Article Text

Extended report
New insight on the Xq28 association with systemic sclerosis
  1. F David Carmona1,
  2. M Carmen Cénit1,
  3. Lina-Marcela Diaz-Gallo1,
  4. Jasper C A Broen2,
  5. Carmen P Simeón3,
  6. Patricia E Carreira4,
  7. José-Luis Callejas-Rubio5,
  8. Vicente Fonollosa3,
  9. Francisco J López-Longo6,
  10. Miguel A González-Gay7,
  11. Nicolas Hunzelmann8,
  12. Gabriela Riemekasten9,
  13. Torsten Witte10,
  14. Alexander Kreuter11,
  15. Jörg H W Distler12,
  16. Rajan Madhok13,
  17. Paul Shiels13,
  18. Jacob M van Laar14,
  19. Annemie J Schuerwegh15,
  20. Madelon C Vonk16,
  21. Alexandre E Voskuyl17,
  22. Carmen Fonseca18,
  23. Christopher P Denton18,
  24. Ariane Herrick19,
  25. Jane Worthington19,
  26. Frank C Arnett20,
  27. Filemon K Tan20,
  28. Shervin Assassi20,
  29. Timothy R D J Radstake2,16,
  30. Maureen D Mayes20,
  31. Javier Martín1,
  32. Spanish Scleroderma Group
  1. 1Instituto de Parasitología y Biomedicina ‘López-Neyra’, CSIC, Granada, Spain
  2. 2Department of Rheumatology & Clinical Immunology, University Medical Center Utrecht, Utrecht, The Netherlands
  3. 3Department of Internal Medicine, Hospital Valle de Hebron, Barcelona, Spain
  4. 4Department of Rheumatology, Hospital 12 de Octubre, Madrid, Spain
  5. 5Department of Internal Medicine, Hospital Clínico Universitario San Cecilio, Granada, Spain
  6. 6Department of Rheumatology, Hospital General Universitario Gregorio Marañón, Madrid, Spain
  7. 7Department of Rheumatology, Hospital Universitario Marqués de Valdecilla, IFIMAV, Santander, Spain
  8. 8Department of Dermatology, University of Cologne, Cologne, Germany
  9. 9Department of Rheumatology and Clinical Immunology, Charité University Hospital, and German Rheumatism Research Centre, a Leibniz Institute, Berlin, Germany
  10. 10Hannover Medical School, Hannover, Germany
  11. 11Department of Dermatology, Venereology, and Allergology, Ruhr University of Bochum, Bochum, Germany
  12. 12Department of Internal Medicine, Institute for Clinical Immunology, University of Erlangen-Nuremberg, Erlangen, Germany
  13. 13Centre for Rheumatic Diseases, Glasgow Royal Infirmary, Glasgow, UK
  14. 14Institute of Cellular Medicine, Newcastle University, Newcastle, United Kingdom
  15. 15Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands
  16. 16Department of Rheumatology, Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands
  17. 17Department of Rheumatology, VU University Medical Center, Amsterdam, The Netherlands
  18. 18Centre for Rheumatology, Royal Free and University College Medical School, London, UK
  19. 19Arthritis Research UK Epidemiology Unit, The University of Manchester, Manchester Academic Health Science Centre, Manchester, UK
  20. 20The University of Texas Health Science Center–Houston, Houston, Texas, USA
  1. Correspondence to Dr F David Carmona, Instituto de Parasitología y Biomedicina López-Neyra. Consejo Superior de Investigaciones Científicas, Parque Tecnológico Ciencias de la Salud. Avenida del Conocimiento s/n, Armilla, Granada 18100, Spain; dcarmona{at}ipb.csic.es FDC, MCC and LMDG contributed equally. TRDJR, MDM and JM share senior authorship

Abstract

Objective To evaluate whether the systemic sclerosis (SSc)-associated IRAK1 non-synonymous single-nucleotide polymorphism rs1059702 is responsible for the Xq28 association with SSc or whether there are other independent signals in the nearby methyl-CpG-binding protein 2 gene (MECP2).

