IMSGC Genome Wide Association Study of Multiple Sclerosis

The goal of this study is to perform a comprehensive allelic and genotypic association analysis of the entire human genome
in multiple sclerosis. The recent definitive linkage genome screen demonstrated that there is no other MS risk gene with an
effect size anywhere near that of the MHC. However, linkage analysis is significantly hampered by reduced power in the face
of heterogeneity and requires multiplex families, which also hampers acquiring an appropriate sample size. In contrast,
genotyping 500K SNPs allows us to survey a significant amount of the genome (we estimate >70%) directly for allelic
or genotypic association. This uses the improved power of association analysis and can also take advantage
of the linkage disequilibrium relationships among SNPs to further increase power (e.g. haplotype analysis). Quality control
and data analysis are significant challenges. We will initially perform substantial QC checks and analyze the data using both
TDT and AFBAC approaches. Multigenic interactions will also be tested using MDR.]]>
The Multiple Sclerosis International Genetics Consortium Diagnostic Criteria FormThe International Multiple Sclerosis Genetics Consortium (IMSGC) Requirements and Procedures for Data SubmissionQuality Control Sample and SNP filtering was done by investigators.TDT association analysis was replicated in pre-compute at NCBI.
Quality control
A total of 1,003 parent-affected child trios were initially identified for
genotyping. After tests for DNA quality and quantity and Mendelian errors
(using the Sequenom fingerprint SNP results), a total of 960 trios were
genotyped using both Affymetrix chips. Results due to systematic errors and
families with false paternity were removed. Preliminary analysis uncovered a
relationship between the BRLMM-generated quality score (this score is the
output for each SNP genotype call by the BRLMM program), and genotypes with
Mendelian errors ( < 1% of genotypes had a quality score > 0.3 as defined in
http://www.broad.mit.edu/gen_analysis/genotyping/brlmm_affy_ncrr.html, but
they contributed over 20% of the genotypes with Mendelian errors). Therefore,
any genotype with a BRLMM quality score > 0.3 was removed. The results then
underwent quality control procedures implemented by a software package (WASP;
Whole-genome Association Study Pipeline) that included the examination of SNP
and sample genotyping efficiency, allele frequencies, checks of Mendelian
errors, and sex-assignment discrepancies (based on available chromosome X
genotypes), and tests of Hardy-Weinberg Equilibrium. We used the Graphical
Representation of Relationships (GRR) program7 to test for relationships
between samples and to detect sample switches, duplications, or contamination.
Potential population stratification was examined as follows. From the 334,923
SNPs, we chose 1000 (with > 1Mb physical spacing) that had maximal chi-square
differences in allele frequency between HapMap populations (500 for the
CEU/YRI comparison, and 500 for the CEU/CHB+JPT comparison). The analysis
International Multiple Sclerosis Genetic Consortium, N Engl J Med 2007;3576
(using the STRUCTURE program8) was seeded with data from unrelated individuals
(60 CEU, 60 YRI, 45 CHB, 45 JPT). No significant stratification was observed.
Further, we used the results of these analyses to determine if any individual
appeared to be of non-European descent. We removed any individual who had a
probability of < 90%, as given by the Structure program, of having all
European-derived parents and grandparents. This procedure removed 34, 21, and
15 individuals from the screening, NIMH, and WTCCC datasets, respectively. All
samples were examined for exclusion using the final SNP dataset (334,923
SNPs). These analyses identified 29 trios with problematic data leaving a
final set of 931 trios for our primary screen.

SNP exclusion criteria
To advance to the whole-genome association study, SNPs had to pass the
following quality control criteria: not mapping to multiple locations in the
genome (3,605 SNPs excluded); a > 95% genotype call rate in the total 2,880
(960 trio) samples (58,435 SNPs excluded); allele frequency (MAF) > 0.005 in
the parents (59,887 SNPs excluded); Hardy Weinberg equilibrium with P > 1 x
10-4 in parents (16,881 SNPs excluded); > 99% genotype call rate if the SNP had
a MAF <10% (due to platform-specific observations demonstrating a difficulty
in calling rare alleles) (74,492 SNPs excluded); and <10 total Mendelian
errors (below the 95% cut-off assuming a binomial distribution for a 0.5%
error rate) (13,136 SNPs excluded). In total, 334,923 SNPs survived these
exclusion criteria and were included in the analysis.
]]>An MS specialist confirmed an MS diagnosis for collection site participants in this study. The diagnosis of MS was made according
to the McDonald criteria after review of all available clinical data. Complete clinical data was available for over 95.0% of patients.
Disease course was recorded for patients at entry to study as either relapsing-remitting (RR), secondary progressive (SP), primary
progressive (PP), or relapsing-progressive (PR) 4. Disability was also assessed at entry with the Expanded Disability Status Scale (EDSS)
5. SPMS was characterized by six months of worsening neurological disability not explained by relapse and measured as a deterioration of
either: i) one or more points on the EDSS in patients with an EDSS less than 6, or ii) a one-half point or more for those with an EDSS > 6.
PPMS was characterized both by 1) progressive clinical worsening for more than 12 months from symptom onset without any relapses, and 2)
abnormal cerebrospinal fluid (CSF) as defined by the presence of two or more oligoclonal bands or elevated IgG index in the CSF. If acute
relapses were superimposed on this steadily progressive course, patients were considered to have PRMS. CIS was characterized as the first,
well-defined neurological event lasting more than 48 hours, and which involved the optic nerve, spinal cord, brainstem, or cerebellum. In
patients presenting with a CIS, the presence of two or more hyper-intense lesions on a T2-weighted MRI sequence was also required for
enrollment into the study. A small number of subjects with clinically isolated syndrome (CIS) were included among the US samples used for
this study. For the purpose of the analysis, they were treated as cases of "relapsing-remitting" cases of MS since they represent one tail
of the distribution of the demyelinating disease spectrum and a majority of them will go on to have a second attack or show evidence of
disease progression.]]>