FULL PAPER IFNG polymorphisms are associated with gender differences in susceptibility to multiple sclerosis

Transcription

1 (2005) 6, 53 6 & 2005 Nature Publishing Group All rights reserved /05 $ FULL PAPER IFNG polymorphisms are associated with gender differences in susceptibility to multiple sclerosis OH Kantarci,9, A Goris 2,9, DD Hebrink, S Heggarty 3, S Cunningham 3, I Alloza 3, EJ Atkinson 4, M de Andrade 4, CT McMurray 5, CA Graham 6, SA Hawkins 7, A Billiau 2, B Dubois 8, BG Weinshenker and K Vandenbroeck 3 Department of Neurology, Mayo Clinic and Foundation, Rochester, MN, USA; 2 Rega Institute for Medical Research, University of Leuven, Belgium; 3 Applied Genomics Group, School of Pharmacy, Queen s University of Belfast, Belfast, UK; 4 Department of Health Sciences Research, Mayo Clinic and Foundation, Rochester, MN, USA; 5 Departments of Pharmacology, Biochemistry and Molecular Biology and Molecular Neuroscience Program, Mayo Clinic and Foundation, Rochester, MN, USA; 6 Northern Ireland Regional Molecular Genetics Laboratory, Belfast City Hospital Trust, UK; 7 Department of Neurology, Royal Victoria Hospital, Belfast, UK; 8 Department of Neurology, University of Leuven, Belgium Interferon-gamma (IFNg) treatment is deleterious in multiple sclerosis (MS). MS occurs twice as frequently in women as in men. IFNg expression varies by gender. We studied a population-based sample of US MS patients and ethnicity-matched controls and independent Northern Irish and Belgian hospital-based patients and controls for association with MS, stratified by gender, of an intron microsatellite [I(76)*CA n ], a single nucleotide polymorphism 3 0 of IFNG [3 0 (325)*G-A] and three flanking microsatellite markers spanning a 8 kb region around IFNG. Men carriers of the 3 0 (325)*A allele have increased susceptibility to MS compared to noncarriers in the USA (P ¼ 0.044; OR: 2.58, 95% CI: ) and Northern Ireland (P ¼ 0.09; OR: 2.37, 95% CI:.0 5.3). There is a nonsignificant trend in the same direction in Belgian men (P ¼ 0.299; OR:.50, 95% CI: ). Men carriers of I(76)*CA 3, which is in strong linkage disequilibrium with the 3 0 (325)*A, have increased susceptibility (P ¼ 0.050; OR: 2.22, 95% CI: ), while men carriers of I(76)*CA 2 have decreased susceptibility (P ¼ 0.022; OR: 0.46, 95% CI: ) to MS in the USA. Similar associations were reported in Sardinia between the I(76)*CA 2 allele and reduced risk of MS in men. Flanking markers were not associated with MS susceptibility. Polymorphisms of IFNG may contribute to differences in susceptibility to MS between men and women. (2005) 6, doi:0.038/sj.gene Published online 27 January 2005 Keywords: multiple sclerosis; gender; interferon gamma (IFNG); polymorphisms; association Introduction Multiple sclerosis (MS) is a complex disease with genetic and environmental influences on susceptibility,2 and possibly on the course of disease. 3 Most susceptibility loci, except perhaps the major histocompatibility complex, contribute only a small effect. 4 8 Association studies are powerful methods to uncover relatively small effects of allelic variants on complex traits. 9 Interferon-gamma (IFNg) stimulates Th- clonal expansion and inhibits Th-2 expansion. The balance between Th- and Th-2 CD4 cells is relevant to autoimmunity and to MS. 0, IFNg expression is increased in experimental allergic encephalomyelitis (EAE) 2 and parallels disease severity. 3 However, Correspondence: Dr BG Weinshenker, Department of Neurology, Mayo Clinic and Foundation, 200 First Street, SW, Rochester, MN 55905, USA. weinb@mayo.edu 9 These two authors contributed equally to this work. Received 9 August 2004; revised November 2004; accepted November 2004; published online 27 January 2005 it is unclear whether the increased IFNg expression is a deleterious or a disease limiting response in EAE, as IFNg receptor knockout mice are more susceptible to EAE than wild-type mice. 4,5 Transgenic mice overexpressing IFNg in the central nervous system under the control of an oligodendrocyte-specific promoter develop extensive primary demyelination compared to wild-type mice. 6,7 IFNg may exert deleterious effects directly on myelinating cells and also through activation of macrophages and microglia. 8 IFNg also induces dendritic retraction and inhibits synapse formation. 9 IFNg expression increases in the 2 weeks preceding attacks of MS. 20,2 Low levels of IFNg expression by lymphocytes at the initiation of treatment with IFNb predict a favorable response to treatment. 22 Treatment with exogenous IFNg is deleterious to patients with MS. 23 However, as is the case in EAE, whether endogenous IFNg expression represents a deleterious or a disease limiting effect in MS is unknown. Genetic variants that affect expression or function of IFNg might favorably or unfavorably influence the susceptibility to and course and severity of MS.

