Transcriptome-level discovery of survival-associated biomarkers and therapy targets in non-small-cell lung cancer

Abstract

Background and purpose: Survival rate of patients with lung cancer has increased by over 60% in the recent two decades. With longer survival, the identification of genes associated with survival has emerged as an issue of utmost importance to uncover the most promising biomarkers and therapeutic targets.

Experimental approach: An integrated database was set up by combining multiple independent datasets with clinical data and transcriptome-level gene expression measurements. Univariate and multivariate survival analyses were performed to identify genes with higher expression levels linked to shorter survival. The strongest genes were filtered to include only those with known druggability.

Key results: The entire database includes 2852 tumour specimens from 17 independent cohorts. Of these, 2227 have overall survival data and 1256 samples have progression-free survival time. The most significant genes associated with survival were MIF, UBC and B2M in lung adenocarcinoma and ANXA2, CSNK2A2 and KRT18 in squamous cell carcinoma. We also aimed to reveal the best druggable targets in non-smokers lung cancer. The three most promising hits in this cohort were MDK, THY1 and PADI2. The established lung cancer cohort was added to the Kaplan-Meier plotter (https://www.kmplot.com) enabling the validation of future gene expression-based biomarkers in both the present and yet unexamined subgroups of patients.

Conclusions and implications: In this study, we established a comprehensive database of transcriptome-level data for lung cancer. The database can be utilized to identify and rank the most promising biomarkers and therapeutic targets for different subtypes of lung cancer.

Keywords: cox regression; gene arrays; gene expression; pharmacology; prognosis; survival.