A Single-Cell Guided Machine Learning Model Predicts Response to Immune Checkpoint Inhibitors in Gastric Cancer

Area of Science:

• Immunology
• Oncology
• Genomics

Background:

• Immune checkpoint inhibitors (ICIs) show limited efficacy in gastric cancer due to drug resistance.
• Identifying cellular and molecular markers of resistance is crucial for improving treatment outcomes.

Purpose of the Study:

• To construct a single-cell transcriptomic atlas of gastric cancer to identify T/NK cell subsets associated with ICI resistance.
• To investigate the mechanisms underlying ICI resistance and identify potential therapeutic targets.

Main Methods:

• Single-cell RNA sequencing (scRNA-seq) to create a gastric cancer atlas.
• Bioinformatic analysis to identify cell populations and gene expression patterns.
• Machine learning model development for predicting immunotherapy response.
• In vitro experiments to validate the role of IRF1.

Main Results:

• A subset of T/NK cells associated with ICI resistance was identified.
• These resistant cells showed impaired MHC-I recognition, early T cell differentiation, and elevated histidine metabolism.
• The transcription factor IRF1 was identified as a suppressor of immune resistance.
• A machine learning model accurately predicted patient responses to immunotherapy across independent cohorts (AUCs 0.75 and 0.73).
• IRF1 demonstrated in vitro ability to inhibit cancer cell invasion and promote apoptosis.

Conclusions:

• Targeting the identified T/NK cell subset or restoring IRF1 function are promising strategies to overcome ICI resistance in gastric cancer.
• The study provides insights into cellular and molecular determinants of immune resistance in gastric cancer.