BRIEF COMMUNICATION A two-stage study on multiple sclerosis susceptibility and chromosome 2q33

(2004) 5, 142 146 & 2004 Nature Publishing Group All rights reserved 1466-4879/04 $25.00 www.nature.com/gene BRIEF COMMUNICATION A two-stage study on multiple sclerosis susceptibility and chromosome 2q33 A Bonetti 1, K Reunanen 1, S Finnilä 2, K Koivisto 3, J Wikström 1, M-L Sumelahti 4, T Pirttilä 5, I Elovaara 6, M Reunanen 7, J Saarela 2,8, L Peltonen 2,8, T Rantamäki 1 and PJ Tienari 1 1 Neuroscience Programme, Department of Neurology, Biomedicum-Helsinki, University of Helsinki, Helsinki University Central Hospital, Haartmaninkatu 4, Helsinki, Finland; 2 Department of Human Genetics, UCLA School of Medicine, Los Angeles, CA, USA; 3 Central Hospital of Seinäjoki, Seinäjoki, Finland; 4 School of Public Health, University of Tampere, Tampere, Finland; 5 Department of Neurology and Neuroscience, University of Kuopio, Kuopio University Hospital, Kuopio, Finland; 6 Department of Neurology, University of Tampere, Tampere University Hospital, Tampere, Finland; 7 Department Neurology, University of Oulu, Oulu University Hospital, Oulu, Finland; 8 Department of Molecular Medicine, National Public Health Institute, Finland We have performed a two-stage study to analyse the association of polymorphism on chromosome 2q33 with multiple sclerosis (MS). In all, 17 markers were analysed in stage-1 in 134 Finnish MS families and the observed associations were tested in stage-2 in 186 MS families. We did not find previously reported allelic or haplotype associations with CTLA4. We obtained a weak signal of two distinct predisposing genes, one proximal the other distal of CTLA4. The putative proximal gene was associated with the marker rs3977 in families lacking HLA-DR2 (P ¼ 0.02 and 0.02) and the other distal gene was associated with D2S1271 in families from a high-risk region in western Finland (P ¼ 0.02 and 0.01). Based on the 43 cm distance and the lack of linkage disequilibrium between these loci, we conclude that the two association signals are independent. Our results provide preliminary evidence for two distinct MS susceptibility genes on 2q33 outside of CTLA4. (2004) 5, 142 146. doi:10.1038/sj.gene.6364049 Published online 15 January 2004 Keywords: ICOS; CD28; coeliac disease; type I diabetes mellitus; Graves disease; HERV-H Introduction Multiple sclerosis (MS) is an inflammatory, principally demyelinating disease of the central nervous system with suspected autoimmune pathogenesis. 1 Both genetic and environmental factors play a role in its aetiology. 2 4 MS is considered a genetically heterogeneous polygenic disease. 3,4 The most consistent genetic finding thus far is the association between MS and HLA-DR2. 5 In several immune-mediated diseases, there is evidence for a predisposing gene on chromosome 2q33. Most studies have been focused on the cytotoxic T-lymphocyteassociated 4 (CTLA4) gene, which codes for a protein involved in the costimulatory downregulation of immune responses. Indeed, the peak of association with type I diabetes and autoimmune thyreoiditis has been shown to occur at a noncoding region 6 kb downstream of CTLA4, which also associates with reduced mrna expression of a soluble splice variant. 6 In addition, associations with CTLA4 have been reported in coeliac disease, 7,8 systemic lupus erythematosus, 9 HLA-DQA1*0501-positive Addison s disease 10 and HLA-DR4-positive rheumatoid arthritis, 11 suggesting a more general role of this genomic region in susceptibility to immune-mediated diseases. Correspondence: Dr PJ Tienari. E-mail: pentti.tienari@hus.fi Received 04 November 2003; revised 12 November 2003; accepted 12 November 2003 In MS, there is evidence from linkage studies 12 and allelic association studies, 12 18 suggesting that chromosome 2q33 would harbour a predisposing gene. A particular candidate has been the CTLA4 gene, 12 18 on which the results have been, however, conflicting. 19 22 Here, we have performed a two-stage association analysis and linkage analysis to evaluate the role of this genomic region in MS susceptibility. Results and discussion In stage-1, we analysed the association between chromosome 2q33 markers and MS in 134 triad families using the transmission disequilibrium test. We tested 17 markers, both microsatellites and SNPs. The marker density was highest across the CD28-CTLA4-ICOS region (eight markers/0.5 Mb). Stage-1 served for setting the hypothesis, therefore corrections for multiple comparisons were not carried out. The results are shown in Figure 1. There were four markers, one SNP (rs3977) and three microsatellites (D2S2396, D2S307 and D2S1271), which associated with MS at Po0.05. None of the markers within the CTLA4 gene were associated with MS. We carried out a more thorough haplotype analysis on the CTLA4 locus since there are five previous reports of allelic or haplotype associations with MS. The associations were reported with þ 49/G, 12,15 651/T, 17 318/

