Transcriptomic associations of a pre-specified 468-gene redox and xenobiotic-metabolism panel across five neurological disorders, with a published-variant inventory
Abstract
Five neurological disorders—Alzheimer’s disease (AD), autism spectrum disorder (ASD), Parkinson’s disease (PD), multiple sclerosis (MS), and amyotrophic lateral sclerosis (ALS)—have distinct clinical and pathological features. I performed independent within-disease random-effects meta-analyses of public transcriptomic cohorts using an immutable 468-gene redox, metal-homeostasis, xenobiotic-metabolism, and barrier panel, and separately inventoried published variants in panel genes for AD, PD and ASD. AD brain, ALS spinal cord/motor neuron, and MS white matter showed disease-specific patterns; in bulk PD substantia nigra two external cohorts supported the high-confidence set in direction and rank but not in per-gene significance. Across the four adult neurodegenerative analyses, 53 genes met the locked high-confidence rule in at least two analyses (Online Resource 3). Seven genes—AKR1C3, CBS, CP, FTL, HMOX1, PINK1, and SLC40A1—met that rule in at least three; CBS was up in AD and PD and down in MS, so the rule does not imply uniform direction or a shared mechanism. HMOX1 was positive in several adult analyses; GSS and GSR transcripts were lower in AD and PD. Adult ASD postmortem brain was a robust FDR null (1 of 392 genes analyzable in at least two cohorts; that gene did not meet the high-confidence rule). Cell-type datasets in MS and PD supply compartment context only. A competitive gene-label permutation test did not show that the 468-gene panel out-enriched power-matched random sets. Inference is gene-level within the locked list, not list-level enrichment. Published variants in panel genes are inventoried for AD, PD and ASD (35 cited entries in the companion manuscripts); they are context for proposed research constructs, not clinical risk scores.
Related articles
Related articles are currently not available for this article.