Proefschrift ramiro by Nicole Nijhuis

ANKYLOSING SPONDYLITIS: assessment and analysis of long-term outcome Sofia Ramiro

ANKYLOSING SPONDYLITIS: ANALYSIS AND ASSESSMENT OF LONG-TERM OUTCOME

Sofia Ramiro 2015

ISBN: 978-94-6108-903-8 Sofia Ramiro was supported by the Fundação para a Ciência e Tecnologia (FCT) grant SFRH/BD/68684/2010

Printing of thesis was financially supported by AbbVie, AMC-UvA, Amgen B.V., ChipSoft, Dutch Arthritis Foundation, Janssen-Cilag B.V., Novartis Portugal, Pfizer B.V., Rijnland Zorggroep, UCB Pharma B.V., which is gratefully acknowledged.

ANKYLOSING SPONDYLITIS: ASSESSMENT AND ANALYSIS OF LONG-TERM OUTCOME

ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus prof. dr. D. C. van den Boom ten overstaan van een door het college voor promoties ingestelde commissie, in het openbaar te verdedigen in de Agnietenkapel op vrijdag 27 februari 2015, te 14:00 uur

door

Maria Sofia Antunes da Cunha Oliveira Ramiro geboren te Lissabon, Portugal

PROMOTIECOMMISSIE Promotores:

Prof. dr. R.B.M. LandewĂŠ

Copromotor:

Dr. A.M. van Tubergen

Universiteit van Amsterdam

Prof. dr. D.M.F.M. van der Heijde Universiteit Leiden Universiteit Maastricht

Overige leden: Prof. dr. J.W.J. Bijlsma

Universiteit van Amsterdam

Prof. dr. D.L.P. Baeten

Universiteit van Amsterdam

Prof. dr. M. Maas

Universiteit van Amsterdam

Prof. dr. A.E.R.C.H. Boonen

Universiteit Maastricht

Faculteit der Geneeskunde

‘All men by nature desire to know’ Aristotle

To my parents To grandmother Nazaré

CONTENT

Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6

Chapter 7 Chapter 8

Chapter 9

General introduction

How to deal with missing items in BASDAI and BASFI Rheumatology (Oxford) 2014;53:374-76

Reference intervals of spinal mobility measures in normal individuals: the mobility study Ann Rheum Dis 2014, March 24. doi: 10.1136/annrheumdis-2013-204953

Neutral lateral fingertip-to-floor distance can be derived from height Ann Rheum Dis 2014;73:1748-9

Hierarchy of impairment of spinal mobility measures in ankylosing spondylitis: 12 year data from the OASIS cohort Submitted

Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study Ann Rheum Dis 2015;74:52-9

Higher disease activity leads to more structural damage in the spine in ankylosing spondylitis: 12 year longitudinal data from the OASIS cohort Ann Rheum Dis 2014;73:1455-61

111

Chapter 10

129

Mechanical stress and smoking may modify the effect of disease activity on radiographic progression in patients with ankylosing spondylitis Submitted

Chapter 11

Summary and conclusions

145

Chapter 12

Samenvatting en conclusies

163

PhD Portfolio

176

List of publications

182

Curriculum Vitae

187

Acknowledgements

188

General Introduction

INTRODUCTION Axial Spondyloarthritis (axSpA) is a chronic rheumatic disease, with predilection for the axial skeleton. Back pain - usually of inflammatory character and reflecting inflammation of the sacro-iliac joints and/or spine - and spinal stiffness, as well as peripheral arthritis, enthesitis and uveitis are among the most common clinical manifestations of patients with axSpA. AxSpA is a broad disease concept that includes 1) non-radiographic axSpA, i.e. patients with signs and symptoms characteristic of the disease as the above-mentioned but without radiographic sacroiliitis; and 2) radiographic axSpA, in which patients also have radiographic sacroiliitis and it corresponds to what is commonly known as Ankylosing Spondylitis (AS). Although we still lack long-term follow-up studies, we know that a substantial proportion of the patients with non-radiographic axSpA progress to AS

. It is important to identify patients timely in order

1,2

to treat them early and in an attempt to influence the course of the disease. The Assessment of SpondyloArthritis international Society (ASAS), an international group of experts in the field of SpA, developed new classification criteria for the disease, covering the full spectrum of axSpA 3. The reason to develop these criteria was to be able to conduct clinical trials in patients reflecting the entire spectrum, including those with non-radiographic axSpA, as well as to facilitate the identification of patients with axSpA in an early stage. These new criteria signal a paradigm shift from AS to axSpA. Disease activity, spinal mobility and radiographic damage are central themes of outcome in axSpA and are included in the core set of outcome domains for patients with axSpA/AS, as defined by ASAS 4. This core set of domains reflects in part the tremendous progress that we have witnessed in the field of AS during the last years: Standardization of outcome assessment is pivotal in obtaining a better understanding of outcome in a chronic disease. Scores for the assessment of radiographic damage have been developed and evaluated 5-9, which enables us now to reliably measure new bone formation, which is characteristic for AS. At the start of the studies performed for this thesis, it was already known that the presence of syndesmophytes is the best predictor of future radiographic progression

. Relationships

10-13

between several outcomes were investigated in the past, showing that inflammatory markers, such as C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR), contribute to some extent to radiographic progression

. Remarkably, though, disease activity, as measured

12-15

with the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), a fully patient-reported index, did not seem to be related to radiographic progression

. The debate about the

12,13

relationship between disease activity and radiographic progression in AS 16 was also fuelled by the repeated observation that tumor-necrosis factor-alpha-inhibiting (TNFi) biologicals, that are very effective in suppressing signs and symptoms of AS, as well as acute phase reactants, did not inhibit radiographic progression in AS 17-19. 10

Chapter 1

Previous studies have suggested male gender and smoking to be associated with radiographic progression

. Radiographic damage, in turn, was shown to be associated with impaired

12,13

spinal mobility

20,21

However, an unequivocal picture of determinants of radiographic progression was failing for

several (often methodological) reasons: 1) disease activity measurement was not optimal; 2) studies were too short; and 3) data-analysis was suboptimal. 1. Disease activity measurement

Recently, assessment of disease activity in axSpA has been improved with the introduction of the AS Disease Activity Score (ASDAS), which now combines patient-reported outcomes and acute phase reactants (either CRP or ESR levels). ASDAS has shown superior psychometric properties compared to the BASDAI 22. 2. Long term data: the Outcome in Ankylosing Spondylitis International Study

The Outcome in Ankylosing Spondylitis International Study, the OASIS cohort, is nowadays the cohort of patients with AS with probably the longest follow-up available according to a predefined protocol. It is a cohort of prevalent cases with AS, starting in 1996 with 217 patients from the Netherlands, Belgium and France 23. Patients have been meticulously followed-up for 12 years. In the first 6 years of follow-up, assessments were completed at least annually, thereafter every 2 years. Assessments included questionnaires, clinical investigations, physical examinations, laboratory assessments, and radiographic assessments of the pelvis, and the cervical and lumbar spine (every 2 years). Because the vast majority of OASIS patients has never been exposed to TNFi, and because of its long standardized follow-up, this cohort provides a unique opportunity to study the â&#x20AC;&#x2DC;natural historyâ&#x20AC;&#x2122; of AS and outcomes on the long-term.

3. Longitudinal data analysis

OASIS offers us the possibility to conduct longitudinal analyses, which have until now scarcely been applied in AS studies. Longitudinal analyses are characterized by using all data collected over time, which also makes them more powerful, due to the more efficient data usage. Most importantly, they adjust for so-called within-subject correlation (between observations), which is essential when analysing data from longitudinal studies and using data from multiple observations in the same patient. Longitudinal analysis allows us to get better insight into the development of an effect over time and also to gain insight into longitudinal relationships between variables. Furthermore, they may to some extent allow a temporal relationship, thus bringing in an element of causality

. For all these reasons, longitudinal analyses are

extremely helpful to investigate the complex relationships between disease activity, spinal mobility, radiographic progression and other determinants. General Introduction

In summary, at the start of the analytical studies performed in the context of this thesis, there was a need to investigate long-term outcome in AS and to investigate longitudinal associations between variables, and there was an appropriate setting (the 12 year OASIS cohort) available allowing us to conduct the appropriate analyses. However, there were still some issues with a few assessments, which had to be solved first. In general terms, the aims of this thesis were: 1. To improve several facets of outcome assessment in AS, in particular disease activity, spinal mobility and radiographic damage, so that these outcomes could be used in the analysis of outcome of AS in the long-term; 2. To analyse the course of radiographic damage over time and get more insight into syndesmophyte formation; 3. To investigate longitudinal relationships between disease activity, spinal mobility and radiographic damage and other potential determinants of these outcomes. Disease activity and function assessment

Frequently disease activity in AS is assessed using the BASDAI, a fully patient-reported instrument. In analogy, physical function is often reported using the Bath AS Functional Index (BASFI). So far, clear instructions about how to deal with missing items in these questionnaires were not available, which could hamper the appropriate performance of the analyses we were planning with regard to this thesis. In chapter 2, we have presented methods to deal with missing data in BASDAI and BASFI. Mobility

The natural course of AS is associated with a progressive impairment in spinal mobility, a hallmark of the disease 25. ASAS recommends the assessment of spinal mobility according to standardized measures

. However, the extent to which spinal mobility measures are

impaired in patients with AS compared to healthy individuals (â&#x20AC;&#x2DC;the normsâ&#x20AC;&#x2122;) were not known. In chapter 3, we describe a study in normal individuals, the MOBILITY-study, aiming at deriving

reference intervals of spinal mobility measures. The idea was to derive percentile curves in analogy to the growth curves for monitoring childrenâ&#x20AC;&#x2122;s growth 27. In chapter 4 we provide an

imputation method for cases in which lateral spinal flexion was inappropriately measured (a frequently seen omission in studies). In chapter 5, finally, a formal comparison of spinal mobility measures between patients with

AS and reference values from normal individuals was performed. The aim was to investigate whether a hierarchy for the impairment of the spinal mobility measures could be established 12

Chapter 1

and to further investigate whether assessing fewer measures could reliably capture â&#x20AC;&#x2DC;full informationâ&#x20AC;&#x2122; regarding spinal mobility, which may facilitate assessment of spinal mobility in daily clinical practice.

Radiographic damage and progression

In a formal comparison of radiographic damage scoring methods, the modified Stoke AS Spine Score (mSASSS), has shown best reliability and sensitivity to change 8. Afterwards, a slightly modified version of the mSASSS, the Radiographic AS Spinal Score (RASSS) has been proposed, that includes the lower thoracic vertebrae 9. In order to obtain the best method for our long-term analysis in the OASIS, a comparison between the mSASSS and RASSS was conducted. This comparative analysis is described in chapter 6. Long-term analysis of radiographic progression has never been performed. In chapter 7 we

have introduced longitudinal data modelling in AS. The course of radiographic damage over time was investigated using generalised estimating equations, taking into account factors that may significantly modify the course of progression. An extension of this basic model was described in chapter 8. The mSASSS quantifies

radiographic damage with a score of 1 per vertebral corner for the presence of erosion, sclerosis and/or squaring and a score of 2 for the presence of syndesmophytes, thus suggesting that syndesmophytes are a natural consequence of erosions, sclerosis and squaring, which was based on pathophysiological considerations. However, it was unknown whether erosion, sclerosis and/or squaring truly precede the development of syndesmophytes. Chapter 8 describes a formal longitudinal analysis to address this research question. Relationship between inflammation and radiographic progression

As mentioned before, the relationship between disease activity and radiographic progression was still poorly defined in AS. An important methodological limitation was that â&#x20AC;&#x201C;in the absence of appropriate long-term cohorts- this relationship was only described in cross-sectional settings. The OASIS cohort provides a better setting to investigate to what extent disease activity causes radiographic progression. Results of a detailed longitudinal analysis in the OASIS cohort are presented in chapter 9. Using the longitudinal model described in chapter

9 as a basic template, chapter 10 finally describes an extension of this basic longitudinal

model, with emphasis on factors that may contribute to explaining radiographic progression in patients with AS, such as mechanical stress and smoking status. Chapter 11 consists of a summary and general discussion on the findings of this thesis. A summary of this thesis in Dutch is provided in Chapter 12.

General Introduction

REFERENCES 1.

Poddubnyy D, Rudwaleit M, Haibel H, et al. Rates and predictors of radiographic sacroiliitis progression over 2 years in patients with axial spondyloarthritis. Ann Rheum Dis 2011;70:1369-74.

Sampaio-Barros PD, Bortoluzzo AB, Conde RA, et al. Undifferentiated spondyloarthritis: a longterm followup. J Rheumatol 2010;37:1195-9.

Rudwaleit M, van der Heijde D, Landewe R, et al. The development of Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis (part II): validation and final selection. Ann Rheum Dis 2009;68:777-83.

MacKay K, Mack C, Brophy S, et al. The Bath Ankylosing Spondylitis Radiology Index (BASRI): a new, validated approach to disease assessment. Arthritis Rheum 1998;41:2263-70. Averns HL, Oxtoby J, Taylor HG, et al. Radiological outcome in ankylosing spondylitis: use of the Stoke Ankylosing Spondylitis Spine Score (SASSS). Br J Rheumatol 1996;35:373-6. Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9. Wanders AJ, Landewe RB, Spoorenberg A, et al. What is the most appropriate radiologic scoring method for ankylosing spondylitis? A comparison of the available methods based on the Outcome Measures in Rheumatology Clinical Trials filter. Arthritis Rheum 2004;50:2622-32. Baraliakos X, Listing J, Rudwaleit M, et al. Development of a radiographic scoring tool for ankylosing spondylitis only based on bone formation: addition of the thoracic spine improves sensitivity to change. Arthritis Rheum 2009;61:764-71.

10. Baraliakos X, Listing J, Rudwaleit M, et al. Progression of radiographic damage in patients with ankylosing spondylitis: defining the central role of syndesmophytes. Ann Rheum Dis 2007;66:910-5. 11. Baraliakos X, Listing J, von der Recke A, et al. The natural course of radiographic progression in ankylosing spondylitis--evidence for major individual variations in a large proportion of patients. J Rheumatol 2009;36:997-1002. 12. Poddubnyy D, Haibel H, Listing J, et al. Baseline radiographic damage, elevated acute-phase

Chapter 1

reactant levels, and cigarette smoking status predict spinal radiographic progression in early axial spondylarthritis. Arthritis Rheum 2012;64:1388-98. 13. van Tubergen A, Ramiro S, van der Heijde D, et al. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23. 14. Kroon F, Landewe R, Dougados M, et al. Continuous NSAID use reverts the effects of inflammation on radiographic progression in patients with ankylosing spondylitis. Ann Rheum Dis 2012;71:1623-9. 15. Haroon N, Inman RD, Learch TJ, et al. The Impact of TNF-inhibitors on radiographic progression in Ankylosing Spondylitis. Arthritis Rheum 2013;65:2645-54. 16. Maksymowych WP, Elewaut D, Schett G. Motion for debate: the development of ankylosis in ankylosing spondylitis is largely dependent on inflammation. Arthritis Rheum 2012;64:1713-9. 17. van der Heijde D, Salonen D, Weissman BN, et al. Assessment of radiographic progression in the spines of patients with ankylosing spondylitis treated with adalimumab for up to 2 years. Arthritis research & therapy 2009;11:R127. 18. van der Heijde D, Landewe R, Einstein S, et al. Radiographic progression of ankylosing spondylitis after up to two years of treatment with etanercept. Arthritis Rheum 2008;58:1324-31. 19. van der Heijde D, Landewe R, Baraliakos X, et al. Radiographic findings following two years of infliximab therapy in patients with ankylosing spondylitis. Arthritis Rheum 2008;58:3063-70. 20. Wanders A, Landewe R, Dougados M, et al. Association between radiographic damage of the spine and spinal mobility for individual patients with ankylosing spondylitis: can assessment of spinal mobility be a proxy for radiographic evaluation? Ann Rheum Dis 2005;64:988-94. 21. Machado P, Landewe R, Braun J, et al. Both structural damage and inflammation of the spine contribute to impairment of spinal mobility in patients with ankylosing spondylitis. Ann Rheum Dis 2010;69:1465-70. 22. van der Heijde D, Lie E, Kvien TK, et al. ASDAS, a highly discriminatory ASAS-endorsed disease activity score in patients with ankylosing spondylitis. Ann Rheum Dis 2009;68:1811-8. 23.

Spoorenberg A, van der Heijde D, de Klerk E, et al. Relative value of erythrocyte sedimentation rate and C-reactive protein in assessment of disease activity in ankylosing spondylitis. J Rheumatol 1999;26:980-4.

24. Twisk J. Applied longitudinal data analysis for epidemiology: a practical guide. Cambridge:

Cambridge University Press; 2003. 25. Carette S, Graham D, Little H, et al. The natural disease course of ankylosing spondylitis. Arthritis Rheum 1983;26:186-90.

age, weight-for-age, weight-for-length, weight-forheight and body mass index-for-age: Methods and development. Geneva: World Health Organization, 2006 (312 pages).

26. Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68 Suppl 2:ii1-44.

27. WHO Multicentre Growth Reference Study Group. WHO Child Growth Standards: Length/height-for-

General Introduction

How to deal with missing items in BASDAI and BASFI Sofia Ramiro, Astrid van Tubergen, Désirée van der Heijde, Filip van den Bosch, Maxime Dougados, Robert Landewé

Rheumatology 2014;53:374-6

Sir, The BASDAI and BASFI are frequently used for the assessment of disease activity and function, respectively, in patients with axial SpA

. However, the calculation of BASDAI

1,2

and BASFI scores is only defined for cases when all the questions (6 for BASDAI and 10 for BASFI) are answered. No official instructions are available on how to deal with missing items (MIs). Theoretically even a single MI precludes calculation of the score. Our aim was to investigate the extent of missing answers in BASDAI and BASFI and to select the best strategy to substitute them. Data from the Outcome in Ankylosing Spondylitis International Study (OASIS), a 12 year follow-up prevalence cohort of patients with AS, were used for this analysis 3. BASDAI (010) and BASFI (0-10) questionnaires were regularly applied

. First, the number of MIs per

1,2

questionnaire was evaluated. Further analyses were restricted to questionnaires with no MIs. Varying numbers of MIs were randomly generated per questionnaire: 1, 2, 3 and 4 MIs (and additionally 5 and 6 for BASFI), which were replaced by five strategies: (i) lowest value of the possible score, i.e. value of 0; (ii) middle value of the possible score, i.e. value of 5; (iii) highest value of the possible score, i.e. value of 10; (iv) mean of the remaining items and (v) median of the remaining items. Additionally, for the BASDAI, one MI was randomly generated in one of the morning stiffness items (question 5 or 6) and replaced by the other. The scores resulting from the different imputation techniques were compared with the original scores, taking various levels of agreement into account. The levels of agreement were derived from the smallest detectable change (SDC), i.e. 1.4 (obtained by dividing the smallest detectable difference of 2 cm

by √2 5). The best imputation techniques were conservatively chosen

as those achieving an agreement >90% within a difference between original and imputed scores of ≤0.7 (half of SDC). This cut-off was chosen in analogy with the European League Against Rheumatism (EULAR) response criteria in RA: moderate response (improvement ≥0.6), defined as half of the good response (improvement ≥1.2) 6,7. A total of 216 patients were included (mean age 44 (S.D. 13) years, mean symptom duration 21 (S.D. 12) years, 71% males, 85% HLA-B27 positive). Over the 12 years, 1747 BASDAI questionnaires were obtained: 1719 (98.4%) had no MI, 24 (1.4%) had one MI, 1 (0.1%) had two MIs, 1 (0.1%) had three MIs and 2 (0.1%) had five MIs. The mean BASDAI, calculated with the complete questionnaires only, was 3.4 (S.D. 2.3). Over the 12 years, 1741 BASFI questionnaires were obtained: 1715 (98.5%) had no MI, 12 (0.7%) one MI, 1 (0.1%) had two MIs, 11 (0.6%) had four MIs, and 2 (0.1%) had six MIs. The mean BASFI was 3.6 (S.D. 2.6). All patients gave their informed consent and the ethics committees from all participating hospitals approved the study. For BASDAI, the agreement between original and imputed scores was highest when one of the morning stiffness items was substituted by the other, with an agreement of 99.7% for 18

Chapter 2

Missings in BASDAI and BASFI

Imputation technique Value 0 Value 5 Value 10 Mean remaining questionnaire items Median remaining questionnaire items Morning stiffness imputed‡ Value 0 Value 5 Value 10 Mean remaining questionnaire items Median remaining questionnaire items Value 0 Value 5 Value 10 Mean remaining questionnaire items Median remaining questionnaire items Value0 Value 5 Value 10 Mean remaining questionnaire items Median remaining questionnaire items Value 0 Value 5 Value 10 Mean remaining questionnaire items Median remaining questionnaire items Value 0 Value 5 Value 10 Mean remaining questionnaire items Median remaining questionnaire items

Level of agreement for BASDAI (%)* Equal Dif ≤|0.3| Dif ≤|0.7| Dif≤|1.0| 15 45 68 81 6 37 76 100 2 11 30 55 20 76 94 98 27 75 92 98 39 91 100 100 6 24 43 53 3 20 45 64 1 2 6 13 57 85 93 13 16 58 83 91 3 15 28 38 2 14 31 43 0 1 3 5 45 71 83 11 10 45 70 81 2 10 20 28 2 12 24 33 0 1 1 2 36 59 72 8 8 36 59 72 ----------------------------------------Dif≤|1.4| 91 100 68 100 99 100 67 84 23 98 96 49 60 10 92 89 38 47 4 84 84 -----------

Level of agreement for BASFI (%)* Equal Dif≤|0.3| Dif≤|0.7| Dif≤|1.0| 24 54 84 100 9 52 100 100 7 23 54 100 32 90 100 100 39 90 99 100 ----14 35 55 70 6 32 64 100 3 8 22 36 81 100 97 25 28 79 99 96 10 28 43 52 3 23 47 63 1 4 10 19 70 93 97 19 22 67 90 96 8 22 35 43 3 17 37 51 1 3 6 11 62 88 96 18 19 60 84 93 6 19 31 36 2 13 31 43 1 2 4 7 16 56 81 91 15 49 75 86 5 17 27 32 2 11 25 35 1 2 3 5 14 53 76 87 13 47 72 84

Dif≤|1.4| 100 100 100 100 100 -85 100 55 100 100 66 89 31 99 99 54 67 19 99 97 45 58 13 97 94 40 48 9 94 92

Figures in italic reflect an agreement >90%, which was the cut-off conservatively chosen by the authors as acceptable for reliable imputation Dif – difference; Value 0, 5, 10 – means that the values 0, 5 or 10 were the values imputed in case of a missing value, respectively; BASDAI – Bath Ankylosing Spondylitis Disease Activity Index; BASFI – Bath Ankylosing Spondylitis Functional Index *Level of agreement: equal – original and imputed scores identical; dif≤|0.3| - a difference of a maximum of 0.3 between the original and imputed scores; dif≤|0.7| - a difference of a maximum of 0.7 between the original and imputed scores; dif≤|1.0| - a difference of a maximum of 1.0 between the original and imputed scores; dif≤|1.4| - a difference of a maximum of 1.4 between the original and imputed scores; ‡Morning stiffness imputed: a missing value was generated for question 5 or 6 of the BASDAI (i.e. morning stiffness) and the value of the remaining (i.e. non-missing) item was imputed

6 missings

5 missings

4 missings

3 missings

2 missings

1 missing

Table 1 – Percentage of agreement between the original and imputed BASDAI and BASFI scores

a difference between scores ≤0.7 (table 1). In practice, this is the same as averaging the remaining items in case of one MI in one of the morning stiffness questions. Further, in the imputation of BASDAI, an agreement >90% for a difference ≤0.7 could only be obtained in questionnaires with one MI (in any of the questions). For these, the best imputation technique was substitution by the mean of the remaining items, reaching an agreement of 94%. For a simplification, we further evaluated the agreement of the average of the remaining five items in case of any MI and it was 91%. Imputing MIs in BASFI resulted in high agreement between the original and imputed scores in up to three MIs imputed (table 1). Similar to the BASDAI, the best imputation technique was substitution by the mean of the remaining items, which again is the same as averaging the remaining items. For a difference between original and imputed scores ≤0.7, an agreement of 100% was obtained for imputation of one MI, 97% for imputation of two MIs and 93% for imputation of three MIs. To our knowledge, this is the first study proposing instructions on how to handle MIs in BASDAI and BASFI based on data. Up to one MI for the BASDAI and three for the BASFI can reliably be imputed. For both the BASDAI and BASFI, we propose averaging the remaining items, as this is easy to perform and makes this substitution strategy suitable for calculation of scores on an individual level and in clinical practice, and independent of the level of disease activity or functional disability. A BASDAI of 4 is frequently used to define eligibility for biologic therapy 8. In case of Ml in the BASDAI, a score can still be calculated and contribute to guide clinical decisions. A commonly used technique to replace missing data is multiple imputation. We decided not to use this simulation-based procedure because its purpose is not to re-create the individual missing values as close as possible to the true ones, but to handle missing data in a way that results in valid statistical inference 9. In this study we did not want to make inferences from relationships between variables, but to derive imputation rules that can be followed at a questionnaire level and be of use for clinical practice. In conclusion, the BASDAI and BASFI can be reliably calculated by averaging the remaining items in case of a maximum of one or three Mls, respectively.

Chapter 2

REFERENCES 1.

Calin A, Garrett S, Whitelock H, et al. A new approach to defining functional ability in ankylosing spondylitis: the development of the Bath Ankylosing Spondylitis Functional Index. J Rheumatol 1994;21:2281-5. Garrett S, Jenkinson T, Kennedy LG, et al. A new approach to defining disease status in ankylosing spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994;21:2286-91.

individual patients: smallest detectable difference or change. Ann Rheum Dis 2005;64:179-82. 6.

van der Heijde DM, van â&#x20AC;&#x2DC;t Hof M, van Riel PL, et al. Development of a disease activity score based on judgment in clinical practice by rheumatologists. J Rheumatol 1993;20:579-81.

van Gestel AM, Prevoo ML, van â&#x20AC;&#x2DC;t Hof MA, et al. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis. Comparison with the preliminary American College of Rheumatology and the World Health Organization/International League Against Rheumatism Criteria. Arthritis Rheum 1996;39:34-40.

Auleley GR, Benbouazza K, Spoorenberg A, et al. Evaluation of the smallest detectable difference in outcome or process variables in ankylosing spondylitis. Arthritis Rheum 2002;47:582-7.

Zochling J, van der Heijde D, Burgos-Vargas R, et al. ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2006;65:442-52.

Bruynesteyn K, Boers M, Kostense P, et al. Deciding on progression of joint damage in paired films of

Rubin DB. Multiple imputation for nonresponse in surveys. New York: Wiley.

Missings in BASDAI and BASFI

Reference intervals of spinal mobility measures in normal individuals: the mobility study Sofia Ramiro, Astrid van Tubergen, Carmen Stolwijk, Désirée van der Heijde, Patrick Royston, Robert Landewé

Ann Rheum Dis 2014, March 24. doi: 10.1136/annrheumdis-2013-204953. [Epub ahead of print]

ABSTRACT Objectives

To establish reference intervals (RIs) for spinal mobility measures as recommended for patients with axial spondyloarthritis, and to determine the effect of age, height and gender on spinal mobility, in normal individuals. Methods

A cross-sectional study (“MOBILITY”) was conducted among normal individuals aged 2069 years. Recruitment was stratified by gender, age (10 year categories) and height (10 cm categories). Eleven spinal mobility measures were assessed. Age specific RIs and percentiles were derived for each measure. Results

393 volunteers were included. All spinal mobility measures decreased with increasing age. Therefore, age specific RIs were developed. The 95% RIs (2.5th and 97.5th percentiles), as well as the 5th, 10th, 25th, 50th, 75th and 90th percentiles for each spinal mobility measure and different ages are presented. Mobility percentile curves were also plotted for each of the measures. For instance, the 95% RI for lateral spinal flexion was 16.2-28.0 cm for a 25year old subject, 13.2-25.0 cm for a 45-year-old subject and 10.1-21.9 cm for a 65-year-old subject. After adjustment for age, there was no need for gender specific RIs, while RIs of some measures are height-adjusted. Conclusion

Age specific RIs and percentiles were derived for each of the spinal mobility measures for normal individuals. These may guide clinicians when assessing the mobility of patients with axial spondyloarthritis. The RIs may serve as cut-off levels for ‘normal’ versus ‘abnormal’, whereas the mobility percentile curves may be used to assess the level of mobility of patients with axial spondyloarthritis.

Chapter 3

INTRODUCTION Axial spondyloarthritis (SpA) is a chronic rheumatic disease that comprises both nonradiographic axial SpA and radiographic axial SpA, the latter known as ankylosing spondylitis (AS). The natural history of axial SpA (axSpA) is associated with a progressive restriction in spinal mobility, a hallmark of the disease 1. Restriction in spinal mobility is one of the most commonly used outcome measures in axSpA follow-up and treatment studies

2,3

and is, at

the same time, a predictor of poor long-term prognosis . The importance of spinal mobility 2

in the follow-up of patients with axSpA is also emphasised by its inclusion in the core set of domains that have been defined by the Assessment of SpondyloArthritis international Society (ASAS) 4. Throughout the last decades spinal mobility has been measured by numerous techniques in patients with AS 5-13. Efforts have been made to homogenise these measures and to identify the most clinically relevant. A composite score was developed, the Bath AS Metrology Index (BASMI), including five measures 14. Subsequently, clinically relevant spinal mobility measures were identified and recommendations were made on assessment in a standardised way for patients with axSpA as incorporated in the ASAS core set 4,14,15. Several studies have been performed among normal individuals with the aim of describing the normal range of spinal movements, and also of identifying factors associated with mobility 5-9,12,13. From these studies, it is known that spinal mobility decreases with age

6,8,9

and also that

there are some differences in spinal mobility with respect to gender . However, the ASAS 6

recommends (slightly) different mobility measures than those previously tested in normal individuals. Consequently, knowledge is lacking on how the ASAS recommended mobility measures behave in normal individuals, which hampers the interpretation of spinal mobility among patients with axSpA. The behaviour of measures in normal individuals is usually established through the determination of reference intervals (RIs). These are commonly used in medical practice for determining unusual or extreme measurements, the premise being that if an individual’s value is outside the interval, further clinical investigation may be required. To construct RIs, a reference sample assumed to consist of “normal” individuals is needed 16,17. The aim of the present study was to derive RIs of spinal mobility measures among normal individuals, assessed as recommended by the ASAS. In order to adequately derive RIs, we first assessed the effect of demographic factors such as age, height, weight and gender on each spinal mobility measure.

Mobility Study

METHODS Study population

A cross-sectional study, MOBILITY, was conducted, aiming at including 400 participants. Participants were recruited at the hospital, universities and also among contacts in the Netherlands and Portugal. Recruitment was stratified for age (10 year categories), height (10 cm categories) and gender in order to guarantee a balanced distribution of these factors in the population, to better take into account their influence on spinal mobility and to enable derivation of more precise estimates for extreme groups (of height or of age), compared to what would occur with a random sample. Female participants were assigned to one of four groups, according to their height: ≤160 cm; 161-170 cm; 171-180 cm; and >180 cm. For male participants, the cut-offs were 10 cm higher, taking into account that men are on average taller than women. According to their age, participants were divided into the following groups: 20-29, 30-39, 40-49, 50-59 and 60-69 years. Across age, height and gender, there were 40 groups, each with 10 individuals. The sample size of 400 participants was mainly based on considerations of convenience, and the requirement that we would like to have at least 10 participants per stratum. Any volunteer agreeing to participate, aged 20-69 years and Caucasian, was a candidate for the study. Exclusion criteria were: (1) history of back surgery; (2) low back pain within the last 3 months that required medical attention; (3) known inflammatory rheumatic disease, osteoporosis, Paget disease or severe osteoarthritis of the spine or hips; (4) more than 3 months’ pregnancy; (5) important scoliosis; (6) symptomatic vertebral disc disease at the moment of the evaluation; (7) hip arthroplasty or any known hip disease; (8) hypermobility syndrome (defined as ≥4 Beighton criteria, which were tested in each participant) 18. All participants gave their informed consent. The study was evaluated by the Ethics Committee of the Maastricht University Medical Center. No formal approval by the Committee was needed, because normal individuals were included and no intervention took place. Spinal mobility measures

Eleven measurements of spinal mobility were performed (measures marked with an asterisk

according to the ASAS recommendations 15), in the following order: tragus-to-wall distance* (TTW), occiput-to-wall distance* (OTW), lateral spinal flexion* (ASAS-LSF), LSF according to the INSPIRE method 19, chest expansion (CE), 10 cm Schober test* (10ST), 15 cm Schober test (15ST) 5, internal hip rotation (IHR)

, intermalleolar distance* (IMD), cervical rotation

lying and cervical rotation sitting*. The best of two tries for each measurement was recorded. All measurements were recorded in centimetres, except for cervical rotation (both lying and sitting) which were recorded in degrees. All measurements were rounded to decimals, except for IHR, IMD and cervical rotation (both lying and sitting) which were rounded to units. 26

Chapter 3

For more details on the measurements, see online supplementary text 1. The BASMI

was calculated and its performance was analysed in normal individuals. The

range of the BASMI is from 0 to10; the higher the score, the worse the mobility. The formula of the BASMI linear was used to compute the score calculated: BASMI-original

and BASMI-modified

and two versions of the score were

. BASMI-original includes TTW, ASAS-

LSF, IMD, 15ST and cervical rotation in the lying position. BASMI-modified

includes two

changes: cervical rotation is measured in sitting position instead of the lying position, and the

15ST is substituted by the 10ST. Spinal mobility measurements were performed by two assessors (SR and CS). The first 100 volunteers were measured by both assessors and their mobility measurements were averaged. After reliability analyses (see online supplementary table S1 and figure S1), agreement between both was considered adequate and each of the assessors proceeded with independently measuring different participants. Statistical analysis

Descriptive statistics (mean, median, SD, range) and box-plots were used to describe the

overall distribution of mobility measurements across age, height and gender categories. Comparison of mobility measurements between groups (eg women vs men) was performed using t-tests or the Mann-Whitney test, as appropriate. Comparison between more than two groups was performed through one-way analysis of variance or the Kruskall-Wallis test, as appropriate. In case of a significant difference between the multiple groups, post hoc analyses were performed to determine significant between-group changes. The association between age, gender, height and each of the spinal mobility measures was investigated through linear regression, adjusted for weight when considered relevant. Univariable followed by multivariable analysis was performed, with forward selection; interactions were also tested. This analysis was intended to identify factors influencing spinal mobility that would eventually need to be taken into account in the development of RIs. Development of age specific Rls Typically, an RI is defined by the range between two centile values, centred around the median on the probability scale. Standard limits commonly used are the 2.5th and 97.5th percentiles, which correspond to a 95% RI. An extensive description of the methods used to define the age specific RIs is provided in the online supplementary text 2. In summary, to define RIs, a parametric regression method based on fractional polynomials was used 16. Age specific 2.5th, 5th, 10th, 25th, 50th, 75th 90th and 97.5th percentile curves were calculated. RIs were derived, adjusted for height, gender and weight, where deemed necessary, that is,

Mobility Study

when this adjustment improved substantially the fit of the data; otherwise, the models for the development of the RIs were kept as simple as possible, which also contributes to their more feasible application in practice. Stata SE V.12 was used.

RESULTS A total of 393 participants were included in the study. All the age, height and gender categories were complete, except for the category of women, aged 60-69 years and >180 cm, due to difficulties in finding eligible volunteers. The descriptive statistics of all measures are presented in table 1. In general, the values are reasonably spread, which reflects the variation in spinal mobility that occurs among normal individuals. Surprisingly, no subject had a BASMI of 0. The lowest BASMI-modified value found in an individual was 0.18; the mean value of all normal individuals was 1.83. Of the five component scores of the BASMImodified, the 10ST was the one with the highest mean (3.4, SD 1.4), and thus contributing most to the impossibility of achieving a BASMI of 0. This was followed by cervical rotation (1.9, SD 1.2). The component score with the lowest average was the TTW (1.1, SD 0.4). Effect of age, gender and height on spinal mobility

Details of each of the spinal mobility measures stratified for age categories, gender or height categories, and corresponding comparisons across categories, are outlined in online supplementary tables and figures S2, S3, S4A and S4B, respectively. The relationship between these factors and each of the spinal mobility measures is summarised in table 2. A significant decrease in all spinal mobility measures was found with increasing age. For example an increase of 10 years was associated with a decrease of 1.5 cm in ASAS-LSF. Age was therefore included in all multivariable models. The last column of table 2 presents our (subjective) judgment on the clinical relevance of the effect based on the magnitude of the effect size. The age effect was considered clinically relevant for almost all the measures. Height was associated with TTW, ASAS-LSF, CE, IMD and IHR, with a higher height being associated with better mobility. For example, every increase in height of 10 cm resulted in an increase in IMD of 4.3 cm (men) or 5.6 cm (women). The effect of height on TTW and ASASLSF was not considered clinically relevant. Gender was associated with some of the spinal mobility measures. Female gender was associated with lower values of CE and cervical rotation sitting, but higher values of IHR.

Chapter 3

11.2 (1.2) 0.2 (1.0) 19.2 (3.7) 33.6 (6.0) 6.9 (2.2) 74 (11) 75 (9) 112 (14) 48 (10) 5.0 (1.0) 6.3 (1.2) 1.47 (0.73) 1.83 (0.74)

Tragus-to-wall, cm

Occiput-to-wall, cm

ASAS lateral spinal flexion, cm

INSPIRE lateral spinal flexion, cm

Chest expansion, cm

Cervical rotation (sitting), º

Cervical rotation (lying), º

Intermalleolar distance, cm

Internal hip rotation, cm

10cm Schober test, cm

15cm Schober test, cm

BASMI-original, 0-10†

BASMI-modified, 0-10‡

1.75 – 1.91

1.39 – 1.55

6.2 – 6.4

4.9 – 5.1

47 – 49

110 – 113

74 - 76

72 – 75

6.7 – 7.1

33.0 – 34.2

18.9 – 19.6

0.1 – 0.3

11.1 – 11.3

95% CI

0.18 – 4.67

0.28 – 4.29

2.2 – 10.5

2.2 – 8.1

20 – 75

61 – 154

36 – 97

38 – 102

1.0 – 13.0

12.0 – 49.2

8.0 – 29.8

0.0 – 8.0

8.9 – 19.5

Min. - max

1.72

1.33

6.2

5.0

111

6.9

34.0

19.3

0.0

11.0

Median

0.97

0.96

1.5

1.2

3.5

7.5

4.9

0.0

1.1

Interquartile range

N = 393; for cervical rotation sitting position N = 323 † BASMI-original: calculated with cervical rotation in lying position and 15cm Schober test ‡BASMI-modified: calculated with cervical rotation in sitting position and 10cm Schober test SD – standard deviation, CI – confidence interval, BASMI – Bath Ankylosing Spondylitis Metrology Index, ASAS – Assessment in Spondyloarthritis international Society

Mean (SD)

Measure

Table 1 – Characteristics of the spinal measurements in normal individuals (aged 20-69 years, men and women)

Mobility Study

Chapter 3

Cervical rotation, lying (degrees)

Cervical rotation, sitting (degrees)

Chest expansion (cm)

INSPIRE Lateral spinal flexion (cm)

ASAS Lateral spinal flexion (cm)

Occiput-to-wall (cm)

Tragus-to-wall (cm)

Measure

-0.35 (-0.41; -0.30) 2.11 (0.38; 3.83) 0.02 (-0.05; 0.09)

Height (cm)

0.12 (0.03; 0.21)

Height (cm)

Gender (female vs male)

-4.13 (-6.43; -1.82)

Gender (female vs male)

Age (years)

-0.35 (-0.42; -0.27)

0.07 (0.06; 0.09)

Height (cm)

Age (years)

-1.16 (-1.58; -0.73)

Gender (female vs male)

0.03 (-0.02; 0.08)

Height (cm) -0.03 (-0.05; -0.01)

-0.92 (-2.10; 0.26)

Age (years)

-0.24 (-027; -0.20)

Gender (female vs male)

0.03 (0.00; 0.06)

Height (cm)

Age (years)

-0.50 (-1.23; 0.23)

Gender (female vs male)

0.01 (0.01; 0.02) -0.15 (-0.17; -0.13)

Height (cm)

Age (years)

-0.24 (-0.43; -0.04)

Gender (female vs male)

0.04 (0.03; 0.05)

Height (cm) 0.02 (0.01; 0.03)

-0.7 (-1.0; -0.5)

Age (years)

0.02 (0.01; 0.03)

Age (years)

Gender (female vs male)

Univariable linear regression β (95% CI)

Table 2 – Effect of age, gender and height on spinal mobility measures

-0.34 (-0.39; -0.29)

-3.83 (-5.89; -1.77)

-0.34 (-0.42; -0.27)

Men: 0.11 (0.08; 0.14) Women: 0.11 (0.09; 0.14)

Men: -0.02 (-0.04; -0.01) Women: --

-0.24 (-027; -0.20)

0.03 (0.00; 0.07)

-0.15 (-0.17; -0.13)

0.02 (0.01; 0.02)

0.02 (0.01; 0.03)

Multivariable linear regression β (95% CI)§

√

Clinically relevant¥

Mobility Study

0.21 (0.05; 0.38) -0.01 (-0.02; -0.00)

Gender (female vs male)

Height (cm)

-0.01 (-0.01; -0.00)

Height (cm) 0.03 (0.03; 0.04)

0.08 (-0.08; 0.24)

Gender (female vs male)

Age (years)

0.03 (0.03; 0.04)

0.00 (-0.01; 0.01)

Height (cm)

Age (years)

-0.33 (-0.57; -0.10)

Gender (female vs male)

0.00 (-0.01; 0.01) -0.02 (-0.03; -0.01)

Height (cm)

Age (years)

-0.01 (-0.02; -0.00) -0.33 (-0.53; -0.13)

Age (years)

0.12 (0.05; 0.20)

Height (cm)

Gender (female vs male)

7.80 (6.04; 9.56)

Gender (female vs male)

0.44 (0.33; 0.55)

Height (cm) -0.14 (-0.21; -0.08)

-1.05 (-3.92; 1.83)

Gender (female vs male)

Age (years)

-0.46 (-0.42; -0.27)

Age (years)

Men: 0.03 (0.02; 0.04) Women: 0.03 (0.03; 0.04)

Men: -0.02 (-0.03; -0.01) Women: -0.02 (-0.03; -0.01)

-0.01 (-0.02; -0.01)

Men: 0.50 (0.37; 0.63) Women: 0.34 (0.21; 0.46)

Men: -Women: -0.17 (-0.25; -0.09)

Men: 0.43 (0.28; 0.58) Women: 0.56 (0.41; 0.72)

Men: -0.36 (-0.48; -0.24) Women: -0.52 (-0.64; -0.40)

√

CI – confidence interval; BASMI – Bath Ankylosing Spondylitis Metrology Index, ASAS – Assessment in Spondyloarthritis international Society § Analyses were adjusted for weight, where needed (tragus-to-wall distance, occiput-to-wall distance, ASAS lateral spinal flexion, chest expansion, cervical rotation (lying), internal hip rotation, 10 cm Schober test, 15 cm Schober test, BASMI-original, BASMI-modified) † BASMI-original: calculated with cervical rotation in lying position and 15cm Schober test ‡BASMI-modified: calculated with cervical rotation in sitting position and 10cm Schober test ¥ Clinical relevance of the effect of each factor on each of the measurements according to the authors’ clinical judgement

BASMI-modified (0-10)‡

BASMI-original (0-10)†

15 cm Schober test (cm)

10 cm Schober test (cm)

Internal hip rotation (cm)

Intermalleolar distance (cm)

Figure 1 – Reference intervals and Percentile Curves for Lateral Spinal Flexion, BASMI and Intermalleolar Distance, all measured as recommended by ASAS. A – Lateral Spinal Flexion; B – BASMI; C – Intermalleolar Distance (measure dependent on height, and here curves are presented for the mean height, height = 175cm). Reference Intervals for other spinal mobility measures can be found in the Online Supplementary Figures S5 and S6 (in the latter, are the different curves for different heights for measures dependent on height)

Chapter 3

Age (and height) specific Rls

For the development of RIs, an equation best fitting the data was derived for each of the

measures (see online supplementary table S5). From those equations, RIs can be computed for any age, and examples were chosen for ages at 10 year intervals (25, 35, 45, 55 and 65 years old). For some measures, namely CE, IMD and IHR, age and height specific percentiles were derived, as these fitted the data best because height influenced those measures. Gender and weight were not taken into account when deriving the RIs, as, after adjustment for age (and in some cases for height), they did not substantially contribute to

improve the fit of the models. Age (and height) specific percentiles for each of the measures are presented in table 3. For instance, for a 25-year old person the 2.5th percentile for ASASLSF was 16.2 cm, which means that 2.5% of the participants had a result below this value. The corresponding 95% RI was 16.2- 28.0 cm. The corresponding percentile curves were derived and are presented in figure 1, and in more detail and in online supplementary figure S5. For the measures for which an adjustment for height was made, RIs were derived for the mean height (175 cm) (table 3); further calculations for other heights (160 cm and 185 cm) are presented in the online supplementary table and figure S6. A potential non-linear relationship between age and spinal mobility measures was taken into account and accommodated when deriving the age-specific percentiles. For measures whose distribution deviated from normal (ie, Gaussian), that is, cervical rotation lying, 10ST, BASMI-original, BASMI-modified, percentile curves were derived through an extended model accommodating the skewness of the data. The relationship between the spinal mobility measure and age was non-linear for the cases of cervical rotation lying, IMD, BASMI-original and BASMI-modified. In these cases, the impairment in mobility with increasing age is even clearer, especially above the age of 50. All 10 year age group graphs for all measures can be found at http://www.asas-group.org.

