Studies that assess interaction between (typically) one genetic variant and a non-genetic factor with respect to a given outcome. These can be useful in several MR contexts. For example, some genetic variants for smoking, alcohol and other risk factors act on a metabolite or mechanism that is caused by the risk factor but does not directly affect levels of the risk factor. In these circumstances, if the risk factor causes the outcome of interest, we would only expect the genetic variant to associate with the outcome in those who have ever smoked / drunk alcohol or been exposed to the risk factor. In those who have never been exposed, we would not expect an association between the gene and the outcome. Thus, a G×E, where E is ever versus never exposure, would provide MR evidence for a causal effect.
Sometimes, G×E can be used to investigate mechanisms. For example, if the genetic variant blocks the postulated mediating mechanism from the exposure to the outcome, then the exposure would not be expected to be associated with the outcome among carriers of the mediator-blocking genotype. G×E can be exploited to mitigate bias due to horizontal pleiotropy in MR. Adequate statistical power typically requires very large samples. Many published studies did not properly adjust for potential confounders.
- Keller MC. Gene x environment interaction studies have not properly controlled for potential confounders: the problem and the (simple) solution. Biol Psychiatry 2014;75:18-24.
- Davey Smith G. Use of genetic markers and gene-diet interactions for interrogating population-level causal influences of diet on health. Genes and Nutrition 2011;6:27-43.