Integrative identification of macrophage-associated candidate biomarkers of chemoradiotherapy response in esophageal squamous cell carcinoma

Esophageal squamous cell carcinoma (ESCC) is a highly aggressive malignancy characterized by substantial variability in response to chemoradiotherapy. Increasing evidence suggests that tumor-associated macrophages (TAMs) influence treatment responses through interactions with the tumor microenvironment (TME). However, the macrophage-associated molecular features related to chemoradiotherapy response in ESCC remain incompletely characterized. This study aimed to identify macrophage-associated biomarkers linked to chemoradiotherapy response in ESCC and explore their potential biological relevance. Expression datasets from the Gene Expression Omnibus (GEO) were utilized to identify differentially expressed genes (DEGs) between ESCC and normal esophageal tissues, as well as between chemoradiotherapy responders and non-responders. Immune infiltration analysis and weighted gene co-expression network analysis were combined to identify gene modules associated with macrophage infiltration. Machine learning methods, including LASSO regression, SVM-RFE, and Boruta, were employed to prioritize candidate biomarkers. Survival analysis and a prediction model based on prioritized genes assessed their prognostic and chemoradiotherapy-response associations. Multiple functional analyses, including GSEA, GSVA, and ceRNA network construction, were conducted to explore biological relevance. Reverse network pharmacology, molecular docking, and molecular dynamics (MD) simulations were performed to hypothesize potential protein–compound interactions. Single-cell RNA-seq datasets were analyzed to characterize cell-type-specific expression patterns, TME context, and cell-cell communication patterns. A total of 53 macrophage-associated DEGs related to chemoradiotherapy response were identified, with KRT14 and MMP10 designated as candidate biomarkers. The prediction model based on KRT14 and MMP10 achieved an AUC of 0.834. Higher KRT14 expression was associated with favorable survival, while both KRT14 and MMP10 correlated with transcriptional programs related to epithelial plasticity, immune regulation, and metabolism. Reverse network pharmacology and molecular docking revealed β-carotene and deoxycholic acid as candidate compounds predicted to interact with MMP10. MD analyses suggested persistent protein–ligand interactions and conformational stability of the corresponding complexes. Single-cell RNA-seq analysis supported the expression of KRT14 and MMP10 in epithelial cells and revealed enhanced fibroblast-epithelial communication in ESCC tissues. KRT14 and MMP10 were identified as macrophage-associated candidate biomarkers related to chemoradiotherapy response in ESCC. Their associations with epithelial cellular states and TME-related processes offer potential biological insights into treatment-response heterogeneity. The computational findings generate testable hypotheses regarding the biological relevance of these biomarkers and predicted compound-target interactions, which require further experimental validation.