Identification of a stable immunosuppressive molecular subtype in esophageal squamous cell carcinoma based on inflammation‐related genes

Abstract

Abstract Inflammation affects the progression and therapeutic response of esophageal squamous cell carcinoma (ESCC). The molecular subtypes based on inflammatory response are still under exploitation for clinical application. Expression of inflammation‐related hallmark genes was extracted from three ESCC datasets for subtype clustering using the non‐negative matrix factorization method. The most stable subtype across datasets was identified using a SubMap analysis. The immune status and molecular characteristics of the stable subtype were explored by comparing with other subtypes, and a classifier was constructed and evaluated using machine learning methods. We identified a subtype of ESCC with a higher inflammatory response score, supported by three datasets, indicating its prevalence. This subtype exhibited a high level of immune infiltration, enriched with CD8+ T cells and various immunosuppressive cells, along with elevated expression of immune checkpoints. It also showed a significant similarity to melanomas responsive to immune checkpoint inhibitor nivolumab, suggesting potential therapeutic benefits. Further analysis revealed that some hub genes in immune pathways likely contribute to the immunosuppressive microenvironment of the subtype and hold prognostic value. We developed an 11‐gene classifier for this subtype with an impressive area under the curve (AUC) of 0.9871. Applying it to external data, we identified a group with a similar immunosuppressive environment and potential response to immune checkpoint inhibitors. Generally speaking, the stable subtype we identified will be helpful to guide clinical diagnosis and treatment for ESCC.

Keywords

• esophageal squamous cell carcinoma
• gene classifier
• inflammation‐related genes
• molecular subtypes
• tumor microenvironment