Integrated analysis of public datasets for the discovery and validation of survival-associated genes in solid tumors

Abstract

Identifying genes with prognostic significance that can act as biomarkers in solid tumors can help stratify patients and uncover novel therapy targets. Here, our goal was to expand our previous ranking analysis of survival-associated genes in various solid tumors to include colon cancer specimens with available transcriptomic and clinical data. A Gene Expression Omnibus search was performed to identify available datasets with clinical data and raw gene expression measurements. A combined database was set up and integrated into our Kaplan-Meier plotter, making it possible to identify genes with expression changes linked to altered survival. As a demonstration of the utility of the platform, the most powerful genes linked to overall survival in colon cancer were identified using uni- and multivariate Cox regression analysis. The combined colon cancer database includes 2,137 tumor samples from 17 independent cohorts. The most significant genes associated with relapse-free survival with a false discovery rate below 1% in colon cancer carcinoma were RBPMS (hazard rate [HR]= 2.52), TIMP1 (HR= 2.44), and COL4A2 (HR= 2.36). The three strongest genes associated with shorter survival in stage II colon cancer include CSF1R (HR= 2.86), FLNA (HR= 2.88), and TPBG (HR= 2.65). In summary, a new integrated database for colon cancer is presented. A colon cancer analysis subsystem was integrated into our Kaplan-Meier plotter that can be used to mine the entire database (https://www.kmplot.com). The portal has the potential to be employed for the identification and prioritization of promising biomarkers and therapeutic target candidates in multiple solid tumors including, among others, breast, lung, ovarian, gastric, pancreatic, and colon cancers.