Tumor microenvironment-based classification for predicting gastric cancer prognosis.

Area of Science:

• Oncology
• Pathology
• Computational Biology

Background:

• The tumor microenvironment (TME), comprising tumor-associated stroma and tumor-infiltrating lymphocytes (TIL), is vital for gastric cancer (GC) prognosis.
• Current clinical application of TME analysis for GC prognosis is limited.

Purpose of the Study:

• To develop and validate a TME-based prognostic model for gastric cancer (GC) using routine histopathology.
• To correlate TME features with specific genetic alterations in GC.

Main Methods:

• Virtual staining and image analysis of H&E-stained slides from 320 GC patients.
• Quantification of tumor-stroma ratio (TSR) and TIL to create a TME-based prediction model (TMEPATH).
• Univariate Cox regression and genomic analysis to link TME features with survival and genetic alterations.

Main Results:

• TMEPATH stratified GC patients into low-, medium-, and high-risk groups with significant survival differences (log-rank P=0.0061).
• Validation in an independent cohort confirmed the prognostic significance of TMEPATH (log-rank P=0.0064).
• TSR, TIL, and TMEPATH showed associations with microsatellite instability, tumor mutation burden, and CDH1 mutations.

Conclusions:

• Classification of GC into three TME subtypes via TSR and TIL offers a robust prognostic tool.
• The TMEPATH model provides valuable prognostic information for GC patient survival.
• This approach integrates histopathology with genomic insights for improved GC prognostication.