Replication of Top Markers of a Genome-Wide Association Study in Multiple Sclerosis in Spain

ORIGINAL ARTICLE Replication of top markers of a genome-wide association study in multiple sclerosis in Spain

ML Cavanillas1,16, O Fernańdez2,16, M Comabella3,16, A Alcina4, M Fedetz4, G Izquierdo5, M Lucas6, MC Ceńit1, R Arroyo7, K Vandenbroeck8,9, I Alloza8, M Garcıá-Barcina10, A Antigu¨ edad11, L Leyva12, CL Go´mez2, J Olascoaga13, D Otaegui14, Y Blanco15, A Saiz15, X Montalbań3, F Matesanz4,16 and E Urcelay1,16 1Immunology Department Hospital Clıńico S. Carlos, Madrid, Spain; 2Inst. de Neurociencias Clıńicas, Hospital Carlos Haya, Ma´laga, Spain; 3Centre d’Esclerosi Mu´ltiple de Catalunya, CEM-Cat, Unitat de Neuroimmunologia Clıńica, Hospital Universitari Vall d’Hebron (HUVH), Barcelona, Spain; 4Instituto de Parasitologıá y Biomedicina ‘Lo´pez Neyra’, C. S. I. C., Granada, Spain; 5Unidad de Esclerosis Mu´ltiple. Hospital Virgen Macarena, Sevilla, Spain; 6Servicio de Biologıá Molecular. Hospital Virgen Macarena, Sevilla, Spain; 7Multiple Sclerosis Unit, Neurology Department, Hospital Clıńico S. Carlos, Madrid, Spain; 8Neurogenomiks Group, Universidad del Paı´s Vasco (UPV/EHU), Leioa, Spain; 9IKERBASQUE, Basque Foundation for Science, Bilbao, Spain; 10Servicio de Gene´tica, Hospital de Basurto, Bilbao, Spain; 11Servicio de Neurologıá, Hospital de Basurto, Bilbao, Spain; 12Laboratorio de Investigacioń, Hospital Civil, Ma´laga, Spain; 13Servicio de Neurologıá, Unidad de Esclerosis Mu´ltiple, Hospital Donostia, San Sebastiań, Spain; 14A´ rea de Neurociencias, Insti. Investigacioń Sanitaria Biodonostia, San Sebastiań, Spain and 15Neurology Service, Hospital Clinic and Institut d’Investigacio´ Biome`dica Pi i Sunyer (IDIBAPS), Barcelona, Spain

Multiple sclerosis (MS) is a chronic inflammatory disease of the central nervous system with presumed autoimmune origin, triggered by genetic and environmental risk factors. A recent genome-wide association study conducted on MS identified new biallelic markers outside the HLA (human leucocyte antigen) region involved in disease susceptibility: rs1109670 (DDEF2); rs1458175 (PDZRN4); rs1529316 and rs2049306 (CSMD1); rs16914086 (TBC1D2); rs1755289 (SH3GL2); rs1841770 (ZIC1); rs651477 (EN1); rs7607490 (TRIB2); rs397020 (C20orf46); rs908821 (SLC25A36); rs7672826 (MGC45800) and rs9523762 (GPC5). We aimed at replicating these top association signals in a Spanish cohort of 2863 MS patients and 2930 sex- and age- matched controls. Only rs9523762 mapping in the GPC5 gene was significantly associated (G allele, P ¼ 1.6 Â 10À5; odds ratio (95% confidence interval) ¼ 1.23 (1.12–1.36)), supporting a role for this proteoglycan in MS predisposition. The independent replication of association signals to validate data generated by genome-wide association scans is a first step in the effort to improve patient care. Genes and Immunity (2011) 12, 110–115; doi:10.1038/gene.2010.52; published online 14 October 2010

Keywords: multiple sclerosis; genetics; GWAS; GPC5

Introduction combination of genetic and environmental risk factors.2,3 The main genetic determinant described to date is the Multiple sclerosis (MS), a chronic inflammatory demye- major histocompatibility complex region identified more linating disorder of the central nervous system, is one of than 30-years ago;4 the HLA (human leucocyte antigen)- the most frequent causes of nontraumatic neurological DRB1*15:01 allele confers by far the strongest effect on disability in young adults. Although inflammation, MS susceptibility.5 Nonetheless, this is not the unique demyelination and axonal degeneration are well char- contributing factor to disease risk, and classical linkage acterized during the course of the disease,1 a compre- and association studies have pointed to an increasing hensive understanding of the etiology of MS and the series of non-major histocompatibility complex genes mechanistic pathways leading to disease progression are involved in MS predisposition. Moreover, during the past still lacking. It is widely accepted, though, that MS is a 2 years, genome-wide association studies (GWAS) have complex and heterogeneous disease resulting from the yielded an extraordinary wealth of data that provided interesting clues about the pathogenesis of immune- mediated diseases including rheumatoid arthritis, inflam- Correspondence: Dr E Urcelay, Immunology Department, Hospital matory bowel disease, type I diabetes among others.6,7 Universitario Clı´nico San Carlos, Martin Lagos s/n, Madrid 28040, Certain pathogenic mechanisms were unambiguously Spain. highlighted, for instance, the importance of autophagy E-mail: [email protected] 16These authors contributed equally to this work. and innate immunity as critical components of the altered 8 Received 22 June 2010; revised and accepted 03 August 2010; host–bacterial interactions leading to Crohn’s disease. published online 14 October 2010 Furthermore, previously unsuspected genes were found Replication of top markers of a genome-wide association study ML Cavanillas et al 111 to influence susceptibility in some immune-mediated Table 1 Demographic and clinical features of the study group diseases although the individual effect of each locus revealed usually modest genetic size effects, consistent Case Control with the polygenic basis of these complex traits.9 Considerable overlap was also shown to exist between Gender, n (%) loci identified as risk factors for the different auto- Female 1907 (66.6) 2004 (68.4) Male 956 (33.4) 926 (31.6) immune diseases, suggesting a shared influence of Female to male ratio 2 2.1 noteworthy genes on the breakdown of self-tolerance. In 2007, the International MS Genetics Consortium Age (year) performed the first GWAS in MS10 and identified the Age at onset involvement of several genes such as the cytokine Mean 29.9±10.35 receptor encoding genes IL2RA and IL7R, in keeping Range 6–68 Age at analysis with the autoimmune nature of the disease. The study Mean 36±12.23 38±10 analyzed more than 12000 subjects in two phases for Range 13–85 10–90 replication purposes. Another GWAS subsequently performed by the Australian and New Zealand MS Disease course no. (%) Genetics Consortium11 confirmed some of the genes RR 2047 (75.2) — previously found, such as IL7R, IL2RA, EVI5-RPL5, SP 472 (17.3) — CLEC16A or CD58, which have been consistently PP 158 (5.8) — 11–13 PR 13 (0.4) — validated in independent studies. CIS 32 (1) — Another GWAS was conducted in 1000 MS patients No details 99 — and an age/gender-matched control group.14 In contrast to the International MS Genetics Consortium study, Abbreviations: CIS, clinically isolated