143 Figure 1 Association of 17 chromosome 2q33 markers with MS. The scale in the CD28-CTLA4-ICOS region has been stretched as shown by the dashed line. The marker map was generated with the aid of the recently published human genome draft sequence (http:// www.ncbi.nlm.nih.gov/) and Marshfield genetic map (http://research.marshfieldclinic.org/genetics/map_markers/maps/). There is a conflict on the location of D2S307 between the Marshfield map (D2S307 positioned distal to D2S1271) and the human draft sequence (D2S307 positioned just distal to CD28 gene). Based on our LD results, its location close to CD28 is more likely. A few of the known genes of the region are positioned to the figure as landmarks. Sara47 microsatellite 23 was tested only in Bothnia families. DNA was extracted from peripheral blood leucocytes using standard procedures. The microsatellites D2S2396, D2S116, D2S307, CTLA4(AT) n, D2S72, D2S1271 and D2S422, and their genotyping protocols were obtained from the genome database (http://gdbwww.dkfz-heidelberg.de/). The CD28(CAA) n was analysed with the primers 5 0 -gag aat cgc ttg aac ctg gc-3 0 and 5 0 -tag aca aat aat cct tca cag ta 3 0. ABI Prism system with fluorescently labelled primers, autoradiography of the polyacrylamide gels using [ 33 P-a]dATP or the LI-COR model 4200 Dual Dye automated DNA sequencing system was used for microsatellite genotyping. The SNPs were obtained from the SNP homepage (http://www.ncbi.nlm.nih.gov/snp/) or previous publications. 9,19 The SNPs were amplified by PCR and genotyped by restriction enzymed digestion and agarose or polyacrylamide gel electrophoresis. The following primers were used. rs16653: 5 0 -taa gtc aca agc aaa atg gg-3 0 and 5 0 -atg cag gga att tcc taa ct 3 0. rs3977: 5 0 -cat aga aat gga aag cag gtg tag-3 0 and 5 0 -gga agt aag tat tag cca gag gag-3 0. rs7224 (synonymous change in exon 13 of the CDC-like kinase 1 gene, CLK- 1): 5 0 -ctc tcg aag aga tct tac aga- 3 0 and 5 0 -taa tga acc aaa tta ccc aaa ctt aa-3 0. rs1594 (synonymous change in exon 9 of the CASP8 and FADDlike apoptosis regulator gene, CFLAR): 5 0 -ttc tgg gag agg cac tgc-3 0 and 5 0 -gtg gtg tca gag ggc cag-3 0. The genotype data were analysed using the TRANSMIT 2.5.2 computer program package. 24 Alleles that were included in the association analysis had to be transmitted from at least 20 informative meioses, corresponding to the Var (O E) value of 5 or more in the TRANSMIT analysis. 24 C þ 49/G haplotype 14 and 318/C þ 49/A(AT) 8 haplotype. 16 Of the above markers, 651C/T was not tested here. However, based on a recent large multinational study, the 651C/T is in complete or near complete linkage disequilibrium (LD) with the 1722C/T (supplementary information B in Ueda et al 6 ) that we tested here. The 651C/T was not associated with MS in one of the previous studies. 16 We analysed all possible two- and three-marker haplotype combinations of the CTLA4 SNPs, the (AT) n as two-marker haplotypes with the SNPs and the (AT) n as a bi- and triallelic marker as suggested previously. 16 No associations at Po0.05 were found (not shown). Thus, within the power limits of the present study, we did not find any allelic or haplotype association with CTLA4. We carried out subpopulation analysis according to HLA type and geographic origin of the patients. The patients were divided into HLA-DR2 vs HLA-DRX and Bothnia vs non-bothnia strata. The rationale for the HLA stratum was based on previous reports on the possible interaction of HLA and 2q33 in disease susceptibility. 10,11,17,18 The Bothnia stratum was based on previous epidemiological 25 and molecular genetic 26 evidence, suggesting that founder effect may contribute to the high frequency of MS in Southern Ostrobothnia (referred here to as Bothnia). The results of the subpopulation analysis suggest that some of the observed associations with the four markers may be dependent on the HLA type or geographic origin (Table 1). Similar subpopulation analysis with the CTLA4 markers/haplotypes did not reveal any associations at Po0.05 level (not shown). The associations that were found in stage-1 were tested in a second set of 186 MS triad families. The results were analysed in the total set of families as well as in the HLA and Bothnia strata (Table 2). There were two associations, which were found in both stage-1 and -2. The association with rs3977t allele was found in the HLA-DRX families (P ¼ 0.02 and 0.02 in stage-1 and -2, respectively), and the association with D2S1271 245 bp allele was found in the Bothnia families (P ¼ 0.02 and 0.01 in stage-1 and -2, respectively). The data sets of stage-1 and -2 are illustrated in Table 3.

