Interpreting Lung Cancer Health Disparity at Transcriptome Level

Area of Science:

• Genomics
• Computational Biology
• Health Disparities Research

Background:

• Lung cancer exhibits significant disparities in incidence and outcomes across racial and sex groups.
• Understanding the genetic underpinnings of these disparities is crucial for developing targeted treatments.
• Existing data imbalance poses challenges for race-based genetic analyses in lung cancer.

Purpose of the Study:

• To identify patient- and cohort-specific biomarker genes contributing to lung cancer health disparities.
• To investigate disparities among African American males (AAMs), European American males (EAMs), African American females (AAFs), and European American females (EAFs).
• To develop a computational framework that overcomes data imbalance issues for accurate genetic analysis.

Main Methods:

• Developed a computational framework using disease conditions (Lung Adenocarcinoma, Lung Squamous Cell Carcinoma, Healthy) instead of race for classification.
• Leveraged explainable AI (SHAP) for local interpretability to identify disparity-related genes.
• Designed four classification tasks: one 3-class (LUAD-LUSC-HEALTHY) and three 2-class (LUAD-LUSC, LUAD-HEALTHY, LUSC-HEALTHY).

Main Results:

• Successfully identified sets of genes and pathways associated with lung cancer health disparities.
• Discovered genetic factors contributing to disparities between AAMs vs. EAMs, AAFs vs. EAFs, AAMs vs. AAFs, and EAMs vs. EAFs.
• Provided a concise list of genes and pathways for further experimental validation.

Conclusions:

• The developed computational framework effectively identifies genetic contributors to lung cancer health disparities.
• The findings offer a foundation for precision medicine approaches to address racial and sex-based inequities in lung cancer.
• The identified biomarkers warrant further investigation in wet lab experiments to validate their role in lung cancer disparities.