ARTHRITIS & RHEUMATISMVol. 44, No. 1, January 2001, pp 61–65© 2001, American College of Rheumatology

Association Between Rheumatoid Arthritis andPolymorphism of Tumor Necrosis Factor Receptor II,

But Not Tumor Necrosis Factor Receptor I, in Caucasians

Anne Barton, Sally John, W. E. R. Ollier, Alan Silman, and Jane Worthington

Objective. Tumor necrosis factor (TNF) is a pow-erful mediator of inflammation in rheumatoid arthritis(RA). In vivo, its acute effects are limited by binding tosoluble receptors (TNFR), suggesting that TNFR genescould be important candidate risk factors. The presentstudy was undertaken to investigate association of poly-morphisms of TNFRI and TNFRII with RA in subjectsin the UK.

Methods. Unrelated Caucasian RA patients (n 5291) and healthy Caucasian controls (n 5 143) weregenotyped for A/G polymorphism in exon 1 of TNFRI.From this sample, 240 of the patients and 137 controlswere also typed for a single-nucleotide polymorphism(SNP) in exon 6 of the TNFRII gene. In followup studies,DNA samples from UK Caucasian RA patients with apositive family history (n 5 149) and UK Caucasianpatients with sporadic RA (n 5 208) were also typed forthe exon 6 TNFRII polymorphism.

Results. TNFRI polymorphism was not associatedwith RA (odds ratio [OR] for GG genotype 0.93, 95%confidence interval [95% CI] 0.54–1.60). For TNFRII, inthe initial study group, patients with RA were signifi-cantly more likely to be positive for both the G allele andGG genotype than were controls (OR for GG genotype2.55, 95% CI 1.11–5.86). The association appeared to beconfined to those with a family history of RA. Thisfinding was replicated in an independent cohort ofpatients with familial RA.

Conclusion. The results of this study provide

evidence of association between an SNP in the TNFRIIgene and RA, the strongest association being observedin patients with a family history. No evidence of associ-ation between RA and TNFRI was demonstrated.

Rheumatoid arthritis (RA) is a chronic inflam-matory disease. Excess concordance in monozygoticcompared with dizygotic twins suggests a genetic com-ponent to RA susceptibility (1). The genetic componentin RA is 50–60% as calculated using Falconer’s estimateof heritability (2). Although HLA is the only major locusthat has been consistently identified to date, a single-locus model for RA has been rejected. HLA is estimatedto contribute 20–50% of the total genetic component ofdisease susceptibility, and it is likely that there are anumber of other genes involved, each making a smallercontribution (3). A better understanding of the geneticsusceptibility factors contributing to RA may lead to theidentification of novel therapeutic targets.

In a complex disease like RA, the detection ofweak genetic effects can be complicated by the fact thatgenetic heterogeneity exists, i.e., different genes willcontribute to disease susceptibility in different people.In families in which more than 1 individual has thedisease, it is likely that the genetic component of diseasesusceptibility is stronger. Previous studies have shownthis to be the case with the major RA genetic suscepti-bility locus, HLA. Probands with a family history of RAare more likely to be DR4 positive and have a youngerage at disease onset (4,5). Studying such multicase RAfamilies may facilitate identification of weak geneticeffects.

Inflammation in RA is mediated by a number ofcytokines, but tumor necrosis factor a (TNFa) appearsto be particularly important. TNFa acts via binding tocell surface receptors I and II (TNFRI and TNFRII).The extracellular portions of both receptors can be shedand the soluble forms still retain the ability to bind to

Supported by the Arthritis and Rheumatism Campaign, UK.Dr. Barton’s work was supported by the Medical Research Council, UK.

Anne Barton, MBChB, Sally John, PhD, W. E. R. Ollier,PhD, Alan Silman, MD, Jane Worthington, PhD: University ofManchester, Manchester, UK.

Address correspondence and reprint requests to Anne Bar-ton, MBChB, ARC-EU, Stopford Building, University of Manchester,Manchester M13 9PT, UK.

Submitted for publication July 19, 2000; accepted in revisedform September 27, 2000.

61

TNFa, thus limiting its acute effects. Soluble TNFR(sTNFR) have a higher affinity for TNFa thanmembrane-bound TNFR. In inflammation, levels ofboth TNFa and TNFR are increased, but the levels ofTNFR are inadequate for complete neutralization ofTNFa. Animal studies using the collagen-induced arthri-tis (CIA) model of inflammatory arthritis have shownthat mice deficient in TNFR have a more severe earlyinflammatory response (6). Furthermore, a modifiedTNFRI gene has been used to suppress CIA in rats (7).Recent clinical studies in which administration ofsTNFRII was used to treat disease in patients with RAhave highlighted its importance in the inflammatoryprocess (8).

The TNFRI gene maps to 12p13 and the TNFRIIgene to 1p36. Whole genome screens (WGS) in rats withadjuvant-induced arthritis (9) and mice with CIA (10)have revealed linkage to regions syntenic to 12p13.Linkage to this region was also detected in a WGS ofEuropean RA affected sibpair (ASP) families (11). Twogenome-wide linkage scans in RA have shown linkage tomarkers mapping to 1p36 (11,12), and linkage has beenreplicated in a second cohort of European HLA-identical RA ASP families (13). One approach to fol-lowing up linkage data is to investigate associationbetween polymorphism in candidate genes mappingwithin regions of linkage and RA. Association studiesprovide a powerful approach to detecting genes withsmall effect. Demonstration of an association of RAeither with specific alleles of a marker (resulting fromlinkage disequilibrium with a potential functional poly-morphism within the gene) or with the functional poly-morphism itself would provide evidence of a role in thepathogenesis of RA. TNFRI and TNFRII are strongcandidate RA genes based on their known biologicfunction and on findings in animal models of inflamma-tory disease. The aim of the current study was to directlytest for association between RA and the TNFRI andTNFRII genes.

PATIENTS AND METHODS

Study design. A case–control (association) study wasperformed to investigate the association of TNFRI and TNF-RII with RA in patients in the UK. Cases with RA werecompared with population controls for both phenotype andgenotype at the 2 loci. Cases were stratified according to thepresence of a family history in a first-degree relative. The studyincluded 2 cohorts of patients: a test group and a verificationgroup.

RA cases. Cases with RA were obtained from 2 mainsources: from the Arthritis Research Council (ARC) National

Repository of patients and families and from local clinics.Patients identified through the National Repository wereexamined by a metrologist using a standardized questionnaireand examination protocol, including completion of a question-naire on family history. In the case of any relatives reported byprobands to have RA, the RA was verified using the sameapproach. For clinic patients, there was no information avail-able regarding presence or absence of a family history. Basedon previous estimates, it can be assumed that up to 4% ofpatients will have a first-degree relative with RA. All RA caseshad disease that satisfied the American College of Rheuma-tology (formerly, the American Rheumatism Association) 1987criteria (14) modified for genetic studies of prevalent cases(15), and all were of UK caucasoid origin. The RA cases weredivided into a test group and a replication group. In the testgroup, 80% of patients had a family history of RA in afirst-degree relative, while the verification group comprised149 probands with a family history of RA (41.7%) and 208without (58.3%).

Controls. As controls, healthy subjects with no historyof inflammatory joint disease were recruited from generalpractice (n 5 67) or from among blood donors (n 5 76). Allcontrols were of UK caucasoid ethnic origin.

Power calculation. Sample sizes were calculated basedon published allele frequencies so that each of the associationstudies had 80% power to detect a gene conferring a genotypicrelative risk of 2.0 at the 5% significance level.

TNFRI restriction fragment length polymorphism(RFLP). A single-nucleotide polymorphism (SNP) at position36 in exon 1 of the TNFRI gene (A/G), which creates arecognition site for the restriction enzyme Msp AII (NewEngland Biolabs, Hertfordshire, UK) when the G allele ispresent but not when the A allele is present (16), has recentlybeen identified. The primer sequences and reaction conditionshave been described (16).

TNFRII RFLP. There is a T/G polymorphism at posi-tion 196 in exon 6 of the TNFRII gene, which creates arecognition site for the restriction enzyme Nla III (NewEngland Biolabs) when the T allele is present but not when theG allele is present (17). This polymorphism results in an aminoacid substitution (methionine to arginine), although the effectof this change on the function of the TNFRII gene is unknown.The primer sequences and reaction conditions are describedelsewhere (17).

Statistical analysis. Allele and genotype frequencies inpatient and control groups were calculated by direct counting.The association of the rarer G allele with RA for both SNPswas assessed by calculating the odds ratio (OR) and 95% con-fidence intervals (95% CI) for the G phenotype and the GGgenotype. P values less than 0.05 were considered significant.

RESULTS

TNFRI polymorphism in patients and controls.DNA was available from 291 RA probands from theARC National Repository (the test group). Seventy-ninepercent were female, 80% had a family history of RA ina first-degree relative, 78% were rheumatoid factor(RF) positive, and 58% had erosive disease. There was

62 BARTON ET AL

no significant difference in the frequency of the G allelebetween patients and controls (OR for GG genotype0.93, 95% CI 0.54–1.60, P 5 0.79; OR for G phenotype0.92, 95% CI 0.60–1.41, P 5 0.69) (Table 1). Allelefrequencies were similar to those reported previously(16). Since RA is a heterogeneous disease, associationsmay be found in subgroups of patients. No associationwith subgroups stratified for sex, RF positivity, presenceof erosions, family history, or age at onset of disease wasidentified. HLA data were available from 137 cases.Stratifying for the presence of the shared epitope did notaffect the results.

TNFRII polymorphism in patients and controls.Test cohort. Based on published allele frequencies, wecalculated that genotyping would have to be performedon 220 RA patients and 110 controls in order to achieve80% power for detection of a gene conferring a geno-typic relative risk of 2.0 at the 5% significance level. Ofthe 291 patients and 143 controls in the test group whowere genotyped in the TNFRI study, 240 patients and

137 controls were also genotyped for the TNFRII exon6 T/G polymorphism. Positivity for both the G allele andthe GG genotype was significantly more frequent inpatients with RA than in controls (OR for GG genotype2.55, 95% CI 1.11–5.86, P 5 0.03; OR for G phenotype1.54, 95% CI 1.01–2.34, P 5 0.046) (Table 2). When thetest group was stratified according to presence or ab-sence of a family history of RA, however, the RA caseswithout a family history showed allele and genotypefrequencies that were similar to control frequencies. Thedifferences in GG genotype and G phenotype frequen-cies were confined to those with a family history of RA(OR for GG genotype 2.89, 95% CI 1.24–6.72, P 50.013; OR for G phenotype 1.81, 95% CI 1.16–2.81, P 50.009).

Verification cohort. DNA was available from 149RA probands from the ARC National Repository whohad a family history of RA in a first-degree relative.Seventy-five percent were female, 84.8% were RF pos-itive, and 75% had erosive disease. DNA was availablefrom an additional 208 individuals with sporadic RA, ofwhom 74% were female, 72% were RF positive, and68% had erosive disease. All were typed for the exon 6TNFRII polymorphism. In this group, too, the frequencyof both the G allele and GG genotype was statisticallysignificantly different in RA cases with a family historyof RA compared with controls (OR for GG genotype3.05, 95% CI 1.28–7.24, P 5 0.011 [Table 2]; OR for Gphenotype 1.65, 95% CI 1.03–2.63, P 5 0.036).

Combined test and verification groups. When datafrom the groups were combined, there was a highly

Table 1. Frequencies of the TNFRI position 36 polymorphism inUK RA patients and controls*

AA AG GG

RA (n 5 291) 95 (32.6) 152 (52.3) 44 (15)Controls (n 5 143) 44 (30.8) 76 (53.1) 23 (16.1)

* Values are the number (%) of patients. Odds ratio (OR) for the GGgenotype 0.93 (95% confidence interval [95% CI] 0.54–1.60, P 5 0.79);OR for the G phenotype 0.92 (95% CI 0.60–1.41, P 5 0.69). For theoverall distribution between the 2 groups, x2 (2 degrees of freedom) 50.178, P 5 0.915). TNFRI 5 tumor necrosis factor receptor I; RA 5rheumatoid arthritis.

Table 2. Frequencies of the TNFRII exon 6 polymorphism in UK RA patients and controls*

TT TG GG

OR, GGgenotype(95% CI) P

Initial studyCombined RA 104 (43.3) 107 (44.6) 29 (12.1) 2.55 (1.1–5.9) 0.032Familial RA 76 (39.4) 91 (47.1) 26 (13.5) 2.89 (1.2–6.7) 0.013Nonfamilial 28 (59.6) 16 (34.0) 3 (6.4) 1.26 (0.3–4.7) 0.74Controls 74 (54.0) 56 (40.9) 7 (5.1)

Replication studyCombined RA 168 (47.1) 152 (42.6) 37 (10.3) 2.15 (0.9–4.8) 0.066Familial RA 62 (41.6) 66 (44.3) 21 (14.1) 3.05 (1.3–7.2) 0.01Nonfamilial 106 (51.0) 86 (41.3) 16 (7.7) 1.55 (0.6–3.8) 0.35Controls 74 (54.0) 56 (40.9) 7 (5.1)

Combined groupsCombined RA 272 (45.6) 259 (43.4) 66 (11.0) 2.30 (1.1–5.1) 0.04Familial RA 138 (40.4) 157 (45.9) 47 (13.7) 2.96 (1.3–6.6) 0.007Nonfamilial RA 134 (52.5) 102 (40.0) 19 (7.5) 1.50 (0.6–3.6) 0.37Controls 74 (54.0) 56 (40.9) 7 (5.1)

* Values are the number (%) of patients. TNFRII 5 tumor necrosis factor receptor II; RA 5 rheumatoidarthritis; OR 5 odds ratio; 95% CI 5 95% confidence interval.

TNFRII POLYMORPHISM AND RA 63

significant difference in the frequency of the G alleleand GG genotype between familial RA cases and con-trols (OR for GG genotype 2.96, 95% CI 1.33–6.58, P 50.007 [Table 2]; OR for G phenotype 1.74, 95% CI1.17–2.59, P 5 0.007).

DISCUSSION

The results of this study provide evidence ofassociation between TNFRII and familial RA. This workextends from previous findings of linkage analysis in RAASP families. Two genome-wide scans in RA ASPfamilies have shown linkage to markers mapping to1p36, close to the TNFRII locus (11,12), and linkage forone of these markers has been replicated in a group ofHLA-concordant European RA ASPs (13). The resultsof the association study presented here suggest that atrue RA susceptibility locus may lie within the TNFRIIgene itself or may be in linkage disequilibrium with theexon 6 polymorphism. No evidence of an associationwith an exon 1 TNFRI polymorphism was detected.

The TNFRII association results could be a false-positive finding, but this is unlikely because 1) there wasa significant increase in the OR for both the GGgenotype and the G phenotype in familial RA casescompared with controls, 2) the association has beenreplicated in independent cohorts of familial RA cases,and 3) the controls were found to be in Hardy-Weinbergequilibrium, with allele frequencies that closely reflectthose reported elsewhere (18).

The difference in association of familial RAversus sporadic RA with the TNFRII polymorphism isnot unexpected. When considering a complex diseasethat exhibits genetic heterogeneity, families in whichmore than 1 individual have the disease are likely tohave a greater genetic susceptibility. In effect, multicasefamilies are likely to concentrate genetic susceptibilityfactors, and probands from such families are an idealgroup to study since there will be a greater chance ofdetecting even weak genetic effects. Differences be-tween familial and nonfamilial RA have been reportedpreviously. Sanders et al found that the frequency ofHLA–DR4 was significantly higher in familial RA casescompared with RA probands without familial RA (4),and this finding has recently been replicated by anothergroup (5).

It may be speculated that polymorphism withinthe TNFR genes could alter binding of ligands such asTNFa or cleavage enzymes, thereby leading to an inap-propriate inflammatory response due to excessive circu-lating TNFa, and hence contributing to RA susceptibil-

ity. One might therefore expect an association with bothTNFRI and TNFRII. However, no difference in allele orgenotype frequencies between RA cases and controlswas detected in studies using an A/G polymorphism inexon 1 of the TNFRI gene. The lack of associationbetween TNFRI and RA may be a false-negative result.The polymorphism examined was nonfunctional, but itwould be expected to be in linkage disequilibrium withany nearby functional polymorphism. Results reportedby Kruglyak suggest that such an assumption may notalways be valid (19). However, another study has failedto demonstrate an association with 2 TNFRI SNPs in apopulation of Dutch RA patients (20). The difference inassociation of TNFRI with RA versus that of TNFRIIwith RA may therefore be real. The receptors areexpressed on different cells, and several lines of evidencesuggest that they may have differing functions in addi-tion to the common function of binding and neutralizingfree TNFa (21).

The T/G SNP in exon 6 of the TNFRII generesults in an amino acid substitution (methionine toarginine), but the effect of this substitution on thefunction of the TNFRII gene is not known. The poly-morphic site lies in the membrane-proximal extracellulardomain of the TNFRII protein and may therefore affectbinding of ligands.

We have found evidence of an association be-tween familial RA and TNFRII, but we have not dem-onstrated an association between RA and TNFRI. Fur-ther work will be needed to attempt to replicate thesefindings in other populations. Interestingly, a recentstudy in a Japanese population failed to demonstrate anassociation between the exon 6 TNFRII polymorphismand RA (22). Functional studies to define the effect ofthe methionine-to-arginine amino acid substitution inthe TNFRII protein are also needed. Information on theeffect of genotype upon response to treatment withsTNFRII would also be of interest since RA does notrespond to this expensive therapy in all patients.

REFERENCES

1. Silman AJ, MacGregor AJ, Thomson W, Holligan S, Carthy D,Farhan A, et al. Twin concordance rates for rheumatoid arthritis:results from a nationwide study. Br J Rheumatol 1993;32:903–7.

2. MacGregor AJ, Snieder H, Rigby AS, Koskenvuo M, Kaprio J,Aho K, et al. Characterizing the quantitative genetic contributionto rheumatoid arthritis using data from twins. Arthritis Rheum2000;43:30–7.

3. Deighton CM, Walker DJ, Griffiths ID, Roberts DF. The contri-bution of HLA to rheumatoid arthritis. Clin Genet 1989;36:178–82.

4. Sanders PA, Grennan DM, Dyer PA, Thomson W, deLange GG.

64 BARTON ET AL

A comparison of clinical and immunogenetic features in familialand sporadic rheumatoid arthritis. J Rheumatol 1987;14:718–22.

5. Laivoranta-Nyman S, Mottonen T, Luukkainen R, Hakala M,Yli-Kerttula U, Hannonen P, et al. Immunogenetic differencesbetween patients with familial and non-familial rheumatoid arthri-tis. Ann Rheum Dis 2000;59:173–7.

6. Mori L, Iselin S, de Libero G, Lesslauer W. Attenuation ofcollagen-induced arthritis in 55kDa TNF receptor type 1(TNFR1)-IgG1-treated and TNFR1-deficient mice. J Immunol1996;157:3178–82.

7. Le CH, Nicolson G, Morales A, Sewell KL. Suppression ofcollagen-induced arthritis through adenovirus-mediated transferof a modified tumor necrosis factor a receptor gene. ArthritisRheum 1997;40:1662–9.

8. Moreland LW, Schiff MH, Baumgartner SW, Tindall EA, Fleisch-mann RM, Bulpitt KJ, et al. Etanercept therapy in rheumatoidarthritis: a randomized, controlled trial. Ann Intern Med 1999;130:478–86.

9. Lorentzen JC, Glaser A, Jacobsson L, Galli J, Fakhrai-rad H,Klareskog L, et al. Identification of rat susceptibility loci foradjuvant-oil-induced arthritis. Proc Natl Acad Sci U S A 1998;95:6383–7.

10. McIndoe RA, Bohlman B, Chi E, Schuster E, Lindhardt M, HoodL. Localization of non-MHC collagen-induced arthritis suscepti-bility loci in DBA/1j mice. Proc Natl Acad Sci U S A 1999;96:2210–4.

11. Cornelis F, Faure S, Martinez M, Prud‘homme JF, Fritz P, Dib C,et al. New susceptibility locus for rheumatoid arthritis suggested bya genome-wide linkage study. Proc Natl Acad Sci U S A 1998;95:10746–50.

12. Shiozawa S, Hayashi S, Tsukamoto Y, Goko H, Kawasaki H,Wada T, et al. Identification of the gene loci that predispose torheumatoid arthritis. Int Immunol 1998;10:1891–5.

13. Cornelis F, for ECRAF, the European Consortium on RA Fam-

ilies. New susceptibility locus on chromosome 1 for rheumatoidarthritis [abstract]. Arthritis Rheum 1998;41 Suppl 9:S242.

14. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS, et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988;31:315–24.

15. MacGregor AJ, Bamber S, Silman AJ. A comparison of theperformance of different methods of disease classification forrheumatoid arthritis: results from an analysis from a nationwidetwin study. J Rheumatol 1994;21:1420–6.

16. Pitts SA, Olomolaiye OO, Elson CJ, Westacoll CI, Bidwell JL. AnMspA1 I polymorphism in exon 1 of the human TNF receptor type1 (p55) gene. Eur J Immunogenet 1998;25:269–70.

17. Al-Ansari AS, Ollier WE, Villarreal J, Ordi J, Teh LS, Hajeer AH.Tumor necrosis factor receptor II (TNFRII) exon 6 polymorphismin systemic lupus erythematosus. Tissue Antigens 2000;55:97–9.

18. Pantelidis P, Lympany PA, Foley PJ, Fanning GC, Welsh KI, duBois RM, et al. Polymorphic analysis of the high-affinity tumournecrosis factor receptor 2. Tissue Antigens 1999;54:585–91.

19. Kruglyak L. Prospects for whole-genome linkage disequilibriummapping of common disease genes. Nat Genet 1999;22:139–44.

20. Kaijzel EL, Bakker AM, van Krugten MV, Huizinga TWJ, Breed-veld FC, Verweij CL. Polymorphisms of tumor necrosis factorreceptor I and II in patients with rheumatoid arthritis. ArthritisRheum 1999;42 Suppl 9:S194.

21. Santee SM, Owen-Schaub LB. Human tumor necrosis factorreceptor p75/80 (CD120b) gene structure and promoter charac-terization. J Biol Chem 1996;271:21151–9.

22. Shibue T, Tsuchiya N, Komata T, Matsushita M, Shiota M, OhashiJ, et al. Tumor necrosis factor a 59-flanking region, tumor necrosisfactor receptor II, and HLA–DRB1 polymorphisms in Japanesepatients with rheumatoid arthritis. Arthritis Rheum 2000;43:753–7.

TNFRII POLYMORPHISM AND RA 65