Answers are grounded in this paper and cite their sources.
Background Blood transcriptomic studies in Alzheimer's disease (AD) often train signatures directly in blood, making it unclear whether signals reflect brain disease programs, blood-cell composition, or cohort structure. Objective To test how neuropathology-defined AD transcriptional programs appear in blood, whether distinct brain programs show a shared peripheral pattern, and whether that pattern is most evident during mild cognitive impairment (MCI). Methods Three directional gene sets were derived before blood analysis from postmortem brain RNA-seq (GSE278723) to represent amyloid-associated, neurodegeneration-related, and core AD changes. These brain-defined programs were projected into five whole-blood cohorts (1975 participants) using directional single-sample Gene Set Enrichment Analysis and assessed for stage-specific patterns, comparison with size-matched random sets, sensitivity to estimated cell composition, dimensionality, held-out classification, and association with Alzheimer's Disease Neuroimaging Initiative (ADNI) genetic and fluid biomarkers. Results Brain-derived AD programs were detected across blood cohorts and showed a shared rather than separable pattern. Scores were highest at MCI, exceeded size-matched random baselines, and were not primarily explained by estimated blood-cell proportions. In GSE63060, P1/P2/P5 scores formed a dominant axis, with PC1 explaining 92.2% of directional half-score variance. A held-out classifier achieved AUC = 0.822, and ADNI analyses showed information beyond APOE ε4 and improved discrimination when blood scores were combined with cerebrospinal fluid pTau181/Aβ 42 . Conclusions Neuropathology-defined AD transcriptomic programs form a shared peripheral axis that is most evident at MCI. This brain-anchored signal adds a systemic transcriptional dimension to genetic and fluid biomarkers and provides a biological framework for studying early AD-related disease in blood.
No takes yet. Share an insight, caveat, or question.
Lu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: