Supplementary Material for: Causal effect of lipoprotein-associated phospholipase A2 activity on ischemic stroke : a Mendelian randomization study

Background
The relationship between ischemic stroke (IS) and lipoprotein-associated phospholipase A2 (Lp-PLA2) activity is still unclear, and there is a dearth of stratified research on the relationship between Lp-PLA2 activity and different IS subtypes. Therefore, Mendelian randomization was used in this study to examine the relationship between genetically proxied Lp-PLA2 activity and the risks of IS and its subtypes.
Methods
Based on information from a meta-analysis of GWAS, which included 13,664 European people, five single-nucleotide polymorphisms related to Lp-PLA2 activity were chosen as instrumental variables. Summary statistics information about MEGESTROKE consortium with the European group (40,585 cases and 406,111 controls) include any ischemic stroke (AIS) (n = 34,217); large-artery stroke (LAS, n=4,373), cardioembolic stroke (CES, n=7,193) and small vessel stroke (SVS, n=5,386). In order to determine the causal relationships between Lp-PLA2 activity and IS as well as its subtypes, the inverse variance weighted (IVW) approach was chosen as the primary analysis. Significant estimates were then tested by sensitivity analysis to rule out heterogeneity and pleiotropy.
Results
IVW showed Lp-PLA2 activity was causally associated with LAS (OR=3.25, 95% CI=1.65-6.41, p=0.0007), but not with other subtypes of stroke. Sensitivity analysis for causal estimates between Lp-PLA2 activity and LAS showed no significant heterogeneity or pleiotropy.
Conclusions
These MR analyses support a causal effect of Lp-PLA2 activity on LAS but not on AIS, CES or SVS, which suggests that serum Lp-PLA2 activity might be a biomarker for prediction of LAS.