Abstract
Purpose:
Glaucoma is a phenotypically and genetically complex neurodegenerative disease that is the second leading cause of blindness worldwide. Known genetic risk loci fail to fully account for the genetic component of glaucoma. Since our prior investigations have implicated the estrogen pathway in primary open angle glaucoma (POAG) (Pasquale LR et al, Molecular Vision), we investigated whether sex-specific risk loci could be detected in an expanded POAG phenotype-genotype dataset.
Methods:
The NEIGHBORHOOD consortium dataset consists of eight separate population samples with genome-wide association data, typed on different genome-wide SNP arrays, imputed to the March 2012 version of the 1000 Genomes with IMPUTE2 and/or MaCH/miniMaC. Each dataset was filtered for minor allele frequency (≥0.05) and imputation quality (≥0.7). Datasets were stratified by gender and study-specific dosage analysis was applied in ProbABEL to evaluate the estimated genotypic probabilities from the imputation step. Study-specific logistic regression models were adjusted for age, significant Eigenvectors, and study-specific covariates. The datasets were then meta-analyzed applying the inverse variance weighted method in METAL and applying genomic control correction.
Results:
The meta-analyzed NEIGHBORHOOD datasets included 1,691 male and 2,160 female POAG cases and 4,367 male and 29,111 female examined normal controls. A total of 7,301,525 variants passed quality control parameters in at least one dataset. Top hits were evaluated in the ANZRAG case control set, which included 563 male and 592 female POAG cases and 1,270 male and 722 female controls. Meta-analysis of the nine datasets revealed male-specific associations attaining genome-wide significance in the ALDH9A1/TMCO1 region of chromosome 1 (rs6426936, P=5.0x10-10), and on chromosome 22 near TXNRD2 (rs35934224, P=7.87x10-10). Interestingly, these associations did not attain genome-wide significance in females (rs6426936 , PFemales=3.58x10-6; rs35934224, PFemales=6.30x10-3), despite a larger sample size.
Conclusions:
Imputation of genome-wide array data extends the genomic coverage beyond what can be interrogated by a single or multiple arrays. Meta-analyzing multiple imputed datasets and investigating subsets therein is useful for studying common diseases with complex inheritance, such as glaucoma, and will help fully define the underlying genetic architecture.