Meta-analysis of GWAS for Cardiovascular Function
After my post yesterday on the JAMA article failing to find an association between 5-HTTLPR serotonin transporter gene, stressful life events and depression, I read another recent JAMA article of a meta-analysis of a huge genome-wide association test of 2.5 million SNPs for echocardiographic measures of cardiovascular structure and function. In this large collaborative effort, the investigators combined Framingham, Rotterdam, MONICA cohorts and several other cardiovascular cohorts in the 1st stage sample (discovery) and examined which SNPs were associated with cardiovascular dysfunction. Then in the 2nd stage sample they assessed which associated SNPS could replicate an association with the phenotype. With that dense of a scan (they used stringent significance criteria for multiple comparisons) they only found a few SNPs which explained only 1-3% of the variance!! How frustrating to have put in such a huge effort of $ and time to get such a little return on the effort. Now the problem is trying to figure out if it is the phenotype (cardiovascular structure, CS) that isn’t that heritable, or whether CS is not controlled by typical genomic polymorphisms (easily could be epigenic mechanism or mitochondrial effect) or if genome-wide scans are just not that adept at identifying small genetic susceptibility variants. I think the issue cuts across all disease/phenotypes and in particular those phenotypes which have strong environmental components. The utility and cost-effectiveness of GWAS studies has been under debate recently in New York Times article and discussed on the Genetic Future blog as well as in a NEJM article last April. The saga will continue, and it will be interesting to see what methods develop to analyze genetic data and if even more novel genetic mechanisms come to light.
