A mendelian randomization study of pro-protein convertase subtilisin kexin 9 inhibition and stroke: evidence for benefits beyond low-density lipoprotein cholesterol lowering

Stroke is the second leading cause of death globally, and recent clinical trials have found that pharmacological proprotein convertase subtilisin kexin 9 (PCSK9) inhibition (PCSK9i) reduces stroke risk. However, the impact of long-term PCSK9 protein inhibition on stroke risk and the complex interplay of PCSK9 with various lipids remains unknown. Our goal was to leverage data on PCSK9 protein levels using drug-target Mendelian randomization (MR) and multivariable MR (MVMR) to investigate the impact of genetically-lowered PCSK9 protein levels on stroke risk. We created genetic instruments using PCSK9 variants associated with plasma PCSK9 concentration (Nprimary=35,559; Nreplication=10,186); low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), and triglycerides (TG; N=1,320,016); apolipoprotein B (ApoB; N=439,125); lipoprotein a (Lp[a]; N=361,194); stroke (Ncases=40,585); ischemic stroke (IS; Ncases=34,217); and stroke subtypes (Ncases≤7,193). We performed drug-target MR investigating relationships between plasma PCSK9 and stroke phenotypes as well as MVMR assessing PCSK9 inhibition on stroke risk while controlling for five major lipid subfractions (LDL-C, HDL-C, TG, ApoB, and Lp[a]). We identified a strong protective association between genetically-lowered PCSK9 protein levels and stroke (odds ratio (OR)=0·86, 95% CI [0·83-0·88], P=1.74 × 10^-28), ischemic stroke (OR=0·88, 95% CI [0·85-0·90], P=1·70 × 10^-18), cardioembolic stroke (OR=0·84, 95% CI [0·76-0·93], P=0·001), and large artery ischemic stroke (OR=0·74, 95% CI [0·68-0·80], P=1·28 × 10^-14). We replicated these findings with complementary MR methods and two genetic instruments from independent data. MVMR analysis accounting for five lipid subfractions suggested that genetically proxied circulating PCSK9 inhibition was protective both for ischemic stroke (OR=0·89, 95% CI [0·82-0·97], P=0·008) and cardioembolic stroke (OR=0·83, 95% CI [0·69-1·00], P=0·04) supplementary to its established effect on lowering lipid levels, while PCSK9 inhibition was protective against large artery atherosclerosis entirely through its lipid pathways. MR estimates were generally consistent across complementary MR methods and tests for heterogeneity and pleiotropy were null, strengthening causal inference. Our data provide new evidence that plasma PCSK9 protein levels inversely associate with stroke risk, resolving discrepancies between randomized clinical trials and previous genomic studies. MVMR suggested pleiotropic relationships of PCSK9 on stroke risk beyond its effects on lipid fractions, underlining that future studies are needed to investigate the protective effect of PCSK9i beyond LDL-C-related pathways. Broadly, our study should reassure clinicians that anti-PCSK9 therapies reduce stroke.