Search tips
Search criteria 


Lancet Neurol. 2008 July; 7(7): 567–569.
PMCID: PMC2696028

Refining genetic associations in multiple sclerosis

International Multiple Sclerosis Genetics Consortium (IMSGC)

Genome-wide association studies involve several hundred thousand markers and, even when quality control is scrupulous, are invariably confounded by residual uncorrected errors that can falsely inflate the apparent difference between cases and controls (so-called genomic inflation).1 As a consequence such studies inevitably generate false positives alongside genuine associations. By use of Bayesian logic and empirical data, the Wellcome Trust Case Control Consortium suggested that association studies in complex disease should involve at least 2000 cases and 2000 controls, at which level they predicted that p values of less than 5×10−7 would more commonly signify true positives than false positives.2

The screening phase of our recent multiple sclerosis genome-wide association study3 involved just 931 trio families and thus fell short of the minimum power recommended by the Wellcome Trust Case Control Consortium. However, the extension phase of our study included 2322 cases, 5418 controls, and 1540 trio families (12 360 individuals in total) and identified three markers exceeding the consortium's threshold—rs6897932 in IL7R (p=2·94×10−7) and rs12722489 and rs2104286 in IL2RA (p=2·96×10−8 and 2·16×10−7 respectively). These markers showed modest levels of significance in the screening phase of the study (p values 0·0058, 0·0013, and 0·0033, respectively). In overlapping4 and independent5 data sets, we simultaneously identified association with IL7R (rs6897932) through a candidate gene approach. IL2RA was suggested as a candidate by its confirmation as a susceptibility gene for type 1 diabetes.2 The extensive linkage disequilibrium between rs12722489 and rs2104286 in the IL2RA gene meant that it was impossible to determine whether one or other locus exerts a primary effect or whether both influence risk.

figure fx1
Human DNA sequence

The three identified loci have several similarities. For each the more common (major) allele increases susceptibility, and in each case the risk exerted by this allele is modest (with odds ratios about 1·2). All three of these single-nucleotide polymorphisms have been studied in the HapMap cohorts and curiously in each case the risk allele is even more common in non-white ethnic groups. Because multiple sclerosis is more common in white people than in other ethnic groups, this reverse pattern of allele frequency is a reminder that these alleles account for only a fraction of the heritable influences on susceptibility.

To refine our understanding of these associations, we typed all three variants in an additional 20 708 individuals in Australia, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, the Netherlands, Norway, Sardinia, Spain, Sweden, and new samples from the UK (webappendix). Together with the 12 360 reported in our original screen this provides a total of 33 068 individuals, including 11 019 unrelated cases, 13 616 controls and 2811 trio families (8433 individuals). All individuals involved in this study gave informed consent under appropriate local ethical approval. Overall genotyping efficiency was 98·4% for rs6897932, 95·4% for rs12722489, and 95·7% for rs2104286. None of the three markers showed any significant evidence for deviation from Hardy-Weinberg equilibrium in the controls although deviation was seen in the cases, as expected for genuine associations (webappendix).

In total, 20 population-specific cohorts (14 case-control and six trio family) were considered. Nominally significant association was observed in eight for rs6897932, in nine for rs12722489, and in 13 for rs2104286. In all but three studies, the risk allele as defined in our original screen (ie, the major allele at each locus) was over-represented in cases. None of these three negative findings (Australia and Ireland for rs6897932, and Holland for rs12722489) was significant. In short, all significant studies were in accordance with the original screen and most in which there was no statistically significant association implicated the major allele as expected. Results for the individual studies are shown in the webappendix.

