A novel miRNA-based signature as predictive tool of survival outcome of colorectal cancer patients

It is great significance of identifying valuable biomarkers for early diagnosis and prognostic prediction of colorectal cancer (CRC) patients. This study aimed at developing and validating a miRNAs-based signature as prognostic tool for CRC patients. The miRNA expression profile of 624 CRC samples (613 tumor tissues and 11 normal tissues) was analyzed, and 523 differentially expressed miRNAs (DEmiRNAs) were identified, in which 191 were downregulated and 332 were upregulated. All patients were randomly divided into a training cohort (N = 308) and an internal validation cohort (N = 200). Using the least absolute shrinkage and selection operator (LASSO) and Cox regression model, a prognostic signature of 10 miRNAs (hsa-miR-149-5p, hsa-miR-193b-5p, hsa-miR-193a-3p, hsa-miR-3677-3p, hsa-miR-29a-3p, hsa-miR-200c-5p, hsa-miR-200a-5p, hsa-miR-6854-5p, hsa-miR-216a-5p and hsa-miR-891a-5p) was developed in the training cohort. The risk score was calculated by the product of the expression level and the coefficients of each miRNA. The prognostic value of 10 miRNAs-based signature for CRC patients was tested and validated. Survival analysis indicated that high-risk patients (> 1.10) had a worse overall survival (OS) than low-risk (= 1.10) patients (5-year OS rate for training cohort: 59.3% vs. 78.9%, p < .001; validation cohort: 48.3% vs. 69.3%, p = .011). The miRNA-based signature was an independent prognostic factor for CRC patients (HR for training cohort:2.476, 95% CI:1.202-5.098, p = .014; HR for validation cohort:2.050, 95% CI:1.087-3.869, p = .027). The AUC values for 3-year and 5-year OS prediction were 0.718 and 0.784 in the training cohort, 0.659 and 0.614 in the validation cohort, respectively. The 10 miRNAs-based signature provided a proper prognostic stratification for CRC patients, and it might be a promising tool for survival prediction.