BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 confers preferential susceptibility to peripheral spondyloarthritis while HLA-B27 preferentially relates to axial spondyloarthritis

For peer review only Journal: BMJ Open

Manuscript ID: bmjopen-2015-009092

Article Type: Research

Date Submitted by the Author: 16-Jun-2015

Complete List of Authors: LONDONO, JOHN; Universidad de La Sabana, Rheumatology; Universidad de Navarra, Department of Internal Medicine Santos, Ana; Universidad de La Sabana, Immunology Peña, Paola; Hospital Militar Central, Rheumatology Calvo, Enrique; Universidad Nacional de Colombia, Radiology Espinosa, Luis; Louisiana State University, LSU Health Sciences Center Reveille, John; University of Texas Health Science Center, USA, Department of Internal Medicine, Division of Rheumatology and Clinical Immunogenetics Vargas-Alarcon, Gilberto; Inst Nacl Cardiol Ignacio, Jaramillo, Calos; Universidad de Los Andes, Department of Biological Sciences Valle-Oñate, Rafael; Hospital Militar Central, Rheumatology

Avila, Mabel; Hospital Militar Central, Dirección de Educación e http://bmjopen.bmj.com/ Investigacion Romero, Consuelo; Hospital Militar Central, Rheumatology Medina, Juan; Universidad de Navarra, Internal Medicine

Primary Subject Rheumatology Heading:

Secondary Subject Heading: Immunology (including allergy)

spondyloarthritis, axial spondyloarthritis, peripheral spondyloarthrtis, HLA- on September 29, 2021 by guest. Protected copyright. Keywords: B27, HLA-B15

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 1 1 2 3 4 5 6 7 8 9 HLA-B15 confers preferential susceptibility to peripheral 10 11 12 spondyloarthritis while HLA-B27 preferentially relates to axial 13 14 15 spondyloarthritisFor peer review only 16 17 18 19 John Londono1, Ana Maria Santos1, Paola Peña1, Enrique Calvo2, Luis R. Espinoza3, 20 21 John D. Reveille4, Gilberto Vargas-Alarcon5, Carlos A. Jaramillo6, Rafael Valle-Oñate7, 22 23 Mabel Avila1,7, Consuelo Romero7 and Juan F. Medina8 24 25 26 27 28 1Department of Rheumatology, University of La Sabana, Chia, Cundinamarca, Colombia, 2Department of 29 Radiology, National University of Colombia, Bogota, Colombia, 3LSU Health Sciences Center, Louisiana 30 4 31 State University, New Orleans, USA, Division of Rheumatology and Clinical Immunogenetics, 32 Department of Internal Medicine, University of Texas Health Science Center, Houston, USA, 5 33 Interventional Genetic Study Group in Cardiovascular Diseases, Department of Molecular Biology, 6

34 National Institute of Cardiology Ignacio Chávez, Mexico DF, Department of Biological Sciences, http://bmjopen.bmj.com/ 35 University of Los Andes, Bogota, Colombia, 7Rheumatology and Immunology Unit, Department of Internal 36 Medicine, Hospital Militar Central, Bogota, Colombia, 8Department of Internal Medicine, School of 37 Medicine and Center for Applied Medical Research (CIMA), University of Navarra and Ciberehd, 38 Pamplona, Spain. 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 Correspondence to: John Londono, MD, Department of Rheumatology, University of La Sabana and 45 Hospital Militar Central, Campus del Puente del Comun, Km. 7, Autopista Norte de Bogotá. CP 250001, 46 Chía, Cundinamarca, Colombia. Tel. (571)8613503; FAX: (571)8615618. Email: 47 48 [email protected] 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 2 1 2 3 Abstract 4 5 Objective. HLAB27 and B15 are associated with spondyloarthritis (SpA). Recent ASAS criteria 6 7 emphasize a distinction between SpA with axial and peripheral pattern. We analyzed whether HLAA, B 8 and DRB1 alleles could associate with these patterns. 9 10 Methods. We studied 100 healthy individuals and 178 patients with SpA according to ESSG criteria. 11 12 Patients were then classified according to ASAS criteria, the axial pattern (axSpA) being defined by 13 ascertained sacroiliitis and the peripheral pattern (pSpA) by enthesitis and/or arthritis in extremities. A 14 15 mixed ax/p patternFor was also considered.peer review only 16 17 Results. Only HLAB27 and B15 alleles were associated with SpA. ASAS criteria for axSpA were met in 18 152 patients (12 with pure axSpA and 140 with a mixed ax/p pattern). When the ASAS peripheral criteria 19 20 were applied, 161 patients met these criteria (13 with pure pSpA and 148 with a mixed ax/p pattern). 21 HLAB27 was found in 83% of patients with axSpA and 43% of ax/pSpA patients according to axASAS. 22 23 HLAB27 occurred in 7% controls but not in any patient with pure pSpA. HLAB15 was encountered in 24 31% of patients with pure pSpA and 20% of ax/pSpA patients according to pASAS criteria. Moreover, 2 25 26 healthy controls, but none of our patients with pure axSpA were positive for HLAB15. 27 28 Conclusion. Our data suggest that the presence of HLAB15 favors the development of pure/mixed 29 peripheral rather than pure axSpA, while HLAB27 promotes a pure/mixed axial disease and exclude a 30 31 peripheral pattern. HLAB15 should be considered in addition to HLAB27 when diagnosing SpA patients 32 according to ASAS criteria. 33

34 http://bmjopen.bmj.com/ 35 36 37 Key words: spondyloarthritis; axial spondyloarthritis; peripheral spondyloarthrtis, HLAB27; HLAB15. 38 39 Strengths and limitations of this study 40 41 • The HLAB15 has been associated with SpA and oligoarthritis in different populations. This study 42 on September 29, 2021 by guest. Protected copyright. 43 clearly establishes this association, through a detailed analysis of the clinical and radiological aspects 44 in a group of patients not AngloSaxon origin. 45 46 • In this study, conducted on a group of Latin American patients with SpA, the presence of HLAB15 47 and HLAB27 clearly influence the pattern of presentation of the disease. The HLAB15 favors a 48 preferably peripheral pattern while the HLAB27 promotes a pure/mixed axial disease and exclude a 49 50 pure peripheral pattern. 51 • HLAB15 should be considered in addition to HLAB27 when diagnosing SpA Colombian patients 52 53 according to ASAS criteria. 54 55 • This study use the ASAS criteria in an alternating manner to discriminate the association of the 56 disease with the HLAB27 and HLAB15. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 3 1 2 3 • The results of this study need to be verified in a research involving a larger number of patients with 4 5 SpA. 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 4 1 2 3 Introduction 4 5 Spondyloarthritis (SpA) comprises a group of inflammatory chronic diseases that share similar clinical 6 7 presentations, radiological findings, HLAB27 association, and positive family history.[1] According to the 8 9 European Spondyloarthropathy Study Group (ESSG) criteria,[2] SpA has been traditionally classified as 10 11 follows: ankylosing spondylitis (AS), reactive arthritis (ReA), undifferentiated spondyloarthritis (uSpA), 12 13 psoriatic arthritis (PsA), and arthritis associated with inflammatory bowel disease (IBD).[3] Its prevalence 14 15 varies from 0.1 toFor 1.4% of thepeer general population review depending on only the geographical region studied.[4] 16 17 Predominantly, SpA begins in individuals younger than 45 years of age and clinically compromises the 18 19 axial spine, peripheral joints, tendons and ligament insertions (entheses), having also extra 20 21 musculoskeletal manifestations in the skin, mucosa, gut and the eyes.[5] 22 23 HLAB27 is the biggest risk factor for the development of SpA, and while it does have a central role 24 25 in its etiopathogenesis, the association is not the same across all ethnic groups.[6] It has been estimated 26 27 that this allele is responsible for up to 28% of the etiology of SpA.[7] But there are other factors that 28 29 impact on the onset of the disease, such as environmental factors, and other genetic factors, including 30 31 those within the Major Histocompatibility Complex (MHC) Class I − HLAB60/B40,[8] B39,[9] B14,[10] 32 33 and B15[11] − and Class II − DRB1[12] − alleles.

34 http://bmjopen.bmj.com/ 35 Current new technologies are allowing more specific examination of disease susceptibility, such as 36 37 those that look at Single Nucleotide Polymorphisms (SNPs), GenomeWide Association Studies (GWAS) 38 39 and genomic resequencing which underscore the polygenic nature of SpA.[13 14] There are more than 40 41 40 genes implicated in the pathogenesis of SpA which may also affect the clinical manifestations, 42 on September 29, 2021 by guest. Protected copyright. 43 including the age of onset and the duration and severity of the disease.[15] 44 45 The diagnosis of SpA is made principally based on clinical, radiologic, and laboratory findings. The 46 47 classification currently proposed by the European League Against Rheumatism (EULAR) and the 48 49 Assessment of SpondyloArthritis international Society (ASAS) shows a high sensitivity and specificity 50 51 (79.5% and 83.3%, respectively).[16 17] According to these entities SpA can have two general clinical 52 53 patterns: axial and peripheral. With these two clinical patterns, there may be a genetic variability that 54 55 accounts for each physical distribution. In fact, data emerging from immunopathological studies and from 56 57 clinical trials suggest that the pathogenesis of axial and peripheral disease may be caused by different 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 5 1 2 3 mechanisms influenced by genetic differences, a situation that underscores the importance of the new 4 5 clinical classification criteria from ASAS for peripheral and axial SpA.[18] The axial criteria rely on two 6 7 major components, imaging and the presence of HLAB27.[19] The latter was adopted because of its high 8 9 sensitivity and specificity and its extensive validation for axSpA. However, the validity of the HLAB27 10 11 component is less defined for pSpA and uSpA. In uSpA with peripheral involvement, for instance, only 12 13 about 40% of patients have HLAB27.[20] On the other hand, previous studies in patients with SpA have 14 15 reported an associationFor with peer HLAB15 in Europe review (Belgium),[21] Northonly Africa (Tunisia),[22] and Latin 16 17 America (Mexico),[23] where the most common presentations of SpA were related to IBD, ReA, and 18 19 uSpA. In particular, peripheral involvement was nearly ubiquitous in these cohorts. This would suggest 20 21 that HLAB15 allele may play an important role in the susceptibility to peripheral SpA.[24] 22 23 The objective of this study was to evaluate the impact of HLAB15 and HLAB27 alleles on the 24 25 susceptibility and the clinical characteristics of the disease (considered according to either ESSG or 26 27 ASAS criteria) in a Colombian population with SpA. 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 6 1 2 3 Patients and methods 4 5 6 Participants and clinical assessment of patients 7 8 9 A group of 178 patients with a diagnosis of SpA according to the ESSG criteria were studied. These 10 11 patients were seen in the Central Military Hospital in Bogota, Colombia, between January 2005 and 12 13 December 2011. All patients were evaluated by a Rheumatologist in the outpatient care department with 14 15 subsequent HLA For genotyping. peerDemographic andreview clinical data, including only disease activity [graded through 16 17 the Bath Ankylosing Spondylitis Disease Activity Index, BASDAI[25] as well as the global disease activity 18 19 assessed by Physician, i.e. the Visual Analog Scale of Physician’s global assessment of disease activity, 20 21 abbreviated as VAS Physician,[26] the VAS global disease activity assessed by patients, abbreviated as 22 23 VAS Patient,[26] and the Visual Analog Scale of Inflammatory Back Pain, abbreviated as VAS IBP,[26] functional 24 25 capacity (graded through the Bath Ankylosing Spondylitis Functional Index, BASFI),[27] inflammatory 26 27 back pain, enthesitis and joints affected were recorded. Recommendations for followup were according 28 29 to the previously validated ASAS guidelines.[26] Each enthesis was evaluated using the Mander’s 30 31 enthesis index.[28] For each patient, laboratory exams that included the erythrocyte sedimentation rate 32 33 were taken.