Methods We analysed a total of 3065 women with SSc and 2630 unaffected controls from five independent Caucasian cohorts. Four tag single-nucleotide polymorphisms of MECP2 (rs3027935, rs17435, rs5987201 and rs5945175) and the IRAK1 variant rs1059702 were genotyped using TaqMan predesigned assays. A meta-analysis including all cohorts was performed to test the overall effect of these Xq28 polymorphisms on SSc.

Results IRAK1 rs1059702 and MECP2 rs17435 were associated specifically with diffuse cutaneous SSc (PFDR=4.12×10−3, OR=1.27, 95% CI 1.09 to 1.47, and PFDR=5.26×10−4, OR=1.30, 95% CI 1.14 to 1.48, respectively), but conditional logistic regression analysis showed that the association of IRAK1 rs1059702 with this subtype was explained by that of MECP2 rs17435. On the other hand, IRAK1 rs1059702 was consistently associated with presence of pulmonary fibrosis (PF), because statistical significance was observed when comparing SSc patients PF+ versus controls (PFDR=0.039, OR=1.30, 95% CI 1.07 to 1.58) and SSc patients PF+ versus SSc patients PF− (p=0.025, OR=1.26, 95% CI 1.03 to 1.55).

Conclusions Our data clearly suggest the existence of two independent signals within the Xq28 region, one located in IRAK1 related to PF and another in MECP2 related to diffuse cutaneous SSc, indicating that both genes may have an impact on the clinical outcome of the disease.

  • Autoimmune Diseases
  • Gene Polymorphism
  • Systemic Sclerosis

Statistics from Altmetric.com

Introduction

Autoimmune diseases are complex polygenic conditions in which multiple susceptibility genes interact with epigenetic and environmental factors for their predisposition and progression. In some cases, part of the genetic component is shared among different immune disorders, suggesting that these pathologies may be influenced by disease-specific and common molecular pathways.1–4 For instance, most of the genetic associations described for systemic sclerosis (SSc), a fibrotic autoimmune disease of skin and internal organs, have also been reported to play a role in the susceptibility to systemic lupus erythematosus (SLE).5 ,6

Remarkably, a wide spectrum of autoimmune diseases shows a significant female preponderance. SSc represents a clear example of a sex biased immune disorder, with women reaching almost 90% of total affected individuals in some populations. Different factors have been proposed to explain this marked sexual dimorphism, including reproductive and sex hormones, fetal microchimerism, and gender differences in the immune system and lifestyle. Nevertheless, definitive evidences are still lacking and cumulative knowledge points to a major role of sex chromosomes in the immune system homeostasis. Due to the statistical complexity of testing for association between phenotype and genetic markers on the sex chromosomes, very few associations of sex-linked genes with the development of autoimmunity have been reported to date.7–9

One of the shared risk loci between SSc and SLE within the non-homologous region of the X chromosome is the interleukin-1 receptor-associated kinase 1 gene (IRAK1),10 ,11 that encodes a serine/threonine protein kinase with a special relevance in the signalling pathways of Toll-like receptors (TLRs)/IL-1R.12 ,13 However, IRAK1 is in the same haplotypic block as the methyl-CpG-binding protein 2 gene (MECP2) on Xq28, which is also associated with SLE,14 ,15 and functional genetic variants of the latter locus may explain the association signals with SLE observed in IRAK1.16

In this study we aimed to evaluate whether the SSc-associated IRAK1 polymorphism rs1059702 (Phe196Ser) described by Dieudé et al10 is the causal variant of the Xq28 association or whether it reflects another association signal from the nearby MECP2.

Patients and methods

Study population

Since the IRAK1/MECP2 genes are located in a sex-linked region, only women were included in the study. Informed written consent from all participants and approval from the local ethical committees were obtained in accordance with the tenets of the Declaration of Helsinki.

