Predicting prognosis in lung adenocarcinoma by predicting TIGIT expression: a pathomics model

Area of Science:

• Oncology
• Immunology
• Genomics
• Pathology

Background:

• Traditional diagnostic methods for lung adenocarcinoma (LUAD) have limited prognostic efficacy.
• T cell immunoreceptor with immunoglobulin and immunoreceptor tyrosine-based inhibitory motif domain (TIGIT) is an emerging biomarker for LUAD.
• There is a need for improved models to predict LUAD patient prognosis.

Purpose of the Study:

• To evaluate TIGIT expression as a biomarker for LUAD.
• To develop a pathological feature-based model for predicting LUAD patient prognosis.
• To correlate pathomics score (PS) with TIGIT expression and patient survival.

Main Methods:

• Analysis of clinical data and pathological images from The Cancer Genome Atlas (TCGA).
• Prognostic value assessment of TIGIT using genetic analysis and gene set enrichment analysis (GSEA).
• Development of a pathomics model using PyRadiomics, feature selection algorithms (RFE, stepwise regression), and logistic regression.
• Evaluation of the pathomics model using ROC curves, calibration, and decision curves.

Main Results:

• TIGIT expression was significantly higher in LUAD tumor tissues and correlated with improved overall survival (OS).
• GSEA identified enrichment in TGF-β and MAPK signaling pathways.
• A pathomics model constructed from four features demonstrated good predictive performance.
• Higher pathomics scores (PS) correlated with TIGIT high expression, improved OS, and increased infiltration of CD8+ T cells and M2 macrophages.

Conclusions:

• TIGIT expression is a significant prognostic biomarker for LUAD.
• The developed pathomics model effectively predicts TIGIT expression and patient prognosis.
• TIGIT and the pathomics model hold potential for improving LUAD patient outcome prediction.