Non-APOE variants predominately expressed in smooth muscle cells contribute to the influence of Alzheimer's disease genetic risk on white matter hyperintensities

Abstract

Introduction: White matter hyperintensity volumes (WMHVs) are disproportionally prevalent in individuals with Alzheimer's disease (AD), potentially reflecting neurovascular injury. We quantify the association between AD polygenic risk score (AD-PRS) and WMHV, exploring single-nucleotide polymorphisms (SNPs) that are proximal to genes overexpressed in cerebrovascular cell species.

Methods: In a UK-Biobank sub-sample (mean age = 64, range = 45-81 years), we associate WMHV with (1) AD-PRS estimated via SNPs across the genome (minus apolipoprotein E [APOE] locus) and (2) AD-PRS estimated with SNPs proximal to specific genes that are overexpressed in cerebrovascular cell species.

Results: We observed a positive association between non-APOE-AD-PRS and WMHVs. We further demonstrate an association between WMHVs and AD-PRS constructed with SNPs that are proximal to genes over-represented in smooth muscles cells (SMCs; β = 0.135, P_FWE < 0.01) and internally replicated (P_{DISCOVERY+REPLICATION} < 0.01).

Discussion: Common AD genetic risk could explain physiological processes underlying vascular pathology in AD. SMC function may offer a treatment target to prevent WMHV-related AD pathophysiology prior to the onset of symptoms.

Highlights: Alzheimer's disease (AD) risk factors such as apolipoprotein E (APOE) ε4, link to increased white matter hyperintensity volume (WMHV). WMHVs indicate vascular risk and neurovascular injury in AD. The broader genetic link between AD risk and WMHV is not fully understood. We quantify AD polygenic risk score (PRS) associations with WMHV, excluding APOE. AD-PRS in smooth muscle cells (SMCs) shows a significant association with increased WMHV.

Keywords: cerebrovascular; polygenic risk score; preclinical; smooth muscles cells; white matter hyperintensities.

FIGURE 1

(A) Whole‐genome AD‐PRS; minus the SMC SNPs and including the APOE region. (B) Cell type–specific AD‐PRS. Beta coefficients in red reflect p _FWE‐CORRECTED < 0.05). Error bars reflect 95% confidence intervals of the beta coefficient for the combined sample (N = 32,087). APOE, apolipoprotein E; BEC, brain endothelial cell; FWE, family‐wise error; SMC, smooth muscle cell.

FIGURE 2

(A) Association between AD‐PRS and WMHV. Red beta coefficients reflect p _FWE‐CORRECTED < 0.05. Error bars reflect 95% confidence intervals of the beta coefficient for discovery (N = 21,360) and replication samples (N = 10,727). (B) The association between SNP effect sizes for AD and WMHV for the whole‐genome AD‐PRS (gray line‐of‐best‐fit), a whole‐genome AD‐PRS (including APOE—blue line‐of‐best‐fit), and SMC AD‐PRS (red line‐of‐best‐fit). Gene IDs reflect gene proximal to SNPs in the SMC AD‐PRS. Each SNP is plotted by coefficient in the AD risk (x axis) versus estimated effect size for WMHV in the independent data set (y axis). (C) The SMC AD‐PRS was associated with WMHV to a higher degree compared to the majority of a 1000 SNP‐set size‐matched, randomly sampled AD‐PRS (two tailed). Vertical red line represents degree of association (absolute standard beta coefficient) between SMC AD‐PRS and WMHV. FWE, family‐wise error; SMC, smooth muscle cell; WMHV, white matter hyperintensity volume.