syndrome; PP, primary the latter focused on the discovery of genes not only progressive; PR, progressive–relapsing; RR, relapsing remitting; associated with MS susceptibility, but also with the SP, secondary progressive. clinical phenotype. Our purpose with the present work is to replicate the association with overall MS of the top (non-HLA) risk markers identified in the GWAS con- by logistic regression analyses adjusted by covariates ducted by Baranzini and collaborators.14 One should (gender and HLA-DRB1*15:01, Table 3), as reported by bear in mind that even a relative large GWAS has a rather Baranzini et al.14 low power to detect any specific marker. The accepted Stratification by the well-known risk factor measure of genome-wide significance (P ¼ 10À7–10À8, HLA-DRB1*15:01, gender or clinical form did not alter equivalent to a P ¼ 0.05 for a classical epidemiological these results (data not shown). study in which only one hypothesis is being tested) warrants corroboration of the relevant genes.15 Even more so when the proposed associations show less Discussion stringent support (Pp10À5), the reported loci need replication in additional data sets before they can be GWA studies explore the entire human genome uncon- considered reliable susceptibility markers. With this aim, strained by previous hypothesis about the nature of a total of 2863 MS patients and 2930 healthy controls genes that influence complex diseases. However, the from Spain, matched by age, gender and ethnicity were massive number of statistical tests performed in this compared in a case-control study. approach presents an unprecedented potential for false positive results.16 Often, GWAS use multistage designs to warrant replication of the original findings, otherwise, Results the confirmation of association signals through large- scale replication is recommended.15 We selected the top (non-HLA) MS risk markers With this limitation in mind, our aim with the present identified in the GWAS conducted by Baranzini et al.14 work was to corroborate the best-associated variants These polymorphisms were genotyped in a Spanish identified in the GWAS by Baranzini et al.14 The authors cohort of MS patients (n ¼ 2863) and controls (n ¼ 2930) defined top-scoring (non-HLA) markers associated with (see Table 1 for the characteristics of the study group). global MS susceptibility, all of them with modest effects No marker showed deviation from Hardy–Weinberg on disease risk, by performing the screen of allele equilibrium. There were no statistically significant frequencies for 551642 SNPs in 978 MS cases and 883 differences in genotyping calling rates between cases controls. Our scan of these markers in an independent and controls for any of these markers. collection of 2863 MS cases and 2930 controls revealed As shown in Table 2, the analysis of the Spanish replication of the GPC5 polymorphism studied, but no samples solely replicated the reported association of other best hit reached significance. Glypicans regulate rs9523762 in the GPC5 gene with MS susceptibility cell–cell signaling and are expressed during brain (G allele: P ¼ 1.6 Â 10À5, odds ratio ¼ 1.23, 95% confidence neurogenesis17 and they have been found in MS active interval ¼ 1.12–1.36). No other associations with the plaques.18 Other polymorphisms in the GPC5 