Towards More Reliable Mendelian Randomisation Investigations
Abstract:
Mendelian randomization is a technique for assessing the causal role of a modifiable risk factor on a disease outcome using genetic data. Recent advances in genome-wide association studies and the increasing availability of publicly available summary data on associations of genetic variants with risk factors and disease outcomes in large sample sizes have enabled powerful Mendelian randomization analyses to be performed relatively quickly and simply. However, these analyses typically use a large number of genetic variants. When at least one of the genetic variants does not satisfy the instrumental variable assumptions, causal estimates will be biased and Type 1 error rates inflated. Two novel methods are presented for obtaining causal inferences from summarized data under weaker assumptions that those of a typical Mendelian randomization investigation. These methods are Egger regression, a method adopted from the meta-analysis literature for dealing with publication bias, and a median-based approach. The talk is illustrated using the example of HDL-cholesterol on coronary artery disease risk: a naive analysis including all genome-wide significant variants suggests a protective effect of HDL-c that is not supported by the biological evidence, whereas the Egger and weighted median approaches suggest a null causal effect.
Admission