Birth Weight and Stroke in Adult Life: Genetic Correlation and Causal Inference With Genome-Wide Association Data Sets.

Objective: Prior studies showed there was an inverse association between birth weight and stroke in adulthood; however, whether such association is causal remains yet known and those studies cannot distinguish between the direct fetal effect and the indirect maternal effect. The aim of the study is to untangle such relationship using novel statistical genetic approaches. Methods: We first utilized linkage disequilibrium score regression (LDSC) and Genetic analysis incorporating Pleiotropy and Annotation (GPA) to estimate the overall genetic correlation between birth weight and stroke. Then, with a set of valid birth-weight instruments which had adjusted fetal and maternal effects, we performed a two-sample Mendelian randomization (MR) to evaluate its causal effect on stroke based on summary statistics from large scale genome-wide association study (GWAS) (n=264,498 for birth weight and 446,696 for stroke). We further validated the MR results with extensive sensitivity analyses. Results: Both LDSC and GPA demonstrated significant evidence of shared maternal genetic foundation between birth weight and stroke, with the genetic correlation estimated to -0.176. However, no fetal genetic correlation between birth weight and stroke was detected. Furthermore, the inverse variance weighted MR demonstrated the maternally causal effect of birth weight on stroke was 1.12 (95% confidence interval [CI] 1.00 - 1.27). The maternal ORs of birth weight on three subtypes of stroke including cardioembolic stroke (CES), large artery stroke (LAS) and small vessel stroke (SVS) were 1.16 (95% CI 0.93 - 1.43), 1.50 (95% CI 1.14 - 1.96) and 1.47 (95% CI 1.15 - 1.87), respectively. In contrast, no fetal causal associations were found between birth weight and stroke or the subtypes. Those results were robust against extensive sensitivity analyses, with Egger regression ruling out the possibility of pleiotropy and multivariable MR excluding the likelihood of confounding or mediation effects of other risk factors of stroke. Conclusion: This study provides empirically supportive evidence on the fetal developmental origins of stroke and its subtypes. However, further investigation is warranted to understand the pathophysiological role of low birth weight in developing stroke.