144 Table 1 Subgroup analysis of 2q33 marker associations with multiple sclerosis (stage-1) Marker/allele (freq) All families (n ¼ 134), HLA-DR2 (n ¼ 76), HLA-DRX (n ¼ 58), Bothnia (n ¼ 35), Non-Bothnia (n ¼ 99), Rs3977/T (0.38) 5.4 (0.02) 0.5 (0.47) 5.3 (0.02) 4.1 (0.04) 2.0 (0.16) D2S2396/155 bp (0.33) 6.2 (0.01) 4.7 (0.03) 2.7 (0.10) 0.7 (0.40) 5.7 (0.02) D2S307/214 bp (0.48) 7.4 (0.007) 4.2 (0.04) 3.5 (0.06) 0.2 (0.70) 7.7 (0.006) D2S1271/245 p (0.44) 6.3 (0.01) 1.9 (0.17) 4.1 (0.04) 5.9 (0.02) 2.5 (0.11) The families were stratified according to HLA (DR2 vs DRX) and parents birth place (Bothnia vs non-bothnia). HLA-DR2 typing was carried out as described. 26 Freq, frequency of the associated allele in parental chromosomes. Table 2 Stage-2 analysis of 2q33 marker associations with multiple sclerosis Marker/allele All families (n ¼ 186), HLA-DR2 (n ¼ 99), HLA-DRX (n ¼ 87), Bothnia (n ¼ 70), Non-Bothnia (n ¼ 116), Rs3977/T 3.1 (0.08) 0.2 (0.64) 5.2 (0.02) 0.4 (0.54) 3.0 (0.08) D2S2396/155 bp 0.3 (0.62) 0.0 (0.99) 0.5 (0.46) 0.1 (0.71) 0.1 (0.72) D2S307/214 bp 1.9 (0.17) 0.1 (0.76) 4.0 (0.05) 1.8 (0.18) 0.6 (0.44) D2S1271/245 bp 2.0 (0.15) 1.1 (0.31) 1.0 (0.32) 6.6 (0.01) 0.0 (0.87) Table 3 Characteristics of the multiple sclerosis family data sets Stage-1 Stage-2 Number of families 134 186 Both parents available 121 85 One parent and sib available 11 20 One parent available 2 81 Patients mean year of birth 1945 (range 1929 1983) Female:male ratio 2.0 2.6 HLA-DR2-positive patients 57% 53% 1958 (range 1934 1982) Two independent Finnish MS family data sets were used in the association analysis. These data sets contained nuclear families with an MS index case diagnosed according to Poser criteria. In the stage- 1 data set case-ascertainment sites were Helsinki (n ¼ 99) and Seinäjoki (n ¼ 35). This is the same data set that has been previously used in our studies on MBP gene and chromosome 19q13. 27,28 Of the families, 12 contained multiply affected offspring, in these cases the oldest affected sibling was chosen for association analysis. There were patients with both RR and PP MS. The number of PP cases was only 14 in stage-1. Hence, the possible association with this subtype of MS 15 could not be analysed. In the stage-2 data set caseascertainment sites were Helsinki (n ¼ 31), Seinäjoki (n ¼ 67), Tampere (n ¼ 13), Kuopio (n ¼ 62) and Oulu (n ¼ 13). LD between pairs of markers was analysed in the 134 MS families of stage-1. This analysis was performed by taking full advantage of the information on chromosomal phase in the families. The observed haplotype distribution in fully informative families was compared, using w 2 test, to the distribution expected on the basis of the marker allele frequencies. Fully informative families were selected by the x1 flag of TRANSMIT. 24 Rare events, where both the observed and expected numbers were below 5, were pooled for the w 2 analysis. Since LD was analysed in a selected material (MS families), the results of the LD analysis cannot be generalised. In all, 19 marker pairs showed global P-values less than 0.001. These pairs of markers were D2S2396/D2S116 (Po10 9 ), D2S116/D2S1271 (Po10 3 ), CD28/D2S307 (Po10 9 ), D2S307/ 318 (Po10 10 ), D2S307/ þ 49 (Po10 7 ), D2S307/(AT) n (Po10 6 ), D2S307/D2S72 (Po10 23 ), D2S307/D2S1271 (Po10 7 ), 1722/D2S72 (Po10 5 ), 318/ þ 49 (Po10 9 ), 318/(AT) n (Po10 6 ), 318/ D2S72 (Po10 41 ), 318/D2S1271 (Po10 6 ), þ 49/(AT) n (Po10 36 ), þ 49/D2S72 (Po10 57 ), (AT) n /D2S72 (Po10 53 ), (AT) n /D2S1271 (Po10 4 ) and D2S72/ D2S1271 (Po10 10 ). There was no LD between rs3977 and D2S1271 or their neighbouring markers, suggesting that the associations with MS, obtained by rs3977 and D2S1271, represent signals of two independent loci. Parametric linkage analysis was performed in 27 multiplex MS families under dominant (D) and recessive (R) models with the penetrance estimates f ¼ 0.05 and 0.76. 27 Five markers were included: D2S2396, D2S116, D2S72, D2S1271 and D2S422 (see Figure 1). No statistically significant evidence for linkage was found. In the total set of families, the maximum two-point lod score was 0.4 (y ¼ 0.2, D/f ¼ 0.05 model) at D2S2396. In HLA- DR2 families (n ¼ 17), the maximum lod score was 0.5 (y ¼ 0.2, D/f ¼ 0.76) at D2S2396. In HLA-DRX families (n ¼ 10), the maximum lod score was 0.6 (y ¼ 0.0, R/ f ¼ 0.05) at D2S72. In Bothnia families (n ¼ 20), the lod scores were negative with all markers at distances y ¼ 0.0 0.2. In non-bothnia families (n ¼ 7), the maximum lod score was 1.21 (y ¼ 0.0 R/f ¼ 0.05) at D2S2396. We have analysed the role of chromosome 2q33 in MS susceptibility. With respect to marker density, this is thus far the most thorough analysis of this genome region in MS. In the association analysis, we obtained the signal of two distinct putative susceptibility genes. Evidence for a proximal rs3977-associated gene was obtained in the HLA-DRX families, and evidence for a distal D2S1271- associated gene was found in the Bothnia families. Based on the large physical and genetic (43 cm) distance of these markers and the lack of LD between these loci, we

consider these two signals independent of each other, that is, indicative of two distinct genes. Our results should, however, be considered preliminary. Although the use of the two-stage design decreases the likelihood of type-1 error, the findings were confined to certain subpopulations and it is important that these findings be confirmed in other data sets. MS is considered a genetically heterogeneous disorder and the analysis of specific subpopulations of patients is gaining more attention. 3,4 One such subpopulation is found in Southern Ostrobothnia in western Finland (Bothnia families). High incidence and prevalence as well as familial clustering of MS has been demonstrated in this region. 29 31 Moreover, we have recently found that the association between MS susceptibility and MBP tetranucleotide repeat polymorphism is restricted to this geographical region that, together with its distinctive population history, suggests that the founder effect contributes to the high risk of MS in this region. 25,26 Thus, there is a justification for analysing Bothnia families separately. The picture with the HLA-DRX subpopulation is less clear. This subpopulation may possess an interacting gene in the non-dr2 HLA haplotype. Alternatively, the lack of DR2 may just mark a genealogical subpopulation of patients without any direct biological interaction with HLA. At present, we cannot resolve this issue. However, certain previous association 32 and linkage signals 27,33,34 have been obtained only in the HLA-DRX MS group, indicating that this HLA dichotomy may be either biologically or genealogically sound. Based on previous linkage studies and association analyses on CTLA4, chromosome 2q33 has been considered a candidate region in MS. Our present results support this candidacy, but do not support the role of a common allelic variant of CTLA4. In type I diabetes and autoimmune thyreoiditis, the chromosome 2q33- mediated disease risk was mapped to a noncoding region 6 kb downstream of the CTLA4 gene. 6 Our results suggest that there may be more than one susceptibility locus for immune-mediated diseases on 2q33. Similar findings were recently obtained in other studies as well. In a Finnish study on coeliac disease, an association was found with ICOS, 8 but not with CTLA4, including the markers implicated in type I diabetes and autoimmune thyreoiditis. 6 In a larger European study on celiac disease, the associated markers differed according to the geographic location, raising the possibility of more than one predisposing allele/locus. 7 Whether the predisposing alleles/loci on 2q33 are the same in MS and celiac disease is unclear, but it is of interest that in both diseases there is evidence for association with other loci than the type I diabetes/autoimmune thyreoiditis locus. Acknowledgements We are indebted to the patients and their families for participation in the study. Ms Seija Ahlbäck is acknowledged for her efforts in recruiting Bothnia families for this study, and Ms Lilja Erola for excellent technical assistance. This study was financially supported by grants from the Finnish Academy (PJT), The Center of Excellence for Disease Genetics of the Academy of Finland (LP), the National Multiple Sclerosis Foundation of USA (LP), the Sigrid Juselius Foundation (PJT and LP), the Helsinki University Central Hospital, the Paulo Foundation and the Finnish Cultural Foundation (PJT). The first two authors contributed equally to this article. References 1 Compston A, Coles A. Multiple sclerosis. Lancet 2002; 359: 1221 1231. 2 Sadovnick AD, Dyment D, Ebers GC. Genetic epidemiology of multiple sclerosis. Epidemiol Rev 1997; 19: 99 106. 3 Compston A, Sawcer S. Genetic analysis of multiple sclerosis. Curr Neurol Neurosci Rep 2002; 2: 259 266. 4 Herrera BM, Ebers GC. Progress in deciphering the genetics of multiple sclerosis. Curr Opin Neurol 2003; 16: 253 258. 5 Olerup O, Hillert J. HLA class II-associated genetic susceptibility in multiple sclerosis: a critical evaluation. Tissue Antigens 1991; 38: 1 15. 6 Ueda H, Howson JM, Esposito L et al. Association of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune disease. Nature 2003; 423: 506 511. 7 Holopainen P, Torinsson Naluai, Moodie S et al. Candidate gene region 2q33 in European families with coeliac disease. Tissue Antigens (in press). 8 Haimila K, Smedberg T, Mustalahti K, Mäki M, Partanen J, Holopainen P. Genetic association of coeliac disease susceptibility to polymorphisms in the ICOS gene on chromosome 2q33. Genes Immun (in press). 9 Hudson LL, Rocca K, Song YW, Pandey JP. CTLA-4 gene polymorphisms in systemic lupus erythematosus: a highly significant association with a determinant in the promoter region. Hum Genet 2002; 111: 452 455. 10 Donner H, Braun J, Seidl C et al. Codon 17 polymorphism of the cytotoxic T lymphocyte antigen 4 gene in Hashimoto s thyroiditis and Addison s disease. J Clin Endocr Metab 1997; 82: 4130 4132. 11 Seidl C, Donner H, Fischer B et al. CTLA4 codon 17 dimorphism in patients with rheumatoid arthritis. Tissue Antigens 1998; 51: 62 66. 12 Ligers A, Xu C, Saarinen S, Hillert J, Olerup O. The CTLA-4 gene is associated with multiple sclerosis. J Neuroimmunol 1999; 97: 182 190. 13 Harbo HF, Celius EG, Vartdal F, Spurkland A. CTLA4 promoter and exon 1 dimorphisms in multiple sclerosis. Tissue Antigens 1999; 53: 106 110. 14 Masterman T, Ligers A, Zhang Z et al. CTLA4 dimorphisms and the multiple sclerosis phenotype. J Neuroimmunol 2002; 131: 208 212. 15 Mäurer M, Ponath A, Kruse N, Rieckmann P. CTLA4 exon 1 dimorphism is associated with primary progressive multiple sclerosis. J Neuroimmunol 2002; 131: 213 215. 16 Kantarci OH, Hebrink DD, Achenbach SJ et al. CTLA4 is associated with susceptibility to multiple sclerosis. J Neuroimmunol 2003; 134: 133 141. 17 Alizadeh M, Babron MC, Birebent B et al. Genetic interaction of CTLA-4 with HLA-DR15 in multiple sclerosis patients. Ann Neurol 2003; 54: 119 122. 18 Rasmussen HB, Kelly MA, Francis DA, Clausen J. CTLA4 in multiple sclerosis. Lack of genetic association in a European Caucasian population but evidence of interaction with HLA-DR2 among Shanghai Chinese. J Neurol Sci 2001; 184: 143 147. 19 Dyment DA, Steckley JL, Willer CJ et al. No evidence to support CTLA-4 as a susceptibility gene in MS families: the Canadian Collaborative Study. J Neuroimmunol 2002; 123: 193 198. 145

146 20 Andreevskii TV, Sudomoina MA, Gusev EI, Boiko AN, Alekseenkov AD, Favorova OO. Polymorphism A/G in position +49 of CTLA4 exon 1 in multiple sclerosis in Russians. Mol Biol (Mosk) 2002; 36: 643 648. 21 Bocko D, Bilinska M, Dobosz T et al. Lack of association between an exon 1 CTLA-4 gene polymorphism A(49)G and multiple sclerosis in a Polish population of the Lower Silesia region. Arch Immunol Ther Exp (Warsz) 2003; 51: 201 205. 22 van Veen T, Crusius JBA, van Winsen L et al. CTLA-4 and CD28 gene polymorphisms in susceptibility, clinical course and progression of multiple sclerosis. J Neuroimmunol 2003; 140: 188 193. 23 Ling V, Wu PW, Finnerty HF et al. Assembly and annotation of human chromosome 2q33 sequence containing the CD28, CTLA4, and ICOS gene cluster: analysis by computational, comparative, and microarray approaches. Genomics 2003; 78: 155 168. 24 Clayton D. A Generalization of the transmission/disequilibrium test for uncertain-haplotype transmission. Am J Hum Genet 1999; 65: 1170 1177. 25 Tienari PJ, Sumelahti M-L, Rantamäki T, Wikström J. Multiple sclerosis in western Finland: Evidence for founder effect. Clin Neurol Neurosurg (in press). 26 Pihlaja H, Rantamäki T, Wikström J et al. Linkage disequilibrium between myelin basic protein (MBP) microsatellite and multiple sclerosis is restricted to a geographically-defined subpopulation in Finland. Genes Immun 2003; 4: 138 146. 27 Reunanen K, Finnilä S, Laaksonen M et al. Chromosome 19q13 and multiple sclerosis susceptibility in Finland: linkage and two-stage association study. J Neuroimmunol 2002; 126: 134 142. 28 Tienari PJ, Kuokkanen S, Pastinen T et al. Golli-MBP gene in multiple sclerosis susceptibility. J Neuroimmunol 1998; 81: 158 167. 29 Wikström J. Studies on the clustering of multiple sclerosis in Finland II: microepidemiology in one high-risk county with special reference to familial cases. Acta Neurol Scand 1975; 51: 173 183. 30 Sumelahti ML, Tienari PJ, Wikström J, Palo J, Hakama M. Regional and temporal variation in multiple sclerosis incidence in Finland during 1979 93. Neuroepidemiology 2000; 19: 67 75. 31 Sumelahti ML, Tienari PJ, Palo Jorma, Wikström J, Hakama M. Increasing prevalence of multiple sclerosis in Finland. Acta Neurol Scand 2001; 103: 153 158. 32 Guerini FR, Ferrante P, Losciale L et al. Myelin basic protein gene is associated with MS in DR4- and DR5-positive Italians and Russians. Neurology 2003; 61: 520 526. 33 Chataway J, Feakes R, Coraddu F et al. The genetics of multiple sclerosis: principles, background and updated results of the United Kingdom systematic genome screen. Brain 1998; 121: 1869 1887. 34 Haines JL, Bradford Y, Garcia ME et al. Multiple susceptibility loci for multiple sclerosis. Hum Mol Genet 2002; 11: 2251 2256.