gene have pathogenesis of the disease were detected in the study been consistently associated with response to interferon population. Statistical power allows detecting the mag- (IFN)-b in MS patients;19,20 however, these polymorphisms nitude of the previously published effects for each (rs10492503, rs9523554/rs9301789) and the one showing single-nucleotide polymorphism (SNP) except for an effect in the present work (rs9523762) are not genetic rs16914086 (Table 2). Similar results were also obtained proxies (r2 ¼ 0.001, D0 ¼ 0.03 and r2 ¼ 0, D0 ¼ 0.06, respec-

Genes and Immunity Replication of top markers of a genome-wide association study ML Cavanillas et al 112 Table 2 Independently replicated genotypic and allelic frequencies of the top (non-HLA) markers previously associated with MS susceptibility in a GWAS (Baranzini et al.14)

SNP Gene Genotype MS Controls P OR (95% CI) OR 80% Published OR (chromosome) Alelles (%) (%) (carrier minor powera (Baranzini allele/minor allele) et al.)14

rs1109670b DDEF2 (2) CC 1170 (52) 1171 (50) CA 901 (40) 969 (42) AA 180 (8) 192 (8) 0.23 0.97 (0.83–1.05) C 3241 (72) 3311 (71) A 1261 (28) 1353 (29) 0.29 0.95 (0.94–1.04) 1.12 1.381 rs651477c EN1 (2) AA 1371 (59) 1440 (60) AG 833 (36) 822 (34) GG 138 (6) 145 (6) 0.37 1.05 (0.94–1.19) A 3575 (76) 3702 (77) G 1109 (24) 1112 (23) 0.51 1.03 (0.94–1.14) 1.13 1.378 rs7607490 TRIB2 (2) GG 2231 (79) 2297 (79) AG 553 (19) 575(20) AA 51(2) 32(1) 0.7 1.02 (0.9–1.17) G 5015 (88) 5169 (89) A 655 (12) 639 (11) 0.35 1.06 (0.94–1.19) 1.18 1.558 rs1841770d ZIC1 (3) TT 646 (32) 709 (33) GT 953 (47) 1078 (49) GG 430 (21) 391 (18) 0.62 1.03 (0.91–1.18) T 2245 (55) 2496 (57) G 1813 (45) 1860 (43) 0.07 1.08 (0.99–1.18) 0.90 0.746 rs7672826 MGC45800 (4) GG 1118 (39) 1180 (41) AG 1333 (47) 1334 (46) AA 401 (14) 393 (13) 0.28 1.06 (0.95–1.18) G 3569 (63) 3694 (63) A 2135 (37) 2120 (36) 0.3 1.04 (0.97–1.13) 1.11 1.366 rs1529316 CSMD1 (8) CC 787 (28) 821 (29) CT 1368 (49) 1402 (49) TT 652 (23) 657(23) 0.69 1.02 (0.91–1.15) C 2942 (52) 3044 (53) T 2672 (48) 2716 (47) 0.65 1.02 (0.95–1.1) 1.11 1.356 rs2049306 CSMD1 (8) AA 741 (26) 773 (27) AC 1339 (47) 1395 (48) CC 742 (26) 718 (25) 0.65 1.03 (0.91–1.16) A 2821 (50) 2941 (51) C 2823 (50) 2831 (49) 0.30 1.04 (0.97–1.12) 1.11 1.362 rs16914086 TBC1D2 (9) GG 2170 (76) 2181(75) GA 639 (22) 670 (23) AA 42 (1) 51 (2) 0.40 0.95 (0.84–1.07) G 4979 (87) 5032 (87) A 723 (13) 772 (13) 0.32 0.95 (0.85–1.06) 0.85 0.886 rs1755289 SH3GL2 (9) CC 1046 (37) 1083 (37) CT 1359 (48) 1358(47) TT 426 (15) 464 (16) 0.79 1.01 (0.91–1.13) C 3451 (61) 3524 (61) T 2211 (39) 2286 (39) 0.74 0.99 (0.92–1.07) 0.90 0.739 rs1458175 PDZRN4 (12) GG 854 (30) 856 (29) GT 1386 (49) 1446 (50) TT 608 (21) 606 (21) 0.65 0.97 (0.87–1.09) G 3094 (54) 3158 (54) T 2602 (46) 2658 (46) 0.98 1.00 (0.93–1.08) 1.11 1.342 rs9523762e GPC5 (13) GG 618 (35) 498 (30) AG 829 (47) 823 (49) AA 300 (17) 365 (22) 0.0003 0.77 (0.66–0.89) G 2065 (59) 1819 (54) A 1429 (41) 1553 (46) 1.6 Â 10À5 0.81 (0.74–0.89) 1.11 1.361

Abbreviations: CI, confidence interval; OR, odds ratio. aOdds Ratios (ORs) as low as those indicated for each polymorphism could be detected at 80% statistical power, a ¼ 0.05; calculated using allelic frequencies. Analyses done excluding bMadrid, cBarcelona, dNorth of Spain or eSouth of Spain heterogeneous populations.

