Macrophage-linked biomarkers flagged for esophageal cancer chemoradiotherapy response in computational study

KRT14 and MMP10 have been identified as macrophage-associated candidate biomarkers linked to chemoradiotherapy response in esophageal squamous cell carcinoma (ESCC), according to a computational study published in Discover Oncology. The research, led by Shanglin Dong and colleagues, combined public gene expression datasets with machine learning to prioritize the two genes, which the authors propose as hypotheses-generating markers rather than validated clinical predictors.

ESCC is a highly aggressive malignancy with substantial variability in how patients respond to chemoradiotherapy. Tumor-associated macrophages, immune cells that populate the tumor microenvironment, are thought to influence treatment responses, but the macrophage-related molecular features tied to chemoradiotherapy response in ESCC have remained incompletely characterized. To address this, the team mined Gene Expression Omnibus datasets to find differentially expressed genes between ESCC and normal esophageal tissue, and between patients who responded to chemoradiotherapy and those who did not.

The analysis pipeline layered several approaches. Immune infiltration analysis and weighted gene co-expression network analysis identified gene modules associated with macrophage infiltration, yielding 53 macrophage-associated differentially expressed genes related to treatment response. Three machine learning methods—LASSO regression, SVM-RFE, and Boruta—then narrowed the field to KRT14 and MMP10. A prediction model built on these two genes achieved an area under the curve of 0.834 in distinguishing responders from non-responders within the analyzed data.

Survival analysis added a prognostic dimension: higher KRT14 expression was associated with favorable survival in the datasets examined. Both genes correlated with transcriptional programs related to epithelial plasticity, immune regulation, and metabolism.

The researchers also explored potential therapeutic angles using purely computational tools. Reverse network pharmacology and molecular docking predicted that the compounds β-carotene and deoxycholic acid could interact with MMP10, and molecular dynamics simulations suggested persistent protein-ligand interactions and conformational stability of the resulting complexes. These are predicted interactions, not experimentally demonstrated binding or activity.

The authors emphasize that the findings are computational and generate testable hypotheses about the biological relevance of these biomarkers and the predicted compound-target interactions. Because the study relies on retrospective public datasets and in silico modeling, the biomarker associations and predicted drug interactions require further experimental and clinical validation before any translation into patient care.

Subject of Research: Computational identification of macrophage-associated biomarkers of chemoradiotherapy response in esophageal squamous cell carcinoma

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

Article References: Dong, S., Li, L., You, S., Ren, Y., & Guo, Y. (2026). Integrative identification of macrophage-associated candidate biomarkers of chemoradiotherapy response in esophageal squamous cell carcinoma. Discover Oncology. https://doi.org/10.1007/s12672-026-06153-z

Image Credits: AI Generated

DOI: 10.1007/s12672-026-06153-z

Keywords: esophageal squamous cell carcinoma, chemoradiotherapy, tumor-associated macrophages, KRT14, MMP10, biomarkers, bioinformatics, machine learning, tumor microenvironment, molecular docking, single-cell RNA-seq, Discover Oncology