Construction of an Esophageal Cancer Prognostic Model Based on Mitochondrial Energy Metabolism-Related Genes and Mechanism of INHBA in Promoting Malignant Progression via the Smad2/3 Pathway

Objective: Esophageal cancer (ESCA) has poor prognosis and lacks reliable biomarkers. This study aimed to construct a mitochondrial energy metabolism-related prognostic model and identify key regulatory genes. Methods: TCGA-ESCA and GSE53625 data were used as training and validation cohorts, respectively. Mitochondrial energy metabolism scores were calculated by ssGSEA. Differential expression analysis and WGCNA identified key genes, followed by consensus clustering, Cox/LASSO regression, enrichment analysis, immune infiltration, TIDE evaluation, and nomogram development. INHBA knockdown or overexpression was performed in KYSE-150 and TE-1 cells to assess malignant phenotypes, mitochondrial function, EMT markers, and Smad2/3 signaling, with SB431542 used for pathway blockade. Results: High mitochondrial energy metabolism scores predicted poorer overall survival. A total of 124 key genes were identified, and three molecular subtypes were established, with Cluster 2 showing the worst prognosis. A four-gene signature comprising COL11A1, INHBA, TNFAIP6, and POSTN stratified patients into high- and low-risk groups in both cohorts. High-risk tumors were enriched in extracellular matrix remodeling, cell adhesion, inflammatory response, hypoxia, angiogenesis, and PI3K-Akt signaling, whereas oxidative phosphorylation was more active in low-risk tumors. The high-risk group had higher TIDE scores, suggesting poorer immunotherapy response. The nomogram incorporating risk group and M stage showed limited discrimination after bootstrap correction. In vitro, INHBA promoted proliferation, migration, invasion, EMT, mitochondrial membrane potential, ATP production, and oxidative phosphorylation. SB431542 partially reversed these effects. Conclusions: This MEMRG-based model predicts ESCA prognosis and immune features. INHBA promotes ESCA progression by activating Smad2/3-mediated EMT and mitochondrial metabolic reprogramming, suggesting its potential as a prognostic biomarker and therapeutic target.