34 http://bmjopen.bmj.com/ 35 The control group consisted of 100 healthy subjects on the list of transplant donors at the Central 36 37 Military Hospital in Bogota, Colombia. These subjects had the same sociodemographic characteristics as 38 our SpA population. Cases and controls were compared for the HardyWeinberg equilibrium as tested 39 40 with the Arlequin® statistical software version 3.5.[29] Allele frequency (AF) among the patient population 41 42 was calculated using Genepop Program.[30] on September 29, 2021 by guest. Protected copyright. 43 44 45 Ethics approval 46 47 48 The study followed the norms established by the Helsinki Declaration, The Guidelines for Good Clinical 49 50 Practice, and the Resolution 8430 (1993) of the Colombian Ministry for Social Protection. Moreover, was 51 52 reviewed and approved by the Ethics Committees of the University of La Sabana, Los Andes University, 53 54 and the Central Military Hospital. Each patient signed an informed consent form, and confidentiality was 55 56 strictly maintained. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 7 1 2 3 Imaging 4 5 6 A plain anterior pelvic xray was taken and Magnetic Resonance Imaging (MRI) of the sacroiliac joints 7 8 with T1 sequence and, fat suppression technique was taken on each patient. This was then read by a 9 10 radiologist with musculoskeletal expertise who was blinded to the clinical picture. An interpretation of the 11 12 xrays was made following the recommendations for population studies in the Atlas of Standard 13 14 Radiographs[31] and the MRI was evaluated looking for acute and chronic changes in the joints.[32] 15 For peer review only 16 17 HLA genotyping 18 19 Genomic DNA was extracted from peripheral blood leukocytes with a Wizard genomic DNA purification 20 21 Kit (Promega, Cat. # A1120). After extraction, samples were adjusted to a DNA concentration of 80 ± 20 22 23 g/L and stored at –80 °C until processing of HLA. Typing was carried out by PCR on the genomic DNA 24 25 using the HLA Sequence Specific Primer HLAABDRSSPtray Kit (BioRad, Cat. # 826 230), which is 26 27 focused on determining HLAA, B, and the DRB1 alleles. PCR products were verified by electrophoresis 28 29 in a 2% agarose gel, and bands were visualized by ultraviolet light photo documentation and further 30 31 analyzed by using the HLASSP Typing V 1.2.0.0. Software (BioRad). 32 33

34 Statistical analyses http://bmjopen.bmj.com/ 35 36 Data are expressed as mean ± SD (except when otherwise indicated). For statistical analysis we used 37 38 SPSS 19.0 (SPSS Inc, USA). For the presentation and analysis of categorical variables, frequency 39 40 distribution, percentages and contingency tables were used. Comparisons were made using the Chi 41 42 square test and Fisher Exact test. For numerical values, central tendency and dispersion measurements on September 29, 2021 by guest. Protected copyright. 43 44 were evaluated. Results were analyzed using Student’s t and Wilcoxon tests. Twotailed P values < 0.05 45 46 were considered statistically significant. The association between disease and HLA alleles was 47 48 determined by Odds Ratio and confidence intervals of 95% (OR: 95% CI). 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 8 1 2 3 Results 4 5 6 Genotype and demographic data 7 8 9 Only HLAB27 and B15 out of all the analyzed HLAA, B and DRB1 alleles were found to be significantly 10 11 associated with SpA. As shown in Table 1, our 178 SpA patients (included in the study because they met 12 13 the ESSG criteria) could therefore be divided into three groups according to their HLAB genotype: 34 14 15 patients (19.1%) For had HLAB15, peer 70 (39.3%) showedreview HLAB27 and 74only (41.6%) possessed other HLAB 16 17 alleles. Patients with HLAB15 and HLAB27 had an earlier age of onset of the disease than patients with 18 19 other HLAB alleles (though differences reached a statistical significance only in the case of patients 20 21 positive for HLAB27; P = 0.008). A male predominance (with a male/female ratio of 3.7/1) was 22 23 documented in the group of 70 HLAB27 positive patients, whereas females were slightly predominating 24 25 (male/female ratio: 0.8/1) in the group of 34 HLAB15 positive patients (Table 1). 26 27 When ASAS classification criteria [16 17] were applied to classify our patients 178 previously 28 29 diagnosed according to the ESSG criteria, 152 patients fulfilled the ASAS criteria for pSpA while 26 did 30 31 not, and 161 patients met the ASAS criteria for pSpA while 17 patients did not. Most of the patients 32 33 meeting the ASAS criteria for axSpA and/or pSpA, manifested a mixed ax/p pattern, though the total

34 http://bmjopen.bmj.com/ 35 number of these patients slightly fluctuated. This fluctuation was dependent on the application to our 36 37 patients of either the axial or peripheral criteria,(Table 1) these two sets of ASA criteria being 38 simultaneously met in 137 patients. Moreover, 12 patients had pure axSpA and 13 showed pure 39 40 pSpA.(Table 1) Noticeably, 100% of the 70 HLAB27 positive patients included in the study exhibited axial 41 42 manifestations of the disease and 100% of the 34 patients positive for HLAB15 had peripheral on September 29, 2021 by guest. Protected copyright. 43 44 manifestations of SpA (both according to ASAS criteria). Among the 12 patients with pure axSpA, 10 of 45 46 them (83%) were positive for HLAB27, and among the 13 patients with pure pSpA, 4 of them (31%) were 47 48 positive for HLAB15. Moreover, HLAB27 was absent in all 13 pSpA patients,(Table 1) though this allele 49 50 did occur in 7% of the 100 healthy control subjects.(not shown) On the other hand, two (2%) healthy 51 52 controls but none of the 12 axSpA patients were positive for HLAB15. Further demographic data and 53 54 clinical presentation related to either ESSG or ASAS criteria and to HLAB27, HLAB15, and other B 55 56 alleles are also shown in Table 1. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 9 1 2 3 Association of HLAB15 and HLAB27 alleles to clinical SpA subsets 4 5 6 When SpA patients were analyzed as a whole we found a statistical association of the disease with HLA 7 8 B15.(Table 2) But when the clinical subsets were considered, HLAB15 was mainly associated with uSpA 9 10 and pSpA forms, whereas HLAB27 was mainly associated with AS (Table 2). The relationships of HLA 11 12 B27 and HLAB15 with clinical forms of presentation according to either the ESSG criteria or the ASAS 13 14 criteria are further summarized in Figure 1. In this figure, the findings that none of the patients with pure 15 For peer review only 16 axial SpA are HLAB15 positive and that HLAB27 does not occurs in patients with pure peripheral SpA 17 18 are graphically represented. This findings appear to strengthen the notion that ASAS criteria to classify 19 20 patients might allowed a better correlation between the clinical presentation of SpA and the HLAB 21 22 genotype than the ESSG criteria. 23 24 25 HLAB15 and HLAB27 alleles and disease characteristics 26 27 The most relevant clinical features and radiological findings in relation to HLA alleles are shown in Figure 28 29 2. Remarkably, patients with HLAB15 had more swollen joints (5.3±5.1 vs 2.3±1.8, P = 0.004) and total 30 31 count of painful enthesitis,(6.0±4.5 vs 4.3±3.3, P = 0.014) and peripheral painful enthesitis,(3.4±2.9 vs 32 33 2.3±2.2, P = 0.050) while positive patients for HLAB27 had greater sacroiliac joint compromise (74.7% vs

34 http://bmjopen.bmj.com/ 35 56.3%, P = 0.043) and painful enthesitis in sacroiliac region.(64.4% vs 36%, P = 0.001) 36 37 38 HLAB15 and HLAB27 alleles and disease activity 39 40 When the disease activity was taken into account, we found that HLAB15 positive patients had a higher 41 42 rate in the BASDAI index, BASFI index and global disease activity assessed by Physician (VAS on September 29, 2021 by guest. Protected copyright. 43 44 Physician) compared to HLAB27 positive patients.(Figure 3) On the other hand, HLAB27 positive 45 46 patients had a higher rate in inflammatory back pain and global disease activity assessed by patient (VAS 47 48 Patient) than HLAB15 positive patients.(Figure 3) Finally, it is noteworthy to mention that HLAB27 49 50 positive patients exhibited a higher erythrocyte sedimentation rate than HLAB15 positive patients (25.7 ± 51 52 24.4 mm/hour versus 16.5 ± 9.8, respectively). 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 10 1 2 3 Discussion 4 5 6 In this study we found that in addition to the strong association of SpA with HLAB27, SpA can be 7 8 associated with another HLAB allele, i.e. with HLAB15. Moreover, the demonstrated increase in the 9 10 prevalence of HLAB15 in SpA patients with peripheral involvement is especially noteworthy. While 11 12 patients with HLAB27 were more likely to be male and to show axial compromise associated with AS, 13 14 HLAB15 positive patients had a lower proportion of men and a more frequent uSpA clinical 15 For peer review only 16 manifestation. 17 18 Patients with either HLAB15 or HLAB27 had similar ages of onset of the disease and duration of 19 20 the disease. From a clinical point of view, HLAB15 patients had an increased number of peripheral joints 21 22 affected than patients positive for HLAB27, while no differences in spinal mobility were observed 23 24 between these two groups of patients. Radiographically, HLAB27 positive patients showed more severe 25 26 sacroiliitis in pelvic xrays and greater acute damage in MRI than patients positive for HLAB15. 27 28 Male gender, axial involvement, early age of onset (<20 yrs), and the presence of HLAB27 are all 29 30 known to be associated with a bad prognosis.[33] In our study, patients with HLAB15 (consisting of 31 32 slightly fewer males than females) exhibited a less severe disease phenotype and more peripheral 33 symptoms as compared to patients positive for HLAB27. But concerning the disease activity, both

34 http://bmjopen.bmj.com/ 35 36 BASDAI and BASFI scores were higher (worse) in HLAB15 positive patients than in those with HLAB27. 37 This may be due to greater functional impairment and higher inflammatory disease activity in a greater 38 39 number of lower extremity joints and enthesitis.[34 35] On the other hand, the acute phase reactants were 40 41 particularly increased in the HLAB27 positive patients. 42 on September 29, 2021 by guest. Protected copyright. 43 HLAB27 has been known for over 40 years as a predisposing factor for SpA, specifically for 44 45 AS.[36] In Caucasians with SpA, 8085% of patients are HLAB27 positive. Overall, however, less than 46 47 5% of the Caucasian patients with this allele go on to develop the disease.[37] This suggests that even 48 49 though HLAB27 appears to play a major part in the disease, it is not enough to completely explain the 50 51 disease susceptibility. In populations other than Caucasians, HLAB27 is less prevalent, occurring in 52 53 between 4060% of patients with SpA.[38] This is seen universally, and it is confirmed in our study as 54 55 well. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 11 1 2 3 In people of AngloSaxon descent, AS is the most common subtype of SpA, while uSpA is the 4 5 predominant subtype in the Latin American mestizo population.[39] This variation could be due to 6 7 associations of the disease with different HLA genes in addition to HLAB27.[40] The new criteria recently 8 9 proposed by ASAS to classify SpA allow for grouping the patients according to severity, onset, and axial 10 11 or peripheral pattern. It is remarkable that in our study, 100% of the HLAB15 positive patients met with 12 13 the ASAS peripheral criteria and 100% of the HLAB27 positives fulfilled the ASAS axial criteria. 14 15 There had beenFor previous peer reports partially review considering the HLAB15 only allele in relation to SpA. Thus, a 16 17 study in a Mexican population, in addition to showing a strong association of HLAB27 with SpA, also 18 19 suggested that prevalence of HLAB15 was increased in the whole SpA cohort.[23] In fact, HLAB15 was 20 21 modestly increased in uSpA but not in AS patients, whereas HLAB27 showed a greater association with 22 23 AS than with uSpA.[23] Likewise, a prospective study in Belgium examined patients with different 24 25 subtypes of SpA, and found that HLAB15 was more frequent in those with uSpA (16/51, 32%) as well as 26 27 in those with a clinical remission of symptoms.[41] HLAB27, however, was more prevalent in patients 28 29 with AS and in those with sustained disease activity. Also in Tunisia HLAB15 was associated with uSpA 30 31 as well as with ReA. In fact, this allele was found in up to 50% of patients with uSpA. As this observation 32 33 is very similar to previous studies done in Latin America, Europe, and Africa, it was suggested that it

