Introduction
Rheumatoid arthritis (RA) [MIM 180300] is a phenotypically heterogeneous, chronic destructive inflammatory disease of synovial joints, with an estimated prevalence of 0.8% in the UK [
1]. A strong genetic component has been determined with heritability estimates of 50 to 60% from twin studies, with up to an additional 50% contribution from environmental and/or physiological risk factors [
2]. Approximately 40% of genetic susceptibility to RA is accounted for by the
HLA-DRB1 alleles encoding the shared epitope (SE), the major RA susceptibility locus [
3], together with the protein tyrosine phosphatase nonreceptor 22 gene (
PTPN22), a second susceptibility gene confirmed in populations of northern European ancestry [
4]. Recent genome-wide association studies and candidate gene studies in RA have been highly successful in both the confirmation of known genetic associations and in highlighting new loci/immunological pathways that warrant further investigation [
5]. The present study focuses on five confirmed RA susceptibility genes/loci -
HLA-DRB1 (6p21),
PTPN22 (1p13),
OLIG3/TNFAIP3 (6q23),
STAT4 (2q32) and
TRAF1/C5 (9q33) - that are associated with RA with low-to-moderate risk in UK patients [
6‐
8]. Some of these loci have been replicated in other Caucasian populations of northern European descent (reviewed in [
5,
9]), although only
HLA-DRB1 SE and
STAT4 have been confirmed in Asian populations [
10‐
12].
RA is characterised by the presence of autoantibodies (rheumatoid factor (RF) and cyclic citrullinated peptide (CCP) antibodies) in a significant majority of patients. Many of the RA susceptibility genes identified to date appear to only be significant in the autoantibody-positive cohorts, although this may be secondary to the increased statistical power in this more prevalent patient subgroup [
13]. If confirmed, this observation may suggest that these loci are influencing susceptibility to autoantibody production, perhaps through the loss of self-tolerance, thus explaining their association with multiple autoimmune disorders. The challenge over the next few years will be to identify whether these genes also influence the inflammatory process in RA
per se. Unravelling the stage in the disease process in which these genes exert their maximum influence on RA pathogenesis will be necessary to fully unveil their clinical significance and reveal those pathways that are potential therapeutic targets or may become clinically useful biomarkers.
In the present study we sought to identify which parts of the RA pathway were affected by these susceptibility genes by studying an RA inception cohort from the UK. The Yorkshire Early Arthritis Register (YEAR) Consortium has made a considerable effort to review all patients presenting with early inflammatory symptoms within the community through the establishment of a rapid access system. This enables the full spectrum of RA to be studied. Confirmation of association in this cohort of newly diagnosed RA would support a contribution of these loci to RA susceptibility per se rather than disease persistence and severity, which potentially confound assessment of genetic susceptibility in cross-sectional secondary care cohorts. Further analyses, albeit at reduced statistical power, would determine whether the association is observed within both the autoantibody-positive (RF or CCP) and the autoantibody-negative (RF or CCP) subgroups, to determine whether the primary association was with autoantibody-positive disease. These two antibodies are highly correlated and much larger cohorts would be required to tease out antibody-specific effects.
We then sought to examine the influence of these loci on disease severity. Analysis of RA severity following the initiation of therapy is fraught with many difficulties since the physician's choice of medication, dose prescribed, toxicities, co-morbidities and psychosocial factors may all influence treatment outcome and persistence of inflammation. We therefore took advantage of our rapid access referral system to assess the influence of these loci on objective markers of joint inflammation (swollen joint count (SJC)), function (Health Assessment Questionnaire (HAQ)) and a measure of articular damage (erosions on plain radiographs of the hands and feet) at RA diagnosis. Although these secondary severity analyses would be of lower statistical power and require replication, we propose that any positive results would reveal those pathophysiological pathways that warrant further investigation to see whether they can be exploited as potential therapeutic targets and prognostic and/or predictive biomarkers that may ultimately guide therapeutic decisions.
Discussion
In the present study we confirmed that the
HLA-DRB1 SE,
PTPN22 and the
OLIG3/TNFAIP3 and
TRAF1/C5 loci were associated with susceptibility to RA in an inception cohort. The effect sizes were comparable with the total UKRAG cohort (
HLA-DRB1 SE (2.1, 2.6),
PTPN22 (1.5, 1.5),
OLIG3/TNFAIP3 (1.2, 1.2) and
TRAF1/C5 loci (1.1, 1.1); early RA compared with total UKRAG cohort [
6‐
8,
14], respectively). The distinction between susceptibility and severity remains difficult and some may argue that the present study still does not sample the full spectrum of RA observed in the community. We believe, however, these findings do support a genuine association with RA susceptibility. Further studies utilising community-based cohorts will ultimately be required to confirm these findings. Of particular interest is the observation that for each of these loci the association was most marked in the autoantibody-positive subgroup and, although the association with
STAT4 was not confirmed, some evidence of a weak association was observed in the subgroup harbouring autoantibodies. Indeed, only
PTPN22 demonstrated any suggestion of association with RA in the autoantibody-negative cohort, which warrants further investigation in a larger inception cohort.
These findings support the mounting evidence that different genetic loci are associated with autoantibody-positive and autoantibody-negative RA. Many of the genes identified to date may therefore predispose to autoimmunity in general, or more specifically the immunological processes involved in the breakdown of self-tolerance and autoantibody production. This is supported by the association of
PTPN22 and the
OLIG3/TNFAIP3 locus with other autoantibody-associated autoimmune diseases, such as systemic lupus erythematosus, Graves disease and type 1 diabetes [
5,
7,
8,
15‐
17], but not with those autoimmune/inflammatory disorders not associated with autoantibody production, such as ulcerative colitis, Crohn's disease and ankylosing spondylitis [
5,
18,
19].
The transcription factor encoded by
STAT4 is downstream of several cytokines that play a crucial role in the development of Th1 and Th17 responses, such as IL-12, IL-15 and IL-23 [
20,
21] and the type I interferons [
21,
22]. Whilst this gene undoubtedly contributes to susceptibility to some autoantibody-associated diseases (RA, systemic lupus erythematosus and type I diabetes), there are recent reports that it may be associated with both clinical forms of inflammatory bowel diseases [
23] - suggesting that rather than contributing to autoantibody production, it may be a common risk factor for inflammatory disease
per se. This is consistent with the apparent association of
STAT4 with both autoantibody-positive and autoantibody-negative RA in the literature [
7,
11,
15,
23], although the numbers were considerably lower for the latter subgroup, in all reported series.
Although we acknowledge there was also reduced power to detect trends in the autoantibody-negative cohort in the current study, we were unable to see any suggestive evidence of association with STAT4, with no substantive skewing of allele or genotype frequencies. Additional studies with increased power to investigate autoantibody-negative RA will be needed to unravel the genes predisposing to RA in this patient subgroup. Power calculations based on replicating the association of PTPN22 with autoantibody-negative disease with an effect size in the range 1.2 to 1.5 revealed that 2,909 cases would be required if one control per case was used, reducing to 2,169 if two controls per case were used. If all five loci were investigated, the number of autoantibody-negative cases required would need to increase to 4,329 and 3,227, respectively - with ~12,000 cases required if the effect size was reduced to 1.1. Such studies will require a concerted international effort and large-scale recruitment of cases from around the globe.
In the current study, we analysed the SNPs displaying the strongest association with RA, after the strongly associated
HLA-DRB1 SE alleles, in the UK Caucasian population [
24]. To date, there have been relatively few genetic loci that have shown consistent association with disease severity in RA. As stated previously, it is very difficult to tease out the influence of drugs and other nongenetic factors when designing these studies. We therefore chose to investigate these susceptibility loci with objective markers of joint inflammation or disease severity (radiographic erosions, SJC, and HAQ) at presentation to the rheumatology department. The presence of radiographic articular erosions is generally accepted as the most objective measure of articular damage that is accessible for all rheumatologists, but there is currently a paucity of studies investigating this early time point.
In this exploratory study we were unable to find any strong evidence that these five susceptibility genes were associated with disease severity measures at baseline. We found some weak evidence to support the association of the
TRAF1/C5 locus with HAQ at baseline, a marker of function, although this would not remain significant after correction for multiple comparisons. Carriers of the minor A allele of rs10818488 at this locus were previously shown to have increased radiographic progression, albeit in a small cohort of 278 CCP-positive individuals [
25], whereas homozygosity for the G allele may be associated with mortality [
26]. We found no evidence, however, that
TRAF1/C5 was associated with erosions in this cohort. It is important to note that due to a high level of linkage disequilibrium between the genes encoding TNF receptor-associated factor 1 and complement component 5, it is currently not possible to unravel which of these two genes at 9q33.2 harbours the causal variant. Both are excellent candidate genes that may perpetuate chronic inflammation in RA. TNF receptor-associated factor 1 is a negative regulator of TNF receptor and Toll-like receptor signalling, and may contribute to the proliferation of T cells, and the complement pathway may directly contribute to immune complex-mediated inflammation [
25,
27]. Intensive resequencing efforts with additional fine-mapping studies in expanded genetic cohorts are currently underway to identify the disease-associated variant at such loci. It is likely that for those loci with very strong and extensive linkage disequilibrium, however, functional studies may ultimately be required to unravel the biological significance of the genetic findings.
Membership of the YEAR Consortium.