2 54 Men develop MS one-third as often as women. While many factors have been invoked to explain the difference in susceptibility to MS between men and women, the reasons are not understood. 24 Whether there are genomic differences that account for reduced susceptibility in men is unknown. Men also have a higher frequency of primary progressive and lower frequency of bout-onset (relapsing remitting and secondary progressive) disease course compared to women. 25 Pelfrey et al 26 demonstrated that PLP-induced IFNg secretion was higher in women than men in both patients with MS and controls. The percentage of IFNg-producing CD3 cells correlates with disease severity in women but not in men. 27 The IFNg gene (IFNG) is located on chromosome 2q and is comprised of four exons. A rare polymorphism in the promoter, 29 a dinucleotide repeat polymorphism in intron [I(76)*CA n ] 30 and a polymorphism in the 3 0 untranslated (UTR) region 3 were previously reported. Recently, multiple promoter polymorphisms, a T insertion in intron, multiple polymorphisms in intron 3 and exon 4 as well as in 3 0 UTR have been reported Some of these polymorphisms are predominately described in African or African- American populations Several groups have studied the association of the intron dinucleotide polymorphism with MS Certain authors of this study (KV, AG) previously reported excess parental transmission of the I(76)*CA 3 allele to Sardinian men with MS who did not carry the HLA DR 3 and 4 alleles. 39 They have recently extended the study to eight additional microsatellite markers within the region. The association in the region was confined to the 8-kilobase interval limited by the centromeric marker D2S33 and the telomeric marker D2S25, an interval in which IL26 and IFNG are the only genes. 42 In a preliminary study, we had independently screened the promoter region, exons, splice sites and 3 0 region of IFNG for polymorphisms in a population-based sample in Olmsted County, MN. We showed that the G allele of a novel single-nucleotide polymorphism (SNP) [3 0 (325)*G-A], which was just downstream of the 3 0 UTR and in significant linkage disequilibrium (LD) with the I(76)*CA 2 allele, was under-represented in men compared to women with MS. 43 In the present study, we sought to confirm the association between IFNG polymorphisms and susceptibility to MS in men, using a larger population-based sample of patients from Olmsted County, MN, and two other populations from Northern Ireland and Belgium. Results The demographics of the hospital-based patients with MS from Northern Ireland and Belgium were similar to the population-based sample of patients from USA (Table ). There were differences in the ratio of women to men in the patient populations (2.5 in MN,.9 in Northern Ireland and.5 in Belgium). The age of onset of MS did not differ between populations and between men and women. The frequency of patients with bout-onset disease was lower in the Northern Irish population (326 of 432, 75.5%) compared to USA (98 of 22, 89.6%) (Po0.00) and Belgium (87 of 208, 89.9%) (Po0.00). Men in all three populations had bout-onset disease less frequently than women. The genetic markers considered in this study are illustrated in Figure. Hardy Weinberg equilibrium (HWE) was evaluated by site in controls for men and women separately. In men, marker D2S33 in the USA population, marker D2S25 in the Belgium population and marker I(76)*CA n in the Northern Irish population deviated from HWE. In women, markers D2S33 and D2S25 in the USA population and marker I(76)*CA n in the Northern Irish population deviated from HWE. The 3 0 (325)*G-A polymorphism did not deviate from HWE in either of the gender groups in any of the populations. Strong LD was present between the three markers D2S250, I(76)*CA n and 3 0 (325)*G-A in cases and controls in all three populations. LD was less strong in the flanking markers D2S33 and D2S25. (Table 2). Considering consecutive two-marker haplotypes, adjusted for intermarker distance, a significant association was found between carriage of haplotypes including I(76)*CA n and 3 0 (325)*G-A and MS in men from USA (Po0.00; Figure 2). Specifically, the CA 2 /G haplotype was associated with decreased susceptibility to MS in USA men (cases 22% vs controls 44%, OR ¼ 0.36, 95% CI: ; Po0.00) while the CA 3 /A haplotype was associated with increased susceptibility to MS in men IFNG IL26 0 kb 5 kb 9 kb 8 kb Cen D2S33 3 (325)*G A I(76)*CA n D2S250 D2S25 Figure Location of markers on chromosome 2. Positions of IFNG and IL26 are shown. Directions of arrows indicate direction of transcription. Tel Table Demographic and clinical aspects of the patient populations stratified by gender Population Total Women Men N % BO Age of onset (mean7s.d.) N % BO Age of onset (mean7s.d.) N % BO Age of onset (mean7s.d.) USA Northern Ireland Belgium Combined %BO ¼ percentage of patients who have bout-onset disease course (relapsing remitting and secondary progressive).