34 http://bmjopen.bmj.com/ 35 could be the result of population migrations.[42] Altogether, the aforementioned data indicate that HLA 36 37 B15 is found in various different ethnic groups, crossing geographical barriers and further validate its role 38 39 in SpA pathogenesis.[43] 40 41 Inflammatory diseases other than SpA have also been associated with HLAB15. A metaanalysis 42 on September 29, 2021 by guest. Protected copyright. 43 in a Caucasian population with acute periodontitis showed a strong association with this allele.[44] 44 45 Moreover, there seems to be a relationship between periodontal disease and the onset of inflammatory 46 47 diseases such as rheumatoid arthritis, which suggests that HLAB15 may be related to acute and chronic 48 49 diseases with local bone involvement.[44] Another notable correlation for HLAB15 is with Behçet’s 50 51 disease in a Moroccan population, where females were the dominant affected population.[45] 52 53 HLAB15 is an HMC Class I molecule encoded by multiple polymorphic alleles leading to at least 7 54 55 serologic subtypes.[43] There had been described variants – with either tyrosine or serine at position 116 56 – which greatly diverge in their interaction (either strong or weak, respectively) with other proteins in the 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 12 1 2 3 endoplasmic reticulum for antigen presentation at the cell surface.[46] Accordingly, it might be expected 4 5 that genetic diversity can contribute to increased or decreased predisposition to autoimmune diseases. In 6 7 the case of HLAB15, recent studies have demonstrated that this molecule as well as HLAB14, B38, 8 9 B39, and B73 all have an unpaired free cysteine (at position 67) in the B pocket of the alpha heavy 10 11 chain, which is very similar to HLAB27. This may allow for the formation of homodimers that could 12 13 activate the innate immune response to produce cytokines implicated in chronic inflammation in the 14 15 pathogenesis of SpA.[47]For peer review only 16 17 Recently, a highdensity genotyping study was carried out to identify multiple risk variants for 18 19 AS.[14] In addition to confirming the previously reported strong association of the disease with SNPs at 20 21 the gene for the endoplasmic reticulum aminopeptidase 1 (ERAP1) that was restricted to HLAB27– 22 23 positive AS patients, this study found a significant association of AS with another member of the 24 25 endoplasmic reticulum aminopeptidase family, ERAP2, which prevailed in AS patients negative for the 26 27 HLAB27 allele.[14] Future highdensity genotyping studies might uncover whether some (or most) of 28 29 these HLAB27–negative patients are positive for the HLAB15 allele. While ERAP1 seems to play a 30 31 critical role in determining the length and sequence of the peptides bound and presented by HLAB27 32 33 associated with AS,[48] it might be attractive to hypothesize that ERAP2 may play a parallel role in the

34 http://bmjopen.bmj.com/ 35 context of HLAB15. 36 37 In conclusion, our study with a Mestizo cohort indicates that HLAB15 is associated with the 38 39 peripheral manifestation of SpA independently of sex or age, with a notable involvement of joints and 40 41 enthesitis and with minor axial compromise. HLAB15 may therefore become an important tool and 42 on September 29, 2021 by guest. Protected copyright. 43 marker for the disease in our population, particularly in those patients with peripheral SpA without HLA 44 45 B27. 46 47 48 49 Acknowledgements 50 51 We are grateful to Maria del Pilar Delgado, MCs, Department of Biological Sciences, University of 52 53 Los Andes, for critical review of the manuscript. 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 13 1 2 3 Funding: This work was supported by the University of La Sabana and the University of Los Andes. 4 5 6 7 8 Author contributions: All authors were involved in drafting the article or revising it critically for important 9 10 intellectual content, and all authors approved the final version to be published. Dr. Londono had full 11 12 access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy 13 14 of the data analysis. 15 For peer review only 16 Study conception and design. Londono, Santos, Medina, VargasAlarcon, Espinoza, Reveille, Jaramillo, 17 18 Avila, Romero, ValleOñate 19 20 Acquisition of data. Londono, Santos, Pena. 21 22 Analysis and interpretation of data. Londono, Santos, Pena, Medina. 23 24

25 26 Disclosure statement: The authors declare no conflicts of interest. No financial support or benefits from 27 28 29 commercial sources, or any additional conflict of interest exist with regard to this work. 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 14 1 2 3 References 4 5 6 1. Khan MA. Update on spondyloarthropathies. Ann Intern Med 2002;136:896-907. 7 2. Dougados M, van der Linden S, Juhlin R, et al. The European Spondylarthropathy Study Group 8 preliminary criteria for the classification of spondylarthropathy. Arthritis and rheumatism 1991;34:1218- 9 27. 10 11 3. Dougados M, van der Heijde D. Ankylosing spondylitis: how should the disease be assessed? Best 12 practice & research. Clinical rheumatology 2002;16:605-18. 13 4. Ehrenfeld M. Geoepidemiology: the environment and spondyloarthropathies. Autoimmunity reviews 14 2010;9:325-9. 15 5. Dougados M, HochbergFor MC. peer Why is the concept review of spondyloarthropathies only important? Best practice & 16 research. Clinical rheumatology 2002;16:495-505. 17 18 6. Gonzalez-Roces S, Alvarez MV, Gonzalez S, et al. HLA-B27 polymorphism and worldwide susceptibility 19 to ankylosing spondylitis. Tissue Antigens 1997;49:116-23. 20 7. Reveille JD. Genetics of spondyloarthritis--beyond the MHC. Nature reviews. Rheumatology 21 2012;8:296-304. 22 8. Robinson WP, van der Linden SM, Khan MA, et al. HLA-Bw60 increases susceptibility to ankylosing 23 spondylitis in HLA-B27+ patients. Arthritis and rheumatism 1989;32:1135-41. 24 9. Yamaguchi A, Tsuchiya N, Mitsui H, et al. Association of HLA-B39 with HLA-B27-negative ankylosing 25 26 spondylitis and pauciarticular juvenile rheumatoid arthritis in Japanese patients. Evidence for a role of 27 the peptide-anchoring B pocket. Arthritis and rheumatism 1995;38:1672-7. 28 10. Diaz-Pena R, Blanco-Gelaz MA, Njobvu P, et al. Influence of HLA-B*5703 and HLA-B*1403 on 29 susceptibility to spondyloarthropathies in the Zambian population. J Rheumatol 2008;35:2236-40. 30 11. Wagener P, Zeidler H, Eckert G, et al. Increased frequency of HLA-Bw62 and Bw35 CREG antigens in 31 HLA-B27 negative ankylosing spondylitis. Z Rheumatol 1984;43:253-7. 32 12. Brown MA, Kennedy LG, Darke C, et al. The effect of HLA-DR genes on susceptibility to and severity 33 of ankylosing spondylitis. Arthritis and rheumatism 1998;41:460-5. 34 http://bmjopen.bmj.com/ 35 13. van Gaalen FA, Verduijn W, Roelen DL, et al. Epistasis between two HLA antigens defines a subset of 36 individuals at a very high risk for ankylosing spondylitis. Ann Rheum Dis 2013;72:974-8. 37 14. International Genetics of Ankylosing Spondylitis Consortium (IGAS), Cortes A, Hadler J, et al. 38 Identification of multiple risk variants for ankylosing spondylitis through high-density genotyping of 39 immune-related loci. Nat Genet 2013;45:730-38. 40 41 15. Diaz-Pena R, Lopez-Vazquez A, Lopez-Larrea C. Old and new HLA associations with ankylosing 42 spondylitis. Tissue Antigens 2012;80:205-13. on September 29, 2021 by guest. Protected copyright. 43 16. Rudwaleit M, van der Heijde D, Landewe R, et al. The Assessment of SpondyloArthritis International 44 Society classification criteria for peripheral spondyloarthritis and for spondyloarthritis in general. Ann 45 Rheum Dis 2011;70:25-31. 46 17. Rudwaleit M, Landewe R, van der Heijde D, et al. The development of Assessment of 47 SpondyloArthritis international Society classification criteria for axial spondyloarthritis (part I): 48 49 classification of paper patients by expert opinion including uncertainty appraisal. Ann Rheum Dis 50 2009;68:770-6. 51 18. Baeten D, Breban M, Lories R, et al. Are spondylarthritides related but distinct conditions or a single 52 disease with a heterogeneous phenotype? Arthritis and rheumatism 2013;65:12-20. 53 19. Rudwaleit M, van der Heijde, D. Landewe, R., Listing J, Akkoc N, et al. The development of 54 Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis 55 (part II): validation and final selection. Ann Rheum Dis 2009;68:777-83. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 15 1 2 3 20. Valle-Onate R, Candia L, Romero-Sanchez C, et al. Epidemiology of spondyloarthritis in Colombia. The 4 American journal of the medical sciences 2011;341:293-4. 5 6 21. Mielants H, Veys EM, Joos R, et al. HLA antigens in seronegative spondylarthropathies. Reactive 7 arthritis and arthritis in ankylosing spondylitis: relation to gut inflammation. J Rheumatol 1987;14:466- 8 71. 9 22. Siala M, Mahfoudh N, Fourati H, et al. MHC class I and class II genes in Tunisian patients with 10 reactive and undifferentiated arthritis. Clin Exp Rheumatol 2009;27:208-13. 11 23. Vargas-Alarcon G, Londono JD, Hernandez-Pacheco G, et al. Effect of HLA-B and HLA-DR genes on 12 13 susceptibility to and severity of spondyloarthropathies in Mexican patients. Ann Rheum Dis 14 2002;61:714-7. 15 24. Reveille JD. AnFor update onpeer the contribution review of the MHC to as susceptibility.only Clinical rheumatology 16 2014;33:749-57. 17 25. Garrett S, Jenkinson T, Kennedy LG, et al. A new approach to defining disease status in ankylosing 18 spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994;21:2286-91. 19 26. Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society 20 21 (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68:ii1-44. 22 27. Calin A, Garrett S, Whitelock H, et al. A new approach to defining functional ability in ankylosing 23 spondylitis: the development of the Bath Ankylosing Spondylitis Functional Index. J Rheumatol 24 1994;21:2281-5. 25 28. Mander M, Simpson JM, McLellan A, et al. Studies with an enthesis index as a method of clinical 26 assessment in ankylosing spondylitis. Ann Rheum Dis 1987;46:197-202. 27 28 29. Excoffier L, Lischer HE. Arlequin suite ver 3.5: a new series of programs to perform population 29 genetics analyses under Linux and Windows. Mol Ecol Resour 2010;10:564-7. 30 30. Rousset F. GENEPOP'007: a complete re-implementation of the genepop software for Windows and 31 Linux. Mol Ecol Resour 2008;8:103-6. 32 31. Bennett PH, Burch TA. The epidemiological diagnosis of ankylosing spondylitis. In: Bennett PH, Wood 33 PHN, eds. Population Studies of the Rheumatic Diseases. Amsterdam: Excerpta Medica, 1968:453-58. 34 32. Rudwaleit M, Jurik AG, Hermann KG, et al. Defining active sacroiliitis on magnetic resonance imaging http://bmjopen.bmj.com/ 35 36 (MRI) for classification of axial spondyloarthritis: a consensual approach by the ASAS/OMERACT MRI 37 group. Ann Rheum Dis 2009;68:1520-7. 38 33. Burgos-Vargas R, Vazquez-Mellado J, Cassis N, et al. Genuine ankylosing spondylitis in children: a 39 case-control study of patients with early definite disease according to adult onset criteria. J Rheumatol 40 1996;23:2140-7. 41 34. Heuft-Dorenbosch L, van Tubergen A, Spoorenberg A, Landewe R, Dougados M, Mielants H, van der 42 Tempel H, van der Heijde D. The influence of peripheral arthritis on disease activity in ankylosing on September 29, 2021 by guest. Protected copyright. 43 44 spondylitis patients as measured with the Bath Ankylosing Spondylitis Disease Activity Index. Arthritis 45 and rheumatism 2004;51:154-9. 46 35. Roussou E, Sultana S. The Bath Ankylosing Spondylitis Activity and Function Indices (BASDAI and 47 BASFI) and their correlation with main symptoms experienced by patients with spondyloarthritis. Clinical 48 rheumatology 2010;29:869-74. 49 36. Reveille JD. The genetic basis of spondyloarthritis. Ann Rheum Dis 2011;70:i44-50. 50 51 37. Reveille JD, Ball EJ, Khan MA. HLA-B27 and genetic predisposing factors in spondyloarthropathies. 52 Curr Opin Rheumatol 2001;13:265-72. 53 38. Valle R, Londono J, Velez P, et al. Outcome of patients with undifferentiated seronegative 54 spondyloarthropathy. Arthritis and rheumatism 1997;40:270. 55 39. Londono J, González L, Ramírez A, et al. Caracterización de las espondiloartropatías y determinación 56 de factores de mal pronóstico en una población de pacientes colombianos. Revista Colombiana de 57 Reumatología 2005;12:195-207. 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 16 1 2 3 40. Sampaio-Barros PD, Bortoluzzo AB, Conde RA, et al. Undifferentiated spondyloarthritis: a longterm 4 followup. J Rheumatol 2010;37:1195-9. 5 6 41. Mielants H, Veys EM, De Vos M, et al. The evolution of spondyloarthropathies in relation to gut 7 histology. I. Clinical aspects. J Rheumatol 1995;22:2266-72. 8 42. Burgos-Vargas R, Clark P. Axial involvement in the seronegative enthesopathy and arthropathy 9 syndrome and its progression to ankylosing spondylitis. J Rheumatol 1989;16:192-7. 10 43. Martinez-Laso J, Herraiz MA, Vidart JA, et al. Polymorphism of the HLA-B*15 group of alleles is 11 generated following 5 lineages of evolution. Human immunology 2011;72:412-21. 12 13 44. Stein JM, Machulla HK, Smeets R, et al. Human leukocyte antigen polymorphism in chronic and 14 aggressive periodontitis among Caucasians: a meta-analysis. Journal of clinical periodontology 15 2008;35:183-92. For peer review only 16 45. Choukri F CA, Himmich H, Hue S, Caillat-Zucman S. HLA-B*51 and B*15 alleles confer predisposition 17 to Behcet's disease in Moroccan patients. Human immunology 2001;62:180-5. 18 46. Turnquist HR TH, Prilliman KR, Lutz CT, Hildebrand WH, Solheim JC. HLA-B polymorphism affects 19 interactions with multiple endoplasmic reticulum proteins. European journal of immunology 20 21 2000;30:3021-8. 22 47. McHugh K, Bowness P. The link between HLA-B27 and SpA--new ideas on an old problem. 23 Rheumatology (Oxford, England) 2012;51:1529-39. 24 48. Chen L, Fischer R, Peng Y, et al. Critical role of endoplasmic reticulum aminopeptidase 1 in 25 determining the length and sequence of peptides bound and presented by HLA-B27. Arthritis & 26 rheumatology (Hoboken, N.J.) 2014;66:284-94. 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 17 1 2 3 Figure Legends 4 5 6 7 8 Fig. 1 SpA Classification Criteria in relation to HLAB15 or HLAB27 9 10 ESSG Criteria (The European Spondyloarthropathy Study Group Classification Criteria for 11 12 Spondyloarthropathy Criteria): AS, ankylosing spondylitis; uSpA, undifferentiated 13 14 spondyloarthritis; ReA, reactive arthritis. ASAS Criteria (The Assessment of SpondyloArthritis 15 For peer review only 16 international Society Classification Criteria): axSpA, axial spondyloarthritis; pSpA, peripheral 17 18 19 spondyloarthritis. 20 21 22 23 Fig. 2 SpA manifestations in patients positive for HLAB15 or HLAB27 24 25 (A) Percentage of patients with joint abnormalities in relation to their genotype: XRay SIJ, 26 27 sacroiliacjoint inflammation (sacroiliitis), grade ≥ 2 as assessed by xray of the pelvis; MRI SIJ, 28 29 acute changes detected by nuclear magnetic resonance imaging in sacroiliac joint; Sacroiliac 30 31 32 Enthesis, enthesitis in sacroiliac region assessed by Mander’s enthesis index. (B) Mean number 33 of joints affected in each patient in relation to the respective HLAB genotype: Swollen Joints, 34 http://bmjopen.bmj.com/ 35 36 mean number of joints that are swollen upon assessment of 66 joints in each patient; Overall 37 38 enthesitis: mean of total entheses (axial and peripheral) affected; Peripheral enthesitis: mean of 39 40 peripheral entheses affected. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 Fig. 3 Grading of disease activity in relation to HLAB15 or HLAB27 46 47 BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; BASFI, Bath Ankylosing 48 49 Spondylitis Functional Index; VAS Patient, Visual Analog Scale of Patient’s global assessment 50 51 of disease activity; VAS Physician, Visual Analog Scale of Physician’s global assessment of 52 53 disease activity; VAS IBP, Visual Analog Scale of Inflammatory Back Pain. Each score ranges 54 55 from 0 to 10; median and 25th and 75th percentiles are shown; *P < 0.0001, **P = 0.009 56 57 58 (Student’s ttest). 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 18 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 19 1 2 3 TABLE 1 Demographic and clinical stratification of SpA patients in relation to their HLAB genotype 4 5 6 HLAB15 HLAB27 Other HLAB Significant Variable 7 (n = 34) (n = 70) (n = 74) P values 8 9 Age ± SD years, mean ± SD (n = 178) 35.7 ± 10.0 35.8 ± 13.6 35.8 ± 13.1 10 a 11 Age of onset years, mean ± SD (n = 178) 28.2 ± 9.4 27.7 ± 10.5 32.6 ± 10.9 0.008 12 13 Disease Duration years, median (IQR) 3.4 (2 – 12) 3.6 (1 – 12) 3.0 (1 – 11.5)

14 b Sex – number of men/women (%) 15/19 (44/56) 55/15 (79/21) 30/44 (40/60) 0.001 15 Men:Women ratio For peer review0.8:1 only3.7:1 1.46:1 16 17 SpA patients classified according to ESSG criteria 18 Patients having AS (n = 91) 13 41 37 c 19 Patients having uSpA (n = 57) 17 15 25 0.006 20 Patients having ReA (n = 30) 4 14 12 21 Total (n = 178) 34 70 74

22 SpA patients classified according to ASAS criteria 23 h d 24 Patients with axSpA manifestations (n = 152) 27 70 55 < 0.0001 0 10 2 e 25 With pure axSpA (n = 12) 0.050 27 60 53 26 With mixed ax/pSpA (n = 140)

27 h f Patients with pSpA manifestations (n = 161) 34 60 67 0.050 28 4 0 9 g With pure pSpA (n = 13) 0.015 29 With mixed ax/pSpA (n = 148) 30 60 58 30 31 Percentages within parenthesis are calculated with respect to the number of patients in each HLAB group. IQR, 32 33 interquartile range; SpA, spondyloarthritis (classified according to the ESSG criteria as AS, ankylosing spondylitis;

34 http://bmjopen.bmj.com/ 35 uSpA, undifferentiated spondyloarthritis and ReA, reactive arthritis; classified according to the ASAS criteria as axSpA, 36 37 axial spondyloarthritis and pSpA, peripherial spondyloarthritis). Shown significant P values correspond as follows: aP = 38 39 0.008, between HLAB27 and other HLAB; bP = 0.001, between HLAB27 and the HLAB15; cP = 0.006, between HLA 40 41 B15 and HLAB27; dP < 0.0001, between HLAB27 and other HLAB; eP = 0.050, between HLAB15 and HLAB27; fP = 42 on September 29, 2021 by guest. Protected copyright. 43 0.050, between HLAB27 and HLAB15; gP = 0.015, between HLAB15 and HLAB27. (n = 152)h, among our 178 44 45 patients satisfying the ESSG classification criteria for SpA [2], only 152 met the ASAS criteria for axSpA [17] and 26 46 47 patients did not; (n = 161)h indicates that 161 patients (but not remaining 17 patients) met the ASAS criteria for pSpA 48 49 [16]. 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 20 1 2 3 TABLE 2 Association of HLAB15 and HLAB27 with SpA subsets 4 5 6 HLAB15 HLAB27 7 Disease OR (95% CI) OR (95% CI) 8 9 SpA 14.9* (3.4 – 90.8) 9.1* (3.8 – 22.8) 10 11 SpA subsets according to ESSG criteria 12 13 AS 13.2* (2.8 – 85.1) 15.7* (6.1 – 41.8) 14 15 uSpA For peer 20.9*review (4.4 – 134.9) only13.8 (1.4 – 11.2) 16 ReA 10.1 (1.7 – 77.3) 11.2* (2.7 – 35.3) 17 18 SpA subsets according to ASAS criteria 19 20 axSpA 8.2* (1.9 35.6) 9.8* (4.3 – 22.5) 21 22 pSpA 15.4* (3.6 – 65.4) 7.0* (3.1 – 16.1) 23 24 OR, odds ratio; 95% CI, 95% confidence intervals. *P < 0.05, patients versus controls by Chisquared 25 26 tests. SpA, spondyloarthritis, classified according to the ESSG European Spondyloarthropathy Study 27 28 Group criteria as AS, ankylosing spondylitis; uSpA, undifferentiated spondyloarthritis and ReA, reactive 29 30 arthritis. SpA classified according to the ASAS Assessment of SpondyloArthritis international Society 31 32 criteria as axSpA, axial spondyloarthritis and pSpA, peripherial spondyloarthritis. 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 21 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 Fig. 1 SpA Classification Criteria in relation to HLAB15 or HLAB27 45 46 ESSG Criteria (The European Spondyloarthropathy Study Group Classification Criteria for 47 48 Spondyloarthropathy Criteria): AS, ankylosing spondylitis; uSpA, undifferentiated 49 50 spondyloarthritis; ReA, reactive arthritis. ASAS Criteria (The Assessment of SpondyloArthritis 51 52 53 international Society Classification Criteria): axSpA, axial spondyloarthritis; pSpA, peripheral 54 55 spondyloarthritis. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 22 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 Fig. 2 SpA manifestations in patients positive for HLAB15 or HLAB27 38 39 (A) Percentage of patients with joint abnormalities in relation to their genotype: XRay SIJ, 40 41 sacroiliacjoint inflammation (sacroiliitis), grade ≥ 2 as assessed by xray of the pelvis; MRI SIJ, 42 on September 29, 2021 by guest. Protected copyright. 43 acute changes detected by nuclear magnetic resonance imaging in sacroiliac joint; Sacroiliac 44 45 46 Enthesis, enthesitis in sacroiliac region assessed by Mander’s enthesis index. (B) Mean number 47 48 of joints affected in each patient in relation to the respective HLAB genotype: Swollen Joints, 49 50 mean number of joints that are swollen upon assessment of 66 joints in each patient; Overall 51 52 enthesitis: mean of total entheses (axial and peripheral) affected; Peripheral enthesitis: mean of 53 54 peripheral entheses affected. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 Fig. 3 Grading of disease activity in relation to HLAB15 or HLAB27

34 http://bmjopen.bmj.com/ 35 BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; BASFI, Bath Ankylosing 36 37 38 Spondylitis Functional Index; VAS Patient, Visual Analog Scale of Patient’s global assessment 39 40 of disease activity; VAS Physician, Visual Analog Scale of Physician’s global assessment of 41 42 disease activity; VAS IBP, Visual Analog Scale of Inflammatory Back Pain. Each score ranges on September 29, 2021 by guest. Protected copyright. 43 44 from 0 to 10; median and 25th and 75th percentiles are shown; *P < 0.0001, **P = 0.009 45 46 (Student’s ttest). 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35

36 37 38 39 Fig. 1 SpA Classification Criteria in relation to HLA-B15 or HLA-B27 40 41 ESSG Criteria (The European Spondyloarthropathy Study Group Classification Criteria for 42 on September 29, 2021 by guest. Protected copyright. 43 Spondyloarthropathy Criteria): AS, ankylosing spondylitis; uSpA, undifferentiated 44 45 46 spondyloarthritis; ReA, reactive arthritis. ASAS Criteria (The Assessment of 47 48 SpondyloArthritis international Society Classification Criteria): axSpA, axial 49 50 spondyloarthritis; pSpA, peripheral spondyloarthritis. 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 26 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ Fig. 2 SpA manifestations in patients positive for HLA-B15 or HLA-B27 35 36 37 (A) Percentage of patients with joint abnormalities in relation to their genotype: X-Ray SIJ, 38 39 sacroiliac-joint inflammation (sacroiliitis), grade ≥ 2 as assessed by x-ray of the pelvis; MRI 40 41 SIJ, acute changes detected by nuclear magnetic resonance imaging in sacroiliac joint; 42 on September 29, 2021 by guest. Protected copyright. 43 Sacroiliac Enthesis, enthesitis in sacroiliac region assessed by Mander’s enthesis index. 44 45 (B) Mean number of joints affected in each patient in relation to the respective HLA-B 46 47 genotype: Swollen Joints, mean number of joints that are swollen upon assessment of 66 48 49 50 joints in each patient; Overall enthesitis: mean of total entheses (axial and peripheral) 51 52 affected; Peripheral enthesitis: mean of peripheral entheses affected. 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 26 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 Fig. 3 Grading of disease activity in relation to HLA-B15 or HLA-B27 29 30 31 BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; BASFI, Bath Ankylosing 32 33 Spondylitis Functional Index; VAS Patient, Visual Analog Scale of Patient’s global assessment

