in all patients, generally observed within two weeks of onset of treatment No improvement of arthritis was observed in seven of the 32 patients, and one patient temporarily relapsed within six weeks, after which treatment with azathioprine was commenced. In reports indicating the time-course, improvement began within six weeks, but many case reports did not specify the interval from treatment onset to improvement. The prevalence of MTX side effects was low; four (12%) patients had to terminate MTX due to oral ulcers,

78 2nd-line treatment in SSpA liver function disturbances and anemia combined with leucopenia. All side effects resolved after withdrawal of treatment One patient experienced peripheral neuropa­ thy, which was not improved after discontinuation of MTX*'. Two other patients had transient liver function disturbances, but treatment could be continued. Cyclophosphamide No controlled studies are available on cyclophosphamide (CCP) in SSpA, and the results of open studies are conflicting. In one study in 12 hospitalized patients with peripheral AS who had a high disease activity and failure of treatment with NSAIDs, 200 mg CCP was administered intravenously every other day during the three weeks of hospitalization, followed by 100 mg weekly orally. Within the first three weeks of treatment, significant improvement was seen in pain, swelling of peripheral joints and ESR (90.3 to 50.6 mm/lst hour). The least satisfactory results were obtained with three patients who had amyloidosis. Side effects included erythrocyturia (n=l) and leucopenia (n=l)74. Similar efficacy was observed in six patients (5 AS, 1 RS), treated with 200-400 mg CCP daily75. However, a lack of efficacy was found in another study of seven AS patients, treated with CPP 200 mg daily up to a total dose of 6 to 8 g, followed by 100 mg daily for 40 weeks76. CCP was used as a local intraarticular treatment for ten patients with peripheral AS. Joints were injected on average 2.6 times. Efficacy was seen in eight patients, although 4 of them relapsed77. CCP combined with steroid treatment was found to be beneficial for articular and extraarticular symptoms, in a 29-year old male AS patient with progressive cardiac failure due to aortitis78.

6-Mercaptopurine A short course of 6-mercaptopurine was efficious in one RS patient with extensive skin lesions33. Azathioprine A 16-week placebo-controlled cross-over study was performed of treatment with azathioprine (AZA) of eight patients with RS, using AZA 1 mg/kg in the first month and 2 mg/kg in the second month. The joint score was reduced during the AZA treatment and increased during the placebo period. NSAJD use could be reduced during AZA therapy. Two patients withdrew, one because of nausea and one because of a lack of effect (placebo)79. In one open uncontrolled study of nine patients with AS who had not responded to NSAIDs, AZA, administered at 200 mg per day up to a total dose of 6 to 8 g, followed by 100 mg daily, was effective. All variables improved. No side effects were mentioned75. In two cases of RS treatment with AZA has been reported, both were fatal due to complications of the disease (amyloidosis) and treatment (infections)64,67.

79 Chapter VI

Levamisole Randomized placebo-controlled studies of treatment with levamisole of 22 patients with AS and 31 patients with RS have been reported. Both studies were conducted over 24 weeks. The dose of levamisole varied from 50 to 150 mg on 2-3 consecuti­ ve days weekly. Early morning stiffness and pain were improved significantly within three months in the group treated with levamisole, as was spinal mobility in a minority of patients80·81. Nine of 53 patients had to terminate treatment due to a rash and gastrointestinal symptoms, liver function disturbances, severe leucopenia (n=2), and agranulocytosis (n^)80,81. Results were contradictory in two open studies, in which leucopenia and allergic skin reactions were observed as side effects'2·". Levamisole was used to treat three patients with RS: the dose was 50 mg on two consecutive days per week in two patients and 150 mg daily in the third. All three patients experienced an almost complete remission of arthritis, ocular, mucocutaneous and urethric symptoms"4. A relation between agranulocytosis and HLA-B27 during levamisole therapy was observed in one study in RA", which was not confirmed by another study86. The high frequency of HLA-B27, on average 75-80%87'89, might suggest an increased risk of severe side effects. However, agranulocytosis was seen only once in the reports available.

Other treatments

Methylprednisolone Recently a double-blind study was published in which 17 AS patients were treated with two different doses of methylprednisolone (MTP) i.V.: 375 and 1000 mg. MTP was administered on three consecutive days and patients were monitored for up to 180 days. No significant differences were observed between the two treatment groups, although there did appear to be a trend to a more marked and longer effect in the high-dose group88. Four open uncontrolled studies with MTP on a total of 21 AS patients (5 peripheral AS) have been reported89"92. In general, a reduction of pain and stiffness was seen after one pulse lasting for up to six weeks. Treatment with several pulses led to improvement in most variables, including pain, stiffness, spinal mobility, ESR, acute phase proteins, IgA and IgM, although not all changes were statistically significant. Efficacy was not influenced by more than three pulses in one course90. One patient with peripheral AS withdrew due to a lack of efficacy91; the effects on peripheral joint activity was not described separately in those patients. Overall, no serious side effects were observed, although one patient developed melena six weeks after MTP treatment with concomitant administration of indome- thacin'8, and two patients experienced attacks of tachycardia, not related to dose88.

80 2nd-line treatment in SSpA

Table 2. X-ray treatment in ankylosing spondylitis

η % of patients follow- duration of improve­ remarks improved up ment (years) Smyth 52 72% 'placebo-controlled' 1941"" Kuhns 98 85% only 20 patients of case-control study of 1946107 370 relapsed 370 arthritic patients Toone 28 68% 1948"" Richmond 160 96% early AS best results 1950"" Desmarais 70 64% - 95% 'placebo-controlled', 1953'°° different doses used study in 102 arthritic patients Sharp 332 70% 56 patients(± 20%) atypical AS did not 19541И relapsed within 6 respond well years Wilkinson 200 84% 15 after б months: 84% spondylitics with more 1958"" after 1 year 43% extensive disease and after 10 years: 22% especially peripheral AS did not respond well

Fulton 573 63% >5 after 5 years: different doses used, 1961'" high dose: 62% different sizes of radia­ low dose: 31% tion fields early AS best results Rosen 68 58% >4 after 1 year: 62% 1962" after 3 year 25%

Sambrook 110 76% > 10 < 10 year 50% had to cross-over with drug 1963"2 be treated again treatment

Mason 243 90% >5 1 - 5 years: 88% relief 1964'04 of symptoms Kinsella 14 the majority less than 2 years matched controls were 1966'» studied Hohl 33 93% 1 88% 1969"° Schuler 200 90% > 11 > 1 year 74% a subgroup of 22 pa­ 1969"" > 2 years: 34% tients was studied for > 3 years: 12% up to 2 years before starting radiation therapy Sinclair 157 95% 2-22 1971'"

81 Chapter VI

Corticotrophin Coiticotrophin (ACTH) has been reported not to be effective in a placebo-control­ led study of 21 patients with AS93. Patients were admitted for two weeks of bedrest, postural and mobilizing exercises and daily i.m. injections of ACTH (20 ГО for seven days, 10 Ш for five days) or placebo (0.5 ml normal saline). Clinical assessments on admission, immediately after discharge and two months thereafter furnished no evidence of advantage to ACTH administration. Side effects were seen in five patients, comprising a weight gain of at least 2.5 kg (n=l), acne (n=l) and reactivation of genital herpes (n=l). The drug treatment could be maintained throughout the study.

Radiation therapy

Several forms of radiation have been used therapeutically in AS, i.e. roentgen therapy, Thorium X, and local radiation therapy. These treatments will be discussed briefly below (we will not discuss use of radioactive baths).

Roentgen therapy Use of radiotherapy in AS was quite frequent in the 1940s, but has been replaced by analgesics and anti-inflammatory drugs. Only open uncontrolled studies have been published. Roentgen therapy was finally abandoned because of serious long- term side effects, particularly an increased risk of malignancies'4'9'. Most studies are thus somewhat dated and describe patients, methods and results in a different way than is common today. All studies report a general improvement, defined by reduction of pain and stiffness, improvement of spinal mobility and chest expansi­ on, a drop of ESR and a reduced consumption of analgesics (see Table If2·100·113.

Thorium X (224Radium) Thorium X, a disintegration product of thorium, has been used in the treatment of AS. Its half-life is 3.85 days and the emitted radiation is largely made up of alpha particles (92%), the remainder being beta and gamma rays114. Hemaman-Johnson reported two AS cases which were treated successfully with i.v. Thorium X115. In Germany the isotope has played an important role in the treatment of AS. Koch found an improvement in 120 patients with AS, including a decrease in ESR and increase of chest expansion. One year after treatment 75% of patients were still free of symptoms"6. Twenty years after treatment these patients were studied retrospectively. Compared with a group of patients who had not received thorium X treatment, efficacy seemed to favor the Thorium X-treated group117, confirmed in other studies on a total of 247 AS patients118'123. No data are presented concerning long-term side effects, and treatment was abandoned.

82 2nd-line treatment in SSpA

Local radiation therapy Local irradiation has been used as therapy in SSpA and HLA-B27 related arthropa­ thies. Improvement was observed in IS out of 21 patients treated; 13 patients experienced complete remission, with two partial responders. Follow-up ranged from two to 43 months124,125.

Discussion

Our review of the literature on second-line treatment in seronegative spondyl- arthropathies has yielded a large number of case reports, but only a few double- blind (placebo-controlled) clinical trials. No definite conclusions can thus be drawn from the available literature, other than the established efficacy of SASP in AS. Furthermore, many of the reports have been biased, as unsuccessful treatments ('negative results') often remain unpublished. In addition, many studies may include a type Π statistical error as the number of patients often was small126. Of all the treatments, only SASP has been studied in sufficient detail to allow conclusions to be drawn. Most35"3*'*0 placebo-controlled double-blind studies of AS indicated improvement in the SASP-treated group. In general no significant difference was seen between peripheral and axial AS, although efficacy was more obvious in peripheral AS. In the studies performed by Kirwan et al41 and Corkill et al38 no beneficial effects in spinal mobility were found during a three-year study, in contrast with an improvement observed in peripheral arthritis; however, all patients had long-standing AS (mean disease duration 20 years). Improvements were primarily observed mainly in pain and severity and duration of morning stiffness. Chest expansion was the only objective variable which was improved in most studies. A meta-analysis43 of five placebo-controlled double-blind studies35'39 allows the conclusion to be drawn that SASP is a safe and effective treatment for AS, albeit disease duration was not evaluated in this meta-analysis. In RA it was found that responders to SASP therapy had significantly shorter disease duration than nonresponders127. Although this observation has not been firmly demonstrated in AS, a trend was seen in that patients with early AS were more likely to respond to SASP than patients with longstanding AS12". In severe RS, methotrexate was found to induce an improvement in arthritis and in mucocutaneous lesions58'73. In AS, MTX was reported to be effective in 13 severe cases54"56. Only a small number of side effects were reported, all resolving after discontinuation of the drug or dose reduction. Methylprednisolone pulse therapy, when administered as three pulses, was reported to be effective in AS88"92. Levamisole was beneficial in AS, but led to intolerable number and severity of side effects8*82. It can be concluded that the efficacy of SASP has been sufficiently established in the treatment of AS, that this will be the drug of first choice, if treatment with

83 Chapter VI

NSAIDs fails. Among other second-line drags in the treatment of SSpA, MTX and AZA show promise, but no other conclusions can be drawn. Due to its severe long- term side effects, radiation therapy has been abandoned, but it might perhaps play a role as a local therapy for refractory entheseopathies124,125. In view of the findings in respect of radiation therapy, local radiation therapy should be reserved for cases of severe complications of SSpA, e.g. spondylodiscitis or enthesitis, which fail to respond to all other treatments. Арал from this, radiosynovio-orthesis in chronic and refractory peripheral arthritis may be introduced in the same way as it is used in RA. Some goals of the treatment of chronic diseases, including the reduction of symptoms and joint destruction, improvement of functional capacity and quality of life, might be achieved using second-line treatment in SSpA. The final balance between efficacy and toxicity can only be determined by studies which still remain to be performed.

Acknowledgements

We would like to thank Professor T. Pincus for critical reviewing of the manu­ script.

References

1. Wright V. Seronegative polyarthritis. A unified concept. Arthritis Rheum 1978;21:619-33 2. Dougados M, Linden vd S, Juhlin R et al. The European spondylarthropathy study group; preliminary criteria for the classification of spondylarthropathy. Arthritis Rheum 1991;34:1218-27 3. Calin A. Spondylarthropathies. Grane & Stratum, Florida, Inc. Orlando, 1984 4. Ball J. Enthesopathy of rheumatoid and ankylosing spondylitis. Ann Rheum Dis 1971;30:213-23 5. Bellamy N. Prognosis in the rheumatic diseases. Kluwer Academic Publishers, Dordrecht, the Netherlands, 1991 6. Black RL, O'Brien WM, Van Scott EJ et al. Methotrexate therapy in psoriatic arthritis. JAMA 1964;189:743-7 7. Kersley OD. Amethopterin (methotrexate) in connective tissue disease - psoriasis and polyarthritis. Ann Rheum Dis 1968;27:64-6 8. Maldyk H, Chwalinska-Sadowska H. Die Ergebnisse der endoxan-therapie bei arthropathia psoriatica und rheumatoider arthritis. Med Welt 1970;6:236-40 9. Feldges DH, Barnes CG. Treatment of psoriatic arthropathy with either azathioprine or metho­ trexate. Rheumatol Rehab 1974;13:120-4 10. Roux H, Maestracci D, Recordicr AM. D-pénicillamine et polyarthrites psoriasiques. Nouv Pr Med 1975;4:1133 11. Baum J, Hurd E, Lewis D et al. Treatment of psoriatic arthritis with 6-mercaptopurine. Arthritis Rheum 1973;16:139-47 12. Graf U, Marbet U, Müller W et al. Cyclosporin A - Wirkungen und Nebenwirkungen bei der Behandlung der chronischen Polyarthritis und der Psoriasisarthritis. Immun Infect 1981 ;9:20-8 13. Kragballe K, Zachariae E, Zachariae H. Methotrexate in psoriatic arthritis: a retrospective study. Acta Dermatovener 1983;63:165-7

84 2nd-line treatment in SSpA

14. Willkens RF, Williams Ш, Ward JR. et al. Randomized, double-blind, placebo controlled tnal of low-dose pulse methotrexate in psoriatic arthritis. Arthritis Rheum 1964,27 376-81 15. O'N Daunt D, Cox NL, Robertson JC et al. Indices of disease activity in psoriatic arthritis J Royal Soc Med 1987,80 556-8 16. Salvanne С, Zizzi F, Macchioni Ρ et al. Clinical response to auranofin in patients with psoriatic arthritis. Clin Rheumatol 1989,8-54-7 17. Newman ED, Perruquet JL, Harrington TM. Sulfasalazine therapy in psoriatic arthritis· clinical and immunologic response. J Rheumatol 1991,18-1379-82 18 Hehwell P, Marchesoni A, Peters M et al. A re-evaluation of the osteoarticular manifestations of psonasis. Br J Rheum 1991,30-339-45 19. Gladmann DD, Shuckett R, Russell ML et al. Psoriatic arthritis- an analysis of 220 patients. Q J Med 1987,62-127-41 20. Moll JMH, Wnght V. Psoriatic arthritis. Seminars Arthr Rheum 1973,3-55-78 21. Torre Alonso JC, Rodriguez Perez A, Ambas Castnllo JM et al. Psoriatic - a clinical, immunologi­ cal and radiological study of 180 patients Br J Rheum 1991,30 245-50 22. Science Citation Index 1990 23. Giordano M. Dauerprophylaxe der rezidivierende 'Spondylitischen Iridocyclitis' durch Antimalari­ ca und nichtsteroidale Antiphlogistica. Ζ Rheumatol 1982,41 105-6 24 Steven MM, Morrison M, Sturrock RD. Penicillamine in ankylosing spondylitis- a double blind placebo controlled tnal. J Rheumatol 1985,12-735-7 25 Bemacka K, Tytman K, Sierakowski S Clinical application of D-penicillamine m ankylosing spondylitis a 9-month study Rev Roum Méd Int 1989,27-295-301 26. Tytman К, Bemacka К, Sierakowski S. D-pemcillamine in the therapy of ankylosing spondylitis Qui Rheumatol 1989,8-419-20 27. Goldmg DN. D-penicillamme in ankylosing spondylitis and polymyositis. Postgrad Med 1974,(suppl.)-62-4 28. Golding DN. D-pemcillamine m spondylitis and sacroilutis, a preliminary study. Scand J Rheum 1975,4(suppl 8) E21-18 29. Scharf Y, Nahir M Penicillamine in ankylosing spondylitis. Arthritis Rheum 1976,19*122 30. Bird HA, Dixon ASJ. Failure of d-pemcillamme to affect peripheral joint involvement m ankylosing spondylitis or HLA-b27-associaled arthropathy Ann Rheum Dis 1977,36 289 31. Jaffe IA. Penicillamine in seronegative polyarthritis. Ann Rheum Dis 1977,36-593-4 32. Rosen PS, Graham DC. Ankylosing (Strtlmpell-Mane) spondylitis (a clinical review of 128 cases) A lit 1962,5 158-233 33. Smith DL, Bennett RM, Regan MG Reiter's syndrome in women. Arthritis Rheum 1980, 23 335- 40 34. Grasedyck K, Schattenkirchner M, BandiUa К. Zur Behandlung der Spondylitis ankylosans mit Auranofin (Ridaura) Ζ Rheumatol 1990,49-98-9 35. Dougados M, Boumier Ρ, Amor В. Sulfasalazine in ankylosing spondylitis a double blind controlled study in 60 patients. Br Med J 1986,293-911-4 36. Feltehus N, Hallgrcn R. Sulfasalazine in ankylosing spondylitis. Ann Rheum Dis 1986, 45 396-99 37. Nissila M, Lehtinen K, Leinsalo-Repo M et al. Sulfasalazine in the treatment of ankylosing spondylitis, a twenty-six-week, placebo-controlled clinical trial Arthritis Rheum 1988,31 1111-6 38. Davis MJ, Dawes FT, Bcswick E et al Sulfasalazine therapy m ankylosing spondylitis its effect on disease activity, immunoglobulin A and the complex immunoglobulin A-alpha-1 -antitrypsin Br J Rheum 1989,28 410-3 39. Corkill MM, Jobanputra P, Gibson Τ et al. A controlled tnal of sulfasalazine treatment of chronic ankylosing spondylitis failure to demonstrate a clinical effect Br J Rheum 1990,29 41-5 40. Taylor HG, Bcswick EJ, Dawes PT Sulfasalazine m ankylosing spondylitis A radiological, clinical and laboratory assessment Clin Rheumatol 1991,10 43-8 41. Kirwan J, Edwards E, Huitfeld В et al The course of established ankylosing spondylitis and the effects of sulfasalazine over 3 years Br J Rheum 1993,32 729-33

