Assessing long-term pleiotropic effects of potential novel triglyceride-lowering medications using variants identified by Mendelian randomization

Abstract

Drugs targeting triglyceride (TG)-associated genes have the potential to improve cardiovascular outcomes. However, we know little regarding the potential additional benefits or deleterious effects of such targeting. As such, in ancestry-specific cohorts of European ancestry (EA) and African ancestry (AA) at Vanderbilt (BioVU), we used PheWAS to test associations between measured TG levels and validated variants previously associated with TG levels in Mendelian randomization studies. We replicated results in All of Us (AoU). In the BioVU EA cohort (n=63,094), 9 of 10 validated SNPs had significant or suggestive associations with lipid and cardiovascular phenotypes, largely consistent with AoU participants of EA (n=97,545). In the BioVU AA cohort (n=12,515) and AoU AA cohort (n=31,710), results were more limited; only 1 of 6 validated SNPs was significantly associated with a lipid or cardiovascular phenotype in either BioVU or AoU, and no significant or suggestive associations were consistent across both cohorts. We detected few secondary effects in either EA or AA BioVU patients, and none were replicated. These results suggest that there may be limited additional benefits, but few deleterious effects, associated with targeting known TG-associated genes. However, these targets may not be as effective for mitigating cardiovascular risk among individuals of AA.

Keywords: Triglycerides; drug side effects; mendelian randomization; phewas.

Figure 1.. Study design.

TG = triglyceride. MR = Mendelian randomization. GWAS = genome-wide association study. MAF = minor allele frequency. EA = European ancestry. AA = African ancestry. PheWAS = phenome-wide association study. BioVU = Vanderbilt University Medical Center biobank. AoU = All of Us Research Program.

Figure 2.. Significant associations between prespecified phenotypes and genetic variants in the BioVU European ancestry cohort, with All of Us replication indicated.

○ = replicated in All of Us. Dark red = p<3.15×10⁻⁵ (0.05/1587) [logistic regression]; highly significantly increased risk associated with TG-increasing allele (full PheWAS Bonferroni-corrected). Medium red = p<0.0024 (0.05/21) [logistic regression]; significantly increased risk associated with TG-increasing allele (Bonferroni-corrected). Light red = p<0.05 [logistic regression]; suggestive significance of increased risk associated with TG-increasing allele. Light blue = p<0.05 [logistic regression]; suggestive significance of decreased risk associated with TG-increasing allele.

Figure 3.. Significant associations between prespecified phenotypes and genetic variants in the BioVU African ancestry cohort, with All of Us replication indicated.

○ = replicated in All of Us. Dark red = p<4.69×10⁻⁵ (0.05/1065) [logistic regression]; highly significantly increased risk associated with TG-increasing allele (full PheWAS Bonferroni-corrected). Medium red = p<0.0024 (0.05/21) [logistic regression]; significantly increased risk associated with TG-increasing allele (Bonferroni-corrected). Light red = p<0.05 [logistic regression]; suggestive significance of increased risk associated with TG-increasing allele. Light blue = p<0.05 [logistic regression; suggestive significance of decreased risk associated with TG-increasing allele.