34 http://bmjopen.bmj.com/ 35 of disease activity; VAS Physician, Visual Analog Scale of Physician’s global assessment of 36 37 disease activity; VAS IBP, Visual Analog Scale of Inflammatory Back Pain. Each score ranges 38 39 from 0 to 10; median and 25th and 75th percentiles are shown; *P < 0.0001, **P = 0.009 40 41 42 (Student’s t-test). on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

Analysis of HLA-B15 and HLA-B27 in spondyloarthritis with peripheral and axial clinical patterns

ForJournal: peerBMJ Open review only Manuscript ID bmjopen-2015-009092.R1

Article Type: Research

Date Submitted by the Author: 08-Sep-2015

Complete List of Authors: LONDONO, JOHN; Universidad de La Sabana, Rheumatology; Universidad de Navarra, Department of Internal Medicine Santos, Ana; Universidad de La Sabana, Immunology Peña, Paola; Hospital Militar Central, Rheumatology Calvo, Enrique; Universidad Nacional de Colombia, Radiology Espinosa, Luis; Louisiana State University, LSU Health Sciences Center Reveille, John; University of Texas Health Science Center, USA, Department of Internal Medicine, Division of Rheumatology and Clinical Immunogenetics Vargas-Alarcon, Gilberto; Inst Nacl Cardiol Ignacio, Jaramillo, Calos; Universidad de Los Andes, Department of Biological Sciences Valle-Oñate, Rafael; Hospital Militar Central, Rheumatology Avila, Mabel; Hospital Militar Central, Dirección de Educación e Investigacion http://bmjopen.bmj.com/ Romero, Consuelo; Hospital Militar Central, Rheumatology Medina, Juan; Universidad de Navarra, Internal Medicine

Primary Subject Rheumatology Heading:

Secondary Subject Heading: Immunology (including allergy)

Keywords: RHEUMATOLOGY, IMMUNOLOGY, GENETICS on September 29, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 1 1 2 3 4 5 6 Analysis of HLA-B15 and HLA-B27 in spondyloarthritis with peripheral 7 8 9 and axial clinical patterns 10 11 12

13 1 1 1 2 3 14 John Londono , Ana Maria Santos , Paola Peña , Enrique Calvo , Luis R. Espinosa , 15 For4 peer review5 only6 7 16 John D. Reveille , Gilberto Vargas-Alarcon , Carlos A. Jaramillo , Rafael Valle-Oñate , 17 1,7 7 8 18 Mabel Avila , Consuelo Romero and Juan F. Medina 19 20 21 22 23 1 2 24 Department of Rheumatology, University of La Sabana, Chia, Cundinamarca, Colombia, Department of 25 3 26 Radiology, National University of Colombia, Bogota, Colombia, LSU Health Sciences Center, Louisiana

27 4 28 State University, New Orleans, USA, Division of Rheumatology and Clinical Immunogenetics, 29 30 Department of Internal Medicine, University of Texas Health Science Center, Houston, USA,

31 5 32 Interventional Genetic Study Group in Cardiovascular Diseases, Department of Molecular Biology, 33 National Institute of Cardiology Ignacio Chávez, Mexico DF, 6Department of Biological Sciences, 34 http://bmjopen.bmj.com/

35 7 36 University of Los Andes, Bogota, Colombia, Rheumatology and Immunology Unit, Department of Internal

37 8 38 Medicine, Hospital Militar Central, Bogota, Colombia, Department of Internal Medicine, School of 39 40 Medicine, University of Navarra and Ciberehd, Pamplona, Spain. 41

42 on September 29, 2021 by guest. Protected copyright. 43

44 45

46 47 Correspondence to: John Londono, MD, Department of Rheumatology, University of La Sabana and 48 49 Hospital Militar Central, Campus del Puente del Comun, Km. 7, Autopista Norte de Bogotá. CP 250001, 50 51 Chía, Cundinamarca, Colombia. Tel. (571)8613503; FAX: (571)8615618. Email: 52 53 [email protected] 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 2 1 2 3 Abstract 4 5 6 Objective. HLAB27 and B15 are associated with spondyloarthritis (SpA). Recent ASAS criteria 7 8 emphasize a distinction between SpA with axial and peripheral pattern. We analyzed whether HLAA, B 9 10 and DRB1 alleles could associate with these patterns. 11 12 Methods. We studied 100 healthy individuals and 178 patients with SpA according to ESSG criteria. 13 14 Patients were then classified according to ASAS criteria, the axial pattern (axSpA) being defined by 15 For peer review only 16 ascertained sacroiliitis and the peripheral pattern (pSpA) by enthesitis and/or arthritis in extremities. A 17 18 combined ax/p pattern was also considered. 19 20 21 Results. Only HLAB27 and B15 alleles were associated with SpA. ASAS criteria for axSpA were met in 22 23 152 patients (12 with isolated axSpA and 140 with a combined ax/p pattern). When the ASAS peripheral 24 25 criteria were applied, 161 patients met these criteria (13 with isolated pSpA and 148 with a combined ax/p 26 27 pattern). HLAB27 was found in 83% of patients with axSpA and 43% of ax/pSpA patients according to 28 29 axASAS. HLAB27 occurred in 7% controls but not in any patient with isolated pSpA. HLAB15 was 30 31 encountered in 31% of patients with isolated pSpA and 20% of ax/pSpA patients according to pASAS 32 33 criteria. Moreover, 2 healthy controls, but none of our patients with isolated axSpA were positive for HLA

34 http://bmjopen.bmj.com/ 35 B15. 36 37 Conclusion. Our data suggest that the presence of HLAB15 favors the development of 38 39 isolated/combined peripheral rather than isolated axSpA, while HLAB27 promotes a isolated/combined 40 41 axial disease and exclude a peripheral pattern. HLAB15 should be considered in addition to HLAB27 42 on September 29, 2021 by guest. Protected copyright. 43 when diagnosing SpA patients according to ASAS criteria. 44 45

46 47 Key words: spondyloarthritis; axial spondyloarthritis; peripheral spondyloarthrtis, HLAB27; HLAB15. 48 49 50 51 52

53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 3 1 2 3 Strengths and limitations of this study 4 5 • The HLAB15 has been associated with SpA and oligoarthritis in different populations. This study 6 7 clearly establishes this association, through a detailed analysis of the clinical and radiological aspects 8 9 in a group of patients not AngloSaxon origin. 10 11 • In this study, conducted on a group of Latin American patients with SpA, the presence of HLAB15 12 13 and HLAB27 clearly influence the pattern of presentation of the disease. The HLAB15 favors a 14 15 preferably peripheralFor pattern peer while the HLAB27review promotes a isolated/combinedonly axial disease and 16 17 exclude a isolated peripheral pattern. 18 19 • HLAB15 should be considered in addition to HLAB27 when diagnosing SpA Colombian patients 20 21 according to ASAS criteria. 22 23 • This study used the ASAS criteria in an alternating manner to discriminate the association of the 24 25 disease with the HLAB27 and HLAB15. 26 27 • The results of this study need to be verified in further research studies involving a larger number of 28 29 patients with SpA. 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 4 1 2 3 Introduction 4 5 6 Spondyloarthritis (SpA) comprises a group of inflammatory chronic diseases that share similar clinical 7 8 presentations, radiological findings, HLAB27 association, and positive family history.[1] According to the 9 10 European Spondyloarthropathy Study Group (ESSG) criteria,[2] SpA has been traditionally classified as 11 12 follows: ankylosing spondylitis (AS), reactive arthritis (ReA), undifferentiated spondyloarthritis (uSpA), 13 14 psoriatic arthritis (PsA), and arthritis associated with inflammatory bowel disease (IBD).[3] Its prevalence 15 For peer review only 16 varies from 0.1 to 1.4% of the general population depending on the geographical region studied.[4] 17 18 Despite the fact that SpA tends to affect males more frequently and severely than females, an increasing 19 20 proportion of female patients is being observed during the last decade among patients with less severe 21 22 clinical manifestations.[5] Predominantly, SpA begins in individuals younger than 45 years of age and 23 24 clinically compromises the axial spine, peripheral joints, tendons and ligament insertions (entheses), 25 26 having also extramusculoskeletal manifestations in the skin, mucosa, gut and the eyes.[6] 27 28 HLAB27 is the biggest risk factor for the development of SpA, and while it does have a central role 29 30 in its etiopathogenesis, the association is not the same across all ethnic groups.[7] It has been estimated 31 32 that this allele is responsible for up to 28% of the etiology of SpA.[8] But there are other factors that have 33 an impact on the onset of the disease, such as environmental factors, and other genetic factors, including

34 http://bmjopen.bmj.com/ 35 36 those within the Major Histocompatibility Complex (MHC) Class I − HLAB60/B40,[9] B39,[10] B14,[11] 37 38 and B15[12] − and Class II − DRB1[13] − alleles. 39 40 Current new technologies are allowing early diagnosis and more specific examination of disease 41 42 susceptibility, such as those that look at Single Nucleotide Polymorphisms (SNPs), GenomeWide on September 29, 2021 by guest. Protected copyright. 43 44 Association Studies (GWAS) and genomic resequencing which underscore the polygenic nature of 45 46 SpA.[14 15] There are more than 40 genes implicated in the pathogenesis of SpA which may also affect 47 48 the clinical manifestations, including the age of onset and the duration and severity of the disease.[16] 49 50 The diagnosis of SpA is made principally based on clinical, radiologic, and laboratory findings. The 51 52 classification currently proposed by the European League Against Rheumatism (EULAR) and the 53 Assessment of SpondyloArthritis international Society (ASAS) shows a high sensitivity and specificity 54 55 (79.5% and 83.3%, respectively).[17 18] According to these entities SpA can have two general clinical 56 57 patterns: axial and peripheral. With these two clinical patterns, there may be a genetic variability that 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 5 1 2 3 accounts for each physical distribution. In fact, data emerging from immunopathological studies and from 4 5 clinical trials suggest that the pathogenesis of axial and peripheral disease may be caused by different 6 7 mechanisms influenced by genetic differences, a situation that underscores the importance of the new 8 9 clinical classification criteria from ASAS for peripheral and axial SpA.[19] The axial criteria rely on two 10 11 major components, imaging and the presence of HLAB27.[20] The latter was adopted because of its high 12 13 sensitivity and specificity and its extensive validation for axSpA. However, the validity of the HLAB27 14 15 component is lessFor defined for peerpSpA and uSpA. review In uSpA with peripheral only involvement, for instance, about 16 17 60% of patients are negative for HLAB27.[21] On the other hand, previous studies in patients with SpA 18 19 have reported an association with HLAB15 in Europe (Belgium),[22] North Africa (Tunisia),[23] and Latin 20 21 America (Mexico),[24] where the most common presentations of SpA were related to IBD, ReA, and 22 23 uSpA. In particular, peripheral involvement was nearly ubiquitous in these cohorts. This would suggest 24 25 that HLAB15 allele may play an important role in the susceptibility to peripheral SpA.[25] 26 27 The objective of this study was to evaluate the impact of HLAB15 and HLAB27 alleles on the 28 29 susceptibility and the clinical characteristics of the disease (considered according to either ESSG or 30 31 ASAS criteria) in a Colombian population with SpA. 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 6 1 2 3 Patients and methods 4 5 6 Participants and clinical assessment of patients 7 8 A group of 178 patients with a diagnosis of SpA according to the ESSG criteria were studied. These 9 10 patients were seen in the Central Military Hospital in Bogota, Colombia, between January 2005 and 11 12 December 2011. All patients were evaluated by a Rheumatologist in the outpatient care department with 13 14 subsequent HLA genotyping. Demographic and clinical data, including disease activity [graded through 15 For peer review only 16 the Bath Ankylosing Spondylitis Disease Activity Index, BASDAI[26] as well as the global disease activity 17 18 assessed by Physician, i.e. the Visual Analog Scale of Physician’s global assessment of disease activity, 19 20 abbreviated as VAS Physician,[27] the VAS global disease activity assessed by patients, abbreviated as 21 22 VAS Patient,[27] and the Visual Analog Scale of Inflammatory Back Pain, abbreviated as VAS IBP,[27] 23 24 functional capacity (graded through the Bath Ankylosing Spondylitis Functional Index, BASFI),[28] 25 26 inflammatory back pain, enthesitis and joints affected were recorded. Recommendations for followup 27 28 were according to the previously validated ASAS guidelines.[27] Each enthesis was evaluated using the 29 30 Mander’s enthesis index.[29] For each patient, laboratory exams that included the erythrocyte 31 32 sedimentation rate were taken. 33

