As genotype will affect exposure over a lifetime, MR can in principle allow for more accurate estimation of the magnitude of a causal effect than a direct assessment taken at a single time point [19] although for the same reason it may over-estimate the likely magnitude of an intervention effect. For example, an intervention delivered in middle age will only partially reduce the lifetime exposure to a risk factor that is estimated from MR analyses. Commonly, the association between a genetic variant and the exposure, and between the genetic variant and the outcome, E7080 manufacturer are estimated in the same sample. However, this may not always be possible if
exposure and outcomes are not
measured in the same samples, or if the exposure has only been measured in a subset of the total sample [20••]. In two sample MR, the genotype-exposure and genotype-outcome associations are estimated in different samples and these estimates then combined to provide an estimate of the causal exposure-outcome association [21•]. As both of these parameters are estimates, the standard error of the exposure-outcome association needs to be adjusted using appropriate methods [20••]. Two sample MR does not usually lead to a substantial loss of statistical power [21•], so this type of design may be a more cost effective approach [20••]. Establishing that an association is causal is valuable in itself, but of potentially greater interest this website is establishing the mechanism through which this causal association operates. It may be possible to investigate causal mechanisms between an exposure and an outcome using a two-step MR approach [22]. This type of analysis requires a genetic
variant which associates with the exposure of interest and a separate genetic variant which associates with the mediating factor of interest. For example, there is growing interest in the role of epigenetic mediators of environmental exposures, but epigenetic markers (as with any other biomarker) are vulnerable to confounding and reverse causality. Here, a genetic proxy for the exposure of interest is used to assess the causal relationship between the environmental exposure and a Thymidylate synthase potential mediator such as methylation (step 1, see Figure 4a). Next, a genetic proxy for the mediator (here, DNA methylation) is used to interrogate the causal relationship between the mediator and the outcome of interest (step 2, see Figure 4b). This approach enables a triangulation of evidence to infer a mediating role for, in this case, methylation in the causal pathway between the environmental exposure and the outcome of interest. It can in principle be applied to other potential mediators (e.g., metabolite levels).