Article Text

Download PDFPDF

Extended report
Rheumatoid arthritis susceptibility genes associate with lipid levels in patients with rheumatoid arthritis
  1. Tracey E Toms1,2,
  2. Vasileios F Panoulas1,
  3. Jacqueline P Smith1,
  4. Karen M J Douglas1,
  5. Giorgos S Metsios1,3,
  6. Antonios Stavropoulos-Kalinoglou1,3,
  7. George D Kitas1,2
  1. 1Department of Rheumatology, Dudley Group of Hospitals NHS Trust, Russells Hall Hospital, Dudley, West Midlands DY1 2HQ, UK
  2. 2Arthritis Research UK Epidemiology Unit, Manchester University, Manchester M13 9PT, UK
  3. 3School of Sport, Performing Arts and Leisure, University of Wolverhampton, Walsall WS1 3BD, UK
  1. Correspondence to Professor George D Kitas, Department of Rheumatology, Dudley Group of Hospitals NHS Trust, Russells Hall Hospital, Pensnett Road, Dudley, West Midlands DY1 2HQ, UK; gd.kitas{at} or g.d.kitas{at}


Introduction Rheumatoid arthritis (RA), a systemic inflammatory disease with complex genetic aetiology, associates with excess cardiovascular morbidity and mortality. Dyslipidaemia, a major cardiovascular risk factor has been reported to predate the onset of RA, thus suggesting a potential genetic link between the two conditions. The authors assessed whether RA susceptibility genes associate with the presence of dyslipidaemia in RA patients.

Methods 400 well-characterised RA patients were included in this cross-sectional study. Fasting lipid profile (total cholesterol, high-density lipoproteins (HDL), low-density lipoproteins (LDL), triglycerides, apolipoproteins (ApoA and ApoB) and lipoprotein (a)) and four RA susceptibility genes (PTPN22, TRAF1/C5, STAT4 and human leucocyte antigen shared epitope (HLA-SE)) were assessed and associations were sought in both univariate and multivariate analyses.

Results Following adjustment for age, sex and erythrocyte sedimentation rate, the G allele of TRAF1/C5 associated with lower total cholesterol (p=0.010), LDL (p=0.022) and ApoB (p=0.014); one or more copies of the shared epitope associated with lower ApoA (p=0.035) and higher ApoB:ApoA ratio (p=0.047); while STAT4 TT homozygotes had higher lipoprotein (a) (p=0.004).

Conclusions RA susceptibility genes (TRAF1/C5, STAT4 and HLA-DRB1-SE) may be involved in the regulation of lipid metabolism in RA patients, thus contributing to cardiovascular disease (CVD) risk and adverse outcome. If these findings are replicated, such genotyping could be used to identify and target for prevention those RA patients most at risk of CVD. It will also be interesting to study the association of these genes with lipid levels in the general population and identify mechanisms to explain the link.

Statistics from

Rheumatoid arthritis (RA) affects approximately 1% of the population1 and carries a significant cardiovascular burden.2 3 The excess cardiovascular risk can be explained both by traditional risk factors (eg, hypertension,4 dyslipidaemia,5 obesity6 and physical inactivity)7 and novel risk factors (eg, systemic inflammation).8 Dyslipidaemia is one of the strongest predictors of cardiovascular disease (CVD) in the general population and is highly prevalent in RA (according to the National Cholesterol Education Programme (NCEP) guidelines) affecting approximately half of all patients.9

The aetiology of RA is complex, with genetic, environmental, infectious and hormonal factors thought to contribute.10 Genetic factors play a significant role in the susceptibility to RA, with twin studies demonstrating an excess disease concordance between monozygotic twins compared with dizygotic twins.11 In addition, RA has a familial tendency with siblings of RA patients found to be at a 2–17 times increased risk of developing RA compared with the general population.12 The first significant discovery in the field of susceptibility genes in RA was made several decades ago with the discovery of the human leucocyte antigen (HLA)-DRB1 gene.13 To date, several specific alleles in the HLA-DRB1 gene have been linked to RA (the shared epitope; SE); however, the frequency of each of these alleles is seen to vary according to ethnicity.14 Although, the HLA-DRB1 gene significantly contributes to the genetic susceptibility of RA it does not fully account for the genetic component of RA.15 Recent advances in both our understanding of genetics and genetic technology have led to the identification of a number of other susceptibility genes including STAT4, TRAF1 and PTPN22.16,,18

Although multifactorial, CVD in RA has been shown to be partly genetically determined. Of interest, several of the RA susceptibility genes may contribute to the excess CVD morbidity and mortality in RA.19 20 It has also been reported that changes in the lipid profile may occur in individuals up to 10 years before the diagnosis of RA, indicating a potential genetic link between RA and dyslipidaemia.21 22

In this study we aimed to assess whether the presence of four different RA susceptibility genes associate with lipid abnormalities in patients with RA.


Four hundred consecutive RA patients fulfilling the American College of Rheumatology criteria23 were recruited between 2004 and 2006 from outpatient clinics held at the Dudley Group of Hospitals NHS Foundation Trust (the Dudley Rheumatoid Arthritis Comorbidity Cohort). The study was granted full ethical approval from the local research ethics committee and all patients gave their informed written consent before enrolment and participation in the study.

All RA patients had extensive baseline data collected, by means of a one-to-one interview, retrospective case note analysis and a full clinical examination performed by a rheumatologist. This facilitated the documentation of patient demographics, disease characteristics, comorbidities, drug use and family history, as well as recording measurements such as height, weight, body mass index and waist circumference. Contemporary disease activity and physical function were assessed using the disease activity score (DAS28)24 and the health assessment questionnaire (HAQ),25 respectively. Insulin resistance was evaluated from fasting glucose and insulin using the homeostasis model of assessment of insulin resistance.26 A detailed drug history was recorded to include, lipid-lowering drugs (statins and fibrates), antihypertensive medications, disease-modifying anti-inflammatory drugs, non-steroidal anti-inflammatory drugs, cyclooxygenase II inhibitors and oral glucocorticoids.

DNA and serum samples obtained at baseline from fasted subjects were analysed in a single laboratory. DNA was extracted from whole blood using the QuickGene-810 system (Fujifilm Corp., Tokyo, Japan).27 This method has been described in detail in previous studies by our group.28 Analysis of the serum from RA patients included: a full blood count, urea and electrolytes, liver function tests, thyroid function tests, glucose, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) and a full lipid profile.

The Vitros 5, IFS chemistry system (Ortho Clinical Diagnostics Inc., Rochester, New York, USA) was used to measure all lipid components; however, total cholesterol, high-density lipoproteins (HDL) and triglycerides were measured using multilayered slides, whereas measurement of low-density lipoproteins (LDL) required a dual chamber package. Apolipoproteins A and B (ApoA and ApoB) and lipoprotein (a) were all measured using an automated Konelab method (Thermo-Electron Clinical Diagnostics, Vantaa, Finland).

Dyslipidaemia was defined by the NCEP criteria29 as one or more of the following: total cholesterol 6.2 mmol/l or greater, HDL less than 1.03 mmol/l, LDL 4.13 mmol/l or greater, triglycerides 1.7 mmol/l or greater or taking lipid-lowering therapy.


The Roche LightCycler 2.0 system (2007c) (Roche Diagnostics Ltd., Burgess Hill, UK) was used to identify the single-nucleotide polymorphisms of STAT4 (rs7574865), TRAF1/C5 (rs3761847) and PTPN22 (rs2476601) using real-time PCR and melting curve analysis. These four susceptibility genes were chosen for analysis over the array of other RA susceptibility genes as they have been shown to contribute the most to the genetic susceptibility of RA30

Primers and probes used for the rs7574865 single-nucleotide polymorphism included, primers: 5′-TACGGATGTCTTTGAAGGTAG–3′ (forward) and 5′-CTTTATAATTTCTTTCT–3′ (reverse), probes: the sensor (G) 5′-AGATAACCACTATTCACATTTT–3′-FLU and the anchor a 25-mer oligonucleotide labelled with LC Red640 and its 5′ end and modified at the 3′ end by phosphorylation to block extension: 5′-LCRED640–CCAACTTTTCATACTTTTACTGCATACACAC–PH. For rs3761847, primers: 5 ′-ACTCCCTTTTAACTGTGTACCCCATA–3′ (forward) and 5′-GCTTAGCCTCTCTGTGCCTCAG–3′ (reverse), probes: the sensor (T) 5′-TCTCCCCTCCAGCCTCAA–3′-FLU and the anchor 5′-LCRED640–ACCACCCTCTCTCTACCTGCTCATTCCCA–PH. For rs2476601, primers: 5′-GCCTCAATGAACTCCTCAAAC–3′ (forward) and 5′-CTGATAATGTTGCTTCAACGGA–3′ (reverse), probes: the sensor (A) 5′-CAGGTGTCCATACAGGAAGTG–3′-FLU and the anchor 5′-LCRED640–GGGGATTTCATCATCTATCCTTGGAGCAGTTG–PH. The risk alleles for each of the RA susceptibility genes were: STAT4 rs7574865 (T allele), TRAF1C5 rs3761847 (G allele), PTNP22 rs2476601 (T allele) and one or more copies of the HLA-DRB1-SE.

The genotyping of the HLA-DRB1-SE (HLA-SE) was performed using reverse line assay sequence-specific oligonulceotide probes with the Dynal RELI sequence-specific oligonucleotide strip detection reagent kit ( Assay results were interpreted using the pattern matching program provided by Dynal (Invitrogen, Paisley, UK). The following alleles were classified as SE positive: DRB1*0101, *0102, *0104, *0401, *0404, *0405, *0408, *0413, *0416, *1001 and *1402.13

Of the 400 RA patients in total we were able to genotype 394 patients for STAT4, 397 patients for PTPN22, 387 patients for TRAF1C5 and 355 patients for the HLA-SE.

For the analysis of lipid parameters (lipid levels and lipid ratios) patients on lipid-lowering therapy were excluded due to their confounding effect.

Statistical techniques

Analysis of the data was carried out using SPSS 15.0. The Kolmogorov–Smirnov test was used to evaluate the distribution of each parameter. Comparisons were performed by analysis of variance, Kruskal–Wallis and χ2 tests for normally distributed, non-normally distributed and categorical variables, respectively. Values were expressed as mean±SD, median (25th to 75th percentile) or percentages, as appropriate. The independence of the associations of lipid parameters with the genotypes was established using a multivariate generalised linear model, whereas a linear regression model was used to establish associations between alleles. All multivariate models were adjusted for multiple comparisons. For all lipid-associated analyses patients on lipid-lowering therapy (statins/fibrates) were excluded due to their confounding effects. All genotype frequencies were found to be in Hardy–Weinberg equilibrium.

The study had at least 80% power at the 5% significance level to detect the differences we demonstrate in lipid levels.


Univariate analysis

Details of the demographics, disease characteristics, comorbidities and cardiovascular risk factors/disease across the genotypes for each (PTPN22, STAT4, TRAF1/C5 and the SE) of the susceptibility genes are summarised in tables 1–4. No statistically significant differences were observed in age, gender, disease activity (DAS28, ESR, CRP) or disease severity (HAQ scores) across the genotypes for PTPN22, STAT4 or TRAF1/C5. However, having one or more copies of the HLA-DRB1-SE associated with significantly higher levels in CRP and ESR. In addition, there were some significant variations in RA characteristics: rheumatoid factor positivity (p<0.001) and anticyclic citrullinated peptide (anti-CCP) positivity (p<0.001) were higher among RA patients with one or more copies of the HLA-DRB1-SE; anti-CCP positivity was higher among RA patients either heterozygous or homozygous for the TRAF1/C5 G allele (p=0.018); disease duration was higher among patients with one or more copies of the HLA-DRB1-SE (p=0.021).

Table 1

Demographic, clinical and laboratory characteristics of the cohort according to the STAT4 genotypes

Table 2

Demographic, clinical and laboratory characteristics of the cohort according to the PTPN22 genotypes

Table 3

Demographic, clinical and laboratory characteristics of the cohort according to the TRAF1C5 genotypes

Table 4

Demographic, clinical and laboratory characteristics across the HLA-DRB1 shared epitope

Three out of the four RA susceptibility genes (HLA-DRB1-SE, TRAF1/C5 and STAT4) examined were found to associate with individual lipid parameters in the univariate analysis but not with NCEP-defined dyslipidaemia. Although associations were observed across the three genotypes, the associations were much stronger when comparing the allelic effects (eg, AA vs G allele (GA or GG)). Comparisons across the three genotypes on lipid parameters demonstrated a significant association between TRAF1/C5 and total cholesterol levels (AA 5.7±1.27 vs AG 5.35±1.06 vs GG 5.35±0.99; p=0.039) and STAT4 and lipoprotein (a) levels (GG 0.08 (0.03–0.19) vs GT 0.10 (0.04–0.24) vs TT 0.19 (0.05–0.49); p=0.013). However, the comparison of allelic effects strengthened these associations and demonstrated a number of other significant associations between lipid parameters and RA susceptibility genes. Patients with one or more copies of the HLA-DRB1-SE were found to have significantly lower levels of ApoA (1.62±0.43 vs 1.76±0.51; p=0.026), a higher ApoB:ApoA ratio (0.66±0.24 vs 0.58±0.21; p=0.024) and a higher total cholesterol:HDL ratio (3.65±1.04 vs 3.33±0.88; p=0.020) than patients with no copies of the SE. Patients heterozygous or homozygous for the G allele of TRAF1/C5 were found to have significantly lower total cholesterol levels (5.35±1.04 vs 5.7±1.27; p=0.011), LDL levels (3.15±1.11 vs 3.48±1.25; p=0.020) and ApoB levels (0.97±0.27 vs 1.05±0.33; p=0.023) than patients homozygous for the A allele. Patients homozygous for the STAT4 T allele had significantly higher levels of lipoprotein (a) (0.185 (0.05–0.49) vs 0.09 (0.03–0.22); p=0.007) compared with patients homozygous or heterozygous for the G allele.

Multivariate analysis

Following adjustment for potential confounders (age, sex, ESR and other significant associations identified in the univariate analysis specifically for each gene) the associations between each of the individual RA susceptibility genes and lipid parameters remained unchanged, except the association of TRAF1/C5 and the total cholesterol:HDL ratio, which lost significance (p=0.066; see table 5).

Table 5

Multivariate analysis of RA susceptibility genes with lipid parameters


We demonstrate for the first time a potential genetic link between RA and lipid parameters independent of inflammation and other RA-specific factors. The potential importance of these findings is far reaching, both for our overall understanding of lipid metabolism and CVD in RA but also for the identification and preventive management of CVD in individual RA patients.

Patients harbouring the AA genotype for TRAF1/C5, the TT genotype of STAT4 or one or more copies of the HLA-DRB1-SE appear to be at most risk of CVD, with these genotypes associating with pro-atherogenic changes in the lipid profile, for example, STAT4 TT genotype associating with increased levels of lipoprotein (a) or the HLA-DRB1-SE associating with increased ApoB:ApoA and total cholesterol:HDL ratios. The identification of patients harbouring these genetic polymorphisms may thus aid screening and aggressive management of lipid-associated CVD risk in RA.

Two papers have reported that changes in the lipid profile occur many years before the onset of RA.21 22 The first study demonstrated that blood donors who later developed RA (n=79) had significantly higher levels of total cholesterol, triglycerides and ApoB, but lower HDL levels than matched controls (n=1071). The second study demonstrated that total cholesterol and LDL levels were significantly lower during the 5 years before the onset of RA in a large population-based incident cohort (577 RA patients and 540 non-RA controls).21 The changes observed in the lipid profile before the onset of RA could be the result of either subclinical inflammation, genetic predisposition or a range of other unknown factors. Interestingly, the study by van Halm et al22 attempted to assess whether inflammatory parameters could account for the magnitude of lipid changes observed. However, they demonstrated that only a very small percentage of the difference in lipid levels between RA patients and controls could be explained by changes in CRP; for example, only 3.6% of the difference in HDL levels between the groups could be explained by CRP concentrations. A further population-based, prospective, nested case–control study failed to demonstrate any difference in lipid levels (total cholesterol, HDL, LDL, triglycerides) between patients who developed inflammatory polyarthritis and controls.31 The results of this study may differ from the results of the previous two studies due to differences in the populations studied (eg, RA vs inflammatory polyarthritis), the size of the population studied and differences in the frequency of genetic polymorphisms, for example, susceptibility genes.

To date there are limited data available on the wider CVD effects of RA susceptibility genes. Current evidence would suggest that patients with two copies of the HLA-DRB1-SE (particularly the HLA-DRB1*01/04 combination) have an increased risk of all-cause and CVD mortality.19 32 In addition, the presence of the HLA-DRB1*0404 allele is associated with decreased endothelium-dependent vasodilatation.33 Although several studies have failed to demonstrate an association between polymorphisms of TRAF1/C5 with CVD mortality in RA,34 35 a further susceptibility variant at the CCL21 locus has been shown to associate with CVD mortality in patients with inflammatory polyarthritis.20

Despite these advances, the pathological mechanisms that may link the presence of susceptibility genes with CVD in RA remain poorly understood. Perhaps the most obvious mechanism is mediation through an inflammatory pathway, as the presence of certain RA susceptibility genes (especially HLA-DRB1-SE) associates with more severe, erosive disease.36 37 However, it is possible that RA susceptibility genes could mediate/partly mediate their effects on CVD in RA through both direct and indirect effects on traditional CVD risk factors. This is the first study to date to have specifically assessed whether RA susceptibility genes associate with one of the key traditional CVD risk factors (dyslipidaemia/lipid parameters). The observation that three out of the four RA susceptibility genes examined associate with individual lipid levels is interesting. However, the diversity of genetic effects on the lipid profile, for example, STAT4 only affects lipoprotein (a) levels, while HLA-DRB1-SE affects total cholesterol, LDL and ApoB levels, would suggest that each susceptibility gene acts independently through specific mechanisms rather than through generic effects on the inflammatory process. This thought is supported by our analysis in which adjustment for inflammation (ESR) in the multivariate models did not influence the strength of association between each of the susceptibility genes and lipid levels. Factors other than inflammation may be important, for example enzymes involved in lipid metabolism. A recent study by Palmino-Morales et al38 failed to demonstrate an association of PTPN22, STAT4 and TRAF1C5 polymorphisms with cardiovascular risk in RA. Although an important study, it is not without limitations, with the authors failing to consider the major susceptibility gene, HLA-SE. It is clear that to establish fully whether RA susceptibility genes are important determinants of CVD in RA, much larger studies with hard endpoints are required.

The extensive characterisation of a large cross-sectional RA population has enabled us to perform an in-depth study of the associations of the RA susceptibility genes, something not always feasible with large genome-wide association studies. However, the study has limitations. First, the cross-sectional design precludes firm conclusions on the directionality or causality of the associations observed. Second, the sample size provided just enough power for most of the differences found. Third, the absence of a normal control group does not allow any conclusions about the potential association of these genes with lipid levels in the general population. Unfortunately, analysis of the Welcome Trust Case Consortium did not offer a means of confirming our findings or overcoming some of the above limitations, due to the lack of stored clinically relevant data, for example, lipid levels and inflammatory markers.39In addition, despite adjustment for inflammation in the multivariate analyses, we are unable to assume a direct link between lipid levels and the susceptibility genes.

In conclusion, this is the first time that significant associations between several RA susceptibility genes and lipid parameters have been observed in patients with RA. These findings may have important implications for both the screening and management of CVD risk in such patients. Further large-scale studies are required to confirm these findings and establish the underlying mechanisms, both in RA and in the general population.



  • Funding This work was supported by an Arthritis Research UK clinical research fellowship grant (grant no 18848 to TET), and an Arthritis Research UK infrastructure support grant (grant no 17682, given to the Dudley Group of Hospitals NHS Foundation Trust, Department of Rheumatology). VFP was supported by a PhD scholarship from the Empirikion Foundation, Athens, Greece.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the Black Country Ethics Committee, UK.

  • 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.