The mediating effect of blood lipid indicators on the causal relationship between diabetes and stroke by two-sample bidirectional Mendelian randomization
Objective To explore the mediating effect of blood lipid indicators[high density lipoprotein cholesterol(HDLC),triglyceride(TG)]on the causal relationship between diabetes and stroke by Mendelian randomization(MR).Methods The single nucleotide polymorphisms(SNPs)from published available genome-wide association study(GWAS)were selected as instrumental variables.The inverse-variance weighted method was used as the main analysis method to explore the causal relationship of diabetes and blood lipid indicators with stroke,in combination with complementary analyses of MR-Egger regression,weighted median,simple mode and weighted mode methods.Reverse MR analysis was used to detect reverse causality of diabetes and blood lipid indicators with stroke.The IVW and MR-Egger methods were used to test Cochran's Q heterogeneity,the MR-Egger regression was used to evaluate the horizontal pleiotropy,and the leave-one-out analysis was adopted to exclude the possible influence of individual SNP on the overall results.Mediation analysis was conducted to investigate the potential mediating effect of blood lipid indicators.Results The result of IVW showed that diabetes(OR=17.017,95%CI 7.377-39.252),HDLC(OR=0.724,95%CI 0.069-0.862),TG(OR=1.137,95%CI 1.047-1.235)were significantly associated with stroke;Diabetes was significantly associated with HDLC(OR=0.652,95%CI 0.501-0.849)and TG(OR=3.046,95%CI 1.467-6.323).The P values were all greater than 0.05 by MR-Egger regression method,indicating that there was no horizontal pleiotropy in genetic instrumental variables.Reverse MR indicated reverse causality was not significant between stroke and diabetes,between HDLC and diabetes,between TG and diabetes.The mediating effect of HDLC in diabetes and stroke was 0.137,and the proportion of mediating effect was 4.851%.The mediating effect of TG in diabetes and stroke was 0.143,and the proportion of mediating effect was 5.031%.Conclusions There was positive causal relationship between diabetes and stroke,and HDLC and TG have a mediating effect in the causal relationship between diabetes and stroke.
DiabetesStrokeMendelian randomizationMediumHigh density lipoprotein cholesterolTriglyceride