We analysed a total of 3065 female SSc patients and 2630 female unaffected controls of Caucasian ancestry, from an initial discovery cohort of Spain (1016 SSc and 1520 controls) and four additional replication cohorts from USA (965 SSc and 489 controls), Germany (490 SSc and 180 controls), The Netherlands (235 SSc and 278 controls) and UK (359 SSc and 163 controls). Since most samples have not been subjected to genome-wide association study platforms, population substructure analysis could not been performed and this may represent a potential limitation. In all cases, SSc patients were classified based on their skin involvement into limited cutaneous SSc (lcSSc) or diffuse cutaneous SSc (dcSSc) according to the criteria by Leroy et al17 A subset of the German cohort (43.73%) overlaps with that included in the study of IRAK1 by Dieudé et al10 To perform the subphenotype analyses, they were also phenotypically characterised accordingly with the presence or absence of anticentromere antibodies (ACA), antitopoisomerase antibodies (ATA) and pulmonary fibrosis (PF). ACA were determined by their characteristic distinctive pattern on HEP-2 cells, and ATAs were detected by passive immunodiffusion against calf thymus extract. PF was diagnosed by high resolution CT in all the European cohorts. However, the PF status of our US cohort was obtained by pulmonary function test (patients were considered to have PF if they showed a forced vital capacity (FVC) less than 70%). Since two different methodologies were used for the diagnosis, the meta-analyses on PF data were performed only in the European cohorts. The clinical characteristics of the analysed cohorts are summarised in table 1.

Table 1

Main clinical features of the female systemic sclerosis patients included in the study. Data are referred to the total of analysed individuals

Single-nucleotide polymorphism selection

We followed a single-nucleotide polymorphism (SNP) tagging strategy using Haploview V.4.2 (http://www.broad.mit.edu/mpg/haploview)18 to identify taggers that may cover all the common genetic variation within MECP2 (r2≥0.8) in the Utah residents with ancestry from northern and western Europe (CEU) population of the HapMap database (http://hapmap.ncbi.nlm.nih.gov/). Four MECP2 polymorphisms were selected with this method: rs3027935, rs17435, rs5987201 and rs5945175 (see online supplementary figure S1). Additionally, we also included in the study the non-synonymous IRAK1 genetic variant rs1059702 (Phe196Ser), which was described as the SNP that best explains the SSc susceptibility IRAK1 haplotype.10 The location of the five genetic variants analysed within the IRAK1/MECP2 region is shown in online supplementary figure S2.

Genotyping methods

DNA samples were obtained from peripheral white blood cells of participants following standard procedures. Predesigned TaqMan 5′ SNP genotyping assays were used to genotype the selected Xq28 genetic variants (assay IDs: rs3027935: C__15765567_10, rs17435: C___2597094_20, rs5987201: C__30089704_10, rs5945175: C__30485633_20, rs1059702: C___8966367_30) in a 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster City, California, USA).

Statistical analyses

The overall statistical power of our study, according to Power Calculator for Genetic Studies 2006,19 is shown in online supplementary table S1.

The Linux software Plink V.1.7 (http://pngu.mgh.harvard.edu/purcell/plink/)20 was used to perform 2×2 contingency tables, χ2 and/or Fisher's exact tests, when appropriate. p Values lower than 0.05 were considered as statistically significant. ORs, and 95% CI, were calculated according to Woolf's method. The Breslow-Day (BD) test was used to estimate the OR heterogeneity among the different cohorts. Combined data were analysed by Mantel-Haenszel tests under fixed effect, or random effect (DerSimonian–Laird) when the BD test reached statistical significance. Benjamini & Hochberg (1995) step-up false discovery rate (FDR) control correction21 for multiple testing was applied to the p values in the independent analysis and the combined meta-analysis.

