A null adult postmortem-brain test of a locked 468-gene panel in autism spectrum disorder, with published maternal-axis architecture as context

Abstract Autism spectrum disorder (ASD) was examined in four primary adult postmortem-brain GEO cohorts (GSE102741, GSE28521, GSE269105, GSE64018) using a frozen 468-gene redox, metal, xenobiotic, and barrier panel. Inverse-variance random-effects meta-analysis (DerSimonian–Laird) produced 392 analyzable genes, one at Benjamini–Hochberg FDR < 0.05 (MAFK, lower in ASD), and no high-confidence genes (FDR < 0.05, measured in ≥ 3 cohorts, direction-concordant in ≥ 70% of cohorts, and nominally significant in ≥ 2 cohorts). Iron/redox–NRF2 genes that are higher in the neurodegenerative companions of this series were near zero and heterogeneous here (HMOX1 + 0.10, I² 94%). Maternal whole blood (GSE148450) and umbilical cord blood (GSE123302) were also null at FDR < 0.05. Across 14 supporting developmental-model and postmortem datasets, per-dataset panel dysregulation tracked the breadth of the engineered lesion (broad regulatory or metabolic lesions 16–47% of panel genes; synaptic-scaffold models ≤ 7%) rather than the ASD label. Because adult brain samples a late, compensated tissue, I interpret the adult-brain null as constraining adult-brain transcriptional scarring, not the maternal–fetal interface. In first- versus third-trimester placenta (GSE247382; no ASD label), 459 of the 468 panel genes were measured and 320 expressed; placental polarity is descriptive context, not an ASD contrast. Conjugation and metal-handling genes were higher in the first trimester, barrier and steroid genes at term, and core NRF2-axis genes changed little. Published placental ASD evidence (a CYP2E1 methylation region in MARBLES; a 111-gene expression association in ELGAN) is cited, not re-analyzed. A compiled inventory of published detoxification-gene associations (MTHFR, CBS, PON1, GSTM1, GSTT1, GSTP1, ALDH2) was checked against the primary records for whose genotype was measured and which outcome was reported; only two entries are maternal-genotype ASD findings, both from one CHARGE analysis and both conditional on absent periconceptional prenatal vitamin use (maternal MTHFR 677 TT, OR 4.5, 95% CI 1.4–14.6; maternal CBS rs234715 GT + TT, OR 2.6, 95% CI 1.2–5.4), and the research formula mGVI is defined on those two. The remaining entries are child genotypes, non-ASD outcomes, interaction terms, or unsupported, and are retained as context only. mGVI, ASD-IDRS, and MFS were not computed on a maternal cohort, and none is a clinical tool. No proband detoxification-related variant in the assembled inventory reaches OR ≥ 1.5 in the GWAS summaries used. This paper is a disease-restricted companion to the methods-and-panel manuscript of this series (11).