Management team: Paul Emery, Philip Conaghan, Ann Morgan, Anne-Maree Keenan, Elizabeth Hensor and Julie Kitcheman at the Section of Musculoskeletal Disease, Leeds Institute of Molecular Medicine, University of Leeds, UK. Mark Quinn at York District Hospital, York, UK.
Consultants: Gareth Huston, Mike Martin, Colin Pease and Sally Cox at the Section of Musculoskeletal Disease, Leeds Institute of Molecular Medicine, University of Leeds, UK. Michael Green and Amanda Isdale at York District Hospital, York, UK. Andrew Gough and Michael Green at Harrogate District Hospital, Harrogate, UK. Richard Reece at the Huddersfield Royal Infirmary, Huddersfield, UK. Lesley Hordon at Dewsbury District and General Hospital, Dewsbury, UK. Philip Helliwell and Richard Melsom at St Luke's Hospital, Bradford, UK. Sheelagh Doherty at Hull Royal Infirmary, Hull, UK. Ade Adebajo at Barnsley District General Hospital, Barnsley, UK. Andrew Harvey, Steve Jarrett and Zunaid Karim at Pinderfields General Hospital, Wakefield, UK. Dennis McGonagle at Calderdale Royal Hospital, Halifax, UK.
SpRs: Victoria Bejarano and Jackie Nam at the Section of Musculoskeletal Disease, Leeds Institute of Molecular Medicine, University of Leeds, UK.
Nurses: Claire Brown, Christine Thomas, David Pickles, Alison Hammond, Belinda Rhys-Evans, Barbara Padwell, Sally Smith and Heather King at the Section of Musculoskeletal Disease, Leeds Institute of Molecular Medicine, University of Leeds, UK. Anne Gill and Julie Green at York District Hospital, York, UK. Beverley Neville at Harrogate District Hospital, Harrogate, UK. Alan Fairclough and Caroline Nunns at Huddersfield Royal Infirmary, Huddersfield, UK. Jill Firth and Linda Sigsworth at St Luke's Hospital, Bradford, UK. Jayne Heard at Hull Royal Infirmary, Hull, UK. Julie Madden and Lynda Taylor at Calderdale Royal Hospital, Halifax, UK.
Laboratory staff: Diane Corscadden, Karen Henshaw, Lubna-Haroon Rashid, Stephen G Martin and James I Robinson at the Section of Musculoskeletal Disease, Leeds Institute of Molecular Medicine, University of Leeds, UK.
Membership of the UKRAG Consortium.
University of Manchester: Stephen Eyre, Anne Hinks, Laura J Gibbons, John Bowes, Edward Flynn, Paul Martin, Wendy Thomson, Anne Barton and Jane Worthington at the arc-Epidemiology Unit, The University of Manchester, UK.
University of Leeds: The YEAR consortium, Stephen Martin, James I Robinson, Ann W Morgan and Paul Emery at the Leeds Institute of Molecular Medicine, Section of Musculoskeletal Disease, University of Leeds, UK.
University of Sheffield: Anthony G Wilson at the School of Medicine & Biomedical Sciences, The University of Sheffield, UK.
University of London: Sophia Steer at the Clinical and Academic Rheumatology, Kings College Hospital NHS Foundation Trust, London, UK.
University of Aberdeen: Lynne Hocking and David M Reid at the Bone Research Group, Department of Medicine & Therapeutics, University of Aberdeen, UK.
University of Oxford: Pille Harrison and Paul Wordsworth at University of Oxford Institute of Musculoskeletal Sciences, Oxford, UK.
Competing interests
LS and HAE are employed by Roche Molecular Systems, Inc. (Pleasanton, CA, USA), provider of HLA-DRB1 and PTPN22 genotyping reagents for a subgroup of subjects analysed in the present study. Otherwise the authors declare that they have no competing interests.
Authors' contributions
AWM conceived and designed this study, contributed to the acquisition of data, undertook the statistical analyses, interpreted the data and wrote the manuscript. JIR contributed to the acquisition of data and drafted part of the manuscript. PGC contributed to the acquisition of data and critically reviewed the manuscript. SGM contributed to the acquisition of data and critically reviewed the manuscript. The YEAR Consortium contributed to the acquisition of data. EMAH contributed to the statistical analysis and critically reviewed the manuscript. MDM drafted part of the manuscript. LS developed the multi-locus SNP genotyping platform and critically reviewed the manuscript. HAE developed the multi-locus HLA-DRB1 genotyping platform and critically reviewed the manuscript. H-CG contributed to the acquisition of data and critically reviewed the manuscript. AB contributed to the acquisition of data and critically reviewed the manuscript. The UKRAG Consortium contributed to the acquisition of data and critically reviewed the manuscript. JW contributed to the acquisition of data and critically reviewed the manuscript. PE contributed to the acquisition of data and critically reviewed the manuscript.