3 Table 2 LD (D 0 ) in the region of IFNG across populations 55 Population Marker D2S (325)*G-A I(76)*CA n D2S250 D2S25 USA D2S (325)*G-A I(76)*CA n D2S D2S Northern Ireland D2S (325)*G-A I(76)*CA n D2S D2S Belgium D2S (325)*G-A I(76)*CA n D2S D2S D 0 values for cases are presented below the diagonal of complete LD (D 0 ¼ ) and D 0 values for controls are presented above the diagonal in italics. The LD block with D 0 values are shown in bold. combined population individual populations -log0(score.max.p.sim) Women Women USA N.Irish Belgian D2S33 3 (325)G A I(76)CA n D2S250 D2S25 D2S33 3 (325)G A I(76)CA n D2S250 -log0(score.max.p.sim) D2S Men Men USA N.Irish Belgian D2S33 3 (325)G A I(76)CA n D2S250 D2S25 D2S33 3 (325)G A I(76)CA n D2S250 D2S25 Figure 2 Comparison of two-marker haplotypes between cases and controls stratified by gender in individual populations and in the combined cohort is presented showing the marker distance and contribution of individual populations to the combined cohort. (cases 58% vs controls 44%, OR ¼.76, 95% CI:.5 2.7; P ¼ 0.03). The associations were weaker for the flanking haplotypes comprised of D2S33 and 3 0 (325)*G-A (P ¼ 0.02) as well as I(76)*CA n and D2S250 (P ¼ 0.02). Association was not evident in haplotypes including D2S250 and D2S25 (P ¼ 0.596), which contain IL26. No association was found for these haplotypes in women with MS (Figure 2). The associa-

4 56 tion was not evident in the other two populations although a weak association with MS in Northern Irish women was found for the markers D2S250 and D2S25 (P ¼ 0.027; Figure 2). In the combined population, accounting for the three populations independent contributions, the association of the haplotypes defined by the two IFNG markers 3 0 (325)*G-A and I(76)*CA n became marginally significant (P ¼ 0.059; Figure 2) in men. The analysis of association of the common alleles by carrier status for the two IFNG markers for individual populations is shown in Table 3. Breslow Day test for the carrier status of 3 0 (325)*A allele (P ¼ 0.05) and for 3 0 (325)*AA genotype (Po0.0) indicated possible heterogeneity in allele distribution between the three populations, suggesting that the interpretation cannot be solely based on the combined population if the results are not significant in individual populations. The association with susceptibility to MS was strongest in men from USA who are homozygous for 3 0 (325)*A (cases 59% vs controls 28%, Po0.00; OR: 3.74, 95% CI: ). MS susceptibility was weakly associated in carriers of 3 0 (325)*A (cases 9% vs controls 79%, P ¼ 0.044; OR: 2.58, 95% CI: ). In Northern Irish men, an association was seen only in men who are carriers of 3 0 (325)*A (cases 89% vs controls 77%, P ¼ 0.09; OR: 2.37, 95% CI:.0 5.3). The results were not significant for Belgian men although the odds ratio was in the same direction as the results in the USA and Northern Irish populations (cases 8% vs controls 74%, P ¼ 0.299; OR:.50, 95% CI: ). The I(76)*CA 3 allele, which is in strong but incomplete LD with 3 0 (325)*A, tended to be associated with susceptibility to MS in USA men (cases 83% vs controls 69% P ¼ 0.050; OR: 2.22, 95% CI: ; Table 3). Similarly, the I(76)*CA 2, which is in strong LD with 3 0 (325)*G, was negatively associated with susceptibility to MS in men from USA (cases 52% vs controls 70%, P ¼ 0.022; OR: 0.46, 95% CI: ). The marker I(76)*CA n was not associated with MS in the other populations, even after stratifying by gender. When the populations of men were combined and the analysis was adjusted for populations, carriers for 3 0 (325)*A (P ¼ 0.00; OR: 2.02, 95% CI:.3 3.4) and I(76)*CA 3 (P ¼ 0.034; OR:.48, 95% CI: ) had increased susceptibility to MS. If the effect of the alleles was considered protective, homozygotes for the 3 0 (325)*G (P ¼ 0.00; OR: 0.49, 95% CI: ) and I(76)*CA 2 (P ¼ 0.05; OR: 0.58, 95% CI: ) alleles had reduced susceptibility to MS. No association with susceptibility to MS in women in any of the populations was observed, either individually or when analyzed by successive two-marker haplotypes (Po0.05; Table 3). The differences in gender distribution of the carrier status in the combined population were evident only in cases but not controls (Figure 3a). Among the markers tested, the differences in men and women were strongest for the 3 0 (325)*A allele (P ¼ 0.00; OR:.92, 95% CI: ; Figure 3a). There was no difference between men and women for being a homozygote for this allele. The difference between men with MS and male controls stratified by population for being a carrier for the 3 0 (325)*A allele is shown in Figure 3b. The association in the combined population is evident in both the Northern Irish and USA populations. The effect persists when the primary progressive cases are removed and bout-onset cases are analyzed independently (Figure 3b). Discussion In this study, we have shown that being a carrier of the A allele of a 3 0 (325)*G-A SNP just 3 0 to IFNG is associated with increased susceptibility to MS in men. This association was significant in two (ie USA and Northern Irish) populations studied, with a similar but nonsignificant trend for the third population from Belgium. The difference in men vs women arises from cases, while the controls have a similar distribution of alleles regardless of gender. The intragenic marker I(76)*CA n is associated with susceptibility to MS in men in a well-controlled population-based sample of MS patients from Olmsted County, MN, USA. Analysis of three additional markers that span a 8 kb region flanking IFNG, which were studied in a Sardinian population using family-based association Table 3 Carrier status of the common alleles of the intragenic markers stratified by gender Population Marker Allele Women Men MS (%) Control (%) OR 95% CI MS (%) Control (%) OR 95% CI USA 3 (325) G 00 (63.7) 226 (72.2) (40.6) 92 (7.9) 0.27*** A 23 (78.3) 252 (80.5) (90.6) 0 (78.9) 2.58* I(76) CA 2 03 (66.9) 209 (68.) (5.7) 89 (70.) 0.46* CA 3 6 (75.3) 220 (7.7) (83.3) 88 (69.3) 2.22* Northern Ireland 3 (325) G 24 (75.4) 77 (72.6)) (73.8) 73 (75.8) A 25 (75.7) 83 (78.3) (89.0) 75 (77.3) 2.37* I(76) CA 2 24 (75.) 77 (72.0) (73.) 65 (66.3) CA 3 80 (63.2) 74 (69.2) (7.7) 69 (70.4) Belgium 3 (325) G 85 (68.5) 60 (75.9) (8.3) 86 (76.) A 03 (83.) 58 (73.4) (8.3) 84 (74.3) I(76) CA 2 8 (64.8) 58 (7.6) (80.2) 84 (73.7) CA 3 86 (68.8) 53 (65.4) (7.6) 69 (60.5) Fisher s exact P-values: *Po0.05, **Po0.0, ***Po0.00.

5 a Cases Controls D2S250*43 43 D2S250*43 I (76)*CA3 3 I (76)*CA 3 3' (325)*AA 3' (325)*A D2S250*43 43 D2S250*43 I (76)*CA3 3 I (76)*CA 3 3' (325)*AA 3' (325)*A P= P= P= P= P= P= P= P= P= P= P= 0.35 P= b Combined.97 P= Bout-onset MS USA Northern Irish Belgium P= P= P= Combined 2.02 P= 0.00 USA 2.58 P= All MS Northern Irish 2.37 P= 0.09 Belgium.50 P= Figure 3 (a) The odds ratios in men vs women for carrier status and homozygotes for three IFNG markers adjusted for contribution of individual populations to the combined cohort are presented for cases and controls. Confidence intervals (solid horizontal bars), odds ratios (numbers over each bar) and P-values (Cochran Maentel Haenszel) are shown. Data for alleles that are positively associated with MS susceptibility are presented. The scale is logarithmic and the vertical axis represents an odds ratio of one. Odds ratios less than one represent reduced risk. (b) The odds ratios in cases vs controls for the carrier status of 3 0 (325)*A allele in men by population are presented after stratification for bout-onset disease course. Confidence intervals (solid horizontal bars), odds ratios (numbers over each bar) and P-values (Fisher s exact test) are shown. The scale is logarithmic and the vertical axis represents an odds ratio of one. Odds ratios less than one represent reduced risk. When comparing bout-onset cases with controls, all controls were used for the North Irish and Belgian populations. For the Mayo population, bout-onset cases were compared to their matched controls. methods, 42 suggests that the association is likely within or in the immediate vicinity of the IFNG locus. Our results were similar to those of the Sardinian population for which Goris et al 42 also reported reduced susceptibility to MS in men carrying haplotypes that comprise the I(76)*CA 2 allele. Given the strong LD between I(76)*CA n and 3 0 (325)*G-A polymorphisms in our study, the association cannot be attributed to any of the individual markers. However, the association appears to be strongest at the 3 0 (325) SNP. Despite homogeneity in LD through the region, the distribution of genotypes with regard to individual markers was not homogeneous among the three populations studied necessitating some caution in interpretation

6 58 of the combined population results. While two-marker haplotype analysis revealed association primarily in men from USA, the association for the 3 0 (325) SNP was strongly associated with MS in men both in USA and Northern Ireland; furthermore, the odds ratio was greater than.0 for this marker in all three populations. Population heterogeneity regarding microsatellite markers among populations could explain the difference especially regarding the haplotype analysis but this would not invalidate the results since two of the three populations demonstrated a significant association and the third population s results were in the similar direction. The Belgian study population may be less homogeneous than the Northern Irish (isolated island population) and the American (carefully matched for ethnicity to controls) populations. Also, the MS prevalence in Belgium (88/ in 99) 44 is lower than in Olmsted County, MN (77/00 000) 45 and Northern Ireland (68/ ), 46 and genetic load may be higher in populations with higher prevalence. Ultimately, lesser homogeneity and different genetic loading may affect the power to detect a weak effect in complex disorders. It is also possible that men who have the haplotype with CA 3 /A alleles have greater susceptibility to MS from a yet to be identified additional susceptibility factor present in the MN and Northern Irish, but not Belgian, populations. Goris et al 42 recently confirmed that a protective effect of the haplotype including I(76)*CA 2 from MS in men is evident in Sardinians only if they also lack the high-risk HLA*DR3 or HLA*DR4 genotypes. Increased MS susceptibility to HLA*DR3 or HLA*DR4 carriers is present only in Sardinians, whereas HLA*DR2 carriers have increased susceptibility to MS in other European populations. 47 IFNg production may be HLA*DR dependent. HLA DR3, 4, 5 and 7 are associated with low, while DR, 2 and 6 are associated with high IFNg production. 48 Whether there is any direct or indirect interaction between HLA*DR alleles and IFNG polymorphisms remains to be elucidated. Other groups have also studied the I(76)*CA n polymorphism for association with MS and other multifactorial diseases. 49 In Sweden, an initial weak association with susceptibility and linkage to MS was detected. 37,39 In a large study from Sweden, Dai et al 4 did not find an association between IFNG polymorphisms with either susceptibility to and severity of MS or with mrna levels of IFNg. In a transmission disequilibrium-based study from France, 40 there was significant undertransmission of I(76)*CA 2 to patients with MS (transmitted: N ¼ 34; nontransmitted: N ¼ 60) when compared to I(76)*CA 3 allele (transmitted: N ¼ 62; nontransmitted: N ¼ 44), supporting our finding even though gender stratification was omitted. German and Italian MS populations did not manifest the gender bias observed in our study. 