A locked 468-gene detoxification panel in Alzheimer's disease brain, liver, and blood: 141 high-confidence cerebral associations including lower CYP46A1 transcript abundance, and a 12-sample hepatic CYP7A1 signal

Detoxification gene expression in Alzheimer's disease (AD) has been examined almost exclusively in brain. Whether the same genes move in the same direction in peripheral organs is unknown. Here a single locked, pre-specified 468-gene redox, metal-homeostasis, xenobiotic-metabolism, and barrier panel is applied to three compartments under one analytic rule. In brain, inverse-variance random-effects meta-analysis (DerSimonian–Laird) across six independent postmortem adult-brain GEO cohorts (GSE33000, GSE132903, GSE173955, GSE159699, GSE5281, GSE48350) yielded 458 analyzable panel genes, 167 at Benjamini–Hochberg FDR < 0.05, and 141 high-confidence genes (FDR < 0.05, measured in at least three cohorts, direction-concordant in at least 70% of cohorts, and nominally significant in at least two cohorts). High-confidence down-regulation included CYP26A1, TXN, GSS, ATP1A3, and SLC30A3; high-confidence up-regulation included metallothionein family members (MT1F, MT1E, MT2A), PPARA, and FOXO1. Heterogeneity was often high (median I² among high-confidence genes 79.3%). Within that high-confidence set, three genes of the cholesterol-to-oxysterol-to-bile-acid pathway were informative in all six cohorts: cholesterol 24-hydroxylase CYP46A1 was down-regulated (−0.10; FDR = 0.035), while sterol 27-hydroxylase CYP27A1 (+0.17; FDR = 0.048) and CYP39A1 (+0.14; FDR = 0.018) were up-regulated. Because brain cholesterol is separated from the circulating pool by the blood-brain barrier and must be converted to an oxysterol to be exported at all, this is a direction-concordant transcript shift in the principal cerebral cholesterol-disposal route, not a measurement of enzyme activity or sterol flux. In liver, primary RNA sequencing of commercially bio-banked human hepatic tissue (7 AD, 5 control; 12 samples) is hypothesis-generating and showed a signature unrelated to the central one. In the primary all-case against all-control contrast, five panel genes reached significance: CYP7A1 up-regulated (log₂ fold-change +5.65; Q = 1.5 × 10⁻⁶), NOTUM (+4.30) and CYP4F22 (+3.00) up, and CYP17A1 (−4.47) and CYP2S1 (−1.77) down. Hepatic change is therefore concentrated on the bile-acid and steroidogenic axes rather than being a uniform induction or loss of xenobiotic-metabolizing capacity. CYP26B1 was down-regulated in every subgroup contrast involving male Alzheimer's donors and none involving female donors (−2.54; Q = 1.5 × 10⁻⁶); other cytochromes reaching significance did so only against the two male controls and are reported as exploratory. In a supporting peripheral-blood contrast (GSE63060, AddNeuroMed, n = 329), the panel signal was dominated instead by mitochondrial transcripts (NDUFA1 −1.10, NDUFS5 −0.92, COX7C −0.79, MRPL51 −0.57), with a lower-magnitude detoxification component. Brain, liver, and blood were independent donor collections, not paired samples from the same individuals. The three compartments therefore do not move together. Central detoxification transcripts show metallothionein induction with reduced glutathione-synthesis and thioredoxin abundance, while the small hepatic cohort is dominated by CYP7A1 rather than a uniform xenobiotic-metabolizing shift. These are transcript-abundance observations and do not establish enzyme activity, flux, or causal direction. The hepatic cohort is small (n = 12) and all hepatic and sex-stratified findings are hypothesis-generating. Published AD-associated variants are tabulated alongside expression rather than merged with it; the Genetic Vulnerability Index, Detoxification Dysregulation Index, and Integrated Detoxification Risk Score are uncalibrated research constructs and are not diagnostic, prognostic, or treatment-selection tools.