Plink V.1.7 was also used to carry out a conditional logistic regression analysis to test for dependence between the associated SNPs. Haplotype analyses of allelic combinations with a frequency higher than 5% in the control groups were performed with Plink V.1.7, Haploview V.4.2 and StatsDirect (V.2.6.6; StatsDirect, Altrincham, UK).

Results

The genotyping success rate was higher than 95% in all analysed cohorts. No statistically significant deviation from Hardy-Weinberg equilibrium (p≤0.01) was observed for every SNP in the controls sets, except for rs3027935 in the Spanish population (see online supplementary table S2). The IRAK1 rs1059702 allele frequencies were similar to those described by Dieudé et al,10 whereas the frequencies of the four analysed MECP2 SNPs were in agreement with the data of the HapMap project.

Analysis of the discovery Spanish cohort

We first tested for association in a large Caucasian cohort of Spanish women (see online supplementary table S3). The statistical analysis of the allele frequencies indicated a significant association after FDR correction between IRAK1 rs1059702 and dcSSc (PFDR=0.028, OR=1.34, 95% CI 1.07 to 1.69). This genetic variant also showed trends towards association when the global SSc cohort and the PF+ patients were compared against the control set (PFDR=0.096, OR=1.16, 95% CI 0.99 to 1.35, and PFDR=0.058, OR=1.35, CI 95% 1.04 to 1.76, respectively). On the other hand, significant FDR-corrected p values were obtained for MECP2 rs17435 in the analyses of global disease (PFDR=0.014, OR=1.24, 95% CI 1.08 to 1.43), dcSSc (PFDR=0.014, OR=1.38, 95% CI 1.12 to 1.70), and PF (PFDR=0.047, OR=1.37, 95% CI 1.08 to 1.74). Other trends were suggested in the global SSc analysis for the MECP2 polymorphisms rs5987201 (PFDR=0.096, OR=1.32, 95% CI 0.97 to 1.79) and rs5945175 (PFDR=0.053, OR=1.50, 95% CI 1.06 to 2.11).

Mantel-Haenszel test

Considering the above described associations, we decided to include in the study three additional Caucasian cohorts from Europe (Germany, The Netherlands and UK) and one from the USA. The results of the independent analyses of these populations are summarised in the online supplementary tables S4–S7.

A meta-analysis including all five cohorts was then performed to increase the statistical power thus allowing a better estimation of disease susceptibility (table 2). The overall analysis yielded significant p values only for IRAK1 rs1059702 and MECP2 rs17435 (no heterogeneity of the ORs among the five populations was observed when applying the BD test in both cases). Regarding the IRAK1 non-synonymous SNP rs1059702, the allele frequencies of the dcSSc subgroup differed significantly from those of the control set in the combined analysis (PFDR=4.12×10−3, OR=1.27, 95% CI 1.09 to 1.47), and trends towards association were evident in the global SSc/control (PFDR=0.070, OR=1.13, 95% CI 1.01 to 1.26) and ATA+/control (PFDR=0.087, OR=1.23, 95% CI 1.04 to 1.46) comparisons, consistent with previously published data.10 Similarly, the MECP2 rs17435 variant reached statistical significant after comparing the global disease group and dcSSc subgroup with the control population (PFDR=2.68×10−3, OR=1.19, 95% CI 1.08 to 1.31, and PFDR=5.26 x10−4, OR=1.30, 95% CI 1.14 to 1.48, respectively). The ATA+/control comparison also suggested a trend for this SNP (PFDR=0.088, OR=1.18, 95% CI 1.01 to 1.38).

Table 2

Combined analysis of the allele frequencies of Xq28 genetic variants in Caucasian systemic sclerosis patients and unaffected controls from Spain, USA, Germany, The Netherlands and UK

To confirm that the genetic association between these two SNPs relies on dcSSc rather than on the global disease, we performed another meta-analysis but comparing this time dcSSc patients against those with lcSSc instead of unaffected controls, thus avoiding a possible effect of having SSc as a confounding variable (table 3). This new combined analysis showed statistically significant differences between dcSSc and lcSSc patients for the allele frequencies of IRAK1 rs1059702 (p=0.022, OR=1.19, 95% CI 1.02 to 1.37) and MECP2 rs17435 (p=0.026, OR=1.16, 95% CI 1.02 to 1.32).