Genes and Immunity Replication of top markers of a genome-wide association study ML Cavanillas et al 113 Table 3 Logistic regression model (adjusted for gender and Concordantly, in our population no replications were carriers of HLA-DRB1*15:01) obtained for the two closely related CSMD1 polymorphisms already mentioned. SNP Minor allele Adjusted P OR (95%CI) One potential explanation for the observed discre- pancy between our study and Baranzini’s original data rs1109670a A 0.962 1.003 (0.902–1.115) could reside in the non-replicated associations being a rs651477 G 0.451 0.961 (0.866–1.066) false-positive results. Alternatively, the linkage disequili- rs7607490 A 0.199 1.091 (0.955–1.247) rs1841770a G 0.137 1.080 (0.976–1.196) brium between the causal variants and the markers rs7672826 A 0.478 1.032 (0.946–1.125) associated in each locus might be different depending on rs1529316 T 0.784 1.012 (0.930–1.101) the population studied. An additional possibility is that rs2049306 C 0.360 0.962 (0.884–1.046) what appears to be a homogeneous MS phenotype rs16914086 A 0.497 0.958 (0.847–1.084) might, in fact, be due to distinct etiologies in different rs1755289 T 0.167 0.941 (0.864–1.026) MS patients in such a way that a genuine effect pertinent rs1458175 T 0.713 1.016 (0.934–1.105) rs9523762a A o0.0001 0.776 (0.695–0.866) to a specific category of patients could pass undetected when analyzing the whole collection.27 Finally, the question of population-specific MS risk genes is as yet Abbreviation: CI, confidence interval. unanswered. Given the lower prevalence of MS in the aAnalysis made without the heterogeneous collection in each case. Mediterranean areas including Spain (prevalence of Bold values indicate significant results. 0.0008, http://www.mult-sclerosis.org/prev_tab.html), it remains to be seen whether the same gene–environ- tively). Given the fact that GPC5 is associated with MS ment interactions or genetic penetrance have a role in MS susceptibility, as well as with the response to IFN-b irrespective of geographic/ethnical background. Stratifi- therapy, it is tantalizing to speculate that the IFN type I cation by gender, clinical forms or by the well-known pathway might have a role in the pathogenesis of the susceptibility factor HLA-DRB1*1501 did not provide disease, at least in a subgroup of patients. clues about any specific effects in a subgroup of patients; Recently, GPC5 rs9523762 results were published in nonetheless, the markers presently studied were African Americans,21 though no significant association reported to influence overall MS risk. Research should with MS was reached (P ¼ 0.154). Interestingly, the risk be undertaken to examine the heterogeneity in the causes allele in African Americans (G) is the same as the one of MS; considering a more refined categorization of cases now found to be associated with MS susceptibility in might change the output.28 Spain, whereas the opposite allele (A) was originally The GWAS reveal the complex nature of MS and the described to be associated in Caucasian individuals.21 current number of loci associated with MS represents The meta analysis of both GPC5 studies in Spanish only a small fraction of the disease heritability, so there and African American subjects (heterogeneity P ¼ 0.43; must be additional genetic contributions to discover.29 As 2 I ¼ 0%) yielded a PMÀHo0.00001; odds ratioMÀH (95% GWAS focus on common genetic variation across the confidence inteval) ¼ 0.83 (0.76–0.90). The initially genome, disease-associated low-frequency polymorph- reported odds ratio (1.361) reflects an effect stronger isms or else copy-number variants would have not been than this one. This well-described phenomenon (the so detected with this approach. Moreover, the etiologic called ‘winner’s curse’) would justify a lack of replication variants are presently ascertained in only a minority of for the other best-associated markers, provided that the loci; therefore, with the fine-mapping of the confirmed original effects were overestimated. Nonetheless, the size genetic loci and the identification of the etiologic risk of our collections would allow us to detect effects even variants, a larger effect might be ascribed to the smaller than those originally reported, as indicated susceptibility genes. New avenues to the identification in Table 2, when applying the standard threshold of of susceptibility genes involved in the pathogenesis of a statistical power (80%, a ¼ 0.05). Thus, evidence is specific disease could be opened through the analysis of mounting for the GPC5 gene being a genuine suscept- the interactions among genes, a strategy that has already ibility factor for MS. Recently, polymorphisms within delivered interesting results.30,31 Certainly, although this gene have been associated with sudden cardiac the most immediate use of GWAS results remains the arrest22 or lung cancer in non-smokers,23 and variants in advance in the understanding of biologic pathways of the adjacent GPC6 gene have also been associated with disease causation,5,32 a robust replication is a crucial first primary sclerosing cholangitis.24 Both genes span around step in this direction and to develop suitable therapeutic 3 Mb in 13q32, a chromosomal region where genomic strategies in the future.33,34 amplification has been reported in different types of tumors.25 Further research is warranted to ascertain the role of these genes in the different pathologies. Materials and methods Two polymorphisms in the CSMD1 gene were among the top associated markers originally found in the GWAS The entire data set studied consisted of 2863 MS patients by Baranzini et al.14: rs1529316 and rs2049306 (r2 ¼ 0.84). diagnosed according to Poser’s criteria35 and 2930 Another polymorphism in this gene, rs1611927, was healthy controls, mostly blood donors and staff, matched recently found associated in a meta-analysis that by age, gender, ethnicity and place of recruitment. included data of the performed GWAS in MS.26 How- Patients and controls were included in the study after ever, the latter variant presents very low correlation with written informed consent and were consecutively the two former CSMD1 variants (r2 ¼ 0.012 and r2 ¼ 0.02, recruited as follows: 460 cases and 462 controls from respectively) and therefore, the meta-analysis results do Hospital Vall d’Hebron and Hospital Clinic (Barcelona, not seem to replicate the associations initially described. Spain); 535 cases and 536 controls from Hospital Clı´nico

Genes and Immunity Replication of top markers of a genome-wide association study ML Cavanillas et al 114 (Madrid, Spain); 777 cases and 706 controls from the 2 Fugger L, Friese MA, Bell JI. From genes to function: the next North of Spain (Hospital de Basurto (Bilbao) and challenge to understanding multiple sclerosis. Nat Rev Hospital de Donostia (San Sebastiań)); 1092 cases and Immunol 2009; 9: 408–417. 1226 from the South of Spain (Hospital Virgen Macarena 3 Ramagopalan SV, Dyment DA, Ebers GC. Genetic epidemiol- (Sevilla), Hospital Carlos Haya (Ma´laga), Hospital Clıńico ogy: the use of old and new tools for multiple sclerosis. (Granada), Hospital Virgen de las Nieves (Granada) and Trends Neurosci 2008; 31: 645–652. Blood Bank (Granada)). The Ethics Committees of these 4 Jersild C, Svejgaard A, Fog T. HL-A antigens and multiple 1 centres approved the study. sclerosis. Lancet 1972; : 1240–1241. 5 Oksenberg JR, Baranzini SE, Sawcer S, Hauser SL. The SNPs were genotyped using the iPLEX Sequenom genetics of multiple sclerosis: SNPs to pathways to patho- MassARRAY platform in the Spanish National Genotyp- genesis. Nat Rev Genet 2008; 9: 516–526. ing Center (CEGEN, Santiago de Compostela, www.cegen. 