In the control groups, major-allele frequency was 64–77% for rs6897932, 77–90% for rs12722489, and 69–83% for rs2104286. However, applying the Breslow-Day test confirms that there is no evidence of heterogeneity of effect across the populations for any of the markers. Thus, although the frequency of the risk allele shows modest variation between white populations, the effects of these alleles are of undoubted relevance (table).6,7

Association testing in combined cohorts

We confirmed linkage disequilibrium between the two polymorphisms in IL2RA (r2=0·5). Conditioning on each marker in turn shows that the association seen at rs12722489 is entirely a consequence of its linkage disequilibrium with rs2104286. This finding confirms that rs2104286 (or another single-nucleotide polymorphism in linkage disequilibrium with it) is the primary association even though it showed less significant association than rs12722489 in the original screen. Testing for association at the genotypic level confirms that the homozygous risk genotype confers a significantly greater risk than the heterozygous genotype for both rs6897932 and rs2104286 (webappendix).

This extension analysis illustrates the value of data sets that are significantly larger than the minimum recommended by the Wellcome Trust Case Control Consortium. Although these data convincingly replicate these associations, they do not establish these particular variants as causative. Fine mapping and functional studies will be required.

The IMSGC contributors—David Booth, Robert Heard, Graeme Stewart (University of Sydney, Institute for Immunology and Allergy Research, Westmead Millennium Institute, Westmead Hospital, Australia), An Goris, Rita Dobosi, Benedicte Dubois (Section for Experimental Neurology, Katholieke Universiteit Leuven, Belgium), Annette Oturai, Helle B Soendergaard, Finn Sellebjerg (MS Research Centre, Department of Neurology, Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark), Janna Saarela, Virpi Leppä, Aarno Palotie, Leena Peltonen (Department of Molecular Medicine, National Public Health Institute and Institute of Molecular Medicine Finland, Helsinki, Finland), Bertrand Fontaine, Isabelle Cournu-Rebeix, (INSERM UMR S 546 and Université Pierre et Marie Curie, Paris, France), Francoise Clerget-Darpoux, Marie-Claude Babron (INSERM UMR S 535 and Universite Paris Sud, Villejuif, France), Frank Weber, Florian Holsboer, Bertram Müller-Myhsok (Max Planck Institute of Psychiatry, Munich, Germany), Peter Rieckmann, Antje Kroner (Department of Neurology, University of Würzburg, Germany), Colin Graham, Koen Vandenbroeck, Stanley Hawkins (Queens University, Belfast, Northern Ireland), Sandra D'Alfonso, Laura Bergamaschi (Department of Medical Sciences and Interdisciplinary Research Center of Autoimmune Diseases, University of Eastern Piedmont, Novara, Italy), Paola Naldi (Department of Neurology, Ospedale Maggiore, Novara, Italy), Franca R Guerini (Laboratory of Molecular Medicine and Biotechnology, Don C Gnocchi Foundation IRCCS, S Maria Nascente, Milan, Italy), Marco Salvetti (Neurology and Centre for Experimental Neurological Therapy, Universitè La Sapienza, Roma, Italy), Daniela Galimberti (Department of Neurological Sciences, Ospedale Maggiore Policlinico, University of Milan, Italy), Rogier Hintzen, Cornelia van Duijn (Department of Neurology and Epidemiology, Erasmus MC, Rotterdam, Netherlands), Åslaug R Lorentzen, Elisabeth G Celius, Hanne F Harbo (Department of Neurology, Ullevål University Hospital and University of Oslo, Norway), Anne Spurkland (Institute of Basal Medical Sciences, University of Oslo, Norway), Francesco Cucca (Dipartimento di Scienze Biomediche, Università di Sassari, Sardinia), Maria Giovanna Marrosu (Dipartimento di Scienze Cardiovascolari e Neurologiche, University of Cagliari, Centro Sclerosi Multipla, Ospedale Binaghi, Sardinia), Manuel Comabella, Xavier Montalban (Unitat de Neuroimmunologia Clínica, CEM-Cat, Hospital Universitari Vall d'Hebron, Barcelona, Spain), Pablo Villoslada (Department of Neurology, University of Navarra, Pamplone, Spain), Tomas Olsson, Ingrid Kockum, Jan Hillert (Department of Clinical Neurosciences, Centre for Molecular Medicine, Karolinska Hospital, Stockholm, Sweden), Maria Ban, Amie Walton, Stephen Sawcer, Alastair Compston (University of Cambridge, Department of Clinical Neuroscience, Addenbrooke's, Hospital, Hills Road, Cambridge, UK), Clive Hawkins, Tania Mihalova (Department of Neurology, Keele University Medical School, Hartshill Campus, Stoke on Trent, UK), Neil Robertson, Gillian Ingram (Department of Neurology, University Hospital of Wales, Heath Park, Cardiff, UK), Philip L De Jager, David A Hafler (Division of Molecular Immunology, Center for Neurologic Diseases, Brigham and Women's Hospital, Boston, MA, USA), John Rioux, Mark Daly (Broad Institute of Harvard University and Massachusetts Institute of Technology, Cambridge, MA, USA), Lisa Barcellos (University of California at Berkeley, Berkeley, CA, USA), Adrian Ivinson (Harvard Center for Neurodegeneration and Repair, Harvard Medical School, Boston, MA, USA), Margaret Pericak-Vance (University of Miami School of Medicine, Miami, FL, USA), Jorge Oksenberg, Stephen L Hauser (University of California at San Francisco, San Francisco, CA, USA), Jacob McCauley, David Sexton, Jonathan Haines (Center for Human Genetics Research, Vanderbilt University Medical Center, Nashville, TN, USA)

None of the IMSGC has any conflict of interest to declare. This work was supported by the National Multiple Sclerosis Society (AP 3758-A-16, RG 2899, FG-1718-A1), grants from the NINDS (NS049477, NS032830, NS26799), AI067152, the NIAID (P01 AI039671), an NMSS Collaborative Research Award (CA 1001-A-14), and the Penates Foundation. This study makes use of data generated by the Wellcome Trust Case Control Consortium. Funding for the project was provided by the Wellcome Trust under award 076113. We thank the National Institutes of Mental Health for generously allowing the use of their genotype data. Work in Finland was supported by Neuropromise EU project grant LSHM-CT-2005-018637 and NIH grant NS43559. Work of the French network REFGENSEP was supported by grants from AFM, ARSEP, and INSERM. The Italian work was supported by the Italian Foundation for Multiple Sclerosis (FISM grants, 2002/R/40 and 2005/R/10); Regione Piemonte (grants 2003 and 2004). The Norwegian Bone Marrow Donor Registry is acknowledged for collaboration in establishment of the Norwegian control material. We also thank the many study participants and their families.

Stephen Sawcer, University of Cambridge, Department of Clinical Neuroscience, Addenbrooke's Hospital, Cambridge CB2 2QQ, UK

Web Extra Material



1. Clayton DG, Walker NM, Smyth DJ. Population structure, differential bias and genomic control in a large-scale, case-control association study. Nat Genet. 2005;37:1243–1246. [PubMed]
2. Welcome Trust Case Control Consortium Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447:661–678. [PMC free article] [PubMed]
3. International multiple Sclerosis Genetics Consortium (IMSGC) Risk alleles for multiple sclerosis identified by a genomewide study. N Engl J Med. 2007;357:851–862. [PubMed]
4. Gregory SG, Schmidt S, Seth P. Interleukin 7 receptor alpha chain (IL7R) shows allelic and functional association with multiple sclerosis. Nat Genet. 2007;39:1083–1091. [PubMed]
5. Lundmark F, Duvefelt K, Iacobaeus E. Variation in interleukin 7 receptor alpha chain (IL7R) influences risk of multiple sclerosis. Nat Genet. 2007;39:1108–1113. [PubMed]
6. Purcell S, Neale B, Todd-Brown K. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81:559–575. [PubMed]
7. Dudbridge F. Pedigree disequilibrium tests for multilocus haplotypes. Genet Epidemiol. 2003;25:115–121. [PubMed]