Table 3

Combined analysis of the allele frequencies of Xq28 genetic variants comparing limited cutaneous systemic sclerosis patients against diffuse cutaneous systemic sclerosis patients from Spain, USA, Germany, The Netherlands and UK

Dependence analysis

Taking into account that both SNPs seemed to be specifically associated with risk to develop dcSSc, a conditional logistic regression analysis considering the populations as covariate was carried out to determine whether one of them could be dependent on the other one (table 4). The pairwise conditioning showed that the association of IRAK1 rs1059702 with dcSSc was explained by that of MECP2 rs17435, because only the latter SNP remained significant in the conditional analysis (rs1059702 conditioned p=0.786; rs17435 conditioned p=0.049). The fact that both polymorphisms were in moderate linkage disequilibrium (r2∼0.60) may explain the dependence.

Table 4

Conditional logistic regression analysis for IRAK1/MECP2 polymorphisms accordingly with diffuse cutaneous systemic sclerosis considering the populations as covariate

Analysis on PF data

Since the PF data of the SSc patients from the USA was estimated using a different methodology, we decided to include in the meta-analysis of this clinical feature only the European cohorts (Spain, Germany, The Netherlands and UK; table 5).

Table 5

Combined analysis of the allele frequencies of Xq28 genetic variants accordingly with pulmonary fibrosis in the study cohorts from Spain, Germany, The Netherlands and UK

In a first step, we compared PF+ SSc patients with the unaffected control group, obtaining significant p values for IRAK1 rs1059702 (PFDR=0.039, OR=1.30, 95% CI 1.07 to 1.58) and MECP2 rs17435 (PFDR=0.039, OR=1.24, 95% CI 1.04 to 1.48). However, only IRAK1 rs1059702 showed a consistent association with PF susceptibility, because the allele frequencies were significantly different between SSc patients with PF and SSc patients without PF (p=0.025, OR=1.26, 95% CI 1.03 to 1.55). On the contrary, in the case to case comparison, MECP2 rs17435 was not associated with PF (p=0.346, OR=1.09, 95% CI 0.91 to 1.31).

Haplotype analysis

Finally, we further explore a possible haplotype effect within this genomic region by performing several allelic combinations. Although the combinations rs1059702*G-rs17435*A and rs1059702*A-rs17435*T showed clear association peaks (see online supplementary table S8), the strength of these associations did not differ from those of IRAK1 rs1059702 and MECP2 rs17435 independently.

Discussion

This study represents an important step forward towards a better understanding of the complex genetic association between the Xq28 genomic region and the predisposition to autoimmunity. Because of the linkage disequilibrium structure of this locus, there is controversy about whether one of the two juxtaposed genes IRAK1 or MECP2 may harbour the causal variant of the described association with SLE and other autoimmune diseases including SSc.10 ,14–16 ,22 IRAK1 was considered as the most likely candidate, mainly due to its crucial involvement in the TLR pathway (in which it modulates the IL1-induced activation of NF-κB) and to the fact that this gene has been related with certain SLE manifestations in mouse models.11–13

