Modifiable Lifestyle Factors and Risk of Stroke: A Mendelian Randomization Analysis

Abstract

Background and purpose: Assessing whether modifiable risk factors are causally associated with stroke risk is important in planning public health measures, but determining causality can be difficult in epidemiological data. We evaluated whether modifiable lifestyle factors including educational attainment, smoking, and body mass index are causal risk factors for ischemic stroke and its subtypes and hemorrhagic stroke.

Methods: We performed 2-sample and multivariable Mendelian randomization to assess the causal effect of 12 lifestyle factors on risk of stroke and whether these effects are independent.

Results: Genetically predicted years of education was inversely associated with ischemic, large artery, and small vessel stroke, and intracerebral hemorrhage. Genetically predicted smoking, body mass index, and waist-hip ratio were associated with ischemic and large artery stroke. The effects of education, body mass index, and smoking on ischemic stroke were independent.

Conclusions: Our findings support the hypothesis that reduced education and increased smoking and obesity increase risk of ischemic, large artery, and small vessel stroke, suggesting that lifestyle modifications addressing these risk factors will reduce stroke risk.

Keywords: Mendelian randomization analysis; body mass index; educational status; genetics; risk factors; smoking; stroke.

Figure 1.

Mendelian randomization results showing causal estimates for association of lifestyle traits with stroke and its subtypes. Colors show magnitude and direction of the association P value (truncated at P<1×10⁻⁸) for causal effect estimates using the inverse-variance weighted Mendelian randomization approach. * indicates significant associations (false discovery rate, q<0.05). AS indicates all stroke; AIS, any ischemic stroke; BMI, body mass index; CES, cardioembolic stroke; ICH, intracerebral hemorrhage; LAS, large artery stroke; SVS, small vessel stroke; and WHR, waist-hip ratio.

Figure 2.

Genetic associations of lifestyle traits and stroke subtypes for significant causal estimates. The associations of stroke and stroke subtypes are shown for (A) education, (B) smoking, (C) waist-hip ratio (WHR), and (D) body mass index (BMI). The associations of each genetic variant with significant (false discovery rate, q<0.05) causal estimates are plotted against their association with the corresponding outcome. Circles represent the associated change in levels of the trait and corresponding increased risk for each variant. Horizontal and vertical lines through each circle represent the corresponding 95% CIs for the genetic associations. Associations were oriented to the effect allele of each trait. Colored lines show the slope (causal estimate) of the trait on the outcome obtained using a variety of different Mendelian randomization (MR) approaches. AS indicates all stroke; AIS, any ischemic stroke; ICH, intracerebral hemorrhage; LAS, large artery stroke; and SVS, small vessel stroke.

Figure 3.

Mendelian randomization (MR) associations between genetically predicted lifestyle factors and stroke subtypes. Results derived from random-effects inverse–variance weighted MR analyses. * indicates significant associations for the causal effect estimates (false discovery rate, q<0.05). Results are shown for (A) all stroke (AS), (B) any ischemic stroke (AIS), (C) intracerebral hemorrhage (ICH), (D) cardioembolic stroke (CES), (E) large artery stroke (LAS), and (F) small vessel stroke (SVS). BMI indicates body mass index; and WHR, wait-hip ratio.