34 The control group consisted of 100 healthy subjects on the list of transplant donors at the Central http://bmjopen.bmj.com/ 35 36 Military Hospital in Bogota, Colombia. These subjects had the same sociodemographic characteristics as 37 38 our SpA population. Cases and controls were compared for the HardyWeinberg equilibrium as tested 39 40 with the Arlequin® statistical software version 3.5.[30] Allele frequency (AF) among the patient population 41 42 was calculated using Genepop Program.[31] on September 29, 2021 by guest. Protected copyright. 43 44 45 46 Ethics approval 47 48 The study followed the norms established by the Helsinki Declaration, The Guidelines for Good Clinical 49 50 Practice, and the Resolution 8430 (1993) of the Colombian Ministry for Social Protection. Moreover, was 51 52 reviewed and approved by the Ethics Committees of the University of La Sabana, Los Andes University, 53 54 and the Central Military Hospital. Each patient signed an informed consent form, and confidentiality was 55 56 strictly maintained. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 7 1 2 3 4 5 6 Imaging 7 8 A plain anterior pelvic xray was taken and Magnetic Resonance Imaging (MRI) of the sacroiliac joints 9 10 with T1 sequence and, fat suppression technique was taken on each patient. This was then read by a 11 12 radiologist with musculoskeletal expertise who was blinded to the clinical picture. An interpretation of the 13 14 xrays was made following the recommendations for population studies in the Atlas of Standard 15 For peer review only 16 Radiographs[32] and the MRI was evaluated looking for acute and chronic changes in the joints.[33] 17 18 19 20 HLA genotyping 21 22 23 Genomic DNA was extracted from peripheral blood leukocytes with a Wizard genomic DNA purification 24 25 Kit (Promega, Cat. # A1120). After extraction, samples were adjusted to a DNA concentration of 80 ± 20 26 g/L and stored at –80 °C until processing of HLA. Typing was carried out by PCR on the genomic DNA 27 28 using the HLA Sequence Specific Primer HLAABDRSSPtray Kit (BioRad, Cat. # 826 230), which is 29 30 focused on determining HLAA, B, and the DRB1 alleles. PCR products were verified by electrophoresis 31 32 in a 2% agarose gel, and bands were visualized by ultraviolet light photo documentation and further 33

34 http://bmjopen.bmj.com/ analyzed by using the HLASSP Typing V 1.2.0.0. Software (BioRad). 35 36 37 38 Statistical analyses 39 40 41 Data are expressed as mean ± SD (except when otherwise indicated). For statistical analysis we used 42 on September 29, 2021 by guest. Protected copyright. 43 SPSS 19.0 (SPSS Inc, USA). For the presentation and analysis of categorical variables, frequency 44 45 distribution, percentages and contingency tables were used. Comparisons were made using the Chi 46 47 square test and Fisher Exact test. For numerical values, central tendency and dispersion measurements 48 49 were evaluated. Results were analyzed using Student’s t and Wilcoxon tests. Twotailed P values < 0.05 50 51 were considered statistically significant. The association between disease and HLA alleles was 52 53 determined by Odds Ratio and confidence intervals of 95% (OR: 95% CI). 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 8 1 2 3 Results 4 5 6 Genotype and demographic data 7 8 Only HLAB27 and B15 out of all the analyzed HLAA, B and DRB1 alleles were found to be significantly 9 10 associated with SpA. As shown in table 1, our 178 SpA patients (included in the study because they met 11 12 the ESSG criteria) could therefore be divided into three groups according to their HLAB genotype: 34 13 14 patients (19.1%) had HLAB15, 70 (39.3%) showed HLAB27 and 74 (41.6%) had other HLAB alleles. 15 For peer review only 16 Patients with HLAB15 and HLAB27 had an earlier age of disease onset than patients with other HLAB 17 18 alleles. However, the earlier age was only significant considering the patients with HLAB27 (P = 0.008). 19 20 A male predominance (with a male/female ratio of 3.7/1) was documented in the group of 70 HLAB27 21 22 positive patients, whereas females were slightly predominating (male/female ratio: 0.8/1) in the group of 23 24 34 HLAB15 positive patients. In this regard, however, the statistical difference between these two groups 25 26 (i.e. HLAB27 and HLAB215 patients) was highly significant (table 1). 27 28 When ASAS classification criteria [17 18] were applied to classify our patients 178 previously 29 30 diagnosed according to the ESSG criteria, 152 patients fulfilled the ASAS criteria for pSpA while 26 did 31 32 not, and 161 patients met the ASAS criteria for pSpA while 17 patients did not. Most of the patients 33

34 meeting the ASAS criteria for axSpA and/or pSpA, manifested a combined ax/p pattern, though the total http://bmjopen.bmj.com/ 35 36 number of these patients slightly fluctuated. This fluctuation was dependent on the application to our 37 38 patients of either the axial or peripheral criteria (table 1), these two sets of ASAS criteria being 39 40 simultaneously met in 137 patients. Moreover, 12 patients had isolated axSpA and 13 showed isolated 41 42 pSpA (table 1). Noticeably, 100% of the 70 HLAB27 positive patients included in the study exhibited axial on September 29, 2021 by guest. Protected copyright. 43 44 manifestations of the disease (mainly back pain) and 100% of the 34 patients positive for HLAB15 had 45 46 peripheral manifestations of SpA (mainly join arthritis). Among the 12 patients with isolated axSpA, 10 of 47 48 them (83%) were positive for HLAB27, and among the 13 patients with isolated pSpA, 4 of them (31%) 49 50 were positive for HLAB15. Moreover, HLAB27 was absent in all 13 pSpA patients (table 1), though this 51 52 allele did occur in 7% of the 100 healthy control subjects (not shown). On the other hand, while two 53 54 healthy controls (2%) were positive for HLAB15, none of the 12 axSpA were positive for this allele. 55 56 Further demographic data and clinical presentation related to either ESSG or ASAS criteria and to HLA 57 58 B27, HLAB15, and other B alleles are also shown in table 1. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 9 1 2 3 Association of HLA-B15 and HLA-B27 alleles to clinical SpA subsets 4 5 6 When SpA patients were analyzed as a whole, we found a statistical association of the disease with both 7 8 HLAB15 and HALB27 (table 2). However, when the clinical subsets were considered, HLAB15 was 9 10 found to be mainly associated with uSpA and pSpA forms, whereas HLAB27 was mainly associated with 11 12 AS (table 2). The relationships of HLAB27 and HLAB15 with clinical forms of presentation according to 13 14 either the ESSG criteria or the ASAS criteria are further summarized in figure 1. This figure graphically 15 For peer review only 16 illustrates our findings that none of the patients with isolated axial SpA are HLAB15 positive and that 17 18 HLAB27 does not occur in patients with just peripheral SpA. These findings appear to strengthen the 19 20 notion that ASAS criteria might allowed a better correlation between the clinical presentation of SpA and 21 22 the HLAB genotype than the ESSG criteria. 23

24 25 HLA-B15 and HLA-B27 alleles and disease characteristics 26 27 28 The most relevant clinical features and radiological findings in relation to HLA alleles are shown in figure 29 30 2. Remarkably, patients with HLAB15 had more swollen joints (5.3±5.1 vs 2.3±1.8, P = 0.004) and total 31 32 count of painful enthesitis (6.0±4.5 vs 4.3±3.3, P = 0.014), and peripheral painful enthesitis (3.4±2.9 vs 33

34 2.3±2.2, P = 0.050), while HLAB27 positive patients had more frequent sacroiliitis of grade ≥ 2 (74.7% vs http://bmjopen.bmj.com/ 35 36 56.3%, P = 0.043) and enthesitis in sacroiliac region (64.4% vs 36%, P = 0.001). 37 38 39 40 41 HLA-B15 and HLA-B27 alleles and disease activity 42 on September 29, 2021 by guest. Protected copyright. 43 When the disease activity was taken into account, we found that HLAB15 positive patients had a higher 44 45 rate in the BASDAI index, BASFI index and global disease activity assessed by Physician (VAS 46 47 Physician) compared to HLAB27 positive patients (figure 3). On the other hand, HLAB27 positive 48 49 patients had a higher rate in inflammatory back pain and global disease activity assessed by patient (VAS 50 51 Patient) than HLAB15 positive patients (figure 3). Finally, it is noteworthy to mention that HLAB27 52 53 positive patients exhibited a higher erythrocyte sedimentation rate than HLAB15 positive patients (25.7 ± 54 55 24.4 mm/hour vs 16.5 ± 9.8, respectively). 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 10 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 10 1 2 3 Discussion 4 5 In this study we found that in addition to the strong association of SpA with HLAB27, SpA can be 6 7 associated with another HLAB allele, i.e. with HLAB15. Noticeably, the prevalence of HLAB15 was 8 9 particularly increased in SpA patients with peripheral involvement. Moreover, while patients with HLAB27 10 11 were more likely to be male and to show axial compromise associated with AS, HLAB15 positive patients 12 13 had a lower proportion of men and a more frequent uSpA clinical manifestation. 14 15 Patients withFor either HLAB15 peer or HLAB27 review had similar ages only of disease onset and comparable 16 17 duration of the disease. From a clinical point of view, HLAB15 patients had an increased number of 18 19 peripheral joints affected than HLAB27 patients, while no differences in spinal mobility were observed 20 21 between these two groups of patients. Radiographically, HLAB27 patients showed more severe 22 23 sacroiliitis in pelvic xrays and greater acute damage in MRI than patients positive for HLAB15. 24 25 Male gender, axial involvement, early age of disease onset (< 20 yrs), and the presence of HLA 26 27 B27 are all known to be associated with a bad prognosis.[34] In our study, patients with HLAB15 28 29 (consisting of slightly fewer males than females) exhibited a less severe disease phenotype and more 30 31 peripheral symptoms as compared to patients positive for HLAB27. Concerning the disease activity in 32 33 our patients, however, both BASDAI and BASFI scores were higher (worse) in HLAB15 positive patients

34 http://bmjopen.bmj.com/ 35 than in those with HLAB27. This may be related to important functional impairment and higher 36 37 inflammatory disease activity in a greater number of lower extremity joints and enthesitis in the HLAB15 38 39 group.[35 36] On the other hand, the acute phase reactants were particularly increased in the HLAB27 40 41 positive patients. 42 on September 29, 2021 by guest. Protected copyright. 43 HLAB27 has been known for over 40 years as a predisposing factor for SpA, specifically for 44 45 AS.[37] In Caucasians with SpA, 8085% of patients are HLAB27 positive. Overall, however, less than 46 47 5% of the Caucasian patients with this allele go on to develop the disease.[38] This suggests that even 48 49 though HLAB27 appears to play a major role in the development of the disease, it is not enough to 50 51 completely explain the disease susceptibility. In fact, in populations other than Caucasians, HLAB27 may 52 53 be less prevalent. Thus, in mestizo population, which had been arising from the mix of the preColumbian 54 55 indigenous individuals with Spanish white and African black subjects, HLAB27 has been reported in 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 11 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 11 1 2 3 between 4060% of patients with SpA.[39 40] This is further confirmed in our present study population, in 4 5 which 70 out of 178 SpA patients (39.3%) were positive for HLAB27. 6 7 In people of AngloSaxon descent, AS is the most common subtype of SpA, while uSpA is the 8 9 predominant subtype in the Latin American mestizo population.[40] This variation could be due to 10 11 associations of the disease with different HLA genes in addition to HLAB27.[41] The new criteria recently 12 13 proposed by ASAS to classify SpA allow for grouping the patients according to severity, onset, and axial 14 15 or peripheral patternFor of the disease.peer It is remarkable review that in our study, only 100% of the HLAB15 positive 16 17 patients met with the ASAS peripheral criteria and 100% of the HLAB27 positives fulfilled the ASAS axial 18 19 criteria. 20 21 There had been previous reports partially considering the HLAB15 allele in relation to SpA. Thus, a 22 23 study in a Mexican population, in addition to showing a strong association of HLAB27 with SpA, also 24 25 suggested that prevalence of HLAB15 was increased in the whole SpA cohort.[24] In fact, HLAB15 was 26 27 modestly increased in uSpA but not in AS patients, whereas HLAB27 showed a greater association with 28 29 AS than with uSpA.[24] Likewise, a prospective study in Belgium examined patients with different 30 31 subtypes of SpA, and found that HLAB15 was more frequent in those patients with uSpA (16/51, 32%) 32 33 as well as in those with a clinical remission of symptoms.[42] HLAB27, however, was more prevalent in

34 http://bmjopen.bmj.com/ 35 those patients with AS and in those with sustained disease activity. Also in Tunisia HLAB15 was 36 37 associated with uSpA as well as with ReA. In fact, this allele was found in up to 50% of patients with 38 39 uSpA. As this observation is very similar to previous studies carried out in Latin America, Europe, and 40 41 Africa, it was suggested that it could be the result of population migrations.[43] The aforementioned data 42 on September 29, 2021 by guest. Protected copyright. 43 indicate that HLAB15 is found in various different ethnic groups, crossing geographical barriers, and 44 45 altogether these data further support the notion that HLAB15 plays a role in SpA pathogenesis.[44] 46 47 Inflammatory diseases other than SpA have also been associated with HLAB15. A metaanalysis 48 49 in a Caucasian population with acute periodontitis showed a strong association with this allele.[45] 50 51 Moreover, there seems to be a relationship between periodontal disease and the onset of inflammatory 52 53 diseases such as rheumatoid arthritis, which suggests that HLAB15 may be related to acute and chronic 54 55 diseases with local bone involvement.[45] Furthermore, HLAB15 was found to be associated with 56 Behçet’s disease in a Moroccan population, where females were the dominant affected population.[46] 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 12 1 2 3 HLAB15 is an HMC Class I molecule encoded by multiple polymorphic alleles leading to at least 7 4 5 serologic subtypes.[44] There had been described variants – with either tyrosine or serine at position 116 6 7 – which greatly diverge in their interaction (either strong or weak, respectively) with other proteins in the 8 9 endoplasmic reticulum for antigen presentation at the cell surface.[47] Accordingly, it might be expected 10 11 that genetic diversity can contribute to increased or decreased predisposition to autoimmune diseases. In 12 13 the case of HLAB15, recent studies have demonstrated that this molecule as well as HLAB14, B38, 14 15 B39, and B73 allFor have an unpairedpeer free cysteine review (at position 67) inonly the B pocket of the alpha heavy 16 17 chain, which is very similar to HLAB27. This may allow for the formation of homodimers that could 18 19 activate the innate immune response to produce cytokines implicated in chronic inflammation in the 20 21 pathogenesis of SpA.[48] 22 23 Recently, a highdensity genotyping study was carried out to identify multiple risk variants for 24 25 AS.[15] In addition to confirming the previously reported strong association of the disease with SNPs at 26 27 the gene for the endoplasmic reticulum aminopeptidase 1 (ERAP1) that was restricted to HLAB27– 28 29 positive AS patients, this study found a significant association of AS with another member of the 30 31 endoplasmic reticulum aminopeptidase family, ERAP2, which prevailed in AS patients negative for the 32 33 HLAB27 allele.[15] Future highdensity genotyping studies might uncover whether some (or most) of

34 http://bmjopen.bmj.com/ 35 these HLAB27–negative patients are positive for the HLAB15 allele. While ERAP1 seems to play a 36 37 critical role in determining the length and sequence of the peptides bound and presented by HLAB27 38 39 associated with AS,[49] it might be attractive to hypothesize that ERAP2 may play a parallel role in the 40 41 context of HLAB15. 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 Conclusions 46 47 Our study with a Mestizo cohort indicates that HLAB15 is associated with the peripheral manifestation of 48 49 SpA independently of sex or age, with a notable involvement of joints and enthesitis and with minor axial 50 51 compromise. HLAB15 may therefore become an important tool and marker for the disease in our 52 53 population, particularly in those patients with peripheral SpA without HLAB27. 54

55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 13 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 13 1 2 3 Acknowledgements 4 5 6 We are grateful to Maria del Pilar Delgado, MCs, Department of Biological Sciences, University of Los 7 8 Andes, for critical review of the manuscript. 9 10 11 12 13 14 Funding: This work was supported by the University of La Sabana and the University of Los Andes. 15 For peer review only 16 Data sharing: No additional data available. 17 18 All authors were involved in drafting the article or revising it critically for important 19 Author contributions: 20 21 intellectual content, and all authors approved the final version to be published. Dr. Londono had full 22 23 access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy 24 25 of the data analysis. 26 27 Study conception and design. Londono, Santos, Medina, VargasAlarcon, Espinoza, Reveille, Jaramillo, 28 29 Avila, Romero, ValleOñate 30 31 Acquisition of data. Londono, Santos, Pena. 32 33 Analysis and interpretation of data. Londono, Santos, Pena, Medina.

