Causal relationships between alzheimer’s disease genetics and brain connectivity alterations: a multi-modal mendelian randomization study with transcriptomic validation of 191 rs-fMRI and 635 DTI neuroimaging traits

Traditional observational magnetic resonance imaging (MRI) studies have revealed changes in brain connectivity in Alzheimer’s disease (AD). However, the findings have been inconsistent due to small sample sizes and potential confounding factors. The genetic effects of AD on the inherent brain activity and connectivity of patients are still not well understood. We utilized summary-level GWAS data for 223,906 Europeans from three large AD cohorts and comprehensive GWAS data for 191 rs-fMRI functional connectivity (FC) traits (n = 34,691) and 635 diffusion tensor imaging (DTI) metrics (n = 33,292) from the BIG Knowledge Portal. A bidirectional two-sample Mendelian randomization (MR) analysis with multiple MR methods was performed to evaluate the causality between AD genetics and genetically predicted whole-brain functional and structural connectivity changes. A series of sensitivity analyses were systematically conducted to assess the pleiotropy, heterogeneity, and outliers. Additionally, SNP-to-gene mapping, enrichment analysis, protein-protein interaction (PPI), single-SNP, and SNP location-based MR were performed to elucidate the molecular mechanisms. To validate our findings, we analyzed an independent cohort from ADNI (n = 30/group) and performed transcriptomic validation using RNA-seq data from 63 samples (32 AD, 31 control). Our MR analysis revealed significant causal associations between AD and specific alterations in fMRI FC, particularly involving the precuneus, occipital lobe, and default mode network. Similarly, AD was causally linked to changes in fractional anisotropy (FA) and mean diffusivity (MD) across distinct white matter fiber tracts. The molecular mechanisms underlying these MRI changes involved polygenic contributions from multiple AD-associated SNPs, primarily those mapped to non-coding regions, in addition to genic SNPs enriched in pathways regulating amyloid-beta clearance and neuroinflammation. External validation using the ADNI cohort confirmed the FC alterations identified through MR. Transcriptomic validation confirmed the significant upregulation of four genes (CTSB, SDC4, CTNND2, and FERMT2) in AD and uncovered three potential AD-associated genes (ITGB1BP1, FBXO33, and RASGEF1C). Our multi-modal MR study elucidated causal links between AD genetics and brain imaging-derived phenotypes (IDPs), with independent validation from both neuroimaging and transcriptomic analyses. These findings enhance understanding of AD etiology and identify potential MRI markers for diagnosis and treatment monitoring.