85 Chapter VI

42. Fraser SM, Sturrock RD. Evaluation of sulfasalazine in ankylosing spondylitis - an interventional study. Br J Rheum 1990;29:37-9 43. Bosi Ferraz M, Tugwell P, Goldsmith CH et al. Meta-analysis of sulfasalazine in ankylosing spondylitis. J Rheumatol 1990;17:1482-6 44. Mielants H, Veys EM, Joos R. Sulfasalazine (ealazopyrin) in the treatment of enterogenic reactive synovitis and ankylosing spondylitis with peripheral arthritis. Clin Rheumatol 1986;5: 80-3 45. Stroehmann I, Wuestenhagen E, Martini M. Die Therapie seronegativer Oligoarthritiden mit Salazopyrin. Ζ Rheumatol 1987;46:79-82 46. Amor В, Khan A, Dougados M et al. Sulfasalazine and ankylosing spondylitis. Ann Interri Med 1984:101:878 47. Zwillich SH, Comer SS, Lee E et al. Treatment of the seronegative spondyloarthropathies with sulfasalazine. J Rheumatol 1988;15(suppl 16):33-9 48. Egsmose, C, Hansen TM, Andersen L et al.. Sulfasalazine in the treatment of reactive arthritis. A 6 months, placebo-controlled trial. Rev Esp Reumatol 1993;20(suppl 1):490 49. Tmavsky K, Peliskova Z, Vacha J. Sulfasalazine in the treatment of reactive arthritis. Scand J Rheum 1988;67(suppl):76-9 50. Peliskova Z, Pavelka К jr., Tmavsky K. A placebo controlled, double-blind study of salazopyrin- EN in refractory reactive arthritis. Scand J Rheum 1990;(suppl 80):45 51. Mielants H, Veys EM. HLA-B27 related arthritis and bowel inflammation. Part I. Sulfasalazine (Salazopyrin) in HLA-B27 related reactive arthritis. J Rheumatol 1985;12:287-93 52. Dougados M, Linden van der S, Leirisalo-Repo M et al. Sulfasalazine in spondylarthropathy: a randomized multicentre double-blind placebo-controlled study. Rev Esp Reumatol 1993;20 (suppl 1):165 53. Dougados M, Nguyen M, Mijiyawa et al. Sulfasalazine in spondylarthropathies. Arthritis Rheum 1989;32:S38 54. Sampaios-Barros PD, Costallat LTL, Femandes SRM et al. The use of methotrexate in the treatment of ankylosing spondylitis: pilot study. Rev Esp Reumatol 1993;20 (suppl 1):485 55. Handler RP. Favorable results using methotrexate in treatment of patients with ankylosing spondylitis. Arthritis Rheum 1989;32:234 56. Bosi Ferraz M, Canica da Silva H, Atra E. Low dose methotrexate with leucovorin rescue in ankylosing spondylitis. J Rheumatol 1991;18:146-7 57. Perry HO, Mayne JG. Psoriasis and Reiter's syndrome. Arch Dermatol 1965;92:129-36 58. Mullins JF, Maberry JD, Stone OJ. Reiter's syndrome treated with folic acid antagonists. Arch Dermatol 1966;96:335-40 59. Färber GA, Forshner JG, O'Quinn SE. Reiter's syndrome, treatment with methotrexate. JAMA 1967;200:181-3 60. Shcremata WA. Reiter's disease. A case with severe and unusual complications. Med Serv J Can 1976;23:1216-32 61. Jetton RL, Duncan WC. Treatment of Reiter's Syndrome with methotrexate. Ann Intern Med 1969;70:349-51 62. Topp JR, Fam AG, Harg GS. Treatment of Reiter's syndrome with methotrexate. Can Med Assoc J 1971;105:1168-70 63. Szanto M, Rubinstein H. Reiter's syndrome treated with methotrexate. JJlin Med J 1972; 142:467- 9 64. Paulus HE, Pearson CM, Pitts W. Aortic insuffiency in five patients with Reiter's syndrome. A detailed clinical and pathologic study. Am J Med 1972;53:464-72 65. Caugbey DE, Wakem CJ. A fatal case of Reiter's disease complicated by amyloidosis. Arthritis Rheum 1973;16:695-700 66. Chu SM. Reiter's syndrome - treatment with methotrexate. Sing Med J 17; 101-3, 1976 67. Chee YC, Chan HL. Reiter's disease, responding to methotrexate. Sing Med J 1977;18:136-8 68. Jaramillo D, Lion W, Cardenas V et al. Reiter's syndrome, immunosuppression and strongyloidia­ sis. Report of a fatal case. J Cutan Pathol 1978;5:200-8 69. Owen ET, Cohen ML. Methotrexate in Reiter's disease. Ann Rheum Dis 1979;38:48-50

86 2nd-line treatment in SSpA

70. Luderschmidt Ch, Balda B-R. Morbus Reiter. Erfolgreiche Kombinations-Chemotherapie mit Methotrexat, Etrinat und Prednisolon. MUnch med Wschr 1983;125:936-40 71. Lally EV, Ho G jr. A review of methotrexate therapy in Reiter Syndrome. Seminars Arthr Rheum 1985;15:139-45 72. Curran JJ, Jamieson TW, Denis C. Therapeutic utility of pulse methotrexate in Reiter's syndrome. Arthritis Rheum 1987;30(suppl 4):S75 73. Allen DE, Kaplan B, Pinnell SR. Levamisole and skin disease, ht J Dermatol 1978;17:287-300 74. Sadowska-Wroblewska M, Oarwolinska H, Maczynska-Rusinak B. A trial of cyclophosphamide in ankylosing spondylitis with involvement of peripheral joints and high disease activity. Scand J Rheum 1986;15:259-64 75. Miehlke K, Kafamik U. Klinische Erfahrungen mit Isophosphamid in der Behandlung der chronischen Polyarthritis. Verh Dtsch Oes Rheumat 1972;2 (suppl 2):415-9 76. Fricke R, Petersen D. Behandlung der Ankylosierenden Spondylitis mit Cyclophosphamid und Azathioprin. Verh Dtsch Ges Rheumatol 1969;1:189-95 77. Häntzschel H, Reinelt D, Otto W. Praktische Erfahrungen mit der intraartikulären Behandlung chronischer rheumatischer Erkrankungen. Ztschr Inn Med 1975;30:698-701 78. Townend JN, Emery Ρ, Davies MK et al. Acute aortitis and aortic incompetence due to systemic rheumatologic disorders. Int J Cardiol 1991;33:253-8 79. Calin A. A placebo controlled, crossover study of azathioprine in Reiter's syndrome. Ann Rheum Dis 1986;45:653-6 80. Goebel KM, Goebel FD, Schubert H. Levamisole-induced immuno-stimulation in spondyl- arthropathies. Lancet 1977;ii:214-7 81. Christensen KD. Treatment of seronegative spondylarthritis with levamisole: a double-blind placebo-controlled study. Int J Immunopharmac 1979;1:147-50 82. Ebner W, Senautka G. Therapie der Spondylitis ankylopœtica mit levamisole. Verhalten klinischer und immunologischer Parameter. Fortschr Med 1983;101:364-8 83. Trabert U, Rosenthal M, Müller W. Therapie entzUndlich-rheumatischer Krankheiten mit Levamisol, einer immunomodulierenden Substanz. Schweiz Med Wschr 1967;106:1293-1301 84. Ippen H, Qadripur DS. Levamisol zur Behandlung von Hautkrankheiten. Dtsch Med Wschr 1975;100:171-80 85. Schmidt KL, Müller-Eckhardt C. Agranulocytosis, levamisole and HLA-B27 in rheumatoid arthritis. Lancet 1977;ii:85 86. Veys EM, Mielants H, Rosenthal M. Agranulocytosis, levamisole, and HLA-B27. Lancet 1977;ii:764 87. Brewerton DA, Caffrey M, Hart FD. Ankylosing spondylitis and HLA-B27. Lancet 1973;i:904-7 88. Peters ND, Ejstrup L. Intravenous methylprednisolone pulse therapy in ankylosing spondylitis. Scand J Rheum 1992;21:134-8 89. Mintz G, Enriquez RD, Mercado U. Intravenous methylprednisolone pulse therapy in severe ankylosing spondylitis. Arthritis Rheum 1981;24:734-6 90. Richter MB, Woo P, Panavi GS et al. Pulse methylprednisolone in ankylosing spondylitis. Arthritis Rheum l982;25(suppl):S13 91. Richter MB, Woo P, Panavi GS et al. The effects of intravenous pulse methylprednisolone on immunological and inflammatory processes in ankylosing spondylitis. Clin and Exp Immunol 1983;53:51-9 92. Ejstrup L, Peters ND. Methylprednisolone pulse therapy in ankylosing spondylitis. Dan Med Bull 1985;32:231-3 93. Wordsworth BP, Pearcy MJ, Mowat AG. In-patient regime for the treatment of ankylosing spondylitis: an appraisal of improvement in spinal mobility and the effects of corticotrophin. Br J Rheum 1984;233:39-43 94. Court Brown WM, Abbatt JD. The incidence of leukaemia in ankylosing spondylitis treated with X-rays. A preliminary report. Lancet 1955;i: 1283-5 95. Graham DC. Leukemia following X-ray therapy for ankylosing spondylitis. Arch Intern Med 1960;105:51-9

87 Chapter VI

96. Court Brown WM, Doll R. Mortality from cancer and other causes after radiotherapy for ankylosing spondylitis. Br Med J 1965,December: 1327-32 97. Kaprovc RE, Little AH, Graham DC. Ankylosing spondylitis. Survival in men with and without radiotherapy. Arthritis Rheum 1980;23-57-61 98. Smith PG, Doll R. Mortality among patient with ankylosing spondylitis after a single treatment course with X-rays. Br Med J 1982,284:449-60 99. Darby SC, Doll R, Gill SK. Longterm mortality after a single treatment course with X-rays in patients treated for AS. Br J Cancer 1978,55; 179-85 100. Desmarais MHL. Radiotherapy in arthritis. Ann Rheum Dis 1953;12:25-8 101. Wilkinson M, Bywaters EGL. Clinical features and course of ankylosing spondylitis. Ann Rheum Dis 1958,17-209-28 102. Sharp J, Easson EC. Deep X-ray therapy in spondylitis. Br Med J 1954;13:619-23 103. Schuier В, Dihlmann. Ergebnisse der Roentgentherapie bei Ankylosierender Spondylits. Vcrh Dtsch Ges Rheumat 1969;1:124-32 104. Mason RM. Spondylitis. Proc R Soc Med 1964,57-533-40 105. Smyth CJ, Freyberg RH, Peck WS. Roentgen therapy for rheumatic disease. JAMA 1941,116: 1995-2001 106. Fulton JS. Ankylosing spondylitis. Clin Radiology 1961;12:132-5 107. Kuhns JG, Morrison SL. Twelve years' experience in roentgenotherapy for chrome arthritis. N Engl J Med 1946,235-399-405 108. Toone EC. Rheumatoid spondylitis- observations on the incidence and response to therapy among veterans of the recent war. Ann Intern Med 1949,30-733-9 109. Richmond JJ. The importance of radiotherapy in the treatment of ankylosing spondylitis. Proc R Soc Med 1950,44-443-7 110. Hohl К. Die Strahlenbehandlung des Morbus Bechterew. Praxis 1969,58-579-85 111. Sinclair RJG. Treatment of rheumatic disorders with special reference to ankylosing spondylitis. Proc R Soc Med 1971;64·1031-8 112. Sambrook DK. Radiotherapy for ankylosing spondylitis. A review. Rheumat 1963,19:30-40 113. Kinsella TD, MacDonald FR, Johnson LG. Ankylosing spondylitis- a late re-evaluation of 92 cases. Canad Med Ass J 1966;95:1-9 114. Dehkan O. Preparation of J34Ra for therapy of ankylosing spondylitis. Health Phys 1978,35-21-4 115. Hemaman-Johnson K. Thorium X m spondylitis and chronic rheumatism. Rheumat 1946,3:21-4 116. Koch W, Reskc W. Die Ergebnisse der intravenösen Thorium X-Bchandlung bei der Spondyl­ arthritis ankylopoetica (M. Bechterew). Strahlentherapie 1952;82:439-57 117. Koch W. Ergebnisse einer zwanzigjährigen parenteralen Thorium X Therapie der knöchern- entzündlichen Wirbelsaulcnvcrsteifung (Morbus Bechterew). Vcrh Dtsch Ges Rheumat 1969; 1: 132-40 118. Laschner W. Ergebnisse und Komplikationen der Thonum-X-Behandlung bei M. Bechterew. Ζ Orthop 1973;lll-743-8 119. Hertel E, Heme J. Sakroiliakalcr Strontium-Umsatz und Thorium-X-Therapie beim Morbus Bechterew. Ζ Orthop 1974;112842-5 120. Bertrand A, Legras B, Martin J. Use of Radium-224 in the treatment of ankylosing spondylitis and rheumatoid synovitis. Health Phys 1978;35-57-60 121. Schmitt E. Eine Langzeitstudie zum therapeutische Effekt des Radium-224 beim Morbus Bechterew. Ζ Orthop 1978; 116-621-4 122. Crone-Munzebrock W. Redeker S, Montz R et al. Ergebnisse der Radium 224-Bchandlung bei Patienten mit ankylosierender Spondylitis. Akta Rheumatol 1981;6-100-2 123. Knop J, Stritzke P, Heller M et al. Ergebnisse einer Radium^-therapie der ankylosierenden Spondylitis (M. Stnlmpell-Mane-Bechterew). Ζ Rheumatol 1982,41-272-5 124. Grill V, Smith M, Ahem M et al. Local radiotherapy for pedal manifestations of HLA-B27-rclated arthropathy. Br J Rheum 1988,27-390-2 125. Mantell BS Radiotherapy for painful heel syndrome. Br Med J 1978;2-90-l 126. Laurent MR, Buchanan WW, Bellamy N. Methods of assessment used in ankylosing spondylitis clinical trials, a review. Br J Rheum 1991,30-326-9

88 2nd-line treatment in SSpA

127. Situnayake RD, McConkey B. Clinical and laboratory effects of prolonged therapy with salazopy- rine, gold or penicillamine, the effects of disease duration on treatment response. J Rheumatol 1990; 17:1269-73 128. McConkey B. Sulfasalazine and ankylosing spondylitis. Br J Rheum 1990;29:2-3

89

Chapter VU

Methotrexate in severe ankylosing spondylitis: an open study

MCW Creemers, MJAM Franssen, LBA van de Putte, FWJ Gribnau, PLCM van Riel

Submitted for publication

MTX in severe AS Summary

Efficacy and toxicity of methotrexate (MTX) in ankylosing spondylitis (AS) were studied in a 36 week open, single observer study. Patients were seen 4-weekly and variables were assessed for disease activity and toxicity. Patients, included, had to have evidence for an active disease and failed to response to nonsteroidal antiin­ flammatory drugs (NSAIDs) and sulfasalazine (SASP). MTX was administered weekly at an oral dose of 7.5 to 15 mg. Patients were treated with MTX for at least 24 weeks, while NSAIDs were kept at a stable dose. Efficacy was evaluated in two ways: by calculating the relative difference of assessed variables between weeks 0 and 24; and by asking the patient at week 24 whether he or she would like to continue treatment with MTX. In total, 11 patients entered the study and nine were évaluable at week 24. A large part of assessed variables showed a good rehtive improvement. Four patients decided to continue MTX; three of them could lower the NSAID dose, and one could stop NSAID treatment. The remaining five patients discontinued MTX treatment, of whom three showed a flare and opted to restart MTX. Side effects seen were mild and reversible: abdominal discomfort (n=2), nausea (n=l), transient oral ulcers (n=l) and reversible liverfunction disturbances (n=3).

Introduction

Ankylosing spondylitis (AS) is a chronic inflammatory condition of the axial skeleton and sacroiliac joints that may lead to spinal ankylosis. Peripheral arthritis and extra-articular features occur in a small proportion of cases and are bad prognostic factors1. Up to now treatment of AS mainly comprises nonsteroidal antiinflammatory drugs (NSAIDs). Of second-line drugs, only of sulfasalazine (SASP)2 efficacy has been established in AS. Clinical significant improvement with weekly administered methotrexate (MTX) has been reported in sixteen patients with severe AS3-*. Aim of this study was to determine the efficacy and toxicity of MTX in patients with severe AS in an open, single observer 36-week interventional study.

Patients and methods

The study was conducted at the rheumatology departments of the University Hospital Nijmegen St. Radboud and the St Maartenshospital Nijmegen, the Nether­ lands, and was started after full approval of both ethical committees.

93 Chapter VII

Patients Patients with severe AS, fullfilling the modified New Yoik criteria', of either sex, aged between 18 and 60 years, were selected. They were excluded in case of: 1. serious systemic disease, malignancies, impaired organ function or serious infec­ tions like tuberculosis; 2. mental disorders; 3. alcohol or drug abuse; 4. pregnancy or breast feeding; 5. history of intestinal disease, Reiters' disease. Severe or refractory AS was denned as a failure on treatment with NSAIDs and SASP; besides, active disease had to be present defined as the presence or persistence of at least two of the following features: a. morning stiffness of at least 30 minutes; b. disturbed sleep due to pain and stiffness; с peripheral arthritis; d. ESR £ 30 mm or CRP ¿ 20 mg/L or IgA ¿3,9 mg/L; e. spinal pain; f. stiffness and pain of thorax at movement or during normal breathing; g. pain in both buttocks during the night or the day. A stable NSAID dose and discontinuation of SASP were required, both for at least four weeks at study entry. Adequate anticonception during and six months after MTX treatment was recommended.

Study design The study was an open, single observer, 36-week study in which MTX was studied for efficacy and toxicity. Patients were seen 4-weekly for 36 weeks and variables for disease activity and toxicity were measured every visit. Medication for non- rheumatic chronic conditions and intercurrent acute illnesses was allowed, except for anti-folate drugs (such as sulfonamide derivatives), allopurinol, immunosuppres­ sive treatment and second-line drugs. MTX was administered orally 7.5 mg/week. At week 12 MTX dose was increased to 15 mg/week in case of lack of response. At week 24 efficacy was evaluated based on patients' global assessment: a. if patients judged that MTX was clearly effective, MTX was kept at a stable dose and the NSAID dose was reduced, and b. if patients judged efficacy of MTX moderate or absent (see below), MTX was discontinued and NSAIDs kept at a stable dose. In case of severe side effects MTX was stopped, in milder cases the dose was decreased or 1 mg folic acid was administered1"10.

Measurements Every visit the following variables were assessed: a) Objective clinical variables: occiput-wall-distance, chest expansion, Schober's 10 cm test", fingertip-to-floor distance, an enthesis index (EI)12, the number of swollen joints, Ritchie Articular Index (RAI)13. Extra-articular manifestations were carefully sought for, b) Subjec­ tive variables: Visual Analogue Scales (VAS) ranging from 0 to 100 mm, corres­ ponding with 'none', c.q. 'very good' to 'has never been worse' were used for spinal pain, chestpain, general wellbeing and tiredness. Next to this, duration of morning stiffness after arising was assessed; c) Functional assessment: a Dutch Functional Index (DFI) for AS14, in essence a modification of the Functional Index of Dougados et al1516; d) Laboratory variables: HLA-B27, ESR, C-reactive protein, complete blood cell count, serum Creatinin, liverfunction tests, albumin, immuno- globin A, G and M, Creatinin Phosphokinase; e) Every visit side effects related to

94 MTX in severe AS

Tabel 1. Patients' characteristics at start of study

number of patients

male : female 8 : 3 HLA-B27 positivily 11 peripheral arthritis 3 enthcsitis S iridocyclitis 2

years* age 35 ± 8 (21-47) disease duration 14 ± 9 (3-26)

*: mean ± standard deviation (range) study medication were assessed by direct questioning or as spontaneously reported complaints. Response The relative difference between weeks 0 and 24 was calculated, and response was defined as follows: good, if improvement ~¿ 50% was present in the majority of variables; moderate, if this improvement was between 15 to 50% and no response in case changes were smaller or absent.

Results

In total, 11 patients were enrolled to the study; all had NSAID failure, nine failed to respond on S ASP treatment and 2 patients could not tolerate S ASP. Patients' characteristics at start of the study are shown in Table 1. After 12 weeks of treatment in nine patients MTX was increased to 15 mg weekly. Generally, response, if present, could be observed between four to 12 weeks of MTX treat­ ment. Two patients dropped out at week 16 because of protocol violating changes in treatment: one due to a flare of the disease, the other due to development of a peptic ulcer, for which oral steroids were stopped as well. After 24 weeks nine patients were évaluable with a mean weekly dose of MTX of 13.3 mg over a period of 24 weeks. Improvement could be seen in most assessed variables (see Table 2 and Figure 1). C-reactive protein and ESR improved equally. Immunoglobin A, elevated in 4 patients, decreased. Anaemia (hemoglobin = Hb: 5.7 to 6.7 mmol/L), present in all three patients with peripheral arthritis, improved at least 1.0 mmol/L. The course of iridocyclitis (3 episodes) appeared not to be influenced by MTX. Efficacy was good in 5 patients: four of them continued MTX treatment and the dose of the NSAID could be reduced (n=3) or stopped (n=l). The fifth patient stopped MTX treatment despite good clinical response, but deteriorated within four weeks for which MTX was readministered. In the remain-

95 Chapter VII

Figure 1. Response of the 9 évaluable patients

spinal pain (VAS) chest expansion

pain (mm VAS)_

lune (weeks)

functional index ritchie articular index

96 MTX in severe AS

Table 2. Objective evaluation of MTX treatment: relative improvement in variables*

variables patients' number 1 3 4 5 6 7 8 9 10

disease characteristics* Ρ Ρ Ρ I

spinal pain ++ + ++ . + + - - ++ chest pain ++ ++ - ++ - ++ - - ++ general wellbeing ++ + ++ ++ + ++ - - - tiredness ++ ++ ++ + - - - - - morning stiffness ++ ++ ++ ++ - - - - ++

occiput-to-wall distance + ++ ++ NA NA _ NA NA NA fingertip-to-floor distance ++ - - ++ - - - - ++ chest expansion ++ - ++ ++ - - - ++ + Schobers' 10 cm test ++ ++ ++ ++ - ++ - - ++

Ritchie Articular Index ++ ++ NA ++ + ++ NA _ NA number of swollen joints NA + NA ++ NA ++ NA NA NA enthcsis index ++ ++ ++ ++ - ++ - - NA

Dutch Functional Index ++ - ++ ++ ++ + - - -

ESR ++ + . ++ NA + NA _ .