DISCUSSION In the present study, age (and in some cases height) specific RIs and also different percentile curves have been derived in normal individuals for the different spinal mobility measures currently used in patients with axSpA. This is, to our knowledge, the first study defining RIs for these measures, in which two trained assessors have performed all the measures in a systematic approach and including all ASAS recommended measures. A wide range of values was found for all spinal mobility measures. This emphasises the variation of mobility across participants, even among normal individuals from a sample in which any subject in whom an â&#x20AC;&#x153;abnormalâ&#x20AC;? spinal mobility could be expected was excluded

Mobility Study

Chapter 3

Cervical rotation, lying (degrees)

Cervical rotation, sitting (degrees)

Chest expansion adjusted for height* (cm)

INSPIRE Lateral spinal flexion (cm)

ASAS Lateral spinal flexion (cm)

Spinal mobility measure

25 35 45 55 65 25 35 45 55 65 25 35 45 55 65 25 35 45 55 65 25 35 45 55 65

Age 16.2 14.7 13.2 11.6 10.1 29.2 26.4 23.6 20.8 18.0 3.6 3.3 3.0 2.7 2.4 62 59 55 52 48 69 66 61 56 48

2.5th

Table 3 â&#x20AC;&#x201C; Reference intervals for spinal mobility measures

17.2 15.6 14.1 12.6 11.1 30.6 27.9 25.1 22.4 19.7 4.2 3.9 3.6 3.3 3.1 65 62 58 55 51 71 68 64 59 51

5th 18.2 16.7 15.2 13.7 12.1 32.2 29.6 27.0 24.3 21.7 4.9 4.6 4.3 4.1 3.8 69 65 62 58 55 73 71 67 62 55

10th 20.1 18.5 17.0 15.5 14.0 35.0 32.5 30.0 27.5 25.0 6.1 5.8 5.5 5.2 5.0 74 71 67 64 60 77 75 72 67 60

22.1 20.6 19.1 17.5 16.0 38.0 35.7 33.4 31.0 28.7 7.4 7.1 6.9 6.6 6.3 81 77 74 70 67 81 79 77 73 66

Percentiles 25th 50th 24.1 22.6 21.1 19.6 18.0 41.1 38.9 36.7 34.6 32.4 8.7 8.5 8.2 7.9 7.6 87 84 80 77 73 85 84 81 78 72

75th 26.0 24.4 22.9 21.4 19.9 43.8 41.8 39.8 37.8 35.8 9.9 9.6 9.4 9.1 8.8 93 89 86 82 79 88 87 85 82 77

90th

28.0 26.5 25.0 23.4 21.9 46.8 45.0 43.2 41.3 39.5 11.3 11.0 10.7 10.4 10.1 99 96 92 89 85 91 91 89 86 82

97.5th

Mobility Study

25 35 45 55 65 25 35 45 55 65 25 35 45 55 65 25 35 45 55 65 25 35 45 55 65 25 35 45 55 65

96 94 90 85 77 32 31 29 28 26 3.4 3.3 3.2 3.1 2.9 4.5 4.2 4.0 3.8 3.6 0.34 0.41 0.55 0.76 1.06 0.66 0.65 0.74 0.94 1.30

99 97 94 89 81 35 34 32 31 29 3.7 3.6 3.5 3.3 3.2 4.8 4.6 4.4 4.2 4.0 0.41 0.49 0.64 0.87 1.21 0.75 0.76 0.86 1.09 1.46

104 102 98 93 85 38 37 36 34 33 4.0 3.9 3.8 3.6 3.5 5.2 5.0 4.8 4.6 4.4 0.50 0.59 0.76 1.02 1.39 0.86 0.89 1.01 1.26 1.66

111 109 105 100 92 44 43 41 40 38 4.6 4.4 4.3 4.2 4.1 5.9 5.7 5.5 5.3 5.1 0.67 0.78 0.97 1.28 1.72 1.05 1.12 1.29 1.58 2.02

119 117 113 108 100 50 49 47 46 44 5.2 5.1 5.0 4.8 4.7 6.7 6.5 6.3 6.1 5.9 0.88 1.02 1.25 1.62 2.16 1.30 1.42 1.64 1.98 2.47

126 125 121 116 108 57 55 54 52 51 5.9 5.8 5.6 5.5 5.4 7.4 7.2 7.0 6.8 6.6 1.14 1.31 1.59 2.04 2.68 1.58 1.76 2.04 2.44 2.99

134 132 128 123 115 62 61 59 58 56 6.6 6.4 6.3 6.2 6.1 8.1 7.9 7.7 7.5 7.3 1.44 1.63 1.97 2.50 3.27 1.87 2.12 2.46 2.92 3.54

141 140 136 131 123 68 67 66 64 63 7.4 7.2 7.1 7.0 6.9 8.9 8.7 8.5 8.3 8.1 1.87 2.11 2.54 3.20 4.15 2.27 2.60 3.03 3.58 4.28

ASAS â&#x20AC;&#x201C; Assessment of Spondyloarthritis international Society; BASMI â&#x20AC;&#x201C; Bath Ankylosing Spondylitis Metrology Index * Adjusted for height means that the equation for the derivation of the percentiles included height. For these percentiles, the mean height (175cm) was used. Percentiles for other height values are presented in the Online Supplementary Table S6

BASMI-modified (0-10)

BASMI-original (0-10)

15 cm Schober test (cm)

10 cm Schober test (cm)

Internal hip rotation adjusted for height* (cm)

Intermalleolar distance adjusted for height* (cm)

following the rigorous eligibility criteria. Surprisingly, no subject had a BASMI value of zero, and being a sample composed of normal individuals, it seems that it is technically (almost) impossible to have a BASMI score of zero. An important age effect was found on all spinal mobility measures. This is in line with previous studies

. Nevertheless, the effect is so striking that we consider it deserves to

6-9

be emphasised. When following patients with axSpA throughout time, this age effect on spinal mobility should be taken into account as it might, to some extent, explain a potential impairment in mobility developing, which should likely not be completely attributed to the disease. Comparison between the results of our study and of previous studies performed in normal individuals is hampered by several factors, such as different recruitment strategies, inclusion criteria and mobility measurement techniques. Furthermore, no other study developed RIs, so we can only compare sample or subgroup mean values. Our values for 10ST lie in between those previously reported

. Moll et al 6, described values of 15ST across age

8,9

categories very similar to ours. In another study, CE values were slightly lower than ours 7. A more recent study among 20-29-year-old Turkish men (n = 1982) reported slightly lower values for CE and higher values for 10ST and 15ST 22. Our study seems to be unique in the recruitment stratified for age, height and gender, in order to obtain a sample in which these factors are balanced. Despite the fact that no reference values had been previously defined, a 5 cm cut-off for the 10ST is frequently used in clinical practice. Several of the participants included in our study had a 10ST below this cut-off, and for some age categories, even the average (and median) value was below this cut-off. This finding should be integrated into clinical practice, so that a value below 5 cm is not necessarily considered abnormal. Age specific RIs and percentiles were derived for each of the spinal mobility measures for normal individuals. For some measures (ie, CE, IMD and IHR), age and height specific RIs and percentiles were derived. The derivation of the RIs confirmed the previous findings of the regression analysis (table 2), and only for these measures was it necessary to adjust the RI for height. For measures dependent on height, the mean height has been used to derive the RI presented in this paper, but in case of a patient with an â&#x20AC;&#x2DC;extremeâ&#x20AC;&#x2122; height, appropriate mobility curves (see online supplementary appendices) can be used. Despite the fact that gender and weight were associated with some of the mobility measures, they did not seem to have an additional contributory effect to explaining spinal mobility values, compared to age (and in some cases also height), and were therefore not taken into account when deriving the RIs. The development of RIs for a given measure is common practice in other areas

Chapter 3

. These

RIs may guide clinicians when assessing the mobility of patients with axSpA. These measures are not diagnostic tests (ie, to diagnose a patient with axSpA on the basis of an “abnormal mobility”). The RIs and percentile curves can be of use in the assessment of patients with axSpA, in a similar approach to use of the reference values for dual-energy x-ray absorptiometry bone mineral density assessments as recommended by the WHO or the use of growth curves for monitoring children’s growth

23,24

. In practice, it may be of

value to assess a patient rigorously when he starts deviating from his ‘personal curve’. This hypothesis needs to be further investigated, but looks to be a promising use of these mobility

curves. The RIs and percentiles may also serve as cut-off levels for “normal” versus “abnormal”. For that purpose, the lowest percentiles are the most important, as they are the ones that may define an important impairment in spinal mobility; therefore more percentiles (ie, 2.5th, 5th, 10th, 25th) are presented in the lower range. The cut-off used for the definition of “abnormal” depends on how specific we want to be. If we want to be specific we can use the lowest cut-off, that is, the 2.5th percentile, which is also the lowest border of the 95% RI, and label only those patients with definitely abnormal spinal mobility. If we want to be sensitive, and identify all the cases with “potentially abnormal spinal mobility”, then we will chose a higher cut-off (eg, 10th or 25th percentile). The ultimate definition of such a cut-off should be based on evidence and should have some prognostic value, which needs to be assessed in further research, having these proposed percentiles as a reference. Some limitations of the present study should be addressed. The sample size is relatively small. Nevertheless, the stratified recruitment gave the study a higher power. Although it is not certain to what extent these results are generalisable to other populations, we do not expect large differences. We accepted any volunteer, but the eligibility criteria -which focused on factors eventually affecting spinal mobility- were strictly assessed to ensure we would have a population composed of individuals with normal spinal mobility. A factor potentially influencing spinal mobility that was not taken into account was each subject’s level of physical activity. However, it is a difficult factor to measure, especially by self-report. Furthermore, RIs are usually derived as being only age-specific and not taking many other factors into account, mainly because of feasibility purposes of their application. The study sample was restricted to Caucasians, which may limit the generalisability of the findings. Nevertheless, this was preferred, in order to control for another factor eventually influencing spinal mobility. Strengths of the study are the solid methodology used for the statistical analysis (analogous to the methodology used for the development of the growth curves by the WHO

); having a homogeneous sample in terms of factors influencing mobility

(especially age and height, but also gender) and also a population of normal individuals; having two well-trained observers assessing a quarter of the participants with good reliability; Mobility Study

and including all ASAS recommended mobility measures. In summary, age (and height) specific RIs and percentiles were derived for each of the spinal mobility measures for normal individuals. These RIs may help clinicians when assessing the mobility of patients with axSpA.

SUPPLEMENTARY DATA Supplementary data are published on the website of the Annals of the Rheumatic Diseases.

Chapter 3

REFERENCES 1.

Carette S, Graham D, Little H, et al. The natural disease course of ankylosing spondylitis. Arthritis Rheum 1983;26:186-90.

Amor B, Santos RS, Nahal R, et al. Predictive factors for the longterm outcome of spondyloarthropathies. J Rheumatol 1994;21:1883-7.

Laurent MR, Buchanan WW, Bellamy N. Methods of assessment used in ankylosing spondylitis clinical trials: a review. Br J Rheumatol 1991;30:326-9.

Macrae IF, Wright V. Measurement of back movement. Ann Rheum Dis 1969;28:584-9.

Moll JM, Wright V. Normal range of spinal mobility. An objective clinical study. Ann Rheum Dis 1971;30:381-6.

Moll JM, Wright V. An objective clinical study of chest expansion. Ann Rheum Dis 1972;31:1-8.

Einkauf DK, Gohdes ML, Jensen GM, et al. Changes in spinal mobility with increasing age in women. Phys Ther 1987;67:370-5.

Fitzgerald GK, Wynveen KJ, Rheault W, et al. Objective assessment with establishment of normal values for lumbar spinal range of motion. Phys Ther 1983;63:1776-81.

10.

Loebl WY. Measurement of spinal posture and range of spinal movement. Ann Phys Med 1967;9:103-10.

11.

Sturrock RD, Wojtulewski JA, Hart FD. Spondylometry in a normal population and in ankylosing spondylitis. Rheumatol Rehabil 1973;12:135-42.

12.

Moll JM, Liyanage SP, Wright V. An objective clinical method to measure spinal extension. Rheumatol Phys Med 1972;11:293-312.

13.

Moll JM, Liyanage SP, Wright V. An objective clinical method to measure lateral spinal flexion. Rheumatol Phys Med 1972;11:225-39.

14.

Jenkinson TR, Mallorie PA, Whitelock HC, et al. Defining spinal mobility in ankylosing spondylitis (AS). The Bath AS Metrology Index. J Rheumatol 1994;21:1694-8.

15.

Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68 Suppl 2:ii1-44.

16.

Royston P, Wright EM. A method for estimating age-specific reference intervals (â&#x20AC;&#x2DC;normal rangesâ&#x20AC;&#x2122;) based on fractional polynomials and exponential transformation. J R Statist Soc A 1998;161:79-101.

17.

Engelen L, Ferreira I, Stehouwer CD, et al. Reference intervals for common carotid intimamedia thickness measured with echotracking: relation with risk factors. European heart journal 2012.

18.

Beighton P, Horan F. Orthopaedic aspects of the Ehlers-Danlos syndrome. J Bone Joint Surg Br 1969;51:444-53.

19.

Gladman DD, Inman RD, Cook RJ, et al. International spondyloarthritis interobserver reliability exercisethe INSPIRE study: II. Assessment of peripheral joints, enthesitis, and dactylitis. J Rheumatol 2007;34:1740-5.

20. Maksymowych WP, Mallon C, Richardson R, et al. Development and validation of the Edmonton Ankylosing Spondylitis Metrology Index. Arthritis Rheum 2006;55:575-82. 21. van der Heijde D, Landewe R, Feldtkeller E. Proposal of a linear definition of the Bath Ankylosing Spondylitis Metrology Index (BASMI) and comparison with the 2-step and 10-step definitions. Ann Rheum Dis 2008;67:489-93. 22. Cidem M, Karacan I, Uludag M. Normal range of spinal mobility for healthy young adult Turkish men. Rheumatology international 2012;32:2265-9. 23. Kanis JA. on behalf of the World Health Organization Scientific Group (2007). Assessment of osteoporosis at the primary health-care level. Technical Report. World Health Organization Collaborating Centre for Metabolic Bone Diseases, University of Sheffield, UK. 2007: Printed by the University of Sheffield. 24. Looker AC, Wahner HW, Dunn WL, et al. Updated data on proximal femur bone mineral levels of US adults. Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA 1998;8:468-89. 25. Chan JW, Castellanos A. Infliximab and anterior optic neuropathy: case report and review of the literature. Graefes Arch Clin Exp Ophthalmol 2010;248:283-7.

Mobility Study

Neutral lateral fingertip-to-floor distance can be derived from height Sofia Ramiro, Astrid van Tubergen, Carmen Stolwijk, Désirée van der Heijde, Robert Landewé

Ann Rheum Dis 2014;73:1748-9.

Spinal mobility is one of the core outcome measures used in axial spondyloarthritis as recommended by the Assessment of SpondyloArthritis international Society (ASAS) 1,2. Among the commonly used spinal mobility measures, lateral spinal flexion (LSF) is considered to be the most sensitive to change and has therefore been included as one of the domains in the ASAS 5/6 response criteria 3,4. LSF is measured as the difference between two marks placed on the thigh, one in a neutral standing position and the other in maximum lateral flexion 2. Alternatively, LSF can be calculated as the difference in the lateral distances between middle fingertip-to-floor (FTF) in the neutral position (neutral FTF) and middle FTF in maximum latero-flexion (maximum flexion FTF) 2. In this second LSF measurement, it may sometimes erroneously and accidentally happen that only the maximum flexion FTF is recorded. This prohibits calculation of the LSF unless neutral FTF could be obtained by another method. Because neutral FTF is a static measure, we hypothesised that it would be highly correlated with height and it would be possible to derive its value from height. Data from the MOBILITY-study were used. Details of this study have been published elsewhere 5. Briefly, this was a cross-sectional study, including healthy volunteers, with recruitment stratified by age, gender and height to assure a balanced distribution. Spinal mobility measures as recommended by the ASAS were performed. These included LSF, for which both components (neutral FTF and maximum flexion FTF) were registered separately. These were measured with the participant standing with heels and back against a wall without bending the knees or trunk. The correlation between neutral FTF and height was established using Pearsonâ&#x20AC;&#x2122;s correlation. Univariable linear regression was performed with neutral FTF as the outcome and height as the independent variable and a regression equation was obtained in order to derive neutral FTF from height. Interactions with age and gender were tested. The difference between the original value for neutral FTF and the derived one was investigated. A total of 393 participants were included. The correlation coefficient between neutral FTF and height was 0.92. Neutral FTF could be derived from height through the following equation: neutral FTF = -4.379 + 0.415 * height (figure 1), which showed a very good fit of the data (R2 = 0.84). The average original neutral FTF was 68.4 cm (SD 5.54). The difference between the original neutral FTF and the calculated value was on average only 0.0000052cm (SD 2.22). The relationship between neutral FTF and height was not influenced by age or gender. We hereby show that neutral FTF can be reliably obtained from height using a simple equation. This allows the calculation of LSF in a situation in which FTF has only been registered in maximum flexion as neutral FTF can be obtained from height. Deriving one anthropometric

Chapter 4

measure from another has frequently been performed for other measures, as, for example, to estimate height from the arm span or from the upper limb anthropometry

. Establishing

6,7

a regression equation to derive one measure from another gives further insight into the relationship between two variables and allows the use of one of the values in replacement of the other in case of missing values. The effect of loss of height on the proposed derivation of neutral FTF from height should be further tested but is expected to have only a significant impact with major reduction in height. This proposed derivation of neutral FTF from height should be tested in another sample. In conclusion, we hereby propose an equation to derive neutral FTF from height, enabling reliable imputation of cases with missing values and, in turn, enabling the calculation of LSF.

Figure 1 - Fingertip-to-floor distance on a neutral standing position over height with corresponding regression

Correction for lateral spinal flexion

REFERENCES 1.

van der Heijde D, Calin A, Dougados M, et al. Selection of instruments in the core set for DCART, SMARD, physical therapy, and clinical record keeping in ankylosing spondylitis. Progress report of the ASAS Working Group. Assessments in Ankylosing Spondylitis. J Rheumatol 1999;26:951-4. Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68 Suppl 2:ii1-44. van der Heijde D, Dijkmans B, Geusens P, et al. Efficacy and safety of infliximab in patients with ankylosing spondylitis: results of a randomized, placebo-controlled trial (ASSERT). Arthritis Rheum 2005;52:582-91.

Chapter 4

Brandt J, Listing J, Sieper J, et al. Development and preselection of criteria for short term improvement after anti-TNF alpha treatment in ankylosing spondylitis. Ann Rheum Dis 2004;63:1438-44.

Ramiro S, van Tubergen A, Stolwijk C, et al. Reference intervals of spinal mobility measures in normal individuals: the mobility study. Ann Rheum Dis 2014.

Mohanty SP, Babu SS, Nair NS. The use of arm span as a predictor of height: A study of South Indian women. Journal of orthopaedic surgery 2001;9:19-23.

Akhlaghi M, Hajibeygi M, Zamani N, et al. Estimation of stature from upper limb anthropometry in Iranian population. Journal of forensic and legal medicine 2012;19:280-4.

Correction for lateral spinal flexion

Hierarchy of impairment of spinal mobility measures in ankylosing spondylitis: 12 year data from the OASIS cohort Sofia Ramiro, Robert Landewé, Désirée van der Heijde, Carmen Stolwijk, Maxime Dougados, Filip van den Bosch, Astrid van Tubergen

Submitted

ABSTRACT Objectives

To investigate 1) which spinal mobility measures (SMMs) are most frequently impaired in patients with AS, 2)whether a hierarchical order of impairment could be established and 3) whether assessing fewer measures sufficiently captures impairment in spinal mobility. Methods

Patients from the Outcome in AS International Study (OASIS) were followed-up for 12 years, with regular spinal mobility measurements. SMMs were considered impaired when falling below predefined cut-offs, derived from normal individuals. The proportion of patients in whom every SMM was impaired was calculated using the baseline observation. In patients with ≥1 impaired SMM, we investigated how often impairment in spinal mobility would be missed if only a fixed number of SMMs was assessed. Analyses were repeated using all 12 year observations. Results

A total of 216 patients were included (70% males). Lateral spinal flexion (LSF) was the most frequently impaired measure, followed by modified Schober(mSchober), tragus-to-wall, cervical rotation, intermalleolar distance and chest expansion, respectively. This hierarchy was strikingly consistent over time, and independent of gender, symptom duration and presence of syndesmophytes. In patients with ≥1 impaired SMM, LSF was impaired most frequently (86%), followed by mSchober (58%). If only LSF was measured, 14% of patients with impairment in any SMM would be missed; if additionally mSchober was measured, 9% would be missed. Conclusion

LSF followed by mSchober are the most frequently impaired mobility measures in AS, reflecting

an earlier involvement of the lumbar spine, followed by involvement of the thoracic and cervical spine. In clinical practice LSF and mSchober suffice to screen impairment in spinal mobility. · · · · ·

Chapter 5

INTRODUCTION Impairment of spinal mobility is widely recognized as an important clinical sign and a hallmark of axial spondyloarthritis (axSpA). This is reflected by its inclusion in the modified New York Criteria for the classification of AS and in the core set of domains for the evaluation of patients with SpA in clinical practice and trials, as defined by the Assessment of SpondyloArthritis international Society (ASAS) 1. A study on the natural course of ankylosing spondylitis (AS) conducted in the pre-biological era described that in the long-term more than half of all patients will develop moderate to severe impairment of spinal mobility 2. Spinal mobility is one of the central outcomes included in observational and treatment studies in axSpA 3,4 and is a predictor of poor prognosis 3. The spinal mobility measures (SMMs) recommended by ASAS are: chest expansion,

modified Schoberâ&#x20AC;&#x2122;s test (mSchober), occiput-to-wall distance (OTW), cervical rotation, and either lateral spinal flexion (LSF) or Bath AS Metrology Index (BASMI) 5,6. The BASMI, in turn, includes LSF, mSchober, cervical rotation, tragus-to-wall distance (TTW) and intermalleolar distance (IMD) 5. LSF and BASMI have shown best discrimination between patients with- and without structural damage 7. A normal LSF is the best predictor of absence of radiographic damage, while an abnormal mSchober best predicts its presence. Furthermore, LSF and cervical rotation more consistently discriminated between active treatment and placebo, and between responders and non-responders in randomized controlled trials with tumor necrosis factor inhibitors

. However, to our knowledge it has never been investigated whether

8,9

specific SMMs are more frequently impaired than others in patients with AS compared to normal individuals, and whether there is a certain hierarchical order in the occurrence of impairment. Such an analysis was hampered by the lack of normal values for SMMs. We have recently described reference intervals and percentile curves for SMMs obtained in normal individuals 10. In the present study, we have used these percentile curves for SMM to perform a formal age-matched comparison with values obtained from patients with AS. The aim of this study was to investigate which SMMs are most frequently impaired in patients with AS, whether a hierarchical order in the occurrence of impairment could be established and whether this would be dependent on characteristics such as symptom duration. Furthermore, we have investigated whether assessing fewer measures could still capture sufficient information for detecting impairment of spinal mobility.

Impairment of spinal mobility

METHODS Study population

For this study data from the Outcome in Ankylosing Spondylitis International Study (OASIS)

were used. OASIS is a prevalence cohort that started in 1996 and included 217 consecutive patients with AS from the Netherlands, Belgium and France 11. Clinical data, including spinal mobility, were collected regularly during 12 years. In the first 2 years, spinal mobility was measured every 6 months, and afterwards every 2 years. All patients were included in the present study. The ethics committees of all participating hospitals have approved the study. All patients have provided written informed consent. Spinal mobility measures

Six SMMs were analysed in this study: chest expansion, mSchober, cervical rotation, LSF, TTW and IMD. Measurements were performed according to the ASAS recommendations and the reference values were obtained using these same assessments 6,10. The best of two tries for each measurement was recorded. All measurements were recorded in centimeters, except for cervical rotation that was recorded in degrees. All measurements were rounded to decimals, except for IMD and cervical rotation, which were rounded to units. Additionally, the BASMI5 (range 0 to 10, with a higher score representing a worse mobility) was computed according to the BASMI linear formula 12. In the first 2 years, LSF has erroneously been recorded as the distance from middle fingertipto-floor on maximum flexion for Dutch patients. This means that the fingertip-to-floor distance in a neutral position was missing which hampers the calculation of the LSF. To replace this missing value and enable the computation of LSF, we derived the neutral fingertip-to-floor distance from height according to an equation proposed for this purpose 13. Impairment in spinal mobility

Percentile curves (2.5th to 97.5th percentiles) were obtained for each of the SMMs from our previous study in normal individuals, the MOBILITY study

. Figure 1 illustrates such

percentile curves for LSF as an example. These curves represent age-specific spinal mobility in normal individuals and its evolution by increasing age. The 2.5th and 97.5th percentile curves obtained in normal individuals correspond with Z-scores of -1.96 and +1.96

. In

patients with AS impaired mobility is expected in comparison with normal individuals. In order to better describe this level of impairment, we have introduced 4 additional parallel curves to the mobility curves, corresponding with Z-scores of -2.50 (a’); -3,00 (b’); -3.50 (c’) and -4,00 (d’) respectively (figure 1). For every observation in OASIS a SMM was defined as impaired if its value was below a certain cut-off (either the 2.5th percentile or one of the 4 help-lines). 50

Chapter 5

OTW was not applied in this study, because percentile curves for this measure could not be derived. The reference value for OTW is zero and the vast majority of normal individuals indeed have an OTW of zero 10. We could have applied a value of zero here, but this value would be a static figure and would not change by increasing age or with the different percentile curves used, as opposed to what happens with the other measures. This would make the evaluation of the hierarchy for the impairment of the SMM across the different cut-offs complex or even impossible. We therefore used the TTW, which is another measure for cervical kyphosis. However, for TTW, no percentile curves had been derived in the MOBILITY study, with the rationale that this measure is strongly dependent on the conformation of the head and would therefore not be so relevant to derive reference intervals. For the purpose of the present

Figure 1 - Lateral spinal flexion in function of age and with the percentile curves and help-lines derived from measurements in normal individuals10 . The solid lines represent the percentile curves derived from normal individuals (percentiles indicated). Dashed lines (aâ&#x20AC;&#x2122; to dâ&#x20AC;&#x2122;) are help-lines, parallel to the percentile curves, derived to accommodate impaired values of patients with ankylosing spondylitis. The dots represent the real measurements of lateral spinal flexion of patients with ankylosing spondylitis (A) at baseline or (B) in all 12 year observations.

Impairment of spinal mobility

study, percentile curves for TTW were derived following the methodology described for the MOBILITY study (for more details see Online Supplementary Text 1 and Figure S1) 10. Subgroup analysis

Impairment of spinal mobility was also investigated in subgroups of patients, taking into account variables that may influence spinal mobility. Subgroups were created based on gender, symptom duration (above and below the median (<18 years vs ≥18 years), as well as in tertiles (<14 years, 14-25.29 years and ≥25.3 years) and the presence of baseline syndesmophytes (0 vs ≥1 and <5 vs ≥5). The scores on syndesmophytes from two readers (separate scores from each of the readers) were used 14. Statistical analysis

We conducted two types of analyses. First, for each of the SMMs, the proportion of patients in which the SMM was impaired according to each of the cut-offs was calculated, using the baseline observation. For each of the cut-offs, measures were ranked according to the proportion of patients with impairment in each of the SMMs. Subsequently, we conducted a similar analysis, but now taking all observations over the 12 years per patient into account. Analyses were repeated in all the subgroups. The main analysis was conducted with patients (respectively observations) in whom assessment of SMM was complete, i.e. all SMMs were assessed. Sensitivity analyses were conducted with patients (respectively observations) in whom any SMM was available. For the second analysis, the number of patients with at least one impaired SMM was calculated, using the baseline observation. Each of the SMMs was defined as impaired if the value was below help-line a’ (first cut-off below normal individuals). Among the patients

with at least one impaired SMM, the proportion of patients with each of the SMMs impaired was calculated. Furthermore, we investigated in how many cases impairment in spinal mobility (defined as impairment in at least one SMM) would be missed if only a fixed number of SMMs was assessed. Subsequently, we conducted the same analysis, but now taking all observations over the 12 years per patient into account. For this second analysis, only patients with complete assessments of SMMs were included. Analyses were done using Stata SE version 12 (Statacorp, College Station, TX, USA).

RESULTS A total of 216 patients with the following baseline characteristics were included: 71% males, mean (SD) age of 44 (13) years, mean symptom duration 21 (12) years, 85% HLA-B27 positive, 68% treated with NSAIDs and none with biological agents. Patients were on average followedup for 7.7 (4.1) years with a mean of 7.2 (2.1) observations per patient with assessment of 52

Chapter 5

SMMs. At baseline, the mean LSF was 9.6(5.8) cm, mSchober 2.8 (1.4) cm, TTW 14.3 (4.8) cm, cervical rotation 65 (23) degrees, IMD 105 (22) cm, chest expansion 4.7 (2.2) cm, and the BASMI 3.8 (1.6). Hierarchy of impairment in spinal mobility

LSF was always the most frequently impaired measure, sequentially followed by mSchober, tragus-to-wall, cervical rotation, IMD and chest expansion (table 1). This order was strikingly similar for all cut-offs (with the only exception being the 2.5th percentile, for which chest expansion was slightly more frequently impaired than IMD). Using as cut-off the help-line aâ&#x20AC;&#x2122; (first cut-off below normal individuals), LSF was impaired in 68% of the patients, followed by mSchober that was impaired in 47% of the patients (table 1). Impairment of BASMI, also

defined according to help-line aâ&#x20AC;&#x2122;, occurred in 49% of the patients. The same hierarchy was

found in the analyses including all 12 year observations (table 1). Figure 1 (panel A at patientlevel, panel B at observation-level) shows the percentile curves and help-lines for LSF and the measured values for LSF (i.e. dots) below each cut-off.

A sensitivity analysis, conducted with patients (respectively observations) in whom any SMM was available, provided similar results (online supplementary table S1). Also, after stratification for symptom duration (median), gender or the presence of baseline syndesmophytes this hierarchy in general persisted, both at the patient level (tables 2 and online supplementary tables S2-S3) and at the observation level (example shown for symptom duration in online supplementary table S4). Stratification for symptom duration in tertiles also retrieved the same results, as well as for the presence of baseline syndesmophytes according to the second reader (data not shown). Impairment in spinal mobility captured by fewer measures

Any impairment in spinal mobility was present in 161 (79%) of the 203 patients and in 1,111 (78%) of 1,422 observations with complete assessment of SMMs. Among patients with at least one impaired SMM, LSF was impaired in 86%, followed by mSchober in 58%. One SMM was impaired in 30% of the patients; 2 SMMs were impaired in 33%; and 3 SMMs were impaired in 17% (table 3). If only LSF was measured, 14% of the patients with impairment in SMMs would be missed since in these patients LSF was normal but another SMM was impaired. If mSchober was measured in addition to LSF, only 9% of patients would be missed. When TTW was measured in addition to LSF and mSchober, only 4% of the patients with impairment in mobility would be missed. Analyses taking into account all 12 year observations yielded similar results (table 3).

Impairment of spinal mobility

Chapter 5

n (%)

622 (44)

439 (31)

320 (23)

386 (27)

Tragus-to-wall distance (cm)

Cervical rotation (degrees)

Intermalleolar distance (cm)

Chest expansion (cm)

4 (2)

22 (11)

38 (19)

139 (10)

207 (15)

349 (25)

490 (34)

696 (49)

919 (65)

32 (2)

152 (11)

274 (19)

342 (24)

571 (40)

804 (57)

0 (0)

12 (6)

28 (14)

63 (31)

4 (0)

88 (6)

215 (15)

453 (32)

661 (46)

OBSERVATION LEVEL (ALL OBSERVATIONS)

17 (8)

27 (13)

44 (22)

43 (21)

78 (38)

106 (52)

n (%)

Below Help-line c'

1 (0)

55 (4)

167 (12)

282 (20)

314 (22)

0 (0)

8 (4)

26 (13)

36 (18)

55 (27)

n (%)

Below Help-line d'

1,422

203

Total number of observations (N)

Legend: red: spinal mobility measure (SMM) most frequently impaired; orange: 2nd SMM most frequently impaired; yellow: 3rd SMM most frequently impaired; very light green: 4th SMM most frequently impaired; light green: 5th SMM most frequently impaired; dark green: SMM least frequently impaired § The first column refers to impairment defined as a value below the 2.5th percentile in normal individuals. The remaining columns refer to cut-offs defined on the basis of help-lines, parallel to the percentile curves, derived to accommodate impaired values of patients with ankylosing spondylitis (see Figure 1). Impairment in spinal mobility is presented both at the patient level (using baseline observation) and observation level (all 12 year observations). *Not possible to derive these values due to mathematical characteristics of the equation for the percentile curves for tragus-to-wall distance

845 (59)

54 (27)

Chest expansion (cm)

Modified Schober's (cm)

39 (19)

Intermalleolar distance (cm)

1,035 (73)

55 (27)

Cervical rotation (degrees)

Lateral Spinal Flexion (cm)

85 (42)

62 (31)

94 (46)

111 (55)

Modified Schober's (cm)

Tragus-to-wall distance (cm)

119 (59)

PATIENT LEVEL (BASELINE ONLY)

Below Help-line b'

Below Help-line a'

138 (68)

Below 2.5th percentile n (%)

149 (73)

Lateral Spinal Flexion (cm)

Table 1 - Impairment of each of the spinal mobility measures in patients with AS defined according to age-specific cut-offs derived from normal subjects §

Impairment of spinal mobility

15 (16)

20 (21)

80 (83)

60 (63)

51 (53)

34 (35)

21 (22)

33 (34)

Intermalleolar distance (cm)

Chest expansion (cm)

Lateral Spinal Flexion (cm)

Modified Schober's (cm)

Tragus-to-wall distance (cm)

Cervical rotation (degrees)

Intermalleolar distance (cm)

Chest expansion (cm)

n (%)

Below Help-line c'

0 (0)

5 (5)

12 (13)

14 (15)

30 (31)

46 (48)

0 (0)

3 (3)

10 (10)

25 (26)

41 (43)

13 (14)

17 (18)

28 (29)

38 (40)

50 (52)

77 (80)

4 (4)

16 (17)

24 (25)

27 (28)

45 (47)

70 (73)

0 (0)

8 (8)

17 (18)

35 (36)

62 (65)

PATIENTS WITH LONGER SYMPTOM DURATION (≥18 YEARS)

4 (4)

9 (9)

14 (15)

21 (22)

41 (43)

58 (60)

Total number of observations (N)

96 96

24 (25) 15 (16)

96 96 96 96 96 96 96 96 96 96

* 10 (10) 2 (2) 0 (0) 28 (29) 18 (19) * 15 (16) 5 (5) 0 (0)

n (%)

Below Help-line d'

Legend: red: spinal mobility measure (SMM) most frequently impaired; orange: 2nd SMM most frequently impaired; yellow: 3rd SMM most frequently impaired; very light green: 4th SMM most frequently impaired; light green: 5th SMM most frequently impaired; dark green: SMM least frequently impaired § The first column refers to impairment defined as a value below the 2.5th percentile in normal individuals. The remaining columns refer to cut-offs defined on the basis of help-lines, parallel to the percentile curves, derived to accommodate impaired values of patients with ankylosing spondylitis (see Figure 1). Impairment in spinal mobility is presented at the patient level (for observation level, all 12 year observations, see Table S2) *Not possible to derive these values due to mathematical characteristics of the equation for the percentile curves for tragus-to-wall distance

29 (30)

18 (19)

Cervical rotation (degrees)

47 (49)

Modified Schober's (cm)

Tragus-to-wall distance (cm)

66 (69)

Below Help-line b'

Below Help-line a'

PATIENTS WITH SHORTER SYMPTOM DURATION (<18 YEARS)

Below 2.5th percentile n (%)

Lateral Spinal Flexion (cm)

Table 2 - Impairment of each of the spinal mobility measures in patients with AS compared to cut-offs defined according to age-specific cut-offs derived from normal subjects and stratified for symptom duration (median) §

Table 3 - Impairment of each of the spinal mobility measures in patients/observations with at least one impaired spinal mobility measure * Patient-level (baseline Observation-level assessment) n (%) n (%) (N = 1,111) (N = 161) Impairment in each of the spinal mobility measures Lateral spinal flexion 138 (86) 919 (83) mSchober 94 (58) 696 (63) Tragus-to-wall distance 62 (39) 490 (44) Cervical rotation 44 (27) 349 (31) Intermalleolar distance 27 (17) 207 (19) Chest expansion 17 (11) 139 (13) Number of spinal mobility measures impaired -1 49 (30) 282 (25) -2 53 (33) 355 (32) -3 27 (17) 203 (18) -4 15 (9) 168 (15) -5 16 (10) 91 (8) -6 1 (1) 12 (1) Impairment in spinal mobility captured by fewer measures Lateral spinal flexion 138 (86) 919 (83) Lateral spinal flexion + mSchober 147 (91) 1,008 (91) Lateral spinal flexion + mSchober + 154 (96) 1,067 (96) tragus-to-wall distance Lateral spinal flexion + mSchober + 160 (99) 1,097 (99) tragus-to-wall distance + cervical rotation Lateral spinal flexion + mSchober + 160 (99) 1,108 (100) tragus-to-wall distance + cervical rotation + intermalleolar distance mSchober: Modified Schoberâ&#x20AC;&#x2122;s * Each of the spinal mobility masures was defined as impaired if the measurement fell below help-line aâ&#x20AC;&#x2122; (first cut-off below normal individuals, see Figure 1). Among the patients (N = 161) or observations (N = 1,111) with at least one impaired spinal mobility measure, the proportion of patients (respectively of observations) with each of the spinal mobility measures impaired was calculated.

DISCUSSION

In the present study we have shown that there is a fixed order of involvement of the spine in AS: LSF and mSchober are the most frequently impaired mobility measures in AS, reflecting an earlier involvement of the lumbar spine in impairment in spinal mobility, followed by involvement of the thoracic and cervical spine. This fixed order of involvement of the spine persists across different subgroups of patients based on gender, disease duration or the presence of baseline syndesmophytes. To our knowledge the order of impairment in SMMs has never been investigated before. Therefore, we cannot compare our findings with others. Previous studies comparing spinal mobility in normal individuals and in patients with AS have focused on a single SMM and were performed in the 1970s15,16. Several factors, most importantly different measurement 56

Chapter 5

techniques, hamper a proper comparison with our results 15,16. The lack of studies comparing impairment in spinal mobility between patients with axSpA and normal individuals can partly be explained by the fact that percentile curves from normal individuals did not exist until recently. With the age-adjusted percentile curves from the MOBILITY study10, we now have better tools to compare spinal mobility between patients with axSpA and normal individuals and across studies. Interestingly, the SMMs most frequently impaired, LSF and mSchober, have also been identified as the ones best discriminating between patients with- and without spinal radiographic damage 7. This reinforces their importance when assessing spinal mobility. Furthermore, LSF is considered the SMM that is most sensitive to change 8. The relationship between LSF (as a single measure) and radiographic damage has been shown to be as

good as the relationship between BASMI (composite measure), and substantially better than the relationship between other SMMs and radiographic damage 7. Therefore, ASAS has recommended in the core set that the LSF can be used instead of the BASMI. Our findings endorse this idea, as we have found a higher proportion of observations with impaired LSF than with impaired BASMI, when comparing patients with normal individuals. Furthermore, LSF has been selected as one of the items of the ASAS 5/6 response criteria 17. From our study, it can be deduced that the detection of impairment of spinal mobility can be done by assessing only two SMMs: LSF and mSchober: Impairment of spinal mobility is only missed in 9% of the cases. We hereby recommend that these two measurements are minimally performed to be informed about impairment in spinal mobility in individual patients with AS. If these two are impaired, it is important to assess additional SMMs to be informed about the total level of spinal involvement. If LSF and mSchober are normal, it is unlikely that other SMMs will be found abnormal. This is particularly useful for daily clinical practice, where it may sometimes be unfeasible, due to time constraints, to perform all SMMs, even though these are recommended by ASAS 1. We were unable to find any study investigating the feasibility of spinal mobility assessment in clinical practice, but it may well be the case that spinal mobility measurement is frequently skipped for time-reasons. Measuring LSF and mSchober only will suffice to screen for impairment of spinal mobility and can make mobility measurements more feasible, and eventually improve its assessment. Of note, this proposal of fewer assessments is to screen impairment in spinal mobility. If impairment in LSF and/ or mSchober is identified, then a full assessment of all SMMs is still recommended. This will ultimately inform the rheumatologist about the severity of the disease and will likely guide clinical decisions, such as referral to physiotherapy and/or rehabilitation medicine. An important limitation of the study is that we have used a prevalence cohort of patients with AS and have neither taken into account the development of impairment of spinal mobility Impairment of spinal mobility

over time, nor whether this also occurs in a fixed order. We have rather analysed impairment in spinal mobility cross-sectionally at multiple time points. We have found a hierarchy in the occurrence of impairments per SMM and it is plausible that impairment in spinal mobility over time occurs in an order described in here. However, this order needs to be prospectively validated. Moreover, a comparison was established between patients with AS and â&#x20AC;&#x2DC;normal valuesâ&#x20AC;&#x2122; derived from normal individuals. We cannot therefore know whether these findings are specific of AS, or whether the same type of impairment would be found in other diseases affecting the spine, e.g. mechanical discopathy. Another possible critique is the imputation of LSF values of Dutch patients during the first 2 years. Even though the imputation method performed very well 13, we cannot fully exclude some possible errors. Strengths of our study are the formal comparison between patients with AS with percentile curves derived from normal individuals. We included a large number of measurements, and the hierarchy proposed persisted across subgroups, which adds credibility to the findings. Confirmation of our findings in another cohort, and also in earlier phases of the development of mobility impairment, e.g. in patients with non-radiographic axSpA, are warranted. If confirmed, it would be interesting to explore what factors are associated with this higher impairment in mobility of the lumbar spine, compared to the cervical or thoracic spine. Inflammation may hypothetically be one of the factors involved, as we know that treatment with TNFi, which controls inflammation and disease activity, is associated with improvement in spinal mobility 8,9. In summary, this study has shown a fixed order for impairment in spinal mobility and provided clinicians with a data-driven approach to screen impairment in spinal mobility by assessing the LSF and mSchober in patients with axSpA.

Chapter 5

REFERENCES 1.

Carette S, Graham D, Little H, et al. The natural disease course of ankylosing spondylitis. Arthritis Rheum 1983;26:186-90.

Amor B, Santos RS, Nahal R, et al. Predictive factors for the longterm outcome of spondyloarthropathies. J Rheumatol 1994;21:1883-7.

Laurent MR, Buchanan WW, Bellamy N. Methods of assessment used in ankylosing spondylitis clinical trials: a review. Br J Rheumatol 1991;30:326-9.

Jenkinson TR, Mallorie PA, Whitelock HC, et al. Defining spinal mobility in ankylosing spondylitis (AS). The Bath AS Metrology Index. J Rheumatol 1994;21:1694-8.

Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68 Suppl 2:ii1-44.

van der Heijde D, Kivitz A, Schiff MH, et al. Efficacy and safety of adalimumab in patients with ankylosing spondylitis: results of a multicenter, randomized, double-blind, placebo-controlled trial. Arthritis Rheum 2006;54:2136-46.

10.

Ramiro S, van Tubergen A, Stolwijk C, et al. Reference intervals of spinal mobility measures in normal individuals: the mobility study. Ann Rheum Dis 2014.

11.

12.

van der Heijde D, Landewe R, Feldtkeller E. Proposal of a linear definition of the Bath Ankylosing Spondylitis Metrology Index (BASMI) and comparison with the 2-step and 10-step definitions. Ann Rheum Dis 2008;67:489-93.

13.

Ramiro S, van Tubergen A, Stolwijk C, et al. Neutral lateral fingertip-to-floor distance can be derived from height. Ann Rheum Dis 2014;73:1748-9.