On the other hand, MECP2 is also emerging as an interesting putative susceptibility gene in autoimmunity. It encodes a nuclear protein that binds specifically to methylated DNA, playing a central role in the transcriptional regulation of methylation-sensitive genes. Interestingly, there is increasing evidence pointing to epigenetic alterations, including those in DNA methylation, as relevant players in the pathogenesis of autoimmune diseases.23 For example, epigenetic dysregulation in T cells seems to be an important component of the SLE pathogenesis.24–26 Additionally, SLE patients containing the MECP2 SLE-risk haplotype show abnormal gene expression profiles of B cells, and this has been suggested to contribute to the interferon (IFN) signature observed in this pathology.15 Moreover, it has also been described in mouse models that MeCP2 protein may regulate the transcriptional activation of IRAK1 through microRNAs.27–29 Strikingly, the risk allele of the IRAK1 variant analysed here (rs1059702) has been recently associated with lower mRNA levels of MECP2 in SLE cases and unaffected controls of multiple ancestral groups, which has led to propose that both genes contribute independently to SLE susceptibility.30 Although no epistatic gene-gene interaction between MECP2 and IRAK1 was observed in our data, it could be possible that genetic variation within any of these two genes may be associated with the expression of the other one. In this regard, experimental or bioinformatic analysis on the association of MECP2 rs17435 and IRAK1 rs1059702 with expression levels of both genes would provide important clues that would help to completely unmask their inter-relation and its relevance in the genetic susceptibility to SSc and other related disorders.

Consistent with the latter idea, our data clearly indicate the existence of independent association signals with different SSc clinical conditions in each gene. One of them, rs17435, is located in the intron 2 of MECP2 and tags the SLE-risk haplotype.15 ,31 The same genetic variant has also been associated with Sjögren's syndrome (SS), one of the most frequent comorbidities in SSc patients.22 ,32 ,33 Specifically, we found a firm association of MECP2 rs17435 with the dcSSc subtype, which is the most severe form of the disease.9 It should be noted that SLE, SS and SSc are related autoimmune disorders that share some of their immunological and genetic component.1 ,2 ,34 ,35 For instance, a type I IFN signature similar to that described in SLE has also been identified in a subset of SSc patients.36 Taking into account the above, it is likely that MECP2 represents a real shared susceptibility factor for these systemic diseases probably affecting the epigenetic control of common pathogenic pathways (eg, the IFN signalling).

In addition, MECP2 could also be one of the genes contributing to the gender bias in autoimmune diseases. X chromosome inactivation in women involves processes that include DNA methylation.37 Considering that this chromosome contains a considerable number of important immune-related genes, such as inhibitor of κ light polypeptide gene enhancer in B cells kinase γ (IKBKG),38 interleukin 2 receptor gamma (IL2RG),39 CD40 ligand (CD40L),40 forkhead box P3 (FOXP3)41 or IRAK1,2 ,13 among others, some authors proposed that an abnormal gene dosage effect of X linked genes due to X chromosome defects may be involved in the female preponderance to develop autoimmunity.8 Supporting this idea, it has been recently reported that demethylation of regulatory elements of CD40L on the inactive X chromosome may be directly involved in the overexpression of this gene observed in CD4 T cells from SSc patients.42 ,43

On the other hand, we also found an independent association between the non-synonymous IRAK1 SNP rs1059702 and susceptibility to develop PF in SSc patients. This association is in agreement with that described by Dieudé et al10 However, these authors suggested that this locus is also associated with the dcSSc subtype and production of ATAs. In this regard, our data indicate that the association of IRAK1 rs1059702 with dcSSc is dependant of that for MECP2 rs17435, possibly due to the moderate linkage disequilibrium of both SNPs and the overlapping between this group of patients and those with pulmonary involvement (figure 1). The trend that we detected in the ATA analysis may also be a consequence of the latter cause. Since presence of ATAs and PF occur preferentially in dcSSc patients,9 it is not easy to determine specific associations with these three phenotypes if only cases versus controls are compared (having one of these features may represent a confounding variable). We observed statistically significant differences between SSc patients with PF and those without PF only for IRAK1 rs1059702 but not for MECP2 rs17435, thus suggesting that only IRAK1 rs1059702 is associated with PF predisposition.

Figure 1

Degree of overlapping between different case subsets from Spain, Germany, The Netherlands and UK. SSc, systemic sclerosis; lcSSc, limited cutaneous systemic sclerosis; dcSSc, diffuse cutaneous systemic sclerosis; PF, pulmonary fibrosis. Note: ‘lcSSc with PF’ and ‘dcSSc with PF’ percentages are referred to the total number of lcSSc and dcSSc cohorts, respectively, with PF data.

Despite MECP2 and IRAK1 showing clear association signals with SSc, it should be noted that these associations are less robust than those described in SLE.11 ,14 It could be speculated that the risk variants in this genomic region may lead to a SLE-like phenotype, as it has been observed for other common susceptibility genes like ITGAM.44

In conclusion, the analysis of our data showed that MECP2 and IRAK1 are independent SSc susceptibility genes that influence distinct clinical phenotypes. MECP2 is associated with the most severe subtype of the disease, consistent with the major role that epigenetics seem to play in this kind of diseases, and IRAK1 may be a crucial gene underlying the pathogenic mechanisms leading to PF, probably by altering the TLR pathway.

Acknowledgments

The authors thank Sofía Vargas, Sonia García and Gema Robledo (from Instituto de Parasitología y Biomedicina ‘López-Neyra’, CSIC, Spain), Julio Charles and Marilyn Perry (University of Texas) for their excellent technical assistance, and all the patients and unaffected controls for kindly accepting their essential collaboration. Banco Nacional de ADN (University of Salamanca, Spain) is thanked for supplying the Spanish controls.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Handling editor Tore K Kvien

  • Collaborators See online supplementary note.

  • Contributors FDC, MCC, LMDG, TRDJR, MDM and JM were involved in the conception and design of the study and revised critically the submitted version of the manuscript. They also contributed in the analysis and interpretation of data. FDC drafted the manuscript. JCAB, CPS, PEC, JLCR, VFP, FJLL, MAGG, NH, GR, TW, AK, JHWD, RM, PS, JMvL, AJS, MCV, AEV, CF, CPD, AH, JW, FCA, FKT, SA, TRDJR and MDM, participated in the collection of samples and clinical information, in the analysis and interpretation of data, and revised critically the manuscript draft. All authors approved the final version of the manuscript.

  • Funding This work was supported by the following grants: JM was funded by GEN-FER from the Spanish Society of Rheumatology, SAF2009–11110 from the Spanish Ministry of Science, CTS-4977, and CTS-180 from Junta de Andalucía, RETICS Program, RD08/0075 (RIER) from Instituto de Salud Carlos III (ISCIII), Spain, within the VI PN de I+D+i 2008–2011 (FEDER), and is sponsored by the Orphan Disease Program grant from the European League Against Rheumatism (EULAR). This study was also funded by PI-0590-2010, from Consejería de Salud y Bienestar Social, Junta de Andalucía, Spain. JLCR and JM are funded by Consejería de Salud, Junta de Andalucía, through PI-0590–2010. FDC was supported by Consejo Superior de Investigaciones Científicas (CSIC) through the program JAE-DOC. TRDJR was funded by the VIDI laureate from the Dutch Association of Research (NWO) and Dutch Arthritis Foundation (National Reumafonds). TW was granted by DFG WI 1031/6.1. Study on USA samples were supported by US National Institutes of Health and National Institute of Arthritis and Musculoskeletal Diseases (NIH-NIAMS) R01-AR-055258, Two-Stage Genome Wide Association Study in Systemic Sclerosis (MDM) and by the NIH-NIAMS Center of Research Translation (CORT) in SSc (P50AR054144) (MDM, FCA, FKT), the NIH-NIAMS SSc Family Registry and DNA Repository (N01-AR-0–2251) (MDM), NIH-KL2RR024149 (SA), K23AR061436 (SA), and the Department of Defence Congressionally Directed Medical Research Programs (W81XWH-07-01-0111) (MDM).

  • Competing interests None.

  • Ethics approval Approval from the local ethical committees of all participating centres were obtained in accordance with the tenets of the Declaration of Helsinki.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.