": relative improvement: ++ improvements 50%; + improvement between 15 and 49%; - no improvement or less than 15%; NA = not affected; : Ρ s peripheral arthritis; I s iridocyclitis. ing four patients MTX was stopped, because of moderate or absent efficacy, and two of them deteriorated within four weeks. Side effects were abdominal discomfort (n=2), nausea (n=l), transient oral ulcers (n=l), reversible liverfunction elevations (n=3), for which two patients experienced benefit of folic acid 1 mg.

Discussion

MTX was studied in severe ankylosing spondylitis (AS) in a 36 week open study. Regarding enthesitis, peripheral arthritis and duration of the disease patients were quite different. Data were not analysed statistically because of this heterogeneity and the small sample size. Evaluation of efficacy was evaluated in two different ways: objective, i.e. the relative difference of assessed variables between weeks 0 and 24 (see Table 2), and subjective by deciding at week 24 to either continue (n=4) or stop MTX (n=5) based on patients' global assessment. One patient could stop NSAID treatment, three could reduce the dose of NSAIDs, and four patients deteriorated after discontinuation of MTX and opted to restart. Side effects were

97 Chapter VII reversible or transient, and no severe side effects were seen. There was no obvious difference in response between patients with peripeihal arthritis and patients with only involvement of the axial skeleton. Thus, although numbers were small, this study provides circumstantial evidence that the majority of patients experienced beneficial effects due to MTX. In the future, a double-blind placebo-controlled study should further corroborate these data.

References

1. Carette S. The natural disease course of ankylosing spondylitis. Arthritis Rheum 83;26t 186-90 2. Bosi Ferraz M, Tugwell P, Goldsmith CH, Atra E. Meta-analysis of sulfasalazine in ankylosing spondylitis. J Rheumatol 1990;17:1482-6 3. Bosi Ferraz M. Canica da Silva H, Atra E. Low dose methotrexate with leucovorin rescue in ankylosing spondylitis. J Rheumatol 1991;18:146-7 4. Handler RP. Favorable results using methotrexate in treatment of patients with ankylosing spondylitis. Arthritis Rheum 1989;32:234 5. Yamane K, Saito С, Natsuda H et al. Ankylosing spondylitis successfully treated with methotrexate. Intern Med 1993;32:53-6 6. Sampaios-Barros PD, Costallat LTL, Femandes SRM et al. The use of methotrexate in the treatment of ankylosing spondylitis: pilot study. Rev Esp Reumatol 1993;20 (suppl. 1):485 7. Van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic criteria for ankylosing spondylitis; a proposal for modification of the New York criteria. Arthritis Rheum 1984;27:361-8 8. Tishler M, Caspi D. Fishel B, Yaron M. The effect of leucovorin (folinic acid) on methotrexate therapy in rheumatoid arthritis patients. Arthritis Rheum 1988;31:906-8 9. Morgan S, Bagott JE, Vaughn WH, Young PK, Austin JV, Krumdieck CL. The effect of folic acid supplementation on the toxicity of low-dose methotrexate in patients with rheumatoid arthritis. Arthritis Rheum 1990;33:9-18 10. Joyce DA, Will RK, Hoffman DM, Laing B, Blackbourn SJ. Exacerbation of rheumatoid arhtritis in patients treated with methotrexate after administration of folinic acid. Ann Rheum Dis 1991:50:91- 34 11. Schober P. Lendenwirbelsaule und Kreuzschmerzen. Muench Med Wschr 1937;84:336 12. Mander MM, Simpson JM, McLellan A, Walker D, Goodacre JA, Dick WC. Studies with an enthesis index as a method of clinical assessment in ankylosing spondylitis. Ann Rheum Dis 1987;46:197-202 13. Ritchie DM, Boyle JSA, Mclnnes JM, Jasani MK, Dalakos TG, Grieveson P, Buchanan WW. Clinical studies with an articular index for the assessment of joint tenderness in patients with rheumatoid arthritis. Q J Med 1968,147:393-406 14. Crecmcrs MCW, Hof van 't MA, Franssen MJAM et al. A Dutch version of the Functional Index for ankylosing spondylitis: development and validation in a long-term study. Br J Rheum 1994;33: 842-6 15. Dougados M, Gueguen A, Nakache J-P, Nguyen N, Mery C, Amor B. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1988;15:302-7 16. Dougados M, Gueguen A, Nakache J-P, Nguyen N, Mery C, Amor B. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1990;17:1254-55

98 Chapter Ш

A Dutch version of the functional index for ankylosing spondylitis: development and validation in a long-term study

MCW Creemers, MA van 't Hof, MJAM Franssen, LBA van de Putte, FWJ Gribnau, PLCM van Riel

Published: British Journal of Rheumatology 1994;33:842-6

Dutch functional index for AS

Summary

A Dutch Functional Index (DFI) for ankylosing spondylitis (AS) has been develo­ ped, containing 37 questions in essence this is a modification of a French Index with additional questions. Internal consistency, reproducibility, criterion and construct validity and sensitivity to change were studied in different groups of AS patients. In a group of 149 patients the questionnaire was completed and internal consistency calculated (Cronbach's α = 0.94) after exclusion of three items. Reproducibility was studied in 39 patients with stable disease; Pearson correlation was 094. Criterion validity, against a 'gold standard' i.e. experts' observation of 25 items, was studied in 19 patients; r = 0.85. In a 48-week NSAID study patients were enrolled after a washout period: DFI scores before and after treatment showed significant improvement (p<0.02). There were 187 DFI with corresponding clinical measurements, correlations varied from -0.30 to 0.68. Estimated measure- ment-remeasurement correlation was 0.87. The DFI is, thus, a potentially useful instrument, which is valid, reliable and sensitive to the effects of NSAID treatment.

Introduction

In addition to measuring disease activity, assessment of functional status can be a useful evaluative technique for arthritis outcome. In RA questionnaires have been developed1"3, focusing on the impairment caused by peripheral arthritis. These questionnaires proved to be valid, reliable and sufficiently sensitive to detect functional change and they have therefore been used to evaluate treatment effects in clinical trials4. Questionnaires for RA cannot be used in AS, as the latter affects mainly the axial skeleton with only a minority of cases experiencing peripheral arthritis. This prompted us to search for a questionnaire dealing specifically with problems experienced by AS patients. A literature search for questionnaires already available and evaluated revealed three approaches in this field: 1. the Ankylosing Spondylitis Assessment Question­ naire (ASAQ), a self-administered questionnaire focusing on spinal mobility and pain9; 2. a Functional Index (FI) based on activities of daily living and scored by the investigator after verbal response by the patient6,7; 3. the Health Assessment Questionnaire for Spondylarthropathies (HAQ-S), in essence the HAQ for RA with five additional questions about driving a car, sitting and carrying heavy loads'. The FI of Dougados et al.6,7 was most in agreement with the guidelines for the develop­ ment of measurement scales910, but was not a self-administered questionnaire, and criterion validity and sensitivity to change over the long-term had not been studied. Since the French FI had to be translated and subsequently validated for the Dutch situation, items concerning static activities and riding a bike were added in order to improve its completeness.

101 Chapter VIM

Table 1. Items of the Dutch Functional Index

Can you: put on shoes # * pull on a pair of trousers # * pull on a pullover # * get into a bath # @ О get out of a bath # @ О wash yourself or take a shower cough or neezc # breath deeply # * turn your head to the left and to the right * • remain standing for 10 minutes # * О sleep on your back # sleep on your stomach # • walk up one flight of stairs # * remain standing for 30 minutes О run # * go Bitting # * get out of a chair # * get into a car # * О creep # * bend to pick up something from the floor # * lie down # * tum in bed # * get out of bed # * perform your housework # @ do your job # @ go shopping get out of a car * О remain sitting for 10 minutes * О remain sitting for 30 minutes О kneel down * squat * ride on your bike a slight hill * ride a bike for 10 minutes * О ride a bike for 30 minutes О lift up a weight of 10 kg to shoulderheight * О lift up a weight of 10 kg above shoulderheight *0 tum left on a bike without getting off the bike * φ

explanation of symbols: #: item out of the French FI @: item out of the French FI, split up into two separate items *: item observed by professionals (group V) • : item excluded after calculation of internal consistency (group I) O: item checked for inconsistency (group I)

Patients and methods

Construction of the questionnaire The FI of Dougados et al.6,7 was translated into Dutch. The back-translation from English into Dutch was without mistakes and the translation was therefore deemed satisfactory. Two questions from the FI, 'to get into and out of a bath' and 'to do your job and perform your housework', were split up into four separate questions (marked in Table 1). Interviews were held with 12 AS patients, two rheumatolo- gists, three physical therapists, an occupational therapist, a nurse and a psycholo­ gist, all professionals were familiar with the problems specific to AS. From this information new items, covering the total field of problems in AS in daily life, were added to the questionnaire: firstly, activities with a specific duration of time such as to remain sitting for 10 and 30 min as AS patients are known to experience increasing difficulties over time in remaining certain postures; secondly, questions

102 Dutch functional index for AS to make the FI moie relevant to the Dutch situation, e.g. to ride a bike. In total, 15 items were added The content validity was optimized by the two rheumatologists, the physical therapists and ergotherapist by seeing whether important areas of problems in AS were missing. The final Dutch Functional Index (DFI) was self- administered, in contrast to the FI6,7, so it would be less time-consuming and avoid interobserver variability. The questionnaire then comprised of 37 items (see Table 1). Response options were as follows: 'possible' (score = 0), 'possible but with difficulty' (score = 2) and 'not possible' (score = 4). If patients selected more than one answer, scores were adapted as follows: 1 = 'possible' plus 'possible but with difficulty' and 3 = 'possible but with difficulty' plus 'not possible'. At the most three items were allowed to be ommitted, and the score of the final index was calculated as the mean of the item scores, thus correcting for missing items. Questionnaires missing more items were not considered for further calculations.

Patients AS patients, selected for a 48-week clinical trial, comparing naproxen and ß- cyclodextrin-piroxicam (Brenn), and AS patients participating in an AS-exercise group, were used for the validation of the DFI. From the available files at two rheumatology departments 234 AS patients, of either sex, aged between 16 and 60 yr, rullfilling the modified New York criteria11 were selected. They were excluded if they had any evidence of: 1. previously experienced side effects on NSATDs; 2. serious systemic disease; 3. impaired organ function, mental disorders; 4. alcohol or drug abuse; 5. pregnancy or breast feeding. Additionally, there had to be active disease after a washout period, evidenced by pain and stiffness requiring use of NSATDs. The following groups were identified: Group ¡internal consistency): patients at one centre (n = 113) who decided not to participate in the clinical trial were asked to complete the DFI once by return post and to register the time required to complete the questionnare. These question­ naires (n = 100) together with those DFI completed after the washout period in the clinical trial (n = 55) were the basis for a study on (internal) consistency of the questionnaire (n = 15S); Group Rfeproducibility): in addition to the clinical trial, patients from the Nijmegen Region participating in an AS-exercise group (n = 67) were also asked to complete the questionnaire twice if they considered their disease to be stable. From 39 of these patients duplicate questionnaires were received. This group was used to study reproducibility of the questionnaire group (n = 39); Group V(alidity): a subgroup of R agreed to actually perform the activities of the questionnaire for a professional judgement in order to study criterion validity of the questionnaire (n = 19); Group Lfongitidunal study): the 55 patients who entered the clinical trial served as a longitudinal follow-up group to study the sensitivity to change (group L, η = 55), to calculate correlations with other clinical variables and for estimation of measurement-remeasurement correlation.

103 Chapter VIM

Table 2. pairs of items checked for inconsistency (group I)

activity/classification of answers % correct* % equal* % misclassifìcation

to get into and out of a bath 18 76 6 remain standing for 10 and 30 minutes 40 60 0 lift up a weight of 10 kg to an above shoulderheight 26 73 1 remain sitting for 10 and 30 minutes 31 69 0 to get into and out of a car 13 83 4 to rìde a bike for 10 and 30 minutes 15 85 0

correct = an increasing score similar to expected increasing difficulty equal = the same score in the pairs of items misclassified = a decreasing score

Measurements Group I: 112 men and 37 women (mean age 40 ± 9 yr, mean disease duration 11 ± 9 yr) completed the DFI sufficiently well, six questionnaires were incorrect. Group R consisted of 24 men and 15 women (mean age 40 ± 9 yr, mean disease duration 10 ± 8 yr) and these patients completed the DFI twice. Thirty patients completed the DFI between 7 to 28 days and nine patients between 29 to 42 days. Group V was a subgroup of R and consisted of six men and 13 women (mean age 39 ± 7 yr, mean disease duration 7 ± 5 yr). Activities related to 25 items (see Table 1) were performed and observed in this group by three rheumatologists and three physical therapists, who served as a 'gold standard'. Before observation of group V, several standardization meetings were held in which patients performed the 25 selected items, observed simultaneously by all observers. In case of disag­ reement of item score between the observers these were discussed and a consensus was reached. For the observation of group R the 25 performances were divided into four separate parts and each part had an almost equal degree of difficulty and required similar physical effort. At least two rest periods of 15 min were between these four parts. The entire observed performance took about 1.5 h for each patient. Patients completed the DFI no more than 10 days before the observation. For blind validation, patients were asked to participate in group V without knowing that the procedures were used for the validation of the DFI. Observers did not know DFI scores of these 19 participating patients. Group L consisted of 55 patients, the DFI being completed every 12 weeks. Repeated measurements were obtained from 51 patients (40 men and 11 women, mean age 39 ± 9 yr, mean disease duration 9 ± 7 yr); 38 patients adequately completed the DFI at weeks 0 and 12 and 29 patients completed the whole study (see Table 3). In addition to the DFI, the following variables were measured; i.e. the modified Schober, fingertip-to-floor distance, chest expansion and pain in the spine and sacroiliac region reported during physical examination, duration of mor-

104 Dutch functional index for AS

Table 3. Correlations of DFI with clinical measurements (group L) (n varying from 31 to 46)

variable/week week 0 week 12 week 24 week 36 week 48

spinal pain during the day 0.56 0.67 0.54 037 0.59

spinal pain during the night 0.50 0.58 0.54 0.70 0.60

morning stiffness 0.19 0.33 -0.04 0.34 0.21

general well-being reported by the patient 0.60 0.62 0.68 0.55 0.64

pain of the back and sacroiliac region 0.45 0.45 0.66 0.34 0.68

general well-being reported by the investigator 0.00 0.51 0.47 0.23 0.26

chest expansion 0.11 0.19 0.02 -0.17 -0.30

fingertip-to-floor- dis tance 0.26 0.39 0.48 0.58 0.74

modified Schober -0.13 -0.21 -0.02 -0.32 -0.69

(Standard error in г < 0.18)

rung stiffness and 100-mm visual analogue scales about backpain at night, backpain during the day and general well-being.

Statistical analysis

Group I: principal component analysis10 was applied to identify any homogenous subgroups of items. Cronbach's α for internal consistency and item evaluation was performed12. The interpretation of Cronbach's α is comparable to the measurement- remeasurement correlation. Pairs of items dealing with the same activity but differing in difficulty, e.g. to sit for 10 and for 30 min (see Table 2), were analysed for inconsistency. The hypothesis was studied that 'in general the more difficult the action was, the higher the item score would be', and the following definitions were used: equal = the same score in the pairs of items, correct = an increasing score, misclassified = a decreasing score.

105 Chapter Vili

Group R: for reproducibility correlation between mean score of two series of measurements was calculated. The mean difference between both measurements was used to test for learning effects (paired t-test) and the corresponding standard deviation (S.D.) was used to calculate measurement errors (S.D.//2). Group V: two scores were calculated of the DFI completed by the patient; a mean total score (patients' total DFI score) and a partial score (patients' partial DFI score) being the mean of the sum of the 25 observed items (as indicated in Table 1). Apart from this, scores assessed by the observer were calculated also as the mean of the sum of the 25 observed items (observers' DFI score). Criterion validity was evaluated using these correlations. Group L: paired t-test was used to test the difference in score between week 0 and 12 for determining sensitivity to change. Construct validity was analysed using Pearson correlations of the DFI with clinical variables. Clinical variables which were statistically significant associated with the DFI were analysed using a multiple regression analysis, the DFI score being the dependent variable and the clinical variables being the independent variables. Multiple R and squared R were calculated. In longitudinal studies quality of variables can be determined by calculation of interperiod correlations. These intercorrelations can be plotted against time interval and a regression line can be calculated. The intersection of this line with the vertical axis (intercept) is an estimation of the direct measurement-remeasurement 13 correlation (r0) and this may be interpreted as a quality measure of a variable . The DFI scores of weeks 12, 24, 36 and 48 were used for this estimation and time intervals were 12 weeks.

Results

Group /: Principal component analysis on the correlation matrix of the 37 items (n = 149) resulted in 10 factors (Eigenvalue > 1). Almost all items loaded strongly on the first factor (explained variance 32%). The other factors, also after rotations only comprised a small number of items. Therefore it was decided to consider the questionnaire as unidimensional. Internal consistency of the DFI (n = 149) was analysed using reliability analysis leading to a Cronbach's α = 0.93. Three items decreased the reliability of the questionnaire (indicated in Table 1) and were excluded, leading to a Cronbach's α = 0.94. One of these three originated from the French FI (marked in Table 1). The median total score on 34 items was 1.0 (range 0 - 2.45, S.D. 0.62). The questionnaire included six pairs of items dealing with the same activity but differing in difficulty and physical effort (see Table 1). Inconsistency was counted as described above. Very few misclassificatìons were seen; 2% on average (see Table 2). The time required to complete the questionnaire was on average less than five minutes.

106 Dutch functional index for AS

Figure 1. Inteiperiod correlations and regression line

correlation 1

0.9

0.8

0.7

0.6' 0 12 3 lime interval [12 weeks)

legend: con-elations shown in Table 4 have been plotted against intermediate time, where one time interval was defined as 12 weeks. The corresponding regression line is calculated and extrapolated to the intersection with the Y-axis. This point of intersection can be interpreted as the measurement-remeasure- ment correlation (r0)

Group R: Patients of this group (n = 39) could be divided into two groups, namely those who completed the DFI twice within 4 weeks and those who completed it twice between 4 to 6 weeks. Pearson correlations of these groups were 0.95 (n = 30) and 0.94 (n = 9) respectively. The correlation of the whole group (n = 39) was 0.94. The mean difference in item scores was -0.02 and 0.12, respective­ ly and measurement errors were 0.16 and 0.18 respectively. Paired t-testing for any learning effect was not significant. Group V: Correlation between patients partial DFI score and observers DFI score was 0.85, being criterion validity of the DFI. The mean observers DFI score was 0.68 (S.D. = 0.06) and the mean patients partial DFI score was 0.82 (S.D. = 0.30). The mean difference between the two scores was 0.10 (S.D. = 0.14). Group L: sensitivity to change was tested using the mean scores of weeks 0 and 12, being 1.01 (S.D. = 0.59) and 0.85 (S.D. = 0.63) respectively. The mean difference between weeks 0 and 12 was 0.31 (S.D. = 0.43, range -0.45 to 0.99). Paired t-testing showed significant effect (P < 0.02), meaning that important clinical changes can be measured using the DFI.

107 Chapter Vili

Correlations between the DFI and clinical variables were calculated for construct validity (see Table 3). There were 187 DFIs adequately completed with all corresponding clinical measurements. The multiple correlation (multiple R) was 0.68 and R2 was 0.47. Fingertip-to-floor distance, pain and general well-being were significantly associated in this analysis. Interperiod correlations were calculated (see Table 4), and the corresponding regression line was calculated. Extrapolation of the intersection with the Y-axis of this regression line resulted in the estimation of the measurement-remeasurement correlation (r0) of 0.87 (see Fig. 1).

Discussion

Three different questionnaires claiming to assess functional status in AS have been published in the literature. In our opinion the FI of Dougados et al.6,7 approached most closely the guidelines for development of measurement scales'10, but it was not available for Dutch AS patients and was not self-administered. Therefore, this questionnaire was translated, modified into a self-administered questionnaire and items concerning activities lacking in the French FI were added. The modified DFI was subsequently validated, similar to the French FI, but with additonal criterion validity and sensitivity to change over the longterm. Measurement errors in group R being 0.16 and 0.18 on a range from 0 to 4, and the accompanying measurement-remeasurement correlations 0.95 and 0.94 respectively indicate good reproducibility. Despite the fact that the time interval between completion of the DFI varied from 7 to 42 days, in patients with stable disease (group R) reproducibility was satisfactory, and it can be concluded that daily fluctuations in disease activity do not influence the score of the DFI in patients with stable disease. The lack of a 'gold standard' of a functional status instrument makes it necessary to test criterion validity in a broader term, as other investigators have done before14"16. Criterion validity was, therefore, tested against a 'gold standard' (expert observation) and compared with reproducibility contained smaller albeit satisfactory correlations. Patients' partial DFI score and observers' DFI score did not differ significantly, although, observers' DFI score was slightly lower. An observer bias may cause this difference in DFI score; pain and difficulty experien­ ced by the patient during performance cannot always be observed. An attempt was made to reduce measurement bias by standardizing observation. Sensitivity to change was studied in the clinical trial. A statistically significant decrease was found in DFI scores at the end of a washout period and after 12 weeks of NSAID treatment. It can be concluded that the DFI is a sufficiently sensitive instrument to detect the effects of NSAID treatment. This is in agreement with the studies of Dougados et al.6,7, who found significant improvement in scores in their 2-week clinical trials.