14.

Ramiro S, Stolwijk C, van Tubergen A, et al. Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study. Ann Rheum Dis 2015;74:52-9.

15.

Moll JM, Wright V. An objective clinical study of chest expansion. Ann Rheum Dis 1972;31:1-8.

16.

Moll JM, Liyanage SP, Wright V. An objective clinical method to measure lateral spinal flexion. Rheumatol Phys Med 1972;11:225-39.

17.

Brandt J, Listing J, Sieper J, et al. Development and preselection of criteria for short term improvement after anti-TNF alpha treatment in ankylosing spondylitis. Ann Rheum Dis 2004;63:1438-44.

Impairment of spinal mobility

Scoring radiographic progression in ankylosing spondylitis: should we use the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) or the Radiographic Ankylosing Spondylitis Spinal Score (RASSS)? Sofia Ramiro, Astrid van Tubergen, Carmen Stolwijk, Robert Landewé, Filip van den Bosch, Maxime Dougados, Désirée van der Heijde

Arthritis Res Ther 2013;15:R14

ABSTRACT Introduction

Radiographic damage is one of the core outcomes in axial SpA and is usually assessed with the modified Stoke Ankylosing Spondylitis (AS) Spine Score (mSASSS). Alternatively, the Radiographic AS Spinal Score (RASSS) is proposed, which includes the lower thoracic vertebrae, under the hypothesis that most progression occurs in these segments. We aimed to compare the mSASSS and RASSS with regard to performance. Methods

Two-yearly spinal radiographs from patients followed in the Outcome in AS International Study (OASIS) were used (scored independently by two readers). A total of 195 patients had at least one radiograph (12 year follow-up) to be included. We assessed the accessibility of vertebral corners (VCs) for scoring, as well as status and 2 year progression scores of both scoring methods. To assess the potential additional value of including the thoracic segment in the score, the relative contribution (in %) to the 2 year total RASSS progression of each spinal segment (cervical, thoracic and lumbar) was determined, and compared to the expected contribution, under the assumption that a balanced segmental progression would occur, proportional to the number of sites per segment. Results

The mSASSS could be scored in a total of 809 radiographs and the RASSS in 78% of these. In 58% of the latter, the score was based on one to two available thoracic VCs scores, and the remaining two to three were imputed because they were missing. There were 520 twoyear mSASSS intervals available, and in 63% of them RASSS progression could be assessed. The mean (SD) 2 year interval progression score (330 intervals) was 2.0 (3.6) for the mSASSS and 2.4 (4.4) for the RASSS, yielding a similar effect size (mSASSS 0.57 and RASSS 0.55). Exclusive progression of the thoracic segment occurred in only 5% of the cases. There was no significant difference between the observed (14%) and expected (16%) contribution to progression of the thoracic segment (P = 0.70). Conclusion

The determination of RASSS for radiographic damage of the spine is frequently impossible or strongly influenced by non-contributory imputation. In comparison to the mSASSS, the contribution of thoracic VCs in the RASSS method is negligible, and does not justify the additional scoring efforts.

Chapter 6

INTRODUCTION Radiographic damage is one of the core outcomes in axial spondyloarthritis (axial SpA) (including both non-radiographic axial SpA and ankylosing spondylitis (AS)) as recommended by the Assessment of SpondyloArthritis international Society (ASAS)1. Cross-sectionally, it is associated with impairment in spinal mobility 2,3 and longitudinally with functional disability 4, emphasizing the importance of assessment. ASAS recommends routine radiography of the lateral cervical and lumbar spine for assessing damage over time, but radiographs should not be repeated more frequently than every 2 years, unless indicated in individual cases, who might show faster progression 5,6. Different scoring methods have been developed to quantify structural damage in axial spondyloarthritis: the Bath AS Radiology Index (BASRI) 7, the Stoke AS Spine Score (SASSS) 8

and a modification of the SASSS, the mSASSS 9. In a formal comparison, the mSASSS has

shown best reliability and sensitivity to change 10. Consequently, it is the preferred scoring method for assessing structural damage in the spine for use in clinical trials, as endorsed by ASAS and Outcome Measures in Rheumatology Clinical Trials (OMERACT) 11. The mSASSS assesses the presence of erosions, sclerosis, squaring, syndesmophytes and bridges at the anterior vertebral corners (VCs) of both the cervical and lumbar spine 9. More recently, a new scoring method, the Radiographic AS Spinal Score (RASSS), has been proposed that includes the lower thoracic vertebrae, under the hypothesis that most progression is found in these segments

. Four thoracic VCs are added and the same

features are scored as for the mSASSS, though with slightly modified scoring rules. The usefulness of the RASSS has not been further evaluated thus far. Hence, it is important to compare both scoring methods, in order to establish the preferred method for the assessment of structural damage as an outcome measure. Outcome measures should be valid in all their aspects. To standardize the nomenclature of validity, the OMERACT filter has been proposed and this includes three aspects: discrimination, truth and feasibility 13. The main objective of the present study was to compare the mSASSS and RASSS with regard to performance, taking the aspects of the OMERACT filter into account.

mSASSS vs RASSS

METHODS Patients and radiographs

Radiographs from patients included in the Outcome in Ankylosing Spondylitis International Study (OASIS) were used

. The OASIS study is a prevalence cohort including 217

14,15

consecutive patients with AS from the Netherlands, Belgium and France that started in 1996. According to protocol, cervical and lumbar spine radiographs were taken biannually for 12 years, with a total of seven possible time points per patient. For the present study, patients were included if they had at least one time point in which at least one of the radiographic damage scores could be calculated. Scoring methods

The two scoring methods used were the mSASSS 9 and the RASSS 12 (table 1). In the mSASSS the anterior VCs of the cervical (lower border of C2 to upper border of T1) and lumbar (lower border of T12 to upper border of S1) segments (a total of 24 VCs) are scored at a lateral view, for the presence of erosion and/or sclerosis and/or squaring (1 point), syndesmophyte (2 points) and bridging syndesmophyte (3 points). The total score ranges from 0 to 72 9. The RASSS is similarly scored as the mSASSS with 3 modifications: 1) inclusion of the lower thoracic spine (lower border of T10 to upper border of T12; total of 28 VCs); 2) erosions are not scored; 3) squaring is not scored in the cervical spine. The RASSS ranges from 0 to 84 12. Table 1 – Description of the mSASSS and RASSS scoring systems mSASSS

RASSS

Cervical spine

Lower border of C2 to upper border of T1

Thoracic spine

Not included

Lower border of T10 to upper border of T12

Lumbar spine

Lower border of T12 to upper border of S1

Range of scoring system

0-72

0-84

Spinal segments assessed

Scoring definitions 0

No change

Erosion, squaring, sclerosis

Squaring only for the thoracic and lumbar segments; no erosions scored; sclerosis scores for all VCs

Syndesmophytes

Bridging syndesmophytes / ankylosis

mSASSS – modified Stoke Ankylosing Spondylitis Spine Score; RASSS – radiographic Ankylosing Spondylitis Spinal Score; C – cervical, T – thoracic; S - sacral

Chapter 6

The radiographs were independently scored according to both scoring methods by two trained experts (SR and CS) who were blinded to demographic and clinical data. Both readers registered all the changes identified in each VC (for example, erosions, sclerosis and squaring) separately so that afterwards both scores could be computed. Because radiographs were taken in different formats during the 12 years of follow-up, enabling the readers to identify the points in time, they were scored with known chronology. All the available films per patient were scored at the same time. Only scores of radiographs with â&#x2030;¤ 3 missing VCs per segment (either cervical or lumbar) were used. For the RASSS, the same rule applied and the four additional thoracic VCs were considered part of the lumbar segment 12. Reliability between the two readers was explored using Bland and Altman analysis 16 on the progression intervals. All radiographs from patients with at least one score being beyond the 95% level of agreement were independently scored by an adjudicator (AvT). Averaged scores per VC of the two primary readers were used.

In adjudicated cases, the score of the primary reader closest to the adjudicator was used. Missing VCs were imputed using an adaptation of the last-observation-carried-forward methodology. First, a missing value for a VC was replaced with the value of the previous observation. Then, the mean spinal segmentâ&#x20AC;&#x2122;s progression score (either cervical or lumbar) per patient was calculated. This was added to the imputed value, in an attempt to more accurately reproduce the true progression. This rule was applied assuring that the score achieved per VC never exceeded a score of 3. Similarly, in case of a score missing in a patient with a score of 0 in the same VC at a subsequent time point, the score of 0 for the previous time point(s) was assumed. If the baseline score of a VC was missing, the same procedure was applied, subtracting the mean segment progression from the score of year 2 for a particular patient. If a value of this VC was also missing at year 2, then an average of the other available VCs from this spinal segment at baseline was used to replace the missing VC(s). Status and progression scores were calculated for both scoring methods. Status scores refer to the score in each of the available time points (at baseline and every 2 years thereafter). Progression scores were calculated as the difference between the status scores of two time points. Two year progression scores refer to the progression occurring within 2 years, that is, status score of one time point minus the status score of the immediately previous time point. Twelve year progression scores were computed as the score at year 12 minus the score at baseline. Use of the OMERACT filter to compare the scoring methods

The mSASSS and the RASSS were judged with respect to the different aspects of the

mSASSS vs RASSS

OMERACT filter: truth, discrimination, and feasibility 13. Feasibility

The feasibility aspect of the OMERACT filter addresses the question: can the measure be applied easily, given constraints of time, money and interpretability? The feasibility of both methods (mSASSS and RASSS) was assessed. Because the RASSS requires a further four additional thoracic VCs to be present in the radiograph of the lumbar spine, the assessment of the ability to obtain both scores is important. The availability of the VCs and the ability to assess the status and the 2 year progression scores of both scoring methods was compared and the number of available VCs out of the four additional VCs included in the RASSS was also investigated. Comparisons were performed calculating a ratio of the available cases for the RASSS over the mSASSS, taking all radiographs into account, but also restricted to 1) patients with a RASSS available at year 12 in order to assess whether the RASSS would perform differently in the subset of patients with a complete follow-up and 2) patients with the first interval between years 0 and 2 available to compare with other results available in the literature for the RASSS 12. Discrimination

The discrimination aspect focuses on the question: does the measure discriminate between situations of interest? This aspect of the OMERACT filter pertains to sensitivity to change and reliability. Inter-observer reliability was assessed for both status and progression scores for both mSASSS and RASSS, by means of Bland and Altman plots

and by calculation of the

smallest detectable change (SDC) for each method. The SDC is the smallest change that can be detected beyond measurement error to determine change in an individual and was calculated as follows

: SDC = 1.96*SD diff /

. SD diff is the standard deviation

(SD) of the set of differences in change scores obtained by two readers; k is the number of readers whose change scores are used (here: k = 2). To obtain insight into sensitivity to change of the methods, the means and SDs of baseline, 2 year and 12 year status scores were assessed. Effect sizes (for all 2 year progression scores) were calculated for both mSASSS and RASSS dividing the mean value of the progression scores by the corresponding standard deviation. Truth

The truth aspect deals with the question: is the measure truthful, does it measure what is intended? Is the result unbiased and relevant? Both mSASSS and RASSS are, to a certain extent, similar, which means that they have a common part of construct validity. Therefore, 66

Chapter 6

we assessed the potential additional value of including the thoracic vertebrae in the RASSS, by determining the relative contribution (in %) to the 2 year total RASSS progression of each spinal segment (cervical, thoracic and lumbar) in comparison to the expected contribution. A balanced segmental progression, proportional to the number of VCs assessed in the RASSS (twelve cervical VCs, four thoracic VCs and twelve lumbar VCs) was assumed. The expected and balanced contribution assumed was 43% (12/28 VCs) for each of the cervical and lumbar segments and 14% (4/28 VCs) for the thoracic segment. Statistical analysis

Descriptive statistics were performed, with continuous variables being presented as mean

(SD) and categorical variables as frequencies. Observed and expected progression rates were compared using the chi-square test and a 5% level of significance was assumed. Stata SE version 11 was used. (Statacorp, College Station, TX, USA)

RESULTS A total of 195 patients had at least one radiograph that could be scored (according to the mSASSS and/or RASSS), 64 had a radiograph that could be scored at year 12 and a total of 520 2 year progression intervals throughout the 12 year follow-up period were available Patients had a mean age of 42.8 (SD 12.4) years, mean disease duration since symptom onset of 20.0 (SD 11.6), mean disease duration since diagnosis of 11 (SD 8.7) years, 71% were males and 84% HLA-B27 positive. Baseline demographic, clinical and radiographic characteristics are summarized in table 2. Feasibility

The mSASSS could be scored in a total of 809 radiographs. The RASSS could be calculated in 78% of these radiographs (n = 629) (tables 3 and 4). In 58% of those, in which the RASSS was calculated, the score was based on one or two available thoracic VC scores and the remaining two to three needed imputation because they were missing, so were in fact noninformative. There were 520 2 year mSASSS interval progression scores available, and in 63% of them a 2 year RASSS interval progression score could be determined. If the availability of all four additional VCs was required, then the RASSS could only be calculated in 226 (36%) radiographs and in 64 (19%) progression intervals. For the subgroups of radiographs from patients with the first 2 year interval available and of radiographs from patients with a RASSS available at year 12, see tables 3 and 4.

mSASSS vs RASSS

Table 2 – Baseline demographic, clinical and radiographic characteristics of the patients included in assessment of the radiographic progression in this study Assessment

N = 195*

Age (years)

42.8 (12.4)

Male gender (%)

138 (71%)

HLA-B27 positive (%)

158 (84%)

Symptoms duration (years)

20.4 (12.9)

Disease duration (years)

11.0 (8.7)

ASDAS-CRP

2.7 (1.0)

BASDAI (0-10)

3.4 (2.0)

BASFI (0-10)

3.2 (2.5)

BASMI (0-10)

3.7 (1.5)

CRP (mg/l) (N = 186) Elevated CRP (%)‡

17.5 (23.5) 96 (52%)

mSASSS (0-72) (N = 177)¥

10.8 (15.2)

RASSS (0-84) (N = 130)

11.8 (16.6)

mSASSS of patients with available RASSS (0-72) (N = 130)

10.1 (14.2)

mSASSS>0 (%) (N = 177)

143 (81%)

RASSS>0 (%) (N = 130)

107 (82%)

mSASSS>0 of patients with available RASSS (%) (N = 130)

106 (82%)

*Data are presented as mean (SD) or n (%). N = 195, representing patients with ≥1 radiograph with the mSASSS evaluable throughout follow-up (not necessarily at baseline available) ‡ The cut-off was 10mg/l for the Dutch patients (n = 122) and 5mg/l for the Belgian (n = 23) and French patients (n = 50) ¥ 18 patients did not have a radiograph with an mSASSS evaluable at baseline, but had a radiograph in which the mSASSS could be calculated at a later time point and were therefore included in the study ASDAS-CRP – Ankylosing Spondylitis Disease Activity Score (C-reactive protein); BASDAI – Bath Ankylosing Spondylitis Disease Activity Score; BASFI – Bath Ankylosing Spondylitis Functional Index; BASMI – Bath Ankylosing Spondylitis Metrology Index; CRP – C-reactive protein; mSASSS – modified Stoke Ankylosing Spondylitis Spine Score; RASSS – radiographic Ankylosing Spondylitis Spinal Score;

Discrimination

The first part of the discrimination aspect is reliability. Figure 1 shows the Bland and Altman plots for the progression scores of both the mSASSS and RASSS. In general, both scores could be reliably performed without clear systematic error. The SDC for the progression scores was 2.9 for the mSASSS and 3.5 for the RASSS. Of all radiographs in which both the mSASSS and RASSS could be determined (n=629), the mean (SD) status score was 15.5 (17.9) units for the mSASSS and 18.0 (20.9) units for the RASSS (table 5). The mean (SD) 2 year progression score, calculated in 330 2 year intervals, was 2.0 (3.6) for the mSASSS and 2.4 (4.4) for the RASSS. The progression score per spinal segment was 1.2 (2.3) for the cervical segment, 0.8 (2.1) for the lumbar segment and 0.4 (1.2) for the thoracic segment in the RASSS. The effect size of the 2 year progression score 68

Chapter 6

mSASSS vs RASSS

520

Available 2 year progression intervals

330

629

RASSS

63%

78%

% RASSS in radiographs with mSASSS

239

341

mSASSS

159

277

RASSS

mSASSS

184

164

81%

67%

297

226

2VCs

3VCs

4VCs

36%

47%

10%

139

59%

73%

% RASSS in radiographs with mSASSS

8% 25%

33%

53%

71 11

13% 7%

10%

% 50%

All baseline radiographs from patients with baseline – year 2 RASSS interval available (n = 134)* %

All radiographs from patients with RASSS available at year 12 (n = 341)*

* n = number of radiographs. mSASSS – modified Stoke Ankylosing Spondylitis Spine Score; RASSS – radiographic Ankylosing Spondylitis Spinal Score; VC – vertebral corner

All radiographs with RASSS evaluable (n = 629)*

1VC only

Availability of the 4VCs added in the RASSS

Table 4 – Feasibility of the RASSS – availability of the four thoracic vertebral corners added to the RASSS

134

RASSS

All baseline radiographs from patients with baseline – year 2 interval available (n = 184)*

All radiographs from patients with RASSS available at year 12 (n = 341)*

* n = number of radiographs mSASSS – modified Stoke Ankylosing Spondylitis Spine Score; RASSS – radiographic Ankylosing Spondylitis Spinal Score; VC – vertebral corner

809

Available status scores (≤3 missing VC per segment)

mSASSS

All radiographs (n = 809)*

Table 3 – Feasibility of the mSASSS vs. RASSS – availability of the status and progression scores

was 0.57 (2.0/3.6) for the mSASSS and 0.55 (2.4/4.4) for the RASSS. For all radiographs from patients with a RASSS evaluable at 12 year follow-up (n=59), the mean (SD) status score of the mSASSS was 18.0 (19.3) and of the RASSS 20.8 (22.4); and in this group the mean (SD) 2 year progression score of the mSASSS was 2.4 (4.5) and of RASSS 2.9 (5.5). Compared to the status scores of all radiographs, the status scores of the radiographs from both patients with a RASSS evaluable at 12 year follow-up and radiographs with the four additional thoracic VCs available were higher (table 5). During the first 2 year follow-up, the progression was 1.9 (4.1) units for the mSASSS and 2.2 (5.0) units for the RASSS in patients with available intervals for both mSASSS and RASSS (n=134).

Figure 1 â&#x20AC;&#x201C; Bland & Altman plots: reliability of the mSASSS and RASSS progression scores Difference against mean for mSASSS and RASSS progression scores of the two readers. The SDC for the progression scores was 2.9 for the mSASSS and 3.5 for the RASSS. mSASSS, modified Stoke Ankylosing Spondylitis Spine Score; RASSS, Radiographic Ankylosing Spondylitis Spinal Score; SDC, smallest detectable change.

Chapter 6

mSASSS vs RASSS

RASSS

3.0 (3.9)

10.1 (13.1)

7.1 (9.8)

7.9 (10.1)

18.0 (20.9)

2.4 (4.4)

0.4 (1.2)

1.2 (2.8)

0.8 (2.1)

1.2 (2.3)

RASSS

3.4 (3.8)

11.4 (13.2)

8.0 (9.9)

9.5 (10.7)

20.8 (22.4)

1.0 (2.4)

1.5 (2.6)

2.4 (4.5)

0.5 (1.2)

1.5 (3.3)

1.0 (2.4)

1.4 (2.9)

2.9 (5.5)

All radiographs with RASSS evaluable and from patients with RASSS available at year 12 and with 2 year RASSS intervals evaluable (n = 159)

8.0 (10.0)

10.0 (10.7)

18.0 (19.3)

mSASSS

All radiographs with RASSS evaluable and from patients with RASSS available at year 12 (n = 277)

2.0 (3.4)

0.9 (2.2) -

0.9 (2.2) 1.6 (3.3)

0.7 (1.7)

0.9 (2.2)

1.0 (2.6)

1.9 (4.1)

2.8 (4.2) 1.1 (1.9)

2.1 (3.3)

4.6 (7.9)

5.2 (8.1)

0.4 (1.4)

1.3 (3.1)

0.9 (2.2)

0.9 (2.9)

2.2 (5.0)

All radiographs with RASSS evaluable from patients with baseline-year 2 interval evaluable (n = 96)

5.5 (8.2)

1.2 (2.0)

All radiographs with four additional VCs available and 2 year RASSS intervals evaluable (n = 64)

3.8 (4.4)

6.6 (10.7)

4.6 (7.9)

8.4 (10.2) 12.2 (13.7)

8.4 (10.2) -

8.3 (10.0)

RASSS 11.8 (16.6)

8.9 (10.0)

mSASSS 10.2 (14.2)

RASSS 20.5 (21.4)

mSASSS

All baseline radiographs with RASSS evaluable from patients with baseline-year 2 interval evaluable (n = 134)

17.3 (18.2)

All radiographs with four additional thoracic VCs available (n = 226)

*Data are expressed as mean (SD). mSASSS – modified Stoke Ankylosing Spondylitis Spine Score; RASSS – radiographic Ankylosing Spondylitis Spinal Score; SD – standard deviation, VC – vertebral corner

Thoracic segment

Lumbar segment

0.8 (2.0)

Cervical segment

Lumbar segment (with thoracic segment included)

2.0 (3.6)

1.2 (2.2)

Total Score

All radiographs with 2 year RASSS intervals evaluable (n = 330)

Thoracic segment

2 year PROGRESSION SCORES

7.1 (9.8)

Lumbar segment

Lumbar segment (with thoracic segment included)

8.4 (10.1)

Cervical segment

mSASSS

15.5 (17.9)

Total Score

STATUS SCORES

All radiographs with RASSS evaluable (n = 629)

Table 5 –– Status and progression mSASSS and RASSS scores for radiographs in which RASSS was evaluable

In patients with an mSASSS evaluable at baseline and at 12 year follow-up (n=64), the mean (SD) 12 year progression was 11.7 (11.5). In 38 out of the 64 patients, the RASSS could be assessed with a mean 12 year progression of 14.2 (14.6) and a corresponding mSASSS progression of 12.2 (12.5) in this group. In patients with a RASSS available at year 12 (n=59), the baseline, 2 year and 12 year RASSS status scores were 11.8 (15.3), 12.3 (13.9) and 28.5 (25.0), respectively. For the mSASSS, the corresponding status scores were 10.6 (13.6), 11.2 (12.4) and 24.5 (21.6), respectively. Interestingly, in a few cases, the RASSS enabled the occurrence of negative progression scores. Taking all radiographs into account, in five VCs in the cervical segment, first sclerosis was considered to be present and scored as 1, but at follow-up an erosion appeared, which was still scored as a 1 in the mSASSS but was scored 0 according to the RASSS scoring rules and, consequently, the progression score for that specific VC was -1 for RASSS. Truth

Exclusive progression in the thoracic segment, which can only be captured by the RASSS

and not by the mSASSS, occurred in only 5% of the 2 year progression intervals (17 intervals out of 330). In 2% of the 2 year progression intervals, a progression of two or more units (possibly indicating new syndesmophyte formation) in the thoracic segment was found. In 25% of the intervals (81 out of 330 intervals), progression occurred exclusively in the cervical segment and in 7% (23 out of 330) in the lumbar segment. However, there were no significant differences between the observed and expected contributions of the thoracic segment to progression (16% vs. 14%, P=0.70), whilst progression was observed more frequently than expected in the cervical spine (55% vs. 43%, P=0.09), and less frequently in the lumbar spine (29% vs. 43%, P=0.04) (table 6).

DISCUSSION The present study shows that the mSASSS remains the most appropriate method for scoring radiographic progression in patients with AS based on feasibility, discrimination and truth aspects of the OMERACT filter. With regard to feasibility, the 2 year progression scores were available for the RASSS in only 63% of the cases in which mSASSS progression intervals could be calculated. In the paper describing the RASSS, the authors reported an availability of 88% of the progression scores within the first 2 years 12, while in our study only in 59% of the cases a RASSS progression score could be calculated in this first 2 year interval. Furthermore, in our study in one third of the radiographs in which the RASSS could be calculated, only one to two thoracic

Chapter 6

Table 6 – Ratio of 2 year progression in each of the spinal segments of the RASSS Relative contribution in each of the spinal segments of the RASSS (in %)

P value for the difference

Expected

Observed

Cervical segment (12 VCs)

43%

55%

0.09

Lumbar segment (12 VCs)

43%

29%

0.04

Thoracic segment (4 VCs)

14%

16%

0.70

*n=330 represents progression intervals RASSS – radiographic Ankylosing Spondylitis Spinal Score NA – not applicable

VCs were accessible , meaning that calculation of the RASSS was based on imputed and therefore non-informative VCs in the lumbar spine. This shows that an important number of radiographs obtained in the Netherlands, France and Belgium does not include the lower part of the thoracic spine, in contrast to what has been suggested for Germany 12. According

to the OMERACT filter, feasibility captures an essential element in the selection of measures, one that may be decisive in determining a measure’s success 13. The worse feasibility of the RASSS compared with the mSASSS jeopardizes its wide use. The RASSS demonstrated a higher mean progression, but an increase in the variance of the progression scores was also observed, resulting in similar effect sizes between mSASSS and RASSS. A higher mean progression was expected because the RASSS includes four additional VCs compared with the mSASSS, so that the RASSS is by definition almost always higher than the mSASSS. Exceptions are only the cases with erosions in any segment or squaring in the cervical spine, which are scored for the mSASSS, but not for the RASSS. Nevertheless, and comparing the progression scores with the limited data available in the literature for the RASSS, our RASSS progression scores were higher. In the first 2 years, we found a progression in the RASSS of 2.2 (5.0), whereas Baraliakos et al. reported a 2 year progression of 1.6 (2.8) 12. Also, our mSASSS progression scores were higher compared with the same study (1.9 (4.1) vs. 0.9 (2.5)) 12. A possible explanation for the difference in progression scores can be the difference in baseline radiographic damage (baseline mSASSS of 8.1 (14.6) in German cohort vs. 10.2 (14.2) in OASIS cohort). It is well known that presence of radiographic damage is a predictor of further and faster progression of radiographic damage

. Other literature on RASSS

18-20

progression scores is currently lacking. However, our mSASSS progression scores can be compared with other available studies. There are reports of 2 year mSASSS progression scores of around 1 mSASSS unit

, 2.5 units

21-23

and 2.6 units (extrapolation to a 2 year

period of the annual progression rate of 1.3 (2.5), and assuming linearity) 20. The differences between scores can be attributed to differences in selection of patients, baseline radiographic damage of patients, conditions in which radiographs were read 24 or the method of imputation

mSASSS vs RASSS

of missing VCs. The increase in the variance around the progression scores resulted in similar effect sizes for both methods (0.57 for the mSASSS and 0.55 for the RASSS), showing that the higher mean progression of the RASSS is offset by the increased noise. For discrimination, both scoring methods seemed to be reliable, however, the SDC for the mSASSS was slightly smaller compared to the RASSS (2.9 vs. 3.5), suggesting that the measurement error with the RASSS is somewhat higher. Reliability of the RASSS could possibly improve by having an additional and separate radiograph to score the thoracic VCs. This could reduce the parallax associated with extending the view of the lumbar radiograph to include the thoracic VCs, but would on the other hand imply higher costs and radiation for the patients. With regard to the truth aspect of the OMERACT filter, we found that most progression occurred in the cervical segment of the spine (55%), followed by the lumbar spine (29%) and only 16% was found in the thoracic vertebrae. Furthermore, the progression in the thoracic vertebrae was not significantly different from what was expected, if progression throughout the spine would occur in a balanced way. In addition, we showed that new syndesmophytes exclusively occurring in the thoracic spine occurred in a maximum of 2% of the intervals. These data should be interpreted with caution, because a progression score of 2 does not always correspond to a new syndesmophyte, but can also mean twice a score of 1 in two separate VCs, reflecting development of squaring or sclerosis. This shows that the RASSS does not capture more progression occurring in the thoracic vertebrae, as was hypothesized by Baraliakos et al. 12. In our study, a 2 year progression in the thoracic vertebrae of 0.4 (1.4) out of a total RASSS progression of 2.2 (5.0)) was found. Baraliakos et al. reported a progression of 0.6 (3.3) out of a total RASSS progression of 1.6 (2.8), indicating a higher contribution from the thoracic VCs to the total RASSS. The reason for this discrepancy between both studies is not entirely clear, but availability of the thoracic VCs for scoring can play a role. Some limitations of the present study should be addressed. Films were obtained throughout a 12 year follow-up, which means that some of them were old and did not have the optimal quality. Nevertheless, this limited the reading of both scoring methods similarly. Furthermore, the findings of this study may not be generalizable to countries where lumbar radiographs routinely include the low thoracic spine, which is not standard procedure in our three countries. In conclusion, the calculation of RASSS for status or progression of radiographic abnormalities in the spine is frequently impossible or strongly influenced by non-contributory imputation. The effect size of both methods is similar. In comparison to the mSASSS, the contribution of thoracic VCs in the RASSS is negligible, and does not justify the additional scoring efforts. The mSASSS remains the most appropriate measure to assess radiographic damage in patients with AS.

Chapter 6

REFERENCES 1.

Machado P, Landewe R, Braun J, et al. Both structural damage and inflammation of the spine contribute to impairment of spinal mobility in patients with ankylosing spondylitis. Ann Rheum Dis 2010;69:1465-70.

Wanders A, Landewe R, Dougados M, et al. Association between radiographic damage of the spine and spinal mobility for individual patients with ankylosing spondylitis: can assessment of spinal mobility be a proxy for radiographic evaluation? Ann Rheum Dis 2005;64:988-94. Landewe R, Dougados M, Mielants H, et al. Physical function in ankylosing spondylitis is independently determined by both disease activity and radiographic damage of the spine. Ann Rheum Dis 2009;68:863-7. Zochling J, van der Heijde D, Burgos-Vargas R, et al. ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2006;65:442-52. Braun J, van den Berg R, Baraliakos X, et al. 2010 update of the ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2011;70:896-904.

MacKay K, Mack C, Brophy S, et al. The Bath Ankylosing Spondylitis Radiology Index (BASRI): a new, validated approach to disease assessment. Arthritis Rheum 1998;41:2263-70.

Averns HL, Oxtoby J, Taylor HG, et al. Radiological outcome in ankylosing spondylitis: use of the Stoke Ankylosing Spondylitis Spine Score (SASSS). Br J Rheumatol 1996;35:373-6.

Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9.

10.

Wanders AJ, Landewe RB, Spoorenberg A, et al. What is the most appropriate radiologic scoring method for ankylosing spondylitis? A comparison of the available methods based on the Outcome Measures in Rheumatology Clinical Trials filter. Arthritis Rheum 2004;50:2622-32.

11.

van der Heijde D, Landewe R. Selection of a method for scoring radiographs for ankylosing spondylitis clinical trials, by the Assessment in Ankylosing Spondylitis Working Group and OMERACT. J Rheumatol 2005;32:2048-9.

12.

Baraliakos X, Listing J, Rudwaleit M, et al. Development of a radiographic scoring tool for ankylosing spondylitis only based on bone formation: addition of the thoracic spine improves sensitivity to change. Arthritis Rheum 2009;61:764-71.

13.

Boers M, Brooks P, Strand CV, et al. The OMERACT filter for Outcome Measures in Rheumatology. J Rheumatol 1998;25:198-9.

14.

15.

Spoorenberg A, de Vlam K, van der Heijde D, et al. Radiological scoring methods in ankylosing spondylitis: reliability and sensitivity to change over one year. J Rheumatol 1999;26:997-1002.

16.

Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10.

17.

Bruynesteyn K, Boers M, Kostense P, et al. Deciding on progression of joint damage in paired films of individual patients: smallest detectable difference or change. Ann Rheum Dis 2005;64:179-82.

18.

van Tubergen A, Ramiro S, van der Heijde D, et al. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23.

19.

Poddubnyy D, Haibel H, Listing J, et al. Baseline radiographic damage, elevated acute-phase reactant levels, and cigarette smoking status predict spinal radiographic progression in early axial spondylarthritis. Arthritis Rheum 2012;64:1388-98.

20. Baraliakos X, Listing J, von der Recke A, et al. The natural course of radiographic progression in ankylosing spondylitis--evidence for major individual variations in a large proportion of patients. J Rheumatol 2009;36:997-1002. 21. van der Heijde D, Landewe R, Baraliakos X, et al. Radiographic findings following two years of infliximab therapy in patients with ankylosing spondylitis. Arthritis Rheum 2008;58:3063-70. 22. van der Heijde D, Landewe R, Einstein S, et al. Radiographic progression of ankylosing spondylitis after up to two years of treatment with etanercept. Arthritis Rheum 2008;58:1324-31. 23. van der Heijde D, Salonen D, Weissman BN, et al. Assessment of radiographic progression in the spines of patients with ankylosing spondylitis treated with adalimumab for up to 2 years. Arthritis research & therapy 2009;11:R127.

mSASSS vs RASSS

24. Wanders A, Landewe R, Spoorenberg A, et al. Scoring of radiographic progression in randomised clinical trials in ankylosing spondylitis: a preference for paired reading order. Ann Rheum Dis 2004;63:1601-4.

Chapter 6

mSASSS vs RASSS

Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study Sofia Ramiro, Carmen Stolwijk, Astrid van Tubergen, Désirée van der Heijde, Maxime Dougados, Filip van den Bosch, Robert Landewé

Ann Rheum Dis 2015;74:52-9

ABSTRACT Objectives

To describe the evolution of radiographic abnormalities of the spine in patients with ankylosing spondylitis (AS). Methods

Patients with AS were followed prospectively with 2 yearly radiographs for 12 years. The modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) was scored by two readers (R1 and R2). New syndesmophytes at uninvolved vertebral corners were computed. Radiographic progression was investigated using generalised estimating equations. Results

809 radiographs (presenting 520 at 2 yearly intervals) from 186 patients (70% men, mean age 43 (SD 12) years, mean 20 (SD 12) years since symptom onset and 83% HLA-B27 positive) were included. Mean mSASSS at baseline was 11.6 (16.2). While the course of progression in individual patients was highly variable, and still occurred in patients with decades of symptom duration, mean 2 year progression was 2.0 (3.5) mSASSS units. Over the entire follow-up, at least one new syndesmophyte was found in 55% (R1) and 63% (R2) of patients (38%(R1) and 39%(R2) of all intervals). In 24% of patients (39% of intervals) there was no progression. A progression â&#x2030;Ľ5 mSASSS units occurred in 22% of patients (or in 12% of intervals). At the group level, a linear time course model fitted the data best, with a constant rate over the entire 12 year interval of 0.98 mSASSS units/year. Radiographic progression occurred significantly faster in men, in HLA-B27 positive patients and in patients with a baseline mSASSSâ&#x2030;Ľ10. Conclusion

Long term radiographic progression in AS is highly variable in the individual patient, more

severe in HLA-B27 positive men and still occurs after decades of disease. At the group level, however, progression in AS follows an approximately linear course.

Chapter 7

INTRODUCTION Axial Spondyloarthritis (axSpA), which includes both non-radiographic axSpA and ankylosing spondylitis (AS), is a chronic inflammatory rheumatic disease that primarily affects the axial skeletal. Likely as a consequence of inflammation, structural damage, mainly characterised by the development of syndesmophytes, may arise over time, which may lead to ankylosis of the spine. It is generally assumed that the disease process is most active in the early phases of the disease and that it becomes quiescent over time. Radiographic damage is one of the core outcomes recommended by the Assessment of SpondyloArthritis international Society (ASAS) for follow-up of patients with axSpA 1. ASAS recommends routine radiography of the lateral cervical and lumbar spine for assessing damage over time in which a 2 year interval is the shortest period defined for reliable assessment of progression 2,3. Syndesmophyte formation in axSpA is not easily retarded by treatment: biological agents have failed to show retardation in structural damage progression

4-6

and only non-steroidal

anti-inflammatory drugs (NSAIDs) have an inhibitory effect, especially in patients with elevated C reactive protein (CRP)

. This emphasises the importance of gaining further insight into

7-9

radiographic progression, in particular about how it naturally evolves during the course of the disease. Radiographic progression has been assessed in a few short-term studies (2 years 10,11

or 4 years 12,13). One study with a duration of 8 years had a retrospective design and only

included hospitalised patients 14. The aim of the present prospective study was to further investigate the evolution of radiographic abnormalities in patients with AS during 12 years of follow-up. The course of radiographic progression over time was investigated by longitudinal modelling of time, taking into account factors that may significantly modify the course of progression.

METHODS Study population

This study included patients from the Outcome in Ankylosis Spondylitis International Study (OASIS), a prevalence cohort including 217 consecutive patients with AS from The Netherlands, Belgium and France that began in 1996 15. Clinical data were collected every 6 months until year 2, then yearly until year 4 and thereafter biannually. Cervical and lumbar spine radiographs were taken biannually during 12 years, with a total of seven possible time points per patient. For the present study, patients were included if they had at least two subsequent time points with an available radiographic damage score so that radiographic

Radiographic damage over time

progression could be evaluated. All patients gave informed consent to be included in the study. Scoring methods

Radiographs were scored using the modified Stoke Ankylosing Spondylitis Spine Score

(mSASSS) 16. According to this method, the anterior vertebral corners (VCs) of the cervical (lower border of C2 to upper border of T1) and lumbar (lower border of T12 to upper border of S1) segments (total of 24 VCs) are scored in the lateral view only for the presence of erosion and/or sclerosis and/or squaring (1 point), syndesmophyte (2 points) and bridging syndesmophyte (3 points). The total score per patient ranges from 0 to 72 16. Two trained experts (SR and CS) independently scored the radiographs, blinded to the demographic and clinical data but with known chronology as this is the most sensitive method . All of the available films per patient were scored at the same time but by both readers

separately. Only scores of radiographs with ≤3 missing VCs per segment (either cervical or lumbar) were used. Individual missing VCs were imputed (see online supplementary text section 1). Reliability between the two readers (R1 and R2) was explored using Bland and Altman analysis 18 on the progression intervals of mSASSS. An independent adjudicator (AvT) scored all of the radiographs from patients with at least one score beyond the 95% level of agreement. Averaged scores per VC of the two primary readers were used and, in case of adjudication, the score of the primary reader closest to the adjudicator. Radiographic abnormalities

Status scores (the score at each of the available time points), and progression scores (the difference between the status scores of two time points) were calculated. Two year progression scores refer to the progression occurring within 2 years- that is, status score of one time point minus the status score of the immediately previous time point. Twelve year progression score was computed as the score at year 12 minus the score at baseline. Progression scores were calculated for all patients included in the study and also for those who had an mSASSS at 12 years (’12 year completers’) in order to trace eventual bias due to loss of follow-up. The contribution (in per cent) of each of the segments (cervical and lumbar) to the 2 year total mSASSS progression was compared with the expected contribution, assuming a balanced distribution (ie, 50% in each of the segments). In order to better understand the magnitude of the progression of damage, the available 2 year progression scores were classified in categories of progression: 0; > 0 and < 1; ≥ 1; ≥ 2; ≥ 1 and < 3; ≥ 3; ≥ 3 and < 5; and ≥ 5 mSASSS units.

Chapter 7

At a reader level, the proportion of 2 year intervals with a given number of new syndesmophytes was calculated, both for all patients and for ’12 year completers’, and for uninvolved VCs (without a previous syndesmophyte or bridge). The proportion of patients in whom at least one new syndesmophyte was identified during follow-up was also estimated. For further details on data collection, see online supplementary text section 2. Statistical analysis

Interobserver reliability was assessed by Bland and Altman plots 18 and by smallest detectable change (SDC) for 2 year progression scores. SDC, based on the 95% limits of agreement, is the smallest change that can be detected in an individual beyond measurement error and was calculated as follows 19: SDC = (1.96 X SDdiff)/(

) where SDdiff is the SD of the

set of differences in change scores obtained by the readers; k is the number of readers (here two). Radiographic progression over time was investigated using generalised estimating equations (GEE). This is a technique for longitudinal analysis which makes use of all available longitudinal

data, allows unequal numbers of repeated measurements and has some robustness against deviation from normality

. GEE corrects for the within subject correlation and for this it

20,21

requires an a priori defined ´working´ correlation structure. In this study, the ´exchangeable´ correlation structure was appropriate because the correlations of the outcome at different time points were approximately equal (Spearman correlation coefficients between 0.80 and 0.97). Time was modelled in linear and non-linear (quadratic, cubic, logarithmic, exponential or square root) modes and the best fit was determined using the lowest quasi-likelihood information criterion. Relevant two-, three- and four-way interactions between clinical and demographic characteristics of the patients with time were explored, and if a relevant interaction (p<0.1) was found, progression over time was assessed in the subgroups after stratification. Interactions were tested with HLA-B27 status, gender, symptom duration, baseline mSASSS, NSAID intake (dichotomous variable and ASAS NSAID index

) and

exposure to tumour necrosis alpha inhibitors (TNFi) during follow-up. An interaction between time and a dichotomous variable reflecting the first and the second 6 years of follow-up was also tested in order to assess whether the progression rate changed over time. Because the dependent variable was mSASSS and the independent variable time, the regression coefficient obtained reflects the progression of mSASSS per year. Stata SE V.12 was used (Statacorp, College Station, Texas, USA).

RESULTS In total, 186 patients in whom paired x-rays were available were included in this study, with Radiographic damage over time

the following baseline characteristics: 70% men; 83% HLA-B27 positive; mean age 43 (SD 12) years; mean time since symptom onset 20 (SD 12) years; mean disease duration 11 (SD 9) years; Bath AS Disease Activity Index 3.4 (2.0); and CRP 17 (23) mg/L (47% had an elevated CRP). At baseline, 68% of patients were exposed to NSAIDs and none to TNFi. Throughout follow-up, 95% of patients were at some time exposed to NSAIDs and 22% to TNFi (but before year 8 of follow-up, this proportion was 5%). Duration of follow-up was, on average, 7.9 (4.0) years per patient (range 2-12 years). A total of 809 radiographs [386 radiographs for the 68 ’12 year completers’] were obtained in which mSASSS could be determined (4.2 (1.7) radiographs per patient (range 2-7)). Interobserver reliability was ‘good’ (see online supplementary figure S1). SDC measured across all progression intervals was 2.9. Progression in mSASSS

Mean (SD) mSASSS at baseline was 11.6 (16.2) [11.2 (15.7) in the ’12 year completers’]. At

baseline, 47% (R1) and 58% (R2) of patients had at least one syndesmophyte present: 9% (R1) and 48% (R2) had a score of 1 in at least one VC. Mean mSASSS progressed gradually, from 11.6 (16.1) at baseline (n = 184) to a mean value of 24.5 (21.7) at 12 years (n = 68). The mean 2 year progression rate (in 520 intervals) was 2.0 (3.5) [2.2 (3.9) for the ‘12 year-completers’]. The mean 2 year progression in the cervical segment was 1.2 (2.2) and mean progression in the lumbar segment was 0.8 (2.0). Progression was significantly higher in the cervical segment (accounting for 66% of progression) than in the lumbar segment (33%) (p=0.022) (table 1). Progression in the cervical segment was not higher in patients with psoriasis (data not shown). In the ‘12 year completers’, mean progression was 11.7 (11.5) (n = 64). Over the entire follow-up period, in 24% of patients and in 18% of the ‘12 year completers’ (and in 39% of the 2 year intervals) there was no progression in mSASSS. Progression ≥1 mSASSS unit occurred in 72% of patients (and in 54% of the 2 year intervals) and progression ≥5 mSASSS units in 22% of patients (and in 12% of the 2 year intervals) (table 2). When taking only the first 2 year interval into consideration, 46% of patients did not show any mSASSS progression, 48% showed progression ≥2 units and 25% showed progression ≥5 units. Development of new syndesmophytes

A new syndesmophyte developed in 38% of all 2 year intervals according to R1 and in 39% according to R2 in previously uninvolved VCs. Throughout follow-up, 55% (R1) and 63% (R2) of patients developed at least one syndesmophyte in previously uninvolved VCs (table 3). For

Chapter 7

1.0 (2.2)

0.8 (2.2)

Cervical segment

Lumbar segment

10.4 (10.5)

20.4 (19.8) 11.8 (11.7)

21.0 (21.0)

0.7 (1.9)

1.1 (1.9)

1.8 (3.2)

Y4-Y6 (N = 93)

1.0 (2.4)

1.5 (2.0)

2.5 (3.8)

Y6-Y8 (N = 52)

10.8 (11.7)

13.7 (11.5)

24.5 (21.7)

Y12 (n = 68)

0.6 (1.1)

1.3 (1.6)

1.9 (2.0) 0.8 (1.8)

1.1 (1.9)

1.9 (3.2)

Y8-Y10 Y10 – Y12 (N = 41) (N = 55)

8.1 (10.2) 9.9 (11.8) 9.2 (11.0)

8.7 (10.3)

16.8 (17.8)

Y10 (n = 82)

6.7 (8.2) Y2-Y4 (N = 50) 2.7 (3.9) 1.5 (2.3) 1.2 (2.2)

6.1 (8.9) 5.2 (6.8) Y0-Y2 (N = 60) 1.9 (4.9) 1.2 (3.2) 0.7 (2.0)

5.9 (8.9) 5.3 (7.8)

17.8 (18.9)

Y6 (n = 47)

0.9 (2.6)

1.1 (1.5)

2.0 (3.4)

Y4-Y6 (N = 42)

8.5 (9.7)

8.3 (10.1) 9.3 (10.3)

15.0 (16.8)

11.3 (14.3)

11.2 (15.7)

Y4 (n = 52)

Y2 (n = 62)

1.4 (3.1)

1.7 (2.1)

3.2 (4.7)

Y6-Y8 (N = 29)

10.0 (11.3)

11.3 (10.7)

21.3 (20.8)

Y8 (n = 38)

10.8 (11.7)

13.7 (11.5)

24.5 (21.7)

Y12 (n = 68)

0.7 (1.2)

1.4 (1.7)

2.1 (2.1)

0.8 (1.8)

1.1 (1.9)

1.9 (3.2)

Y8-Y10 Y10 – Y12 (N = 31) (N = 55)

10.2 (11.4)

12.0 (11.3)

22.1 (21.4)

Y10 (n = 55)

Radiographs from ‘12-year completers’ (n = 386)* Y0 (n = 64)

§ Status scores refer to all patients with an mSASSS evaluable at each time point, independent of the availability of a 2 year progression score mSASSS, modified Stoke Ankylosing Spondylitis Spine Score *Radiographs from patients with an mSASSS at year 12 evaluable (n = 386)

0.9 (1.9)

1.5 (2.9)

2.4 (4.1)

1.8 (3.6)

7.3 (9.7)

Total mSASSS

6.3 (9.3)

7.7 (9.7)

15.0 (16.7)

Y0-Y2 Y2-Y4 (N = 164) (N = 115)

5.6 (9.0)

Lumbar segment

6.5 (9.2)

12.8 (16.5)

PROGRESSION SCORES

6.0 (8.9)

11.6 (16.1)

Y8 (n = 66)

All radiographs (n = 809)

Y0 Y2 Y4 Y6 (n = 184) (n = 177) (n = 126) (n = 106)

Cervical segment

Total mSASSS

STATUS SCORES §

Table 1 – Status and progression based on the modified Stoke Ankylosing Spondylitis Spine Score

Radiographic damage over time

Chapter 7

152 (29) 161 (31) 121 (23) 58 (11) 63 (12)

≥2 mSASSS units (mean progression)¥

≥1 and <3 mSASSS units

≥3 mSASSS units

≥3 and <5 mSASSS units

≥5 mSASSS units

41 (22)

32 (17)

73 (39)

60 (32)

86 (46)

133 (72)

21 (31)

12 (18)

33 (49)

21 (31)

39 (57)

54 (79)

1 (1)

12 (18)

No of patients from the ’12 year completers’ during all 2 year intervals (maximum progression)* (n (%)) (n = 68)

mSASSS, modified Stoke Ankylosing Spondylitis Spine Score § Categories of progression are not mutually exclusive * Number and proportion of patients whose maximum 2 year progression score, during follow-up (up to 12 years), met the criterion of each of the rows ** This category of progression existed because of imputations, which made some status scores, and hence progression scores have decimal values ¥ Mean progression taking all 2 year intervals into account

282 (54)

8 (4)

45 (24)

204 (39) 0 (0)

No of patients during all 2 year intervals (maximum progression)* (n (%)) (n = 186)

No of intervals out of all 2 year intervals (n (%)) (n = 520)

≥1 mSASSS unit

>0 and <1 mSASSS units**

0 mSASSS units

Progression§

Table 2 – Progression of radiographic damage over time

41 (25)

16 (10)

42 (26)

20 (12)

36 (22)

78 (48)

11 (7)

75 (46)

No of patients during the first 2 year interval (Y0 Y2) (n (%)) (n = 164)

Radiographic damage over time

2 (0) 1 (0) 3 (1) 0 (0) 1 (0) 1 (0)

0 (0)

1 (0)

2 (0)

4 (1)

Reader 2 109 (63)

Reader 1

96 (55)

109 (71) 16 (10) 19 (12) 4 (3) 2(1) 2 (1) 0 (0) 0 (0) 0 (0) 0 (0) 1 (1) 0 (0)

141 (59) 48 (20) 28 (12) 9 (4) 6 (3) 2 (1) 1 (0) 2 (1) 2 (1) 1 (0) 0 (0) 0 (0)

43 (69)

Reader 1

0 (0)

1 (1)

0 (0)

2 (1)

5 (3)

18 (12)

23 (15)

102 (67)

Reader 2

Reader 2 51 (33)

Reader 1 44 (29)

Reader 1 46 (74)

First 2 year interval (Y0 - Y2) with at least 1 uninvolved VC * (n (%)) (n = 153)

Reader 1

First 2 year interval (Y0 - Y2) with at least 1 uninvolved VC * (n (%)) (n = 153)

Reader 2

All patients with at least 1 uninvolved VC and with mSASSS available at Y12 (n (%) (n = 62)

1 (0)

0 (0)

2 (1)

1 (0)

2 (1)

5 (2)

12 (5)

25 (10)

48 (20)

141 (59)

Reader 1

All intervals with at least 1 uninvolved VC * in ’12 year completers’** (n (%)) (n = 240)

*uninvolved VC is a VC without a previous syndesmophyte or bridge, at risk for developing a new syndesmophyte, according to both readers **Patients with mSASSS at year 12 available mSASSS, modified Stoke Ankylosing Spondylitis Spine Score; VC, vertebral corner

No of patients with new syndesmophytes

4 (1)

9 (2)

15 (3)

52 (11)

95 (20)

286 (61)

Reader 2

All patients with at least 1 uninvolved VC (N (%)) (n = 173)

23 (5) 10 (2)

56 (12)

75 (16)

292 (62)

Reader 1

No of new syndesmophytes

All intervals with at least 1 uninvolved VC* (n (%)) (n = 468)

Table 3 – Development of new syndesmophytes over time

the ’12-year completers’, similar results were obtained for the proportion of 2 year intervals with a new syndesmophyte, and in total 69% (R1) and 74% (R2) of patients developed at least one syndesmophyte over the 12 years. Taking the first 2 year interval into consideration, 29% (R1) and 33% (R2) of patients developed at least one new syndesmophyte. Radiographic progression over time

At the individual patient level, radiographic progression was highly variable. In figure 1, mSASSS is plotted as a function of follow-up time (figure 1A) and as a function of symptom duration (figure 1B). Variable progression rates are seen within and across patients, independent of follow-up time and symptom duration (and age - see online supplementary figure S2). This cohort included patients with very short symptom duration as well as patients

Figure 1 – Radiographic progression at the patient level. (A) Modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) per patient, plotted as a function of follow-up time. (B) mSASSS plotted as a function of duration since symptom onset

Chapter 7

with more than 40 years of symptom duration. This visual analysis shows that radiographic progression is not only seen in young patients with early disease but also in patients with longstanding disease. At the group level, however, a linear time function fitted the observed data best (figure 2). Time was positively associated with radiographic damage, at a rate of 0.98 mSASSS units/ year. Radiographic progression occurred significantly faster in men than in women (1.11 vs 0.69 mSASSS units/year), in HLA-B27 positive than in HLA-B27 negative patients (1.03 vs 0.70 mSASSS units/year) and in patients with a baseline mSASSSâ&#x2030;Ľ10 (which was the median value) compared with those with a baseline mSASSS<10 (1.44 vs 0.69 mSASSS units/year). HLA-B27 positive men (but not women) had a significantly higher progression than HLA-B27 negative men (1.18 vs 0.69 mSASSS units/year) (interaction of HLA-B27, gender and time: p=0.17).

Figure 2 - Radiographic progression at the group level. mSASSS, modified Stoke Ankylosing Spondylitis Spine Score.

HLA-B27 positive men with a baseline mSASSS â&#x2030;Ľ10 had a higher progression than HLA-B27 negative men with a baseline mSASSS <10 (1.56 vs 0.75 mSASSS units/year) (interactions p<0.04). This effect was not found in women. Progression occurred fully independently of disease and symptom duration (table 4). This table also shows that subgroups of women or HLA-B27 negative patients were small, as usually seen in AS cohorts. Progression was higher in patients that were ever exposed to TNFi compared with those who were not (1.54 vs 0.82 mSASSS units/year; interaction p=0.041). Progression was independent of treatment with NSAIDs during follow-up, spondylarthritis

Radiographic damage over time

Table 4 – Modelling of radiographic progression over time

Time (years)

Univariable regression analysis*β (95% CI)

p value**

N***

0.052

195

0.98 (0.84; 1.12)

Time X HLAB27

202

Progression rate if HLAB27 negative

0.70 (0.48; 0.90)

Progression rate if HLAB27 positive

1.03 (0.88; 1.19)

164

Time X gender

0.005

202

Progression rate in women

0.69 (0.57; 0.80)

Progression rate in men

1.11 (0.93; 1.29)

144

Time X HLAB27 in women

0.958

Time X HLAB27 in men

0.044

139

Progression rate in men HLAB27 negative

0.69 (0.34; 1.03)

Progression rate in men HLAB27 positive

1.18 (0.98; 1.38)

120

Time X Baseline mSASSS

<0.001

Progression rate in baseline mSASSS<10

0.69 (0.63; 0.75)

Progression rate in baseline mSASSS≥10

1.44 (1.18; 1.70)

Time X baseline mSASSS in women

124 59 <0.001

Progression rate in women with baseline mSASSS<10

0.49 (0.42; 0.56)

Progression rate in women with baseline mSASSS≥10

1.41 (0.97; 1.85)

Time X baseline mSASSS in men

183

55 44 11

<0.001

128

Progression rate in men with baseline mSASSS<10

0.81 (0.73; 0.90)

Progression rate in men with baseline mSASSS≥10

1.45 (1.16; 1.74)

Time X baseline mSASSS in HLAB27 negative

0.013

Progression rate in HLA-B27 negative with baseline mSASSS<10

0.75 (0.63; 0.86)

Progression rate in HLA-B27 negative with baseline mSASSS≥10

0.54 (-0.20; 1.28)

Time X baseline mSASSS in HLAB27 positive

28 22 6

<0.001

149

Progression rate in HLA-B27 positive with baseline mSASSS<10

0.68 (0.61; 0.75)

Progression rate in HLA-B27 positive with baseline mSASSS≥10

1.54 (1.27; 1.81)

Time X baseline mSASSS in men HLAB27 negative

0.011

Progression rate in HLA-B27 negative men with baseline mSASSS<10

0.80 (0.63; 0.98)

Progression rate in HLA-B27 negative men with baseline mSASSS≥10

0.45 (-0.40; 1.29)

Time X baseline mSASSS in men HLAB27 positive

<0.001

107

Progression rate in HLA-B27 positive men with baseline mSASSS<10

0.82 (0.72; 0.92)

Progression rate in HLA-B27 positive men with baseline mSASSS≥10

1.56 (1.25; 1.87)

Time X disease duration

0.701

Time X symptom duration

0.214

* Because the dependent variable was the mSASSS and the independent variable was time, the regression coefficient obtained reflects the progression of mSASSS per year ** p values are only shown for the interactions between time and other variables. In the presence of a significant interaction, results are stratified, and the progression in each of the subgroups presented through the regression coefficient. *** All patients with an mSASSS evaluable at each time point independently of the availability of a 2 year progression score were included in the models mSASSS, modified Stoke Ankylosing Spondylitis Spine Score

Chapter 7

related manifestations (psoriasis, inflammatory bowel disease and uveitis) and the presence of a positive spondyloarthritis family history. Progression was not different between the first and the second 6 years of follow-up.

DISCUSSION The present study, performed in a prevalence cohort of patients with AS, showed that long term radiographic progression in AS is more severe in HLA-B27 positive men, and –as expected- in patients with a higher level of radiographic damage present at baseline. Not less than 60% of all patients developed at least one new syndesmophyte over a period of up to 12 years. Radiographic progression follows an approximately linear course at the group level, with a remarkably stable progression rate of approximately 1 mSASSS unit per year when modelled by GEE. More progression was found in the cervical than in the lumbar spine, which is in line with what has been previously reported in an analysis of OASIS with limited follow-up 23.

The remarkably stable group progression of 1 mSASSS unit per year in AS covers up a highly variable -and to a large extent unpredictable- course of progression of radiographic damage in individual patients. Approximately 25% of patients showed no progression, 25% showed a high level of progression (arbitrarily defined as at least one 2 year interval with progression ≥ 5 mSASSS units) and the remaining patients showed progression rates of about 2 units/2 years. But more importantly, simple visualisation of individual progression curves plotted against duration of symptoms revealed that periods of steep progression and relative quiescence may alternate, and this process of oscillation may occur both in ‘early AS patients’ as well as in ‘advanced’ patients with ages greatly more than over 60 years. Likely the most important observation of this longitudinal study is that radiographic progression is entirely independent of disease or symptom duration. This observation has implications for clinical practice where a vested opinion says that AS may extinguish over time (which is now demonstrably false), as well as for the necessity of intervening with radiographic progression: 40 years with continuous slow progression or intervals with accelerated progression will sooner or later lead to a situation with clinically important radiographic damage 24. We have shown previously that mean progression in OASIS was 1 unit per 2 years while in the current analyses we found a mean progression of 1 unit per year. This at first contradictory result can be fully explained by the difference in reading with blinded time points in the previous analyses and known time points in the current analysis. We have described this effect of (un)blinding time order in a separate paper comparing 2 year progression scores in OASIS 17. However, the progression rate reported in this study (2 mSASSS units every 2

Radiographic damage over time

years) is in line with what has been previously reported in other studies. There are reports of 2 year mSASSS progression scores of approximately 1 mSASSS unit 4-6,11, 1.5 units 25, 2.5 units 14

and 2.6 units (extrapolation to a 2 year period of the annual progression rate of 1.3)

whereas for the same interval (first 2 year interval) we found a progression rate of 1.8 units.

The differences between the scores can be attributed to differences in selection of patients, baseline radiographic damage, reading conditions

or method of imputation of missing

VCs. Another study described radiographic damage over time, but only in 20 patients, using the Bath AS Radiology Index 26, and thus not allowing a formal comparison 27. Furthermore, the mSASSS has been shown to be the most appropriate measure to assess radiographic damage in patients with AS and thus our selection 13,28. The proportion of patients developing new syndesmophytes over time has not been reported in many studies. Poddubnyy et al reported 11% of patients with AS with at least one new syndesmophyte and Baraliakos et al reported 19% of patients, both within a 2 year period. In our study, we found 29-33% (according to each of the readers) of patients with at least one new syndesmophyte over the same period. These differences can also be explained by the varying presence of baseline syndesmophytes, in all studies the strongest variable to be associated with further progression of damage 11,12,14. The long term follow-up of our study allows us to conclude that 60% of all patients develop at least one new syndesmophyte over up to 12 years, implying that syndesmophytes will occur in the vast majority of patients and not only in those with severe radiographic progression. This study is unique in that it addresses the long-term ´natural´ course of radiographic progression. The vast majority of OASIS patients has never been exposed to TNFi. Remarkably, in those that have been exposed to TNFi, we found a higher progression rate than in patients that had never been exposed. Likely this is ‘confounding by indication’: the TNFi exposed patients would be those with more active disease and subsequently higher progression rates. Interestingly, NSAID use, which has been associated with impaired progression

7,8

did not have any impact on radiographic progression over time. Some limitations of the present study should be addressed. Over the 12 years, almost twothirds of the patients were lost to follow-up, and selection bias cannot be ruled out. We have presented our results for all patients as well as for the ’12 year completers’ and we have found them to be concordant in every analysis, thus making it unlikely that non-random attrition influenced our observations. OASIS is a prevalence cohort that, despite having a good representation of the spectrum of AS and of patients of varying ages, includes mainly patients with established AS with a relatively high level of baseline damage. Therefore, these results may not be generalisable to patients with short symptom duration. Further, women

Chapter 7

and HLA-B27 negative patients (especially in combination) do not occur in high numbers in AS cohorts, and estimates of radiographic progression in these subgroups are therefore rather unstable. Although interobserver reliability of the mSASSS was good, agreement for the individual VC scores between readers at a given time point was worse, mainly because of the long follow-up. However, in most of those cases, readers scored the same new syndesmophytes but at different time points, which is not necessarily the same as non-agreement: if a reader observes the subtle appearance of a new syndesmophyte in a time series of x-rays, he has to decide at which time point that syndesmophyte will first be scored as positive. This methodological phenomenon has precluded an analysis of new syndesmophytes agreed upon by both readers at the same time point. Such an approach would artificially and substantially reduce the number of new syndesmophytes over time. In summary, over a follow-up period of up to 12 years, progression of structural damage in AS patients is largely unpredictable, often alternating with periods of acceleration and

quiescence, and may occur in early as well as in very advanced patients with AS at older ages. About 60% of all patients developed at least one new syndesmophyte over a period of up to 12 years. Radiographic progression may be underestimated if the data are presented at the group level, as a linear average rate of 1 mSASSS unit per year. In HLA-B27 positive men in particular and in patients with a higher level of damage already present, physicians should be alert to future radiographic progression.

SUPPLEMENTARY DATA Supplementary data are published on the website of the Annals of the Rheumatic Diseases.

Radiographic damage over time

REFERENCES 1.

Zochling J, van der Heijde D, Burgos-Vargas R, et al. ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2006;65:442-52.

Braun J, van den Berg R, Baraliakos X, et al. 2010 update of the ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2011;70:896-904.

van der Heijde D, Landewe R, Baraliakos X, et al. Radiographic findings following two years of infliximab therapy in patients with ankylosing spondylitis. Arthritis Rheum 2008;58:3063-70.

van der Heijde D, Landewe R, Einstein S, et al. Radiographic progression of ankylosing spondylitis after up to two years of treatment with etanercept. Arthritis Rheum 2008;58:1324-31.

van der Heijde D, Salonen D, Weissman BN, et al. Assessment of radiographic progression in the spines of patients with ankylosing spondylitis treated with adalimumab for up to 2 years. Arthritis research & therapy 2009;11:R127.

Wanders A, Heijde D, Landewe R, et al. Nonsteroidal antiinflammatory drugs reduce radiographic progression in patients with ankylosing spondylitis: a randomized clinical trial. Arthritis Rheum 2005;52:1756-65.

Poddubnyy D, Rudwaleit M, Haibel H, et al. Effect of non-steroidal anti-inflammatory drugs on radiographic spinal progression in patients with axial spondyloarthritis: results from the German Spondyloarthritis Inception Cohort. Ann Rheum Dis 2012;71:1616-22.

Kroon F, Landewe R, Dougados M, et al. Continuous NSAID use reverts the effects of inflammation on radiographic progression in patients with ankylosing spondylitis. Ann Rheum Dis 2012;71:1623-9.

10.

Baraliakos X, Listing J, Rudwaleit M, et al. Progression of radiographic damage in patients with ankylosing spondylitis: defining the central role of syndesmophytes. Ann Rheum Dis 2007;66:910-5.

11.

12.

van Tubergen A, Ramiro S, van der Heijde D, et al.

Chapter 7

Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23. 13.

14.

Baraliakos X, Listing J, von der Recke A, et al. The natural course of radiographic progression in ankylosing spondylitis--evidence for major individual variations in a large proportion of patients. J Rheumatol 2009;36:997-1002.

15.

16.

Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9.

17.

Wanders A, Landewe R, Spoorenberg A, et al. Scoring of radiographic progression in randomised clinical trials in ankylosing spondylitis: a preference for paired reading order. Ann Rheum Dis 2004;63:1601-4.

18.

Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307-10.

19.

Bruynesteyn K, Boers M, Kostense P, et al. Deciding on progression of joint damage in paired films of individual patients: smallest detectable difference or change. Ann Rheum Dis 2005;64:179-82.

20. Twisk J. Applied longitudinal data analysis for epidemiology: a practical guide. Cambridge: Cambridge University Press; 2003. 21. Twisk JW. Longitudinal data analysis. A comparison between generalized estimating equations and random coefficient analysis. Eur J Epidemiol 2004;19:769-76. 22. Dougados M, Simon P, Braun J, et al. ASAS recommendations for collecting, analysing and reporting NSAID intake in clinical trials/ epidemiological studies in axial spondyloarthritis. Ann Rheum Dis 2011;70:249-51. 23. van Tubergen A, van der Heijde D, Dougados M, et al. Are syndesmophytes most prevalent in the lumbar or in the cervical spine in patients with ankylosing spondylitis and do they develop in a specific direction? Rheumatology 2012;51:1432-9. 24. Landewe R, Dougados M, Mielants H, et al.

Physical function in ankylosing spondylitis is independently determined by both disease activity and radiographic damage of the spine. Ann Rheum Dis 2009;68:863-7. 25. Baraliakos X, Haibel H, Listing J, et al. Continuous long-term anti-TNF therapy does not lead to an increase in the rate of new bone formation over 8 years in patients with ankylosing spondylitis. Ann Rheum Dis 2014;73:710-5. 26. MacKay K, Mack C, Brophy S, et al. The Bath Ankylosing Spondylitis Radiology Index (BASRI): a new, validated approach to disease assessment.

Arthritis Rheum 1998;41:2263-70. 27. Brophy S, Mackay K, Al-Saidi A, et al. The natural history of ankylosing spondylitis as defined by radiological progression. J Rheumatol 2002;29:1236-43. 28. Ramiro S, van Tubergen A, Stolwijk C, et al. Scoring radiographic progression in ankylosing spondylitis: should we use the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) or the Radiographic Ankylosing Spondylitis Spinal Score (RASSS)? Arthritis research & therapy 2013;15:R14.

Radiographic damage over time

Erosions and sclerosis on radiographs precede the subsequent development of syndesmophytes at the same site: a 12 year prospective follow-up of the OASIS study in patients with ankylosing spondylitis Sofia Ramiro, Astrid van Tubergen, Désirée van der Heijde, Carmen Stolwijk, Gideon Bookelman, Maxime Dougados, Filip van den Bosch, Robert Landewé

Arthritis Rheumatol 2014;66:2773-9

ABSTRACT Objective

To analyse whether erosions, sclerosis and squaring assessed on radiographs precede the development of syndesmophytes in patients with ankylosing spondylitis (AS). Methods

Patients with AS from the Outcome in Ankylosing Spondylitis International Study (OASIS) cohort were followed up for 12 years, with radiographs obtained every 2 years. Two readers (reader 1 and reader 2) scored radiographs according to the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) and recorded abnormalities per vertebral corner. Progression from erosions, sclerosis or squaring to (bridging) syndesmophytes was investigated using multilevel longitudinal (autoregressive and time lagged) regression analysis. Interactions with reader and spinal region were investigated. Results

The analysis included 211 patients (71% male, 85% HLA-B27 positive) with a mean age of 43 (SD 13) years and a symptom duration of 21 (SD 12) years. A total of 921 radiographs were included, with 20,509 (reader 1) and 20,568 (reader 2) vertebral corners evaluable. Erosions were found in 1% and 2.5%, sclerosis in 0.3% and 1.7%, squaring in 6.5% and 5.1%, and syndesmophytes in 25% and 27%, by reader 1 and reader 2, respectively. The odds ratio for a new syndesmophyte to occur within 2 years at the same vertebral corner if erosions, sclerosis, or squaring were present was 2.0 [95% confidence interval (95% CI) 1.7-2.3] for the cervical and lumbar spine together, 3.1 (95% CI 2.5-3.9) for the cervical spine, and 1.3 (95% CI 1.0-1.6) for the lumbar spine. When vertebral corners with erosions, sclerosis, and squaring were analysed separately, this effect was statistically significant for erosions and for sclerosis, but not for squaring. Conclusion

Erosions and sclerosis occur infrequently in patients with AS, but when they occur, they

precede the development of a new syndesmophyte, and are therefore important to assess.

Chapter 8

INTRODUCTION Ankylosing spondylitis (AS) is known for the development of new bone, forming syndesmophytes in the spine. Syndesmophytes most importantly contribute to radiographic progression, and it is their presence that forecasts further damage. Syndesmophytes are associated with impairment in mobility, worse function, and impairment in quality of life 1. In addition, the pathophysiology of syndesmophyte formation is incompletely understood. Hence, there is an important need to gain further insight into the development of new syndesmophytes. Radiographic damage is one of the core outcomes recommended by the Assessment of SpondyloArthritis international Society (ASAS) for the follow-up of patients with AS 2, and the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) has been shown to be the most appropriate measure to quantify radiographic damage in patients with AS

. For

3,4

this method, the anterior vertebral corners of the cervical and lumbar spine are scored for the presence of syndesmophytes (bridging or not) and erosions, sclerosis or squaring 3. To date, insight into the presence of erosions, sclerosis, or squaring and their contribution to the development of new syndesmophytes is limited. Some authors have proposed to exclude

erosions from the mSASSS because of their rare occurrence 5,6, but it is still unknown whether or not erosions, sclerosis, or squaring play a role in the development of syndesmophytes. The aim of the present study was to analyse the presence of erosions, sclerosis, and squaring and evaluate if they precede the development of new syndesmophytes in 2 year intervals.

METHODS Study population

For this study, data from the Outcome in Ankylosing Spondylitis International Study (OASIS)

were used. This is a prevalence cohort including 217 consecutive patients with AS from The Netherlands, Belgium and France that started in 1996 7. Clinical data were collected every 6 months until year 2, then yearly until year 4 and every 2 years thereafter. Cervical and lumbar spine radiographs were obtained every 2 years during a period of 12 years, with a total of 7 possible time points per patient. For the present study, patients were included if they had at least one radiograph available. The ethics committees from all participating hospitals approved the study, and informed consent was obtained from all patients. Scoring methods

Radiographs were scored using the mSASSS 3. According to this method, the anterior

vertebral corners of the cervical (lower border of C2 to upper border of T1) and lumbar

Erosions, sclerosis and syndesmophytes

(lower border of T12 to upper border of S1) segments (total of 24 vertebral corners) are scored at a lateral view only, for the presence of an erosion and/or sclerosis and/or squaring (1 point), syndesmophyte (2 points), and bridging syndesmophyte (3 points). Squaring was considered to be present when a piece of paper was held in the anterior face of the vertebra and more than 50% of the anterior face would be touching the paper. Readers were trained to only score squaring in the cervical spine when they were certain about it, as a somewhat squared configuration of cervical vertebrae may also be seen in normal individuals. Syndesmophytes were scored if the angle between the newly formed bone and a horizontal line extending from the vertebral surface (upper or lower) was >45ยบ 8. The total possible score per patient ranges from 0 to 72 3. Two trained readers (SR and CS) (reader 1 and reader 2) who independently scored all available radiographs per patient, were blinded with regard to demographic and clinical data, but not with regard to chronology, since this is the most sensitive method 9. The readers separately recorded erosions, sclerosis, and squaring (or any combination of the 3) per vertebral corner. Readers were trained by experts in the field and over multiple training sessions. Scoring of the radiographs started only after a good agreement between readers could be assured. Radiographic abnormalities

For this study, the individual scores for each of the vertebral corners were used.

Syndesmophytes were defined as bridging or not, i.e., a score of 2 or 3 in the mSASSS per vertebral corner. Status scores (the score at each of the available time points) and change scores (the difference between the status scores of 2 subsequent time points) were calculated. For the change scores, erosions, sclerosis, and squaring were combined, so that the change between the individual scores of the mSASSS could be obtained, as follows: no abnormality to erosions, sclerosis, or squaring (from score 0 to 1), no abnormality to (bridging) syndesmophyte (from score 0 to either 2 or 3), and erosions, sclerosis, or squaring to (bridging) syndesmophyte (from score 1 to either 2 or 3). In order to analyse the temporal association between new syndesmophytes and erosions, sclerosis, or squaring, and if combinations of erosions, sclerosis, or squaring were found (i.e. any combination of erosions and/or sclerosis and/or squaring) in one vertebral corner, each abnormality was considered separately.

Statistical analysis

Frequencies of the different radiographic changes at the vertebral corner level were computed.

In order to investigate whether a new syndesmophyte developed from erosions, sclerosis, or squaring at the same vertebral corner, a multilevel analysis was conducted. Autoregressive 100

Chapter 8

generalised estimating equations (GEE) were applied making use of all longitudinal data 10

. Autoregressive analysis can be considered equivalent to analysis of change scores per

interval, in this case the development of new syndesmophytes, per time interval. The unit of analysis was the vertebral corner, and the levels were ‘time point’, ‘spinal region,’ (cervical or lumbar), and ‘reader’. Models were time lagged, i.e., the presence of erosions, sclerosis, or squaring at one time point and the subsequent development of syndesmophytes 2 years later were assessed. Also, models with longer time lags (i.e., 4 and 6 years) were tested. Interactions with reader and spinal region were investigated, and if a relevant interaction (set at P < 0.1) was found, models were run in subgroups. GEE adjusts for the within-subject correlation and requires an a priori defined ´working´ correlation structure. In this study the ´exchangeable´ correlation structure was found to be appropriate and was chosen. In order to trace potential bias due to loss of follow-up, analysis was performed with observed data and with missing values replaced by last observation carried forward (LOCF). In case of LOCF, both the outcome and the predictor radiographic damage scores can only underestimate the true progression. A comprehensive set of sensitivity analyses was performed. First, the analysis was repeated

on vertebral corners in which both readers agreed with respect to the development of a syndesmophyte during the follow-up period, independently of the time point in which this was scored. Second, the association between concordantly scored (when both readers see the same abnormality) erosions, sclerosis, or squaring on the one hand, and concordantly scored syndesmophytes on the other hand was investigated in different manners, as follows: concordant erosions, sclerosis, or squaring versus ‘any’ (i.e., at least one of the readers sees an abnormality) syndesmophytes; concordant erosions, sclerosis, or squaring versus concordant syndemophytes; and any erosions, sclerosis, or squaring versus concordant syndemophytes. In this latter analysis, the factor ‘time’ was omitted as a variable, and the analysis was performed using GEE on vertebral corners with adjustment for patient correlation (multilevel approach). Stata SE version 12 (StataCorp) was used to perform the analyses.

RESULTS Patient characteristics

In total, 211 patients with at least 1 radiograph were included in this study. Of these patients, 71% were men, and 85% were HLA-B27 positive. Their mean age was 43 (SD 13) years. The mean time since symptom onset was 21 (SD 12) years and time since diagnosis was 12 (SD 9) years. During follow-up, which was on average 7.8 (SD 4.1) years per patient, 921 radiographs were obtained (205 at baseline, 199 at year 2, 141 at year 4, 127 at year 6, 78 at Erosions, sclerosis and syndesmophytes

101

year 8, 92 at year 10, and 79 at year 12), giving a mean number of 4.4 (SD 1.8) radiographs per patient. At least one 2 year change score was available in 207 patients (98%).

Evaluation of vertebral corners

A total of 20,509 vertebral corners (reader 1) and 20,568 vertebral corners (reader 2) were evaluable. Of these, erosions were scored in 1% by reader 1 (in 2.5% by reader 2), sclerosis in 0.3% by reader 1 (in 1.7% by reader 2), squaring in 6.5% by reader 1 (in 5.1% by reader 2), and (bridging) syndesmophytes in 25% by reader 1 (in 27% by reader 2) (table 1). In 96% (95% for reader 2) of all vertebral corners over all 2 year intervals, no change in the vertebral corner score was found. In 3% (2% for reader 2) of the vertebral corners over the intervals, a (bridging) syndesmophyte developed from a normal vertebral corner and in 1% (both readers) of the vertebral corners over the intervals from erosions, sclerosis, or squaring. Erosions, sclerosis, or squaring or any combination developed from a normal vertebral corner in 1% (both readers) of the vertebral corners over the 2 year intervals (table 2). Examples of the development of syndesmophytes from erosions, sclerosis, or squaring are shown in Supplementary Figures 1 and 2, available on the Arthritis & Rheumatology web site at http:// onlinelibrary.wiley.com/doi/10.1002/art.38775/abstract.

Development of syndesmophytes from vertebral corners with erosions, sclerosis and/ or squaring

In the multivariable and longitudinal analysis, the presence of erosions, sclerosis, and/or squaring conveyed, compared with a normal vertebral corner, an increased risk, with an odds ratio (OR) of 2.0 (95% confidence interval [95%CI] 1.7-2.3), for the development of a new syndesmophyte over a 2-year period (table 3). There was a significant interaction between erosions, sclerosis, and/or squaring and spinal region, with a higher OR for the cervical spine (OR 3.1 [95%CI 2.5-3.9]) than for the lumbar spine (OR 1.3 [95%CI 1.0-1.6]). There was no significant interaction with reader. When analysed at the level of the individual feature erosions, sclerosis, or squaring, the presence of erosions (OR 2.1 [95%CI 1.6-2.8]) and the presence of sclerosis (OR 5.1 [95%CI 3.9-6.8]), but not the presence of squaring (OR 1.3 [95%CI 1.0-1.6]) showed a statistically significant OR for new syndesmophytes. An interaction with spinal region was found for squaring only. In the cervical spine, squaring was significantly associated with new syndesmophytes (OR 8.5 [95%CI 2.1-34.7]), but this was not found in the lumbar spine (OR 1.1 [95%CI 0.8-1.4]). The models were also constructed with the LOCF imputed data, and the results were very 102

Chapter 8

Bridging syndesmophyte

*Values are the number (%).

1,782 (9)

3,288 (16)

Syndesmophyte

0 (0)

Erosions, sclerosis and squaring 3,090 (15)

2,541 (12)

2 (0)

40 (0)

22 (0)

6 (0) 21 (0)

Erosions and squaring

Sclerosis and squaring

13 (0)

1039 (5)

355 (2)

0 (0)

1327 (7)

Squaring only

Erosions and sclerosis

55 (0)

Sclerosis only

513 (3)

12,953 (63)

13,820 (67) 210 (1)

Normal

Erosions only

Reader 2 (n = 20,568)

Reader 1 (n = 20,509)

All vertebral corners

0 (0)

0 (0) 1,681 (17) 1,374 (14)

0 (0) 1,142 (12) 1,566 (16)

1,722 (16)

640 (6)

6 (0) 21 (0)

0 (0)

1,716 (16)

860 (8)

2 (0)

40 (0)

18 (0)

0 (0)

1,037 (10)

1,324 (12)

33 (0)

6,992 (64)

2 (0) 4 (0)

0 (0)

4 (0)

7,104 (66)

13 (0)

185 (2)

31 (0) 3 (0)

480 (5)

206 (2)

Reader 2 (n = 10,868)

170 (2)

5,961 (61)

6,717 (70)

Reader 1 (n = 10,845)

Lumbar spine

24 (0)

Reader 2 (n = 9,700)

Reader 1 (n = 9,664)

Cervical spine

Table 1 â&#x20AC;&#x201C; Descriptive analysis of all vertebral corners according to the individual items included in the modified Stoke Ankylosing Spondylitis Spine Score over the entire followup period (up to 12 years)*

Erosions, sclerosis and syndesmophytes

103

Table 2 â&#x20AC;&#x201C; Change in radiographic abnormalities over time* Change**

Reader 1

Reader 2

Observed No change

13,182 (96)

13,160 (95)

No abnormality to erosions, sclerosis or squaring

92 (1)

197 (1)

No abnormality to (bridging) syndesmophyte

373 (3)

335 (2)

Erosions, sclerosis or squaring to (bridging) syndesmophyte

128 (1)

115 (1)

28,035 (98)

28,101 (97)

97 (0)

223 (1)

LOCF No change No abnormality to erosions, sclerosis or squaring No abnormality to (bridging) syndesmophyte

487 (2)

395 (1)

Erosions, sclerosis or squaring to (bridging) syndesmophyte

139 (1)

136 (1)

* Radiographs were from up to 12 years of follow-up for each 2 year interval and were scored by 2 readers. Analyses were conducted with observed data and with missing data imputed with last observation carried forward (LOCF). Values are the number (%). ** No abnormality to erosions, sclerosis or squaring represents a change in the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS), at the vertebral corner level, from 0 to 1. No abnormality to (bridging) syndesmophyte represents a change in the mSASSS from 0 to either 2 or 3. Erosions, sclerosis or squaring to (bridging) syndesmophyte represents a change in the mSASSS of 1 to either 2 or 3.

similar: OR 1.9 (95%CI 1.6-2.3) for erosions, sclerosis, and/or squaring, 2.0 (95%CI 1.5-2.6) for erosions, 4.8 (95%CI 3.7-6.2) for sclerosis, and 1.2 (95%CI 0.9-1.5) for squaring. Table 3 also shows the results of the longitudinal analysis when the time lag was increased to 4 years and 6 years, respectively. The strength of the association decreased as the time lag increased for erosions, sclerosis, and/or squaring taken together, for erosions alone and, to some extent, for sclerosis alone. For erosions, sclerosis, and/or squaring, the OR for a time lag of 2 years was 2.0, decreasing to 1.8 for a 4-year lag and to 1.6 for a 6-year lag. An association between squaring and syndesmophyte formation was not confirmed. A sensitivity analysis including only those syndesmophytes that were scored concordantly by both readers yielded similar results (see Supplementary table 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38775/abstract). A subsequent set of sensitivity analyses assessing the effects of concordantly scored erosions, sclerosis, or squaring on concordantly scored syndesmophytes (in comparison with several scenarios in which at least one of the readers scored the abnormality) also confirmed the results of the main analysis (see Supplementary table 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38775/ abstract).

104

Chapter 8

Erosions, sclerosis and syndesmophytes

3.9 (2.8 - 5.5)

1.1 (0.8 – 1.4) 1.6 (1.2 - 2.1) 1.0 (0.6 - 1.4)

- Lumbar spine

- Reader 1

- Reader 2

1.3 (1.1 - 1.6)

4.2 (3.2 - 5.4)

1.3 (1.0 - 1.5)

4.2 (3.2 - 5.5)

1.4 (1.1 - 1.9)

1.6 (1.4 - 1.8)

1.8 (1.5 - 2.1)

1.7 (1.4 - 2.2)

Same model, with a time lag of 6 years in the main predictor

Same model, with a time lag of 4 years in the main predictor

* Values are the odds ratio (OR) (95% confidence interval [95% CI]) for the main predictor of the model. All models are longitudinal, autoregressive (i.e., adjusted for the presence of syndesmophytes seen on the previous radiograph, obtained 2 years before), and time lagged (i.e., the effect of the predictor - erosions, sclerosis, squaring - corresponds to one previous radiograph, obtained 2 years before the outcome). All models are adjusted for time (in years), reader (2 readers) and spinal region (cervical versus lumbar), except for cases in which a positive interaction with these variables was found and results of stratification are presented, as indicated. GEE = generalised estimating equations; mSASSS = modified Stoke Ankylosing Spondylitis Spine Score.

8.5 (2.1 – 34.7)

- Cervical spine

Factor: Squaring

15.3 (9.2 - 25.5)

No significant interaction

- Reader 2 1.3 (1.0 – 1.6)

5.1 (3.9 – 6.8)

2.1 (1.6 – 2.8)

No significant interaction

- Reader 1

- Lumbar spine

- Cervical spine

Factor: Sclerosis

- Reader 2

- Reader 1

- Lumbar spine

- Cervical spine

Factor: Erosions

- Reader 2

- Reader 1

3.1 (2.5 – 3.9) 1.3 (1.0 – 1.6)

Same model, but with results stratified for spinal region or reader if significant interactions

- Lumbar spine

2.0 (1.7 – 2.3)

Multivariable GEE models with syndesmophytes as the outcome

- Cervical spine

Factor: Erosions, sclerosis and/or squaring (score of 1 in mSASSS)

Stratification

Table 3 - Longitudinal effect of erosions, sclerosis and squaring on the development of new syndesmophytes over 12 years*

105

DISCUSSION The present study shows that erosions, sclerosis, and squaring occur infrequently in patients with AS, representing in total 8-10% of the scores in the vertebral corners (according to both readers). However, if present, they often precede the development of a new syndesmophyte (OR of 2 compared with a normal vertebral corner). This association is statistically significant for erosions (OR 2) and sclerosis (OR 5), but not for squaring (OR 1.3). These findings do not endorse the proposals from Baraliakos et al 5 and Maksymowych 6 to remove erosions from the mSASSS due to their rare occurrence. We confirm that erosions are infrequent (occurring in 1-3% of the vertebral corners), but when they are present, it is more likely, compared to a normal vertebral corner, that a syndesmophyte will develop within 2 years (OR 2.1). The results with regard to squaring were at first glance more ambiguous than those for erosions and sclerosis, because we found a strong, significant effect of squaring on syndesmophyte formation in the cervical spine (OR 8.5). However, these findings resulted from only 2 and 3 vertebral corners in which squaring was scored by reader 1 and reader 2, respectively (of > 9600 cervical vertebral corners) and are therefore not reliable. Our data confirm that squaring in the cervical spine is difficult to assess and very infrequent 11,12. However, in these analyses we only investigated the effect of squaring on the development of syndesmophytes, and we do not know whether squaring is related to other outcome measures. Furthermore, before deciding on removing any feature from the mSASSS, all aspects of validity of the instrument should be tested. This was not the objective of present analysis. Our aim, rather, was to investigate whether erosions, sclerosis, or squaring were predictive of the development of syndesmophytes. Therefore, in contrast to an earlier proposal to remove squaring from the cervical spine from the mSASSS, we believe that the mSASSS should remain as it has been validated, since it has been proven to be the most adequate method for quantitative assessment of structural damage in patients with AS 3-5. The longitudinal character of the analysis allows an interpretation of the importance of time (e.g., time lags), while within-patient correlation is appropriately adjusted for, and the multilevel type of analysis allows assessment at the level of the vertebral corner. Both types of adjustments add to the credibility of the results. The time lag analysis unequivocally demonstrates that erosions, sclerosis, and/or squaring (especially erosions and sclerosis) are most closely associated with the development of new syndesmophytes if they occur in close temporal relation to the new syndesmophytes, and that the relationship becomes weaker with increasing time lag. This can be seen as supportive of a predictive association.

106

Chapter 8

To our knowledge, this is the first study to evaluate the role of abnormalities scored as a 1 in the mSASSS, and our findings help provide insight into the process of development of new syndesmophytes. The fact that erosions and sclerosis led to the formation of syndesmophytes may implicate these changes as part of the development process of syndesmophytes. This is consistent with current theories of the relationship between inflammation and new bone formation in AS. Sieper et al

proposed that structural damage occurs in two steps: first,

inflammation causes erosive structural damage, and then tissue repair takes place, with subsequent ossification. According to Schett et al 14, a mechanical or inflammatory trigger leads to an inflammatory response and afterward a repair response starts, perpetuating a bone anabolic response

. It seems that any ongoing new bone formation may not be

stopped. As shown in this study, erosions and sclerosis are part of this ongoing process, and sclerosis seems to be the first step in the bone proliferation process. A parallelism can be established between these findings and the relationship between fatty lesions in magnetic resonance imaging (MRI) and syndesmophytes 15. The large majority of new syndesmophytes develop from vertebral corners without inflammation seen on MRI, but a vertebral corner with inflammation is predictive of the development of a new syndesmophyte. This is also what was found in the present study: the majority of syndesmophytes develop

in previously uninvolved vertebral corners, but the presence of erosions, sclerosis, or squaring is predictive for the development of a new syndesmophyte. It is also possible that inflammation and its consequences, in this case erosions or sclerosis, always precede the development of a syndesmophyte, but that the 2 year interval between the radiographs was too long to capture all the steps of the process. Some limitations of this study need to be addressed. Readers were aware of the chronological order of the radiographs and could be alert to identifying a syndesmophyte after having scored erosions, sclerosis, or squaring. In other words, the dependent variable (new bone formation) was not assessed independently of the predictor variable. This feature may have influenced the results to some extent. However, readers were not explicitly trained to be alert to syndesmophytes under the above-described conditions, and they were unaware of the research question of the present study when scoring. Moreover, no substantial differences were found between the readers with regard to scoring. Analyses were performed with the scores of the readers considered independently (which focuses on highest possible sensitivity), as well as with concordant reader scores (which focuses on the highest possible specificity). We found similar results in 2 independently obtained data sets (reader 1 and reader 2), as well as in the analysis of concordant findings. The analysis of concordant observations, presented here as sensitivity analysis, yielded essentially similar ORs for erosions, sclerosis, or squaring versus syndesmophytes, which is supportive of the main analysis and indicates that reader variation is not responsible for the associations. The entire Erosions, sclerosis and syndesmophytes

107

set of analyses adds to the credibility and robustness of the data. Still, it appears very difficult to obtain high-level agreement with regard to subtle radiographic abnormalities such as syndesmophytes. A substantial amount of disagreement is inherent in the long follow-up in combination with chronological scoring. In most cases, readers did score the same new syndesmophyte, but they assigned the new syndesmophyte to different time points. This is not necessarily the same as nonagreement, but will artificially reduce agreement in change scores when calculated over the same time intervals. Furthermore, the validated instrument used in the present study is the mSASSS, and reliability should be evaluated for the whole instrument. Reliability of the data used from OASIS was found to be good for both status scores and progression scores 16. Lastly, radiographs were obtained every 2 years, and in this long interval some steps in the sequence occurring until bone formation might be missed (e.g., formation of an erosion or sclerosis). The results might be different in another cohort, especially if radiographs were analysed more frequently. In summary, erosions and sclerosis often precede the development of syndesmophytes. This finding fuels the hypothesis that the development of structural damage in AS starts with osteodestruction, followed by osteoproliferation.

SUPPLEMENTARY DATA Supplementary data are published on the website of Arthritis & Rheumatology.

108

Chapter 8

REFERENCES 1.

Machado P, Landewe R, Braun J, et al. A stratified model for health outcomes in ankylosing spondylitis. Ann Rheum Dis 2011;70:1758-64.

Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9.

Maksymowych WP. Controversies in conventional radiography in spondyloarthritis. Best practice & research Clinical rheumatology 2012;26:839-52.

Baraliakos X, Listing J, Rudwaleit M, et al. Progression of radiographic damage in patients with ankylosing spondylitis: defining the central role of syndesmophytes. Ann Rheum Dis 2007;66:910-5.

10.

Twisk J. Applied longitudinal data analysis for epidemiology: a practical guide. Cambridge: Cambridge University Press; 2003.

11.

Kim TJ, Kim HS, Joo KB, et al. Do we really need to evaluate entire cervical spines for squaring score in modified stoke ankylosing spondylitis spinal score? J Rheumatol 2008;35:477-9.

12.

Ward MM, Learch TJ, Weisman MH. Cervical vertebral squaring in patients without spondyloarthritis. J Rheumatol 2012;39:1900.

13.

Sieper J, Appel H, Braun J, et al. Critical appraisal of assessment of structural damage in ankylosing spondylitis: implications for treatment outcomes. Arthritis Rheum 2008;58:649-56.

14.

Schett G, Rudwaleit M. Can we stop progression of ankylosing spondylitis? Best practice & research Clinical rheumatology 2010;24:363-71.

15.

van der Heijde D, Machado P, Braun J, et al. MRI inflammation at the vertebral unit only marginally predicts new syndesmophyte formation: a multilevel analysis in patients with ankylosing spondylitis. Ann Rheum Dis 2012;71:369-73.

16.

Ramiro S, Stolwijk C, van Tubergen A, et al. Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study. Ann Rheum Dis 2015;74:52-9.

Erosions, sclerosis and syndesmophytes

109

Higher disease activity leads to more structural damage in the spine in ankylosing spondylitis: 12 year longitudinal data from the OASIS cohort Sofia Ramiro, Désirée van der Heijde, Astrid van Tubergen, Carmen Stolwijk, Maxime Dougados, Filip van den Bosch, Robert Landewé

Ann Rheum Dis 2014;73:1455-61.

ABSTRACT Objectives

To analyse the long-term relationship between disease activity and radiographic damage in the spine in patients with ankylosing spondylitis (AS). Methods

Patients from the Outcome in AS International Study (OASIS) were followed-up for 12 years, with 2 yearly clinical and radiographic assessments. Two readers independently scored the x-rays according to the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS). Disease activity measures include the Bath AS Disease Activity Index (BASDAI), AS Disease Activity Index (ASDAS)-C-reactive protein (CRP), CRP, erythrocyte sedimentation rate (ESR), patient’s global assessment and spinal pain. The relationship between disease activity measures and radiographic damage was investigated using longitudinal, autoregressive models with 2 year time lags. Results

184 patients were included (70% males, 83% HLA-B27 positive, mean (SD) age 43 (12) years, 20 (12) years symptom duration). Disease activity measures were significantly longitudinally associated with radiographic progression. Neither medication nor the presence of extraarticular manifestations confounded this relationship. The models with ASDAS as disease activity measure fitted the data better than models with BASDAI, CRP or BASDAI+CRP. An increase of one ASDAS unit led to an increase of 0.72 mSASSS units/2 years. A ‘very high disease activity state’ (ie, ASDAS>3.5) compared with ‘inactive disease’ (ie, ASDAS<1.3) resulted in an additional 2 year progression of 2.31 mSASSS units. The effect of ASDAS on mSASSS was higher in males versus females (0.98 vs -0.06 mSASSS units per ASDAS unit) and in patients with <18 years vs ≥18 years symptom duration (0.84 vs 0.16 mSASSS units per ASDAS unit). Conclusion

This is the first study showing that disease activity contributes longitudinally to radiographic progression in the spine in AS. This effect is more pronounced in men and in the earlier phases of the disease.

112

Chapter 9

INTRODUCTION The relationship between inflammation and new bone formation in ankylosing spondylitis (AS) has been extensively debated over the past couple of years

. Studies with tumour

1,2

necrosis factor inhibitors (TNFi), which have profound effects on disease activity in AS, have failed to document inhibition of radiographic damage progression, thus challenging the dogma in rheumatology that inflammation causes damage 3-5. Existing data are inconclusive: In several observational studies up to 4 years duration, the acute phase reactants (APR) C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) have found to be somehow associated with radiographic progression in AS 6-9, but the patient-reported outcome (PRO) Bath AS Disease Activity Index (BASDAI)) has not 6,7. These studies, either positive or negative, have been cross-sectional in nature, implying that they at best provide evidence for a generic association between disease activity and radiographic progression at the group level. Only a true longitudinal analysis, however, will provide insight into whether an increase in disease activity in a patient will be followed by an increase in radiographic progression in that patient. Several reasons may explain why associations between disease activity and radiographic progression are not easily demonstrated: First, radiographic progression in AS is slow and

is difficult to measure. Patients have to be followed- up for a long period to detect sufficient progression of damage, while measurement error often jeopardises a reliable progression signal. Second, thus far disease activity in these analyses has either been measured with PROs (eg, BASDAI) or with APR (eg, CRP), providing disparate results. The recently developed AS Disease Activity Index (ASDAS) combines PROs and APR and has demonstrably better psychometric properties than the BASDAI and the CRP 10,11. ASDAS is therefore an attractive measure for disease activity in sophisticated longitudinal analyses. Third, associations of interest have usually been applied in databases of modest patient numbers and short followup duration, while usually only baseline and endpoint radiographic scores were exploited, making conventional analyses less robust. A cohort with AS patients with long-term followup and multiple repeated outcome observations will allow a true longitudinal analysis and provide more statistical power, thus creating more robust results. In the past, we have been able to demonstrate a true longitudinal relationship between disease activity and radiographic progression in rheumatoid arthritis

. Using similar

methodology, we here present a true longitudinal analysis in a 12 year follow-up cohort of patients with AS. The aim was to investigate the longitudinal relationship between disease activity measured by various assessments and radiographic progression in patients with AS.

Disease activity and structural damage

113

METHODS Study population

For this study, data from the Outcome in Ankylosing Spondylitis International Study (OASIS)

were used. OASIS is a prevalence cohort including 217 consecutive patients with AS from The Netherlands, Belgium and France, which has started in 1996 13. Clinical and radiographic (cervical and lumbar spine) data were collected every 2 years during 12 years. Patients were included in the present study if they had at least two sets of radiographs and data on disease activity available. All patients have given informed consent, and the ethics committees from all participating hospitals have approved the study. Radiographic damage

Radiographs were scored using the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) 14. According to this method, the anterior vertebral corners (VCs) of the cervical and lumbar segments (total of 24 VCs) are scored on a lateral view only for the presence of an erosion and/or sclerosis and/or squaring (one point), syndesmophyte (two points) and bridging syndesmophyte (three points). The total score ranges from 0 to 72 14. Two well-trained readers (SR and CS) have independently scored all available radiographs per patient, blinded to demographic and clinical data, but with known chronology, as this is the most sensitive method 15. Average scores of the two readers were used. A total of 809 radiographs (184 at baseline; 177 at year 2; 126 at year 4; 106 at year 6; 66 at year 8; 82 at year 10; and 68 at year 12) were obtained in which the mSASSS could be determined. Details of the radiograph readings, imputation of missing VCs and reliability have been reported elsewhere 16. Disease activity

Disease activity measures include the BASDAI (0-10) 17, the ASDAS (which was calculated using the CRP (mg/L)) 11, CRP, ESR, patient’s global assessment of disease activity (0-10) and spinal pain (0-10). Disease activity states were computed using the ASDAS cut-off levels

: inactive disease (ASDAS <1.3), moderate disease activity (1.3 ≤ASDAS <2.1),

high disease activity (2.1 ≤ASDAS ≤3.5) and very high disease activity (ASDAS >3.5). In analogy, disease activity states based on BASDAI were computed, and because there are no well-defined cut-offs, we chose the following: BASDAI <2, 2 ≤BASDAI <4, 4 ≤BASDAI ≤6 and BASDAI >6. Other potentially contributing factors

Factors potentially influencing the relationship between disease activity and radiographic

damage were also taken into account. These included symptom duration, gender, HLA-B27 114

Chapter 9

status, baseline mSASSS, extra-articular manifestations (EAM) and treatment. Information on the presence of EAM, that is, uveitis, psoriasis or inflammatory bowel disease, was collected at every visit, and each EAM was considered to be persistent after its first presentation. Non-steroidal anti-inflammatory drugs (NSAIDs), disease modifying antirheumatic drugs (DMARDs) and/or TNFi were recorded as dichotomous variables (yes/no) at every visit. NSAID intake was also computed, for every 2 year interval, according to the NSAID index proposed by the Assessment of SpondyloArthritis international Society (ASAS)

, which

takes into account the type of drug, dose and duration of treatment, varying from 0 to 100 (100 means daily NSAID intake in full dose). Modelling and statistical analysis

The relationship between disease activity measures and radiographic damage was investigated using generalised estimating equations (GEE), which is suitable to elucidate longitudinal relationships, and makes use of all available data

. GEE corrects for the

20,21

within-subject correlation, and for this it requires an a priori defined ´working´ correlation structure. On the basis of the covariance matrix, the ´exchangeable´ correlation structure was most appropriate because the correlations of the outcome at different time points were approximately equal (Pearson correlation coefficients between 0.80 and 0.99). In a first set of analyses (baseline analysis), we have investigated using GEE whether disease

activity determined at baseline was associated with the development of radiographic damage over time (disease activity x time interaction). In a second set of analyses (longitudinal analysis), the longitudinal relationship between disease activity measures and mSASSS over time was analysed in autoregressive marginal models with 2 year time lags using GEE. Autoregressive analysis adjusts a variable’s absolute value at a particular time point (t) (here: mSASSSt) for the variable’s absolute value at the

previous time point (t-1) (here: mSASSSt-1) and can be considered synonymous to modelling

progression over time. A ‘2 year time lag’ here means that a disease activity measure at the start of a 2 year interval (eg, ASDASt) was related to radiographic progression during the

consecutive 2 year interval (here: mSASSSt was modelled by ASDASt-1 and by mSASSSt-1). Interactions were tested and if proven significant (here p<0.1), analyses were repeated in

subgroups. The following longitudinal models that differed with regard to the disease activity measure were run: (1) with ASDAS as a continuous measure; (2) with ASDAS categories; (3) with BASDAI as a continuous measure and CRP; (4) with BASDAI categories and CRP; (5) with patient global as a continuous measure and CRP; and (6) with spinal pain as a continuous measure and CRP. Disease activity and structural damage

115

The main model included only variables that were either (1) associated with mSASSS or (2) confounded the relationship between disease activity and radiographic progression. A sensitivity analysis was performed including all variables that, from a clinical perspective, could be contributory to explaining the relationship between ASDAS and mSASSS. Goodness-of-fit statistics (quasi-likelihood under the independence model criterion (QIC)) were used to get an impression about ‘how good the model fits the data’, with lower QICs reflecting better data fit. Analyses were done using Stata SE V. 12 (Statacorp, College Station, Texas, USA).

RESULTS In total, 184 patients (70% males, 83% HLA-B27-positive) with at least one 2 year mSASSS interval and disease activity measures at the start of that interval available were included in the analyses (table 1). These patients had similar baseline characteristics as those included in the entire OASIS cohort (see online supplementary table S1). Patients that were followedup until year 12 were not different from those initially included in the study (see online supplementary table S1).

Table 1 - Baseline demographic, clinical and radiographic characteristics Assessment Age [years] Male gender (%) HLA-B27 positive (%) Symptoms duration [years] Disease duration [years] ASDAS-CRP BASDAI [0-10] CRP [mg/l] Elevated CRP (%)‡ ESR [mm/h] Patient's global assessment of disease activity [0-10] Spinal pain [0-10] mSASSS [0-72] mSASSS >0 (%) NSAIDs (%) Tumour necrosis factor alpha inhibitors (%)

N = 184 mean (SD) or n (%) 43 (12) 129 (70%) 149 (83%) 20 (12) 11 (9) 2.6 (1.0) 3.4 (2.0) 17.4 (23.3) 85 (48%) 14.0 (15.0) 3.7 (2.7) 3.5 (2.3) 10.8 (15.2) 140 (81%) 125 (68%) 0 (0%)

‡The cut-off was 10mg/L for the Dutch patients and 5mg/L for the Belgian and French patients ASDAS-CRP, Ankylosing Spondylitis Disease Activity Score (C-reactive protein); BASDAI, Bath Ankylosing Spondylitis Disease Activity Score; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score; NSAIDs, non-steroidal anti-inflammatory drugs

116

Chapter 9

A total of 39 (21%) patients had 2 sets of x-rays, 32 (17%) had 3 sets, 31 (17%) had 4 sets of x-rays, 32 (17%) had 5 sets, 29 (16%) had 6 sets and 21 (11%) had all 7 sets of x-rays. During follow-up, 95% of the patients were at some time exposed to NSAIDs and 22% (5% before year 8) to TNFi. The mean (SD) follow-up duration per patient was 7.9 (4.0) years, and the mean (SD) number of radiographs was 4.2 (1.7) (range 2-7). The vast majority (87%) of the intervals between evaluable x-rays were 2 years, in less than 1% was the interval longer than 6 years and on average the interval was 2.4 years. Baseline analysis

On average patients had a progression of 1.9 mSASSS units/2 years. All separate disease activity measures assessed at baseline except spinal pain were significantly associated with the course of radiographic damage over time (table 2). Patients with inactive disease (ASDAS <1.3) at baseline had an average progression of 0.71 mSASSS units/2 years, whereas patients with very high disease activity (ASDAS >3.5) at baseline had a progression of 3.1 mSASSS units/2 year progression. A similar relationship was found for BASDAI: patients with a BASDAI <4 at baseline had an average progression of 1.4 mSASSS units/2 years, while patients with a BASDAI ≥4 (2.7 mSASSS units/2 years) or those with a BASDAI >6 (2.0 mSASSS units/2 years) at baseline experienced more

progression over time. Longitudinal analysis

All disease activity measures, except ESR as a separate variable, were longitudinally associated with radiographic progression (table 3). The autoregressive time lagged nature of the analysis is congruent with the following longitudinal interpretation: ‘An increase of one ASDAS unit in an individual patient is expected to lead to an increase of 0.72 mSASSS units over the next two years’. Further, ‘a patient with very high disease activity (ASDAS >3.5) may in comparison to a patient with inactive disease (ASDAS <1.3) expect an additional progression of 2.3 mSASSS units in the subsequent two years.’ Similarly, ‘an increase of one BASDAI unit is associated with an increase of 0.21 mSASSS units, and an increase of 10mg/L in CRP is associated with an increase of 0.2 mSASSS units’. Models with ESR instead of CRP were similar, except for the fact that ESR itself was not a statistically significant contributor in any analysis. The model with ASDAS as a continuous score had lowest QICs of all models, illustrating Disease activity and structural damage

117

Table 2 – Effect of disease activity at baseline on the course of radiographic progression over time Progression rate per 2 year interval* p value** β (95% CI) Time

1.95 (1.67; 2.23)

ASDAS-time interaction

N*** 202

0.001

200

Baseline ASDAS<1.3

0.71 (0.08; 1.35)

Baseline ASDAS≥1.3 & <2.1

1.30 (0.85; 1.76)

Baseline ASDAS≥2.1 & ≤3.5

1.90 (1.44; 2.35)

Baseline ASDAS>3.5

3.05 (2.34; 3.76)

BASDAI-time interaction

44 0.023

202

Baseline BASDAI<4

1.41 (1.07; 1.75)

120

Baseline BASDAI≥4

2.71 (2.18; 3.23)

CRP-time interaction

0.010

201

Negative baseline CRP‡

1.73 (1.42; 2.04)

101

Positive baseline CRP‡

2.20 (1.73; 2.67)

ESR-time interaction

0.005

Baseline ESR<20

1.72 (1.41; 2.04)

Baseline ESR≥20

3.01 (2.39; 3.62)

Patient's global assessment - time interaction 1.35 (1.00; 1.70)

Baseline patient's global assessment≥4

2.82 (2.36; 3.28)

Spinal pain - time interaction

47 0.001

Baseline patient's global assessment<4

201 149 202 111 90

0.147

201

* Because the dependent variable was the modified Stoke in Ankylosing Spondylitis Spine Score (mSASSS) and the independent variable was time, the regression coefficient (ß) obtained reflects the progression of mSASSS per 2 years ** p-values are only shown for the interactions between time and other variables. In the presence of a significant interaction, results are stratified, and the progression in each of the subgroups is presented through the regression coefficient. *** All patients with an mSASSS evaluable at each time point independently of the availability of a 2 year progression score were included in the models ASDAS, Ankylosing Spondylitis Disease Activity Score; BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate. ‡The cut-off was 10mg/L for the Dutch patients and 5mg/L for the Belgian and French patients

that the model with ASDAS as a disease activity measure has best captured variation in radiographic scores (better than the model with BASDAI and CRP, for instance). The visual relationship between ASDAS and mSASSS is shown in figure 1. In this modelling exercise, figure 1A, a stable ASDAS level over time was assumed, and progression curves were plotted for different ASDAS levels. Starting with an mSASSS of 10.8 at baseline (representing the study population’s mean value), the average patient with ‘inactive disease’ (plotted as ASDAS of 1.0) would reach an mSASSS of 15.8 (5 units progression) after 12 years, whereas the average patient with ‘very active disease’ (plotted as ASDAS of 4.0) would reach 118

Chapter 9

Disease activity and structural damage

Multivariable regression 1 ß (95% CI) (N = 183¥) 1.03 (1.01; 1.05) 0.72 (0.41; 1.04) § § § § § § § § § § § ** ** ** ** ** ** ** ** **

Analysis in separate models ß (95% CI)* (N = 174 - 185) -0.72 (0.41; 1.04)

0.57 (-0.56; 1.69)

0.91 (-0.17; 1.99)

2.31 (1.11; 3.51)

0.24 (0.09; 0.40)

-0.19 (-0.98; 0.60)

1.13 (0.30; 1.97)

1.11 (0.12; 2.11)

0.02 (0.01; 0.04) 0.02 (-0.01; 0.04) 0.20 (0.08; 0.32)

0.25 (0.12; 0.38) 0.47 (-0.47; 1.41)

0.25 (-0.94; 1.44) 1.03 (-0.19; 2.25) 0.58 (-0.10; 1.26) 0.00 (-0.01; 0.01) -0.11 (-0.89; 0.66) 0.26 (-0.81; 1.33) -0.25 (-1.68; 1.18) -0.17 (-0.86; 0.52)

Model with ASDAS continuous

§ -0.29 (-1.09; 0.52)

0.21 (0.06; 0.37) §

** ** ** ** ** ** ** **

§ **

§ § §

2.31 (1.11; 3.51)

1.06 (0.21; 1.90) 0.82 (-0.19; 1.84) 0.02 (0.01; 0.04) § § § ** ** ** ** ** ** ** ** **

§ § 0.02 (0.00; 0.04) § § § ** ** ** ** ** ** ** ** **

0.91 (-0.17; 1.99)

0.57 (-0.56; 1.69)

Model with ASDAS Model with BASDAI Model with BASDAI categorical continuous and CRP categorical and CRP Multivariable Multivariable Multivariable regression 2 regression 3 regression 4 ß (95% CI) ß (95% CI) ß (95% CI) (N = 183¥) (N = 184) (N = 184) 1.03 (1.01; 1.05) 1.03 (1.01; 1.05) 1.03 (1.01; 1.05) § § §

** ** ** ** ** ** ** **

§ **

0.02 (0.00; 0.04) § 0.17 (0.05; 0.29)

** ** ** ** ** ** ** **

0.22 (0.00; 0.04) **

0.02 (0.00; 0.04) § §

Multivariable regression 6 ß (95% CI) (N = 184) 1.03 (1.01; 1.05) §

Multivariable regression 5 ß (95% CI) (N = 184) 1.03 (1.01; 1.05) § §

Model with back pain and CPR

Model with Patient global and CRP

QIC of the model 5439 5458 5665 5579 5664 5473 *All models are time lagged (2 years of time lag) and autoregressive (i.e, adjusted for mSASSS in the 2 years before). ß reflect progression in mSASSS units over 2 years ** Not selected during multivariable regression analysis (p≥0.05) § Not included in the model ¥ One patient had a missing in one of the items of the ASDAS (BASDAI question 3) and therefore ASDAS could not be calculated ASDAS, Ankylosing Spondylitis Disease Activity Score; BASDAI, Bath Ankylosing Spondylitis Activity Index; CRP, C-reactive protein; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score; ESR, erythrocyte sedimentation rate; DMARDs, disease modifying antirheumatic drugs; NSAIDs, non-steroidal anti-inflammatory drugs; QIC, Quasi-likelihood under the Independence model Criterion.

Previous mSASSS (0-72) ASDAS ASDAS disease activity states - Moderate vs inactive (≥1.3 and <2.1 vs <1.3) - High vs inactive (≥2.1 and ≤3.5 vs <1.3) - Very high vs inactive (>3.5 vs <1.3) BASDAI (0-10) BASDAI disease activity states - Moderate vs inactive (≥2 and <4 vs <2) - High vs inactive (≥4 and ≤6 vs <2) - Very high vs inactive (>6 vs <2) CRP (mg/l) ESR (mm/h) Patient's global assessment (0-10) Pain (0-10) HLA-B27 (positive vs negative) Biologics DMARDs NSAIDs NSAID score (0-100) Uveitis Inflammatory bowel disease Psoriasis Extra-articular manifestation

Variable

Table 3 - Longitudinal relationship between disease activity and radiographic damage

119

an mSASSS of 29.8 units (19 units progression). In figure 1B, a modelled example is shown reflecting how (longitudinal) fluctuation in ASDAS may lead to (longitudinal) fluctuation in mSASSS progression in a typical patient. A significant interaction was found between ASDAS and gender (p=0.007), as well as ASDAS and symptom duration (p=0.011). The effect of ASDAS on mSASSS was higher in males versus females: an increase of one ASDAS unit led to an increase of 0.98 and -0.06 mSASSS units/2 years, respectively. The effect of ASDAS on mSASSS was also higher in patients with symptom duration below the median (18 years) versus above the median: an increase of one ASDAS unit led to an increase of 0.84 vs 0.16 mSASSS units/2 years respectively (table 4). The same type of effect modification was found in the models with other disease activity

Figure 1 - Longitudinal relationship (modelled) between disease activity (ASDAS) and radiographic damage (mSASSS) The graphs are two visualisations (solutions) of the same multivariable regression equation with the parameter estimates shown in table 4: mSASSS timepoint t = -0.27 + 1.03 * mSASSS t-1 + 0.72 ASDAS t-1. The mSASSS at baseline (10.8 units) was assumed to be the intercept ( mSASSS at t0). In (A), four equidistant ASDAS values, each reflecting a certain disease activity state, were chosen and a stable ASDAS level over time was assumed. In (B), fluctuating ASDAS levels over time (in a decreasing pattern) were assumed (closed triangles) ASDAS, Ankylosing Spondylitis Disease Activity Score; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score

120

Chapter 9

Table 4 - Longitudinal relationship between ASDAS and radiographic damage in different strata Stratification for

Stratum analysed

Gender

ASDAS in males (n = 129)

Multivariable regression models (auto-regressive and time lagged) ß (95%CI)* 0.98 (0.56; 1.39)

ASDAS in females (n = 54) ¥ Symptom duration§

-0.06 (-0.40; 0.27)

ASDAS in symptom duration <18years (n = 92)

0.84 (0.42; 1.26)

ASDAS in symptom duration ≥18years (n = 80)

0.16 (-0.22; 0.54)

Gender and symptom ASDAS in women, symptom duration <18years duration§ (n = 24)

0.21 (0.03; 0.38)

ASDAS in women, symptom duration ≥18years (n = 29)

-0.06 (-0.63; 0.51)

ASDAS in men, symptom duration <18years (n = 68)

0.94 (0.43; 1.45)

ASDAS in men, symptom duration ≥18years (N = 51)

0.36 (-0.12; 0.84)

* Results refer to different multivariable models. Stratifications were made due to significant interactions. The ß indicates the progression in mSASSS per unit increase in ASDAS per 2 years. In italic are the strata in which there is a significant effect of disease activity (as measured by the ASDAS) on radiographic damage (as measured by the mSASSS) ¥ One patient had a missing in one of the items of the ASDAS (BASDAI question 3) and therefore ASDAS could not be calculated § Symptom duration was not available for 12 patients who could therefore not be included in this stratified analysis ASDAS, Ankylosing Spondylitis Disease Activity Score; BASDAI, Bath AS Disease Activity Index; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score

measures. Effect modification or confounding was not found for HLA-B27, drug use (NSAIDs, TNFi) and the presence of any EAM (table 3). Additional information on drug use over time can be found in the online supplementary table S2. A sensitivity analysis including HLA-B27, NSAID-score and TNFi-usage provided similar results than the main analysis (see online supplementary table S3). When –as a sensitivity analysis - analyses were repeated using only data from follow-up until year 4, results were similar for most disease activity measures, but the longitudinal association between CRP and radiographic progression was not statistically significant anymore.

DISCUSSION In the present study, we have shown for the first time that disease activity is longitudinally associated with radiographic progression in AS. The effect of disease activity on radiographic damage is stronger in men than in women, and in the earlier phases of the disease. Although the longitudinal association is seen with all measures of disease activity, the ASDAS,

Disease activity and structural damage

121

combining PROs and APR in a fixed relationship, showed the best performance. Previous studies have found only a relatively weak association between radiographic progression and APR, but not with radiographic progression and PROs such as BASDAI . Further, these studies provided cross-sectional analyses that do not allow a longitudinal

6-9

interpretation

. We have demonstrated here for the first time that the effect of disease

6,7

activity on radiographic damage is actually rather impressive and longitudinal: per one ASDAS unit increase a 0.7 mSASSS units progression in 2 years can be expected. This longitudinal relationship was pertinent over the entire 12 years of follow-up. An important observation here is that the models with ASDAS as a disease activity measure outperformed the models with all other disease activity measures, which importantly adds to the validity of ASDAS as the disease activity measure of choice. The fact that significant longitudinal associations with radiographic progression could also be established for other disease activity measures (among which PROs that showed inconclusive results in conventional cross-sectional analyses in the past) than ASDAS alone points to the power of longitudinal databases and appropriate longitudinal analysis. While the observation of a link between disease activity and radiographic progression is rather new, we acknowledge that disease activity (inflammation) by far does not fully explain radiographic damage in AS: even without measurable disease activity there is still considerable radiographic progression. It can be deduced from the model that patients who have inactive disease for the entire 12 years still can expect a progression of 5 mSASSS units on average. Radiographic progression without measurable clinical disease activity in AS is different from radiographic progression in patients with RA in remission in which progression is almost absent. It seems as if increased bone formation in AS is partly constitutive and partly inflammation-dependent. A parallel with findings from MRI studies is appealing here: MRI inflammation at the vertebral corner to some extent forecasts the development of syndesmophytes, but most new syndesmophytes still develop in sites not previously active on MRI 22. The effect of disease activity on radiographic progression seems to be far stronger in men than in women. Only in men we found an additional progression of about 1 mSASSS unit per 2 year interval per additional unit of ASDAS. This male-female disparity is an attractive explanation for the frequently observed more severe radiographic damage in males 16. Such a statement should be taken with some consideration, though, since less than one-third of the included patients were women with on average little structural progression, which complicates the search for an external factor that modulates progression 16. We have also found a stronger effect of disease activity on radiographic progression in

122

Chapter 9

the (relatively) early phase of the disease. A clear explanation is lacking, but one might speculate on the effects of more strenuous physical exercise in the early phases of the disease as compared to later phases. A relationship between mechanical forces on the spine and syndesmophyte formation has been postulated in the past

, and one may

reason that younger people with shorter disease duration will usually perform more intensive occupational physical activities with stronger forces acting on the spine. Since male patients may often have physically more demanding jobs than females, this speculative explanation may extend to the demonstrated male-female disparity. Of note, short and long symptom duration in our analysis is only a convenience distinction since we have stratified by the median of symptom duration, which was already 18 years. It is obvious that more research should give resolution here. What are the consequences of our observations? It was the unexpected lack of inhibition of structural damage by TNFi

3-5

that has triggered

the debate about the link of inflammation and bone formation. Recent new observational studies have suggested a protective effect of TNFi on radiographic progression 9,24, but are not beyond methodological argumentation 25. The results of the current study could light-heartedly be interpreted as if inhibition of disease

activity (by, eg, TNFi) will automatically lead to inhibition of radiographic progression. Such a spurious â&#x20AC;&#x2DC;causal-chain-interpretationâ&#x20AC;&#x2122; would ignore the evidence that TNFi have more effects than only suppression of TNF-induced inflammation. For example, TNFi have been suggested to be responsible for bone formation in animal models 26, so that negative effects on unrelated bone-forming processes could counterbalance a potentially positive effect on inflammation. The negative results of TNFi with regard to syndesmophyte formation in three independent studies with an untreated control group 3-5 are compatible with such a scenario and cannot easily be ignored by sparse information from observational studies 9,24, although former trials were of short duration. Further, trials with NSAIDs in AS in fact have proven a TNF-independent mechanism of action being responsible for NSAID-induced inhibition of syndesmophyte formation 27. The biggest challenge for the near future will therefore be to design a clinical study that takes all these different aspects into consideration. This study has limitations that should be discussed. As a cohort study, it suffers from losses to follow-up, which may result in bias-by-completion. However, our main study question relates to an association between disease activity and radiological progression rather than to find the best estimate of absolute progression. It is unlikely that the relationship between Disease activity and structural damage

123

disease activity and progression per se is different in patients completing 12 years of followup and in those that discontinue early for whatever reasons. Furthermore, the sample size of this long-term observational study is not large enough to split the data into several relevant subgroups (eg, females only, or HLAB27-negative patients). Strengths of our study are the uniquely long follow-up of patients with AS according to a predefined protocol of data collection and the longitudinal analyses we conducted, which make use of all data collected over time, are more powerful and may to some extent allow a temporal relationship, thus bringing in an element of causality. Analyses incorporated adjustments for existing structural damage at every time point, the strongest known determinant of subsequent progression 6,7,28. Another important strength of this study, already discussed, is the consistency of the results across the different disease activity measures analysed in this study. In summary, this study has demonstrated that inflammation in AS indeed leads to new bone formation. The next question will be whether lowering disease activity by drugs with various modes of action will result into less radiographic damage.

SUPPLEMENTARY DATA Supplementary data are published on the website of the Annals of the Rheumatic Diseases.

124

Chapter 9

REFERENCES 1.

Lories RJ, Dougados M. Inflammation and ankylosis: still an enigmatic relationship in spondyloarthritis. Ann Rheum Dis 2012;71:317-8.

Maksymowych WP, Elewaut D, Schett G. Motion for debate: the development of ankylosis in ankylosing spondylitis is largely dependent on inflammation. Arthritis Rheum 2012;64:1713-9.

14.

Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9.

van der Heijde D, Landewe R, Einstein S, et al. Radiographic progression of ankylosing spondylitis after up to two years of treatment with etanercept. Arthritis Rheum 2008;58:1324-31.

15.

16.

Ramiro S, Stolwijk C, van Tubergen A, et al. Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study. Ann Rheum Dis 2015;74:52-9.

van der Heijde D, Landewe R, Baraliakos X, et al. Radiographic findings following two years of infliximab therapy in patients with ankylosing spondylitis. Arthritis Rheum 2008;58:3063-70.

17.

Garrett S, Jenkinson T, Kennedy LG, et al. A new approach to defining disease status in ankylosing spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994;21:2286-91.

18.

Machado P, Landewe R, Lie E, et al. Ankylosing Spondylitis Disease Activity Score (ASDAS): defining cut-off values for disease activity states and improvement scores. Ann Rheum Dis 2011;70:47-53.

van Tubergen A, Ramiro S, van der Heijde D, et al. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23.

19.

Kroon F, Landewe R, Dougados M, et al. Continuous NSAID use reverts the effects of inflammation on radiographic progression in patients with ankylosing spondylitis. Ann Rheum Dis 2012;71:1623-9.

Dougados M, Simon P, Braun J, et al. ASAS recommendations for collecting, analysing and reporting NSAID intake in clinical trials/ epidemiological studies in axial spondyloarthritis. Ann Rheum Dis 2011;70:249-51.

Haroon N, Inman RD, Learch TJ, et al. The Impact of TNF-inhibitors on radiographic progression in Ankylosing Spondylitis. Arhtiris Rheum 2013;65:2645-54.

10.

Lukas C, Landewe R, Sieper J, et al. Development of an ASAS-endorsed disease activity score (ASDAS) in patients with ankylosing spondylitis. Ann Rheum Dis 2009;68:18-24.

11.

van der Heijde D, Lie E, Kvien TK, et al. ASDAS, a highly discriminatory ASAS-endorsed disease activity score in patients with ankylosing spondylitis. Ann Rheum Dis 2009;68:1811-8.

12.

Welsing PM, Landewe RB, van Riel PL, et al. The relationship between disease activity and radiologic progression in patients with rheumatoid arthritis: a longitudinal analysis. Arthritis Rheum 2004;50:2082-93.

13.

Spoorenberg A, van der Heijde D, de Klerk E, et al. Relative value of erythrocyte sedimentation rate

and C-reactive protein in assessment of disease activity in ankylosing spondylitis. J Rheumatol 1999;26:980-4.

20. Twisk J. Applied longitudinal data analysis for epidemiology: a practical guide. Cambridge: Cambridge University Press; 2003. 21. Twisk JW. Longitudinal data analysis. A comparison between generalized estimating equations and random coefficient analysis. Eur J Epidemiol 2004;19:769-76. 22. van der Heijde D, Machado P, Braun J, et al. MRI inflammation at the vertebral unit only marginally predicts new syndesmophyte formation: a multilevel analysis in patients with ankylosing spondylitis. Ann Rheum Dis 2012;71:369-73. 23. Ward MM, Reveille JD, Learch TJ, et al. Occupational physical activities and long-term functional and radiographic outcomes in patients with ankylosing spondylitis. Arthritis Rheum 2008;59:822-32. 24. Baraliakos X, Haibel H, Listing J, et al. Continuous long-term anti-TNF therapy does not lead to an increase in the rate of new bone formation over 8 years in patients with ankylosing spondylitis. Ann Rheum Dis 2014;73:710-5. 25. Machado P. Anti-TNF and new bone formation in

Disease activity and structural damage

125

ankylosing spondylitis - the controversy continues. Arthritis Rheum 2013;65:2537-40. 26. Zwerina J, Tuerk B, Redlich K, et al. Imbalance of local bone metabolism in inflammatory arthritis and its reversal upon tumor necrosis factor blockade: direct analysis of bone turnover in murine arthritis. Arthritis research & therapy 2006;8:R22. 27. Wanders A, Heijde D, Landewe R, et al. Nonsteroidal antiinflammatory drugs reduce radiographic progression in patients with ankylosing spondylitis: a randomized clinical trial. Arthritis Rheum 2005;52:1756-65.

126

Chapter 9

28.

Baraliakos X, Listing J, Rudwaleit M, et al. Progression of radiographic damage in patients with ankylosing spondylitis: defining the central role of syndesmophytes. Ann Rheum Dis 2007;66:910-5.

Disease activity and structural damage

127

Mechanical stress and smoking may modify the effect of disease activity on radiographic progression in patients with ankylosing spondylitis Sofia Ramiro, Robert Landewé, Astrid van Tubergen, Annelies Boonen, Carmen Stolwijk, Maxime Dougados, Filip van den Bosch, Désirée van der Heijde

Submitted

ABSTRACT Objectives

To investigate the complex relationship between inflammation, mechanical stress and radiographic progression in patients with AS, using job type as a proxy for continuous mechanical stress. Methods

Patients from OASIS were followed-up for 12 years, with 2 yearly assessments. Two readers independently scored the x-rays according to the mSASSS. Disease activity was assessed by the ASDAS-CRP. The relationship between ASDAS and spinal radiographic progression was investigated with longitudinal analysis, with job type at baseline (physically demanding (‘blue collar’) vs. sedentary (‘white collar’) labor) as a potential factor influencing this relationship. The effects of smoking status and socio-economic factors were also investigated. Results

In total, 184 patients were included in the analyses (70% males, 83% HLA-B27 positive, 39% smokers, 48% blue-collar workers (65/136 patients in whom data on job type were available)). The relationship between disease activity and radiographic progression was significantly and independently modified by job type: In ‘blue-collar’ workers vs. ‘white collar’ workers every additional unit of ASDAS resulted in an increase of 1.2 vs. 0.2 mSASSS units/2 years (p=0.014 for the difference between blue collar and white collar workers). In smokers vs. non-smokers every additional unit of ASDAS resulted in an increase of 1.9 vs. 0.4 mSASSS units/2 years. Conclusion

Physically demanding jobs may amplify the driving effects of inflammation on radiographic

progression, thus supporting the theory that mechanical stress leads to bone formation in AS. Smoking and personal income are likely classic confounders of this relationship but a separate detrimental effect of smoking on radiographic progression could not be excluded.

130

Chapter 10

INTRODUCTION In ankylosing spondylitis (AS) radiographic progression of the spine is faster in males, in HLA-B27 positive patients, in patients with active disease and in those that already have signs of damage 1-5. We have recently shown that disease activity is longitudinally associated with radiographic progression 5. This effect is amplified in males and in patients with shorter symptom duration 5. Part of syndesmophyte formation was constitutive; without any signs of disease activity patients still had some progression 5. A clear and overarching explanation of syndesmophyte formation in AS is still lacking, but – in analogy with osteophyte formation in osteoarthritis 6 - there may be a role for mechanical stress. Mice that were tail suspended, so that mechanical loading on paws was low, showed less bone formation than those kept in regular cages 7, which may be proof-of-concept for the proposition that mechanical strain drives new bone formation in spondyloarthritis (SpA). In addition, sparse data have suggested that occupations with physically demanding activities are associated with more radiographic damage 8. A detailed analysis of how mechanical stress and inflammation may interact in explaining radiographic progression has never been conducted, because lifetime mechanical stress is difficult to quantify and because the role of inflammation in driving syndesmophyte formation has long been obscure. But it is rational to postulate that radiographic progression is the resultant of an interplay between different factors (among which disease activity and

mechanical stress), rather than the consequence of a sole factor. Now we have recently established the potential contribution of inflammation to radiographic progression in the Outcome in Ankylosing Spondylitis International Study (OASIS cohort), using the AS Disease Activity Score (ASDAS) as a proxy for inflammation, and also have formulated the idea of ‘constitutive progression’ (independent of inflammation)5, we think the OASIS cohort may be the appropriate template to study the interplay between inflammation and mechanical stress. In light of the problem to quantify ‘long-term mechanical stress’ we have chosen ‘job type’ (physically demanding vs. sedentary) as the most convenient proxy for ‘lifetime mechanical stress’ on the spine. We have also taken potential confounders into consideration: Smoking status was shown to be predictive of radiographic progression in patients with axial SpA 3. It is attractive to hypothesize that smoking is associated with ‘job type’, but smoking may also have an independent effect on radiographic progression. Likewise, socio-economic factors such as personal income may interfere because they may be associated with ‘job type’ 9-11. We have performed a detailed analysis focusing on the effects of ‘job type’, smoking status Mechanical stress and structural damage

131

and socio-economic status on radiographic progression in AS.

METHODS Study population

Data from the OASIS were used. OASIS is a prevalence cohort starting (1996) with 217

consecutive patients with AS from the Netherlands, Belgium and France

. Clinical and

radiographic (cervical and lumbar spine) data were collected every 2 years during 12 years. Patients were included in the present study if they had at least two subsequent radiographs and data on disease activity as well as on occupational activities and/or smoking status and/ or socio-economic factors (educational level, personal income and/or family income) were available. All patients gave informed consent and the ethics committee from all participating hospitals approved the study. Radiographic damage

Radiographs were scored using the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS), which ranges from 0 (no damage) to 72 (maximal damage) 13. Two well-trained readers (SR and CS) independently scored all available radiographs per patient, blinded to demographic and clinical data, but with known chronology 14. Mean reader-scores were used. Details of the reading methodology have been reported elsewhere 1. Disease activity and occupational activity, smoking status and socio-economic factors

The disease activity measure of choice was the ASDAS 15 which best reflects the association between disease activity and radiographic progression 5. Patient-reported occupational activities (collected at baseline by an open question) were used as a proxy for unmeasured lifetime mechanical stress on the spine. ‘Job type’ was determined by consensus, without knowledge of disease activity and/or radiographic damage. Two job types were distinguished: ‘blue collar’ and ‘white collar’ jobs, which are common circumscriptions for manual jobs (that imply more physical labor) and more sedentary jobs (that imply less physical activities) 16,17. Examples of ‘blue collar workers’ are craftworkers, laborers and transportation operatives. Examples of ‘white collar workers’ are managers, administrative workers, teachers and engineers. The analyses were performed under the assumptions that 1) ‘job type’ at baseline reflects ‘lifetime job type’; and 2) ‘blue collar jobs’ are associated with more mechanical stress on the spine than ‘white collar jobs’. Smokers versus non-smokers were distinguished based on baseline smoking status. If baseline information was not available, patients were retrospectively inquired in order to minimize missing data. 132

Chapter 10

Socio-economic factors were collected at baseline and included 1) educational level (collected in 7 categories and dichotomized into higher professional and university education versus any other level of education; 2) baseline gross monthly personal income and family income (collected in 10 categories and dichotomized at €1,588 (personal income) and €3,176 (family income) respectively. These amounts demarcate income classes 1-4 vs. 5-10. Statistical analysis

The template for the analysis of the effects of ‘job type’ on radiographic progression was based on the longitudinal analysis that has been presented in detail elsewhere 5. In brief, mSASSS over time (t) (mSASSSt) was longitudinally modeled using generalised estimating

equations (GEE), assuming an exchangeable working correlation structure for mSASSS in order to adjust for within-patient correlation 5.

In the first model (referred to as the DIRECT model since we investigated the direct effect of ‘job type’ on radiographic progression) ‘job type’ was introduced by testing the interaction of ‘job type’ and ‘time’ on mSASSSt. Using similar methodology, we investigated the effects of smoking and the socio-economic factors, separately and in combination with ‘job type’.

Second, the potential of the ‘job type’ to modify the established relationship between disease activity and mSASSSt 5 (referred to as the INDIRECT model as we investigated the effect of

‘job type’ on the relationship between ASDAS and radiographic progression) was analysed

in our autoregressive model with 2 year time lags as proposed previously 5. Briefly, a ‘2

year time lag’ here means that disease activity at the start of a 2 year interval (ASDASt) was

associated with radiographic progression during the subsequent 2 year interval (or: mSASSSt was modeled by ASDASt-1 and by mSASSSt-1 (autoregressive component)). ’Job type’ was

tested in interaction with the longitudinal variable ‘ASDASt’. Similar INDIRECT models

were run for smoking and socio-economic factors. In the presence of relevant interactions (arbitrarily defined as interactions with P<0.1) analyses were repeated in subgroups.

Figure 1 represents graphically the DIRECT and INDIRECT models that were conducted to investigate the effects that an ‘exposure’ (e.g. occupational activity) could have on radiographic damage. Note that the consequences of both models are fundamentally different. Goodness-of-fit statistics (Quasi-likelihood under the Independence model Criterion (QIC)), with lower QICs reflecting better data fit, were used to select the best models. Due to missing information regarding occupational activity, smoking status and the different socio-economic factors, analyses were first conducted in all patients with each of the variables available. Next, sensitivity analysis in patients with complete data (missing family income was allowed) was

Mechanical stress and structural damage

133

Figure 1 - Different scenarios that explain the effect that an external factor (here occupational activity - blue vs. white collar - used as an example) could hypothetically have on radiographic damage. A - Occupational activity as a predictor of the course of mSASSS over time, modifying this evolution over time; B - occupational activity as a factor modifying the relationship between ASDAS and mSASSS Graphs represent hypothetical scenarios and not real data. mSASSS - modified Stoke Ankylosing Spondylitis Spine Score; ASDAS - Ankylosing Spondylitis Disease Activity Score

conducted. None of the other measured variables in OASIS appeared to be confounders of the relationship between disease activity and radiographic progression as shown in previous analyses 5 and were omitted from this analysis. Analyses were done using Stata SE version 12 (StataCorp, College Station, TX, USA).

RESULTS In total, 184 patients (70% males, 83% HLA-B27 positive) were included in the analyses (table 1). These patients had similar baseline characteristics as those included in the entire OASIS cohort, and patients that were followed-up until year 12 were similar to those initially included in the study 5, just as those included in the sensitivity analysis (N = 85)) (Online Supplementary Table S1). Of the 136 patients with baseline occupational activity available, 65 (48%) were assigned a ‘blue collar’ job. Expectedly, ‘blue collar’ workers (86%) were more often of male gender than ‘white collar’ workers (63%), and men (82%) were more often smokers than women (63%). Importantly, ‘blue collar’ workers had a higher level of ASDAS than ‘white collar’ workers, and a higher mSASSS at baseline (table 1). ‘White collar’ workers had a higher level of education and a higher personal income. The DIRECT model: effects of ‘job type’ on radiographic progression

Radiographic progression was slightly but significantly higher in ‘blue collar’ than in ‘white-

collar’ workers: 2.18 mSASSS units/2 years (95% confidence interval (CI) 1.52; 2.84) vs. 1.82 mSASSS units/2 years (95% CI 1.54; 2.11) (P=0.05 for the difference between blue and

134

Chapter 10

Mechanical stress and structural damage

2.6 (1.0) 3.4 (2.0)

ASDAS-CRP

BASDAI [0-10]

NSAIDs (%)

49 (39%) 65 (48%)

Smoker (%)

Blue collar worker (%)

32 (52%) 14 (35%) 17 (33%) --

4 (9%) 23 (51%) --

12 (17%)

1 (2%) 17 (28%)

50 (78%)

6.4 (8.6) 50 (70%)

25 (53%)

34 (50%)

30 (48%) 11.0 (14.7) 50 (81%)

3.4 (1.9) 18.9 (23.0)

2.9 (1.9) 15.7 (23.0)

3.7 (2.0) 18.2 (21.6)

46 (71%)

2.8 (1.0)

2.4 (0.9)

2.9 (1.0)

23 (58%)

6 (21%)

16 (38%)

1 (2%)

36 (73%)

31 (72%)

9.9 (15.9)

8 (6)

9 (8)

16 (10)

18 (9) 11 (8)

17 (10)

43 (88%)

40 (82%)

38 (11)

Smokers* N = 49 mean (SD) or n (%)

22 (39%)

10 (21%)

23 (34%)

9 (12%)

54 (69%)

60 (80%)

9.3 (13.7)

37 (49%)

17.9 (25.9)

3.2 (2.2)

2.6 (1.1)

11 (9)

20 (11)

61 (78%)

49 (63%)

42 (12)

Non-smokers* N = 78 mean (SD) or n (%)

‡The cut-off was 10mg/l for the Dutch patients and 5mg/l for the Belgian and French patients § Baseline occupational activity was missing for 48 patients (6 retired, 25 work-disabled, 4 housewives, 2 not working for own choice, 3 students, 1 unemployed, and 7 with missing baseline occupational activity missing) ASDAS-CRP, Ankylosing Spondylitis Disease Activity Score (C-reactive protein); BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; CRP, C-reactive protein; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score; NSAIDs - non-steroidal anti-inflammatory drugs * Baseline smoking status was missing for 57 patients

56 (35%) 21 (19%)

Monthly personal income ≥€1588 (%)

Monthly family income ≥€3176 (%)

14 (8%)

140 (81%) 125 (68%)

mSASSS >0 (%)

University education (%)

85 (48%) 10.8 (15.2)

Elevated CRP (%)‡

mSASSS [0-72]

17.4 (23.3)

11 (9)

CRP [mg/l]

20 (12)

61 (88%)

Disease duration [years]

45 (63%)

56 (86%) 50 (79%)

129 (70%) 149 (83%)

Male gender (%)

HLA-B27 positive (%)

Symptoms duration [years]

41 (11)

40 (12)

43 (12)

Age [years]

White-collar jobs§ N = 71 mean (SD) or n (%)

Blue-collar jobs§ N = 65 mean (SD) or n (%)

N = 184 mean (SD) or n (%)

Assessment

Table 1 - Baseline demographic, clinical and radiographic characteristics of all patients and stratified by baseline smoking status and by baseline occupational activity

135

white collar workers). When investigating the effect of occupational activities on radiographic progression in subgroups of males and females, statistical significance was lost. Smoking status (p=0.22), education (p=0.44), personal income (p=0.99) and family income (p=0.80) were not directly associated with radiographic progression. The INDIRECT model: Modification of the relationship between disease activity and radiographic progression by ‘job type’

The relationship between ASDAS and radiographic progression was significantly and importantly dependent on occupational activity (p=0.014): An increase of one ASDAS-unit led to an increase of 1.2 mSASSS units per 2 years in ‘blue collar workers vs. 0.2 mSASSS units per 2 years in ‘white collar’ workers (table 2). Similar effects were found when the analysis was done with ‘smoking’ or ‘personal income’ instead of ‘job types’ as explanatory factors, but not with ‘educational level’ or ‘family income’. Note that the disparate effects of ‘job type’ on this relationship (but also the effects of smoking and personal income) are only seen in men, and not in women: An association between ASDAS and radiographic progression in women was almost absent 5. In a subsequent analysis we have tried to disentangle the presumed effects of ‘job type’ and ‘smoking’ on the relationship between disease activity and radiographic progression. In the subgroup of smokers, 2 year progression per additional ASDAS-unit in ‘blue-collar’ workers (1.52 (95% CI 0.29, 2.74) mSASSS units) was slightly higher - but not significantly different from - progression in ‘white collar’ workers (1.24 (95%CI 0.09, 2.40) mSASSS units). In the subgroup of non-smokers, ‘blue collar’ workers had a 2 year progression of 0.61 (95% CI 0.18; 1.05) mSASSS progression per additional ASDAS unit increase. Unfortunately, the model did not reach convergence for ‘white collar’ workers (35 patients). Further subgroup analyses (in gender and/or personal income strata) were impossible because of small subgroups. When comparing the fit of the models (with each of the interaction terms included) the model with ‘smoking status’ had the best fit (QIC 4484), followed by the model with ‘job type’ (QIC 4565), personal income (QIC 5486) and finally education (QIC 5714). Note that the influence of smoking on the association between ASDAS and radiographic progression was statistically stronger than the effect of ‘job type’ on this relationship. Sensitivity analysis in patients with all variables available (except for family income due to higher number of missing values) provided similar results (Online Supplementary Table S2).

136

Chapter 10

Mechanical stress and structural damage

≥€3176 (n = 21)

<€3176 (n = 90)

≥€1588 (n = 56)

<€1588 (n = 105)

‘University’ (n = 14)

‘Non-unversity’ (n = 167)

‘White collar’ (n = 71)

‘Blue collar’ (n = 65)

Non-smokers (n = 78)

Smokers (n = 49)

0.445

0.059

0.364

0.014

<0.001

0.16 (-0.13; 0.44) -0.60 (-1.59, 0.40) -0.08 (-0.43; 0.28) -0.04 (-0.39; 0.30) -0.74 (-1.82; 0.34) -0.20 (-0.66; 0.25) -0.21 (-0.90; 0.48) -0.25 (-0.80; 0.30)

0.44 (0.02; 0.86) 1.47 (0.81; 2.14) 0.35 (-0.30; 1.01) 1.00 (0.55; 1.44) 0.81 (-3.15; 4.78) 1.31 (0.66; 1.96) 0.18 (-0.24; 0.59) 0.77 (0.27; 1.27) 0.36 (-0.22; 0.94)

0.35 (0.04; 0.65) 1.19 (0.58; 1.79) 0.20 (-0.23; 0.64) 0.74 (0.41; 1.07) -0.18 (-1.91, 1.55) 0.93 (0.45, 1.41) 0.14 (-0.21, 0.50) 0.49 (0.09, 0.89) 0.15 (-0.35, 0.65)

-0.15 (-0.93; 0.63)

0.47 (-0.12; 1.06)

2.15 (1.01; 3.30)

Women 2 year increase in mSASSS per oneASDAS unit increase [units, (95% CI)]

1.94 (1.00, 2.87)

Men 2 year increase in mSASSS per oneASDAS unit increase [units, (95% CI)]

2 year increase in mSASSS per oneASDAS unit increase [units, (95% CI)]

Overall group p-value for the interaction

*All models are time lagged (2 years of time lag) and autoregressive (i.e. adjusted for mSASSS in the 2 years before). Progression per subgroup is expressed in mSASSS units over 2 years per one-ASDAS unit increase § Subgroup analysis was conducted in all patients with the variable of each of the subgroup analyses available, which means that due to missing values some patients were not included in some of the subgroup analyses. Numbers of included patients can be seen in front of the corresponding stratum. ASDAS - Ankylosing Spondylitis Disease Activity Score; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score

Monthly gross family income

Monthly gross personal income

Education

Job type

Smoking

Subgroup §

Table 2 - Effects of disease activity on radiographic progression in subgroups*

137

DISCUSSION In the present study we have proposed scientific arguments for the hypothesis that long-term physically demanding activities, here operationalized by ‘blue collar job type’, amplify the detrimental effects of inflammation on bone (syndesmophyte) formation in AS. We reiterate that the effect of ‘job type’ on radiographic progression was investigated in two different manners (figure 1): First, we have investigated whether ‘job type’ was associated with the course of radiographic progression itself (the DIRECT model). In fact, this turned out to be not the case: While progression was initially found to be higher in ‘blue collar’ than in ‘white collar’ workers, this contrast disappeared when repeating the analysis separately in males and females. We have already reported that radiographic progression was higher in men

and ‘blue collar’ work is primarily performed by men. The conclusion of this

analysis is therefore that strenuous physical activities likely do not have a strong DIRECT effect on radiographic progression. An intriguing alternative explanation could be that it is physical activity rather than gender that determines radiographic progression in AS, under the assumption that the intensity of physical labor is overall higher in men than in women regardless of ‘job type’. This study will not give further resolution. The second type of analyses (the INDIRECT models), however, showed an interference of ‘job type’ with the reported association of ASDAS and radiographic progression 5: ‘Blue collar’ work indeed amplified the effect of ASDAS on radiographic progression in comparison with ‘white collar’ work. This means that lifetime strenuous physical activities may amplify the detrimental effects of inflammation on radiographic progression. Implications could be far stretching: the commonly given advice to patients with AS to regularly exercise in order to optimize mobility and quality of life (which is supported by a Cochrane review

and

implemented in guidelines 19) may need to be revised, as at least certain types of exercises or mechanical stress on the spine seem to amplify progression of structural damage. If confirmed, this would imply that physically demanding labor should better be discouraged. There is argumentation in the literature to support our findings: The amplifying effects of ‘blue collar’ job type on the association between disease activity and radiographic progression can be a consequence of increased strenuous mechanical forces exerted on the spine. This is in concordance with what has recently been shown in animal models, in which mechanical strain has led to new bone formation 7. Unfortunately this study does not provide sufficient explanation for several reasons. First, mechanical strain has also been shown to directly result in inflammation in animal models 7. It is plausible that physically demanding jobs in patients with AS may lead to more ‘true’

138

Chapter 10

inflammation of the spine, and therefore to truly higher ASDAS levels, which may explain more progression. The higher ASDAS levels we found in patients with ‘blue collar’ jobs are in agreement with such a hypothesis. Second, we have found similar (or even stronger) interactions with ‘smoking’ and with ‘socioeconomic determinants’ with regard to the association between ASDAS and radiographic progression. On the other hand ‘job type’, ‘smoking’ and ‘socio-economic status’ are strongly related. ‘Blue collar’ workers, for example, are more frequently smokers than ‘white collar’ workers 20 and ‘blue collar’ workers have on average less income than ‘white collar’ workers. The model with ‘smoking’ had a better data-fit (lower QIC) than the model with ‘job type’. This raises the question of ‘what is the determinant and what is the confounder’? Is it ‘job type’ that eventually results in more progression, and is ‘smoking’ an epidemiological confounder; is it ‘smoking’ that is the primary cause of progression, and is ‘job type’ the confounder; or do both factors independently convey detrimental effects (figure 2)? In the appreciation that this study will not give a final verdict, biologic plausibility may give some resolution: Smoking has been

Figure 2 - Factors influencing the relationship between disease activity (as measured with the ASDAS) and radiographic progression (as measured with the 2 year mSASSS progression) and possible relationships between them A - hypothesis 1: occupational activity modifies the relationship between disease activity and radiographic progression and this effect might be confounded by the effect of gender, smoking status and/or low socioeconomic status which can for example be measured with education, personal income, family income. B - hypothesis 2: smoking status modifies the relationship between disease activity and radiographic progression and this effect might be confounded by the effect of gender, occupational activity and/or low socio-economic status ASDAS - Ankylosing Spondylitis Disease Activity Score; mSASSS, modified Stoke Ankylosing Spondylitis Spine Score

Mechanical stress and structural damage

139

associated in rheumatic diseases with worse outcomes, particularly in rheumatoid arthritis, where it provides an attractive explanation in the debate of citrullinization of peptides

While smoking has been associated with several factors of SpA, such as early onset of back pain, more disease activity and more magnetic resonance imaging inflammation 22-24, and it has also been associated with spinal damage in AS in one study 3, an attractive biological explanation is still lacking, or at least unproven 3. Mechanical stress, on the other hand, has been brought into relation with inappropriate bone formation in several conditions such as SpA

and osteoarthritis 6. It is therefore much more attractive and plausible to suggest

that mechanical stress (here artificially dichotomized into ‘blue collar’ vs. ‘white collar’ jobs) is the causative provoking factor, and ‘smoking’ -known to be associated with ‘blue collar’ work- the confounder, rather than vice-versa. An explanation that cannot be refuted is that both ‘job type’ and ‘smoking’ are independently contributory. In this discussion we propose that ‘personal income’, and to a lesser extent ‘education’ and ‘family income’, whilst being proven modifiers of the relationship between ASDAS and radiographic progression, do not have biological plausibility, and should be considered as confounders, id est: ‘personal income’ and others are subordinate to ‘job type’. In analogy with many rheumatic diseases, the association of socio-economic variables with outcome of disease also here is intriguingly present 25-28. As far as we know, the role of occupational activities, smoking or socio-economic factors on the course of radiographic progression over the long-term has not been investigated previously, and we cannot compare our findings with others. This study has several additional limitations to be mentioned. The sample size of this observational study is not large enough to allow subtle but complex relationships in critically relevant subgroups (eg: smokers vs. non-smokers and men vs. women). On the other hand, larger cohorts without biological treatment but more meticulous follow-up than this one will likely never be conducted. An important limitation is that we assumed the dichotomy of ‘blue-’ vs. ‘white collar’ jobs as being representative of a high versus low level of mechanical stress on the spine. We cannot exclude the possibility that ‘white collar’ workers follow more thoroughly the physicians’ recommendations to intensively exercise and thus compensate lower levels of physical activity during working hours with higher levels of exercise. However, regardless of what epidemiological mechanism may have worked against the effects of ‘job type’, the effects are significant and relevant. Another limitation of this study is that we have modeled ‘job type’ at baseline as being representative of ‘job type’ during follow-up. Patients may have changed jobs, for example

140

Chapter 10

as a consequence of their disease, and ‘blue collar’ workers may have accepted ‘white collar’ jobs. However, such an effect would have deflated rather than inflated the contrast, and is not an appropriate explanation for the differences found. What impact do the findings in this study and the previous one 5 have with regard to explaining syndesmophyte formation in AS? Previously we have identified important determinants of syndesmophyte formation in AS patients: It occurs primarily in HLA-B27 positive male patients, reiterating the constitutive (genetic) component 1. Disease activity (inflammation) does have an influence on the rate of progression, but primarily in (genetically) susceptible (male) patients 5. In addition we have now made likely that –among other factors still to be identified – lifetime physical activities during working hours may amplify the detrimental effects of inflammation on radiographic progression. Whether smoking is an independent modifier of this relationship or only a confounder still needs to be elucidated.

Mechanical stress and structural damage

141

REFERENCES 1.

13.

Baraliakos X, Listing J, Rudwaleit M, et al. Progression of radiographic damage in patients with ankylosing spondylitis: defining the central role of syndesmophytes. Ann Rheum Dis 2007;66:910-5.

Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9.

14.

15.

van Tubergen A, Ramiro S, van der Heijde D, et al. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23.

van der Heijde D, Lie E, Kvien TK, et al. ASDAS, a highly discriminatory ASAS-endorsed disease activity score in patients with ankylosing spondylitis. Ann Rheum Dis 2009;68:1811-8.

16.

Schreuder KJ, Roelen CA, Koopmans PC, et al. Job demands and health complaints in white and blue collar workers. Work 2008;31:425-32.

17.

Gaudette LA, Richardson A, Huang S. Which workers smoke? Health reports 1998;10:35-45 (ENG); 35-47 (FRE).

18.

Dagfinrud H, Kvien TK, Hagen KB. Physiotherapy interventions for ankylosing spondylitis. The Cochrane database of systematic reviews 2008:CD002822.

19.

Braun J, van den Berg R, Baraliakos X, et al. 2010 update of the ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2011;70:896-904.

Ramiro S, van der Heijde D, van Tubergen A, et al. Higher disease activity leads to more structural damage in the spine in ankylosing spondylitis: 12year longitudinal data from the OASIS cohort. Ann Rheum Dis 2014;73:1455-61.

van der Kraan PM, van den Berg WB. Osteophytes: relevance and biology. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society 2007;15:237-44.

Jacques P, Lambrecht S, Verheugen E, et al. Proof of concept: enthesitis and new bone formation in spondyloarthritis are driven by mechanical strain and stromal cells. Ann Rheum Dis 2014;73:437-45.

Ward MM, Reveille JD, Learch TJ, et al. Occupational physical activities and long-term functional and radiographic outcomes in patients with ankylosing spondylitis. Arthritis Rheum 2008;59:822-32.

Tehranifar P, Liao Y, Ferris JS, et al. Life course socioeconomic conditions, passive tobacco exposures and cigarette smoking in a multiethnic birth cohort of U.S. women. Cancer causes & control : CCC 2009;20:867-76.

10.

Franks P, Jerant AF, Leigh JP, et al. Cigarette prices, smoking, and the poor: implications of recent trends. American journal of public health 2007;97:1873-7.

11.

12.

142

Ramiro S, Stolwijk C, van Tubergen A, et al. Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study. Ann Rheum Dis 2015;74:52-9.

Sherriff NS, Coleman L. Understanding the needs of smokers who work as routine and manual workers on building sites: results from a qualitative study on workplace smoking cessation. Public health 2013;127:125-33. Spoorenberg A, van der Heijde D, de Klerk E, et al. Relative value of erythrocyte sedimentation rate and C-reactive protein in assessment of disease activity in ankylosing spondylitis. J Rheumatol

Chapter 10

1999;26:980-4.

20. Kotz D, West R. Explaining the social gradient in smoking cessation: itâ&#x20AC;&#x2122;s not in the trying, but in the succeeding. Tobacco control 2009;18:43-6. 21. Klareskog L, Malmstrom V, Lundberg K, et al. Smoking, citrullination and genetic variability in the immunopathogenesis of rheumatoid arthritis. Seminars in immunology 2011;23:92-8. 22. Chung HY, Machado P, van der Heijde D, et al. Smokers in early axial spondyloarthritis have earlier disease onset, more disease activity, inflammation and damage, and poorer function and healthrelated quality of life: results from the DESIR cohort. Ann Rheum Dis 2012;71:809-16. 23. Ward MM, Hendrey MR, Malley JD, et al. Clinical and immunogenetic prognostic factors for radiographic severity in ankylosing spondylitis. Arthritis Rheum 2009;61:859-66. 24. Kaan U, Ferda O. Evaluation of clinical activity and functional impairment in smokers with ankylosing spondylitis. Rheumatology international 2005;25:357-60. 25. Mackenbach JP, Stirbu I, Roskam AJ, et al. Socioeconomic inequalities in health in 22 European countries. The New England journal of medicine 2008;358:2468-81.

26. Callahan LF, Cleveland RJ, Shreffler J, et al. Associations of educational attainment, occupation and community poverty with knee osteoarthritis in the Johnston County (North Carolina) osteoarthritis project. Arthritis research & therapy 2011;13:R169.

28. Brennan SL, Turrell G. Neighborhood disadvantage, individual-level socioeconomic position, and self-reported chronic arthritis: a cross-sectional multilevel study. Arthritis Care Res (Hoboken) 2012;64:721-8.

27. Mackie SL, Taylor JC, Twigg S, et al. Relationship between area-level socio-economic deprivation and autoantibody status in patients with rheumatoid arthritis: multicentre cross-sectional study. Ann Rheum Dis 2012;71:1640-5.

Mechanical stress and structural damage

143

Summary and conclusions

SUMMARY AND CONCLUSIONS The studies described in this thesis cover aspects of the assessment of Ankylosing Spondylitis (AS) and analysis of its long-term outcome. More specifically, in these studies the following topics were addressed. First, several facets of outcome assessment in AS were discussed and improved, in particular pertaining to disease activity, spinal mobility and radiographic damage. Second, the course of radiographic damage over time was analysed and better insight was obtained into the process of syndesmophyte formation. Finally, longitudinal relationships between disease activity, spinal mobility and radiographic damage and other potential determinants of these outcomes were investigated. The studies were conducted in two cohorts: the Outcome in AS International Study (OASIS cohort) and the MOBILITY study. OASIS is a cohort of patients with prevalent AS, in which 217 patients from the Netherlands, Belgium and France have been included in 1996, and have been followed-up for 12 years 1. Clinical and radiographic data were collected at least every 2 years (during the first years of follow-up more often). Assessments included questionnaires, clinical investigations, physical examinations including spinal mobility measurements, laboratory assessments, and radiographic assessments of the pelvis, and the cervical and lumbar spine. The MOBILITY study was a cross-sectional study of normal individuals in whom a normal spinal mobility was expected and patients with conditions possibly affecting spinal mobility (e.g. back surgery and low back pain) were excluded. Recruitment was stratified by age, gender and height so that the population was balanced with respect to these factors, expected to influence spinal mobility. Participants were assessed once for their spinal mobility using a series of spinal mobility measurements. In this final chapter we will summarize the main findings of the studies comprised in this thesis and also share our vision on research challenges in the field for the upcoming years. Disease activity and function assessment

We started in chapter 2 with the proposal of a method for dealing with missing items in two

patient-reported instruments; one to measure disease activity, the Bath AS Disease Activity Index (BASDAI), and the other to measure functional disability, the Bath AS Functional Index (BASFI). This study was conducted using the 12 year OASIS data. Up to one missing item for the BASDAI and three for the BASFI could be reliably imputed by averaging the remaining items. These imputation techniques yielded an agreement >90% and resulted in a difference between the original and the imputed scores of â&#x2030;¤0.7 (which is half of the smallest detectable change). We considered this acceptable and have therefore recommended this technique. It was compared with other imputation methods, such as imputation of the median, and lowest, middle or highest value, but the average of the remaining items consistently performed 146

Chapter 11

best. So far, no clear instructions on how to deal with missing items were available and the proposed method is easy and can also easily be applied in clinical practice. Especially the BASDAI is frequently used for clinical decision-making: a BASDAI of 4 is a common cut-off to define eligibility for biologic therapy 2. With the proposed solution, one missing item will no longer preclude the calculation of the BASDAI and, hence, the score can still be taken into account to guide clinical decisions. Similarly, even when up to 3 items are missing can the BASFI still be calculated at an individual level, which can also contribute for clinical decisions, given the importance of functional disability as an outcome. Furthermore, we could hereby verify that missing data in BASDAI and BASFI items in our 12 year data from OASIS occurred very rarely (<2% of observations), facilitating longitudinal analysis (see below). Mobility

Thusfar spinal mobility, while being a central outcome in AS, had not yet been fully addressed in normal individuals. The Assessment of SpondyloArthritis international Society (ASAS), an international group of experts in the field of spondyloarthritis, recommends the assessment of spinal mobility measures, both in clinical practice and in clinical trials with therapeutic interventions 3. The recommended measures are: chest expansion, modified Schober’s test, occiput-to-wall distance, cervical rotation, and either lateral spinal flexion or Bath AS Metrology Index (BASMI)

. The BASMI, in turn, includes lateral spinal flexion, modified

3,4

Schober, cervical rotation, tragus-to-wall distance and intermalleolar distance 4. In normal individuals (‘norms’), ‘normal values’ or ‘reference values’ for the different spinal mobility measures, commonly used in the assessment of patients with AS, were not known. We have conducted a study on normal individuals, the MOBILITY study, rigorously excluding any

subject in whom ‘abnormal’ spinal mobility could be expected, so that ‘reference values’ for each spinal mobility measure could be derived (chapter 3). In anticipation of interference by

age, gender and height, MOBILITY had a ‘latin-square design’. Recruitment was stratified for the three above-mentioned factors, which means that the study population is balanced

with respect to those factors, and which further conveys the best power even with a relatively small sample, meaning that subtle relationships can be more efficiently identified. We first investigated the effect of age, height and gender on the different spinal mobility measures. A significant decrease in all spinal mobility measures was found with increasing age. Increasing height was associated with higher tragus-to-wall distance, lateral spinal flexion, chest expansion and intermalleolar distance. Female gender was associated with lower values of chest expansion or cervical rotation. In chapter 3, age- (and in some cases height-) specific

reference intervals and percentiles for each of the spinal mobility measures are presented.

Although gender was associated to some extent with some spinal mobility measures, its additional value in the reference intervals did not outweigh the increasing complexity of the model and of the consequent reference intervals. Age-specific, and in some cases heightSummary and conclusions

147

specific, reference intervals fitted the data best. Percentile curves were derived in analogy to the growth curves for the monitoring of children’s growth 5. The proposed reference intervals and percentile curves represent a clear way forward in the outcome assessment of AS, as we now know how spinal mobility measures perform in normal individuals and we can better interpret the same measures in patients with AS. An important age effect on all spinal mobility measures was found, which emphasizes that this needs to be taken into account when following patients with AS: likely some part of the impairment in mobility seen throughout the years is not attributable to the disease, but to normal aging. The proposed percentile curves also provide a tool for further research in patients with AS. They can be used as a reference of ‘normal values’ when comparisons between patients and normal individuals are necessary. Furthermore, percentile curves may have their value in the follow-up of patients with AS in order to find out if a patient increasingly deviates from his ‘personal curve’ over time. Whether this application has clinical relevance needs to be tested. From the same MOBILITY study we have derived an imputation method for lateral spinal flexion (LSF) measurement, namely where LSF has been inappropriately recorded (which often happens). Lateral spinal flexion can be calculated as the difference in the lateral distances between middle fingertip-to-floor in the neutral position and middle fingertip-tofloor in maximum latero-flexion 6. If neutral fingertip-to-floor is missing, and only fingertip-tofloor distance at maximum flexion is recorded, true lateral spinal flexion cannot be calculated. This problem of inappropriate recording of an otherwise correctly performed measure was frequently found in OASIS and likely in other cohorts too, and hampers the apt calculation of LSF. We have shown that neutral fingertip-to-floor, which is a static measure, can be reliably approximated by height using a simple equation (chapter 4). This imputation method

showed a very good fit of the data (R2 = 0.84).

This imputation method enabled us to ‘recover’ several inappropriately recorded LSF values in OASIS. Finally, we could perform a formal comparison of spinal mobility between patients with AS, using the 12 year OASIS data, and normal individuals, using the percentile curves. Our aim was to analyse if there is a hierarchical order for the occurrence of impairment of spinal mobility measures. This has never been properly investigated, since cut-off levels for normal vs. abnormal were lacking. As described in chapter 5, we have considered the spinal mobility

measures recommended by ASAS. Spinal mobility measures were considered ‘impaired’ based on the cut-offs from the age-specific percentile curves derived from normal individuals.

We have found that there is a clear fixed order in which the spine in AS gets involved: Lateral spinal flexion and modified Schober are most frequently impaired in patients with AS, followed by tragus-to-wall distance, then cervical rotation, then intermalleolar distance 148

Chapter 11

and at last chest expansion. This observation fuels the hypothesis of earliest involvement of the lumbar spine, followed by involvement of the thoracic spine and then the cervical spine. This fixed order of involvement of the spine was found across different patient groups, defined based on either gender, disease duration or the presence or extent of baseline syndesmophytes, all being factors potentially influencing spinal mobility. In addition, from this study we have also concluded that impairment in spinal mobility, defined as an impairment in at least one spinal mobility measure, can be detected by only measuring lateral spinal flexion and modified Schober. Such a simple screening strategy would miss only 9% of patients with an impairment of spinal mobility, while saving a lot of time involved to perform the full set of spinal mobility measurements. If one of the measures of lateral spinal flexion and modified Schober is impaired, a full set of mobility measures should be performed to be informed on the extent of spinal mobility impairment. This may have important implications for daily clinical practice where time is limited. Radiographic damage and progression

At the start of the studies described in this thesis, the modified Stoke AS Spine Score (mSASSS)

was considered the most reliable and sensitive to change scoring method to

assess structural damage in AS 8. After this formal comparison between scoring methods had taken place, a slightly modified version of the mSASSS, the Radiographic AS Spinal Score (RASSS) 9 was developed. The RASSS - in addition to the mSASSS - includes the lower thoracic vertebrae, under the hypothesis that most progression is found in these segments 9. Before starting our long-term analysis of radiographic damage in OASIS, we have conducted a comparison between the two radiographic scores. As described in chapter 6, both

scoring methods were compared with respect to the validation criteria proposed by the Outcome Measures in Rheumatoid Arthritis Clinical Trials (OMERACT): 1) truth; (or: does the instrument measure what is intended? is the result unbiased and relevant?); 2) discrimination (or: does the measure discriminate between situations of interest?); and 3) feasibility (or: can the measure be applied easily, is it feasible to use in the context of clinical trials?)

Regarding the truth aspect, we have found that progression in the thoracic vertebrae was not significantly different from what was expected under the assumption that progression occurs in a balanced manner; in other words, progression was not more prominent in the thoracic spine. In terms of discrimination, both scoring methods showed similar effect sizes for progression despite a higher mean level of progression detected by the RASSS (which was expected because of the additional thoracic vertebrae included). Similar effect sizes for both measures, however, imply that the increased signal detected with the RASSS was offset by increased â&#x20AC;&#x2DC;noiseâ&#x20AC;&#x2122; (higher level of variance). Both methods showed similar reliability. Feasibility of the RASSS was strongly hampered by a lower availability (36% less compared to the mSASSS) of lower thoracic vertebrae on x-rays. Furthermore, in one third of the Summary and conclusions

149

radiographs in which the RASSS could be calculated, only one to two thoracic vertebral corners were accessible, which means that the calculation of the RASSS was based on imputed -and thus non-informative- vertebral corners in the lumbar spine (up to 3 missing corners per spine segment (cervical or lumbar) were allowed; if more vertebral corners were missing, the radiograph could not be included). In one third of the cases where RASSS was computed, the missing vertebral corners included the thoracic vertebrae. In particular feasibility aspects of the RASSS compared to the mSASSS fall short, and this is in our opinion not outweighed by better discrimination/reliability. In conclusion, therefore, the mSASSS remained to us the most appropriate scoring method to assess radiographic progression in patients with AS. Consequently, for all the studies including radiographic damage (chapters 7-10), which

were conducted with 12 year data from OASIS, we have used radiographic scores obtained by the mSASSS. In chapter 7, we have investigated the course of radiographic damage over

time. Previous findings have been confirmed by us and radiographic progression was found

to take place at a relatively ‘slow pace’. Progression is highly variable at the individual patient level 11. Approximately one quarter of the patients in OASIS did not show progression; one quarter had a high level of progression (arbitrarily defined as at least one 2 year interval with a progression of ≥5 mSASSS units); and half of the patients had an intermediate progression rate (≥2 mSASSS units per 2 years). Remarkably, radiographic progression seems to be entirely independent of disease- or symptom duration. Periods of steep progression can be found at any time in the course of the disease, even decades after disease onset, and after periods of relative or complete quiescence. Furthermore, 60% of the patients have developed at least one syndesmophyte over a period of 12 years, implying that syndesmophytes may occur in the vast majority of patients and not only in a bad-prognosis subgroup. So, even though radiographic progression may occur at a relatively slow pace, the chronic character of the disease and the rather unpredictable course in an individual patient assure an accrual of substantial damage over time in most patients. Since damage will –among others- lead to impairment in functional ability 12, this may have direct clinical consequences. Using longitudinal data analysis, we have investigated for the first time the course of radiographic damage over time at the group level. We have used generalised estimating equations (GEE) and found an approximately linear course of progression of radiographic damage, with a remarkably stable progression rate of approximately 1 mSASSS unit per year. Radiographic progression was found to be more severe in HLA-B27 positive men (1.2 mSASSS units per year) and, as expected, in patients with a higher level of radiographic damage present at baseline (1.4 mSASSS units per year in patients with a baseline mSASSS ≥10 mSASSS units, the population median). In these groups of patients, physicians should be particularly alert for radiographic progression. 150

Chapter 11

In chapter 8 we have analysed syndesmophyte formation in more detail. We were

particularly interested in analyzing whether erosion, sclerosis and/or squaring as assessed

on radiographs precede the development of syndesmophytes. These abnormalities are all scored at the vertebral corner level and are part of the mSASSS method. But there is also an important pathophysiological connotation. The presence of erosion, sclerosis and/or squaring in a vertebral corner is scored as 1 and syndesmophytes are scored as 2 (or 3, if bridging), presuming that the former precede the latter, but this had never been formally investigated. Using the 12 year OASIS data, we have conducted a multilevel analysis, in order to adjust for the different levels (i.e. 24 vertebral corners) occurring in the same patient (within-patient correlation). In order to correctly assess whether erosions, sclerosis and/or squaring indeed precede syndesmophyte (bridging or not) formation, we have used 2 year time lags between the detection of erosion, sclerosis and/or squaring on the one hand, and the detection of a syndemophyte at the same level on the other hand. We have found that, while erosion, sclerosis and/or squaring were rather infrequently found (they form only 8-10% of the scores in the vertebral corners), their occurrence often precedes the detection of a new syndesmophyte (Odds Ratio (95% CI): 2.0 (1.7 - 2.3)). When split up, erosions (2.1 (1.6-2.8)) and sclerosis (5.1 (3.9 - 6.8)) but not squaring were statistically significantly associated with subsequent new syndesmophyte formation. According to a hypothesis formulated by Sieper et al., structural damage may start with osteodestruction (erosion), being a consequence of inflammation, followed by a repair response with ossification, the osteoproliferative phase 13. Our findings are in support of this theory that links inflammation to new bone formation via an intermediate step of osteodestruction in AS. Another theory postulates that a mechanical or inflammatory

trigger may lead to a local inflammatory response triggering a subsequent repair reaction perpetuating a bone anabolic response 14. As postulated in chapter 8 erosion and sclerosis may reflect parts of this ongoing process where erosion reflects destruction, and sclerosis may be the first step of the bone proliferation

process. On the basis of these findings we hypothesize that erosions or sclerosis may precede the development of a new syndesmophyte. The 2 year interval of the radiographs that we had to commit to in this study may be too long so that we have missed the ‘real sequence’ of the process of syndesmophyte formation. For similar reasons a hypothetical order of the chain ‘erosion-sclerosis-syndesmophyte’ could not be investigated properly. Whether this sequence can be investigated in near future is doubtful since it may require frequent radiographs in large groups of patients.

Summary and conclusions

151

Relationship between inflammation and radiographic progression

A main aim of this thesis was to explore the longitudinal relationship between inflammation, as measured by disease activity, and radiographic progression. Before the studies of this thesis were conducted, a relationship between C-reactive protein and 2 year radiographic progression had been described

, but such a relationship could not be secured using

the standard measures for AS disease activity

. This fact coincided with the negative

15,16

studies on inhibition of radiographic progression by tumor-necrosis factor-alpha-inhibiting (TNFi) biologicals

. Both observations fuelled theories postulating an ‘uncoupling’ of

17-19

inflammation and bone formation

. The OASIS study provided an excellent setting to

20,21

unravel this enigmatic relationship. It combined a long follow-up, with up to 7 sets of 2 yearly radiographs per patient, with an extensive assessment of disease activity not only with BASDAI, but also with the AS Disease Activity Score (ASDAS). Such a detailed longitudinal set-up allows longitudinal data-analysis that helps in detecting subtle relationships. As described in chapter 9, we have first investigated whether baseline disease activity was associated with the development of radiographic damage over time (baseline analysis). We

have found that baseline disease activity significantly modified the course of radiographic damage over time. This means that patients who have a higher disease activity at a given time point will have a higher level of progression of structural damage over the following years. Patients with inactive disease (ASDAS < 1.3) at baseline had an average progression of 0.7 mSASSS units every 2 years, whereas patients with very high disease activity (ASDAS > 3.5) at baseline had a progression of 3.1 mSASSS units/2 years. A similar relationship was found with increasing levels of baseline BASDAI, but the ‘dose-effect’ was less evident: patients with a BASDAI <4 at baseline had an average progression of 1.4 mSASSS units/2 years, while patients with a BASDAI ≥4 (2.7 mSASSS units/2 years) or those with a BASDAI >6 (2.0 mSASSS units/2 years) at baseline experienced more progression over time. In a second set of analyses (longitudinal analysis), time lagged autoregressive models were used, which technically enabled us to investigate the relationship between disease activity at a given time point and radiographic progression in the subsequent 2 years. It has for the first time been shown that disease activity unequivocally contributes to radiographic progression in the spine in AS: an increase in disease activity is followed by an increase in radiographic progression. The effect of disease activity on radiographic damage is actually rather impressive: an increase of one ASDAS unit in an individual patient is expected to lead to an increase of 0.7 mSASSS units progression over the next 2 years. Further, a patient with very high disease activity (ASDAS > 3.5) may in comparison to a patient with inactive disease (ASDAS < 1.3) expect an additional progression of 2.3 mSASSS units in the subsequent 2 years. Similar conclusions could be drawn for all disease activity measures used (i.e. ASDAS, BASDAI, CRP and/or patient’s global assessment of disease activity). Models with ASDAS 152

Chapter 11

as a disease activity measure showed a better fit than models with all other disease activity measures. Furthermore, there was a clearer ‘dose-effect relationship’ in the analyses with ASDAS: Unlike BASDAI states of disease activity, increasing ASDAS disease activity states 22

were best associated with increased radiographic progression. This superb performance

of ASDAS in comparison to other disease activity measures adds to the validity of ASDAS as the disease activity measure of choice. Moreover, the described relationship between disease activity and radiographic progression may provide an additional argument to pursue treat-to-target in AS/axial spondyloarthritis (SpA), in accordance with what has recently been proposed 23. Having established the relationship between disease activity and radiographic progression, we were then particularly interested in identifying factors that may influence this relationship. First, we have demonstrated that this relationship was more pronounced in men and in patients with shorter symptom duration (less than the median of 18 years) (chapter 9): While

an increase of one ASDAS unit led to an increase of 0.98 mSASSS units/2 years in men, it did not measurably affect 2 year progression in women. Progression rates in women were particularly low, which may reflect a minimal progression in this group of patients but also a more unstable regression coefficient because the group of women was small. Similarly, in patients with shorter symptom duration (< 18 years) an increase of one ASDAS unit led to an increase of 0.84 mSASSS units/2 years, compared to only 0.16 mSASSS units/2 years in patients with longer symptom duration: It looks as if the detrimental effects of inflammation on radiographic progression extinguishes over time, a finding that may not be unexpected to experienced clinicians.

Although disease activity undoubtedly leads to radiographic progression, it should be emphasized that an important part of radiographic progression occurs in patients without any measurable disease activity (chapter 9). This underlines that syndesmophyte formation

in AS is still not yet a fully explained process. One of the still unknown contributory factors

to explaining radiographic progression is mechanical stress. Mechanical forces may also have a role in osteophyte formation in osteoarthritis 24. The OASIS cohort allowed us to further investigate the contribution of work-related mechanical forces in explaining radiographic damage since information on paid work has been collected routinely (chapter 10). In this

study we have taken ‘job type’ (a physically demanding job versus a more sedentary job) as a proxy for ´life-time mechanical stress’. Several confounding factors, such as smoking and socio-economic status, were taken into account. The longitudinal model described in chapter 9 was used as the template for the analysis. In chapter 10, we have shown that longterm physically demanding activities, operationalized as ‘physically demanding (‘blue collar’)

job type amplified the detrimental effects of disease activity on radiographic progression, in comparison with a more sedentary (‘white collar’) job type: In ‘blue-collar’ workers versus Summary and conclusions

153

‘white collar’ workers every additional unit of ASDAS resulted in an increase of 1.2 vs. 0.2 mSASSS units/2 years (p=0.014 for the difference). These findings are in support of recent observations in animal models, showing that ‘mechanical strain’ leads to new bone formation 25

. However, similar effects on the relationship between disease activity and radiographic

progression were found for smokers and for patients with a lower socio-economic status (mainly measured as lower personal income). In a subsequent analysis we have tried to disentangle the effects of ‘job type’ and ‘smoking’ by analysis in relevant subgroups: In the subgroup of smokers, there was slightly (but not significantly) more effect of ASDAS on progression in ‘blue collar’ workers (1.5 mSASSS units 2 year progression per ASDAS unit) than in ‘white collar’ workers (1.2 mSASSS units). In the subgroup of non-smokers, ‘blue collar’ workers had a 2 year progression of 0.6 mSASSS units per ASDAS unit, but the model did not provide a final solution for ‘white collar’ workers (too few remaining patients and too little progression). So this study cannot provide final resolution and larger studies are needed. It may be that smoking and socio-economic status should be considered as confounders of a relationship that includes the relevance of mechanical forces rather than as (pathophysiological) determinants themselves. The main reason for this is ‘biological plausibility’

. It is well known that ‘blue collar’ workers are more frequently smokers than

‘white collar’ workers

and that ‘blue collar’ workers have on average a lower income than

‘white collar’ workers (and our findings were consistent with this). Implications of findings and a perspective on future research

In the studies described in this thesis a lot of focus has been laid on two important themes in the broader field of axial SpA: 1. The assessment and interpretation of spinal mobility in AS 2. The link of inflammation and inappropriate bone formation in AS Both themes will be discussed below with regard to their impact on future research in the broader field of axial SpA. Of importance here is that our main observations pertain to patients with AS, but that recent research has focused more on patients with axial SpA, that includes AS but is not synonymous. The assumption –that will be on the research agenda for many aspects of the disease anyway– here is that findings in AS can be generalised to the entire field of axial SpA. 1. The value of spinal mobility assessment

Historically, spinal mobility has always been a very appealing topic in the field of AS, likely because it could be measured in so many different manners, including so many different

parts of the spine. But ‘multitude’ is not necessarily an advantage. Many of the proposed 154

Chapter 11

spinal mobility measures lack sufficient inter-reader reliability; we do not know if spinal mobility measures can be used to monitor within-patient change; until recently, we did not know about spinal mobility in the ‘normal population’; a multitude of available assessments invites to pick out the most convenient one, and so on. Apart from these metric characteristics of spinal mobility assessments, there is a fundamental flaw in the available knowledge regarding ‘truth aspects’ of measurements. Do they really measure what they intend to measure, and are impairments truly relevant with regard to longterm outcome? Do changes in the level of impairment, eg. induced by medicines, truly reflect an improvement in health? It is this type of questions that deserve attention in the research of patients with axial SpA. Now we slowly deviate from the narrow disease AS towards the broader disease axial SpA, and tend to pick up patients far earlier than before, due to improved classification and diagnosis, we will find more and more patients with still normal spinal mobility measures. Using the MOBILITY-based normal values, we are better suited to follow-up these patients with regard to spinal mobility, determine if the occurrence or worsening of spinal mobility impairment has clinical relevance in axial-SpA patients, find out if treatment-induced change is also clinically relevant, and so on. May be more important, even, is to expand on the observations made in OASIS suggesting to us that there is a strict order in the measures that get consecutively impaired. Careful prospective analysis of existing cohorts with patients with axial SpA with short symptom duration will inform us if this order in AS extends to axial SpA. That is important information

not only from a clinical perspective but also in order to guide pathophysiological research and research on imaging: One may hypothesize that the order in which measurements ‘fall out’ has implications for the determination of the most promising site of interest in pathophysiological research, and in this way ‘simple assessment’ of spinal mobility may guide basic scientific research in axial SpA. On a second note, it would be wise to seek consensus about which measures should be used in busy clinical practice, but also in research studies, in order to optimally make use of the limited time. It may even be possible to design a better index for measuring spinal mobility impairment. 2. Inflammation and syndesmophyte formation

The link we have proven in this thesis between inflammation and structural progression has clearly filled an unexplained lacuna in our knowledge about the consequences of inflammatory process in AS.

