Shared Genetic Architecture Between Kidney Function and Alzheimer Disease Across Ancestries

Abstract

Epidemiological studies have consistently shown that chronic kidney disease is associated with increased Alzheimer disease risk. However, the underlying genetic architecture connecting these two conditions remains largely unexplored beyond genome-wide correlation analyses. Here, we conducted the first comprehensive, multi-ancestry, large-scale genetic investigation to identify shared genetic components between kidney function and Alzheimer disease. We leveraged large-scale genome-wide association study summary statistics for estimated glomerular filtration rate (N{approx}1.5 million European, N{approx}145,000 African ancestry) and late-onset Alzheimer disease (N=63,926 and N=398,058 in two European cohorts; N=9,168 in African ancestry) corrected for competing risk bias. We deployed a novel analytical framework integrating linkage disequilibrium score regression and polygenic risk score analysis, local analysis of [co]variant association, conjunctional false discovery rate analysis with Bayesian colocalization and fine-mapping, and bidirectional cis-Mendelian randomization to identify vertical pleiotropy. Despite the absence of genome-wide genetic correlation (rg {approx} 0, p > 0.1), local genetic analysis uncovered striking regional heterogeneity. Sixteen pleiotropic loci were identified in individuals of European ancestry (conjunctional false discovery rate < 0.05), including APOE, PICALM, SPI1, and EFTUD1, alongside 15 loci with significant local genetic correlations. Fine-mapping revealed that most pleiotropic loci harbored distinct causal variants for kidney function and Alzheimer disease, indicating horizontal pleiotropy. An APOE {varepsilon}4-defining allele (rs429358) was the sole variant with shared causality across both traits. We identified vertical pleiotropy using cis-Mendelian randomization at the PICALM and EFTUD1 loci, providing evidence that kidney function-related genetic variants can causally affect Alzheimer disease risk at specific genomic loci. In contrast, loci such as CD2AP, MAT1A, and SYMPK demonstrated horizontal pleiotropy, reflecting shared upstream biological pathways rather than direct causal mediation. Notably, APOE was the only pleiotropic locus shared between European and African ancestry groups, underscoring marked ancestry-specific genetic architectures with critical implications for risk prediction and therapeutic translation. Alzheimer disease and kidney function share genetic components at specific loci rather than genome-wide, with mixed directional effects and horizontal pleiotropy explaining the absent global correlation despite strong local signals. At a subset of loci, we identified directional effects linking kidney genetic determinants to Alzheimer disease risk using cis-Mendelian randomization, supporting a complex kidney-brain genetic axis. Most overlap reflects horizontal pleiotropy, with limited loci showing vertical pleiotropy. APOE was the only shared locus across ancestries, underscoring ancestry-specific architectures with implications for risk prediction. The multi-scale approach used here also provides a methodological framework for dissecting complex disease relationships missed by traditional genome-wide analyses.