Machine Learning Defines TIME Subtypes Linking EGFR Mutations to Immune Profiles in LUAD

Overview

A machine learning-integrated single-cell transcriptomic analysis of 153 lung adenocarcinoma (LUAD) samples identified distinct tumor immune microenvironment (TIME) subtypes associated with EGFR mutation status. EGFR-mutant tumors showed immunosuppressive profiles enriched with TIGIT+ regulatory T cells, neutrophils, and macrophages, whereas wild-type tumors exhibited immune-active profiles with abundant cytotoxic and memory immune cells.

Background

Lung adenocarcinoma (LUAD) is the most common subtype of non-small cell lung cancer, with a high prevalence of EGFR mutations, especially among Asian females. EGFR-mutant LUAD typically exhibits an immunosuppressive tumor microenvironment characterized by fewer cytotoxic T cells and lower PD-L1 expression, contributing to poor responses to immune checkpoint inhibitors. Single-cell RNA sequencing combined with machine learning offers a powerful approach to dissect the complex immune landscapes and heterogeneity within LUAD, enabling better understanding of immune differences between EGFR-mutant and wild-type tumors.

Data Highlights

Key Findings

• EGFR-mutant LUAD tumors are enriched with immunosuppressive cells including TIGIT+ regulatory T cells, neutrophils, and macrophages.
• Wild-type LUAD tumors contain higher levels of immune-active cells such as ZNF683+ CD8+ tissue-resident memory T cells, diverse memory B cells, and FGFBP2+ CD16high natural killer (NK) cells.
• Non-negative matrix factorization identified five distinct TIME subtypes, with EGFR-mutant patients clustering into immunosuppressive profiles associated with worse clinical outcomes.
• FGFBP2+ NK cells exhibit cytotoxic functions and enhance the efficacy of PD-1 blockade immunotherapy, as confirmed by flow cytometry and mouse models.
• Machine learning-based immunogenomic analysis effectively delineates immune heterogeneity and provides a framework for precision immunotherapy in LUAD.

Clinical Implications

Understanding the distinct immune microenvironment subtypes in EGFR-mutant versus wild-type LUAD can guide personalized immunotherapy strategies. The identification of immunosuppressive TIME profiles in EGFR-mutant tumors highlights the need for novel therapeutic approaches beyond conventional immune checkpoint inhibitors. Enhancing cytotoxic NK cell activity, particularly FGFBP2+ NK cells, may improve immunotherapy responses in EGFR-mutant LUAD patients.

Conclusion

This comprehensive machine learning-driven single-cell analysis reveals that EGFR mutations in LUAD are associated with distinct immunosuppressive tumor microenvironments, which correlate with poor prognosis and reduced immunotherapy efficacy. These insights provide a foundation for developing tailored immunotherapeutic interventions targeting specific TIME subtypes in EGFR-mutant lung adenocarcinoma.

References

1. Author/Source/Year -- Machine Learning Uncovers TIME Subtypes Correlating EGFR Mutations with Immune Profiles in Lung Adenocarcinoma