This link is close to ‘dogmatic’ in chronic inflammatory

Summary and conclusions

155

rheumatology: No-one ever disputes the statement that inflammation leads to radiographic progression in diseases such as rheumatoid arthritis (RA) and psoriatic arthritis. While there is ample pathophysiological argument to explain an absence of such a relationship in AS (eg. radiographic progression is bone formation rather than bone destruction) many clinical investigators will feel ‘reassured’ that a disease like AS also behaves like a classic inflammatory rheumatic disease. But not only that: The proof of such a connection opens up the gate again to further investigate the effects of profound suppressors of inflammation on syndesmophyte formation. Fortunately, we have those ‘suppressors’ available for research, but classic trials in AS have shown that TNFi-drugs do not measurably inhibit 2 year radiographic progression. The question that awaits resolution is whether ‘no effect on progression’ is simply true, or whether a methodological ‘deficit’ is at the basis of this finding. In our attempts to clarify a link between inflammation and radiographic progression we have developed a longitudinal model for analysis that can perfectly serve as a template to further investigate this link in patients with axial SpA and potentially to investigate the effects of drugs on this link. Like we have found ‘job type’ (blue collar work vs white collar work) to be of influence on the longitudinal relation between disease activity and radiographic progression, we could also investigate the longitudinal effects of medicines

(such as non-steroidal

anti-inflammatory drugs (NSAIDs) or TNFi or non-TNFi biologicals) on this relationship. What we need for that is a sufficiently long follow-up (eg. 10 years or more) and repeated measurements of structural abnormalities. This analytical template is unprecedented in many ways. First, it has allowed us to investigate the contributory role of ‘mechanical forces’ on the relationship between disease activity and syndesmophyte formation, but also of other, at first glance unrelated, factors such as smoking and socio-economic factors. Apparently, the analytical model that we have developed works! The contributory role of ‘mechanical forces’, in our studies approximated by ‘job type’, sets the stage for in-depth research into the pathophysiological mechanisms behind syndesmophyte formation. Does syndesmophyte formation indeed resemble osteophyte formation in osteoarthritis? To what extent is the pathophysiology of osteoarthritis the same as that of axial SpA? Is syndesmophyte formation indeed an inappropriate reaction of the vertebral bodies to mechanical stress? But such a complicated association may also have implications for clinical practice: Since decennia we have recommended patients with AS to intensively perform regular exercises, in order to improve or maintain mobility. If true, the proven associations between disease activity, mechanical forces and radiographic progression may result exactly into what doctors try to avoid: Development of structural damage. Careful analysis in well-designed cohorts with meticulous follow-up will have to give resolution. 156

Chapter 11

Second, the template we have proposed may serve to disentangle complex pathophysiological relationships that have an epidemiological basis. The best example is the effect we have established for ‘smoking’ being an effect modifier of the relationship between disease activity and radiographic progression. This effect was similar to the effect of ‘job type’. But insufficient statistical power prevented us from investigating into detail which factor is truly important: Mechanical forces, smoking or both? Obviously, these factors are not independent: Patients with a ‘blue collar’ job type more often have a smoking history, and the same is true for ‘socioeconomic status’. We should not forget that ‘socioeconomic status’ is a determinant of many important predictive relationships in rheumatology. May be the truth is that ‘socioeconomic status’ is a reflection of some other biologically more plausible factor that awaits careful analysis to be disentangled from its confounders. Third, the longitudinal model we have developed out of necessity to prove a subtle relationship in a disease with very slow progression may have value in other inflammatory disease in- and outside rheumatology. Usually, prediction models have been relatively simple coupling a ‘baseline factor’ (such as disease activity) to subsequent radiographic progression. Such models suffice if this relationship is rather strong, as for disease activity vs. radiographic progression in RA. Sometimes, though, it may prevent investigators from factoring-in sufficient ‘detail’: The observation that ‘mechanical forces’ modify the relationship between disease activity and radiographic progression in AS has important implications for the pathophysiology. Here, ‘confounding’ and effect modification loose their negative connotations, and open up interesting new angles. In analogy, the application of our analytical template in cohorts of patients with RA, undifferentiated arthritis or ‘pre-RA’ may help unraveling important pathophysiological mechanisms. We will work out one potentially

relevant example. Suppose, the question to be addressed is if the presence of a certain genetic polymorphism predisposes to a ‘bad outcome’ defined as radiographic progression above a certain level in patients presenting with undifferentiated arthritis (UA). Usually, one investigates the presence vs the absence of such a factor with regard to a relevant outcome in a large cohort of patients with UA using logistic regression that allows adjusting for multiple factors that may also influence that outcome. Such an analysis is only successful if the contribution of the genetic factor is very ‘strong’, which is hardly if ever the case in this type of research. In the analytical template we have proposed here, we would make use of prior knowledge establishing the relationship between disease activity and radiographic progression. We would establish that relationship in a longitudinal cohort preferably including multiple coupled observations of disease activity and radiographic damage per patient. In such a setup we would investigate the interaction between disease activity and the genetic factor on Summary and conclusions

157

radiographic progression rather than the influence of the genetic factor itself on radiographic progression (with or without adjustment for disease activity). Such a set-up implies a paradigm-shift in predictive research of chronic diseases. But in our experience this approach is far more suitable to find subtle but relevant predictors in an arena in which appropriate treatment will prevent you to study the natural outcome of disease and its predictors. In summary, the studies reported in this thesis have emphasized two important themes: 1) the value of spinal mobility assessment; and 2) the relevance and implications of the established link between inflammation and syndesmophyte formation in AS. We have described here how these findings may affect our current thinking about the pathophysiology and clinical care of patients with axial SpA, and how they may provide guidance to future clinical epidemiological and pathophysiological studies in the field of axial SpA.

158

Chapter 11

REFERENCES 1.

12.

Landewe R, Dougados M, Mielants H, et al. Physical function in ankylosing spondylitis is independently determined by both disease activity and radiographic damage of the spine. Ann Rheum Dis 2009;68:863-7.

Zochling J, van der Heijde D, Burgos-Vargas R, et al. ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis 2006;65:442-52.

13.

Sieper J, Appel H, Braun J, et al. Critical appraisal of assessment of structural damage in ankylosing spondylitis: implications for treatment outcomes. Arthritis Rheum 2008;58:649-56.

14.

Schett G, Rudwaleit M. Can we stop progression of ankylosing spondylitis? Best practice & research Clinical rheumatology 2010;24:363-71.

15.

16.

van Tubergen A, Ramiro S, van der Heijde D, et al. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23.

17.

Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68 Suppl 2:ii1-44.

18.

Creemers MC, Franssen MJ, vanâ&#x20AC;&#x2122;t Hof MA, et al. Assessment of outcome in ankylosing spondylitis: an extended radiographic scoring system. Ann Rheum Dis 2005;64:127-9.

van der Heijde D, Landewe R, Einstein S, et al. Radiographic progression of ankylosing spondylitis after up to two years of treatment with etanercept. Arthritis Rheum 2008;58:1324-31.

19.

van der Heijde D, Landewe R, Baraliakos X, et al. Radiographic findings following two years of infliximab therapy in patients with ankylosing spondylitis. Arthritis Rheum 2008;58:3063-70.

20. Maksymowych WP, Elewaut D, Schett G. Motion for debate: the development of ankylosis in ankylosing spondylitis is largely dependent on inflammation. Arthritis Rheum 2012;64:1713-9.

Jenkinson TR, Mallorie PA, Whitelock HC, et al. Defining spinal mobility in ankylosing spondylitis (AS). The Bath AS Metrology Index. J Rheumatol 1994;21:1694-8.

WHO Multicentre Growth Reference Study Group. WHO Child Growth Standards: Length/height-forage, weight-for-age, weight-for-length, weight-forheight and body mass index-for-age: Methods and development. Geneva: World Health Organization, 2006 (312 pages).

10.

Boers M, Brooks P, Strand CV, et al. The OMERACT filter for Outcome Measures in Rheumatology. J Rheumatol 1998;25:198-9.

11.

21. Lories RJ, Dougados M. Inflammation and ankylosis: still an enigmatic relationship in spondyloarthritis. Ann Rheum Dis 2012;71:317-8. 22. Machado P, Landewe R, Lie E, et al. Ankylosing Spondylitis Disease Activity Score (ASDAS): defining cut-off values for disease activity states and improvement scores. Ann Rheum Dis 2011;70:47-53. 23. Smolen JS, Braun J, Dougados M, et al. Treating spondyloarthritis, including ankylosing spondylitis and psoriatic arthritis, to target: recommendations of an international task force. Ann Rheum 2014;73:616. 24. van der Kraan PM, van den Berg WB. Osteophytes:

Summary and conclusions

159

relevance and biology. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society 2007;15:237-44. 25. Jacques P, Lambrecht S, Verheugen E, et al. Proof of concept: enthesitis and new bone formation in

160

Chapter 11

spondyloarthritis are driven by mechanical strain and stromal cells. Ann Rheum Dis 2014;73:437-45. 26. Kotz D, West R. Explaining the social gradient in smoking cessation: itâ&#x20AC;&#x2122;s not in the trying, but in the succeeding. Tobacco control 2009;18:43-6.

Summary and conclusions

161

Samenvatting en conclusies

SAMENVATTING EN CONCLUSIES Axiale spondyloartritis (axSpA) is een chronische reumatische ziekte, die gekenmerkt wordt door ontstekingen van de gewrichten die het bekken verbinden met de wervelkolom en/of ontstekingen van de gewrichten van de wervelkolom zelf. AxSpA presenteert zich meestal met rugpijn, die vaak ’s nachts optreedt. De rugpijn verbetert met bewegen, maar niet met rust, en is geassocieerd met stijfheid. Daarnaast komen andere symptomen voor, zoals gewrichtsontstekingen (die gepaard gaan met pijn en/of zwelling), oogontstekingen (uveïtis genoemd) en ontstekingen in de darmen (die gepaard gaan met diarree en geassocieerd zijn met een inflammatoire darmaandoening die bekend staat als de ziekte van Crohn). AxSpA is een breed concept dat enerzijds bestaat uit niet-radiografische axSpA, waarbij patiënten de karakteristieke klachten hebben zoals hierboven beschreven maar zonder radiografische sacroiliitis (schade aan de gewrichten van het bekken die zichtbaar is op röntgenfoto’s), en anderzijds uit radiografische axSpA, waarbij patiënten radiografische sacroiliitis hebben en dat ook wel bekend staat als Ankyloserende Spondylitis (AS) of de ziekte van Bechterew. Alhoewel we nog geen studies hebben met een lange follow-up duur, weten we dat een belangrijk deel van de patiënten met niet-radiografische axSpA uiteindelijk AS zullen ontwikkelen. De afgelopen jaren is de manier waarop we patiënten met axSpa evalueren, ook wel ‘outcome assessment’ genoemd, enorm verbeterd. De Assessment of SpondyloArthritis international Society (ASAS), een internationale groep van experts op het gebied van SpA, heeft gedefinieerd welke uitkomstmaten geëvalueerd zouden moeten worden bij patiënten met axSpA/AS. Dit heeft mede de vooruitgang bepaald die er geboekt is op het gebied van axSpA gedurende de laatste jaren. Het standaardiseren van de te evalueren uitkomstmaten is namelijk onmisbaar in het beter begrijpen van de gevolgen van een chronische ziekte. Ziekteactiviteit (de mate van ontsteking en klachten ten gevolge van de ziekte), beweeglijkheid van de wervelkolom en radiografische schade (schade aan botten en/of gewrichten zichtbaar op röntgenfoto’s, bijvoorbeeld in de wervelkolom) zijn belangrijke uitkomstmaten bij axSpA. De structurele schade die te zien is op röntgenfoto’s van de wervelkolom bij AS wordt met name veroorzaakt door botnieuwvorming, de zogenaamde syndesmofyten. Dit is botvorming in de hoek van een wervel, waardoor een overbrugging kan ontstaan tussen twee opeenvolgende wervels. Er zijn scores voor het beoordelen van radiografische schade van de wervelkolom (gestandaardiseerde manieren om bot- of gewrichtsschade op een röntgenfoto te beoordelen) ontwikkeld en getest, wat ons in staat stelt om op een betrouwbare manier botnieuwvorming te meten. Aan het begin van de studies die zijn gedaan in het kader 164

Chapter 12

van dit proefschrift, was al bekend dat de aanwezigheid van syndesmofyten de belangrijkste voorspeller is van toekomstige radiografische progressie, of in andere woorden: de aanwezigheid van botnieuwvormingen leidt tot verdere botnieuwvorming. De verbanden tussen verschillende uitkomstmaten waren ook onderzocht, en uit dit onderzoek bleek dat ontstekingswaarden die we kunnen meten in het bloed, zoals ‘C-reactive protein’ (CRP) en bezinking (BSE), tot op bepaalde hoogte ook bijdragen aan radiografische progressie. Opvallend is echter dat ziekteactiviteit gemeten met behulp van de Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), een volledig door patiënten gerapporteerde index (een vragenlijst die ingevuld wordt door een patiënt, waarbij de antwoorden samen een score vormen, die de ernst van de klachten weerspiegelt), niet gerelateerd bleek te zijn aan radiografische progressie (oftewel botnieuwvorming). Hierdoor werd het verband tussen ziekteactiviteit en radiografische progressie in twijfel getrokken. ‘Biologicals’ zijn relatief nieuwe medicijnen, die zeer effectief zijn in het onderdrukken van zowel de klachten van AS, als ook de ontstekingswaarden (CRP en BSE). Deze middelen vertraagden echter niet de radiografische progressie, waardoor het verband tussen ontsteking en botnieuwvorming ook in twijfel werd getrokken. Andere studies hebben eerder laten zien dat het mannelijk geslacht en roken beiden zijn geassocieerd met radiografische progressie. Radiografische schade op zijn beurt, bleek geassocieerd te zijn met een verminderde beweeglijkheid van de wervelkolom. Er waren verschillende oorzaken waardoor er geen duidelijk beeld ontstond wat de belangrijkste factoren zijn die radiografische progressie bepalen: 1) het meten van ziekteactiviteit was niet optimaal; 2) studies waren te kort; en 3) data-analyse was suboptimaal. 1.

Het meten van ziekteactiviteit.

Recent werd het meten van ziekteactiviteit verbeterd door het introduceren van de AS Disease Activity Score (ASDAS), waarin nu door patiënten gerapporteerde uitkomsten en ontstekingswaarden in het bloed (CRP of BSE) worden gecombineerd. Het is bewezen dat de ASDAS als meetinstrument voor ziekteactiviteit beter presteert dan de BASDAI (een alleen door patiënten gerapporteerde uitkomstmaat). 2. Lange termijn data: de ‘Outcome in Ankylosing Spondylitis International Study (OASIS)’. OASIS is tegenwoordig waarschijnlijk het cohort van patiënten met AS met de langste follow-up volgens een vooraf gedefinieerd protocol. Het is een cohort van patiënten met AS, dat gestart werd in 1996 met toentertijd 217 patiënten uit Nederland, België en Frankrijk. Patiënten werden 12 jaar lang zorgvuldig gevolgd. In de eerste 6 jaar van de follow-up werden Samenvatting en conclusies

165

de evaluaties minstens eens per jaar gedaan en daarna elke twee jaar. Deze evaluaties bestonden uit vragenlijsten, klinisch onderzoek, lichamelijk onderzoek, laboratoriumtesten, en röntgenfoto’s van het bekken en de neken lendenwervels (elke twee jaar). Omdat het grootste deel van de patiënten uit OASIS nooit is behandeld met ‘biologicals’, en vanwege de lange, gestandaardiseerde follow-up duur, biedt dit cohort een unieke mogelijkheid om het natuurlijk beloop van AS en de gevolgen op de lange termijn van deze ziekte te onderzoeken. 3. Longitudinale data-analyse. OASIS biedt ons de mogelijkheid om longitudinale analyses te doen, die tot nu toe nauwelijks waren gedaan in studies naar AS. Longitudinale analyse is een techniek voor het analyseren van data, waarbij alle data die verzameld is over een lange tijd wordt gebruikt. Dit maakt deze analyses zeer krachtig, omdat data op een efficiënte manier wordt gebruikt. Deze analyse heeft ook de belangrijke eigenschap dat er wordt gecorrigeerd voor de zogenaamde ‘within-subject correlation’ (namelijk dat uitkomstmaten van patiënten op één punt in de tijd gecorreleerd zijn met dezelfde uitkomstmaat van dezelfde patiënt op een ander punt in de tijd), hetgeen essentieel is bij het analyseren van data van longitudinale studies en bij het gebruiken van data van verschillende observaties in dezelfde patiënt. Longitudinale analyse geeft ons de mogelijkheid om een beter inzicht te krijgen in de ontwikkeling van een effect over de tijd, en om inzicht te krijgen in longitudinale verbanden tussen verschillende parameters. Daarnaast kunnen we in dit soort analyses tot op een bepaalde hoogte een tijdsgerelateerd effect zien, waardoor we dus kunnen kijken naar een oorzakelijk verband (een bepaald kenmerk leidt tot een gevolg op een later tijdstip). Om al deze redenen zijn longitudinale analyses zeer behulpzaam in het onderzoek naar complexe verbanden tussen ziekteactiviteit, beweeglijkheid van de wervelkolom, radiografische progressie en andere determinanten. Samengevat, was er aan het begin van de analytische studies die zijn gedaan in het kader van dit proefschrift de noodzaak om lange termijn gevolgen van AS en longitudinale verbanden tussen verschillende parameters te bestuderen, waarvoor een passende setting beschikbaar was (namelijk het 12-jaars OASIS-cohort) waarin we de geschikte analyses konden uitvoeren. Dit proefschrift

De studies die zijn beschreven in dit proefschrift hebben betrekking op de evaluatie van

AS en de uitkomsten op lange termijn van deze ziekte. In meer detail werden de volgende onderwerpen onderzocht: Ten eerste werden verschillende aspecten van het meten van uitkomsten bij AS besproken en verbeterd, in het bijzonder met betrekking tot ziekteactiviteit, beweeglijkheid van de wervelkolom en radiografische schade. Ten tweede werd het beloop 166

Chapter 12

van radiografische schade over de tijd geanalyseerd en werd een beter inzicht verkregen in het proces van syndesmofytvorming. Tenslotte werden longitudinale relaties tussen ziekteactiviteit, beweeglijkheid van de wervelkolom, radiografische schade en andere potentiële determinanten van deze uitkomstmaten onderzocht. De studies zijn in twee cohorten uitgevoerd: het OASIS cohort (dat hierboven is beschreven) en de MOBILITY studie. De MOBILITY studie onderzocht normale individuen bij wie een normale beweeglijkheid van de wervelkolom werd vermoed; deelnemers met aandoeningen die deze beweeglijkheid mogelijk konden beïnvloeden (zoals rugoperaties en lage rugpijn) werden geëxcludeerd. Per leeftijdscategorie en lichaamslengte werden gelijke aantallen deelnemers geïncludeerd, apart voor mannen en vrouwen, zodat de populatie gelijk verdeeld was wat betreft deze factoren, omdat we verwachtten dat geslacht, leeftijd en lichaamslengte de beweeglijkheid van de wervelkolom beïnvloeden. De beweeglijkheid van de wervelkolom werd één keer onderzocht bij alle deelnemers met behulp van een aantal maten van beweeglijkheid van de wervelkolom. Beoordeling van ziekteactiviteit en functie

In hoofdstuk 2 beschrijven we een methode hoe met missende gegevens in twee door

patiënten gerapporteerde meetinstrumenten om te gaan: één om ziekteactiviteit te meten,

de BASDAI, en de ander om de mate van functioneren te meten, de ‘Bath AS Functional Index’ (BASFI). Dit onderzoek werd uitgevoerd door gebruik te maken van data uit OASIS. Één missend onderdeel (van de zes) van de BASDAI en tot drie missende onderdelen (van de 10) van BASFI konden vervangen worden door de overgebleven onderdelen te middelen. Imputatie door middel van deze techniek resulteerde in goede overeenkomst met de originele data. Tot nu toe waren er geen duidelijke instructies hoe om te gaan met missende onderdelen en de voorgestelde methode is gemakkelijk en kan eenvoudig worden

toegepast in de dagelijkse praktijk. Beweeglijkheid

Tot nu toe was beweeglijkheid van de wervelkolom nooit goed onderzocht bij normale individuen, alhoewel het een centrale uitkomstmaat is bij AS. De aanbevolen metingen om beweeglijkheid van de wervelkolom te onderzoeken bij patiënten met axSpa zijn: ‘chest expansion’ (uitzetting van de borstkas), de gemodificeerde test van Schober, ‘occiput-towall distance’ (afstand tussen het hoofd en de muur), ‘cervical rotation’ (draaien met de nek), en ofwel ‘lateral spinal flexion’ (opzij buigen) ofwel de ‘Bath AS Metrology Index’ (BASMI). De BASMI is een samengestelde maat van de ‘lateral spinal flexion’, de gemodificeerde test van Schober, ‘cervical rotation’, ‘tragus-to-wall distance’ (afstand tussen de oorschelp en de muur) en de ‘intermalleolar distance’ (afstand tussen de binnenkant van de enkels).

Samenvatting en conclusies

167

In het kort is ‘chest expansion’ een maat voor de beweeglijkheid van de borstkas, door het verschil te meten tussen de diameter bij maximale in- en uitademing. Met de gemodificeerde test van Schober kwantificeren we de beweeglijkheid van de lage rug wanneer iemand vooroverbuigt. De ‘occiput-to-wall distance’ en ‘tragus-to-wall distance’ zegt iets over de kromming van het bovenste gedeelte van de wervelkolom en of iemand een toegenomen kromming heeft (een zogenaamde kyfose) waardoor de wervelkolom niet recht kan worden gemaakt. ‘Cervical rotation’ is een belangrijke maat om functie en beweeglijkheid van de nek te meten. ‘Lateral spinal flexion’ kwantificeert de beweeglijkheid van de rug wanneer iemand opzij buigt naar links en naar rechts. De ‘intermalleolar distance’ tenslotte is de afstand tussen de twee enkels bij maximaal spreiden van de benen, en is een maat voor de functie van de heupen. We kenden de ‘normaalwaarden’ of ‘referentiewaarden’ niet van normale individuen voor de verschillende maten van beweeglijkheid van de wervelkolom die vaak worden gebruikt bij het onderzoeken van patiënten met AS. We hebben een studie gedaan bij normale individuen, de MOBILITY studie, waarbij we iedereen excludeerden die mogelijk een ‘abnormale’ beweeglijkheid van de wervelkolom kon hebben, zodat we ‘referentiewaarden’ voor elke maat van beweeglijkheid van de wervelkolom konden afleiden (hoofdstuk 3). Een

gelijk aantal deelnemers, zowel mannen als vrouwen, in leeftijdscategorieën van 10 jaar en

per 10 cm lichaamslengte werden geïncludeerd, zodat de populatie gelijk verdeeld was wat betreft deze factoren, omdat we verwachtten dat deze kenmerken de beweeglijkheid van de wervelkolom beïnvloeden Eerst onderzochten we het effect van leeftijd, lichaamslengte en geslacht op de verschillende maten van beweeglijkheid van de wervelkolom. Er werd een significante afname gezien in alle maten voor beweeglijkheid van de wervelkolom bij een toenemende leeftijd. Een grotere lichaamslengte was geassocieerd met een grotere ‘tragus-to-wall distance’, ‘lateral spinal flexion’, ‘chest expansion’ en ‘intermalleolar distance’. Vrouwen hadden gemiddeld een lagere waarde voor ‘chest expansion’ en ‘cervical rotation’ dan mannen. In hoofdstuk 3, worden leeftijds- (en soms lichaamslengte-) specifieke

referentie intervallen en percentielen gepresenteerd voor elke maat van beweeglijkheid

van de wervelkolom. De curves voor de percentielen zijn vergelijkbaar met de groeicurves waarmee de groei van kinderen wordt gevolgd. De voorgestelde referentie intervallen en percentielcurves zijn een concrete stap vooruit in het meten van uitkomsten bij AS, omdat we nu weten hoe de beweeglijkheid van de wervelkolom is bij normale individuen en we daardoor de uitkomsten van deze maten beter kunnen interpreteren bij patiënten met AS. Uit deze studie bleek ook een belangrijk effect van leeftijd op alle maten van beweeglijkheid van de wervelkolom, hetgeen benadrukt dat hier rekening mee gehouden moet worden wanneer men patiënten opvolgt met AS: een deel van de beperkingen in beweeglijkheid die gedurende de jaren optreedt moet dus niet toegeschreven worden aan de ziekte, maar

168

Chapter 12

is het gevolg van normale veroudering. De voorgestelde percentielcurves zijn ook een instrument voor verder onderzoek in patiënten met AS. De curves kunnen gebruikt worden als een referentie van ‘normaalwaarden’ wanneer men patiënten en normale individuen wil vergelijken. Daarnaast kunnen deze curves ook waardevol zijn in het opvolgen van patiënten met AS, om zo te zien of een patiënt in de loop van de tijd toenemend afwijkt van zijn of haar ‘eigen curve’. Of deze toepassing relevant is in de klinische praktijk zal nog moeten worden onderzocht. In dezelfde MOBILITY studie hebben we ook een imputatie methode ontwikkeld voor de maat ‘lateral spinal flexion’, voor het geval deze maat op een verkeerde manier wordt genoteerd (hetgeen vaak voorkomt). ‘Lateral spinal flexion’ wordt berekend door het verschil te nemen tussen de afstand van de middelvinger tot de vloer in een neutrale positie en de afstand van de middelvinger tot de vloer bij maximaal opzij buigen. Wanneer de afstand van vingertop tot vloer in neutrale positie mist, en alleen de afstand van vingertop tot vloer bij maximaal opzij buigen wordt vastgelegd, kan de echte ‘lateral spinal flexion’ niet worden berekend. Het probleem dat deze, op zich goed gemeten maat, verkeerd was genoteerd, werd vaak gezien in OASIS en dit zal waarschijnlijk ook in andere cohorten gebeuren, en belemmert het berekenen van de ‘lateral spinal flexion’. Wij hebben aangetoond dat de neutrale vingertop tot vloer afstand, een statische uitkomstmaat is en betrouwbaar kan worden benaderd met behulp van de lichaamslengte in een eenvoudige vergelijking (hoofdstuk 4). Deze imputatie

methode bleek goed overeen te komen met de data, en stelde ons in staat om meerdere onjuist vastgelegde waarden voor ‘lateral spinal flexion’ te “herstellen” in OASIS. Tenslotte hebben we de beweeglijkheid van de wervelkolom vergeleken tussen patiënten met AS, met behulp van 12-jaars data uit OASIS, en normale individuen, met behulp van de percentielcurves. Ons doel was om te analyseren of er een vaste volgorde bestaat in de

afname van de maten van beweeglijkheid van de wervelkolom. Dit is nooit goed onderzocht, omdat er geen afkapwaarden bestonden voor normaal versus abnormaal. Zoals beschreven in hoofdstuk 5, werd alleen gekeken naar de maten voor beweeglijkheid van de wervelkolom

die worden aanbevolen door ASAS, zoals eerder uitgelegd. De maten voor beweeglijkheid van de wervelkolom werden ‘verminderd’ genoemd op basis van de afkapwaarden van de leeftijdsspecifieke percentielcurves afgeleid van normale individuen. We ontdekten dat er inderdaad een duidelijke vaste volgorde bestaat waarin de wervelkolom bij AS wordt aangedaan: ‘lateral spinal flexion’ en de gemodificeerde test van Schober zijn het meest frequent aangedaan bij patiënten met AS, gevolgd door de ‘tragus-to-wall distance’ (een maat die vergelijkbaar is met de ‘occiput-to-wall distance’), daarna ‘cervical rotation’, dan ‘intermalleolar distance’ en tenslotte ‘chest expansion’. Deze observatie ondersteunt de hypothese dat de lendenwervels het eerst zijn aangedaan (de onderrug), gevolgd door de borstwervels en daarna de nekwervels (het bovenste deel van de rug). Deze vaststaande Samenvatting en conclusies

169

volgorde waarin de wervelkolom wordt aangetast werd gevonden onafhankelijk van geslacht, ziekteduur of de aanwezigheid of uitgebreidheid van syndesmofyten op baseline, hetgeen allemaal factoren zijn die mogelijk een invloed uitoefenen op de beweeglijkheid van de wervelkolom. Daarnaast konden we ook uit deze studie concluderen dat verminderde beweeglijkheid van de wervelkolom, gedefinieerd als een afname van tenminste één maat van beweeglijkheid van de wervelkolom, kan worden opgespoord door alleen de ‘lateral spinal flexion’ en de gemodificeerde test van Schober te meten. Deze simpele screeningsstrategie zou maar 9% van de mensen missen die een verminderde beweeglijkheid hebben van de wervelkolom, terwijl veel tijd kan worden bespaard doordat niet alle maten hoeven te worden gemeten. Wanneer de ‘lateral spinal flexion’ of de gemodificeerde test van Schober verminderd is, moeten wel alle andere uitkomstmaten worden gemeten om de uitgebreidheid van de vermindering in beweeglijkheid van de wervelkolom te kwantificeren. Deze bevinding kan belangrijke implicaties hebben voor de dagelijkse klinische praktijk, waar de tijd om een patiënt te onderzoeken beperkt is. Radiografische schade en progressie

De ‘modified Stoke AS Spine Score’ (mSASSS) werd algemeen gezien als de beste scoringsmethode om structurele schade in AS te beoordelen en daarom door ASAS geselecteerd. In de mSASSS worden scores van zowel de nek als de lendenwervels meegenomen. Later werd een licht gemodificeerde versie van de mSASSS ontwikkeld, namelijk de ‘Radiographic AS Spinal Score’ (RASSS). De RASSS kijkt, in tegenstelling tot de mSASSS, ook naar de onderste borstwervels, vanwege de hypothese dat de meeste progressie gevonden wordt in dit segment van de wervelkolom. Voordat we een longitudinale analyse gingen doen van de radiografische schade in OASIS, hebben we deze twee radiografische scoringsmethodes met elkaar vergeleken (hoofdstuk 6). We

vonden dat progressie in de borstwervels vergelijkbaar was met de progressie in de neken lendenwervels; in andere woorden, progressie was niet meer prominent aanwezig in de thoracale wervelkolom. Beide scoringsmethodes lieten een vergelijkbaar vermogen zien om progressie op te pikken, alhoewel de RASSS een hogere progressie liet zien (wat ook werd verwacht gezien de additionele borstwervels die in deze score zijn geïncludeerd). Dit impliceert dat het toegenomen ‘signaal’ dat wordt opgepikt met de RASSS samengaat met meer ‘ruis’. Dit betekent dat het moeilijker was voor lezers om het met elkaar eens te worden over de hoeveelheid radiografische progressie in de borstwervels. Beide methodes waren even betrouwbaar (dat wil zeggen: dezelfde resultaten werden gevonden wanneer dezelfde foto door dezelfde lezer twee keer of door twee verschillende lezers werd gescoord). De haalbaarheid van de RASSS werd in belangrijke mate belemmerd door een lagere beschikbaarheid van de onderste borstwervels op röntgenfoto’s. Met name de haalbaarheid van de RASSS schiet tekort in vergelijking met de mSASSS, en dit werd niet

170

Chapter 12

gecompenseerd door een beter onderscheidend vermogen of een betere betrouwbaarheid. Samengevat bleef de mSASSS wat ons betreft dus de meest geschikte scoringsmethode om radiografische progressie te beoordelen in patiënten met AS. Daarom hebben we voor alle studies waarin gekeken wordt naar radiografische schade (hoofdstuk 7-10), gebaseerd op de 12-jaars data uit OASIS, de radiografische scores

gebruikt die werden verkregen met behulp van de mSASSS. In hoofdstuk 7 hebben

we het beloop van radiografische schade over de tijd onderzocht, in andere woorden,

de ontwikkeling van botnieuwvorming. Zoals reeds uit de literatuur bekend was, bleek radiografische progressie relatief langzaam te ontstaan. Progressie is zeer variabel op het niveau van de individuele patiënt. Ongeveer een kwart van de patiënten in OASIS liet helemaal geen progressie zien; een kwart had veel progressie (arbitrair gedefinieerd als tenminste 5 mSASSS units toename in een interval van 2 jaar); en de helft van de patiënten had een gemiddelde progressie (tenminste 2 mSASSS units toename per 2 jaar). Opmerkelijk is dat radiografische progressie volledig onafhankelijk blijkt te zijn van ziekte- of symptoomduur. Periodes van snelle progressie kunnen op elk moment van de ziekte gezien worden, zelfs tientallen jaren na het begin van de ziekte en na periodes van relatieve of complete remissie. Verder ontwikkelde 60% van de patiënten tenminste één syndesmofyt in een periode van 12 jaar, wat impliceert dat syndesmofyten ontstaan bij het merendeel van de patiënten en niet alleen in de subgroep van patiënten met een slechte prognose. Dus, alhoewel radiografische progressie relatief langzaam ontstaat, zorgt het chronische karakter van de ziekte en het onvoorspelbare beloop in een individuele patiënt uiteindelijk voor een accumulatie van significante schade bij de meeste patiënten. Omdat schade onder andere leidt tot een verminderd vermogen om de dagelijkse activiteiten te beoefenen, heeft dit directe klinische consequenties.

Door gebruik te maken van longitudinale data analyse hebben we voor het eerst het beloop van radiografische schade over de tijd in een groep patiënten kunnen onderzoeken. We vonden een bij benadering lineair verloop van de progressie van radiografische schade, met een opmerkelijk stabiele snelheid van progressie van ongeveer 1 mSASSS unit per jaar. Radiografische progressie bleek sneller te zijn in mannen met de genetische marker HLA-B27 (1,2 mSASSS units per jaar) en, zoals verwacht, in patiënten met meer radiografische schade op baseline (1,4 mSASSS units per jaar in patiënten met op baseline ≥10 mSASSS units, het mediaan van de populatie). In deze groep patiënten moeten artsen extra alert zijn op radiografische progressie. In hoofdstuk 8 hebben we de vorming van syndesmofyten in meer detail geanalyseerd. We waren vooral geïnteresseerd in de vraag of erosies, sclerose en/of ‘squaring’, zoals

kan worden gezien op röntgenfoto’s, voorafgaan aan de ontwikkeling van syndesmofyten. Samenvatting en conclusies

171

Deze afwijkingen worden allemaal gescoord op de hoeken van de wervels en zijn onderdeel van de mSASSS methode. De aanwezigheid van erosies, sclerose en/of ‘squaring’ op de hoek van een wervel wordt gescoord als 1 en syndesmofyten als 2 (of 3, wanneer ze overbruggend zijn), er van uitgaande dat dit een opeenvolgend proces is, maar dit was nooit onderzocht. Met behulp van de 12-jaars data uit OASIS vonden we dat, alhoewel erosies, sclerose en/of ‘squaring’ niet vaak voorkwamen (ze vormen slechts 8-10% van de scores op de hoeken van de wervels), hun aanwezigheid vaak voorafging aan de ontdekking van een nieuwe syndesmofyt: een hoek van een wervel met erosie, sclerose en/of ‘squaring’ had een odds van 2 om op die plek binnen 2 jaar een syndesmofyt te vormen. Wanneer ze apart werden bekeken, bleken erosies en sclerose, maar niet ‘squaring’, geassocieerd te zijn met de vorming van nieuwe syndesmofyten. Onze bevindingen ondersteunen theorieën die ontsteking in AS koppelen aan botnieuwvorming, met osteodestructie (dus erosies) als tussenstap. Relatie tussen ontsteking en radiografische progressie

Een belangrijk doel van dit proefschrift was om de longitudinale relatie tussen ontsteking, gemeten door middel van ziekteactiviteit, en radiografische progressie te exploreren. De OASIS studie was een excellente setting om deze geheimzinnige relatie te ontrafelen. Deze studie had namelijk een combinatie van een lange follow-up, met tot 7 sets van tweejaarlijkse röntgenfoto’s per patiënt, met een uitgebreide evaluatie van ziekteactiviteit, niet alleen met BASDAI, maar ook met ASDAS. Zo’n gedetailleerde longitudinale opzet maakt longitudinale data analyse mogelijk, waardoor subtiele relaties kunnen worden gedetecteerd. Zoals beschreven in hoofdstuk 9 hebben we eerst onderzocht of ziekteactiviteit op baseline geassocieerd was met het ontwikkelen van radiografische schade op termijn (baseline

analyse). We vonden dat ziekteactiviteit bij baseline een significante invloed had op het beloop van radiografische schade over de tijd. Dit betekent dat patiënten die een hogere ziekteactiviteit hadden op een bepaald tijdspunt, de jaren daarna een snellere progressie van structurele schade hadden. Patiënten met inactieve ziekte (ASDAS < 1,3) op baseline, hadden een gemiddelde progressie van 0,7 mSASSS units elke 2 jaar, terwijl patiënten met een heel hoge ziekteactiviteit (ASDAS > 3,5) op baseline een gemiddelde progressie hadden van 3,1 mSASSS units per 2 jaar. Een vergelijkbare relatie werd gevonden met een toenemende BASDAI op baseline, maar de ‘dosis-effect-relatie’ was minder evident: patiënten met een BASDAI < 4 op baseline hadden een gemiddelde progressie van 1,4 mSASSS units per 2 jaar, patiënten met een BASDAI ≥ 4 op baseline 2,7 mSASSS units per 2 jaar en patiënten met een BASDAI > 6 op baseline 2,0 mSASSS units per 2 jaar. In een tweede set analyses werden verfijnde longitudinale analyse methoden gebruikt, waardoor we de relatie tussen ziekteactiviteit op een specifiek tijdspunt en radiografische 172

Chapter 12

progressie in de daaropvolgende twee jaar konden onderzoeken. Voor de eerste keer konden we aantonen dat ziekteactiviteit ontegenzeglijk bijdraagt aan radiografische progressie in de wervelkolom bij AS: een toename in ziekteactiviteit wordt gevolgd door een toename in radiografische progressie. Het effect van ziekteactiviteit op radiografische schade is zelfs behoorlijk indrukwekkend: een toename van één ASDAS unit in een individuele patiënt leidt tot een verwachte toename van 0,7 mSASSS units progressie in de daaropvolgende twee jaar. Daarbij kan een patiënt met een zeer hoge ziekteactiviteit (ASDAS > 3,5) in vergelijking met een patiënt met een inactieve ziekte (ASDAS < 1,3) een additionele progressie van 2,3 mSASSS units verwachten in de komende twee jaar. Vergelijkbare conclusies konden worden getrokken voor alle maten van ziekteactiviteit die gebruikt waren (namelijk ASDAS, BASDAI, CRP en/of ‘de algemene indruk over de mate van ziekteactiviteit door de patiënt’). Modellen met ASDAS als een maat voor ziekteactiviteit waren beter dan modellen met alle andere maten voor ziekteactiviteit. Verder was er een duidelijkere ‘dosis-effect-relatie’ in de analyses met ASDAS: in tegenstelling tot de verschillende stadia van ziekteactiviteit bij BASDAI, waren toenemende stadia van ziekteactiviteit bij ASDAS het best geassocieerd met een toename in radiografische progressie. De uitstekende prestatie van ASDAS in vergelijking met andere maten van ziekteactiviteit draagt bij aan de validiteit van ASDAS als maat van ziekteactiviteit van eerste keus. Bovendien kan de beschreven relatie tussen ziekteactiviteit en radiografische progressie een aanvullend argument zijn om te streven naar een strikte controle van de ziekteactiviteit. Na het vaststellen van een relatie tussen ziekteactiviteit en radiografische progressie, waren we met name geïnteresseerd in het identificeren van factoren die deze relatie kunnen beïnvloeden. Ten eerste hebben we laten zien dat deze relatie meer uitgesproken was bij mannen en in patiënten met een kortere symptoomduur (minder dan de mediaan van 18

jaar) (hoofdstuk 9): terwijl een toename van één ASDAS unit leidde tot een toename van

0,98 mSASSS units per 2 jaar bij mannen, had het geen meetbaar effect op de tweejaars progressie bij vrouwen. De snelheid van progressie bij vrouwen was erg laag, wat zou kunnen wijzen op een minimale progressie bij deze groep patiënten, maar ook op een instabieler regressie coëfficiënt, doordat de groep vrouwen klein was. Bij patiënten met een kortere symptoomduur (< 18 jaar) leidde een toename van één ASDAS unit tot een toename van 0,84 mSASSS units per twee jaar, in vergelijking met maar 0,16 mSASSS units per 2 jaar bij patiënten met een langere symptoomduur. Alhoewel ziekteactiviteit ontegenzeglijk leidt tot radiografische progressie, moet worden benadrukt dat een belangrijk deel van radiografische progressie optreedt bij patiënten zonder meetbare ziekteactiviteit (hoofdstuk 9). Dit onderstreept dat syndesmofytvorming in AS nog steeds voor een deel een onbegrepen proces is. Een van de onbekende

mogelijke bijdragende factoren die radiografische progressie zouden kunnen verklaren Samenvatting en conclusies

173

is mechanische stress. Het OASIS cohort bood ons de mogelijkheid om de bijdrage van werk-gerelateerde mechanische krachten aan radiografische progressie te onderzoeken, doordat informatie over betaald werk routinematig was vastgelegd (hoofdstuk 10). In deze studie hebben we ‘type werk’ (fysiek zwaar werk versus een meer zittend beroep) genomen

als maat voor ‘levenslange mechanische belasting’. Er werd daarbij rekening gehouden met verschillende mogelijk interfererende variabelen, zoals roken en socio-economische status. Het longitudinale model dat werd beschreven in hoofdstuk 9 werd gebruikt als

sjabloon voor de analyse. In hoofdstuk 10 hebben we laten zien dat fysiek zwaar werk

voor een lange periode, de eerder beschreven effecten van ziekteactiviteit op radiografische progressie versterken, in vergelijking met een meer zittend beroep (zoals ambtenaar): bij

‘fysiek belastend werk’ versus ‘niet-fysiek belastend werk’ resulteerde elke bijkomende ASDAS unit in een toename van 1,2 versus 0,2 mSASSS units per 2 jaar. Deze bevindingen ondersteunen recente observaties in diermodellen, die laten zien dat ‘mechanische stress’ leidt tot botnieuwvorming. Vergelijkbare effecten op de relatie tussen ziekteactiviteit en radiografische progressie werden echter ook gevonden voor rokers en patiënten met een lage socio-economische status (gemeten als laag persoonlijk inkomen). In een latere analyse hebben we geprobeerd het effect van ‘type werk’ en ‘roken’ te ontwarren door analyses uit te voeren in relevante subgroepen: in de subgroep van rokers was er een licht toegenomen effect van ASDAS op progressie bij personen met ‘fysiek belastend werk’ (1,5 mSASSS units progressie per 2 jaar per ASDAS unit) in vergelijking met personen met ‘niet-fysiek belastend werk’ (1,2 mSASSS units per 2 jaar per ASDAS unit). In de subgroep van nietrokers, hadden personen met ‘fysiek belastend werk ’ een tweejaarlijkse progressie van 0,6 mSASSS units per ASDAS unit. Maar in andere subgroepen bleven er te weinig patiënten over en de progressie was te klein om een verder onderscheid te maken tussen roken en type werk. Hiervoor zijn grotere studies vereist. Samengevat, hebben de studies in dit proefschrift twee belangrijke thema’s beschreven: 1) de waarde van de evaluatie van beweeglijkheid van de wervelkolom; en 2) de relevantie en implicaties van de vastgestelde relatie tussen ontsteking en syndesmofytvorming in AS. Deze studies leveren een bijdrage aan onze huidige denkwijze over de pathofysiologie en klinische zorg voor patiënten met axiale SpA, en kan aanknopingspunten bieden voor toekomstig klinisch epidemiologische en pathofysiologische studies in het veld van axiale SpA.