6 Donnelly P. Progress and challenges in genome-wide associa- org). All DNA was received in the CEGEN centre for tion studies in humans. Nature 2008; 456: 728–731. genotyping in 96-well plates; the concentration of double- 7 Lettre G, Rioux JD. Autoimmune diseases: insights from genome- stranded DNA was assessed using PicoGreen (Molecular wide association studies. Hum Mol Genet 2008; 17:R116–R121. Probes, Inc., Eugene, OR, USA). As quality control, two 8 Rioux JD, Xavier RJ, Taylor KD, Silverberg MS, Goyette P, trios of CEPH samples were genotyped (CEPH Na10830, Huett A et al. Genome-wide association study identifies new Na10831, Na12147, Na10860, Na10861 and Na11984). susceptibility loci for Crohn disease and implicates autophagy The genotypes of these samples resulted in no detection in disease pathogenesis. Nat Genet 2007; 39: 596–604. 9 Wellcome Trust Case Control Consortium. Genome-wide of mendelian inconsistencies and corresponded with the association study of 14 000 cases of seven common diseases ones deposited in HapMap. On the other hand, a 10% of and 3000 shared controls. Nature 2007; 447: 661–678. random samples were subjected to re-genotyping. The 10 Hafler DA, Compston A, Sawcer S, Lander ES, Daly MJ, resulting data were concordant with an average accuracy De Jager PL et al. Risk alleles for multiple sclerosis identified of 499.9%. We performed quality filtering of both by a genomewide study. N Engl J Med 2007; 357: 851–862. samples and SNPs to ensure robust association testing. 11 Australia and New Zealand Multiple Sclerosis Genetics We analyzed only SNPs with high quality data (495% Consortium (ANZgene). Genome-wide association study genotype call rate and Hardy–Weinberg equilibrium identifies new multiple sclerosis susceptibility loci on chromo- P-value40.05 in healthy control subjects). The poly- somes 12 and 20. Nat Genet 2009; 41: 824–828. morphism rs397020 did not satisfy the quality control 12 Hoppenbrouwers IA, Aulchenko YS, Janssens AC, Ramagopalan SV, Broer L, Kayser M et al. Replication of and therefore it was excluded from further analysis and CD58 and CLEC16A as genome-wide significant risk genes for rs908821 did not conform to the chip design restrictions. multiple sclerosis. J Hum Genet 2009; 54: 676–680. The homogeneity of DNA samples from different 13 Rubio JP, Stankovich J, Field J, Tubridy N, Marriott M, locations and Mantel–Haenszel analyses performed to Chapman C et al. Replication of KIAA0350, IL2RA, RPL5 investigate the association of previously reported SNPs and CD58 as multiple sclerosis susceptibility genes in (Review Manager (RevMan) 5.0 (Copenhagen: The Australians. Genes Immun 2008; 9: 624–630. Nordic Cochrane Centre, The Cochrane Collaboration, 14 Baranzini SE, Wang J, Gibson RA, Galwey N, Naegelin Y, 2008)) are shown in Supplementary Tables 1 and 2. The Barkhof F et al. Genome-wide association analysis of suscept- present work has been designed as a replication study; ibility and clinical phenotype in multiple sclerosis. Hum Mol therefore P-values are not corrected. Power calculations Genet 2009; 18: 767–778. were performed by using http://www.sph.umich.edu/ 15 Ioannidis JP, Thomas G, Daly MJ. Validating, augmenting and refining genome-wide association signals. Nat Rev Genet 2009; csg/abecasis/cats/. 10: 318–329. 16 Pearson TA, Manolio TA. How to interpret a genome-wide association study. JAMA 2008; 299: 1335–1344. Conflict of interest 17 Luxardi G, Galli A, Forlani S, Lawson K, Maina F, Dono R. Glypicans are differentially expressed during patterning and The authors declare no conflict of interest. neurogenesis of early mouse brain. Biochem Biophys Res Commun 2007; 352: 55–60. 