Bidirectional Mendelian randomization analyses reveal APOE status mediates the effects of statin medications on Alzheimer’s disease

Abstract Background There is currently no unanimous conclusion on the statin medications and Alzheimer’s disease (AD) risk, which has long been one of the most concerns in lipid‐lowering and anti‐atherosclerosis treatment. We performed a bidirectional Mendelian Randomization (MR) to evaluate whether there were causal links between statin medications and AD. Method Single nucleotide polymorphisms (SNPs) identified from genome‐wide association studies (GWAS) were used as instrumental variables. One GWAS of statin medications (N = 462,933) and AD (N = 63,926) with summary statistics available was included in the MR analysis. All statistical analyses were conducted using the TwoSampleMR package with default parameters. Inverse variance weighted (IVW) was used as the primary method to calculate the causal estimates (β), accompanied by two complement methods, MR‐Egger and weighted median. Pleiotropic effects and heterogeneity driving causal association were also assessed. Result IVW results showed that genetically predicted use of statins was causally associated with increased risk of AD (atorvastatin: β = 33.67, P = 0.048; simvastatin: β = 4.88, P = 0.044). However, after removing SNPs inside the apolipoprotein E (APOE) gene region (CHR19:45,111,942‐45,711,941 based on GRCh37/hg19), the causal links of statin medications on AD disappeared ( Figure 1A ). Two SNPs, rs429358 (C;C) and rs7412 (C;C) known as APOE4 mainly contributed to the controversial data involved in use of statins and AD risk. The results of reverse causal analysis showed that patients with AD tended to have a higher exposure to statins (atorvastatin: β = 0.005, P = 0.001; simvastatin: β = 0.012, P = 0.004). Consistently, when discarding the SNPs within the APOE gene region, no relationship was found between diagnosis of AD and statins exposure ( Figure 1B ). The directional horizontal pleiotropy and heterogeneities found in MR‐PRESSO global and MR‐Egger regression test were largely disappeared after excluding SNPs inside APOE cluster region. Conclusion Our results suggested that the inconsistent association between statins use and AD risk observed in previous observational investigations might be mainly attributed to various APOE status (APOE2, APOE3, and APOE4). Thus, it reminds clinicians that statins can be used with confidence in lipid‐lowering and anti‐atherosclerosis and are unlikely to affect AD risk.