Extended Data Fig. 1: Additional Plots for Lipid Lowering Drug Target MR using LDL Downstream Biomarker Data, for which there was no reductive effect on the positive control.

Additional plots for inverse variance-weighted MR (Mendelian randomization) for lipid-lowering drug targets using large-scale European genome-wide association study (GWAS) summary statistics. The cis-acting gene region was selected for each target from a LDL-c (a downstream biomarker) GWAS (n = 188578) as the exposure with five vascular dementia outcomes; WMH (white matter hyperintensities, n = 50559), iFA (inversed fractional anisotropy, n = 31125), MD (mean diffusivity, n = 31147), LS (lacunar stroke, n = 6030/248929 cases/controls), VaD (vascular dementia diagnosis, n = 7008/899672 cases/controls). CAD (coronary artery disease) was used as a positive control (n = 122733/424528 cases/controls) and is indicated in red. The additional target presented here did not show the expected lowering effect on the positive control (Pval>0.1) and as only a single IV was used for analysis a Wald ratio was calculated. Data are presented as the causal MR estimate as the centre and error bars indicating a +/- a 95% confidence interval (CI), with the minima corresponding to the lower 95% CI bound and the maxima representing the upper 95% CI bound. The number of instrumental variables (IVs) used in each analysis is given. The F-statistic (Fstat) is also provided as a measure of instrument strength. Pval denotes uncorrected p-values; multiple-comparison corrected p-values (applied at differing levels) are available in the supplement. LDL-c (low-density lipoprotein cholesterol), SD (standard deviation).