174

Chapter 12

Samenvatting en conclusies

175

PHD PORTFOLIO Name PhD student: Sofia Ramiro PhD period: 2011 - 2014 Name PhD supervisor: prof. dr. R. LandewĂŠ Courses Factor analysis in London (London School of Hygiene and Tropical Medicine), UK Statistical analysis with missing data in London (London School of Hygiene and Tropical Medicine), UK European League Against Rheumatism (EULAR) Epidemiology Course in Berlin, Germany Advanced Programming with Stata in London (London School of Hygiene and Tropical Medicine), UK Predictive models in Amsterdam (Vrije Universiteit), the Netherlands Presentations Performance of the ankylosing spondylitis disease activity score (ASDAS) in patients under biological therapies in daily practice - results from the Portuguese register Reuma.pt ACR/ARHP Annual Scientific Meeting in Chicago, USA (poster presentation) Predictive factors of response at 12 weeks in patients with ankylosing spondylitis starting biological therapies - results from the Portuguese register - Reuma.pt Portuguese Congress of Rheumatology in Vilamoura, Portugal (oral presentation) EULAR Annual European Congress of Rheumatology in Berlin, Germany (poster presentation) 8th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) The Mobility Study Assessment of SpondyloArthritis international Society in Amsterdam, The Netherlands (oral presentation) How to deal with missing items in BASDAI and BASFI? 8th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) Feasibility of radiographic damage scores in ankylosing spondylitis EULAR Annual European Congress of Rheumatology in Berlin, Germany (poster presentation) Radiographic score for AS: mSASSS vs RASSS 8th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) ACR/ARHP Annual Scientific Meeting in Washington D.C., USA (poster presentation) Evolution of radiographic damage in ankylosing spondylitis over 12 years of follow-up: a longitudinal analysis 8th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation)

176

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

1 ECTS

2013

0.5 ECTS

2011

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

ACR/ARHP Annual Scientific Meeting in Washington D.C., USA (poster presentation) EULAR Annual European Congress of Rheumatology in Madrid, Spain (poster presentation) Spinal mobility measures are dependent on age - the Mobility Study EULAR Annual European Congress of Rheumatology in Berlin, Germany (poster presentation) 8th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) Spinal mobility measures in normal individuals - the Mobility Study ACR/ARHP Annual Scientific Meeting in Washington D.C., USA (poster presentation) Symposium Inflammation & Pain, Espinho, Portugal (poster presentation) Factors influencing the diagnosis of osteoporosis - data from EpiReumaPt Portuguese Congress of Rheumatology in Vilamoura, Portugal (oral presentation) The Mobility Study Assessment of SpondyloArthritis international Society in Houston, USA (oral presentation) Reference intervals of spinal mobility measures in normal individuals - the Mobility Study ACR/ARHP Annual Scientific Meeting in Washington D.C., USA (poster presentation) Symposium Inflammation & Pain, Espinho, Portugal (oral presentation) EULAR Annual European Congress of Rheumatology in Madrid, Spain (poster presentation) Erosions and sclerosis, but not squaring, predict the development of new syndesmophytes: a 12-year longitudinal analysis EULAR Annual European Congress of Rheumatology in Madrid, Spain (oral presentation) ACR/ARHP Annual Scientific Meeting in San Diego, USA (oral presentation) EULAR Recommendations for the treatment of RA - Safety aspects: a systematic literature review EULAR Annual European Congress of Rheumatology in Madrid, Spain (oral presentation) Assessing the standards of training for rheumatology fellows across Europe EULAR Annual European Congress of Rheumatology in Madrid, Spain (oral presentation) Higher disease activity leads to more damage in the early phases of ankylosing spondylitis: 12-year data from the OASIS cohort ACR/ARHP Annual Scientific Meeting in San Diego, USA (oral presentation) EULAR Annual European Congress of Rheumatology in Paris, France (poster presentation) Disease activity in male smokers has a >10-fold amplified effect on radiographic damage in comparison with female non-smokers in ankylosing spondylitis ACR/ARHP Annual Scientific Meeting in San Diego, USA (poster presentation) EULAR Annual European Congress of Rheumatology in Paris, France (poster presentation)

2012

0.5 ECTS

2013

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2013

0.5 ECTS

2012

0.5 ECTS

2013

0.5 ECTS

2012

0.5 ECTS

2013 2013

0.5 ECTS 0.5 ECTS

2013

0.5 ECTS

2013

0.5 ECTS

2013

0.5 ECTS

2013

0.5 ECTS

2013

0.5 ECTS

2014

0.5 ECTS

2013

0.5 ECTS

2014

0.5 ECTS

177

NVR (Nederlandse Vereniging voor Reumatologie) Najaarsdagen in Arnhem, the Netherlands (poster presentation) 9th International Congress on Spondyloarthropathies in Gent, Belgium (oral presentation) Measurement of lateral spinal flexion and Schober is sufficient to be informed about spinal mobility in patients with ankylosing spondylitis: 12 year OASIS results ACR/ARHP Annual Scientific Meeting in San Diego, USA (poster presentation) EULAR Annual European Congress of Rheumatology in Paris, France (poster presentation) NVR (Nederlandse Vereniging voor Reumatologie) Najaarsdagen in Arnhem, the Netherlands (poster presentation) 9th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) Spinal mobility gets impaired in a fixed order in patients with ankylosing spondylitis: 12 year OASIS results ACR/ARHP Annual Scientific Meeting in San Diego, USA (poster presentation) EULAR Annual European Congress of Rheumatology in Paris, France (poster presentation) NVR (Nederlandse Vereniging voor Reumatologie) Najaarsdagen in Arnhem, the Netherlands (poster presentation) 9th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) A physically demanding job may amplify the effect of disease activity on the development of syndesmophytes in patients with ankylosing spondylitis EULAR Annual European Congress of Rheumatology in Paris, France (oral presentation) NVR (Nederlandse Vereniging voor Reumatologie) Najaarsdagen in Arnhem, the Netherlands (oral presentation) 9th International Congress on Spondyloarthropathies in Gent, Belgium (poster presentation) ACR/ARHP Annual Scientific Meeting in Boston, USA (oral presentation) Conferences EULAR Annual European Congress of Rheumatology in London, UK ACR/ARHP Annual Scientific Meeting in Chicago, USA Spondyloarthritis Forum in Porto, Portugal Systemic Lupus Erythemathosus Forum in Lisboa, Portugal Portuguese Congress of Rheumatology in Vilamoura, Portugal EULAR Annual European Congress of Rheumatology in Berlin, Germany NVR (Nederlandse Vereniging voor Reumatologie) Najaarsdagen in Arnhem, the Netherlands 8th International Congress on Spondyloarthropathies in Gent, Belgium ACR/ARHP Annual Scientific Meeting in Washington D.C., USA Symposium Inflammation & Pain, Espinho, Portugal EULAR Annual European Congress of Rheumatology in Madrid, Spain ACR/ARHP Annual Scientific Meeting in San Diego, USA Portuguese Congress of Rheumatology in Albufeira, Portugal EULAR Annual European Congress of Rheumatology in Paris, France

178

2014

0.5 ECTS

2014

0.5 ECTS

2013

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2013

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2011 2011 2012 2012 2012 2012 2012

1.0 ECTS 1.25 ECTS 0.25 ECTS 0.25 ECTS 0.25 ECTS 1.0 ECTS 0.25 ECTS

2012 2012 2013 2013 2013 2014 2014

0.25 ECTS 1.25 ECTS 0.25 ECTS 1.0 ECTS 1.25 ECTS 0.25 ECTS 1.0 ECTS

NVR (Nederlandse Vereniging voor Reumatologie) Najaarsdagen in Arnhem, 2014 the Netherlands 9th International Congress on Spondyloarthropathies in Gent, Belgium 2014 ACR/ARHP Annual Scientific Meeting in Boston, USA 2014

0.25 ECTS 1.25 ECTS

Seminars and workshops Improving your graphs and tables for publication and presentation

2013

0.2 ECTS

20102011

0.5 ECTS

20112012 20122013

0.5 ECTS

Other Member of the International Steering Committee for the 3E initiative on pain management by pharmacotherapy in inflammatory arthritis. Fellow conducting systematic literature review. Member of the Portuguese Steering Committee for the 3E initiative on how to diagnose and manage gout Member of the EULAR Task Force for the EULAR recommendations for the management of rheumatoid arthritis. Fellow conducting systematic literature review. Member of the EULAR Task Force for the EULAR recommendations for the management of psoriatic arthritis. Fellow conducting systematic literature review. Working Group Member of the Emerging EULAR Network

Invited Lectures New biologic treatments in spondyloarthritis at the Spondyloarthritis Forum in Porto, Portugal Basic ideas of statistics and practical analysis of a dataset at the Portuguese Congress of Rheumatology in Vilamoura, Portugal EpiReumaPt - Portuguese Epidemiologic Study of Rheumatic Diseases at the Portuguese Congress of Rheumatology in Vilamoura, Portugal Spondyloarthritis Preceptorship for Novartis in Amsterdam, The Netherlands Methodological challenges in the implementation of a cohort study at the Systemic Lupus Erythemathosus Forum, Lisboa, Portugal Modelling Longitudinal Data at EULAR Annual European Congress of Rheumatology in Paris, France OASIS: the benefits of a 12-year longitudinal cohort at ARC meeting, Amsterdam, the Netherlands From ankylosing spondylitis to axial spondyloarthritis, Rijnland Ziekenhuis, Leiderdorp, the Netherlands Scientific support Polina Putrik, Impact of socioeconomic gradients within and between countries on health of patients with rheumatoid arthritis: lessons from QUEST-RA, Maastricht University Medical Center Polina Putrik, Inequities in access to biologic and synthetic DMARDs across 46 European countries, Maastricht University Medical Center Polina Putrik, Variations in criteria regulating treatment with reimbursed biologic DMARDs across European countries, Maastricht University Medical Center Carmen Stolwijk, Aspects of validity of the self-administered comorbidity questionnaire in patients with ankylosing spondylitis, Maastricht University Medical Center

0.25 ECTS

0.5 ECTS

2014ongoing

0.5 ECTS

2011ongoing

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012

0.5 ECTS

2012 2012

0.5 ECTS 0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2014

0.5 ECTS

2012

1 ECTS

20112013 20112013

1 ECTS

20112013

1 ECTS

179

Carmen Stolwijk, Comparison of tests for lumbar flexion and hip function in patients with axial spondyloarthritis and in normal individuals, Maastricht University Medical Center Ivette Essers, Characteristics associated with the presence and development of extra-articular manifestations in ankylosing spondylitis: 12year results from OASIS, Maastricht University Medical Center Ivette Essers, Do extra-articular manifestations influence outcome in ankylosing spondylitis? 12 year results from OASIS, Maastricht University Medical Center Filipe Araújo, Outcomes assessed in trials of gout and accordance with OMERACT-proposed domains: a systematic literature review, Centro Hospitalar de Lisboa Ocidental, Lisboa, Portugal José Castillo, Work outcome in patients with ankylosing spondylitis: results of 12-year follow-up of OASIS, Maastricht University Medical Center Milla Kviatkovsky, The minimum clinically important improvement and patient acceptable symptom state in BASDAI and BASFI for patients with ankylosing spondylitis, Leiden University Medical Center Antje Neuen, The additional impact of musculoskeletal diseases on the association of multimorbidity with health and health care costs, Maastricht University Medical Center Antje Neuen, Impact of multimorbidity and, particularly of musculoskeletal diseases, on work outcome, Maastricht University Medical Center Polina Putrik, Socio-economic inequities: musculo-skeletal diseases compared to other chronic diseases, Maastricht University Medical Center Polina Putrik, People with lower education, inactive working status and female gender are more likely to have musculoskeletal disease, Maastricht University Medical Center Polina Putrik, Patients with rheumatoid arthritis with lower education and living in poorer countries have higher disease activit: results from the multinational COMORA study, Maastricht University Medical Center Polina Putrik, Level of education is associated with access to biologic DMARDs even in a country with highly developed social welfare (Norway), Maastricht University Medical Center Polina Putrik, Patients with rheumatoid arthritis from wealthier countries perform better on clinical disease activity measures, but tend to show worse person reported outcomes, Maastricht University Medical Center Monika Hifinger, Levels of Fatigue Are Dependent on Country of Residence in Rheumatoid Arthritis: An Analysis Among 3920 Patients from 17 Countries (the COMORA study), Maastricht University Medical Center Monika Hifinger, A discrete choice experiment among rheumatologists to assess how physicians weight the core attributes of available RA treatments across Europe. Do economic considerations and patient´s preferences play a central role?, Maastricht University Medical Center Ana Gherghe, Longitudinal association of the different types of radiographic damage with physical function in patients with rheumatoid arthritis - analysis of the RAPID trials, Leiden University Medical Center Maureen Turina, A psychometric analysis of outcome measures in trials of peripheral spondyloarthritis, Academic Medical Center Féline Kroon, Non-steroidal anti-inflammatory drugs (NSAIDs) for axial spondyloarthritis (ankylosing spondylitis and non-radiographic axial spondyloarthritis), Leiden University Medical Center

180

20112013

1 ECTS

20112013

1 ECTS

20122014

1 ECTS

20112013

1 ECTS

20112014 20122014

1 ECTS

20122014

1 ECTS

20122014 20122014 2012ongoing

1 ECTS

20132014

1 ECTS

20132014

1 ECTS

2013ongoing

1 ECTS

2013ongoing

1 ECTS

2013ongoing

1 ECTS

20132014

1 ECTS

20122014 20122014

1 ECTS

1 ECTS 1 ECTS

1 ECTS

Victoria Navarro, Relationship between types of radiographic damage and disability in patients with rheumatoid arthritis in the EURIDISS cohort: a longitudinal study, Leiden University Medical Center Victoria Navarro, Relationship between sacro-iliac inflammation on magnetic resonance imaging and clinical disease activity measures in axial spondyloarthritis: a longitudinal analysis in the DESIR cohort, Leiden University Medical Center

20132014

1 ECTS

2014

1 ECTS

Grants Doctoral Grant from the Portuguese Foundation for Science and Technology

2011

Awards and Prizes BES Rheumatology Prize 2011 Best research project at the scientific evening from the Rijnland Ziekenhuis

2011 2014

181

LIST OF PUBLICATIONS 1.

Fonseca JE, Silva JA, Canhão H, Santos MJ, Barcelos A, Inês L, Costa ML, Rodrigues M, Bernardo A, Cordeiro A, Cravo AR, Ribeiro A, Teixeira A, Malcata A, Faustino A, Ribeiro C, Nour D, Araújo D, Sousa E, Mariz E, Ramos F, Vinagre F, Ventura FS, Sequeira G, Santos H, Branco JC, Ramos J, Santo JE, Costa JA, da Silva JA, Miranda L, Sampaio L, Afonso MC, Saavedra MJ, Cunha MI, Queiroz MV, Couto M, Pinto P, Valente P, Abreu P, Figueira RO, Ramiro S, Nóvoa T, Bernardes M; Grupo de Estudo de Artrite Reumatóide da Sociedade Portuguesa de Reumatologia. Practical guide for the use of biotechnological therapies in rheumatoid arthritis. Acta Reumatol Port 2009;34:395-9

2. da Silva JAP, Ramiro S, Pedro S, Rodrigues A, Vasconcelos JC, Benito-Garcia E. Patients’ and physicians’ priorities for improvement. The case of rheumatic diseases. Acta Reumatol Port 2010;35:192-9 3. Ramiro S, Tubergen A, Landewé RBM. RAPID and FAST4WARD trials: Certolizumab pegol for rheumatoid arthritis. Exp Rev Clin Immunol 2010;6:713-20 4. Ramiro S, Canhão H, Branco JC. EpiReumaPt Protocol - Portuguese epidemiologic study of the rheumatic diseases. Acta Reumatol Port 2010;35:384-90 5. Ramiro S, Machado P, Singh JA, Landewé RBM, da Silva JAP. Applying science in practice: the Optimization of Biological Therapy in Rheumatoid Arthritis. Arthritis Res Ther 2010;12:220-7 6. Ribeiro A, Machado P, Ramiro S, Duarte C, Mourão AF, Bogas M, Costa L, Bernardes M, Santos MJ, Fonseca JE, Silva JAP, Canhão H. Portuguese recommendations “How to diagnose, follow up and treat peripheral undifferentiated arthritis”: Systematic literature review and expert opinion. Acta Reumatol Port 2010;35:475-496 7. Ramiro S, Radner H, van der Heijde D, van Tubergen A, Buchbinder R, Aletaha D, Landewé R. Combination therapy for pain management in inflammatory arthritis (Protocol). Cochrane Database of Systematic Reviews 2010, Issue 12 8. Radner H, Ramiro S, Buchbinder R, Landewé R, van der Heijde D, Aletaha D. Pain management for inflammatory arthritis and gastrointestinal or liver comorbidity (Protocol). Cochrane Database of Systematic Reviews 2010, Issue 12 9. Roque R, Ramiro S, Cordeiro A, Gonçalves P, da Silva JC, Santos MJ. Development of amyloidosis in patients with rheumatoid arthritis under TNF-blocking agents. Clin Rheumatol 2011;30:869-70 10. Ramiro S, Radner H, van der Heijde D, van Tubergen A, Buchbinder R, Aletaha D, Landewé R. Combination therapy for pain management in inflammatory arthritis (Review). Cochrane Database of Systematic Reviews 2011, Issue 10 11. Bandinelli F, Bijlsma J, Ramiro S, Pia E, Goekoop-Ruiterman YP, Sivera F, Molto-Revilla A, Marinescu C, Rimar D, Norkuviene E, Mayer M, Duarte C, da Silva JA, Cutolo M. Rheumatology education in Europe: results of a survey of young rheumatologists. Clin Exp Rheumatol 2011;29:843-5 12. Roque R, Ramiro S, Vinagre F, Cordeiro A, Godinho F, Santos MJ, Gonçalves P, da Silva JC. Mixed cryoglobulinemia. Acta Reumatol Port 2011;36:298-303 13. Ramiro S, Roque R, Vinagre F, Cordeiro A, Tavares V, van Tubergen A, da Silva JC, Landewé R, Santos MJ. Biologicals and switch in Rheumatoid Arthritis throughout time – are we being more aggressive? Acta Reumatol Port 2011;36:234-42 14. Barbosa L, Ramiro S, Roque R, Gonçalves P, da Silva JC, Santos MJ. Patient’s satisfaction with the rheumatology day care unit. Acta Reumatol Port 2011;36: 377-84 15. Fonseca JE, Bernardes M, Canhão H, Santos MJ, Quintal A, Malcata A, Neto A, Cordeiro

182

A, Rodrigues A, Mourão AF, Ribeiro AS, Cravo AR, Barcelos A, Cardoso A, Vilar A, Brana A, Faustino A, Silva C, Duarte C, Araújo D, Nour D, Sousa E, Simões E, Godinho F, Brandão F, Ventura F, Sequeira G, Figueiredo G, Cunha I, Matos JA, Branco J, Ramos J, Costa JA, Gomes JA, Pinto J, Silva JC, Silva JA, Patto JV, Costa L, Miranda LC, Inês L, Santos LM, Cruz M, Salvador MJ, Ferreira MJ, Rial M, Queiroz MV, Bogas M, Araújo P, Reis P, Abreu P, Machado P, Pinto P, André R, Melo R, Garcês S, Cortes S, Alcino S, Ramiro S, Capela S, Portuguese Society of Rheumatology. Portuguese guidelines for the use of biological agents in rheumatoid arthritis - October 2011 update. Acta Reumatol Port 2011;36:385-8 16. Mourão AF, Fonseca JE, Canhão H, Santos MJ, Bernardo A, Cordeiro A, Cravo AR, Ribeiro A, Teixeira A, Barcelos A, Malcata A, Faustino A, Duarte C, Ribeiro C, Nour D, Araújo D, Sousa E, Mariz E, Ramos F, Vinagre F, Ventura FS, Sequeira G, Santos H, Branco JC, Gomes JA, Silva JA, Ramos J, Santo JE, Costa JA, Silva JA, Ribeiro JS, Inês L, Miranda L, Sampaio L, Costa ML, Rodrigues M, Afonso MC, Cunha MI, Saavedra MJ, Queiroz MV, Couto M, Bernardes M, Bogas M, Pinto P, Valente P, Coelho P, Abreu P, Cortes S, Pimenta S, Ramiro S, Figueira R, Nóvoa T, Grupo de Estudo de Artrite Reumatóide da Sociedade Portuguesa de Reumatologia. Practical guide for the use of biological agents in rheumatoid arthritis - December 2011 update. Acta Reumatol Port 2011;36:389-95 17. van Tubergen A, Ramiro S, van der Heijde D, Dougados M, Mielants H, Landewé R. Development of new syndesmophytes and bridges in ankylosing spondylitis and their predictors: a longitudinal study. Ann Rheum Dis 2012;71:518-23 18. Radner H, Ramiro S, Buchbinder R, Landewé R, van der Heijde D, Aletaha D. Pain management for inflammatory arthritis and gastrointestinal or liver comorbidity (Review) Cochrane Database of Systematic Reviews 2012, Issue 18 19. Whittle SL, Colebatch AN, Buchbinder R, Edwards CJ, Adams K, Engelbrecht M, Hazlewood G, Marks JL, Radner H, Ramiro S, Richards BL, Tarner IH, Aletaha D, Bombardier C, Landewé RB, Müller-Ladner U, Bijlsma JW, Branco JC, Bykerk VP, Pinheiro GRC, Catrina A, Hannonen P, Kiely P, Leeb B, Lie E, Martinez-Osuna P, Montecucco C, Ostergaard M, Westhovens R, Zochling J, van der Heijde D. Multinational evidencebased recommendations for pain management by pharmacotherapy in inflammatory arthritis: integrating systematic literature research and expert opinion of a broad panel of rheumatologists in the 3e Initiative. Rheumatology(Oxford) 2012;51:1416-25 20. Laires PA, Canhão H, Araújo D, Fonseca JE, Machado P, Mourão AF, Ramiro S, Romeu JC, Santos MJ, Silva I, Silva JA, Sousa E, Tavares V, Gouveia N, Branco JC. CoReumaPt Protocol: The Portuguese Cohort of Rheumatic Diseases. Acta Reumatol Port 2012;37:18-24 21. Machado P, Bogas M, Ribeiro A, Costa J, Neto A, Sepriano A, Raposo A, Cravo AR, Vilar A, Furtado C, Ambrósio C, Miguel C, Vaz C, Catita C, Nour D, Araújo D, Vieira-Sousa E, Teixeira F, Brandão F, Canhão H, Cordeiro I, Gonçalves I, Ferreira J, Fonseca JE, Silva JA, Romeu J, Ferreira J, Costa L, Maurício L, Cunha-Miranda L, Parente M, Coutinho M, Cruz M, Oliveira M, Salvador MJ, Santos MJ, Pinto P, Valente P, Abreu P, Roque R, Ramiro S, Capela S, Las V, Barcelos A. Portuguese recommendations for the use of biological therapies in patients with psoriatic arthritis. Acta Reumatol Port 2012;37:26-39 22. Machado P, Bernardo A, Rodrigues A, Malcata A, Nour D, Vieira.Sousa E, Godinho F, Pimentel F, Canhão H, Santos H, Cunha I, Fonseca JE, Costa J, Cunha-Miranda L, Maurício L, Cruz M, Santos MJ, Bernardes M, Bogas M, Valente P, Ramiro S, Barcelos A. Portuguese recommendations for the use of biological therapies in patients with axial spondyloarthritis - December 2011 update. Acta Reumatol Port 2012;37:40-7 23. Ramiro S, Radner H, van der Heijde D, Buchbinder R, Aletaha D, Landewé R. Combination therapy for pain management in inflammatory arthritis. J. Rheumatol Suppl 2012;90:47-55 183

24. Radner H, Ramiro S, van der Heijde D, Landewé R, Aletaha D. How do gastrointestinal or liver comorbidities influence the choice of pain treatment in inflammatory arthritis? J. Rheumatol Suppl 2012;90:74-80 25. Silva I, Mourão AF, Ramiro S, Rodrigues A, Miguel C, Sampaio L, Faustino A, Vaz C, Canhão H, Cruz M, Neto A, Cravo AR, Ribeiro A, Brandão A, Peixoto D, Medeiros D, Santos FA, Teixeira F, Gonçalves I, Silva JP, Gomes JM, Maurício L, Simões E, Salvador MJ, Silva M, Nero P, Roque R, Barros R, André R, Falcão S, Serra S, Tavares V, Branco JC. Portuguese Recommendations for pain management by pharmacotherapy in inflammatory arthritis. Acta Reumatol Port 2012;37:160-74 26. Putrik P, Sokka T, Ramiro S, Boonen A. Impact of socioeconomic gradients within and between countries on health of patients with rheumatoid arthritis: lessons from QUEST RA. Best Pract Res Clin Rheumatol 2012;26:705-20 27. Ramiro S, van Tubergen A, Stolwijk C, Landewé R, van den Bosch F, Dougados M, van der Heijde D. Scoring radiographic progression in ankylosing spondylitis: should we use the modified Stoke Ankylosing Spondylitis Spine Score (mSASSS) or the Radiographic Ankylosing Spondylitis Spinal Score (RASSS)? Arthritis Res Ther 2013;15:R14 28. Barbosa L, Ramiro S, Santos MJ, Canas Da Silva J. Applicability of the EULAR recommendations on the role of the nurse in the management of chronic inflammatory arthritis in Portugal. Acta Reumatol Port 2013;38:186-91 29. Putrik P, Ramiro S*, Kvien TK, Sokka T, Pavlova M, Uhlig T, Boonen A on behalf of the Working Group “Equity in access to treatment of rheumatoid arthritis in Europe”. Inequities in access to biologic and synthetic DMARDs across 46 European countries. Ann Rheum Dis 2014;73:198-206 (*co-first authorship) 30. Smolen JS, Landewé R, Breedveld FC, Buch M, Burmester G, Dougados M, Emery P, Gaujoux-Viala C, Gossec L, Nam J, Ramiro S, Winthrop K, de Wit M, Aletaha D, Betteridge N, Bijlsma JW, Boers M, Buttgereit F, Combe B, Cutolo M, Damjanov N, Hazes JM, Kouloumas M, Kvien TK, Mariette X, Pavelka K, van Riel PL, Rubbert-Roth A, Scholte-Voshaar M, Scott DL, Sokka-Isler T, Wong JB, van der Heijde D. EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2013 update. Ann Rheum Dis 2014;73:492-509 31. Ramiro S, van Tubergen A, van der Heijde D, van den Bosch F, Dougados M, Landewé R. How to deal with missing items in BASDAI and BASFI. Rheumatology (Oxford) 2014;53:374-76 32. Gaujoux-Viala C, Nam J, Ramiro S, Landewé R, Buch MH, Smolen JS, Gossec L. Efficacy of conventional synthetic disease-modifying antirheumatic drugs, glucocorticoids and tofacitinib: a systematic literature review informing the 2013 update of the EULAR recommendations for management of rheumatoid arthritis. Ann Rheum Dis 2014;73:510-5 33. Nam JL, Ramiro S, Gaujoux-Viala C, Takase K, Leon-Garcia M, Emery P, Gossec L, Landewe R, Smolen JS, Buch MH. Efficacy of biological disease-modifying antirheumatic drugs: a systematic literature review informing the 2013 update of the EULAR recommendations for the management of rheumatoid arthritis. Ann Rheum Dis 2014;73:516-28 34. Ramiro S, Gaujoux-Viala C, Nam JL, Smolen JS, Buch M, Gossec L, van der Heijde D, Winthrop K, Landewé R. Safety of synthetic and biological DMARDs: a systematic literature review informing the 2013 update of the EULAR recommendations for management of rheumatoid arthritis. Ann Rheum Dis, 2014;73:529-35 35. Fonseca JE, Gonçalves J, Araújo F, Cordeiro I, Teixeira F, Canhão H, da Silva JA, Garcês S, Miranda LC, Ramiro S, Roxo A, Pimentel-Santos FM, Tavares V, Neto A, Sepriano A, Malcata A, Faustino A, Silva C, Ambrósio C, Duarte C, Miguel C, Barcelos 184

F, Santos H, Cunha I, Ramos JC, Gomes JA, Pimentão JB, Costa L, Maurício L, Silva M, Bernardes M, Bogas M, Coelho PC, Monteiro P, Aguiar R, André R, Leitão R, Pimenta S, Meirinhos T, Fernandes S, Las V, Castelão W; Sociedade Portuguesa de Reumatologia. The Portuguese Society of Rheumatology position paper on the use of biosimilars. Acta Reumatol Port 2014;39:60-71. 36. Ramiro S, van der Heijde D, van Tubergen A, Stolwijk C, Dougados M, van den Bosch F, Landewé R. Higher disease activity leads to more structural damage in the spine in ankylosing spondylitis: 12-year longitudinal data from the OASIS cohort. Ann Rheum Dis 2014;73:1455-61 37. Araújo F, Cordeiro I, Teixeira F, Rovisco J, Ramiro S, Mourão AF, Costa JA, Pimentão JB, Malcata A, Santos MJ, Branco JC. Portuguese Recommendations for the Diagnosis and Management of Gout. Acta Reumatol Port 2014;39:158-71 38. Ramiro S, van Tubergen A, Stolwijk C, van der Heijde D, Landewé R. Neutral lateral fingertip-to-floor distance can be derived from height. Ann Rheum Dis 2014;73:1748-9 39. Ramiro S, van Tubergen A, van der Heijde D, Stolwijk C, Bookelman G, Dougados M, van den Bosch F, Landewé R. Erosions and sclerosis on radiographs precede the subsequent development of syndesmophytes at the same site: A 12-year prospective follow-up of the OASIS study in patients with ankylosing spondylitis. Arthritis Rheumatol 2014;66:2773-9 40. Stolwijk C, van Tubergen A, Ramiro S, Essers I, Blaauw M, van der Heijde D, Landewé R, van den Bosch F, Dougados M, Boonen A. Aspects of validity of the self-administered comorbidity questionnaire in patients with ankylosing spondylitis. Rheumatology (Oxford) 2014;53:1054-64 41. Putrik P, Ramiro S*, Kvien TK, Sokka T, Uhlig T, Boonen A on behalf of the Working Group “Equity in Clinical Eligibility Criteria for RA treatment Working Group”. Variations in criteria regulating treatment with reimbursed biologic DMARDs across European countries. Are differences related to country’s wealth? Ann Rheum Dis 2014;73:2010-21 (*co-first authorship) 42. Ramiro S, Stolwijk C, van Tubergen A, van der Heijde D, Dougados M, van den Bosch F, Landewé R. Evolution of radiographic damage in ankylosing spondylitis: a 12 year prospective follow-up of the OASIS study. Ann Rheum Dis 2015;74:52-9 43. Ramiro S, van Tubergen A, Stolwijk C, van der Heijde D, Royston P, Landewé R. Reference intervals of spinal mobility measures in normal individuals: the mobility study. Ann Rheum Dis published 24 Mar 2014. doi: 10.1136/annrheumdis-2013-204953. [Epub ahead of print] 44. Navarro-Compán V, Landewé R, Provan SA, Odegård S, Uhlig T, Kvien TK, Keszei AP, Ramiro S, van der Heijde D. Relationship between types of radiographic damage and disability in patients with rheumatoid arthritis in the EURIDISS cohort: a longitudinal study. Rheumatology (Oxford) published 26 Jul 2014. doi:10.1093/rheumatology/keu284. [Epub ahead of print] 45. Stolwijk C, Ramiro S, Vosse D, Landewé R, van der Heijde D, van Tubergen A. Comparison of tests for lumbar flexion and hip function in patients with axial spondyloarthritis and in normal individuals. Arthritis Care Res published 3 Sep 2014. doi: 10.1002/acr.22464. [Epub ahead of print] 46. Frank-Bertoncelj M, Hatemi G, Ospelt C, Ramiro S, Machado P, Mandl P, Gossec L, Buch MH. Mentoring of young professionals in the field of rheumatology in Europe: results from an EMerging EUlar NETwork (EMEUNET) survey. Clin Exp Rheumatol published 8 Sep 2014. [Epub ahead of print] 47. Essers I, Ramiro S, Stolwijk C, Blaauw M, Landewé R, van der Heijde D, Van den Bosch F, Dougados M, van Tubergen A. Characteristics associated with the presence and 185

development of extra-articular manifestations in ankylosing spondylitis: 12-year results from OASIS. Rheumatology (Oxford) published 17 Sep 2014. doi:10.1093/rheumatology/ keu388. [Epub ahead of print] 48. Araújo F, Cordeiro I, Ramiro S, Falzon L, Branco JC, Buchbinder R. Outcomes assessed in trials of gout and accordance with OMERACT-proposed domains: a systematic literature review. Rheumatology (Oxford) published 13 Nov 2014. doi:10.1093/ rheumatology/keu424. [Epub ahead of print] 49. Ramiro S, Landewé R, van der Heijde D, Stolwijk C, Dougados M, van den Bosch F, van Tubergen A. Hierarchy of spinal mobility measures in ankylosing spondylitis: 12-year data from the OASIS cohort. (Submitted) 50. Ramiro S, Landewé R, van Tubergen A, Boonen A, Stolwijk C, Dougados M, van den Bosch F, van der Heijde D. Mechanical stress and smoking may modify the effect of disease activity on radiographic progression in patients with ankylosing spondylitis. (Submitted) 51. Ramiro S, Landewé R, van der Heijde D, Harrison D, Collier D, Michaud K. Discontinuation rates of biologics in patients with rheumatoid arthritis: are TNF inhibitors different from non-TNF inhibitors? (Submitted) 52. Sivera F, Ramiro S*, Cikes N, Dougados M, Gossec L, Kvien TK, Lundberg IE, Mandl P, Moorthy A, Panchal S, da Silva J, Bijlsma JW. Differences and similarities in rheumatology specialty training programs across European countries. (Submitted) (*cofirst authorship) 53. Putrik P, Ramiro S, Keszei A, Hmamouchi I, Dougados M, Uhlig T, Kvien TK, Boonen A. Lower education and living in countries with lower wealth are associated with higher disease activity in RA: results from the multinational COMORA study. (Submitted) 54. Putrik P, Ramiro S*, Chorus AM, Keszei A, Boonen A. Socio-economic inequities in health among patients with musculoskeletal diseases compared to other chronic diseases. (Submitted) (*co-first authorship) 55. Turina M, Ramiro S*, Baeten DL, Mease P, Paramarta JE, Song I, Pangan A, Landewé R. A psychometric analysis of outcome measures in trials of peripheral spondyloarthritis. (Submitted) (*co-first authorship) 56. Kroon F, van der Burg L, Ramiro S, Landewé R, Buchbinder R, Falzon L, van der Heijde D. Non-steroidal anti-inflammatory drugs (NSAIDs) for axial spondyloarthritis (ankylosing spondylitis and non-radiographic axial spondyloarthritis). (Submitted)

186

CURRICULUM VITAE Sofia Ramiro was born on May 23rd, 1982 in Lisbon, Portugal. She studied at the German School in Lisbon, where in 2000 she finished high school and obtained her Abitur diploma. In the same year she started studying Medicine at the Faculty of Medical Sciences, New University of Lisbon (currently Nova Medical School). During her studies she did some rotations abroad (Cambridge, UK; Charlottesville, US; Rochester, US). She obtained her medical degree in 2006. In 2007 she worked as a resident at Hospital Egas Moniz. In 2008 she started her training to become a rheumatologist at Hospital Garcia de Orta. In September 2009 she interrupted her clinical training to do a research fellowship in Maastricht, The Netherlands. At the same time, she took a Master in Epidemiology at the University of Maastricht, where she graduated cum laude in 2011. She continued her research work that resulted in this thesis at the Academic Medical Center, University of Amsterdam. During this research period, she followed several courses in statistics and epidemiology, particularly focusing on longitudinal analyses. In 2013 she decided to restart her training in rheumatology, this time at the Leiden University Medical Center (under the supervision of Prof. Dr. Tom Huizinga). She is currently working in internal medicine, the first part of her training, at the Rijnland Ziekenhuis (under the supervision of Dr. Janssen and Dr. Anten).

187

ACKNOWLEDGEMENTS This journey is now reaching its end and it would never have been possible, and certainly never as pleasant or memorable as it has been, without the help of many people. These years have been at least as intense in research work as rich in events and changes in my life that have contributed to my personal development and growth. Even the language in which I speak to several of the people who are named here has changed. For simplicity, to avoid the cacophony of three languages and to ensure that everyone can read these acknowledgements, I will address everyone in English. I would like to start by thanking all the patients and health professionals involved in OASIS who throughout these many years have provided an essential contribution to science with this unique cohort. It is important to emphasize that this research would not have been possible without them. Also a special thanks to all the volunteers who allowed their spinal mobility to be measured for one of the research projects. Combining volunteers from Portugal and the Netherlands - several of which are my friends or friends of my friends - in a research project that I have started from scratch, made it even more special. My dear promotors, Prof. Dr. R. LandewĂŠ and Prof. Dr. D. van der Heijde, dear Robert and DĂŠsirĂŠe, there are really no words to describe how I feel about what we have shared in the last years. Your vast knowledge is inspiring, your enthusiasm for research is contagious, your personal and career guidance a privilege and the way you care about your PhD students including their personal life is heartwarming. Being your PhD student means having a unique opportunity to grow as a researcher and as a person: with all one may receive from you, the sky is truly the limit. You have an unprecedented combination of skills, and even more importantly, a gift for guiding people which extends far beyond their scientific work. Would I have ever been asked to choose the ideal PhD supervisors, I could not have made a better choice. You know you have changed my life and that, among others, you have been my inspiration and motivation to stay in the Netherlands: I am and will always be immensely grateful for this. I really hope this is just the first chapter of many we will write together. Dear co-promotor, Dr. A. van Tubergen, dear Astrid, thank you very much for sharing your experience with me during these last years, for your critical inputs, for always making time for discussions and for reviewing in such detail and in such a timely manner everything I was working on. Writing a paper seems now somewhat less complex than a few years ago and you have definitely contributed to that. I am thankful for the opportunity you have given me of starting co-supervising other people with you, which made me realize how rewarding this can be.

188

During this period I have in some way ‘belonged’ to different places and I will not forget the support I have always received from all of them. All started with my department in Portugal (Hospital Garcia de Orta). Dr. Canas, I will never forget that you allowed me to interrupt my clinical training at a time that may not have been the most convenient for the department. A wholehearted ‘thank you’ for fully supporting me and giving me all the opportunities and encouragement to go further, develop research skills abroad and even stay abroad. Raquel and Filipe, your close friendship and continuous presence in my daily life despite physical distance are amazing and make me feel close to ‘home’. Together with Dr. Viviana, Dr. Pedro and Sandra: what a great group of friends I have! Thanks for always finding a way to spend a good time together during my stays in Lisbon, nice memories! Thank you to the whole Maastricht team - rheumatologists, trainees, PhD students, nurses, secretaries -, who warmly hosted me during my first years in the Netherlands. To Annelies and Polina, and since more recently also to Monika and Antje, with whom I had fascinating and stimulating brainstorms, raising promising ideas for the upcoming future! To Carmen and Ivette. After sharing a lot of work (and frustrations) with OASIS, it feels really good to look back at what we have done together. Carmen, unforgettable adventure: measuring spinal mobility and the meeting of the ‘club of tall people’. To Marloes and Caroline. You were the reason to stop at the trainees’ room always for a nice chat. To the whole amC team - rheumatologists, researchers, PhD students, secretary - thank you very much for making my integration so easy, despite the fact that I was working from a distance. Thank you for making me feeling part of the team. Dear Prof. Dr. J.W. Bijlsma, dear Hans, throughout these years I have had the opportunity to work closer with you in our project on training in rheumatology and it has been a privilege to learn from you how to turn a challenge into an opportunity. Dear Prof. Dr. D. Baeten, dear Dominique, a particular thanks for including me in a project from the amC, making me feel more involved and bringing me into closer contact with Maureen and Jacky. To both of you, thanks for always being available to help me, particularly Jacky, with all the advice for the final phase of this journey. My next stop – to which I already officially ‘belong’ - is the LUMC. Prof. Dr. T. Huizinga, dear Tom, thank you for entrusting me this challenge. To all rheumatologists, trainees, PhD students (including old ones) and secretaries: thank you for already making me feel a part of the group. Many friends are -or have been- at the department and have particularly

189

contributed to make me quickly feel at home when moving to Leiden: Rosaline, Manouk, Pauline, Iris, Karen, Melek, Diane, Emilia, César, Felice and Ana. Thank you! The last parts of my thesis were written while already doing my training in rheumatology, or more specifically the rotation of internal medicine, at the Rijnland Ziekenhuis. Dr. Janssen and Dr. Anten, thank you very much for having accepted the challenge to guide me through this adventure of working as a clinician in the Netherlands. This has turned out to be quite demanding and I want to thank all internists and colleagues for having helped me during this adaptation process and making me feel increasingly at ease seeing patients, despite the additional difficulties that the Dutch language pose on me. A special thanks to all Portuguese rheumatologists who have always supported me. Pedro, my companion, our last years have had a lot in common and it was always of great help to discuss plans and share uncertainties with you. The PhD journey, full of surprises and challenges, is reaching its end for both of us: what comes next? Prof. Pereira da Silva, to you I owe the path I have finally chosen. My destiny became Maastricht instead of London, and I cannot emphasize too much how thankful I am for that. Prof. Jaime Branco, Lena, Dr. João, Dr. Maria José, Fernando, Patrícia, Filipa, Ana, Nélia, and many others, thank you for always keeping me close to the Portuguese rheumatology and for keeping me actively involved in research projects, for your continuous support and your friendship. No matter where I am, Portuguese rheumatology will always be in my heart. Throughout these years I have had the privilege to work in research projects with many people from different places in the world and experience one of the biggest rewards from research: being stimulated and learning from others. Dear Prof. Dr. M. Dougados, dear Maxime, and dear Prof. Dr. F. van den Bosch, dear Filip, I have enjoyed very much working together with you; thank you very much for your wise comments and suggestions to all my papers. Prof. R. Buchbinder, dear Rachelle, it is impressive what I have learnt from you, and especially how much fun we had all the time! I am sure I will end up visiting you in Australia. Dr. K. Michaud, dear Kaleb, thanks for helping me survive with Stata and not having to surrender to SPSS, for the many things I have learnt with you and for always finding time to discuss ideas. Prof. Dr. J. Smolen, dear Josef, thank you for giving me the opportunity to participate in initiatives of treatment recommendations, rewarding challenges with an impressive work rhythm.

190

Prof. Dr. T. Kvien, dear Tore, Prof. Dr. T. Uhlig, dear Till, thank you both for the fruitful collaborations, which will hopefully continue. Dear Louise Falzon, for your precious help with search strategies in all systematic literature reviews, thank you very much. Prof. Dr. P. Royston, dear Patrick, after literally stalking you in London, we managed to produce some nice reference intervals and percentile curves for spinal mobility. Thank you for your willingness to help and share knowledge, despite the busy agenda and the fact that I was a complete stranger demanding your time. Dear Francisca Sivera, what an adventure our project in training in rheumatology has become! Best learning point: nothing is really impossible! And (one of) the best part(s) is certainly doing it with you and having so much fun together. Working in EMEUNET should not be forgotten and especially several friends with whom it has been a real pleasure to work: Laure Gossec, Daniel Aletaha, Maya Buch, Anna Moltó, Peter Mandl, Caroline Ospelt, Christian Beyer, Helga Radner and many others: It is impressive to look back and see how EMEUNET started and what it has become in such a short time! My dearest Victoria Navarro, always making jokes and making me laugh. Leiden, Maastricht, Spain/Portugal, and so much more in common, could only result in awesome moments together and a very good friendship. Years in Maastricht were very rich in meeting good friends for life. Maria, Gonzalo, Milla, Isabel, Laura, Paulo, Inês, Hugo, all my dear ex-pats, we shared so many good moments, passed through similar difficulties and challenges while being abroad. You have left and spread throughout the world, but the most important is that memories and true friendships persist. Féline and Lennart, my ‘Dutch family’, first in Maastricht, now in Leiden together, thank you for all you have taught me in Dutch, for being so patient (and not giving up on speaking in Dutch with me from the beginning) and for your precious support. To the friends at home who have kept me close despite the distance, I deeply appreciate the e-mails, messages, calls, visits, more than you will ever know. Rita, Rute, André, Ricardo, Eduardo, Maria, Ana, Kika, Matilde, Filipa, António, Tomás, Nuno, Susana, Manel, Chico, Catarina and many others: it is impossible to describe the feeling of seeing friendships unchanged despite the fact that I am already abroad for more than 5 years. I hope it continues the same for many more, independently of where we all are.

191

My dear paranymfs, Rosaline and Raquel, your continuous support is priceless, your friendship invaluable. I am very lucky I can have such special friends. These years, and especially wrapping up the thesis, would have been much harder without you. Last, but not least, my dearest family. My parents, brothers, uncles and aunts (grand-aunt included) give me a fantastic support. The way in which you believe in my career gives me strength to continue and work harder, makes going abroad much easier than what I thought. Thank you for stimulating me to do this! It is a pity that grandmother NazarĂŠ cannot witness this, but I am sure she would be as enthusiastic. Nuno and Paulo, my dear brothers, always ready to listen to my dilemmas and to give good advice. Dear Mom and Dad, I could never imagine I could be abroad but still feel so close home and receive so much daily help from you. You make my difficult decisions easy by unconditionally supporting me and being amazingly available always for everything. Impossible even to name all you do for me. Mom, you even inserted data from patients into a dataset, several days in a row! I will never forget it. A deep thank you. For all I have written, and much more that has been left unsaid about all the special people around me (unfortunately it has been impossible to name everyone), and how they have contributed to make the most of these last years, I feel privileged and lucky! Muito obrigada, thank you very much, hartelijk bedankt!

192

193