Genotyping in rheumatoid arthritis: a game changer in clinical management?

Abstract

Rheumatoid arthritis (RA) is a genetically complex disease of immune dysregulation characterized by painful inflammation of synovial joints. Despite advances in its management afforded by biologic drug development, efforts to improve outcomes for patients are confounded by the condition's heterogeneous pathobiology, and consequent variability in therapeutic responses. Great strides have been made in understanding the genetic epidemiology of rheumatoid arthritis since its association with the HLA locus was established in the 1980s, with over 100 additional disease-associated variants now confirmed through cumulative genome-wide association studies. Yet translation of this new knowledge for patient benefit – whether as a route to predicting disease risk, drug development or personalized medicine – has been slow. To address this, collaborating teams of interdisciplinary scientists will need to pool resources, including ever larger, well-characterized patient cohorts and sophisticated biostatistical approaches. Recent advances suggest that the fruits of these endeavors are beginning to come within reach.

Keywords:

• epidemiology
• genetics
• genotyping
• management
• rheumatoid arthritis

Genetic variation plays an important role in the etiology of rheumatoid arthritis (RA), contributing up to two-thirds of the risk of its development in Caucasians [1], and compounded by environmental risk factors such as smoking. Polymorphism at the HLA-DRB1, -DPB1 and -B loci [2], together with over 100 additional RA-associated loci confirmed through cumulative genome-wide association studies (GWASs) over 7 years [3–5], account for up to half of this ‘heritability.’ Yet, for all of the biological ‘leads’ that this explosion in new knowledge represents, its impact upon clinicians’ ability to improve the lives of patients with RA or, indeed, individuals at risk of developing the disease in the future, has yet to be felt. It is entirely understandable therefore that we should enquire after the investment made on our behalf into unraveling the genetic epidemiology of RA. When will the long-promised dividend be paid? Or is it just a case of ‘Jam tomorrow …but never jam today?’

Population screening?

An overarching objective of complex disease genetics is the ability to accurately predict individuals’ risk of developing chronic conditions within a population. It is hoped that targeted intervention might then delay or prevent disease onset, thus reducing both individual suffering and societal burden. Attempts to develop weighted genetic risk scores (GRSs) for RA may become informative under certain circumstances. For example, while prediction models incorporating particular HLA alleles and non-HLA variants yield modest discriminatory utility between RA patients and controls (AU ROC 0.72–0.79), this increases among males where smoking status is also considered (AUC 0.86) [6,7]. Nonetheless, GRSs remain insufficiently accurate for general population screening, and given the low prevalence of the disease (about 0.01), such an approach is unlikely ever to prove economical in practice. Recent paradigms for RA development acknowledge that a preclinical phase of the disease begins with measurable immune dysregulation many years before the onset of clinical symptoms. Future GRSs for RA risk prediction should build on this new understanding, being developed for application in subsets of individuals with higher a priori probabilities of disease development – for example, first-degree relatives of those already diagnosed with the condition and/or individuals without arthritis who are positive for anti-citrullinated peptide autoantibodies [8]. They might also focus on individuals with environmental risk factors, such as overweight smokers. Finally, given overlapping genetic architectures between autoimmune conditions, the utility of the GRS approach for predicting RA in patients with other conditions, such as thyroiditis, could be explored. Large-scale genetic studies are now needed in prospectively followed cohorts ‘at risk’ of RA development; if RA risk proves to be sufficiently predictable, such work will catalyze research into preventative interventions.

Drug discovery?

The development, and burgeoning use, of biologic drugs over the past two decades has seen a revolution in RA management. For example, agents that target proinflammatory pathways (anti-TNF, anti-IL-6R) and cellular interactions such as T-cell costimulation (CTLA4-Ig) are now established in the clinic [9]. The corresponding, albeit largely subsequent, discovery of RA susceptibility loci that implicate these successfully targeted disease mechanisms has been a remarkable feature of the GWAS era, with variants at TNFAIP, TRAF1, CFLAR, IL6R, TYK2, CD28 and CTLA4 providing obvious examples [3–5]. However, gratifying such corroborative data may be in terms of pathophysiological understanding, genetic studies have not, to date, helped to identify novel therapeutic targets. However, recent studies focusing on biologically relevant pathways rather than individual genetic loci have provided some encouraging insights [10,11]. In a groundbreaking investigation, Okada et al. integrated publicly available drug target–gene data with findings of a novel GWAS meta-analysis [4]. While highlighting established RA treatments, their network analysis also identified cyclin-dependent kinase inhibitor drugs, already used in cancer treatment, as putative repressors of synovial fibroblast proliferation. Preclinical studies reinforce their conclusions and appear to support their analytical approach [12].

Nonetheless, genotyping studies have in general been slow to specify causative polymorphisms and precise mechanisms of association are, in most cases, obscure. An understanding of how genetic variation influences the expression of genes and their protein products, the extent to which this is impacted by epigenetic and/or environmental factors, and the phenotypic consequences for human disease at a cellular level is, however, starting to become unraveled. In our view, this is an essential next step before the potential of genotyping analysis as a route to drug discovery can be fully exploited. Such studies are now being actively pursued.

Precision medicine?

The bewildering heterogeneity of RA confounds clinicians on many levels, not least when it comes to selecting an effective treatment from the ever-expanding therapeutic armamentarium. Clinical responses to the standard first-line agent, low-dose methotrexate, are highly variable, with remission being achieved in around 30%, but with the remainder subject to delays in disease control that may result in permanent joint damage and disability. The situation for biologic drugs is similar, and biomarkers that stratify the RA population into subgroups or ‘endotypes,’ each of which associates with responsiveness to a particular therapeutic strategy, are therefore keenly sought. Studies that address this at the level of genetic variation have proved challenging to perform, tending to suffer as a result of insufficient power, differing definitions of therapeutic efficacy, confounding factors or a combination thereof [13]. A few seemingly robust findings include association between the reduced folate carrier 1 variant and methotrexate efficacy [14] and between the 1q23 locus and etanercept efficacy (with the CD84 gene being functionally implicated) [15]. Moving forward concerted efforts must be made to collate large cohorts of RA patients, better characterized for multiple clinical covariates, in order that high-powered and optimally controlled GWASs might identify robust predictors of therapeutic efficacy. However, as with drug discovery, genetic markers alone may prove insufficient to characterize endotypes and most efforts in this field are being directed at combining SNP data with additional large datasets, including cell phenotype, transcriptional, proteomic, metabolomic and epigenetic analyses. The combining of such large and disparate datasets itself presents a significant and entirely new challenge.

Similarly, comprehensive studies should now also consider genotypic predictors of disease severity. Radiographic joint damage is an independent marker of functional decline in patients already diagnosed with RA, and the rate of destruction therefore serves as a useful objective measure of disease severity in this population [16]. Accordingly, the heritability of accelerated, ‘severe’ disease has been estimated at 45–58% [17], and the identity of contributory genetic variants is expected to shed light on mechanisms of tissue damage. For example, the effects – both detrimental and protective – of particular amino acid combinations encoded at the HLA-DRB1 locus in determining RA severity are increasingly well characterized [18], and variants at CD40, IL2RA and TRAF1 loci (already associated with disease susceptibility) are further implicated. Moreover, recent data suggest that the propensity of RANK/RANK-ligand and MMP3 pathways to mediate joint destruction in RA may be genetically determined, respectively, at the osteoprotegerin and sperm-associated antigen 16 (SPAG16) loci [19,20]. Hence, the large interventional studies required to test the hypothesis that the treatment of newly diagnosed RA might, in the future, also be rationally tailored to reflect these genotypic risk factors of disease severity remain some way off, but are now warranted.

Summary

For all the many valuable lessons we have learned from genotyping, it cannot yet be lauded as a game changer in the clinical management of RA, and the journey toward that end has so far proved convoluted. However, it is pertinent to consider whether part of the disappointment reflects inappropriate expectations raised by nonexperts. RA is a complex disease with non-Mendelian inheritance and we should not therefore have expected large, single gene effects on diagnosis, prognosis or therapeutic outcomes. Nonetheless, genetic analyses have undoubtedly shed bright lights on RA etiology and pathogenesis, and the combination of genetic data with various additional datasets is highly likely to influence future practice. Current trajectories that include rapid technological advances in ‘omic’ technologies, the accrual of large, well-characterized clinical cohorts by interdisciplinary teams of collaborating scientists and, critically, the development of the bioinformatics solutions necessary for interrogating the complex datasets that follow suggest that the wait may be coming to an end. RA patients and their doctors must continue to be patient. But we shall have jam, and soon.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.