Genetics Journal Club
Cara Skraban, MD
Clinical Genetics Fellow
February 12, 2015
JAMA Neurology Feb 2015
Myasthenia Gravis
• Autoimmune disorder of neuromuscular transmission
• Characterized by muscle fatigability
• Typically mediated by antibodies against nicotinic
acetylcholine receptors (AChRs) or against related
proteins at NM junction
– Muscle-specific tyrosine kinase (MuSK)
– Lipoprotein receptor-related protein 4
– Agrin
• Bimodal affected populations
– Young women
– Older men
Myasthenia Gravis
Genetics Factors of MG
• HLA locus is the most strongly associated risk factor for
disease
• Previous GWAS studies
– Major histocompatibility complex class II
– Protein tyrosine phosphatase nonreceptor type 22
(PTPN22)
– TNFAIP3 interacting protein 1 (TNIP1)
• Gene studies have suggested association of cytotoxic Tlymphocyte-associated protein 4 gene (CTLA4)
• Patients often have a family history of autoimmune
disease
• 5% of patients have a positive family history of MG
following an AD inheritance
Patients
• Patients attending MG clinics at 14 centers throughout
North America (972 patients)
–
–
–
–
Diagnosed by a neurologist specializing in MG
Onset of symptoms after 18 yo
Non-Hispanic white race
Diagnosed clinically and confirmed with anti-AChR
antibodies
– Samples collected using Oragene DNA Saliva Collection kits
• Control Cohort (1977 patients)
– Downloaded genotype data from dbGAP
– Neurologically normal individuals
– Matched for race and ethnic group, not age and sex
Replication Cohort
• 423 Italian patients with AChR-positive MG
• 467 Italian neurologically normal controls
• Matched to the case cohort for race/ethnic
group but not for age or sex
• Blood samples collected
Patients
Genome-wide Genotyping
• Genotyped in the Laboratory of Neurogenetics,
National Institute of Aging, using
HumanOmniExpress BeadChips (Illumina)
– Assay 730,525 SNPs across the genome
• Control cohort previously genotyped at the
Center for Inherited Disease Research at Hopkins
on HumanOmni1-Quad BeadChips (Illumina)
• Analyses were confined to the 677,673
autosomal SNPs that were common to both chips
Genotyping Bias
• To exclude possibility of genotyping bias arising from
different sources of DNA, they compared from two
patients:
– whole genome genotyped data (Illumina) generated using
paired DNA samples extracted from blood
– DNA extracted from saliva using Oragene DNA Saliva Collection
system
– Concordance rate >99.99%
– None of discordant SNPs were located within significantly
associated loci
• Exclude genotyping bias from using amplified DNA, they
compared from 94 samples:
– Sanger sequencing data generated using DNA samples that were
amplified
– Data generated using genomic, unamplified DNA
– Concordance rate 100% for both rs601006 and rs9271850
Genotyping in the Replication Cohort
• RS231770, rs4263037, rs9270986
– Taqman genotyping assays
– Scanned on an ABI 7900HT Real-Time PCR
• Rs601006 and rs9271850
– Sequencing using Big-Dye Terminator version 3.1
sequencing kit
– Run on an ABI 3730xl DNA analyzer
– Analyzed with Sequencher software and Mutation
Surveyor
Statistical Analysis: Genome-wide
Association
• Statistical analyses were performed using R
statistical software
• Standard quality-control procedures; Exclusion
of the following:
– SNP call rates of less than 95%
– Non-European ancestry
– Cryptic relatedness- identity-by-descent > 0.1
– Minor allele frequency <0.01 in the control cohort
– Hardy-Weinberg equilibrium P < 0.001 in the
control cohort
Imputation
• Markov chain-based Haplotyper to impute
genotypes
– Imputed by a two-stage design
– Confirmed accuracy of imputation for most associated
SNPs for the 972 MG patients
• Taqman genotyping for rs231770
• Sanger sequencing for rs601006 and rs9271850
• High concordance for all: 99.8%, 98%, 100%
• 8,114,394 SNPs available for analysis
• 513,081 genotyped SNPs
• 7,601,313 imputed SNPs
Statistical Analysis Continued
• P values calculated using logistic regression
modeling
– First two principle components used as covariates
to compensate for any residual population
stratification.
– Principle components were generated using
Genome-wide Complex trait analysis software
package implementation of eigenstrat
– Threshold of 5.0 x 10-8 for genome wide
significance after Bonferroni correction
Probability Analysis and Heritability
Estimates
• Density estimation was used to generate
posterior probabilities of developing MG based
on sex and age
• Genome-wide Complex Trait Analysis
– Used to compare each case series to control
individuals (all cases, early-onset, late-onset)
– Compared two separate sets of SNPs
• All genotyped SNPs
• Only those within 1 MB from the loci identified as genomewide significant in the discovery phase
• Only SNPs passing quality control were used to evaluate the
heritability
Loci Showing Genome-Wide
Association with Myasthenia Gravis
Quartile-Quartile Plot
All cases
Bimodal Sex Distribution
Early and Late Onset Cases
Replication Cohort
• 3 SNPs from the risk loci identified in the
overall cohort for genotyping in the replication
cohort of 423 Italian AChR antibody-positive
MG cases and 467 controls.
• Strongest signals
– rs9270986 in the intergenic region between HLADRB1 and HLA-DQA1
– rs231770 located 3.3 kb upstream of CTLA4
Combined Analysis
Early-onset Replication Cohort
Late-onset Replication Cohort
Summary of Results
• Overall case-control cohort
– CTLA4 (rs231770)
– HLA-DQA1 (rs9271871)
– TNFRSF11A (rs4263037)
• Replicated for CTLA4 and HLA-DQA1 in the
Italian cohort
Summary of Results
• Early and late-onset disease have distinct, but
overlapping, genetic architecture
– Genetic variation within TNFSRF11A locus drives
susceptibility to disease among older cases
– Different haplotypes across the same HLA region on
chromosome 6 were identified in early and late-onset
cases
– CTLA4 exerts significant effect regardless of age at
symptom onset, suggesting it plays a central role in
generating the aberrant autoimmune response that
leads to neuromuscular junction dysfunction
HLA-DQA1
CTLA4
TNFSF11A
• 4.5-kDa receptor activator of nuclear factor-K
B expressed on the surface of antigenpresenting dendritic cells.
• Important regulator of the interaction
between T cells and dendritic cells that is
essential for immune surveillance and
regulation of specific immunity
Study Limitations
•
•
•
•
Sample size
Possible population stratification
Lack of age and sex match controls
AChR antibody positive only patients (85%
MG)
• Different results from previous studies
– Different signal in MHC region
– No association of PTPN22 and TNIP1