108 Dutch functional index tor AS

Table 4. Correlation matrix of DFI scores* (Group L)

week 12 week 24 week 36 week 48 week 12 1.00 (41) week 24 0.80 1.00 (34) (41) week 36 0.70 0.87 1.00 (28) (28) (33) week 48 0.79 0.83 0.86 1.00 (28) (29) (29) (36)

* the number of questionnaires adequately completed is placed between brackets

Correlations of DFI scores with clinical variables varied from poor to moderate (see Table 4); this was also found by Dougados et al.6,7 as well. Spinal pain, well- being and fingertip-to-floor distance showed moderate correlations. In a multiple regression analysis general well-being, finger-tip-to-floor distance and pain during physical examination were significant and about 47% of variance in the DFI could be explained by these three variables. It can be concluded that the DFI is a potentially useful instrument, which is valid, reliable and sensitive to the effects of NSAID treatment, is easy to administer and requires little time for completion.

Acknowledgements

We wish to thank Dr. W. van Lankveld (psychologist), Mrs. A. Smeijers (nurse), Mrs. S. Terwindt (occupational therapist), Mrs. A. de Boer, Mrs. W. Rutten and Mrs. E. Kok-Truijens (physiotherapists) for providing information for the develop­ ment of additional items. The latter three also for their observation for criterion validation, Mr. M. Bardoel (physiotherapist) for providing facilities and Mrs. N. Pluijmackers (secretary) for her organizational assistance during the measurements of criterion validity.

References

1. Meenan RF, Gertman PM, Mason JH. Measuring health status in arthritis: The Arthritis Impact: Measurement Scales. Arthritis Rheum 1980;23:146-52 2. Fries JF, Spitz PW, Kraines RG. Measurement of patient outcome in arthritis. Arthritis Rheum 1980;23:137-45

109 Chapter Vili

3. Pinci» Τ, Summey JA, Soraci SA, Wallston KA, Hummon NP. Assessment of patients satisfaction in activities of daily living using a modified Stanford Health Assessment Questionnaire. Arthritis Rheum 1985,26· 1346-53 4. Heijde van DMFM, Riel van PLCM, Putte van de LBA. Sensitivity of a Dutch health assessment questionnaire in a trial comparing hydroxychloroquine versus sulphasalazine. Scand J Rheumatol 1990,9-407-12 5. Nemeth R, Smith F, Elswood J, Calm A. Ankylosing spondylitis - an approach to measurement of seventy and outcome* ankylosing spondylitis assessment questionnaire (ASAQ) - a controlled study. Br J Rheumatol 1987,26(suppl 1)69-70 6. Dougados M, Gueguen A, Nakache J-P et al Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1988,15*302-7 7. Dougados M, Gueguen A, Nakache J-P et al Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1990,17-1254-5 8. Daltroy LH, Larson MG, Roberts WN, Liang MH. A modification of the health assessment questionnaire for the spondyloarthropathies. J Rheumatol 1990,17-946-50 9. Bombardier C, Tugwell P. Methodological considerations in functional assessment J Rheumatol 1987,14(suppl. 15)6-10 10. Streiner DL, Norman GR. Health measurement scales. A practical guide to their development and use. Oxford. Oxford University Press; 1989 11. Van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic entena for ankylosing spondylitis, a proposal for modification of the New York entena. Arthritis Rheum 1984,27 361-8 12. Cronbach LJ. Coefficient alpha and the internal structure of tests. Psychomctnka 1951; 16-297-334 13. Hof van 't MA, Kowalski CJ. In Prahl Anderson B, Kowalski CH, Heyendaal P, eds. A mixed longitudinal interdisciplinary study of growth and development New York' Academic Press, 1979,161-72,387-91 14. Fries JF, Spitz P, Krames RG, Holman HR. Measurement of patient outcome in arthritis. Arthritis Rheum 1980;23· 137-45 15 Ekdahl C, Eberhardt K, Andersson SL Svensson B. Assessing disability in patients with rheumatoid arthritis Use of a Swedish version of the Stanford Health Assessment Questionnaire. J Rheumatol 1988,17-263-71 16. Sullivan FM, Eagers RC, Lynch K, Barber JH. Assessment of disability caused by rheumatic diseases in general practice. Ann Rheum Dis 1987,46-598-600

110 Chapter IX

Disease activity in ankylosing spondylitis: selection of a core set of variables and a first step in the development of a disease activity score

MCW Creemers, MA van 't Hof, MJAM Franssen, LBA van de Putte, FWJ Gribnau, PLCM van Riel

Submitted for publication

Disease activity in AS

Summary

For assessment of disease activity in ankylosing spondylitis (AS) a large number of variables are available. The aim of this study was to evaluate validity of commonly used variables, to select a core set of valid variables for disease activity, and finally compute an AS disease activity score (AS-DAS). Data of two longitudinal studies were used. Principal component analysis and reliability analysis resulted in 11 factors: cervical mobility, lumbar flexion, subjective complaints, functional index (FI), enthesis index (EI), inflammatory response, IgA, IgM, root joints, swollen joints and spinal mobility. Based on discriminating power, reproducibility and correlation with disease duration, seven single variables were selected. In a subsequent discriminant analysis an AS-DAS was computed of five variables, i.e. subjective complaints, FI, EI, root joints and C-reactive protein, which should be validated in the future. A core set of process variables solves the problem of multiple testing in clinical trials, and improves comparability.

Introduction

The evaluation of disease activity in ankylosing spondylitis (AS) is a complex and multifactorial problem. Therefore a large number of different variables are being used for evaluation of patients with AS. Guidelines of the ARA1 and EULAR2 for clinical trials in AS have been developed. However, of this large number of variables validity, reproducibility and their mutual relation has not been studied3,4, with the exception of some variables3. Laboratory variables, for example C-reactive protein (CRP), ESR and IgA are not abnormal in every patient with active dis­ ease4'1*, and primarily appear to be associated with peripheral arthritis15'16 and extra­ articular features17. Another major problem is the lack of sensitivity to change, especially in those variables measuring spinal mobility2, possibly caused by the fact that commonly used variables are a combination of momentary disease activity, i.e. process of the disease, and of the result, i.e. outcome of the disease. As a result of these problems, in clinical trials in AS a wide variety of variables is used, making comparison between different studies difficult. The aim of this study was to evaluate the validity of commonly used variables for the assessment of disease activity in AS. Data of two clinical longitudinal prospective studies, in which patients had to have a high disease activity upon study entry, have been used for this. The results of this evaluation brings within reach the selection of a core set of variables, wich are reliable, reproducible and sensitive to change. Finally, this might lead to the development of a disease activity score for AS.

113 Chapter IX

Table 1. Patients' characteristics at baseline

NSAID study MTX study (n=59) (n=ll)

age (years)* 39 ± 9 35 ± 8 sex ratio male : female 45 : 14 (76% : 24%) 8 : 3 (73% : 27%) disease duration (years)* 10 ± 8 14 ± 9

number of patients with: HLA-B27 55 93% 11 100% peripheral arthritis 3 5% 3 3% iridocyclitis 2 3% 2 18%

* : mean ± standard deviation

Patients and methods Patients AS patients, fulfilling the modified New York criteria", of either sex, aged between 18 and 60 years, were selected. Study I was a 48-week NSAID study (n=59), comparing ß-cyclodextrin-piroxicam and naproxen. In study II patients with severe AS (n=ll) were treated with low-dose methotrexate (MTX) for at least 24 weeks. Exclusion criteria of both studies were: 1. serious systemic disease and or impaired organ function; 2. depression or other mental disorders; 3. alcohol or drug abuse; 4. pregnancy and breast feeding. Additional exclusion criteria for study I were major side effects caused by NSAEDs and active peptic ulcer disease in the past, and for study Π a history of intestinal disease, Reiters' disease or chronic infection. Patients' characteristics of both clinical studies are presented in Table 1. Active disease had to be present at the start of both studies: in study I, after a washout period of NSAIDs, a clear need for the use of analgesics or NSAIDs had to be present, next to at least two of the following features: a. back pain during the night and/or the day; b. morning stiffness of at least 15 minutes; с pain and stiff­ ness in both buttocks; d. pain and stiffness of the chest. In study П, additional to failure to treatment with NSAIDs and sulfasalazine, patients had to have at least two of the following features: a. morning stiffness of at least 30 minutes; b. disturbed sleep due to pain and stiffness; с peripheral arthritis; d. ESR ¿ 30 mm or CRP S 20 mg/L or IgA > 3,9 mg/L; e. spinal pain; f. stiffness and pain of the chest at movement or during normal breathing; g. pain in both buttocks during the night and/or the day.

Measurements Patients were seen at weeks 0, 2, 4, 8, 12, 18, 24, 30, 36, 42 and 48 for study I and 4-weekly for study П. For measurement of disease activity the following varia-

114 Disease activity in AS

Table 2. Median, percentiles 10 and 90 of assessed variables

Variable plO median p90 skewness skewness after transformation

spinal pain night (VAS) 6 37 82 0.19 - spinal pain day (VAS) 6 46 82 0.04 - general wellbeing (VAS) 10 46 84 0.14 - morning stiffness (minutes) 0 60 144 2.34 0.08 (sqrt)

fingertip-to-floor distance (cm) 0 16 41 0.62 - lumbar flexion index (cm) 17.0 20.0 22.0 -0.40 - modified Schober index (cm) 17.0 20.0 21.5 -0.25 - Schober 10 cm index (cm) 11.0 14.0 15.0 -0.37 - lateral flexion good side (%) 1.75 4.17 9.38 0.83 - lateral flexion bad side (%) 2.20 5.14 10.17 0.69 - occiput-to-wall distance (cm) 0 1 10 1.75 0.62 (sqrt) chest expansion (cm) 2.0 3.5 6.0 1.02 -0.76 (In)

cervical mobility (degrees) anteflexion 16 52 71 -0.30 - retroflexion 20 42 68 0.11 - lateral flexion good side 13 32 48 0.05 - lateral flexion bad side 14 40 58 -0.04 - rotation good side 30 62 80 -0.69 - rotation bad side 40 70 4 -1.08 -

enthesis index 0 7 29 1.97 0.34 (sqrt) root joint index 0 0 2 3.28 1.29 (d-sqrt)

Dutch functional index 0.24 1.12 2.17 0.20

number of swollen joints 5.40 2.26 (d-sqrt)

ESR (mm/lst hour) 6 15 53 1.92 -0.09 (In) C-reactive protein (mg/L) 4 19 51 3.61 -0.20 (In) platelets (lO'/L) 217 274 405 1.53 0.26 (In) IgA (mg/L) 1.61 2.74 5.15 0.77 - IgO (mg/L) 8.50 11.28 17.60 3.19 1.04 (In) IgM (mg/L) 0.80 1.74 2.74 0.72 -

VAS ш visual analogue scale; ESR = erythrocyte sedimentation rate; (In) = logaritimic transforma­ tion; (sqrt) » square root transformation; (d-sqrt) = double square root transformation bles were assessed at every visit: a) Clinical variables: * occiput-to-wall distance, measured in cm with the patient standing as erect as possible with heels and back against the wall; * chest expansion: the difference in cm between the circumference of the chest at nipple line on full inspiration and full expiration;

115 Chapter IX

* Schober 10 cm index19; * Modified Schober 15 cm index5; * Lumbar flexion index20; * fingertip-to-floor distance: the distance in cm between the third finger and the floor with the patient bending forward maximally, without flexing the knees; * lumbar lateral flexion21, in the percentage of the body height; * the number of swollen joints; * enthesis index22, maximal score being 90; * root joint index: pain of shoulders and hips by palpation and/or passive move­ ment was assessed with pain graded from 0 to 3. The sum of the scores being the index; * mobility of the cervical spine (in degrees): rotation using a goniometer, and lateral flexion, anteflexion and retroflexion using a hydrogoniometer. b) Subjective variables: * spinal pain during the day, spinal pain during the night and general wellbeing were assessed by using 100 mm VAS; * duration of morning stiffness after arising, recorded in minutes with a maximum of 360 minutes. c) Laboratory variables: * ESR, C-reactive protein, complete blood cell count, immunoglobin A, G and M; d) Functional assessment: A Dutch functional index for AS23, in essence a modification of the functional index of Dougados et al24,25 was completed at week 0 and every 12 weeks. Score maximally 4. Variables which have been measured at both sides, i.e. lumbar lateral flexion, cervical rotation and cervical lateral flexion, were reported as a good and bad score for each patient irrespective of the side.

Definition of disease activity and selection of records Patients' records were selected guided by the definitions of high and low disease activity. Disease activity was defined low, in study I if patients had used a stable dose of NSAIDs for at least 24 weeks, and in study Π if patients continued MTX after 24 weeks. Disease activity was defined high at week 0 of both studies, with the exception that some patients of the NSAID study, of whom pain and stiffness did not deteriorate enough after a washout period for NSAIDs, were excluded for this study. Finally, 101 records of 65 patients were selected for further analysis: 46 records with a low disease activity and 55 with a high disease activity. For 36 patients both a low and a high disease activity record was available. Ranges of variables of these 101 records are shown in Table 2.

116 Disease activity in AS

Statistical analysis

Variables used in multivariate statistical analysis have to be normally distributed. If necessary, assessed variables, were, therefore, transformed to obtain normality. To find subgroups of variables, principal component analysis was performed on all variables. For reliability testing Cronbach's a26 was calculated of the factors. A forward stepwise discriminant analysis was performed for determination of factors discriminating significantly between a high and a low disease activity. Because the number of visits in the two studies was different, resulting in different amount of records per patient, only the selected 101 patients' records were used for factor analysis and discriminant analysis. Records of patients of whom both a high and low disease activity record was available were analysed using paired t-tests. Of these paired observations the discriminating power was expressed as the standardi­ zed difference (difference/pooled standard deviation). AU records of the 59 patients of the NSATD study were used for analysis of the reproducibility of variables, calculated with an interperiod correlation matrix and the corresponding regression line was calculated for the estimation of measurement-remeasurement correlation 27 (r0) . Because variables might be a combination of process and outcome of the disease, the relation with disease duration was analysed. For this, Spearman rank correlations were computed of disease duration and all assessed variables of weeks 0 and 12 of study I. To separate process and outcome variables, those variables having a significant correlation with the disease duration, were considered mainly assessing outcome. Finally, a set of process variables was then selected and an index for disease activity calculated using discriminant analysis.

Results

Factor analysis and reliability Principal component analysis was performed on 27 variables of the 101 selected records. Seven factors were found with eigenvalues larger than 1 (see Table 3) and a cumulative explained variance of 76%. After varimax rotation, four variables, i.e. fingertip-to-floor distance, occiput-to-wall distance and lumbar lateral flexion loaded moderately on factor 1 as well as on factor 2, and were put into a new factor (factor ПІ). Factor 3 was composed out of morning stiffness, spinal pain during the day and during the night, general wellbeing, the enthesis index and the functional index. The first four variables were considered patients' assessed variables and were grouped together in factor Ша. The questionnaire (functional index) and the 'assessor' variable (enthesis index) were used as separate factor, Factor ШЪ and Шс respectively. Variables were scaled, and scaling factors were chosen to equalize the standard deviations of variables within each factor. Factor values then were calculated as the

117 Chapter IX

Table 3. Factors out of prinicipal component analysis and reliability analysis

Factor analysis Reliability analysis

Factor 1 Factor I 'CERVICAL MOBILITY' cervical anteflexion anteflexion " retroflexion retroflexion " rotation g rotation g " rotation b rotation b " lateralflexion g lateralflexion g " lateralflexion b lateralflexion b

Factor 2 Factor II 'LUMBAR FLEXION' modified Schober modified Schober lumbar flexion index lumbar flexion index Schober 10 cm index Schober 10 cm index

Factor 3 Factor Ша 'SUBJECTIVE COMPLAINTS' spinal pain day spinal pain day spinal pain night spinal pain night general wellbeing general wellbeing morning stiffness morning stiffness functional index cnthesis index Factor IHb FUNCTIONAL INDEX

Factor Шс ENTHESIS INDEX

Factor 4 Factor IV •INFLAMMATORY RESPONSE' ESR ESR CRP CRP IgO IgG platelets platelets

Factor 5 Factor Va IgA IgA IgM Factor Vb IgM

Factor 6 Factor VI ROOT JOINTS root joint index

Factor 7 Factor VII SWOLLEN JOINTS swollen joints

Factor 8* Factor ІІГ 'SPINAL MOBILITY' fingertip-to-floor distance fingertip-to-floor distance occiput-to-wall distance occiput-to-wall distance lumbar lateralflexion g lumbar lateralflexion g lumbar lateralflexion b lumbar lateralflexion b

g = good side; b = bad side; * : variables which loaded moderately both on factor 1 and factor 2

118 Disease activity in AS mean of the variables corresponding with the factor. Cronbacb's α was calculated for the composed factors (1, 2, Ша, 4, S, 8). All variables within a factor gave a positive contribution to the reliability, except for factor 5 (Cronbach's α = 0.47), which subsequently was splitted up into two single factors (factors Va and Vb). Cronbach's α of factors 1, 2, 4, 8 and ΠΙ varied from 0.82 to 0.97 and may be considered as reliable. Finally there were 11 factors: cervical mobility (factor I), lumbar flexion (factor Π), 'subjective complaints' (factor Ша), the functional index (factor ШЬ), the enuresis index (factor Шс), 'inflammatory response' (factor Г ), IgA (factor Va), IgM (factor Vb), the root joints (factor VI), swollen joints (factor VI) and 'spinal mobility' (factor Ш).

Discriminant analysis Paired t-testing was performed on the factor values of the records of 36 patients with observations both with a high and low disease activity. For further analysis factor I was excluded because no significant change was found (p=0.16). The number of swollen joints also showed no significant change (p=0.06); as this was possibly due to the small number of patients with peripheral arthritis, we did not exclude this factor. The stepwise forward discriminant analysis showed five significant steps entering: 'subjective complaints' (F Ша), 'inflammatory response' (F TV), root joints (F VI), enthesis index (F Шс) and IgA (F Va). Table 4 shows the relative importance of factors and variables used in the discriminant analysis. Canonical correlation was 0.64 and 80% was correctly classified.

Table 4. Correlations of eleven factors with discriminant score and discriminating power

Factor correlations with standardized difference discriminant score (72 paired records) (101 records)

subjective complaints' (F Ша) 0.84 1.17 enthesis index* (F Шс) 0.63 0.89 functional index (F ¡lib) 0.56 0.69 spina] mobility (F ІП) -0.42 0.46 root joints* (F VI) 0.38 0.50 lumbar flexion(F П) 0.30 0.44 inflammatory response" (F Г ) -0.24 0.33 IgM (F Vb) 0.17 0.40 swollen joints (F П) 0.12 0.34 IgA' (F Va) 0.10 0.25 cervical mobility (F I) 0.16

(F) : Factor (see Table 3) *: factors selected stepwise in discriminant analysis

119 Chapter IX

The discriminating power, computed as the standardized differences from paired t-testing of the 36 patients, and the correlations with the discriminant score (101 records) gave similar information about the factors used in the analyses (see Table 4).

Disease duration Variables and factorvalues from the NSAID study of weeks 0 and 12 and corre­ sponding differences were used for calculation of Spearman rank correlations with disease duration. Only correlations statistically significant are shown in Table 5.

Reproducibility In longitudinal studies quality of variables can be determined by calculation of interperiod correlations. These intercorrelations can be plotted against rime interval and a regression line can be calculated. The intersection of this line with the vertical axis (intercept) is an estimation of the direct measurement-remeasurement correlation (r0) and this may be interpreted as a quality measure of a variable. Estimated r0 of variables belonging to composed factors are shown in Table 5.

Selection of a core set of variables The selection of usefull process variables is based on discriminating power (standardized difference), reproducibility and a lack of correlation with disease duration. Factors and variables correlating with disease duration were considered assessing primarily outcome and not selected. For reasons of simplicity and practicality, only one single variable out of a composed factor was selected. The selection of factor and variables was based on the largest discriminating power (standardized difference) and reproducibility (measurement-remeasurement correla­ tion, r0), and variables with standardized differences less than 0.30 were not taken into account. Selected process variables are thus (marked in table 5): 1. spinal pain during the day (VAS); 2. lumbar flexion index; 3. root joints; 4. enthesis index; 5. swollen joints; 6. functional index and 7. CRP.

Composition of a disease activity score A forward stepwise discriminant analysis on the seven selected process variables resulted in a correlation with the discriminant score of 0.07 for swollen joints, while the remaining six variables had correlations of at least 0.35. Therefore, this variable was left out of consideration. The subsequent discriminant analysis showed a positive coefficient for the lumbar flexion index and a negative correlation with the discriminant score. Because of this discrepancy, this variable was considered being unstable, i.e. its discriminating power is completely expressed by the other (interrelated) variables. Finally, the Ankylosing Spondylitis Disease Activity Score (AS-DAS) was computed with a discriminant analysis using five variables, namely spinal pain during the day (VAS), enthesis index, functional index, root joints and CRP. Three variables had to be transformed to obtain normality for this analysis.

120 Disease activity in AS

Table 5. Analyses used for selection of variables

selected discriminating reproducibility correlations with for AS-DAS power disease duration

factors and venables standardized estimated r0 week 0 week 12 difference (n=36) (n=59)

'CERVICAL MOBILITY' 0.16 anteflexion 0.11 0.77 -0.32" retroflexion 0.01 0.81 rotation g 0.29* 0.82 rotation b 0.29* 0.80 lateral flexion g 0.06 0.87 lateral flexion b 0.05 0.86

'LUMBAR FLEXION' 0.44"" lumbar flexion index (X) 0.63"" 0.88 modified Schober index 0.30"" 0.87 Schober 10 cm index 0.35"' 0.85

•SUBJECTIVE COMPLAINTS' 1.17"" spinal pain day X 1.06"" 0.65 spinal pain night 0.83"" 0.63 general wellbeing 1.11"" 0.58 morning stiffness 0.75"" 0.55

ENTHESIS INDEX X 0.89" 0.74

FUNCTIONAL INDEX 0.69" 0.92

•INFLAMMATORY RESPONSE' 0.33"" ESR 0.17 0.87 CRP X 0.72 0.74 IgG 0.23 0.86 platelets 0.02 0.85

IgA 0.25"" 0.94

IgM 0.40"' 0.95 -0.28"

ROOT JOINTS 0J0* 0.67

SWOLLEN JOINTS (X) 034Hio6 0.80

'SPINAL MOBILITY' 0.46"" -0.30* -0.37" fingertip-to-fioor distance 0.44"" 0.94 occiput-to-wall distance 0.31'" 0.89 0.27* 0.27* lumbar lateral flexion g 0.31"" 0.94 -0.33" lumbar lateral flexion b 0.31"" 0.94 -0.27' -0.35·' chest expansion 0.34"· 0.86

p-value:* p<0.05, **· p<0.01; *·*· p<0.005; *»**· p<0.001 ESR = erythrocyte sedimentation rate, CRP = C-reactive protein; b = bad side, g = good side

121 Chapter IX

О Sa

s .s о 1 as s sr

q Ö Ο *n η О s # + I «g гч с^ òo г- M S' та .S о-

о •"ΐ « M •« + -" — Ό tS О 5

J* а s? 0\ л-: «« α. о о 8 si ν -3 I II .S2 ее m œ о vi 8 ö о ö *ό <ή и

§ * I II •β t: ST .11 .k' 13 .Ξ ο ο ο ο ο 00 я о •3 II II 9* S S" S1 1 ρ; S e •л о 5 -л i 3 S .a B-JS 81 'illa £ S'S-0

122 Disease activity in AS

Correlations with the final score, means and standard deviations of the separate variables and coefficients for the final formula are shown in Table 6. The canonical correlation was 0.67 and 82% was correctly classified in this final discriminant analysis.

Discussion

Judgement of disease activity in rheumatic diseases is a complex process and forms the basis of clinical decisions. Assessed variables are often a combination of process and outcome of the disease. In AS commonly used variables contain a lot of interpretation problems as stated in the introduction. In the past, some indices of disease activity in AS have been proposed, based on clinical judgement28"32. However, the majority of these indices have not been validated. Recently, Salaffi et al33 analysed data of 45 AS patients in order to identify groups of variables out of the large number available. Three groups of variables were identified, measuring disease activity, damage and functional status, the latter being a combination of disease activity and disease outcome. In our study longitudinal data of AS patients were analysed to identify those variables which discriminate for disease activity. Since no 'gold standard' for disease activity in AS is available, predefined criteria for high and low disease activity were used. Two groups of patients' records were formed; one group with high and one group with low disease activity. For evaluation of reproducibility, discriminating power and the correlation of variables and factors with disease duration, a predefined analysis strategy34, including principal component analysis to form subgroups of reliable, interrelated variables, reliability of factors was used. This evaluation resulted in a selection of 7 process variables. Finally, a disease activity score was computed of five variables, derived from a discrimant analysis. The core set of seven process variables encompassed lumbar flexion index, spinal pain during the day, enthesis index, functional index, CRP, root joints and the number of swollen joints. In a discriminant score for the construction of a disease activity score two variables had to be excluded: a. 'swollen joints' because of a low correlation with the discriminant score; and b. 'the lumbar flexion index', which appeared to be an instable variable, i.e. its discriminating power is com­ pletely expressed by the other (interrelated) variables. The five variables of the AS- DAS are part of the whole spectrum of clinical, subjective and laboratory variables. It is striking that pain assessed in three of these variables, i.e. two, gives most weight to the AS-DAS. In conclusion, the proposed AS-DAS should be validated further. For standardi­ zation purposes the selected core set of process variables should be used in future clinical trials in AS, resulting in improvement of comparability of different studies. Additionally, the problem of multiple statistical testing will be solved since only a reduced number of variables, and eventually a single composed variable, i.e. the AS-DAS, have to be analysed for evaluation of disease activity.

123 Chapter IX

References

1. Guidelines for the Clinical Evaluation of Anti-Inflammatory and Anti-Rheumatic drugs (Adults and Children), US Department of Health and Human Services, Public Health Service, Food And Drug Administration, April 1988, p. 38-40 2. Guidelines for the Clinical Investigation of Drugs Used in Rheumatic Diseases, European Drug Guideline Sene 5, World Health Organization, Regional Office for Europe, Copenhagen, European League Against Rheumatism, March 1985, ρ 12 3. Laurent MR, Buchanan WW, Bellamy N. Methods of assessment used in ankylosing spondylitis clinical trials- a review. Br J Rheum 1991,30 326-9 4. Rigby AS, Silman AJ Outcome assessment in clinical tnals of ankylosing spondylitis. Br J Rheum 1991,30 321-5 5. Moll JMH, Wnght V. Normal range of spinal mobility. Ann Rheum Dis 1971,30-381-6 6. Espinoza LR, Gaylord SW, Bocanegra TS, Vasey FB, Germain BF. Circulating immune complexes in the seronegative spondyloarthropathies. Clin Imm Immunopath 1982,22 384-93 7. Laurent MR, Panayi GS. Acute-phase proteins and serum immunoglobulins in ankylosing spondylitis. Ann Rheum Dis 1983,42-524-8 8. Vinje O, Moller Ρ, Mellbye J. Immunological variables and acute-phase reactants in patients with ankylosing spondylitis (Bechterew's syndrome) and their relatives. Clin Rheumatol 1984,3 501-14 9. Cowling P, Ebringer R, Cawdell D, Ishn M, Ebringer A. C-reactivc protein, ESR and klebsiella in ankylosing spondylitis. Ann Rheum Dis 1980,39*45-9 10. Scott DGI, Ring EFJ, Bacon PA Problems in the assessment of disease activity in ankylosing spondylitis. Rheumat Rehab 1981,20-74-80 11. Calgunens M, Swinburne L, Shincbaum R, Cooke EM, Wnght V. Secretory IgA- Immune defence pattern in ankylosing spondylitis and klebsiella. Ann Rheum Dis 1981,40-600-4 12 Hickling P, Tumbull L, Dixon JS The relationship between disease activity, immunoglobulins and lymphocyte sub-populations m ankylosing spondylitis. Rheum Rehab 1982,21*145-50 13 Sanders KM, Hertzman A, Escobar MR, Littman BH. Correlation of immunoglobulin and C- reactive protein levels in ankylosing spondylitis and rheumatoid arthritis. Ann Rheum Dis 1987,46-273-6 14. Collado A, Sanmarti R, Brancos MA et al Correlation of Immunoglobulin and C-reactive protein levels in ankylosing spondylitis. Ann Rheum Dis 1987,46-719-20 15 Kendall MI, Lawrence DS, Shuttlcworth GR, Whitfield AGW. Hematology and biochemistry of ankylosing spondylitis. Br Med J 1973,23 235-7 16. Dixon JS, Bird HA, Wnght V. A comparison of serum biochemistry m ankylosing spondylitis, seronegative and seropositive rheumatoid arthritis Ann Rheum Dis 1981,40 404-8 17. Struthers GR, Lewin IV, Stanworth DR. IgA-alpha, antitrypsin complexes in ankylosing spondylitis. Ann Rheum Dis 1989,48-30-4 18. Linden van der S, Valkenburg HA, Cats A. Evaluation of diagnostic entena for ankylosing spondylitis, a proposal for modification of the New York entena. Arthritis Rheum 1984,27-361-8 19. Schober P. Lendenwirbelsaule und Kreuzschmerzen. Muench Med Wschr 1937,84 336 20. Adnchem JAM, Korst van der JK. Assessment of the flexibility of the lumbar spine. Sc and J Rheum 1973,2 87-91 21. Domján L, Nemes Τ, Bálmt GP, Tóth Ζ, Gomor Β. A simple method for measuring lateral flexion of the dorsolumbar spme. J Rheumatol 1990,17 663-5 22. Mander MM, Simpson JM, McLellan A et al. Studies with an enthesis index as a method of clinical assessment in ankylosing spondylitis. Ann Rheum Dis 1987,46 197-202 23. Creemers MCW, Hof van 't MA, Franssen MJAM et al. A Dutch version of the functional index for ankylosing spondylitis. Development and validation in a long-term study Br J Rheum 1994,33*842-6 24. Dougados M, Gueguen A, Nakache J-P et al. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1988,15-302-7

124 Disease activity in AS

25. Dougados M, Gueguen A, Nakache J-P et al. Evaluation of a functional index and an articular index in ankylosing spondylitis. J Rheumatol 1990;17·1254-55 26. Cronbach LI. Coefficient alpha and the internal structure of tests. Psychometnka 1951,16 297-334 27. Hof van 't MA, Kowalski CJ. In· Prahl Anderson B, Kowalski CH, Heyendaal P, eds. A mixed longitudinal interdisciplinary study of growth and development. New York: Academic Press, 1979; 161-72,387-91 28. Calm A, McShane D, Powers R. Objective measurements m evaluating drug therapy m ankylosing spondylitis. Curr Ther Res 1978.24:838-42 29. Franssen MIAM, Gribnau FWI, Putte van de LBA. Assessment of disease activity m ankylosing spondylitis- the common clinical criteria and an index of disease activity. Thesis 1985, Catholic University Nijmegen, the Netherlands 30. Peelers AJ, Wall Bake van den AWL, Albada-Kuipers OA et al. IgA containing immune complexes and hematuria in ankylosing spondylitis. A prospective longitudinal study. J Rheumatol 1988,15' 1662-7 31. Kennedy LG, Edmunds L, Calm A. The natural history of ankylosing spondylitis. Docs it bum out'' J Rheumatol 1993,20-688-92 32. Garrett SL, Kennedy LG, Whitelock HC et al. A new approach to defining disease status m AS- the Bath ankylosing spondylitis disease activity index (BASDAI). Br J Rheum 1993:32(suppl 2)-23 33. Salaffi F, Carom M, Brccciaroli D, Cervini С. Assessment of ankylosing spondylitis- a first step in the development of a composite index. Rev Esp Reumatol 1993,20(suppl l)-489 34. Heijde van de DMFM, MA van 't Hof, PLCM van Riel et al. Judging disease activity m clinical practice in rheumatoid arthritis- fust step in the development of a disease activity score. Ann Rheum Dis 1990,49-916-20

125

Chapter Χ

A radiographic scoring system and identification of variables measuring structural damage in ankylosing spondylitis

MCW Creemers, MJAM Franssen, MA van 't Hof, FWJ Gribnau, LBA van de Putte, PLCM van Riel submitted for publication

A radiographic scoring system in AS

Summary

A radiographic scoring system for ankylosing spondylitis (AS) was developed and evaluated in a 48-week clinical study, in which ß-cyclodextrin-piroxicam was compared with naproxen, comprising a modified lumbar spine score (LSS), a cervical spine score (CSS), an enthesis score (ES) and Larsen hip score (HS). In study I radiographs of patients with AS, not participating in the clinical study, were scored by two observers to study reproducibility. The interobserver correlations of the CSS and LSS were larger than 035, corresponding intraobser- ver errors were satisfactory. In addition, the intraobserver correlation and error were studied of ES and HS, and were satisfactory. However there was a statistical­ ly significant difference between the scores of the ES (t-test). In study II, radiographs made at week 0 and 48 of 57 patients were scored, blinded for treatment and patient, by a single observer in a sequential way. A significant change was found from week 0 to 48. There were no differences in radiographic scores and change in radiographic score between the two treatments (t-test), nor between responders and non-responders. Differences in radiographic scores between men and women were evaluated, and no significant differences could be found if corrected for disease duration. Pearson correlations of radio­ graphic scores and clinical variables assessed in the clinical study revealed several significant correlations, and these variables might be considered variables measu­ ring structural damage. This extended radiographic scoring is reproducible and sensitive to change. No differences were found between ß-cyclodextrin-piroxicam and naproxen, responders and non-responders, and men and women.

Introduction

To evaluate outcome in AS radiological scoring systems of the lumbar spine and the sacroiliac joints have been published by Taylor et al.1 But next to sacroiliitis, the hallmark lesion in AS, the whole axial spine, entheses and root joints are frequently involved2"*. Therefore, additional to radiographic scoring of the lumbar spine, these other sites should also be taken into account as well in the evaluation of outcome in AS. The aim of this study was to develop and evaluate a more extensive radiologi­ cal scoring system for AS, in which several entheses, the hip joints and the cervical spine were encompassed. This extended radiological scoring system was used in a 48-week longitudinal study of patients with AS to describe radiographic features, and to evaluate sensitivity to change. In addition to this, the relationship with commonly used clinical variables was studied.

129 Chapter Χ

Table 1. Patients' characteristics at the start of the clinical trial

number of paticns (n«57)

sex ratio male : female 43 : 14 HLA-B27 54 peripheral arthritis 3 age 39 ±9 disease duration* 10 ±8

*: years, mean ± standard deviations

Patients and methods

The study was conducted at the departments of rheumatology of the University Hospital Nijmegen, St. Radboud and of the St Maartenshospital Nijmegen, the Netherlands, and started after full approval of both ethical committees was obtain­ ed.

Radiographs Study I Plain X-rays of the pelvis (anterior-posterior film) and lateral view of the lumbar and cervical spine were used for the development of the radiographic scoring systems and for standardization of the two observers (MF, MC). Radiographs of 370 patients (not included in study Π) with AS fulfilling the modified New York criteria5, attending the rheumatology departments, were used. To study reproducibi­ lity, a separate set of 95 X-rays of the pelvis, 46 lateral views of the lumbar spine and 26 lateral views of the cervical spine were scored.

Study II Plain X-rays of the pelvis (anterior-posterior film), and lateral view of lumbar and cervical spine were made at weeks 0 and 48 of patients with AS fulfilling the modified New York criteria5, who participated in a nonsteroidal antiinflammatory drug (NSAID) trial. This trial was a single observer double-blind randomised study in which ß-cyclodextrin-piroxicam (SA Nycomed Christiaens BV, Belgium) was compared with naproxen (SA Nycomed Christiaens BV, Belgium). Radiographs of both week 0 and 48 were available of 57 out of 59 patients. Patients' characteristics are shown in Table 1. All radiographs were scored blinded with respect to the patient and treatment in a sequential way by a single observer. All patients had to have an active disease after a washout period for NSAIDs. Patients were sequentially assigned to one of the treatment groups, balanced* for age, sex, peripheral arthritis, extent and duration of the disease. Clinical measure­ ments were performed by a single observer (MC), generally at the same time of the day, at start of the medication (week 0) and at weeks 2, 4, 8, 12, 18, 24, 30, 36, 42

130 A radiographic scoring system in AS and 48. The following measurements were assessed at every visit: a. objective clinical variables: occiput-to-wall-distance, chest expansion, Schober 10 cm test7, lumbal lateral flexion', the number of swollen joints (n=S3), enthesis index', root joint index (pain of shoulders and hips by palpation and/or passive movement was assessed with pain graded from 0 to 3, the sum of the score being the index), mobility of the cervical spine in all three planes; b. subjective variables: duration of morning stiffness, and 100 mm Visual Analogue Scales (VAS) for spinal pain during the day, during the night and general wellbeing; с laboratory variables: HLA-B27, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), complete blood cell count, immunoglobin A, G and M; d. functional assessment: a Dutch Functional Index for AS10 was completed at week 0 and every 12 weeks thereafter.

Radiographic scoring system a. lumbar spine: the scoring system, developed by Taylor et al.4 was applied to the lower border of the 12th thoracic vertebra, all five lumbar vertebrae and the upper border of the sacrum were viewed on the lateral X-ray. The corresponding nominal scoring system: 0 = no abnormality; 1 = erosion, sclerosis or squaring; 2 = syndesmophyte, and 3 = total bony bridging at each site. Radiological abnormalities, which were not related to AS such as osteophytes, were not considered for scoring. Sites which were not clearly visible on the radiograph were not considered for scoring. If more than three scoring sites were missing, radiographs were not taken into account. Additional to the definitions of Taylor et al.1 squaring was defined as present if the surface of the vertebra was convex; or if a line could Actively be drawn with a transparent ruler, from the upper to the lower border of a vertebral body, and if this line overlayed 50% or more with the surface of the vertebra. This added definition of squaring approaches the method of Ralston et al.11 In order to deal efficiently with missing observati­ ons the total score (range 0-36) of all radiographs was calculated as 12 times the mean score of all scoring sites; b. cervical spine: this scoring system was identical to that of the lumbar spine. The lower border of the 2nd cervical vertebra up to and including the upper border of the first thoracic vertebra were viewed for this on the lateral X-ray. Due to the original straight shape of the lateral surface of the third cervical vertebra it was decided that this vertebra was not scored for squaring, although erosions and sclerosis were scored; с hip joints: the scoring system of the hips developed by Larsen et al.12 was applied. The anterior posterior X-ray of the pelvis was viewed. This scoring system uses the overall appearance of each hip was graded from 0 to S accor­ ding to standard radiographs (0 = normal conditions; and 5 = mutilating abnor­ mality and gross bone deformation). The total score is the sum of the grades of the two hips (range 0-10);

131 Chapter Χ d. entheses: the following 4 entheses were viewed on the anterior-posterior film of the pelvis: the ischial tuberosities, the iliac alae, and the minor and major trochanter of the femoral bone. They were scored: 0 = no abnormalities; and 1 = irregular new bone formation ('whiskering'), or osseous erosions with ill-defined bony margins and reactive sclerosis. In order to deal efficiently with missing observations the total score was eight times the mean score of the different scoring sites, with at most two entheses allowed not to be scored. e. pubic symphysis: on the anterior-posterior view of the pelvis both sites of the pubic symphysic joint were scored separately: 0 = no abnormalities; and 1 = irregular new bone formation ('whiskering'), or osseous erosions with ill-defined bony margins and reactive sclerosis, or (sub)luxation. The total score (range 0-2) was the sum of the score of the two sites of the joint. Radiographs of patients in study I were used for the development of the radio­ graphic scoring systems. After standardization of the two observers, radiographs, which had not been used previously, were scored to study reproducibility.

Statistical analysis

Reproducibility (Study I) was analysed by calculation of interobserver correlation and interobserver duplicate error (VEdj2/2n) of the different scores. To find systema­ tical differences between the two observers t-tests were applied on the scores of the two observers. Intra-observer correlation and intraobserver error were only studied additionally if the interobserver correlation was less than 0.95. For the intraobserver study radiographs were scored a second time, without knowledge of the previous score. If necessary variables were transformed to obtain normality. To study potential differences in radiographic scores, several groups of patients were studied (Study Π): a. the two treatment groups; b. men and women; and с patients who completed the whole study, i.e. responders, and patients who dropped out because of either inefficacy or side effects, i.e. non-responders. The Wilcoxon signed rank-test was used for analysis. In order to avoid multiple testing and thus decrease the risk of false significancies, two radiographic scores were composed: 1. the spinal score which was the sum of the lumbar and cervical radiographic score (range 0 - 72); and 2. a weighted overall score (range 0 - 124) of the spinal score (range 0 - 72) + 3.6 times the hip score (range 0-10) + 2 times the enuresis score (range 0 - 8). Additionally, to study the cervical spine involvement in the different groups, this score was analysed similarly. A 3-way ANOVA has applied additional­ ly to the scores of men and women to correct p-values for disease duration. In order to analyse the influence of HLA-B27 positivity, sex and disease duration on the scoring systems of the lumbar spine, cervical spine, entheses and hips (Study П) a multiple regression analysis has been done. Pearson correlations between the spinal score and the overall score and assessed clinical variables, which were transformed if necessary to obtain normality, were

132 A radiographic scoring system in AS

ö ci s •« я о s •9 Ά s. о •a 3 3 σ* « —· I Б η «i S η # № ^ nei о о f J TS Ss I e 'β .5 я δν <τ* os г*- ö о о' о"

б e? χ —' 4.^ >«^ Чя*" "W ε S α ** s о fr i i к о s δ о tí tí > и -9 .s n ST Ç4 о? ~ •ä о »Л — сч о 1 Ιέ •ч·*· "w* W "W -н-н-н-н-н-н-н-н p~ oo — Л Г- oo © © 8 «-«»dòri·" I

^ ίο ín ^ -"«4-«t4"-<4-*f4 "8 i! во « «п — •ι •| !» α, 1

α ti .8 S..Ï g .2 .1 ею .a о s- ° « fr Ö "Я ь -3 •s .•ая dì α. о. ¡.fri ее 1 -IЁ -I =| 1 ¿3 Η 3 s 3

133 Chapter Χ calculated (Study Π). Those variables which revealed significant correlations with the overall and spinal score both at weeks 0 and 12 were considered as variables measuring structural damage.

Results

Study I In total 46 X-rays of the lumbar spine, 26 of the cervical spine and 96 of the pelvis have been scored for the evaluation of reproducibility. Table 2 shows the means, standard deviations, interobserver correlations, duplicate error and p-value of t-testing of the different radiographic scoring systems. Because the hip and enthesis score had interobserver correlations less than 0.9S, the intraobserver correlation and intraobserver duplicate error of these two scoring systems were studied additionally (see Table 2).

Study II Table 3 shows the number of patients at week 0 with radiological involvement of the cervical and lumbar spine, the hips and entheses. To get an impression about the distribution of the radiological involvement in the cervical and lumbar spine, the percentages of the upper and lower border with scores 0 and 3 are shown in

Table 4. Total bony bridging was mainly present from Cj to Th„ Q to C3, C3 to

C4, and from Thi2 to L,. The 12th thoracic, the 1st lumbar and the 4th cervical vertebrae showed most frequently radiological involvement

Table 4. Percentages of the presence of score 0 and 3 per vertebra in study Π

score score cervical 0 3 lumbar 0 3 vertebra vertebra

Qi 84 5 Th„l 44 11 C311 74 5 L,u 51 11 C,l 75 5 L,l 75 5

C4u 25 5 L,u 72 5 Qi 30 2 W 83 2 CjU 65 2 L,u 65 2 C,l 68 2 L,l 77 5

Qu 74 2 L4u 72 5

Qi 72 2 L41 79 5 C,u 60 2 L,u 79 5 C71 60 7 L,l 82 4 Th,!! 50 12 S,u 93 4

Explanations of abbreviations' С = cervical; Th = thoracic, L =• lumbar; S » sacral;u = upper margin; 1 = lower margin

134 A radiographic scoring system in AS

Table 5. Median (percentiles 10 and 90) of cervical spine, spinal and overall score and change in score of men and women and the two treatments at weeks 0 and 48

radiographic score week men (n » 43) women (n a 14) p-valuc corrected p-value

cervical spine score 0 3.3(1.0, 11.6) 2.2(0, 11.6) 0.31 0.70 48 5.0 (2.0, 14.9) 2.2 (0, 14.4) 0.24 0.54 change' 1.1 (0, 4.5) 0 (0, 3.7) 0.25

spinal score 0 6.0 (1.4, 25.6) 3.6 (0, 16.0) 0.09 0.22 48 9.0 (2.0, 30.8) 4.7 (0, 17.7) 0.05 0.15 change* 2.0 (0, 5.7) 1.1 (0, 3.8) 0.15

overall score 0 9.6 (2.4, 29.7) 4.6 (0, 17.0) 0.02 0.09 48 13.0 (4.4, 38.6) 7.0 (1.1, 20.2) 0.02 0.053 change' 3.1 (0.4, 8.0) 2.1 (0, 4.5) 0.15

ß-cyclodextrin- naproxen p-value piroxicam

spinal score 0 5.7 (0.9, 32.3) 6.0 (0, 20.6) 0.79 48 8.0(1.8,33.4) 9.0 (1.1, 22.8) 0.76 change' 1.1 (0, 4.5) 0 (0,3.7) 0.89

overall score 0 7.2(1.8,32.3) 9.6 (2.0, 22.7) 0.87 48 10.2 (3.6, 32.3) 12.8 (3.6, 28 J) 0.79 change* 2.0 (0, 5.7) 1.1 (0, 3.8) 0.47

Explanations of symbols: percentiles 10 and 90 are placed between brackets, ': the change from weeks 0 to 48

The mean change in scores between week 0 and 48 of all patients was 1.45 (range 0 - 6.0) in the cervical spine, 1.06 (0 - 5.0) in the lumbar spine, 0.08 (0 - 1.0) in the hips, and 0.32 (0 - 2.0) in the enthesis score. All changes were statisti­ cally significant (p < 0.001). Radiological progression in the cervical spine was seen in 23 patients, in the lumbar spine in 24 patients, and in the enthesis score in 17 patients. The spinal score and overall score at weeks 0 and 48, and the change in these scores were calculated for the two treatment groups, for responders and non- responders, and for men and women. No statistically significant differences were found between the two treatment and between responders and non-responders. The mean disease duration for men and women was 11 years (standard deviation = sd = 8 years) and six years (sd = 4 years) respectively,whic h was statistically different (p = 0.015). T-testing of the scores of men and women revealed a statistically significant difference, which was no longer present after correction for disease duration (see Table 5). None of the calculated changes between the different groups

135 Chapter Χ

Table 6. Clinical variables with significant Pearson correlations (57 patients)

variables overall score spinal score week 0 week 12 week 0 week 12

chest expansion -0.47"· -0.56"· -0.Э6" -0.43"· fingertip-to-floor distance 0.29* 0.28* occiput-to-wall distance 0.41"· 0.42"· 0.37"' 0.36"' lumbar flexion index -0.42"· -0.53"· -0.41"· -0.5 Γ" lumbar lateral flexion -0.40*" -0.45*" •0.4 Γ* -0.44"* cervical anteflexion -0.28* •0.41"* -0.40'" cervical retroflexion -0.35* -0.37"· -0.37" -0.36" cervical lateral flexion -0.45*" -0.48"· -0.35* -0.41*·· cervical rotation -0.32* -0.50"· -0.29* -0.49""

Explanation of symbols: p-values: ' « ρ < 0.05; " = ρ < 0.01; '" = ρ < 0.005 showed a statistical significant difference. Table S shows the scores and change in scores for men and women, and for the two treatments. To evaluate the influence of disease duration, sex and HLA-B27 positivity on the radiographic scores, a multiple regression analysis was done on all six scores at week 48. Long disease duration and male sex showed an unfavourable significant influence on the enuresis score. HLA-B27 did not reach significance in any of the scores. Pearson correlations between all assessed clinical variables at weeks 0 and 12 and the spinal and overall score were analysed, and those which revealed statistical significant correlations are shown in Table 6.

Discussion

A scoring method for radiographic assessment in AS has been evaluated in this study. Previously published radiographic scoring systems of the hips13 and of the lumbar spine4 have been used. The scoring system of the lumbar spine has been modified and made applicable for the cervical spine as well. Next to this, a scoring system for entheses of the pelvis and femural bone was developed and evaluated. During the standardization procedure many difficulties were experienced with the pubic symphysis score, reason to exclude this score. The radiological evaluation of the sacro-iliac joint was left out of consideration, as the presence of sacroiliitis is fundamental to diagnosis of AS. The interobserver correlations of the lumbar and cervical spine scores were good (r > 0.95). Although the cervical spine score showed a statistically significant difference between the scores of the two observers, the interobserver duplicate error of 0.55 in a range from 0 to 36 can be considered as relatively small. The enthesis and hip scores have been scored twice by the two observers, because interobserver

136 A radiographic scoring system in AS correlations were less than 0.95. The intraobserver correlations of both scores were satisfying (see Table 2). As it concerns the enthesis score one observer is obligate for this part of the scoring system. In our group of patients the thoracolumbar area and the lower lumbar vertebrae most frequently showed total bony bridging. In the past Resnick et al.13 have reported a predominant involvement of the thoracolumbar area with subsequent ascent. The same study13 has found that the upper levels of the cervical spine most frequently were involved, together with the cervicothoracic area. AU these findings- were confirmed by our data. However Braunstein et al.'4 have reported a spinal ascend in the cervical spine. Comparison of squaring is made impossible, because different definitions for squaring of vertebrae have been used in other studies1115, or the frequency of squaring has not been reported1. Pelvic whiskering has been reported to be present in 44% of cases13, without reporting patients' characteristics. Since it is plausible that radiographic involvement and progression increases with disease duration, we cannot compare out data with those published by others. The radiographic scores progressed statistically significant during the first year of follow-up, although only half of the patients showed changes. However, it should be noted that the radiographs have been scored in sequential order with each other, so only a positive difference, i.e. progression, is possible. In previous studies it has been reported that radiological involvement and progression was less severe in women13,14,1', whereas others did not find these differences15,17. Apart from the differences in severity, some studies have found an increased involvement of the cervical spine in women13,1'. In our study significant differences between men and women have been found in the spinal score and overall score, but neither in the cervical spine score nor in the change in scores over 48 weeks. However, if we corrected for disease duration, no significant differences between men and women could be found. Thus, it might be concluded that disease progression and radiographic involvement is similar in men in women in this study. Kennedy et al.19 recently have reported that onset of disease in women is later than in men, previously already reported by Resnick et al.13. Our data might confirm this, however we have studied only 57 patients. Chest expansion, occiput-to-wall distance, lumbar flexion index, lumbar lateral flexion and cervical anteflexion, retroflexion, lateral flexion and rotation revealed significant correlations with radiographic scores at weeks 0 and 12 and might be identified variables measuring structural damage. In future studies evaluation should contain a much longer follow-up time. The enthesis score should be evaluated anew, and possibly modified, after extensive standardization. The identification of a subset of variables which show highly significantly correlations with radiographic scores is important for interpretation of results in future studies, since these variables might not be sensitive to change regarding disease activity. In conclusion, this extensive radiographic scoring system is a very useful method to assess radiographic damage in AS.

137 Chapter Χ

References

I. Taylor HG, Wardle Τ, Beswick EJ, Dawes PT. The relationship of clinical and laboratory measure­ ments to radiological change in ankylosing spondylitis. Br J Rheum 1991:30-330-5 2 Moll JMH. Ankylosing spondylitis Churchill Livingstone, Edingburgh, 1980 3. Carette S. The natural disease course of ankylosing spondylitis. Arthritis Rheum 1983,26-186-90 4. Mau W, Zeidler H, Mau R et al. Clinical features and prognosis of patients with possible ankylo­ sing spondylitis Results of a 10-year follow-up. J Rheumatol 1988,15-1109-14 5. Linden van der S, Valkenburg HA, Cats A Evaluation of diagnostic entena for ankylosing spondylitis, a proposal for modification of the New York entena Arthritis Rheum 1984,27-361-8 6. Pocock SJ Sequential treatment assignment with balancing for prognostic factors in the controlled clinical tnal. Biometrics 1975,36 81-90 7 Schober P. Lendenwirbelsaule und Kreuzschmerzen. Muench Med Wschr 1937,84-336 8. Domján L, Nemes Τ, Bálint GP, Tóth Ζ, Gomör Β. A simple method for measuring lateral flexion of the dorsolumbar spine. J Rheumatol 1990,17 663-5 9. Mander MM, Simpson JM, McLellan A et al. Studies with an enthesis index as a method of cluneal assessment in ankylosing spondylitis. Ann Rheum Dis 1987,46-197-202 10. Creemers MCW, Hof van 't MA, Franssen МММ et al. A Dutch version of the functional mdex for ankylosing spondylitis. Development and validation in a long-term study. Br J Rheum 1994,33 842-6 II. Ralston SH, Urquhart GDK, Brzeski M, Sturrock RD. A new method for the radiological assess­ ment of vertebral squaring in ankylosing spondylitis. Ann Rheum Dis 1992,51-330-3 12. Larsen A, Dale K, Eck M. Radiographic evaluation of rheumatoid arthritis and related conditions by standard reference films Acta Radiol 1977,18-481-91 13 Resnick D, Dwosh IL, Goergen TG et al. Cluneal and radiographic abnormalities m ankylosing spondylitis- a comparison in men and women Radiology 1976,119 293-7 14. Braunstein EM, Martel W, Moidel R. Ankylosing spondylitis in men and women- a clinical and radiographic comparison. Radiology 1982,144-91-4 15. Spencer DG, Park WM, Dick HM, Papazoglou SN, Buchanan WW Radiological manifestations m 200 patients with ankylosing spondylitis- correlation with cluneal features and HLA B27. J Rheumatol 1979,6305-15 16 McBryde AM, McCollum DE. Ankylosing spondylitis in women. The disease and its prognosis. N Carol Med J 1973.34 34-7 17. Jiménes-Balderas FJ, Mintz G. Ankylosing spondylitis' cluneal course in women and men. J Rheumatol 1993,20-2069-72 18 Tyson TL, Thompson WAL, Ragan C. Mane-StrUmpell spondylitis m women. Ann Rheum Dis 1953,12-40-42 19. Kennedy LG, Will R, Calm A. Sex ratio m the spondylarthropathies and its relationship to phenotypic expression, mode of inheritance and age at onset J Rheumatol 1993,20-1900-4

138 Chapter XI

Summary and conclusions

Summary and conclusions

Ankylosing spondylitis (AS) is a chronic inflammatory disease mostly affecting the sacroiliac joints and the spine. This inflammation may often lead to ankylosis, resulting in a limited mobility of the back and eventually in a decreased functional capacity. Structural damage, in fact the final result of the inflammatory process, can partly be seen on X-rays as ankylosis and syndesmophyte formation. Treatment mainly consists of chronic intermittent administration of nonsteroidal antiinflam­ matory drugs (NSAIDs), however a growing interest is seen for 2nd line treatment (such as sulfasalazine, methotrexate, hydroxychloroquine), since these drugs might influence the course of the disease. In this thesis, several aspects of first and second-line treatment in AS were studied. In addition, attention was paid to assessment of the disease.

ß-cyclodextrin-piroxicam versus naproxen

The comparison of ß-cyclodextrin-piroxicam and naproxen was evaluated in 59 patients (ß-cyclodextrin-piroxicam: η = 29; naproxen: η = 30) with AS during 48 weeks. Efficacy and toxicity are described in chapter 2. In general there was no difference in efficacy and toxicity between ß-cyclodextrin-piroxicam and naproxen. At the end of the study 38 patients still used the study medication (21 patients treated with ß-cyclodextrin-piroxicam, 17 patients treated with naproxen). The mean drug survival time is 41 weeks (standard deviation = sd = 14 weeks) for ß-cyclo­ dextrin-piroxicam, 36 weeks (sd = 16 weeks) for naproxen, which is not statistical­ ly significant different. Comparison between the two treatments of commonly used clinical variables over 48 weeks and the mean change from week 0 to week 2 did not reveal any difference in efficacy.

Gastrointestinal toxicity of NSAIDs

In chapter 3 gastrointestinal (GI) toxicity of the two NSAIDs is described. For this purpose the HemoQuant Assay has been used, a method which measures haem- derived porphyrins. Advantages of this HemoQuant Assay are that it is a non­ invasive technique, and that it evaluates the entire GI tract. Evaluation of the entire GI tract is of crucial importance, because NSAIDs can cause (severe) mucosal injury in the small and large intestines, and additional to this, the frequency of inflammatory bowel disease is known to be increased in AS. The amount of faecal blood loss did not differ between ß-cyclodextrin-piroxicam and naproxen over 48 weeks, however the amount of faecal blood loss at week 4 was increased compared to the other measurement occasions, i.e. weeks 0, 24 and 48. This increase indicates a possible mechanism of gastric adaptation, as reported in previous studies with aspirin.

141 Chapter XI

Pseudoporphyria due to naproxen In chapter 4 we describe three cases of pseudopoiphyria due to naproxen, of whom two patients participated in the NSAID study. All literature on pseudopor- phyria due to naproxen has additionally been reveiwed.

Pharmacokinetic interaction of ß-cyclodextrin-piroxicam with H2-receptor antagonists

The possible interaction between ß-cyclodextrin-piroxicam and the two H2-receptor antagonists, Cimetidine and ranitidine, has been studied at steady state conditions of all drugs. Two groups of each eight healthy male volunteers were included in the study, and during 40 days blood and urine samples have frequently been taken. Absorption, metabolism and excretion of ß-cyclodextrin-piroxicam did not change during concomitant administration of the two H2-receptor antagonists. Pharmaco­ kinetics of Cimetidine and ranitidine did not change either during ß-cyclodextrin- piroxicam administration (chapter 5 and appendix).

Second-line treatment

In chapter 6 a review of the literature on second-line treatment in seronegative spondylarthropathies is reported. Second-line treatment reviewed included hydroxy­ chloroquine, gold, d-penicillamine, sulfasalazine, methotrexate, cyclophosphamide, 6-mercaptopurine, azathioprine, cyclosporin and levamisole as well as radiothera­ py, methylprednisolone and corticotrophin. Only sulfasalazine has been studied in sufficient detail to allow definitive conclusions, but methotrexate and azathioprine may be promising drugs. Radiation therapy led to short-term improvement in AS, but was abandoned because of severe long-term side effects.

Methotrexate in ankylosing spondylitis

Methotrexate (MTX), therefore, was studied in 11 patients with severe AS in an open study of 36 weeks (chapter 7). Evaluation of efficacy was evaluated in two different ways for objectivity, i.e. the relative difference of assessed variables between weeks 0 and 24, and by deciding at week 24 either to continue or to stop MTX based on patients' global assessment. In total nine patients were évaluable at week 24: one patient could stop NSAID treatment, three could reduce the dose of NSATDs, and three patients deteriorated after discontinuation of MTX and opted to

142 Summary and conclusions restart. Side effects were reversible or transient, and no severe side effects were seen.

Functional assessment in ankylosing spondylitis

Chapter 8 describes the development and evaluation of Dutch Functional Index for the assessment of functional status in patients with AS. For this purpose a French Functional Index was translated, and items concerning static activity and bicycle riding, typically Dutch, have been added. In 150 patients with AS fulfilling the modified New York criteria, reliability has been studied (cronbach's α = 0.95). Reproducibility and criterion validity were studied in two other groups of patients with AS (n = 39 and η = 19, respectively) and considered good. Finally, in the NSAID study the questionnaire was completed at the start of the study medication and every 12 weeks thereafter in order to study sensitivity to change on the longterm and for evaluation of the measurement-remeasurement correlation. In conclusion, the Dutch Functional Index, comprising 34 questions, turned out to be a questionnaire, which is reliable, reproducible, valid and sensitive to change for assessment of functional status in patients with AS.

Assessment of process variables of disease activity

As discussed before, the assessment of disease activity comprises many difficulties. Because of this, all data of the two clinical studies, i.e. the NSAID study and the MTX study, have been used to evaluate 27 commonly used clinical variables (chapter 9). For evaluation of reproducibility, discriminating power and the correlation of variables and factors with disease duration, a predefined analysis strategy was used, which included principal component analysis to form subgroups of reliable, interrelated variables. This evaluation resulted in a selection of seven process variables. The core set of seven process variables encompassed lumbar flexion index, spinal pain during the day, enthesis index, functional index, C- reactive protein (CRP), root joints and the number of swollen joints. In subsequent discriminant analysis two variables had to be excluded, namely the number of swollen joints and the lumbar flexion index for the computation of a disease activity score (AS-DAS). The formula for the AS-DAS is: AS-DAS = 0.238*(Dutch Functional Index) + 1.13*sqrt(enthesis index) + 0.318*(spinal pain during the day) + 6.54*ln(CRP) + 5.245*double square root(root joints). This AS- DAS has to be validated in the future. A core set of process variables for disease activity in AS improves standardization in future studies, thus improving comparability of different clinical studies. The problem of multiple testing, caused by the large number of variables, is solved, especially if in future the validated AS-

143 Chapter XI

DAS can be used as a single, composed variable for evaluation in clinical studies in AS.

Radiographic scoring systems

Structural damage, in fact the result of the disease, can be seen on X-rays. To evaluate structural damage in AS, a radiographic scoring system is described in chapter 10. This radiographic scoring system is composed of: 1. a radiographic scoring system of the lumbar spine, in essence a modification of a previously published radiographic scoring system for the lumbar spine; 2. a similar system for the cervical spine; 3. the Larsen score for the hip; and 4. a radiographic enthesis score of the ischial tuberosities, the iliac alae, and the minor and major trochanter of the femoral bone. This extended radiographic scoring system has been validated. Reproducibility and duplicate errors are satisfactory. Subsequently X-rays made at weeks 0 and 48 of the NSAID study have been scored with this system. The change in scores from weeks 0 to 48 was highly statistically significant, indicating that the radiographic system is sensitive to change. The mean change from week 0 to 48 as well as the scores at weeks 0 and 48 reveal no statistically significant difference between ß-cyclodextrin-piroxicam and naproxen, respondéis and non- responders, and men and women.

144 Appendix

Ion-pair solid-phase extraction of Cimetidine from plasma and subsequent analysis by high- performance liquid chromatography

FGM Rüssel, MCW Creemers, Y. Tan, PLCM van Riel, FWJ Gribnau

Accepted for publication: Journal of Chromatography B, June 1994

Solid-phase extraction of Cimetidine and analysis bij HPLC Summary

An improved method is described for the solid-phase extraction of Cimetidine from plasma or serum with subsequent analysis by high-performance liquid chromatog­ raphy (HPLC). New aspects of the method include protein precipitation with metaphosphoric acid (5%wlv), followed by selective adsorption of Cimetidine and the internal standard ranitidine on the surface of a solid-phase phenyl (PH Bond Elut) column, using octanesulfonate as an ion-pairing agent. Separation was achieved on a LiChrosorb RP-18 column with a mobile phase consisting of acetonitrile - 0.01 M phosphate buffer pH 3.0 containing 0.005 M octanesulfonate (22:78 vlv). Intra-assay coefficient of variation varied between 0.7 and 4.0%. The procedure provides cleaner and more stable samples and a better recovery (90 ± 23%) and sensitivity (limit of detection 5 nglml and limit of quantitation 25 nglml) as compared to previous methods.

Introduction

A large number of high-performance liquid chromatographic methods (HPLC) for the quantitation in biological fluids of Cimetidine, an H2-receptor antagonist widely used for the treatment of gastric and duodenal ulcers, has been published. For the sample clean-up of plasma or serum one of the investigators used acetonitrile for the precipitation of protein1. Some investigators employed single or multiple liquid- liquid extraction procedures2"9, and others made use of solid-phase extraction10"15. None of the authors who employed the solid-phase extraction technique for the sample clean-up of plasma or serum took account of the protein binding of Cimeti­ dine, which is about 13 to 26%16. This is probably one of the reasons why some authors have reported an absolute recovery as low as 59%12 and 73%u. Another important reason for the low recovery, and in most papers high coefficient of variations, may be heterogeneity and channeling in the solid-phase bed. However, when the solid-phase extraction technique is used optimally, an absolute recovery of 90% with negligible interferences should always be achievable. The mean particle size of the commonly used bonded silica sorbents is 40 urn17. The nominal porosity of most of the sorbents is 60 Â, adequate for compounds with molecular weights up to approximately 15,000. Molecules larger than this (molecular weight of human serum albumin = 69,000) are excluded from the 60 Â pores and are exposed to too little of the surface area of the sorbent for sufficient interactions with the sorbent functional groups. As a consequence, they pass through these sorbents without being retained. Cimetidine bound to albumin will be lost in this manner. Therefore, we incorporated a precipitation step in our solid-phase extraction method using metaphosphoric acid. This does not add much time to the procedure, as precipitation of the protein can be carried out in a large series of samples. A second new element introduced in our procedure is ion-pair solid-phase extraction.

147 Appendix

After deproteinizatìon the pH of the resulting supernatant is acidic. Under this condition Cimetidine and ranitidine (internal standard) are principally not retained by non-polar solid-phase extraction columns. In order to cause retention of both compounds on PH (Phenyl) columns we added, octanesulfonate, an ion-pairing agent, to the supernatant. Both protein precipitation and ion-pair solid-phase extraction provide cleaner samples and a better recovery and sensitivity as compa­ red with previously published methods.

Experimental

Reagents and materials Cimetidine and ranitidine hydrochloride were purchased from Sigma (St. Louis MO, USA). Potassium dihydrogen phosphate, disodium hydrogen phosphate 2- hydrate, orthophosphoric acid, metaphosphoric acid were all of analytical grade (Merck, Darmstadt, Germany). Methanol and acetonitrile were HPLC grade (Merck), and 1-octanesulfonic acid sodium salt was from Janssen Chimica (Beerse, Belgium). Vac Elut manifold and Bond Elut PH (Phenyl) columns (1 ml capacity, to which a 4 ml Bond Elut reservoir was attached using a Bond Elut adaptor) were manufactured by Analytichem International (Habor City, CA, USA). Available reverse osmosis water (Millipore) was further purified with a MilliQ system (Millipore) before use. Human blood samples, stored at -20 °C, were obtained from the local blood bank.

Standard solutions Stock solutions of Cimetidine (1 mg/ml) and ranitidine (1 mg/ml) were made with 0.01 M phosphate buffer pH 3.0. Working standard solutions were prepared each time by diluting the stock solutions with the same buffer. A 100 ul aliquot of the working standard solutions was used. Stock solutions of Cimetidine and ranitidine in 0.01 M phosphate buffer pH 3.0 were stable for at least 6 months when stored in the dark at 4 °C.

Apparatus and chromatographic conditions The chromatography system consisted of a Spectra Physics P2000 binary gradient pump, a Hewlett Packard S um ODS Hypersil guard column (20 χ 2.1 mm), a Hewlett Packard 5 urn LiChrosorb RP-18 analytical column (200 χ 4.6 mm) and a Spectra Physics AS 3000 autosampler with a built-in column heater. The mobile phase, acetonitrile - 0.01 M phosphate buffer pH 3.0 containing 0.005 M octanesul­ fonate (22:78 v/v) maintained under a helium sparge during use, was delivered at a rate of 1 ml/min, the resulting pressure being 8.0 MPa. The column effluent was monitored with a Spectra Physics UV1000 variable wavelength detector set at 228 nm. The signal was processed by a Spectra Physics SP4400 integrator. The column heater was set at 40 °C and the injection volume was 10 ul (50 ul at concentrations lower than 0.05 ц^ті).

148 Solid-phase extraction of Cimetidine and analysis bij HPLC Sample pretreatment Into a plastic tube were pipetted successively, 0.5 ml plasma, 100 ul internal standard (5 mg ranitidine per 100 ml 0.01 M phosphate buffer pH 3.0) and 2 ml of a 5% (w/v) metaphosphoric acid solution. The tube was vortex-mixed for 5 s. After standing for 5 min at room temperature, the tube was again vortex-mixed for 5 s. Therafter, the tube was centrifuged at 2000g and 20 °C for 10 min. The supernatant was transferred into another plastic tube, and 0.8 ml 0.3 M NajHPO^ solution containing 0.02 M octanesulfonate was added. The tube was then vortex-mixed for 5 s. The prepared sample was now ready to be loaded onto the PH (Phenyl) solid- phase extraction column. Extraction procedure Extraction of Cimetidine and ranitidine (internal standard) from plasma was achieved by the use of a bonded silica solid phase extraction column (Bond Elut PH, 1 ml capacity). The column was conditioned prior to use by drawing three column volumes (appr. 3 ml) of methanol followed by a similar volume of 0.01 M phosphate buffer pH 3.0 containing 0.005 M octanesulfonate through the column. The prepared sample was loaded onto and subsequently drawn through the column. The column was then washed with four column volumes of 0.01 M phosphate buffer pH 3.0 containing 0.005 M octanesulfonate. The cover of the manifold was then removed and the blunt nose stainless-steel needle of the Vac Elut cover was wiped with a tissue to remove drops of washing solution. The Vac Elut rack holding a 2-ml glass sampling tube, was placed under the column. Cimetidine and ranitidine were eluted from the column with 0.75 ml of an elution mixture, acetonitrile - 0.01 M phosphate buffer pH 3.0 containing 0.005 M octanesulfonate (30:70 v/v). To make the elution strength of the collected eluate equal to that of the mobile phase, 275 ul 0.01 M phosphate buffer pH 3.0, containing 0.005 M octane­ sulfonate, was added. This was done to prevent peak broadening whenever larger volumes (50 ul) had to be injected onto the column. The eluate was then analysed as described. The eluate was found to be stable for more than 2 weeks at room temperature and more than 4 weeks stored at 4 °C.

Results and discussion Chromatography Figure 1 shows typical chromatograms for blank plasma and for blank plasma spiked with Cimetidine and the internal standard (I.S.) ranitidine. The retention times of Cimetidine and ranitidine were 5.85 and 7.57 min, respectively, while an endogenous peak had a retention time of 6.37 min. Without the presence of the ion- pairing agent octanesulfonate retention times of Cimetidine and ranitidine were 2.87 and 3.28 min. The efficiency of the analytical column used to obtain the chromatogram in Fig. 1

149 Appendix

Fig. 1. Chromatograms obtained for blank plasma and blank plasma spiked with 1.25 pg/ml Cimetidine (1) and 1.25 pg/ml ranitidine (2). Injection volume = 10 pi.

Ш Ш о о с in to in ja P- о Ξ о X S3 xJ3 CM to см «о II II I!E

'

Calculation The Cimetidine concentration in a sample was determined from a standard curve of peak height ratio versus Cimetidine concentration. Whenever a sample containing Cimetidine was measured, a standard curve was generated by adding different amounts of Cimetidine to blank plasma. A linear relationship was found between the peak height ratio of Cimetidine to ranitidine (Y) and the plasma Cimetidine concentration (X), as given by the equation Y = 0.841X + 0.002 (r = 0.9999, л = 6) for the plasma Cimetidine concentration range 0.25 - 5 pg/ml (LS. = 5 pg/ml) and Y = 8.472X + 0.009 (r = 0.9997, η = 5) for the plasma Cimetidine concentration range 0.025 - 0.25 pg/ml (LS. = 0.5 pg/ml).

150 Solid-phase extraction of Cimetidine and analysis bij HPLC

Recovery Overall recovery with the extraction procedure was determined by comparing the peak heights of Cimetidine and ranitidine obtained after injection of non-extracted standard solutions with peak heights obtained after injection of extracted plasma containing equal concentrations of both compounds. The absolute recovery of Cimetidine and ranitidine is 90% and independent of the concentration with a coefficient of variation (C.V.) equal or less than 2.3% (Table 1)

Efficiency of the deproteinization procedure To investigate the efficiency of protein precipitation by 5% (w/v) metaphosphoric acid, 0.3 ml aliquots of a series of ten spiked plasma samples were deproteinized. After centrifugation the volume of the supernatant was measured. The volume was found to be 2.43 ml (C.V. = 0.8%, n=10), which is 90% of the initial volume (0.5 ml plasma + 2 ml 5% metaphosphoric acid + 100 ul standard solution + 100 ul internal standard solution). From these results it was concluded that Cimetidine and ranitidine existed completely free in the deproteinized sample, 90% in the superna­ tant and 10% in the pellet.

Sensitivity, precision and applicability The limit of detection (two times baseline noise) of the method described was 5 ng/ml. Table 2 shows the precision and accuracy of Cimetidine measurement in plasma. The intra-assay C.V. varied between 4.0 and 0.7% over a Cimetidine concentration range of 0.063 - 2.500 pg/ml. At the limit of quantitation (five times limit of detection) the C.V. was 14.1%. Various drugs that are often concurrently administered with Cimetidine were examined for possible interference with the assay. Plasma samples from patients who were treated with therapeutic doses of acetylsalicylic acid, amiloride, atenolol, azapropazone, diclofenac, furosemide, hydrochlorothiazide, indomethacin, naproxen, omeprazole, oxazepam, paracetamol, phenylbutazone, pinracam, prednisone, salazopyrine, and salicylic acid were examined. None of the drugs showed interfer-

Table 1. Recovery of Cimetidine and ranitidine from plasma'

Compound Concentration Recovery C.V. (ц$/т1) (%) (%)

Cimetidine 0.25 90 2.3

2.50 90 2.3

Ranitidine 0.50 89 2.2

2.50 91 1.2

'For all concentrations, η = 6: C.V. = coefficient of variation.

151 Appendix

Table 2. Precision and accuracy of Cimetidine analysis in spiked plasma samples'

Spiked Observed C.V. Accuracy concentration concentration (%) (%) (Mg/ml) (Hg/ml)

0.025 0.025 14.1 100.0

0.063 0.064 4.0 98.4

0.125 0.128 3.9 97.7

0.188 0.194 2.7 96.9

0.250 0.245 2.3 102.0

0.625 0.638 2.6 98.0

1.250 1.298 0.7 96.3

1.875 1.839 1.1 102.0

2.500 2.503 0.7 99.9

'For all concentrations, л = 5; C.V. = coefficient of variation.

ing peaks at theretention times fo r Cimetidine and ranitidine. The H2-antagonists nizatidine and metiamide do not interfere, whereas theretention time o f famotidine coincides with that of Cimetidine.

Conclusion The Cimetidine method described here for plasma or serum is a significant im­ provement over previously published methods in terms of sensitivity, recovery, stability and cleanliness of the final sample. It should prove to be valuable for clinical monitoring of plasma or serum levels and detailed pharmacokinetic studies.

References

1. Kunitani MG, Johnson DA, Upton RA, Riegelman S. J Chromatogr 1981;224:156 2. Randolph WC, Osborne VL, Walkcnstein SS, Intoccia AP. J Pharm Sci 1977;66:1148 3. Lee RM, Osborne PM. J Chromatogr 1978; 146:354 4. Ziemniak JA, Chiarmonte DA, Schentag JJ. Clin Chem 1981;27:272 5. Fleitman J, Torosian O, Perrin JH. J Chromatogr 1982:229:255 6. Mihaly GW, Cockbain S, Jones DB, Hanson RG, Smallwood RA. J Pharm Sci 1982;71:590 7. Guay DRP, Bockbrader HN, Matzke GR. J Chromatogr 1982,228:398 8. Boutagy J, More DG, Munro IA, Shenfield GM. J Liq Chromatogr 1984:7:1651 9. Abdel-Rahim M, Ezra D. Peck C, Lazar J. Clin Chem 1985;31:621-3

152 Solid-phase extraction of Cimetidine and analysis bij HPLC

10. Kozma M, Vereczkey L. J Chromatogr 1983:273:223 11. Nitsche V, Maschcr H. J Chromatogr 1983;273:449 12. Bartlett JM, Segelman AB. J Chromatogr 1983;255:239 13. Lin Q, Lensmeycr GL, Larson FC. J Anal Toxicol 1985;9:161 14. Chiou R, Stubbs RJ, Bayne WF. J Chromatogr 1986;377:441 15. Strong HA, Spino M. J Chromatogr 1987:422:301 16. Somogyi A, Gugler R. Clin Pharmacokinet 1983;8:463 17. Sorbent Extraction Technology, ed. Home van КС. Analytichem International, Harbor City, USA

153

Samenvatting en conclusies

Een Nederlandse samenvatting van dit proefschrift voor niet-medici

Spondylitis ankylopoëtica

Spondylitis ankylopoëtica (SA), ook wel de ziekte van Bechterew genoemd, is een chronisch reumatische ziekte. 'Spondylitis' betekent een ontsteking van de gewrichtjes van de wervelkolom, en 'ankylopoëtica' dat deze ontsteking kan leiden tot verstijving van deze gewrichtjes. Het sacro-iliacale gewricht, tussen darmbeen en heiligbeen in het bekken, ondergaat een soortgelijke ontsteking en verstijving. De oorzaak van SA is niet bekend. Wel is het duidelijk dat er een erfelijke aanleg tot het krijgen van de ziekte aanwezig is in de vorm van een bepaalde 'bloed­ groep', het HLA-B27. Doorgaans manifesteert SA zich tussen het 18e en 35e levensjaar. De ziekte komt bij mannen drie keer zoveel voor als bij vrouwen. De belangrijkste symptomen van SA zijn pijn en stijfheid. De pijn zit voorna­ melijk in de (onder)rug, is vroeg in de ochtend op z'n hevigst, en gaat gepaard met stijfheid. Doorgaans verdwijnt deze pijnlijke stijfheid na beweging. Ook kan er pijn en stijfheid van de borstkas optreden, evenals ontstekingen van de overgangen van bot naar pees - enthesis genoemd - en van gewrichten van de ledematen. Al deze verschijnselen bij elkaar kunnen een verminderd functioneren veroorzaken. Ook buiten het bewegingsapparaat kan SA zich manifesteren, enkele voorbeelden hiervan zijn een regenboogvliesontsteking en aften in de mond en op de geslachts­ organen. Bovendien komen chronische ontstekingsziekten van de darm (ziekte van Crohn, colitis ulcerosa) en psoriasis vaker voor. Op de röntgenfoto is blijvende, irreversibele gewrichtsschade zichtbaar, zoals: a) totale verbening van het sacro-iliacale gewricht; b) verkalking van de randen van de tussenwervelschijven, waardoor de wervels als het ware aan elkaar vast­ groeien; c) kalkafzetting en een aanvreten van het bot (zgn. erosies) op de plaats van de entheses; en d) erosies en ernstige beschadiging van gewrichten van de ledematea Geneesmiddelen, gebruikt bij de behandeling van SA, zijn de zogeheten niet- steroïde anti-inflammatoire geneesmiddelen (NS AID's), ook wel eerste lijns anti­ reumatica genoemd. NSATD's hebben een pijnstillend en ontstekingsremmend effect. Helaas veroorzaken ze frequent bijwerkingen van het maag-darm kanaal, variërend van misselijkheid tot zelfs een maagbloeding. Tweede lijns anti-reumati- ca worden al geruime tijd gebruikt bij andere reumatische aandoeningen, en het lijkt dat ze dan het beloop van de ziekte in gunstige zin beïnvloeden. Ze werken alleen ontstekingsremmend en daardoor hebben ze ook hun weerslag op de pijn en stijfheid. Hun werking treedt doorgaans pas na enkele maanden gebruik op, en daarom worden ze in combinatie met NS AID's voorgeschreven. Tot voor een

155 Samenvatting en conclusies

aantal jaren geleden was het vrij ongebruikelijk om voor een zo 'onschuldige' aandoening als SA tweede lijns anti-reumatica voor te schrijven. Er is echter een groeiende interesse voor deze geneesmiddelen in de behandeling van SA. Naast medicamenteuze behandeling speelt oefentherapie en fysiotherapie een belangrijke rol. Immers, om verstijving in voorovergebogen houding te voorkomen, is een goede houding en goed gebruik van de rug noodzakelijk. Ademhalingsoefe­ ningen kunnen nuttig zijn als de patiënt pijn aan de borstkas heeft, die bij elke ademteug gevoeld wordt. Uiteindelijk kan de aandoening, ondanks behandeling met medicamenten, leiden tot bewegingsbeperkingen en een verminderd functione­ ren. Maatschappelijke begeleiding en ergotherapie kunnen dan nodig wordea

Dit proefschrift

In dit proefschrift is aandacht besteed aan zowel de eerste als de tweede lijns anti- reumatica bij de behandeling van SA. Tevens is onderzoek verricht naar de wijze van meten van ziekte-acüviteit, blijvende schade, en functionaliteit (wat kunnen patiënten nog aan lichamelijke activiteiten, doorgaans een optelsom van ziekte- activiteit en restschade ten gevolge van ontsteking). De verschillende studies met hun conclusies zullen hieronder worden samengevat.

ß-cyclodextrine-piroxicam (Brexine) versus naproxen in SA Twee NSAID's die veel gebruikt worden zijn piroxicam en naproxen. Door Piroxi­ cam te verbinden met de suiker, ß-cyclodextrine, wordt piroxicam sneller opgeno­ men in de bloedbaan, waardoor er mogelijk minder maag-darm bijwerkingen zouden kunnen optreden. Bij 59 SA-paüënten zijn de effectiviteit (werkzaamheid) en toxiciteit (mate van bijwerkingen en nadelige effecten) van ß-cyclodextrine- piroxicam vergeleken met die van naproxen in een 48 weken durende studie (hoofdstuk 2). Patiënten werden in totaal 12 keer gezien op de polikliniek. Om in aanmerking te komen voor deelname aan dit onderzoek moesten patiënten een duidelijk actieve ziekte hebben. Dit werd bekeken door de patiënten voor de start van het onderzoek te laten stoppen met het door hun gebruikte NSAID, en dan de ernst van hun klachten te beoordelen. Om tot een zo eerlijk mogelijke beoordeling van deze twee geneesmiddelen te komen, is het onderzoek dubbel-blind uitgevoerd. Dit betekent dat zowel de patiënt als de behandelend arts, c.q. de onderzoeker, niet wisten met welk van de twee geneesmiddelen de patiënt behandeld werd. Beide NSAID's lieten een significante verbetering zien na de start van de behandeling. Er was geen verschil in effectiviteit en toxiciteit tussen de twee NSAID's.

Maag-darm bijwerkingen van NSAID's Hoofdstuk 3 beschrijft hoe in bovenstaand NSAID-onderzoek de maag-darm toxi­ citeit nader werd onderzocht. Naast het aantal maag-/darmklachten, is de mate van bloedverlies in het maag-/darmkanaal gemeten. De hoeveelheid bloedverlies in het

156 Samenvatting en conclusies maag-Maimkanaal geeft aan in hoeverre bet slijmvlies van het maag-/darmkanaal geïrriteerd is, d.w.z. hoe meer bloedverlies des te meer irritatie. Het bloedverlies werd in ontlasting gemeten, die verzameld was gedurende een periode van 48 uur, bij de start van het NSAID (week 0) en na 4, 24 en 48 weken behandeling. Er waren geen verschillen tussen de beide NSAID's in het aantal en de ernst van optredende maag-/darmbijwerkingen, en in de mate van bloedverlies. Het bloed­ verlies bij SA-paüënten was aanmerkelijk hoger (gemiddeld 1600 ml op jaarbasis) dan bij gezonde vrijwilligers die NSAID's gebruiken (gemiddeld 400 ml per jaar). Dit verschil zou kunnen berusten op de veelvuldig bij SA voorkomende chronische ontstekingsziekten van de darm, die ook gepaard kunnen gaan met bloedverlies. Opmerkelijk was dat er een toename van bloedverlies van week 0 naar week 4 te zien was, en een daling van week 4 naar week 24 en 48. Dit wijst op een mogelijk aanpassingsmechanisme van het maagslijmvlies bij chronisch NSAID-gebruik.

Pseudoporfyrie ten gevolge van naproxen In het NSAID-onderzoek ontwikkelden twee patiënten een niet vaak voorkomende huidbijwerking van naproxen - pseudoporfyrie. Pseudoporfyrie is een huidaandoe­ ning die blaren geeft bij aan zonlicht blootgestelde huid, waarbij met name bet gelaat, de hand- en voetrug aangedaan zijn. De blaren genezen met littekenvorming en aldaar verminderde pigmentarie. Na het stoppen van naproxen verdwijnt de aandoening. In hoofdstuk 4 wordt alle beschikbare literatuur over pseudoporfyrie ten gevolge van naproxen kort samengevat, en wordt de ziektegeschiedenis van drie patiënten beschreven.

Farmacokinetische interactie tussen ß-cyclodextrine-piroxicam en twee H2-receptor antagonisten NSAID's kunnen maagzweren veroorzaken, die dan met een ander geneesmiddel, een Hj-receptor antagonist, behandeld worden. H2-receptor antagonisten remmen de maagzuurproduktie. Omdat reumapatiënten niet langdurig zonder NSAID's kunnen, worden de twee geneesmiddelen vaak samen voorgeschreven. Dit kan leiden tot interactie. Hoofdstuk 5 beschrijft een onderzoek bij gezonde vrijwilligers naar de interactie van ß-cyclodextrine-piroxicam met twee frequent voorgeschreven Hj- receptor antagonisten, Cimetidine (Tagamet) en ranitidine (Zantac). De farmacoki- netiek - opname, verdeling, omzetting door de lever en uitscheiding door de nier - van piroxicam werd niet gewijzigd door de H2-receptor antagonisten. Dit betekent dat er geen interactie bestaat tussen ß-cyclodextrine-piroxicam en Cimetidine en ranitidine. In de appendix wordt een nieuwe methode beschreven om de concen­ tratie van Cimetidine in bloed en urine te bepalen.

Tweede lijns behandeling Een literatuuronderzoek naar de tweede lijns behandeling van spondylarthropathie- en, de groep ziekten waartoe SA behoort, is beschreven in hoofdstuk 6. Tot de tweede lijns behandeling worden gerekend: hydroxy-chloroquine, goud, d-penicil-

157 Samenvatting en conclusies

lamine, sulfasalazine, methotrexaat, cyclofosfamide, 6-mercaptopurine, azathiopri- ne, Ciclosporine, levamisol, alsmede radiotherapie en hoimoontoediening in de vorm van prednison of corticotrofine. Alleen naar sulfasalazine (salazopyrine) waren voldoende onderzoeken gedaan om een conclusie te kunnen trekken omtrent de werkzaamheid bij spondylarthropathieën. Methotrexaat en azathioprine leken veelbelovend in diverse rapportages. Radiotherapie liet een snel effect zien, maar bleek op de lange termijn een verhoogd risico op kanker, waaronder met name bot- en bloedkanker, te geven.

Methotrexaat bij de behandeling van ernstige SA Na dit literatuuronderzoek is besloten om methotrexaat (MTX) te bestuderen in een kleine groep van patiënten met ernstige SA, die niet genoeg baat hadden bij behandeling met NSAID's en salazopyrine. Patiënten werden in een open onder­ zoek gedurende 24 weken behandeld met één toediening MTX per week, terwijl de dosis van het NSAID constant werd gehouden. Twee patiënten vielen uit, en in week 24 bleek er een verbetering te zien te zijn bij het merendeel van de paü- entea Patiënten mochten zelf beslissen aan de hand van het door hun ervaren effect of ze wilden doorgaan met de behandeling met MTX. Vier patiënten besloten door te gaan met MTX, bij drie van hen kon de dosis van het NSAID worden verminderd en een patiënt kon volledig stoppen met het NSAID. De overige vijf patiënten besloten te stoppen met MTX. Bij drie van deze vijf verergerde de ziekte daarna zodanig dat ze opnieuw startten met MTX. Bijwer­ kingen waren alle mild en voorbijgaand.

Functionaliteit bij SA Hoofdstuk 7 beschrijft de ontwikkeling en evaluatie van een Nederlandse Functi­ onele Index om de functionaliteit bij SA-patiënten te meten. Voor dit doel werd een bestaande Franse vragenlijst vertaald. Omdat deze Franse vragenlijst niet alle aspecten van functionaliteit leek te bevatten, zijn er vragen toegevoegd met betrekking tot statische activiteiten, zoals 30 minuten zitten of staan, alsmede vragen rondom een typisch Nederlandse activiteit: fietsen. De Nederlandse vragen­ lijst bevat in totaal 34 vragen over onder meer het zichzelf aankleden, wassen, voorover bukken, iets optillen of oprapen, een bepaalde tijd staan of zitten enzovoort. Om te onderzoeken of deze Nederlandse Functionele Index reprodu­ ceerbaar was, daadwerkelijk mat wat patiënten konden en gevoelig genoeg was voor veranderingen, werden verschillende groepen patiënten onderzocht. Conclude­ rend bleek deze Nederlandse Functionele Index een betrouwbaar, reproduceerbaar en gevoelig instrument te zijn om functionaliteit te meten bij SA-patiënten.

Ziekte-activiteit bij SA Zoals voorafgaand beschreven, zijn pijn en stijfheid de belangrijkste symptomen bij SA. Het is algemeen bekend dat deze twee symptomen moeilijk objectiveerbaar zijn. Dit blijkt ook uit het grote aantal beschikbare metingen. Zo zijn er talloze

158 Samenvatting en conclusies manieren ontwikkeld om de beweeglijkheid van de wervelkolom te bepalen. In het bloed kan de bezdnking (een maat voor ontsteking) verhoogd zijn, evenals de concentratie van bepaalde eiwitten zoals het CRP en het IgA, maar dat is niet zo bij elke SA-patiënt. Een probleem bij al deze variabelen is dat het onduidelijk is in hoeverre ze nu daadwerkelijk ziekte-activiteit meten, dan wel restschade. Bijvoorbeeld bij pijn is de beweeglijkheid van de wervelkolom afgenomen, maar dat is ook het geval als de wervelkolom is verstijfd zonder dat dit pijnlijk is. In hoofdstuk 8 is beschreven hoe uit 27 veelgebruikte variabelen een selectie gemaakt kon worden van een kleine groep variabelen die met name ziekte-activiteit meten. De waarnemingen bij de patiënten uit de NSAID- en de MTX-studie, zijn onderverdeeld in twee groepen, namelijk die met hoge en die met lage ziekte-activiteit. Vervolgens is er met diverse statistische methodes bekeken hoe deze twee groepen zich van elkaar onderscheidden. Het bleek dat er een groepje variabelen geselecteerd kon worden, dat het beste het verschil weergaf tussen hoge en lage ziekte-activiteit Uiteindelijk kon er met een bepaalde statistische analyse een formule gemaakt worden voor een ziekte-activiteitsscore bij SA (AS-DAS). Met behulp van deze formule kan de ziekte-activiteit als een getal worden weergegeven, waarbij een hoog getal overeenkomt met een hoge ziekte-activiteit, en andersom. De formule maakt gebruik van vijf variabelen: a) de Nederlandse Functionele Index; b) een enthesis index (een maat voor pijn aan bot-peesaanhechtingen); c) het CRP (een maat voor ontsteking); d) rugpijn overdag (gemeten op een schaal van 0 tot 10 cm); en e) de 'wortel-gewrichts' index (een maat voor pijn van heupen en schouders). De AS- DAS is een eerste aanzet om met minder variabelen ziekte-activiteit bij SA te meten, echter de AS-DAS zal in de toekomst geëvalueerd moeten worden net zoals met de Nederlandse Functionele Index is gebeurd.

Röntgenscore bij SA Op de röntgenfoto is de blijvende schade ten gevolge van de ontsteking bij SA zichtbaar, dit wordt röntgenschade genoemd. Het uiteindelijk gewenste resultaat van een behandeling zou in het ideale geval een uitblijven van schade moeten zijn. Om röntgenschade en -progressie van de ziekte nauwkeurig vast te stellen is een meetmethode nodig. Daarom wordt in hoofdstuk 9 de ontwikkeling en evaluatie van een röntgenscore methode voor SA beschreven, bestaande uit een score van de nek- en lendewervelkolom, de entheses van het bekken en de heupgewrichten. In het NSATD-onderzoek (hoofdstuk 2) zijn hiervoor röntgenfoto's gemaakt bij de start van het onderzoek (week 0) en in week 48. Er wordt verondersteld dat met NSAID's voornamelijk een symptomatische behandeling wordt gegeven, zonder dat er iets verandert aan het beloop van de ziekte. Zodoende mag aangenomen worden dat er progressie van de ziekte is ondanks behandeling met NSAID's. Een statistisch significante toename in röntgenscore van week 0 naar week 48 kon worden vastgesteld. Hieruit blijkt dat deze röntgenscore-methode gevoelig genoeg is om veranderingen te meten. Er werd geen verschil gezien in de mate van röntgenprogressie tussen de beide bestudeerde NSAID's, ß-cyclodextrine-piroxicam

159 Samenvatting en conclusies en naproxen, tussen mannen en vrouwen, en tussen degenen die wel en die niet goed reageerdeno p de behandeling met NS AID's.

160 Dankwoord

het schrijven van dit dankwoord bleek vergelijkbaar met de sleutelpassage van een moeilijke beklimming mogelijk of onmogelijk.

Daarom: BEDANKT, iedereen die bewust of onbewust, direct of indirect, een steentje heeft bijgedragen aan de berg die resulteerde in dit proefschrift!!!

Deze berg steentjes beschouwend wil ik een aantal mensen met name noemen: Allereerst alle patiënten, deelnemers van de Bechterew-oefengroep te Malden en proefpersonen, zonder wie dit proefschrift er niet zou zijn. Vervolgens de mede­ werkers van de polikliniek interne geneeskunde van het Academisch Ziekenhuis Nijmegen (AZN) St. Radboud en de St. Maartenskliniek (SMK), bedankt voor jullie inzet en de prettige sfeer. Het spijtserum van de patiënten van de SMK werd verzorgd door de medewerkers van het laboratorium aldaar. Alle IgA en CRP bepalingen werden verricht op het centraal klinisch-chemisch laboratorium van het AZN. Thijs van Doom ontwierp de 'helm' en de 'schuifmaat'. Nauwgezet werden alle gegevens in de computer ingevoerd door Carien Versteegden en Sytske Nawijn. Erik Brummelkamp schreef hiervoor de benodigde programma's en stond me met raad en daad bij als er gerekend en geanalyseerd moest worden. Hulp bij de ontwikkeling en validering van de Nederlandse Functionele Index kreeg ik van: Dr. W. (Wim) van Lankveld, Antoinette Smeijers, Susan Terwindt, Ans de Boer, Wilma Rutten, Els Kok-Truijens en Nan Pluijmackers. Annie van Schaijck leerde mij de HemoQuant techniek, en Caria Bessems verricht­ te met grote nauwkeurigheid deze bepalingen. Dr. A. (Albert) Tangerman was altijd op de achtergrond aanwezig. De afdeling farmacologie wil ik bedanken voor het gastvrij verlenen van onderdak voor mij en mijn proefpersonen. Yuen Tan ontwikkelde de analyse methode voor Cimetidine, en voerde alle analyses uit van Cimetidine en ranitidine van de vele plasma- en urinemonsters. Dr. F. (Frans) Rüssel, bedankt voor je duidelijke uitleg van de farmacokinetiek, je positieve ideeën over deze 'negatieve' studie, en ook de vele gezellige momenten. Dr. P. (Piet) van Riel en Dr. M. (Marcel) Franssen leverden het oorspronkelijke idee voor het NSAID-onderzoek. Ik wil jullie allebei bedanken voor de ruimte die ik kreeg om mijn eigen ideeën in dit onderzoek te verwerken. Marcel introduceerde mij in de reumatologie, met name voor wat betreft de spondylitis ankylopoëtica, en scoorde stapels röntgenfoto's. Piet, bedankt voor de vele ontelbare gaatjes en gaten in je drukke agenda. Ik bewonder je geduld als het mij allemaal weer eens niet snel genoeg ging, en je dan toch op je eigen manier alles bekeek. Dr. M. (Martin) van 't Hof, jouw statistische blik op de kliniek werkte erg objecti­ verend, en voor mij bovenal motiverend om te proberen de verschillende analyses te snappen. Vele uren hebben we samen doorgebracht achter de computer, ze waren nimmer saai, ik kwam altijd met plezier. Mijn promotores, Prof. Dr. L. (Leo) van de Putte en Prof. Dr. F. (Frank) Gribnau, bedankt voor het in mij gestelde vertrouwen. Leo, jouw kritische blik op de

161 verschillende artikelen en de discussies, ruimte gevend voor mijn mening, heb ik gewaardeerd. Frank, zonder jouw enthousiasme voor de combinatie van weten­ schappelijk onderzoek en kliniek, zou ik er nooit over hebben gedacht om zelf onderzoek te gaan doen. Het is me hartstikke goed bevallen. Jouw positieve benadering van welk resultaat dan ook haalde doorgaans snel mijn negativisme weg. Alle collegae en medewerkers van de afdelingen reumatologie van het AZN en de SMK wil ik bedanken voor de prettige sfeer, Marlou Prevoo, Anke van Gestel en mijn keldergenoot Jan Stolk vooral bedankt voor het meedenken en jullie begrip bij al mijn ups en downs. Bert Thomassen ontwierp de omslag, bedankt voor je artistieke bijdrage aan dit proefschrift. Op het laatst werd op voortreffelijke wijze de lay-out en het correctiewerk overge­ nomen door Ants Creemers en René de Bruin. Ants, bedankt voor je kritische en speurende blik naar typefouten en andere schoonheidsfoutjes. René, de lay-out en alle tabellen zien er schitterend uit. Bedankt daarvoor, maar nog meer voor alle momenten dat je er 'zomaar' was.

162 Curriculum vitae

Maijonne Creemers werd op 17 september 1963 geboren in Venlo. Na het behalen van het diploma Gymnasium ß aan het SL Thomascollege te Venlo, startte zij in 1981 met de studie geneeskunde. Het doctoraalexamen weid in november 1986 behaald. Van september 1987 tot en met mei 1990 liep zij co-assistentschappen, met onder meer een facultatief co-schap oncologie (Uniklinik te Keulen, Prof. dr. H.O. Klein) en een co-schap gezondheidszorg in ontwikkelingslanden (Quthing, Lesotho). Het artsexamen werd in juni 1990 behaald. In oktober 1990 startte zij met het onderzoek dat in dit proefschrift resulteerde, uitgevoerd op de afdeling reumatologie van het Academisch Ziekenhuis Nijmegen (AZN) St. Radboud, in samenwerking met de afdeling reumatologie van de St. Maartenskliniek Nijmegen, de afdeling gastro-enterologie AZN St. Radboud, farmacologie en de medisch statistische afdeling van de Katholieke Universiteit Nijmegen (KUN). Tijdens het onderzoek was zij in de gelegenheid zich verder te bekwamen in methodologie en statistiek door het volgen van de basiscursus biometrie en bijbehorende keuzeblok- ken voor AIO's (verzorgd door de medisch statistische afdeling, KUN). Tevens doorliep zij met goed gevolg de 'writing course for junior researchers' (verzorgd door de vakgroep Engels, KUN). Sedert juli 1994 is zij werkzaam als Assistent- Geneeskundige-In-Opleiding (AGIO) binnen de afdeling interne geneeskunde van het Canisius-Wilhelmina Ziekenhuis te Nijmegen (opleider: dr. R.W. de Koning).

163

STELLINGEN

Behorende bij het proefschrift Clinical and methodological studies in ankylosing spondylitis

I De toename van bloedverlies in de tractus digestivus bij patiënten met spondylitis ankylopoëtica - die nog wordt vergroot door het gebruik van eerste-lijns antireumatica (NS AID's) - is een van de argumenten om op salazopyrine over te schakelen. dit proefschrift; J. Hayllar, Arthritis Rheum. 1994;8:1146-50

Π Men dient zich af te vragen of het feit dat pseudoporfyrie ten gevolge van naproxen zelden wordt gezien het gevolg is van daadwerkelijk zeldzaam voorkomen of van onderdiagnostisering. eigen waarneming, dit proefschrift

Ш Het is frappant dat, hoewel de effectiviteit van tweede-lijnsgeneesmiddelen bij invaliderende auto-immuunziekten sinds 1960 in dubbelblinde klinische studies is bestudeerd, deze benadering pas 20 jaar later bij spondylitis ankylopoëtica aan de orde kwam. eigen waarneming, dit proefschrift

TV Op basis van het aantal gemaakte cervicale-wervelkolomfoto's zou men verwachten dat de prevalentie van nekklachten bij patiënten met spondylitis ankylopoëtica laag is; het tegendeel is het geval. eigen waarneming

V Bij patiënten met een continu hoge ziekte-activiteit gedurende een lange periode, kan verbetering na interventie in een open studie niet alleen verklaard worden door 'regression to the mean'.

VI Door het overigens terecht grote belang dat aan statistische significantie wordt gehecht, dreigt de klinische relevantie als punt van beschouwing vaak uit het oog verloren te worden. eigen waarneming VII Omdat een praktizerend medicus altijd leidinggevende taken heeft, reeds te beginnen in de patiëntenzorg, zou een managementcursus een verplicht onderdeel moeten zijn van de opleiding.

Ш Werktijden naar de letter van de dienaangaande Algemene Maatregel van Bestuur kunnen mede verwezenlijkt worden door niet iedereen de bevoegdheid te geven op elk moment het sein van de arts-assistent te bellen.

EX Zolang vrouwen uit eigen beweging kiezen voor part-time werken omwille van kinderen worden mannen zelden gedwongen die keuze te maken.

X De belangstelling voor stellingen is omgekeerd evenredig met de tijd die het kost om ze te maken.

Nijmegen, 12 december 1994

Marjonne Creemers

ISBN 90-9Θ07674-3