Abstract
Colorectal cancer (CRC) is one of the leading causes of cancer-related mortality worldwide. The current TNM (Tumor, Node, and Metastasis) classification approach is suboptimal in determining the prognosis of CRC patients. The prognosis for CRC is affected by a variety of features that are present at the initial diagnosis. Herein, we performed a systematic exploration and established a novel five-panel gene signature as a prognostic and early diagnosis biomarker after performing differential gene expression analyses in five independent in silico CRCs cohort and independently validating it in one clinical cohort, using immunohistochemistry. Four genes (BDNF, PTGS2, GSK3B, and CTNNB1) were significantly upregulated and one gene (HPGD) was significantly downregulated in primary tumor tissues compared with adjacent normal tissues throughout all the five in silico datasets. The univariate CoxPH analysis yielded a five-gene signature that accurately predicted overall survival (OS) and recurrence-free survival (RFS) in the in silico training (AUC = 0.73 and 0.69, respectively) and one independent in silico validation cohort (AUC = 0.69 and 0.74, respectively). This five-gene signature demonstrated significant associations with poor OS in independent clinical validation cohorts of colon cancer (CC) patients (AUC = 0.82). Intriguingly, a risk stratification model comprising of the five-gene signature together with TNM stage and gender status achieved an even superior AUC of 0.89 in the clinical cohorts. On the other hand, the circulating mRNA expression of the upregulated four-gene signature achieved a robust AUC = 0.83 with high sensitivity and specificity as a diagnosis marker in plasma from CRC patients. We have identified a novel, five-gene signature as an independent predictor of OS, which in combination with TNM stage and gender offers an easy-to-translate and facile assay for the personalized risk-assessment in CRC patients.