Cross-cohort transcriptomic integration links neoadjuvant chemotherapy response to an inflammatory macrophage-epithelial program in muscle-invasive bladder cancer

Background Only a subset of patients with muscle-invasive bladder cancer (MIBC) achieves a major pathological response to cisplatin-based neoadjuvant chemotherapy (NAC). The cellular sources of pretreatment transcriptional programs associated with response remain incompletely resolved. Methods Two pretreatment clinical response cohorts, GSE212810 (21 responders and 18 non-responders) and GSE247185 (8 pathological complete responders and 5 non-pCR patients), were analyzed for cross-cohort response association, while the treatment-naive single-nucleus cohort GSE169379 (25 patients) was used for cellular attribution. Gene-level response statistics were calculated separately in each clinical cohort and combined using a directional sample-size-weighted Stouffer procedure. Hallmark programs were evaluated by preranked gene-set enrichment analysis. Concordant pathway leading edges were compared, and genes present in at least three leading edges were retained as a 20-gene core before cellular localization. Patient-within-sample comparisons avoided cellular pseudoreplication. Regimen restriction, leave-one-sample-out analyses, and adjustment for a basal-minus-luminal expression proxy were used as sensitivity analyses. Results TNF-alpha signaling via NF-kB was the strongest response-associated program in the cross-cohort meta-ranking (normalized enrichment score [NES] 2.574; permutation false discovery rate [FDR] < 1/5,000). Enrichment was concordant in GSE212810 (NES 1.713; FDR 0.00836) and GSE247185 (NES 2.539; FDR < 1/3,000). Interferon-gamma response, inflammatory response, epithelial-mesenchymal transition, and IL6/JAK/STAT3 signaling were also positively enriched, whereas MYC targets V1 and E2F targets were negatively enriched in the meta-ranking. The five positively enriched leading edges yielded a 20-gene core represented in at least three programs. All 20 genes were detected in GSE169379. Patient-paired analysis localized this core to inflammatory macrophages (21 patients; mean paired difference 0.319; FDR 1.3x10^-5) and CDH12-positive epithelial cells (25 patients; mean paired difference 0.199; FDR 2.67x10^-4). PDPN-positive fibroblasts showed a positive but multiplicity-adjusted nonsignificant difference (15 patients; FDR 0.0603), while FAP- and PDGFRB-positive fibroblasts were unsupported. TNF/NF-kB enrichment remained positive in every leave-one-sample-out iteration in both response cohorts. After cohort standardization and adjustment for a basal-minus-luminal proxy, response remained associated with the inflammatory core (beta 0.534; HC3 P = 0.0353). Conclusions Conclusions: Pretreatment MIBC response to NAC was associated with a coordinated inflammatory program rather than a single-gene marker. Independent single-nucleus analysis localized the shared core predominantly to inflammatory macrophages and CDH12-positive epithelial cells. These observational findings require validation in larger prospective cohorts and functional studies.