Title: An integrative informatics approach towards precision medicine in lung cancer

ABSTRACT

Clinical cancer genomics is at the forefront of precision medicine where diverse treatment strategies are adopted, including targeted therapy, chemotherapy, and immunotherapy. Choosing proper treatment options requires an elaborate analysis of patient characteristics typically based on multi-omics data. In this talk, I will present two multi-omics studies performed to realize the precision medicine in lung adenocarcinoma (LUAD). The first project focuses on devising patient stratification schemes to recommend patient-specific treatment options based on exome and transcriptome data from 111 LUAD patients in never-smoker Korean females. Our patient stratification method was able to identify somatic driver aberrations and expression signatures that could be utilized for cancer immunotherapy. In the second project, we developed a computational model for predicting treatment response to PD-1/PD-L1 inhibitors in LUAD patients. We analyzed the mutation and expression data from 85 LUAD patients who received anti-PD-1 (Pembrolizumab) treatment. We found that TMB was a good predictor (AUC=0.8) for the PD-L1 negative patients (n=20). For the PD-L1 positive patients (n=65), we built an ensemble model of 100 XGBoost learning machines where gene set activities and cell type composition, both inferred from transcriptome sequencing data, were used as input features. The transcriptome-based model achieved the highest performance of AUC=0.91 and highlighted the contribution of T cell activities, tumor microenvironments, and novel signatures of negative T cell regulators. Our data and analysis results provides valuable insights for developing biomarkers and strategies to realize precision medicine in LUAD.