39 As discussed for the Belgian population differences in genetic loading, heterogeneity of cases and controls and the indirect effects of these two factors on power may account for some of these differences. Interestingly, the IFNG intron CA polymorphism was shown to be a modifier gene in tuberous sclerosis-2; the CA 2 allele is protective against kidney angiomyolipomas, particularly in men. 50 This finding suggests that IFNG may be relevant to gender-associated disease manifestations beyond MS. Other microsatellites in this general region have been associated with susceptibility to rheumatoid arthritis in women. 5 Pravica et al 52 and others have shown that the I(76)*CA 2 allele is associated with increased expression of IFNg mainly in homozygotes, although a trend toward increased expression was observed in heterozygotes, findings confirmed by others Pravica et al 56 also reported that an SNP in intron [I(759)*T-A] isin tight LD with I(76)*CA n. The high-secretory I(76)*CA 2 allele is in strong LD with I(759)*T. Homozygosity for the I(759)*A allele is associated with a 3.8-fold increase in risk of tuberculosis in Spain. Stimulated production of IFNg by peripheral mononuclear cells from I(759)*A homozygotes is depressed compared with I(759)*T carriers. 57 This polymorphism was also found to be strongly associated with susceptibility to tuberculosis in South Africa both in populationbased and family-based association studies. 58 LD between these markers and, as demonstrated in our study, between I(76)*CA n and the 3 0 (325)*G-A SNP, may lend a functional relevance to our observations. Alternatively, a yet to be identified polymorphism in LD with the relevant haplotypes as described could be responsible for the effect. The apparent protective effect of CA 2 /G haplotype, which would be expected to be associated with higher IFNg expression, seems to conflict with the hypothesis that higher IFNg expression leads to an increased Th- cell response and consequently greater susceptibility to MS. 59 On the other hand, some effects of IFNg-related inflammation such as inhibition of the accumulation of activated T cells by limiting antigen-induced proliferation and/or by enhancing apoptosis can actually be protective in autoimmune disorders as recently discussed by O Shea and Paul 60 and Martino et al. 6 Recent evidence by Pelfrey et al 26 and Nguyen et al 27 suggests that there is stronger Th-/Th-2 bias in women than men with MS. The difference in frequency of the alleles between men and women may also reflect interaction between the disease, the genotype and gender, because the differences in genotype frequency are not evident in the controls. A functional explanation for how the IFNG polymorphism associations we revealed might influence gender bias in susceptibility to MS might be sought in hormonal effects on IFNg production. Estrogen (7-betaestradiol) markedly increases transcription from the IFNg promoter and the level of dehydroepiandrosterone correlates with IFNg secretion. 62,63 Estriol treatment of nonpregnant women with MS decreases IFNg levels and suppresses the development of gadolinium enhancing lesions in the brain. 64 No evidence currently exists to link the current polymorphisms with this effect of the sex steroids on IFNg expression. IFNg expression and effects vary by gender in a variety of experimental models and disease states other than MS. Myelin basic protein (MBP)-specific T cells from male mice were shown to be less encephalitogenic than MBPspecific cells from female mice in EAE. 65 In that study, IL- 2 and IFNg production were lower in males than in females after antigen-specific stimulation of draining lymph node cells, while the IL-0 and IL-4 production were similar. In the predominantly Th-2 cell-mediated Leishmania mexicana infection, female mice known to be resistant to infection express more IFNg in lymph nodes than male mice. 66 These findings suggest a predominant

7 Th- response in females, which is protective in this instance. Female mice produce more IFNg than male mice in picornavirus and encephalomyocarditis virus infections and in response to PPD II in BCG-sensitized mice IFNg deficiency markedly decreases atherosclerotic lesions in male mice, but has no effect in female mice. 7 Male IFNg or IFNg receptor knockout mice infected with herpes simplex virus type- have a higher frequency of reactivation and a higher associated mortality than wild-type mice, whereas there is no difference between knockout and wild-type females. 72 Taken together with previous studies, our study provides further evidence that there is an association between IFNG polymorphisms and susceptibility to sporadic MS. Genetic variation in IFNg may in part account for gender differences in susceptibility to MS. Patients and methods Patient populations The data set consisted of three independently ascertained populations of patients and controls from USA, Northern Ireland and Belgium. The demographic details of these populations are illustrated in Table. The patient population from MN was unique in that it consisted of 22 patients (64 men and 57 women) representing 79% ascertainment of two overlapping population-based prevalence cohorts from 99 73,74 from Olmsted County, MN. A total of 442 controls (28 men and 34 women) matched for ethnicity, age and gender were selected from 4000 Mayo Clinic patients from whom DNA samples had been obtained. The details of the patient population and ethnicity matching are described elsewhere. 75 The population from Northern Ireland consisted of 432 cases (47 men and 285 women) and 205 controls (98 men and 07 women) and the population from Belgium consisted of 208 cases (82 men and 26 women) and 96 controls (4 men and 82 women). Controls from these populations were clinic-based and not specifically matched to cases for ethnicity, gender and age. The institutional review boards at all three participating institutions approved the study. Selection of markers and genotyping The same markers were selected as those that had been studied in the previous studies in Sardinia and MN. 42,43,76 We evaluated two markers within or immediately adjacent to IFNG, a CA repeat microsatellite in intron [I(76)*CA n ] and an SNP just 3 0 of the poly A site [3 0 (325)*G-A] (rs# ); these markers are identified herein according to their position relative to the first base of the exon or intron in which they are found. We also studied three flanking microsatellite markers spanning a 8 kb region around IFNG that includes IL26 (D2S33, D2S250 and D2S25 from centromere to telomere). The microsatellite markers were sized by automated sequencing utilizing DNA sequencers (Applied Biosystems, ABI300 model). The SNP was analyzed by a fluorescence polarization-based single base extension method (Analyst HT, Molecular Devices). 77 Equivocal samples are studied with a primer in the reverse orientation on the amplicon to ensure agreement with results of the other primer. The relative location of the markers studied is shown in Figure. Statistical analysis LD in cases and controls through the region containing IFNG was analyzed using the D 0 measure. We studied association for individual markers by genotype; differences in carrier frequencies for cases and controls, stratified by gender, were assessed using Fisher s exact statistics for contingency tables. The Breslow Day test was used to first check for homogeneity of the odds ratios for the three populations, and then the Cochran Mantel Haenzel statistic was used to combine results of the three populations. Note that the odds ratio confidence limits are conservative and do not correspond exactly to the P-values listed. Haplotype frequencies were imputed for individuals with ambiguous haplotypes by an expectation-maximization algorithm and compared between cases and controls using a test for associations between traits and haplotypes where phase is ambiguous using the haplo.score function within Splus. 78 Simulated P-values were estimated to help eliminate spurious significant results that may result from multiple testing. 78 Consecutive two-marker haplotypes were used to further compare cases and controls using the max-stat statistic and the corresponding permutation P-values. Haplotypes with expected frequencies of less than 5% were removed from the analyses to increase model stability. All analysis was completed using the SAS 79 and Splus software packages. 80 Acknowledgements The National Institute of Health, grant no. NS45442, and the National MS Society, grant no A-6-02, to B Weinshenker, K Vandenbroeck and C Pelfrey have supported this study. The Northern Ireland HPSS R&D Office supported K Vandenbroeck, grant RRG.5 RSG 726. A Goris was a research assistant of the Fund for Scientific Research Flanders (FWO Flanders). The Mayo Foundation, grant no. DK A2, and National Institutes of Health, grant no. MH-56207, supported C McMurray. The University Hospital Research Council, Leuven, Belgium, supported B Dubois. References Oksenberg JR, Hauser SL. New insights into the immunogenetics of multiple sclerosis. Curr Opin Neurol 997; 0: Dyment DA, Sadovnick AD, Ebers GC. Genetics of multiple sclerosis. Hum Mol Genet 997; 6: Kantarci OH, de Andrade M, Weinshenker BG. Identifying disease modifying genes in multiple sclerosis. J Neuroimmunol 2002; 23: Ebers GC, Kukay K, Bulman DE et al. A full genome search in multiple sclerosis. Nat Genet 996; 3: Haines J, Pericak-Vance M, Seboun E, Hauser S, and the Multiple Sclerosis Study Group. A complete genomic screen for multiple sclerosis underscores a role for the major histocompatibility complex. Nat Genet 996; 3: Sawcer S, Jones HB, Feakes R et al. A genome screen in multiple sclerosis reveals susceptibility loci on chromosome 6p2 and 7q22. Nat Genet 996; 3: Kuokkanen S, Sundvall M, Terwilliger JD et al. A putative vulnerability locus to multiple sclerosis maps to 5p4 p2 in a region syntenic to the murine locus Eae2. Nat Genet 996; 3:

8 60 8 Chataway J, Feakes R, Coraddu F et al. The genetics of multiple sclerosis: principles, background and updated results of the United Kingdom systemic genome screen. Brain 998; 2: Risch N, Merikangas KR. The future of genetic studies of complex human diseases. Science 996; 273: Billiau A. Interferon-gamma in autoimmunity. Cytokine Growth Factor Rev 996; 7: Navikas V, Link H. Review: cytokines and the pathogenesis of multiple sclerosis. J Neurosci Res 996; 45: Issazadeh S, Mustafa M, Ljungdahl A et al. Interferon gamma, interleukin 4 and transforming growth factor beta in experimental autoimmune encephalomyelitis in Lewis rats: dynamics of cellular mrna expression in the central nervous system and lymphoid cells. J Neurosci Res 995; 40: Renno T, Lin JY, Piccirillo C, Antel J, Owens T. Cytokine production by cells in cerebrospinal fluid during experimental allergic encephalomyelitis in SJL/J mice. J Neuroimmunol 994; 49: 7. 4 Willenborg DO, Fordham S, Bernard CC, Cowden WB, Ramshaw IA. IFN-gamma plays a critical down-regulatory role in the induction and effector phase of myelin oligodendrocyte glycoprotein-induced autoimmune encephalomyelitis. J Immunol 996; 57: Espejo C, Penkowa M, Saez-Torres I et al. Treatment with anti-interferon-gamma monoclonal antibodies modifies experimental autoimmune encephalomyelitis in interferongamma receptor knockout mice. Exp Neurol 200; 72: Horwitz MS, Evans CF, McGavern DB, Rodriguez M, Oldstone MB. Primary demyelination in transgenic mice expressing interferon-gamma. Nat Med 997; 3: Renno T, Taupin V, Bourbonniere L et al. Interferon-gamma in progression to chronic demyelination and neurological deficit following acute EAE. Mol Cell Neurosci 998; 2: Popko B, Baerwald KD. Oligodendroglial response to the immune cytokine, interferon gamma. Neurochem Res 999; 24: Kim IJ, Beck HN, Lein PJ, Higgins D. Interferon gamma induces retrograde dendritic retraction and inhibits synapse formation. J Neurosci 2002; 22: Beck J, Rondot P, Catinot L, Falcoff E, Kirchner H, Wietzerbin J. Increased production of interferon gamma and tumor necrosis factor precedes clinical manifestation in multiple sclerosis: do cytokines trigger off exacerbations? Acta Neurol Scand 988; 78: Dettke M, Scheidt P, Prange H, Kirchner H. Correlation between interferon production and clinical disease activity in patients with multiple sclerosis. J Clin Immunol 997; 7: Petereit HF, Nolden S, Schoppe S, Bamborschke S, Pukrop R, Heiss WD. Low interferon gamma producers are better treatment responders: a two-year follow-up of interferon beta-treated multiple sclerosis patients. Mult Scler 2002; 8: Panitch HS, Hirsch RL, Haley AS, Johnson KP. Exacerbations of multiple sclerosis in patients treated with gamma interferon. Lancet 987; : Duquette P, Pleines J, Girard M, Charest L, Senecal-Quevillon M, Masse C. The increased susceptibility of women to multiple sclerosis. Can J Neurol Sci 992; 9: Thompson AJ, Polman CH, Miller DH et al. Primary progressive multiple sclerosis. Brain 997; 20 (Part 6): Pelfrey CM, Cotleur AC, Lee JC, Rudick RA. Sex differences in cytokine responses to myelin peptides in multiple sclerosis. J Neuroimmunol 2002; 30: Nguyen LT, Ramanathan M, Weinstock-Guttman B, Baier M, Brownscheidle C, Jacobs LD. Sex differences in in vitro proinflammatory cytokine production from peripheral blood of multiple sclerosis patients. J Neurol Sci 2003; 209: Zimonjic DB, Rezanka LJ, Evans CH, Polymeropoulos MH, Trent JM, Popescu NC. Mapping of the immune interferon gamma gene (IFNG) to chromosome band 2q4 by fluorescence in situ hybridization. Cytogenet Cell Genet 995; 7: Giedraitis V, He B, Hillert J. Mutation screening of the interferon-gamma gene as a candidate gene for multiple sclerosis. Eur J Immunogenet 999; 26: Ruiz-Linares A. Dinucleotide repeat polymorphisms in the interferon-gamma (IFNG) gene. Hum Mol Genet 993; 2: Wu S, Muhleman D, Comings DE. G5644A polymorphism in the interferon-gamma (IFNG) gene. Psychiatry Genet 998; 8: Bream JH, Carrington M, O Toole S et al. Polymorphisms of the human IFNG gene noncoding regions. Immunogenetics 2000; 5: Iwasaki H, Ota N, Nakajima T et al. Five novel singlenucleotide polymorphisms of human interferon gamma identified by sequencing the entire gene. J Hum Genet 200; 46: Ackerman H, Udalova I, Hull J, Kwiatkowski D. Evolution of a polymorphic regulatory element in interferon-gamma through transposition and mutation. Mol Biol Evol 2002; 9: Bream JH, Ping A, Zhang X, Winkler C, Young HA. A single nucleotide polymorphism in the proximal IFN-gamma promoter alters control of gene transcription. Genes Immunity 2002; 3: Chevillard C, Henri S, Stefani F, Parzy D, Dessein A. Two new polymorphisms in the human interferon gamma (IFN-gamma) promoter. Eur J Immunogenet 2002; 29: He B, Xu C, Yang B, Landtblom AM, Fredrikson S, Hillert J. Linkage and association analysis of genes encoding cytokines and myelin proteins in multiple sclerosis. J Neuroimmunol 998; 86: Vandenbroeck K, Opdenakker G, Goris A, Murru R, Billiau A, Marrosu MG. Interferon-gamma gene polymorphism-associated risk for multiple sclerosis in Sardinia. Ann Neurol 998; 44: Goris A, Epplen C, Fiten P et al. Analysis of an IFN-gamma gene (IFNG) polymorphism in multiple sclerosis in Europe: effect of population structure on association with disease. J Interferon Cytokine Res 999; 9: Reboul J, Mertens C, Levillayer F et al. Cytokines in genetic susceptibility to multiple sclerosis: a candidate gene approach. French Multiple Sclerosis Genetics Group. J Neuroimmunol 2000; 02: Dai Y, Masterman T, Huang WX et al. Analysis of an interferon-gamma gene dinucleotide-repeat polymorphism in Nordic multiple sclerosis patients. Mult Scler 200; 7: Goris A, Heggarty S, Marrosu MG, Graham C, Billiau A, Vandenbroeck K. Linkage disequilibrium analysis of chromosome 2q4 5 in multiple sclerosis: delineation of a 8-kb interval around interferon-gamma (IFNG) that is involved in male vs female differential susceptibility. Genes Immun 2002; 3: Kantarci OH, Hebrink DD, Atkinson EJ, McMurray CT, Weinshenker BG. A comprehensive screen for genetic variation in the interferon-gamma gene in multiple sclerosis. Neurology 2000; 54 (Suppl 3): A van Ooteghem P, D Hooghe MB, Vlietinck R, Carton H. Prevalence of multiple sclerosis in Flanders, Belgium. Neuroepidemiology 994; 3: Mayr WT, Pittock SJ, McClelland RL, Jorgensen NW, Noseworthy JH, Rodriguez M. Incidence and prevalence of multiple sclerosis in Olmsted County, Minnesota, Neurology 2003; 6: McDonnell GV, Hawkins SA. An epidemiologic study of multiple sclerosis in Northern Ireland. Neurology 998; 50:

9 47 Marrosu MG, Murru MR, Costa G et al. Multiple sclerosis in Sardinia is associated and in linkage disequilibrium with HLA DR3 and DR4 alleles. Am J Hum Genet 997; 6: Petrovsky N, Harrison LC. HLA class II-associated polymorphism of interferon-gamma production. Implications for HLA-disease association. Hum Immunol 997; 53: Vandenbroeck K, Goris A. Cytokine gene polymorphisms in multifactorial diseases: gateways to novel targets for immunotherapy? Trends Pharmacol Sci 2003; 24: Dabora SL, Roberts P, Nieto A et al. Association between a high-expressing inteferon-gamma allele and a lower frequency of kidney angiomyolipomas in TSC2 patients. Am J Hum Genet 2002; 7: Vandenbroeck K, Cunningham S, Goris A et al. Polymorphisms in the interferon-gamma/interleukin-26 gene region contribute to sex bias in susceptibility to rheumatoid arthritis. Arthritis Rheum 2003; 48: Pravica V, Asderakis A, Perrey C, Hajeer A, Sinnot PJ, Hutchinson IV. In vitro production of IFN-g correlates with CA repeat polymorphism in the human IFN-g gene. Eur J Immunogenet 999; 26: Hoffmann SC, Stanley EM, Cox ED et al. Ethnicity greatly influences cytokine gene polymorphism distribution. Am J Transplant 2002; 2: Hutchings A, Guay-Woodford L, Thomas JM et al. Association of cytokine single nucleotide polymorphisms with B7 costimulatory molecules in kidney allograft recipients. Pediatr Transplant 2002; 6: Miyake K, Nakashima H, Akahoshi M et al. Genetically determined interferon-gamma production influences the histological phenotype of lupus nephritis. Rheumatology 2002; 4: Pravica V, Perrey C, Stevens A, Lee JH, Hutchinson IV. A single nucleotide polymorphism in the first intron of the human IFN-gamma gene: absolute correlation with a polymorphic CA microsattelite marker of high IFN-gamma production. Hum Immunol 2000; 6: Lopez-Maderuelo D, Arnalich F, Serantes R et al. Interferongamma and interleukin-0 gene polymorphisms in pulmonary tuberculosis. Am J Respir Crit Care Med 2003; 67: Rossouw M, Nel HJ, Cooke GS, van Helden PD, Hoal EG. Association between tuberculosis and a polymorphic NFkappaB binding site in the interferon gamma gene. Lancet 2003; 36: Young HA. Regulation of interferon-gamma gene expression. J Interferon Cytokine Res 996; 6: O Shea JJ, Paul WE. Regulation of T(H) differentiation controlling the controllers. Nat Immunol 2002; 3: Martino G, Adorini L, Rieckmann P et al. Inflammation in multiple sclerosis: the good, the bad, and the complex. Lancet Neurol 2002; : Fox HS, Bond BL, Parslow TG. Estrogen regulates the IFNgamma promoter. J Immunol 99; 46: Verthelyi D, Klinman DM. Sex hormone levels correlate with the activity of cytokine-secreting cells in vivo. Immunology 2000; 00: Sicotte NL, Liva SM, Klutch R et al. Treatment of multiple sclerosis with the pregnancy hormone estriol. Ann Neurol 2002; 52: Kim S, Voskuhl RR. Decreased IL-2 production underlies the decreased ability of male lymph node cells to induce experimental autoimmune encephalomyelitis. J Immunol 999; 62: Satoskar A, Al-Quassi HH, Alexander J. Sex-determined resistance against Leishmania mexicana is associated with the preferential induction of a Th-like response and IFN-gamma production by female but not male DBA/2 mice. Immunol Cell Biol 998; 76: Ishikawa R, Bigley NJ. Sex hormone modulation of interferon (IFN) alpha/beta and gamma production by mouse spleen cell subsets following picornavirus infection. Viral Immunol 990; 3: McFarland HI, Bigley NJ. Sex-dependent, early cytokine production by NK-like spleen cells following infection with the D variant of encephalomyocarditis virus (EMCV-D). Viral Immunol 989; 2: Huygen K, Palfliet K. Strain variation in interferon gamma production of BCG-sensitized mice challenged with PPD II. Importance of one major autosomal locus and additional sexual influences. Cell Immunol 984; 85: Dabora SL, Roberts P, Nieto A et al. Association between a high-expressing interferon-gamma allele and a lower frequency of kidney angiomyolipomas in TSC2 patients. Am J Hum Genet 2002; 7: Whitman SC, Ravisankar P, Daugherty A. IFN-gamma deficiency exerts gender-specific effects on atherogenesis in apolipoprotein E / mice. J Interferon Cytokine Res 2002; 22: Han X, Lundberg P, Tanamachi B, Openshaw H, Longmate J, Cantin E. Gender influences herpes simplex virus type infection in normal and gamma interferon-mutant mice. J Virol 200; 75: Rodriguez M, Siva A, Ward J, Stolp-Smith K, O Brien P, Kurland L. Impairment, disability, and handicap in multiple sclerosis: a population-based study in Olmsted County, Minnesota. Neurology 994; 44: Pittock SJ, Mayr WT, McClelland RL et al. Change in MSrelated disability in a population-based cohort: a 0-year follow-up study. Neurology 2004; 62: Kantarci OH, Hebrink DD, Achenbach SJ et al. A populationbased association study of Fas and FasL polymorphisms in multiple sclerosis. J Neuroimmunol 2004; 46: Goris A, Marrosu MG, Vandenbroeck K. Novel polymorphisms in the IL-0 related AK55 gene (chromosome 2q5). Genes Immun 200; 2: Kwok PY. SNP genotyping with fluorescence polarization detection. Hum Mutat 2002; 9: Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. Score tests for association between traits and haplotypes when linkage phase is ambiguous. Am J Hum Genet 2002; 70: SAS/STAT User s Guide, Version 8. SAS Institute Inc., Cary, NC, InsightfulCorp. S-plus Reference Manual, Version 6. Seattle, WA,