Macrophage migration inhibitory factor - Nature

Genes and Immunity (2003) 4, 487–491 & 2003 Nature Publishing Group All rights reserved 1466-4879/03 $25.00 www.nature.com/gene

Macrophage migration inhibitory factor (MIF) gene polymorphism is associated with susceptibility to but not severity of inflammatory polyarthritis A Barton1, R Lamb1, D Symmons1, A Silman1, W Thomson1, J Worthington1 and R Donn1 Arthritis Rheumatism Campaign Epidemiology Research Unit, University of Manchester, Manchester, UK

1

The aim of the study was to investigate whether polymorphisms of macrophage migration inhibitory factor (MIF) determine susceptibility to or severity of inflammatory polyarthritis (IP). Genotypes for a single-nucleotide polymorphism (MIF-173*G/C) and a tetranucleotide (CATT)n repeat mapping to the promoter region of the MIF gene were compared between UK Caucasian IP cases (n ¼ 438) and controls (n ¼ 343). Both polymorphisms were also investigated for association with features of disease activity and severity at baseline and by 5 years. The MIF-173*C allele (OR 1.7, 95% CI 1.3–2.4, P ¼ 1.8 104) and the CATT7 allele (OR 1.5, 95% CI 1.0–2.1, P ¼ 0.02) were found to be associated with increased susceptibility to IP. Furthermore, presence of the haplotype containing both associated polymorphisms was associated with a three-fold increase risk of developing IP. No association with disease severity or activity either at baseline or by 5 years was detected for either of the promoter polymorphisms studied. In conclusion, MIF is a susceptibility gene for the development of IP. The same alleles previously reported to be associated with susceptibility to juvenile idiopathic arthritis account for the increased risk. The promoter polymorphisms of MIF, investigated in this study, do not influence the severity of disease outcome by 5 years. Genes and Immunity (2003) 4, 487–491. doi:10.1038/sj.gene.6364014 Keywords: MIF; inflammatory polyarthritis; rheumatoid arthritis; genetics; erosions

Introduction Twin studies suggest that there is a substantial genetic contribution to rheumatoid arthritis (RA) susceptibility, and family studies have shown that first-degree relatives of RA probands are at increased risk1 of disease development. However, the increased risk in probands recruited from a community-based setting is modest compared with hospital-recruited patients.2 As hospitalbased patients are likely to have more severe disease, this suggests that genetic factors may play a greater role in determining disease severity and outcome. Previous studies have shown that the combinations of clinical factors alone are only moderately predictive of outcome and, therefore, the role of genetic factors is increasingly being investigated.3 The identification of genetic factors that would allow the prediction of which patients with inflammatory polyarthritis (IP), at presentation, are likely to suffer a severe disease course would be a major clinical advance as it would allow those patients to be selected to receive early, aggressive treatment. Macrophage inhibitory factor (MIF) is a ubiquitously expressed protein that has proinflammatory, hormonal and enzymatic activities (reviewed in Lolis4). It anatagoCorrespondence: Dr A Barton, ARC-EU, Stopford Building, University of Manchester, Manchester M13 9PT, UK. E-mail: [email protected] Funding: Arthritis Research Campaign Received 4 March 2003; revised 26 March 2003; accepted 14 May 2003

nises the effect of glucocorticoids and, therefore, plays a central role in determining the magnitude of the inflammatory response. Raised levels of MIF expression have been reported in RA serum and synovial fluid, and anti-MIF antibodies have been reported to delay the onset and severity of animal models of inflammatory arthritis.5–7 Two functional promoter polymorphisms of the MIF gene have been described. Donn et al8 identified a singlenucleotide polymorphism (SNP) in the 5’ region of the MIF gene (MIF-173*G/C). Functional studies have shown, both in vivo and in vitro, that the mutant allele (MIF-173*C) is associated with increased MIF protein production.9 Baugh et al10 have reported that a tetranucleotide (CATT)n repeat, mapping upstream of the MIF-173*G/C SNP, is also associated with functional regulation of MIF expression. The allele with five repeats of the tetranucleotide (CATT5) was found to be associated with reduced MIF promoter activity. Furthermore, homozygosity for this allele was associated with less severe RA. In that study, the RA patients all had longstanding disease and were recruited from a hospital setting. As such patients are more likely to have severe disease overall, which can hamper separation of susceptibility from severity factors. Furthermore, these patients may not be representative of those presenting to primary-care physicians with early IP when knowledge of genetic predictors of outcome would be most useful. The aims of this study, therefore, were to examine the role of the MIF-173*G/C SNP and CATT tetranucleotide repeat in predicting either disease susceptibility and/or

MIF promoter polymorphisms and association with IP A Barton et al

488

disease severity (by 5 years) using a population primarycare inception cohort of subjects with new onset IP followed prospectively.

Results Patients Clinical characteristics of the 438 IP cohort are shown in Table 1. In all, 65% of patients were female. A subset (n ¼ 289) of the more severe patients had X-rays 1 year after presentation, of whom 97 (33%) were erosive at that time. All patients had X-rays at 5 years at which point 42.4% were erosive. Among those who were erosive at 5 years, the total Larsen score ranged from 2 to 138. Association with susceptibility MIF-173*G/C SNP. No deviation from Hardy–Weinberg equilibrium was seen in the control panel (P ¼ 0.75). Significant evidence for association of MIF and IP was found when the MIF-173*G/C genotypes were compared between the IP patients and the controls (Table 2). Specifically, the MIF-173*C allele was associated with an increased susceptibility to IP (OR C allele 1.73, 95% CI 1.3-2.4, P ¼ 1.8 104). The association was maintained in both shared epitope positive (n ¼ 245, P ¼ 2.6 104) and negative (n ¼ 161, P ¼ 4 103) IP patients. CATT tetranucleotide repeat. Four alleles of 5-, 6-, 7- and 8-length repeats were detected with frequencies very

Table 1 Characteristics of the cohort with IP Patient characteristics Age (years): mean (s.d.) Gender (female): n (%) RF positive - baseline n (%) - 5 years n (%) RA criteria positive - baseline: n (%) - 5 years: n (%) Copies of shared epitope - 0: n (%) - 1: n (%) - 2: n (%) Erosive at 5 years: n (%) Median Larsen score at 5 years in patients with erosions (IQR)

55.4 298 113 166 197 343 176 199 60 194 25

IQR=interquartile range.

(14.4) (65) (25.8) (36.4) (45.0) (74.9) (40.5) (45.7) (13.8) (42.4) (4–30)

close to those reported previously in a control population (allele frequencies for 5-, 6-, 7- and 8-length repeats in the control population ¼ 25.2, 65.8, 8.7 and 0.3%, respectively).10 No deviation from Hardy–Weinberg expectations was observed in the control population (P ¼ 0.55). Presence of the 7-length tetranucleotide (CATT7) repeat was associated with susceptibility to IP (Table 2) and this association was present in both shared epitope positive (n ¼ 243; P ¼ 0.04) and negative (n ¼ 161; P ¼ 0.03) IP patients. Haplotype analysis. Recent work by Alourfi et al11 has shown an interactive effect of the MIF-173*C and CATT alleles. Frequencies of the tetranucleotide repeat and MIF-173*G/C haplotypes (estimated using the EHPLUS program) are shown in Table 3. Neither in this nor in a previously reported cohort has the haplotype CATT5MIF-173*C been observed.9 In order to examine any possible contribution of the CATT5 allele with disease susceptibility, the CATT5-MIF-173*G haplotype was studied, but no association was found (Table 3). The CATT7-MIF-173*C haplotype occurred approximately three times more commonly in cases than controls (Table 3). Association with severity MIF-173*G/C SNP. No association of the MIF-173*C allele with the presence of erosions or Larsen score at 1 or 5 years; or with SJC, TJC, RF or HAQ score at baseline or 5 years was detected. Adjusting for possible treatment effects did not affect the findings (Table 4). CATT tetranucleotide repeat. Presence of the CATT5, and particularly homozygosity for this allele, has been reported previously to be associated with less-severe RA using a composite score to assess RA severity.10 No association of the CATT5 allele with the absence of erosions, Larsen score, SJC, TJC, seronegativity or HAQ score was detected in this inception cohort of patients with IP (Table 4). There was a trend for the presence of the CATT5 allele to be associated with a lower TJC at baseline (P ¼ 0.06), but not at 5 years. Indeed, the presence of this allele showed a trend towards a higher Larsen score at 5 years (P ¼ 0.06). No association with homozygosity for the CATT5 allele was found (data not shown). As the CATT7 allele was found to be associated with susceptibility, association with measures of disease

Table 2 Genotype frequencies for two polymorphisms mapping to the MIF gene in IP cases and controls Polymorphism

Genotype

MIF-173*G/C

GG GC CC 5,5 6,6 7,7 5,6 5,7 6,7 5,8 6,8

(CATT)n repeat

Genes and Immunity

Cases (%) 280 131 15 20 173 4 126 30 67 1 2

(65.7) (30.8) (3.5) (4.7) (40.9) (0.9) (29.9) (7.1) (15.8) (0.2) (0.5)

Controls (%) 258 73 3 21 150 5 115 15 35 1 1

(77.2) (21.9) (0.9) (6.1) (43.7) (1.5) (33.5) (4.4) (10.2) (0.3) (0.3)

Comparison w2(2df) by genotype (14.5), P=0.001 OR MIF-173*C allele=1.73 (1.3–2.4) P=1.8 104 OR CATT7 allele=1.48 (1.0–2.1) P=0.02

P=0.21 P=0.13 P=0.12 P=0.89 T=1.26; T=1.50; T=1.54; T=0.14; P=0.41 P=0.49 P=0.74 P=0.79 T=0.81; T=0.69; T=0.33; T=0.27;

All analyses have been adjusted for the possible confounding effect of treatment on outcome.

P=0.55 P=0.41 P=0.85 P=0.77 T=0.59; T=0.82; T=0.19; T=0.30; T=0.90; P=0.37 T=1.16; P=0.25 T=0.33; P=0.74 T=0.32; P=0.75

T=1.91; P=0.06 T=1.08; P=0.28 T=1.08; P=0.28 T=0.58; P=0.56

T=0.74; P=0.46 T=1.09; P=0.27 T=0.84; P=0.40 T=0.34; P=0.74 T=0.97; P=0.33 T=1.28; P=0.20 T=0.46; P=0.56 T=0.90; P=0.37

T=0.99; P=0.33 T=1.06; P=0.29

0.87(0.7–1.2); P=0.36 1.00(0.4–1.4); P=0.97 1.04(0.8–1.4); P=0.81 0.90(0.7–1.2); P=0.49 0.76(0.5–1.1); P=0.12 0.95(0.7–1.4); P=0.80

T=1.31; P=0.19 T=0.15; P=0.88 T=0.51; P=0.61 T=1.03; P=0.30 T=1.27; P=0.21 T=1.54; P=0.06

1.09(0.7–1.6); P=0.66 0.90(0.7–1.2); P=0.46

1.04(0.71.6); P=0.85 0.66(0.4–1.1); P=0. 10

In the most extensive analysis of the role of the MIF gene in IP to date, we have shown that polymorphisms within the gene are associated with susceptibility to, but not severity of, disease. There are several lines of evidence to support the MIF gene as a strong candidate arthritis susceptibility gene. Firstly, in experimental animal models of inflammatory arthritis, neutralisation of MIF markedly suppresses the inflammatory response.5–7 Secondly, MIF protein within the synovial fluid of rheumatoid arthritis (RA) patients has been found to be elevated 5- to 10-fold higher than that seen in osteoarthritis patients and normal volunteers.12 Thirdly, MIF is known to induce macrophage tumour necrosis factor a (TNFa) release and nitric oxide production, as well as being involved in T-cell activation, all of which play a role in RA pathogenesis. Fourthly, immunostaining of MIF within the RA synovium is strongly correlated with disease activity, as measured by CRP.13 Finally, a previous study has investigated the role of the MIF CATT tetranucleotide repeat polymorphism in RA, and reported that the 5-length repeat (CATT5) was associated with less severe disease.10 In this study, we have investigated the role of polymorphism within the MIF gene and IP. We have found that the MIF-173*C allele and the CATT7 are

1.05(0.7–1.5); P=0.81 1.06(0.7–1.6); P=0.76

Discussion

T=1.00; P=0.32 T=1.09; P=0.28

Haplotype analysis As haplotype analysis showed an association of the CATT7-MIF-173*C haplotype with susceptibility, this haplotype was also investigated with regard to measures of disease severity (Table 4). No association with severity parameters (presence of erosions, Larsen score, SJC, TJC, RF or HAQ score) was found (Table 3). Previous work by Baugh et al has an found association with milder disease in the presence of the CATT5 allele.10 This allele only occurs on a CATT5-MIF-173*G hapolytpe and so this was investigated for association to measures of disease severity and activity. However, no association with the presence of erosions, Larsen score, SJC, TJC, RF or HAQ score was detected (Table 4).

T=0.32; P=0.75 T=1.34; P=0.18

severity was also explored. However, no association with severity by the presence of this allele was detected (Table 4). Analysis by the presence of the 6- and 8-length repeat alleles similarly showed no association with measures of disease severity and adjustment for possible confounding effects of treatment did not alter these findings (data not shown).

1.17(0.8–1.7); P=0.41 1.07(0.8–1.4); P=0.66

Only common haplotypes (>1% frequency) are reported. The OR for presence of the CATT7–MIF-173*C haplotype vs other haplotypes combined is shown.

1.35(0.9–2.1); P=0.18 1.33(0.9–1.9); P=0.11

OR 3.16 (1.7–6.3), P=1 104

1.36(0.8–2.4); P=0.30 1.07(0.7–1.7); P=0.76

(60.2) (24.3) (6.7) (4.1)

0.78(0.5–1.3); P=0.33 0.96(0.7–1.4); P=0.84

206 83 23 14

489

Erosions (OR (95% CI); P-value) by 1 year (n=289) 5 years (n=438) Larsen score (Mann–Whitney) by 1 year (n=289) 5 years (n=438) RF positive (OR (95% CI); P-value) by Baseline (n=113) 5 years (n=166) SJC (Mann–Whitney) at Baseline (n=406) 5 years (n=176) TJC (Mann–Whitney) at Baseline (n=411) 5 years (n=194) HAQ (Mann–Whitney) at Baseline (n=378) 5 years (n=363)

(58.1) (22.7) (6.0) (12.2)

CATT–MIF-173*G

246 96 25 52

Comparison

CATT7–MIF-173*C

CATT6-MIF–173*G CATT5-MIF–173*G CATT6-MIF–173*C CATT7-MIF–173*C

Controls (%)

CATT5

Cases (%)

CATT7

Haplotype

MIF-173*C

Frequencies of estimated haplotypes in cases with IP compared with controls (estimated using EHPLUS)

activity

Table 3

Table 4 Association of the MIF-173*C allele, the CATT7 allele, the CATT5 allele, the CATT7–MIF-173*C haplotype and the CATT57–MIF-173*G haplotype with measures of disease severity and


Genes and Immunity


490

associated with an approximately two-fold increased risk of IP susceptibility. The association appears independent of shared epitope status as it was maintained in both SE positive and negative IP patients for both associated alleles. These two alleles were found to be in strong LD in the IP cases studied (P ¼ 1.4 106), and the presence of both associated polymorphisms on the CATT7–MIF– 173*C haplotype confers a three-fold increased risk of susceptibility to IP. It is particularly interesting that these are the same alleles of MIF that have been shown to be both associated and linked with juvenile idiopathic arthritis (JIA), the most common form of chronic arthritis that presents before the age of 16 years.14 Furthermore, the CATT7 and the MIF-173*C have each been found to contribute to JIA susceptibility independently. It will now be necessary to replicate the association of MIF with IP in other cohorts. While an association between MIF and IP susceptibility has been seen, no association with MIF polymorphism and severity was observed. This was the case whether measures of disease severity or activity at either baseline or by 5 years were considered. One previous study has investigated the role of the CATT tetranucleotide repeat polymorphism in determining disease severity in RA, and reported that the 5-length repeat (CATT5) was associated with less severe disease.10 We have not replicated this finding. Studying prevalent RA cases recruited from a hospital setting may not be representative of patients with early IP presenting in the community. It is in this group of IP patients that the identification of genes that determine prognosis is likely to have greatest impact as it will allow better targeting of aggressive therapies. Therefore, we investigated an inception cohort of community-based IP patients. Furthermore, we did not restrict analysis to RA patients as the ACR criteria perform poorly in early disease.15 However, it should be noted that the majority of these patients subsequently satisfied ACR criteria for RA, and restricting analysis to patients satisfying ACR criteria for RA by 5 years did not affect the results (data not shown). In patients with early IP, no association of CATT5 with a number of measures of disease severity (presence of erosions, Larsen score), activity (tender and swollen joint counts) and disability (HAQ score) was detected. Specifically, when the MIF alleles that were associated with an increased risk of IP susceptibility were examined, no association with severity was seen. In support of the hypothesis that MIF polymorphism may be associated with susceptibility rather than severity, the previously reported association of MIF with JIA is known to persist across all clinical subgroups of JIA and is not restricted to more severe subtypes.14 Our data show an association of MIF gene polymorphisms and susceptibility to IP, but the mechanism behind this is currently unknown. It is possible that the MIF polymorphisms act in concert with other susceptibility factors to lower the threshold at which an arthritic response is triggered. Alternatively, MIF may be involved in persistence, but not magnitude of an inflammatory response, thereby explaining the lack of association with parameters of severity. Further investigation of MIF polymorphisms in cohorts of IP patients with documentation of remission/persistence of disease will be required to decipher these possible effects.

Genes and Immunity

In summary, we have found MIF polymorphisms to be associated with IP susceptibility, but not with disease severity. Several investigators have suggested the use of anti-MIF therapy.16–18 Our data support such a use in the potential treatment of patients with IP.

Patients and methods Study design An association study was performed to investigate whether the presence of the MIF polymorphisms played a role in determining disease susceptibility and disease outcome in a large inception cohort of patients with IP (of which RA is a major subset). Frequencies of the polymorphisms and estimated haplotypes were compared in IP patients and controls. Association of the polymorphisms with the presence or absence of radiological erosions, the Larsen score and rheumatoid factor (RF), as a marker of future severity, was assessed. In addition, association with markers of disease activity and disability at baseline and at 5 years was investigated. Patients DNA samples were available from 438 patients from a primary-care inception cohort of IP recruited from the Norfolk Arthritis Register (NOAR). Details of the case ascertainment procedure have been described in more detail previously.19 Briefly, all cases with IP (defined as swelling of two or more joints lasting for four or more weeks) within the region formerly known as the Norwich Health Authority are notified and assessed by a research nurse using a standard questionnaire and examination. Baseline clinical data are recorded and blood taken for RF measurement and DNA extraction. Patients are reviewed annually and at each assessment, scored as to whether ACR classification criteria for RA20 are satisfied, using a cumulative evaluation as described previously.21 Patients who receive an alternative clinical diagnosis at any stage of follow-up, other than psoriatic arthritis, RA or postviral arthritis, are excluded. All patients had radiographs of the hands and feet 5 years after diagnosis. Details regarding radiographic policy and XR scoring are described elsewhere.22 Briefly, X-rays of the hands and feet are scored by two observers using the Larsen method.23 A third observer arbitrates in cases of disagreement. A subset with more severe disease had X-rays at 1 year.24 Controls DNA from unrelated Caucasian controls, with no history of IP, was available (n ¼ 343). These individuals were recruited from general practice or were blood donors and comprised the same individuals as were genotyped as part of a study to investigate association with JIA.9 The controls had all previously been genotyped for the MIF173*G/C polymorphism, but only 92 had been genotyped for the CATT tetranucleotide repeat. Genotyping Genotyping for the MIF-173*G/C SNP and the CATT tetranucleotide repeat was performed as described previously.9


Statistical analysis Frequency of MIF-173*G/C and the (CATT)n tetranucleotide alleles were compared between IP cases and controls. In order to determine whether the polymorphisms contributed to disease severity frequencies were also compared between firstly, IP cases who had developed erosions by 5 years with those that had not and, secondly those who were RF positive at 5 years with those who remained seronegative. Odds ratios and 95% confidence intervals were calculated. The two-sample (Mann–Whitney) test was used to test for an influence of the carriage of the alleles on the following disease features: the number of tender (TJC) and swollen joints (SJC) at baseline and 5 years, the health assessment questionnaire (HAQ) score at baseline and 5 years and Larsen score at 1 and 5 years. As patients with more severe disease are more likely to receive treatment, treatment effects may mask real associations. Therefore, disease-modifying antirheumatic therapy (DMARD) and steroid treatment by 5 years was used as a covariate in all analyses in order to adjust for potential confounding. Linkage disequilibrium between the two polymorphisms was investigated using EHPLUS.25 In order to investigate haplotype effects of combinations of (CATT)n and MIF-173*G/C alleles, alleles of the (CATT)n repeat were first regrouped as a biallelic variable for each of the alleles vs the rest. SNPhap (David Clayton, Cambridge Institute for Medical Research, Cambridge, UK) was used to assign haplotypes of the specific (CATT)n repeat alleles and the MIF-173*G/ C alleles. Haplotype frequencies were compared using EHPLUS25 for the outcome variables listed above.

References 1 MacGregor AJ, Snieder H, Rigby AS et al. Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins. Arthritis Rheum 2000; 43: 30–37. 2 Thomson W, Harrison B, Oilier B et al. Quantifying the exact role of HLA–DRB1 alleles in susceptibility to inflammatory polyarthritis: results from a large, population-based study. Arthritis Rheum 1999; 42: 757–762. 3 Brennan P, Harrison B, Barrett E et al. A simple algorithm to predict the development of radiological erosions in patients with early rheumatoid arthritis: prospective cohort study. BMJ 1996; 313: 471–476. 4 Lolis E. Glucocorticoid counter regulation: macrophage migration inhibitory factor as a target for drug discovery. Curr Opin Pharmacol 2001; 1: 662–668. 5 Mikulowska A, Metz CN, Bucala R, Holmdahl R. Macrophage migration inhibitory factor is involved in the pathogenesis of collagen type II-induced arthritis in mice. J Immunol 1997; 158: 5514–5517. 6 Santos L, Hall P, Metz C, Bucala R, Morand EF. Role of macrophage migration inhibitory factor (MIF) in murine antigen-induced arthritis: interaction with glucocorticoids. Clin Exp Immunol 2001; 123: 309–314. 7 Leech M, Metz C, Santos L et al. Involvement of macrophage migration inhibitory factor in the evolution of rat adjuvant arthritis. Arthritis Rheum 1998; 41: 910–917. 8 Donn RP, Shelley E, Oilier WE, Thomson W. A novel 50 flanking region polymorphism of macrophage migration inhibitory factor is associated with systemic-onset juvenile idiopathic arthritis. Arthritis Rheum 2001; 44: 1782–1785.

491 9 Donn R, Alourfi Z, De Benedetti F et al. Mutation screening of the macrophage migration inhibitory factor gene: positive association of a functional polymorphism of macrophage migration inhibitory factor with juvenile idiopathic arthritis. Arthritis Rheum 2002; 46: 2402–2409. 10 Baugh JA, Chitnis S, Donnelly SC et al. A functional promoter polymorphism in the macrophage migration inhibitory factor (MIF) gene associated with disease severity in rheumatoid arthritis. Genes Immun 2002; 3: 170–176. 11 Alourfi Z, Donn R, Ray D. Macrophage migration inhibitory factor (MIF) expression is differentially regulated by a functional interaction between two promoter polymorphisms. Rheumatology (Oxford), 2003; 42 (suppl): 6. 12 Onodera S, Nishihira J, Iwabuchi K et al. Macrophage migration inhibitory factor up-regulates matrix metalloproteinase-9 and -13 in rat osteoblasts. Relevance to intracellular signaling pathways. J Biol Chem 2002; 277: 7865–7874. 13 Morand EF, Leech M, Weedon H, Metz C, Bucala R, Smith MD. Macrophage migration inhibitory factor in rheumatoid arthritis: clinical correlations. Rheumatology (Oxford) 2002; 41: 558–562. 14 Donn RP, Zeggini E, Lamb et al. Linkage and association of macrophage migration inhibitory factor (MIF) with juvenile idiopathic arthritis. Rheumatology (Oxford), 2003; 42 (Suppl.) 82. 15 Dugowson CE, Nelson JL, Koepsell TD. Evaluation of the 1987 revised criteria for rheumatoid arthritis in a cohort of newly diagnosed female patients. Arthritis Rheum. 1990; 33: 1042–1046. 16 Leech M, Metz C, Bucala R, Morand EF. Regulation of macrophage migration inhibitory factor by endogenous glucocorticoids in rat adjuvant-induced arthritis. Arthritis Rheum 2000; 43: 827–833. 17 de Jong YP, Abadia-Molina AC, Satoskar AR et al. Development of chronic colitis is dependent on the cytokine MIF. Nat Immunol 2001; 2: 1061–1066. 18 Denkinger CM, Denkinger M, Kort JJ, Metz C, Forsthuber TG. In vivo blockade of macrophage migration inhibitory factor ameliorates acute experimental autoimmune encephalomyelitis by impairing the homing of encephalitogenic T cells to the central nervous system. J Immunol 2003; 170: 1274–1282. 19 Symmons DP, Barrett EM, Bankhead CR, Scott DG, Silman AJ. The incidence of rheumatoid arthritis in the United Kingdom: results from the Norfolk Arthritis Register. Br J Rheumatol 1994; 33: 735–739. 20 Arnett FC, Edworthy SM, Bloch DA et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988; 31: 315–324. 21 MacGregor AJ, Bamder S, Silman AJ. A comparison of the performance of different methods of disease classification for rheumatoid arthritis. Results from an analysis from a nationwide twin study. J Rheumatol 1994; 21: 1420–1426. 22 Bukhari M, Harrison B, Lunt M, Scott DG, Symmons DP, Silman AJ. Time to first occurrence of erosions in inflammatory polyarthritis: results from a prospective communitybased study. Arthritis Rheum 2001; 44: 1248–1253. 23 Larsen A, Dale K, Eek M. Radiographic evaluation of rheumatoid arthritis and related conditions by standard reference films. Ada Radiol Diagn (Stockh) 1977; 18: 481–491. 24 Bukhari M, Lunt M, Harrison BJ, Scott DG, Symmons DP, Silman AJ. Rheumatoid factor is the major predictor of increasing severity of radiographic erosions in rheumatoid arthritis: results from the Norfolk Arthritis Register Study, a large inception cohort. Arthritis Rheum 2002; 46: 906–912. 25 Curtis D, Sham PC. Model-free linkage analysis using likelihoods. Am J Hum Genet 1995; 57: 703–716.

Genes and Immunity