18 van Horssen J, Bo L, Dijkstra CD, de Vries HE. Extensive extracellular matrix depositions in active multiple sclerosis Acknowledgements lesions. Neurobiol Dis 2006; 24: 484–491. We thank patients with multiple sclerosis and control 19 Byun E, Caillier SJ, Montalban X, Villoslada P, Fernandez O, et al. subjects for making this study feasible. Financial support Brassat D Genome-wide pharmacogenomic analysis of the response to interferon beta therapy in multiple sclerosis. for the study was provided by: Ministerio de Ciencia e Arch Neurol 2008; 65: 337–344. Innovacioń-Feder (SAF2009-11491), Fondo de Investiga- 20 Cenit MD, Blanco-Kelly F, de las Heras V, Bartolome M, de la cioń Sanitaria FIS (RETICS-REEM RD07/0060, PI0902105, Concha EG, Urcelay E et al. Glypican 5 is an interferon-beta PI081636, PI070353) Junta de Andalucıá (P07-CVI-02551), response gene: a replication study. Mult Scler 2009; 15: 913–917. Fundacioń Ilundain, and Ikerbasque, the Basque Founda- 21 Johnson BA, Wang J, Taylor EM, Caillier SJ, Herbert J, Khan tion for Science (Bilbao). SNP genotyping services were OA et al. Multiple sclerosis susceptibility alleles in African provided by the Spanish ‘Centro Nacional de Genotipado’ Americans. Genes Immun 2010; 11: 343–350. CEGEN-USC, www.cegen.org. 22 Arking DE, Reinier K, Post W, Jui J, Hilton G, O’Connor A et al. Genome-wide association study identifies GPC5 as a novel genetic locus protective against sudden cardiac arrest. PLoS One 2010; 5: e9879. References 23 Li Y, Sheu CC, Ye Y, de Andrade M, Wang L, Chang SC et al. Genetic variants and risk of lung cancer in never smokers: 1 Frohman EM, Racke MK, Raine CS. Multiple sclerosis–the a genome-wide association study. Lancet Oncol 2010; 11: plaque and its pathogenesis. N Engl J Med 2006; 354: 942–955. 321–330.

Genes and Immunity Replication of top markers of a genome-wide association study ML Cavanillas et al 115 24 Karlsen TH, Franke A, Melum E, Kaser A, Hov JR, Balschun T 30 Gourh P, Agarwal SK, Divecha D, Assassi S, Paz G, et al. Genome-wide association analysis in primary sclerosing Arora-Singh RK et al. Polymorphisms in TBX21 and STAT4 cholangitis. Gastroenterology 2010; 138: 1102–1111. increase the risk of systemic sclerosis: evidence of possible 25 Yu W, Inoue J, Imoto I, Matsuo Y, Karpas A, Inazawa J. GPC5 gene-gene interaction and alterations in Th1/Th2 cytokines. is a possible target for the 13q31-q32 amplification detected in Arthritis Rheum 2009; 60: 3794–3806. lymphoma cell lines. J Hum Genet 2003; 48: 331–335. 31 Perdigones N, Vigo AG, Lamas JR, Martı´nez A, Balsa A, 26 De Jager PL, Jia X, Wang J, de Bakker PI, Ottoboni L, Pascual-Sacedo D et al. Evidence of epistasis between Aggarwal NT et al. Meta-analysis of genome scans and TNFRSF14 and TNFRSF6B polymorphisms in patients with replication identify CD6, IRF8 and TNFRSF1A as new multiple sclerosis susceptibility loci. Nat Genet 2009; 41: 776–782. rheumatoid arthritis. Arthritis Rheum 2010; 62: 705–710. 27 Weiner HL. The challenge of multiple sclerosis: how do we 32 Hirschhorn JN. Genomewide association studies–illuminating cure a chronic heterogeneous disease? Ann Neurol 2009; 65: biologic pathways. N Engl J Med 2009; 360: 1699–1701. 239–248. 33 Carroll WM. Oral Therapy for Multiple Sclerosis – Sea Change 28 Zuvich RL, McCauley JL, Pericak-Vance MA, Haines JL. or Incremental Step? N Engl J Med 2010; 362: 456–458. Genetics and pathogenesis of multiple sclerosis. Semin 34 Lopez-Diego RS, Weiner HL. Novel therapeutic strategies Immunol 2009; 21: 328–333. for multiple sclerosis–a multifaceted adversary. Nat Rev Drug 29 Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Discov 2008; 7: 909–925. Hunter DJ et al. Finding the missing heritability of complex 35 Poser CM. Revisions to the 2001 McDonald diagnostic criteria. diseases. Nature 2009; 461: 747–753. Ann Neurol 2006; 59: 727–728.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity