The genetics of rheumatoid arthritis: risk and protection in different stages of the evolution of RA

Abstract

There is now a general consensus that RA has a spectrum of disease stages that can begin many years before the onset of clinical symptoms. It is widely thought that understanding the complex interplay between genetics and environment, and their role in pathogenesis, is essential in gaining further insight into the mechanisms that drive disease development and progression. More than 100 genetic susceptibility loci have now been identified for RA through studies that have focused on patients with established RA compared with healthy controls. Studying the early preclinical phases of disease will provide valuable insights into the biological events that precede disease and could potentially identify biomarkers to predict disease onset and future therapeutic targets. In this review we will cover recent advances in the knowledge of genetic and environmental risk factors and speculate on how these factors may influence the transition from one stage of disease to another.

Introduction

The heritability of RA has been shown from twin studies to be ∼60%. Since 2007, rapid advances in technology underpinning the use of genome-wide association studies (GWASs) have allowed the identification of hundreds of genetic risk factors for many complex diseases. There are now >100 genetic loci that have been associated with RA (Table 1). The heritability of RA suggests that a large proportion of the disease could be the result of contributing environmental risk factors. It is widely thought that understanding the complex interplay between genetics and environment, and their roles in pathogenesis, is essential to gain further insight into the mechanisms that drive disease development and progression.

Table 1.

Genetic loci associated with susceptibility to RA

Chromosome	SNP	Gene	Reference: association with ACPA-positive RA (f)a	Ethnicity	Reference: association with ACPA-negative RA	Reference: association with RA severity	Reference: association with IP (e)	Reference: association in individuals at risk of RA [(a) and (b)]
1	rs2228145	IL6R	[1]	Caucasians
1	rs2240336	PADI4	[1]	Caucasians
1	rs883220	POU3F1	[1]	Caucasians
1	rs798000	CD2	[2]	Caucasians
1	rs2843401	MMEL1	[2]	Caucasians				[3]
1	rs2014863	PTPRC	[2]	Caucasians
1	rs2476601	PTPN22	[2]	Caucasians	[4]	[5, 6]	[7]
1	rs10494360	FCGR2A	[2]	Caucasians
1	rs2105325	loc100506023	[8]	Caucasians
1	rs28411352	MTF1-INPP5B	[8]	Caucasians
1	rs227163	TNFRSF9	[8]	Asian
1	rs10430455	FCRL3	[9]	Asians
1	rs6682654	CD244	[10]	Asians
2	rs10209110	AFF3	[2]	Caucasians		[5, 6]
2	rs34695944	REL	[2]	Caucasians
2	rs6546146	SPRED2	[2]	Caucasians
2	rs13426947	STAT4	[2]	Caucasians	[4]
2	rs1980422	CD28	[2]	Caucasians
2	rs11571302	CTLA4	[2]	Caucasians
2	rs11900673	B3GNT2	[11]	Asians
2	rs6732565	ACOXL	[8]	Caucasians
2	rs6715284	CFLAR-CASP8	[8]	Caucasians
2	rs10175798	LBH	[8]	Caucasians
3	rs35677470	DNASE1L3	[2]	Caucasians
3	rs3806624	EOMES	[8]	Caucasians
3	rs9826828	IL20RB	[8]	Caucasians
3	rs4452313	PLCL2	[8]	Caucasians
4	rs78560100	IL2, IL21	[2]	Caucasians
4	rs932036	RBPJ	[2]	Caucasians
4	rs2867461	ANXA3	[11]	Asians
4	rs13142500	CLNK	[8]	Caucasians
4	rs2664035	TEC	[8]	Caucasians
5	rs71624119	ANKRD55	[2]	Caucasians	[4]
5	rs39984	GIN1	[2]	Caucasians	[4]
5	rs657075	CSF2	[11]	Asians
6	rs59466457	CCR6	[2]	Caucasians
6	rs6911690	PRDM1	[2]	Caucasians
6	rs629326	TAGAP	[2]	Caucasians		[5, 6]
6	rs6920220	TNFAIP3	[2]	Caucasians	[4]
6	rs72928038	BACH2	[12]	Caucasians
6	rs12529514	CD83	[11]	Asians
6	rs2233434	NFKBIE	[11]	Asians
6	rs2234067	ETV7	[8]	Caucasians
6	rs9378815	IRF4	[8]	Caucasians
6	rs9373594	PPIL4	[8]	Asian
6	amino acid position 9	HLA-DPB1	[13]	Caucasians
6	amino acid position 9	HLA-B	[13]	Caucasians	[14, 15]
6	amino acid position 74	HLA-DRB1	[13]	Caucasians
6	amino acid position 71	HLA-DRB1	[13]	Caucasians
6	amino acid position 11	HLA-DRB1	[13]	Caucasians	[14, 15]
6	amino acid position 77	HLA-A	[15]	Caucasians
7	rs3807306	IRF5	[2]	Caucasians
7	rs4272	CDK6	[8]	Caucasians
7	rs67250450	JAZF1	[8]	Caucasians
8	rs4840565	BLK	[2]	Caucasians	[4]
8	rs678347	GRHL2	[8]	Caucasians
8	rs1516971	PVT1	[8]	Caucasians
8	rs998731	TPD52	[8]	Caucasians
9	rs10739580	TRAF1	[2]	Caucasians		[5, 6]		[3]
9	rs2812378	CCL21	[2]	Caucasians
10	rs12764378	ARID5B	[1]	Caucasians
10	rs2275806	GATA3	[2]	Caucasians
10	rs947474	PRKCQ	[2]	Caucasians
10	rs10795791	IL2RA	[2]	Caucasians
10	rs12413578	10p14	[8]	Caucasians
10	rs726288	SFTPD	[8]	Asian
10	rs2671692	WDFY4	[8]	Caucasians
10	rs793108	ZNF438	[8]	Caucasians
11	rs595158	CD5	[1]	Caucasians
11	rs4938573	DDX6	[1]	Caucasians
11	rs570676	TRAF6	[2]	Caucasians
11	rs3781913	PDE2A-ARAP1	[11]	Asians
11	chr11:107967350	ATM	[8]	Caucasians
11	rs4409785	CEP57	[8]	Caucasians
11	rs73013527	ETS1	[8]	Caucasians
11	rs968567	FADS1-FADS2-FADS3	[8]	Caucasians
12	rs10683701	KIF5A	[2]	Caucasians		[5, 6]
12	rs773125	CDK2	[8]	Caucasians
12	rs10774624	SH2B3-PTPN11	[8]	Caucasians
13	rs9603616	COG6	[8]	Caucasians
14	rs911263, rs1950897	RAD51L1/RAD51B	[12]	Caucasians
14	rs2841277	PLD4	[11]	Asians
14	rs3783782	PRKCH	[8]	Asian
15	rs8043085	RASGRP1	[1]	Caucasians
15	rs8026898	TLE3	[1]	Caucasians
16	rs13330176	IRF8	[1]	Caucasians
16	rs4780401	TXNDC11	[8]	Caucasians
17	rs12936409	IKZF3	[1]	Caucasians
17	rs72634030	C1QBP	[8]	Caucasians
17	rs1877030	MED1	[8]	Caucasians
18	rs7234029	PTPN2	[16]	Caucasians
18	rs2469434	CD226	[8]	Asian
19	rs34536443	TYK2	[1]	Caucasians
19	chr19:10771941	ILF3	[8]	Caucasians
20	rs6032662	CD40	[2]	Caucasians
21	rs2834512	RCAN1	[1]	Caucasians
21	rs9979383	RUNX1	[1]	Caucasians
21	rs73194058	IFNGR2	[8]	Caucasians
21	rs1893592	UBASH3A	[8]	Caucasians
21	rs2075876	AIRE	[17]	Asians
22	rs3218251	IL2RB	[2]	Caucasians
22	rs909685	SYNGR1	[8]	Caucasians
22	rs1043099	GATSL3	[18]	Caucasians
22	rs5754217	UBE2L3	[19]	Caucasians
X	rs13397	IRAK1	[1]	Caucasians
X	chrX:78464616	P2RY10	[8]	Asian

The table shows genetic loci associated with susceptibility to RA. References are shown for the association of loci with ACPA-positive RA. If loci have also been associated with ACPA-negative RA, RA severity, IP or individuals at risk of disease, the references for these studies are also shown. Ethnicity describes the population in which the association was observed. ^aEULAR disease phases are shown in parentheses. SNP: single nucleotide polymorphism; IP: inflammatory polyarthritis; EULAR: European League Against Rheumatism.

There is now a general consensus that RA has a spectrum of disease stages that can begin many years before the onset of clinical symptoms. Data showing the presence of autoantibodies and indicators of activation of the immune system years before disease onset indicate the presence of a long preclinical phase of disease potentially influenced by environmental factors [20, 21]. This preclinical phase results in a continuum that eventually crosses a threshold leading to the manifestation of clinical symptoms and ultimately joint damage. It is hypothesized that genetic markers associated with disease, in combination with stochastic environmental risk factors, influence the transition from one disease stage to another.

This concept is relatively new, however, and as a result the majority of genetic and environmental studies to date have focused on patients with established RA, according to the 1987 ACR criteria for RA [22], compared with healthy controls, and few studies have used cases in the early stages of disease or even in the preclinical phases of disease onset. In more recent years, research has focused on the earlier stages of disease [20, 21, 23, 24]. This has been helped by the ACR/European League Against Rheumatism (EULAR) classification criteria published in 2010, which include earlier features of the disease, such as autoantibody status [25], although it is only recently that researchers have begun establishing cohorts that will allow investigation of the earliest stages of disease. An initial challenge is definition of the disease phases; thus the EULAR Study Group for Risk Factors for RA recently published recommendations to classify individuals into different stages of disease for recruitment into prospective studies [26].

Prior to the development of the EULAR recommendations, RA patients have typically been split into two groups based on autoantibody status [26]. Evidence from studies of ACPA-positive and negative disease subsets suggest that they are two distinct diseases with different underlying pathogenesis; this is discussed in more detail later. Studying patients who may be at risk of disease due to genetic and environmental risk factors gives researchers the chance to identify significant biological changes that may occur in the preclinical phase of RA, potentially providing invaluable opportunities for new treatment strategies.

ACPAs are of particular interest, as these autoantibodies are highly specific for RA and can be found in ∼50% of early RA patients [27], making ACPA an important early clinical biomarker. In addition, the presence of anti-CCP antibodies has been shown to predict disease severity and radiological damage [28, 29], meaning that ACPA could also be used as a biomarker for patients with a more severe disease phenotype and to select those patients eligible for more aggressive treatment [30].

It is not clear how individuals progress from a pre-RA ACPA-positive state with no symptoms to clinical RA. Not all individuals with ACPA develop RA; therefore other triggers may be required to make the transition.

It has also been shown that the number of citrullinated antigens identified by ACPA increases exponentially leading up to disease onset. This is known as epitope spreading [31, 32]. The expansion of the ACPA repertoire that occurs before the manifestation of clinical disease could potentially be used to predict the impending onset of disease. Although no specific single antigen or target has been identified, it may be that certain antigens contribute to an initial break in immune tolerance, resulting in epitope spreading and an immune response to other antigens. This could then result in the transition from preclinical to clinically apparent disease. In this review we will discuss the recent advances in knowledge of genetic susceptibility to RA and the study of environmental risk factors and will speculate on how these factors may influence the transition from one stage of disease to another. Reference made to EULAR disease phase refers to the EULAR disease classification defined in Gerlag et al. [26].

Genetics of seropositive RA [EULAR disease phase (f)]

At least two-thirds of the risk of RA is thought to be conferred by genetic risk factors [33]. The largest genetic risk factor for RA lies within the human leucocyte antigen (HLA) class II region and encodes the HLA-DRB1 molecule [34, 35]. The specific alleles of HLA-DRB1 that have been associated with RA encode a conserved amino acid sequence that lies in the antigen binding groove of the antigen presenting molecule. This conserved sequence is referred to as the shared epitope (SE) [36]. It is thought that the presence of different alleles in the SE influence the interaction between the class II molecule and the T cell receptor (TCR) or antigen, and could therefore directly affect the efficiency of antigen presentation [36].

A more recent analysis of the HLA region in RA carried out by Raychaudhuri et al. [13], showed that three amino acid positions (11, 71 and 74) in HLA-DRB1 and a single amino acid position (position 9) in HLA-B and HLA-DPB1, account for the association of the MHC with anti-CCP positive RA. A key finding of this study was the association of amino acids 11 and 13, both in tight linkage disequilibrium, which lie outside the classical SE region of DRB1 but within the peptide binding grooves. Recent evidence has suggested that the HLA class II locus is not associated with risk of becoming ACPA+, but with the risk of progressing from ACPA+ to ACPA+ RA [37, 38]. This supports our hypothesis that the HLA class II locus is not directly involved in the formation of ACPA responses, but rather in its maturation, possibly via T cells providing help to ACPA-producing B-cells before the precipitation of disease

The largest genetic association with RA outside the HLA region lies within the protein tyrosine phosphatase non-receptor 22 (PTPN22) gene [39–44]. With an odds ratio (OR) of ∼1.6, PTPN22 encodes lymphoid tyrosine phosphatase which is a negative regulator of signal transduction from the TCR [45]. Mutations in the PTPN22 gene are associated with multiple autoimmune diseases and have been shown to confer opposite effects [40, 41, 43–53]. Several studies investigating the function of this variant have had conflicting results with some suggesting it results in a gain of function [54–58]. and others suggesting that the R620W polymorphism actually leads to a loss of function [58, 59].

Since the identification of PTPN22, fine mapping of linkage peaks [60], GWAS [11, 18, 61–66] and meta analyses [1, 2] have led to the identification of many more genetic risk factors for RA. There is an emerging picture of shared genetic overlap between autoimmune diseases [67, 68]. A recent large scale fine mapping study called the Immunochip project leveraged this remarkable genetic overlap to its advantage by designing a custom Illumina single nucleotide polymorphism (SNP) genotyping array containing ∼200 000 SNPs at 186 loci previously implicated in 12 autoimmune diseases, providing the opportunity for fine mapping of autoimmune disease loci [1]. For RA, the Immunochip study was carried out in 11 475 cases of European descent and 15 870 controls. These data were then combined with previous independent GWAS data (additional 2363 cases, and 17 872 controls) in a meta-analysis. 14 new susceptibility loci were identified; additionally the fine mapping approach allowed the peak of association to be refined to a single gene for 19 loci [1]. A later follow up study of loci reaching suggestive significance in the Immunochip study identified a further two novel loci in independent datasets [12].

The most recent large genetic studies have increased the number of RA susceptibility loci to over 100 across multiple populations (Table 1) [8, 69]. Plenge et al. performed a large scale trans-ethnic GWAS Meta analysis of >100 000 individuals of European and Asian ancestry and assessed over 10 million SNPs. Outside of the MHC 100 loci were shown to explain 5.5% and 4.7% of the heritability of RA in Europeans and Asians respectively.

The majority of the genetic loci identified to date confer small effect sizes and therefore a substantial proportion of the heritability of RA remains undetected. The missing heritability could lie in epigenetic changes (heritable changes in gene activity not related to changes in the DNA sequence), gene-gene or gene-environment interactions, or even in rare variants although recent studies have suggested the contribution of rare variants to complex disease may be negligible [70]. As well as risk variants for RA, it is likely that there may also be protective variants and although some of the identified RA genetic loci confer ORs below one, the protective effect of genetic variants for RA is poorly understood with the exception of specific HLA alleles. A meta-analysis investigating HLA-DRB1 alleles showed HLA-DRB1*13 to be associated with protection in ACPA-positive RA [71].

Genetic susceptibility to seronegative RA

Genetic studies of RA have largely been carried out in anti CCP positive patients, and many of the RA loci identified to date have stronger effects in this subgroup of patients [1]. Although ACPA negative RA patients make up ∼30% of the RA population this subset of patients remains understudied. Many people consider anti-CCP positive and anti-CCP negative RA to be two distinct diseases, however this debate is ongoing [72–74]. The heritability of the two subgroups has been estimated to be similar at 68% and 66% respectively although the contribution of the SE to this heritability was found to be significantly lower in anti-CCP negative disease (18% in anti-CCP positive, 2.4% in anti-CCP negative) [75]

Two recent studies of ACPA negative patients and healthy controls identified a peak of association in the HLA region which was determined to be accounted for by two amino acid positions HLA-DRB1 position 11 and HLA-B position 9 [14, 15]. These two positions have also been shown to be associated with ACPA positive RA; however the specific amino acid residues at each position which confer risk were different between the two disease subsets. Interestingly the presence of a serine residue at HLA-DRB1 position 11 conferred risk of ACPA negative disease but was protective against ACPA positive disease [15]. The study by Raychaudhuri et al. also identified a new association at position 77 in HLA-A in ACPA positive patients [15]. These studies increase the evidence that ACPA positive and ACPA negative RA are two distinct diseases.

It is now well known that other ACPA antibodies and fine specificities may contribute to RA, for example Lundberg et al. [76] tested four fine specificities of ACPA in RA patients, they observed different associations between genetic and environmental risk factors with RA depending on the ACPA specificity, with the HLA-DRB1 SE, PTPN22 and smoking showing the strongest association with the RA subset defined by the presence of antibodies to citrullinated α-enolase. Therefore it must be recognised that testing for the presence of anti-CCP antibodies in order to define the ACPA status of a patient does not definitively define an individual as ACPA negative and may result in heterogeneity in so called anti-CCP negative patient cohorts.

Recently it has been identified that breaking tolerance to post-translationally modified proteins in arthritis is not exclusively confined to citrullination. ACPA recognize proteins only after the enzymatic conversion of the amino acid arginine by PAD-enzymes to become the amino acid citrulline. It is likely that proteins that have undergone a different type of posttranslational modification are also recognized by auto-antibodies. One of these other post-translational processes is carbamylation, where the amino acid lysine is changed to become homocitrulline. Smoking can enhance carbamylation [77] and extensive carbamylation is especially thought to occur during (chronic) inflammation [77]. As smoking and chronic inflammation are important in the context of RA it is possible that carbamylation could be taking place in the inflamed synovium.

The post-translationally modified amino acids citrulline and homocitrulline are very similar in structure. The resemblance between the two modifications and the likely presence of carbamylated proteins (CarPs) in the joint was the motivation to test for the presence of antibodies directed against CarPs. Using, a novel assay that specifically detects the presence of antibodies directed against CarPs (anti-CarPs) [78], 43% of RA patients were found to have antibodies directed against CarPs. Importantly these anti-CarP antibodies were not only present in ACPA positive but also in ACPA negative RA patients, suggesting that antibodies recognizing one modification do not necessarily cross-react with the other modification [79]. Anti-CarP antibodies were also associated with a higher rate of joint damage however, further studies are required to determine the contribution of these antibodies to disease. This finding highlights the possibility that seronegative RA does not exist rather that we have as yet failed to identify all possible RA associated autoantibodies. It is not known why some people develop these antibodies and how they may contribute to disease. The question remains if these antibodies could be used to predict progression towards clinical disease in individuals at risk, or if patients should be tested for anti-CarP alongside anti-CCP. As indicated above international replication studies are needed to confirm and expand observations discussed.

Several non HLA RA susceptibility genes identified in anti-CCP positive disease have also been shown to be associated with anti-CCP negative disease (TNFAIP3, GIN1/C5orf30, STAT4, ANKRD55/IL6ST, BLK and PTPN22) [4]. The high heritability of anti-CCP negative RA and the low number of associated loci highlights the need for further large well powered GWAS studies to identify novel seronegative disease specific loci. Further investigation into the overlapping associations will be required in these two subgroups of disease.

Genetics of inflammatory polyarthritis [EULAR disease phase (e)]

It is difficult to speculate on the impact of a locus on transition between disease stages as the mechanism of the majority of RA susceptibility loci has not yet been investigated. Studying early inflammatory polyarthritis (IP) or patients with undifferentiated arthritis (UA) may help determine the genetic factors that are involved in the pathogenesis and prognosis of IP and UA. Some IP and UA patients will ultimately progress to RA or even PsA, however, some individuals will remain undifferentiated indefinitely. Several disease susceptibility loci have been associated with disease severity (TRAF1/C5, KIF5A, PTPN22, AFF3, TAGAP) [5, 6], suggesting that these loci may represent a tipping point allowing disease to progress one stage further and that individuals without these loci may maintain mild disease.

There have been a limited number of studies investigating the genetics of IP patients [80–84]. A study of 680 patients with IP from the Norfolk Arthritis Register (NOAR) and 286 controls showed a modest association between the SE alleles of HLA-DRB1 and IP [85]. The OR (1.8) for the association of the SE alleles with IP was much smaller than that typically observed with RA (OR 2–3).

The modest association of a SNP in the PTPN22 locus has been shown with UA [86]. A more recent study of this locus showed significant association of the PTPN22*1858T allele with IP and the strength of the effect was similar to that observed in RA [7]. A study in a Dutch population has shown that the PTPN22*1858T allele could not predict progression from UA to RA [87]. The new EULAR recommendations should allow further study of this group of patients.

Studies of individuals at risk of RA [EULAR disease phases (a) and (b)]

The ability to identify individuals in the preclinical phase of RA is challenging. Therefore studying individuals who are at high risk of developing RA in the future due to known risk factors such as family history provides a unique opportunity to study the biological events that precede disease development.

El-Gabalawy et al. [3] investigated non-HLA genes for association with RA in a North American Native (NAN) population previously shown to be predisposed to RA due a high prevalence of RA, multicase families and a high background frequency of HLA-DRB1 risk alleles. The authors tested 21 non-HLA SNPs previously associated with RA and showed HLA-DRB1 to be the major genetic risk factor for RA in this population. Additionally, the presence of the SE and the minor allele of MMEL1-TNFRSF14 significantly reduced RA risk, whereas TRAF1-C5 increased the risk, showing that additional risk factors outside of the HLA can contribute to disease risk in this predisposed population.

A limited number of studies have been carried out in first-degree relatives (FDRs) of RA patients who could be considered an at-risk population. One study showed that FDRs had a higher prevalence of RA autoantibodies than healthy controls, and individuals who were ACPA positive and rheumatoid factor positive had the highest prevalence of joint symptoms [88]. It is possible that some of these patients, particularly individuals with positive autoantibody status, may develop disease in the next few years [89] and therefore would benefit from symptom monitoring to allow early treatment. A study in the same NAN population showed an increase in the levels of multiple cytokines and CRP in FDRs compared with controls [90]. Other studies in the relatives of RA patients have also identified the presence of ACPAs in healthy relatives [91] and demonstrated that the fine specificity of ACPAs in healthy FDRs and their related RA patient can be different [92].

To gain the true benefit of studying these at-risk populations, prospective cohorts of FDRs will need to be established for long-term follow-up.

Environmental risk factors [EULAR disease phase (b)] and gene–environment interactions in RA

The heritability of RA (50–60%) suggests that a large proportion of disease could be due to environmental risk factors [33, 93]. More studies regarding environmental risk factors are required, although identifying environmental risk factors can be challenging due to recall bias and difficulties distinguishing cause from effect.

However, the identification of environmental risk factors presents an interesting challenge, as prevention strategies based on avoidance of exposure to risk would be attractive. The possibility that RA starts outside the joints raises many questions about the role of environmental factors. Could an environmental influence be responsible for the initiation of autoimmunity? Or could an environmental factor be the trigger to drive the transition to clinical disease? It is likely to be the environmental contribution to disease that interacts with a susceptible individual’s genetic component to alter disease course and progression. Understanding the influence of the environment may help us to understand the initial phases of disease and how these can be altered.

Several environmental risk factors for RA have been implicated in the development of disease, although few of these have substantial evidence [94–105]. A recent prospective study identified pack-years of smoking, diabetes mellitus, BMI and parity as risk factors for developing RA or IP, while alcohol, higher social class and breastfeeding were associated with a decreased risk of RA or IP [106], although conflicting results have been reported regarding alcohol [107].

Smoking is the most well-established environmental risk factor for RA [101, 108–110], and several studies have shown a strong interaction between smoking and the alleles of the HLA-DRB1 SE. Studies in the Swedish Epidemiological Investigation of Rheumatoid Arthritis (EIRA) cohort have shown that individuals who have ever smoked and carry SE alleles are at increased risk of developing RF- or ACPA-positive RA [111–113]. This effect was not observed in ACPA-negative RA. This gene environment interaction was replicated in a Danish case–control study [114]. However, attempts to replicate this interaction in three North American cohorts with RA were unsuccessful and only weak evidence for a gene environment interaction was found in one cohort [North American Rheumatoid Arthritis Consortium (NARAC)] [115].

Further studies by Klareskog et al. [111] have demonstrated the presence of citrullinated proteins in bronchoalveolar lavage cells from smokers, allowing the possibility that citrullination of proteins in the lung caused by smoking may, in the presence of HLA-DRB1 SE alleles, result in RA-specific immune reactions to citrullinated proteins. However, citrullination has been found throughout the body, so it is unclear how citrullination in the lungs may be important in the development of ACPA.

As smoking has been shown to increase citrullination [111], it is hypothesised that smoking could promote autoimmunity to citrulline residues [116]. Presentation of these citrullinated peptides to T cells would then initiate proliferation, differentiation, cytokine production and the formation of T cell memory against citrullinated peptides.

Predicting disease risk

The identification of genetic and environmental risk factors for RA has raised the question as to whether these risk factors are sufficient to predict individuals who are at risk of disease. Indeed, one of the ultimate aims of genetic research is to predict who will develop disease before symptom onset and joint erosion/destruction begins. However, the modest effects of the loci identified by GWASs have left their predictive ability in question. There have been several studies in recent years that have used genetic risk factors, and in some cases environmental risk factors, to try to predict disease risk [106, 117–123]. In general these models have demonstrated the ability to identify a subset of individuals at high risk with relatively high specificity (80–90%), but with relatively low sensitivity (∼40%), resulting in a significant proportion of high-risk individuals being misclassified as not at risk. Stratifying RA into disease stages and profiling the genetics of individuals in each group in order to identify distinct profiles may provide insight into which loci are involved in specific processes and inform targeting of more aggressive treatments to a particular subgroup of patients.

Due to the low prevalence of RA, whole population screening of individuals to identify those at risk is not economical. However, risk prediction may prove more fruitful in targeted subsets of individuals who are classified as at risk due to the presence of a family history of disease or environmental risk factors and is the premise of many primary prevention studies.

It should also be noted that the genetic markers identified to date only account for about half of the heritability of RA. Further studies to account for a larger proportion of the heritability of disease and the addition of further risk factors such as environmental factors, biomarkers, and clinical predictors will almost certainly improve the power of predictive models and may prove more clinically useful.

The genetic susceptibility variants identified to date strongly implicate several immune pathways in the development of RA [1]. In addition to this, there is already significant overlap between the targets of several approved RA therapies and identified susceptibility genes for RA (CTLA4: abatacept; TYK2: tofacitinib; IL-6: tocilizumab) demonstrating the potential of genetics not only to identify biological pathways that lead to RA, but also that these pathways can be targeted and lead to the successful treatment of disease symptoms.

Rheumatology key messages.

• Cohorts representing all phases of RA are essential for studies to define risk of progression.

• Prospective studies of individuals with genetic/environmental risk factors will help in understanding preclinical RA.

• ACPA-negative RA patients will be further classified by fine specificities of autoantibodies.