34 http://bmjopen.bmj.com/ 35 36 37 Disclosure statement: The authors declare no conflicts of interest. No financial support or benefits from 38 39 commercial sources, or any additional conflict of interest exist with regard to this work. 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 14 1 2 3 References 4 5 6 1. Khan MA. Update on spondyloarthropathies. Ann Intern Med 2002;136:896-907. 7 2. Dougados M, van der Linden S, Juhlin R, et al. The European Spondylarthropathy Study Group 8 preliminary criteria for the classification of spondylarthropathy. Arthritis and rheumatism 1991;34:1218- 9 27. 10 11 3. Dougados M, van der Heijde D. Ankylosing spondylitis: how should the disease be assessed? Best 12 practice & research. Clinical rheumatology 2002;16:605-18. 13 4. Ehrenfeld M. Geoepidemiology: the environment and spondyloarthropathies. Autoimmunity reviews 14 2010;9:325-9. 15 5. Haroon NN, PatersonFor JM, peer Li P, et al. Increasingreview proportion ofonly female patients with ankylosing 16 spondylitis: a population-based study of trends in the incidence and prevalence of AS. BMJ open 17 18 2014;4:e006634. 19 6. Dougados M, Hochberg MC. Why is the concept of spondyloarthropathies important? Best practice & 20 research. Clinical rheumatology 2002;16:495-505. 21 7. Gonzalez-Roces S, Alvarez MV, Gonzalez S, et al. HLA-B27 polymorphism and worldwide susceptibility 22 to ankylosing spondylitis. Tissue Antigens 1997;49:116-23. 23 8. Reveille JD. Genetics of spondyloarthritis--beyond the MHC. Nature reviews. Rheumatology 24 2012;8:296-304. 25 26 9. Robinson WP, van der Linden SM, Khan MA, et al. HLA-Bw60 increases susceptibility to ankylosing 27 spondylitis in HLA-B27+ patients. Arthritis and rheumatism 1989;32:1135-41. 28 10. Yamaguchi A, Tsuchiya N, Mitsui H, et al. Association of HLA-B39 with HLA-B27-negative ankylosing 29 spondylitis and pauciarticular juvenile rheumatoid arthritis in Japanese patients. Evidence for a role of 30 the peptide-anchoring B pocket. Arthritis and rheumatism 1995;38:1672-7. 31 11. Diaz-Pena R, Blanco-Gelaz MA, Njobvu P, et al. Influence of HLA-B*5703 and HLA-B*1403 on 32 susceptibility to spondyloarthropathies in the Zambian population. J Rheumatol 2008;35:2236-40. 33 12. Wagener P, Zeidler H, Eckert G, et al. Increased frequency of HLA-Bw62 and Bw35 CREG antigens in 34 http://bmjopen.bmj.com/ 35 HLA-B27 negative ankylosing spondylitis. Z Rheumatol 1984;43:253-7. 36 13. Brown MA, Kennedy LG, Darke C, et al. The effect of HLA-DR genes on susceptibility to and severity 37 of ankylosing spondylitis. Arthritis and rheumatism 1998;41:460-5. 38 14. van Gaalen FA, Verduijn W, Roelen DL, et al. Epistasis between two HLA antigens defines a subset of 39 individuals at a very high risk for ankylosing spondylitis. Ann Rheum Dis 2013;72:974-8. 40 41 15. International Genetics of Ankylosing Spondylitis Consortium (IGAS), Cortes A, Hadler J, et al. 42 Identification of multiple risk variants for ankylosing spondylitis through high-density genotyping of on September 29, 2021 by guest. Protected copyright. 43 immune-related loci. Nat Genet 2013;45:730-38. 44 16. Diaz-Pena R, Lopez-Vazquez A, Lopez-Larrea C. Old and new HLA associations with ankylosing 45 spondylitis. Tissue Antigens 2012;80:205-13. 46 17. Rudwaleit M, van der Heijde D, Landewe R, et al. The Assessment of SpondyloArthritis International 47 Society classification criteria for peripheral spondyloarthritis and for spondyloarthritis in general. Ann 48 49 Rheum Dis 2011;70:25-31. 50 18. Rudwaleit M, Landewe R, van der Heijde D, et al. The development of Assessment of 51 SpondyloArthritis international Society classification criteria for axial spondyloarthritis (part I): 52 classification of paper patients by expert opinion including uncertainty appraisal. Ann Rheum Dis 53 2009;68:770-6. 54 19. Baeten D, Breban M, Lories R, et al. Are spondylarthritides related but distinct conditions or a single 55 disease with a heterogeneous phenotype? Arthritis and rheumatism 2013;65:12-20. 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 15 1 2 3 20. Rudwaleit M, van der Heijde, D. Landewe, R., Listing J, Akkoc N, et al. The development of 4 Assessment of SpondyloArthritis international Society classification criteria for axial spondyloarthritis 5 6 (part II): validation and final selection. Ann Rheum Dis 2009;68:777-83. 7 21. Valle-Onate R, Candia L, Romero-Sanchez C, et al. Epidemiology of spondyloarthritis in Colombia. The 8 American journal of the medical sciences 2011;341:293-4. 9 22. Mielants H, Veys EM, Joos R, et al. HLA antigens in seronegative spondylarthropathies. Reactive 10 arthritis and arthritis in ankylosing spondylitis: relation to gut inflammation. J Rheumatol 1987;14:466- 11 71. 12 13 23. Siala M, Mahfoudh N, Fourati H, et al. MHC class I and class II genes in Tunisian patients with 14 reactive and undifferentiated arthritis. Clin Exp Rheumatol 2009;27:208-13. 15 24. Vargas-AlarconFor G, Londono peer JD, Hernandez-Pacheco review G, et al. Effect only of HLA-B and HLA-DR genes on 16 susceptibility to and severity of spondyloarthropathies in Mexican patients. Ann Rheum Dis 17 2002;61:714-7. 18 25. Reveille JD. An update on the contribution of the MHC to as susceptibility. Clinical rheumatology 19 2014;33:749-57. 20 21 26. Garrett S, Jenkinson T, Kennedy LG, et al. A new approach to defining disease status in ankylosing 22 spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. J Rheumatol 1994;21:2286-91. 23 27. Sieper J, Rudwaleit M, Baraliakos X, et al. The Assessment of SpondyloArthritis international Society 24 (ASAS) handbook: a guide to assess spondyloarthritis. Ann Rheum Dis 2009;68:ii1-44. 25 28. Calin A, Garrett S, Whitelock H, et al. A new approach to defining functional ability in ankylosing 26 spondylitis: the development of the Bath Ankylosing Spondylitis Functional Index. J Rheumatol 27 28 1994;21:2281-5. 29 29. Mander M, Simpson JM, McLellan A, et al. Studies with an enthesis index as a method of clinical 30 assessment in ankylosing spondylitis. Ann Rheum Dis 1987;46:197-202. 31 30. Excoffier L, Lischer HE. Arlequin suite ver 3.5: a new series of programs to perform population 32 genetics analyses under Linux and Windows. Mol Ecol Resour 2010;10:564-7. 33 31. Rousset F. GENEPOP'007: a complete re-implementation of the genepop software for Windows and 34 Linux. Mol Ecol Resour 2008;8:103-6. http://bmjopen.bmj.com/ 35 36 32. Bennett PH, Burch TA. The epidemiological diagnosis of ankylosing spondylitis. In: Bennett PH, Wood 37 PHN, eds. Population Studies of the Rheumatic Diseases. Amsterdam: Excerpta Medica, 1968:453-58. 38 33. Rudwaleit M, Jurik AG, Hermann KG, et al. Defining active sacroiliitis on magnetic resonance imaging 39 (MRI) for classification of axial spondyloarthritis: a consensual approach by the ASAS/OMERACT MRI 40 group. Ann Rheum Dis 2009;68:1520-7. 41 34. Burgos-Vargas R, Vazquez-Mellado J, Cassis N, et al. Genuine ankylosing spondylitis in children: a 42 case-control study of patients with early definite disease according to adult onset criteria. J Rheumatol on September 29, 2021 by guest. Protected copyright. 43 44 1996;23:2140-7. 45 35. Heuft-Dorenbosch L, van Tubergen A, Spoorenberg A, Landewe R, Dougados M, Mielants H, van der 46 Tempel H, van der Heijde D. The influence of peripheral arthritis on disease activity in ankylosing 47 spondylitis patients as measured with the Bath Ankylosing Spondylitis Disease Activity Index. Arthritis 48 and rheumatism 2004;51:154-9. 49 36. Roussou E, Sultana S. The Bath Ankylosing Spondylitis Activity and Function Indices (BASDAI and 50 51 BASFI) and their correlation with main symptoms experienced by patients with spondyloarthritis. Clinical 52 rheumatology 2010;29:869-74. 53 37. Reveille JD. The genetic basis of spondyloarthritis. Ann Rheum Dis 2011;70:i44-50. 54 38. Reveille JD, Ball EJ, Khan MA. HLA-B27 and genetic predisposing factors in spondyloarthropathies. 55 Curr Opin Rheumatol 2001;13:265-72. 56 39. Valle R, Londono J, Velez P, et al. Outcome of patients with undifferentiated seronegative 57 spondyloarthropathy. Arthritis and rheumatism 1997;40:270. 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 16 1 2 3 40. Londono J, González L, Ramírez A, et al. Caracterización de las espondiloartropatías y determinación 4 de factores de mal pronóstico en una población de pacientes colombianos. Revista Colombiana de 5 6 Reumatología 2005;12:195-207. 7 41. Sampaio-Barros PD, Bortoluzzo AB, Conde RA, et al. Undifferentiated spondyloarthritis: a longterm 8 followup. J Rheumatol 2010;37:1195-9. 9 42. Mielants H, Veys EM, De Vos M, et al. The evolution of spondyloarthropathies in relation to gut 10 histology. I. Clinical aspects. J Rheumatol 1995;22:2266-72. 11 43. Burgos-Vargas R, Clark P. Axial involvement in the seronegative enthesopathy and arthropathy 12 13 syndrome and its progression to ankylosing spondylitis. J Rheumatol 1989;16:192-7. 14 44. Martinez-Laso J, Herraiz MA, Vidart JA, et al. Polymorphism of the HLA-B*15 group of alleles is 15 generated followingFor 5 lineages peer of evolution. Humanreview immunology 2011;72:412-21. only 16 45. Stein JM, Machulla HK, Smeets R, et al. Human leukocyte antigen polymorphism in chronic and 17 aggressive periodontitis among Caucasians: a meta-analysis. Journal of clinical periodontology 18 2008;35:183-92. 19 46. Choukri F CA, Himmich H, Hue S, Caillat-Zucman S. HLA-B*51 and B*15 alleles confer predisposition 20 21 to Behcet's disease in Moroccan patients. Human immunology 2001;62:180-5. 22 47. Turnquist HR TH, Prilliman KR, Lutz CT, Hildebrand WH, Solheim JC. HLA-B polymorphism affects 23 interactions with multiple endoplasmic reticulum proteins. European journal of immunology 24 2000;30:3021-8. 25 48. McHugh K, Bowness P. The link between HLA-B27 and SpA--new ideas on an old problem. 26 Rheumatology (Oxford, England) 2012;51:1529-39. 27 28 49. Chen L, Fischer R, Peng Y, et al. Critical role of endoplasmic reticulum aminopeptidase 1 in 29 determining the length and sequence of peptides bound and presented by HLA-B27. Arthritis & 30 rheumatology (Hoboken, N.J.) 2014;66:284-94. 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 17 1 2 3 Figure Legends 4 5 6 7 Figure 1 Spondyloarthritis (SpA) Classification Criteria in relation to HLAB15 or HLAB27. The European 8 9 Spondyloarthropathy Study Group Classification Criteria for Spondyloarthropathy (ESSG) Criteria): 10 11 ankylosing spondylitis (AS), undifferentiated spondyloarthritis (uSpA), and reactive arthritis (ReA); The 12 13 Assessment of SpondyloArthritis international Society (ASAS) Criteria: axial spondyloarthritis (axSpA) and 14 15 peripheral spondyloarthritisFor (pSpA). peer review only 16 17 18 19 Figure 2 Spondyloarthritis (SpA) manifestations in patients positive for HLAB15 or HLAB27. (A) 20 21 Percentage of patients with joint abnormalities in relation to their genotype: sacroiliacjoint inflammation 22 23 grade ≥ 2 as assessed by xray of the pelvis (XRay SIJ), acute changes detected by nuclear magnetic 24 25 resonance imaging in sacroiliac joint (MRI SIJ), enthesitis in sacroiliac region assessed by Mander’s 26 27 enthesis index (Sacroiliac Enthesis). (B) Mean number of joints affected in each patient in relation to the 28 29 respective HLAB genotype: Swollen Joints, mean number of joints that are swollen upon assessment of 30 31 66 joints in each patient; Overall enthesitis: mean of total (axial and peripheral) entheses affected; 32 33 Peripheral enthesitis: mean of peripheral entheses affected.

34 http://bmjopen.bmj.com/ 35 36 37 Figure 3 Grading of disease activity in relation to HLAB15 or HLAB27. Bath Ankylosing Spondylitis 38 39 Disease Activity Index (BASDAI), Bath Ankylosing Spondylitis Functional Index (BASFI), Visual Analog 40 41 Scale of Patient’s global assessment of disease activity (VAS Patient), Visual Analog Scale of Physician’s 42 on September 29, 2021 by guest. Protected copyright. 43 global assessment of disease activity (VAS Physician, ), Visual Analog Scale of Inflammatory Back Pain 44 45 (VAS IBP). Each score ranges from 0 to 10; median and 25th and 75th percentiles are shown; *P < 46 47 0.0001, **P = 0.009 (Student’s ttest). 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 18 1 2 3 Table 1 Demographic and clinical stratification of SpA patients in relation to their HLAB genotype 4 5 6 HLAB15 HLAB27 Other HLAB Significant 7 Variable 8 (n = 34) (n = 70) (n = 74) P values 9 10 11 Age ± SD years, mean ± SD (n = 178) 35.7 ± 10.0 35.8 ± 13.6 35.8 ± 13.1 a 12 Age of onset years, mean ± SD (n = 178) 28.2 ± 9.4 27.7 ± 10.5 32.6 ± 10.9 0.008 13 14 Disease Duration years, median (IQR) 3.4 (2 – 12) 3.6 (1 – 12) 3.0 (1 – 11.5) 15 For peer review only b 16 Sex – number of men/women (%) 15/19 (44/56) 55/15 (79/21) 30/44 (40/60) 0.001 17 Men:Women ratio 0.8:1 3.7:1 1.46:1 18 19 20 SpA patients classified according to ESSG criteria 21 22 Patients having AS (n = 91) 13 41 37 23 c 24 Patients having uSpA (n = 57) 17 15 25 0.006 25 Patients having ReA (n = 30) 4 14 12 26 27 Total (n = 178) 34 70 74 28 29 30 SpA patients classified according to ASAS criteria 31 h 27 70 55 d 32 Patients with axSpA manifestations (n = 152) < 0.0001

33 0 10 2 e With isolated axSpA (n = 12) 0.050

34 http://bmjopen.bmj.com/ 27 60 53 35 With combined ax/pSpA (n = 140) 36 37 38 h f Patients with pSpA manifestations (n = 161) 34 60 67 0.050 39 4 0 9 g 40 With isolated pSpA (n = 13) 0.015 41 30 60 58 42 With combined ax/pSpA (n = 148) on September 29, 2021 by guest. Protected copyright. 43 44 45 46 Percentages within parenthesis are calculated with respect to the number of patients in each HLAB group. IQR, 47 48 interquartile range; SpA, spondyloarthritis (classified according to the ESSG criteria as AS, ankylosing spondylitis; 49 50 uSpA, undifferentiated spondyloarthritis and ReA, reactive arthritis; classified according to the ASAS criteria as axSpA, 51 axial spondyloarthritis and pSpA, peripherial spondyloarthritis). Shown significant P values correspond as follows: aP = 52 53 0.008, between HLAB27 and other HLAB; bP < 0.001, both between HLAB27 and HLAB15, and between HLAB27 54 55 c and other HLAB alleles; there was no significant difference between HLAB15 and other HLAB; P = 0.006, between 56 57 d e HLAB15 and HLAB27; P < 0.0001, between HLAB27 and other HLAB; P = 0.050, between HLAB15 and HLAB27; 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 19 1 2 3 fP = 0.050, between HLAB27 and HLAB15; gP = 0.015, between HLAB15 and HLAB27. (n = 152)h, among our 178 4 5 patients satisfying the ESSG classification criteria for SpA [2], only 152 met the ASAS criteria for axSpA [18] and 26 6 7 patients did not; (n = 161)h indicates that 161 patients (but not remaining 17 patients) met the ASAS criteria for pSpA 8 9 [17]. 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

HLA-B15 and preferential susceptibility to pSpA 20 1 2 3 Table 2 Association of HLAB15 and HLAB27 with SpA subsets 4 5 6 HLAB15 HLAB27 7 Disease 8 OR (95% CI) OR (95% CI) 9 10 11 SpA 14.9* (3.4 – 90.8) 9.1* (3.8 – 22.8) 12 13 SpA subsets according to ESSG criteria 14 15 AS For peer review13.2* (2.8 – 85.1) only15.7* (6.1 – 41.8) 16 uSpA 20.9* (4.4 – 134.9) 13.8 (1.4 – 11.2) 17 18 ReA 10.1 (1.7 – 77.3) 11.2* (2.7 – 35.3) 19 20 SpA subsets according to ASAS criteria 21 22 axSpA 8.2* (1.9 35.6) 9.8* (4.3 – 22.5) 23 24 pSpA 15.4* (3.6 – 65.4) 7.0* (3.1 – 16.1) 25 26 OR, odds ratio; 95% CI, 95% confidence intervals. *P < 0.05, patients versus controls by Chisquared 27 28 tests. SpA, spondyloarthritis, classified according to the ESSG European Spondyloarthropathy Study 29 30 Group criteria as AS, ankylosing spondylitis; uSpA, undifferentiated spondyloarthritis and ReA, reactive 31 32 arthritis. SpA classified according to the ASAS Assessment of SpondyloArthritis international Society 33

34 criteria as axSpA, axial spondyloarthritis and pSpA, peripherial spondyloarthritis. http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 190x142mm (300 x 300 DPI)

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 23 BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 190x142mm (300 x 300 DPI)

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 23 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2015-009092 on 11 November 2015. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 190x142mm (300 x 300 DPI)

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml