Genome-Wide DNA Methylation and Gene Expression Profiling Characterizes Molecular Subtypes of Esophagus Squamous Cell Carcinoma for Predicting Patient Survival and Immunotherapy Efficacy.

Abstract

Simple SummaryEsophageal squamous cell carcinoma (ESCC) represents roughly 85–90% of all esophageal carcinoma patients in China. Immunotherapy is used to treat an increasing number of ESCC patients in clinical practice. This study aims to understand the molecular heterogeneity and the tumor immune microenvironment of ESCC for designing novel immunotherapies to improve response and outcomes. We identified two molecular subtypes associated with prognosis, immune-related pathways, and tumor microenvironment. In an independent cohort of Chinese ESCC patients treated with immunotherapy, the response rate of the S1 subtype is significantly higher than the S2 subtype. These findings provide a new perspective on the molecular subtyping for ESCC and a biological rationale for novel therapeutic intervention in a specific subgroup of ESCC that could potentially be translated into clinical practice both diagnostically and therapeutically to benefit ESCC patients.Background: Immunotherapy is recently being used to treat esophageal squamous cell carcinoma (ESCC); however, response and survival benefits are limited to a subset of patients. A better understanding of the molecular heterogeneity and tumor immune microenvironment in ESCC is needed for improving disease management. Methods: Based on the DNA methylation and gene expression profiles of ESCC patients, we identify molecular subtypes of patients and construct a predictive model for subtype classification. The clinical value of molecular subtypes for the prediction of immunotherapy efficacy is assessed in an independent validation cohort of Chinese ESCC patients who receive immunotherapy. Results: We identify two molecular subtypes of ESCC (S1 and S2) that are associated with distinct immune-related pathways, tumor microenvironment and clinical outcomes. Accordingly, S2 subtype patients had a poorer prognosis. A 15-gene expression signature is developed to classify molecular subtypes with an overall accuracy of 94.7% (89/94, 95% CI: 0.880–0.983). The response rate of immunotherapy is significantly higher in the S1 subtype than in the S2 subtype patients (68.75% vs. 25%, p = 0.028). Finally, potential target drugs, including mitoxantrone, are identified for treating patients of the S2 subtype. Conclusions: Our findings demonstrated that the identified molecular subtypes constitute a promising prognostic and predictive biomarker to guide the clinical care of ESCC patients.