<![CDATA[Background. Major depressive disorder (MDD) and death by suicide may involve transcriptional and cellular shifts in the dorsolateral prefrontal cortex (DLPFC). Bulk transcriptomes mix within-cell regulation with tissue composition, so signals may not reproduce. Objective. To identify covariate-adjusted DLPFC co-expression modules associated with MDD and suicide, relate them to cell-type signatures, and assess external reproducibility. Methods. This secondary analysis of public datasets used GSE101521 (29 controls, 21 MDD donors who died by suicide [MDD-S], and 9 who died otherwise [MDD-NS]). After covariate residualization, the 5,000 most variable genes entered signed weighted gene co-expression network analysis. Eigengenes were tested with empirical-Bayes models and false-discovery rate (FDR) correction. GSE144136 supplied cell signatures and donor-level composition; scores were tested in 15 matched GSE54568 pairs. Results. The brown module (509 genes; inflammatory enrichment) was nominally lower in MDD (β = −0.499; p = 0.045; FDR = 0.316) and MDD-S (β = −0.570; p = 0.039; FDR = 0.274) than controls; blue showed the largest MDD-S versus MDD-NS estimate (β = −0.807; p = 0.015; FDR = 0.105). No module or bulk-deconvolved cell fraction reached FDR < 0.05. In GSE144136 (male donors), MDD-S showed lower relative astrocyte (FDR = 0.002) and higher excitatory and inhibitory neuronal (both FDR = 0.024) composition. No module replicated in GSE54568; two of five were directionally concordant. Conclusion. Inflammatory and oligodendrocyte-related modules lacked multiplicity-adjusted or external support for phenotype associations. Donor-level composition gave stronger yet constrained evidence. The findings prioritize immune-glial and neuronal hypotheses for replication, not biomarker use.]]>
Copyrights © 2026
