Pleiotropy-robust Mendelian Randomization

Abstract

Background: The potential of Mendelian Randomization studies is rapidly expanding due to (i) the growing power of GWAS meta-analyses to detect genetic variants associated with several exposures, and (ii) the increasing availability of these genetic variants in large-scale surveys. However, without a proper biological understanding of the pleiotropic working of genetic variants, a fundamental assumption of Mendelian Randomization (the exclusion restriction) can always be contested. Methods:We build upon and synthesize recent advances in the econometric literature on instrumental variables (IV) estimation that test and relax the exclusion restriction. Our Pleiotropy-robust Mendelian Randomization (PRMR) method first estimates the degree of pleiotropy, and in turn corrects for it. If a sample exists for which the genetic variants do not affect the exposure, and pleiotropic effects are homogenous, PRMR obtains unbiased estimates of causal effects in case of pleiotropy.Results: Simulations show that existing MR methods produce biased estimators for realistic forms of pleiotropy. Under the aforementioned assumptions, PRMR produces unbiased estimators. We illustrate the practical use of PRMR by estimating the causal effect of (i) cigarettes smoked per day on Body Mass Index (BMI); (ii) prostate cancer on self-reported health, and (iii) educational attainment on BMI in the UK Biobank data.Conclusions: PRMR allows for instrumental variables that violate the exclusion restriction due to pleiotropy, and corrects for pleiotropy in the estimation of the causal effect. If the degree of pleiotropy is unknown, PRMR can still be used as a sensitivity analysis.

Beyond plausibly exogenous

“…It thereby provides a well‐informed way to perform sensitivity analyses in IV estimation. In a companion epidemiological paper (van Kippersluis and Rietveld, ), we applied this idea in the context of genetic variants as instrumental variables. Here we apply the approach in general IV settings, and illustrate our procedure by estimating the effect of (a) attending Catholic high school on schooling outcomes and (b) the number of children on female labour supply.…”

“… An intuitive example is discussed in van Kippersluis and Rietveld ( ), where certain genetic variants are used as IVs for prostate cancer to study its effect on self‐reported health. Since prostate cancer naturally is only a risk factor among males, the first stage among females is zero by construction.…”

On the test of exclusion restriction in linear instrumental variable regressions

“…The idea is to identify a sample over which it is known, preferably by design, that the candidate IV is not related to the treatment. For example, van Kippersluis and Rietveld discussed an application where certain genetic variants are used as IVs for prostate cancer to study its effect on self‐reported health. Because prostate cancer is a risk factor only for males, females provide such a sample.…”

Detecting and correcting for bias in Mendelian randomization analyses using gene-by-environment interactions

“…Conley et al(17) emphasise the potential trade-off between instrument strength and degree of IV3 violation in putting forward the notion of plausibly endogeneity, whilst further works underlining the utility of using instrument-covariate interactions have also emerged, such as those of Gennetian et al and Small(18, 19). It is also worth noting that a similar method, entitled Pleiotropy Robust Mendelian Randomization (PRMR), has recently been put forward with a similar intuitive framework(20). However, the PRMR approach requires a subgroup for which the instrument and the exposure are independent to be observed, potentially limiting the applicability of the approach in contrast to approaches which estimate the expected association at a hypothetical independent subgroup.…”