A Lipoylation–PDH–TCA Transcriptional Deficit in the Alzheimer’s Disease Cortex — Donor-level Multi-cohort Evidence with Neuronal-composition, Disease-specificity and Matched-null Controls

Cuproptosis, a copper-dependent form of regulated cell death acting through lipoylated tricarboxylic acid (TCA) cycle proteins, has been proposed as a molecular link between type 2 diabetes and Alzheimer’s disease. We asked whether cuproptosis effector genes are dysregulated in a cell-type-specific manner in the human Alzheimer’s disease cortex, and whether any such signal survives controls for neuronal composition and disease specificity. Single-nucleus data from 21 prefrontal cortex donors were aggregated to donor-level pseudobulk within each cell type. Four independent bulk cohorts, each contributing one observation per donor, entered a random-effects meta-analysis. Adjustment for estimated neuronal content, competitive and expression-matched null gene-set testing, and comparison against Huntington’s disease and vascular dementia were applied as controls. The donor was the unit of replication throughout. No gene–cell-type pair survived correction across the 75 prespecified tests; PDHA1 in inhibitory neurons showed the largest nominal reduction (− 0.418, q = 0.829). In donor-level meta-analysis PDHA1 was lower in Alzheimer’s disease in all four cohorts (− 0.457, 95% CI − 0.712 to − 0.201, I 2 = 0%) but did not survive correction (q = 0.086). A post hoc lipoylation/pyruvate dehydrogenase/TCA module did reach the adjusted threshold (− 0.245, 95% CI − 0.363 to − 0.127, q = 0.035), whereas the copper modules pooled to zero and moved in opposite directions across cohorts. Comparable reductions occurred in Huntington’s disease and vascular dementia, and the disease-control contrasts were not significant. These transcriptional associations do not demonstrate cuproptosis, are not specific to Alzheimer’s disease, and are hypothesis-generating only.