A novel miRNA-based model for predicting the 3-year recurrence risk of hepatocellular carcinoma following liver transplantation.

BACKGROUND

Tumor recurrence is one of the strongest survival-limiting factors for patients with hepatocellular carcinoma (HCC) receiving liver transplantation (LT). This study aimed to develop a novel microRNA (miRNA)-based model for predicting the risk of HCC recurrence within 3 years following LT.

METHODS

Based on two public datasets [including the GSE30297 dataset and The Cancer Genome Atlas-liver hepatocellular carcinoma (TCGA-LIHC) dataset], the core miRNAs related to HCC recurrence following LT were determined by synthetically using differential expression analysis, Kaplan-Meier method and least absolute shrinkage and selection operator (LASSO)-logistic regression. Multivariate logistic regression was utilized to calculate the coefficients of each miRNA for model construction. Models were comprehensively evaluated by concordance index, calibration plots, decision curve analysis, and receiver operating characteristic curves.

RESULTS

Five core miRNAs (including miR-454, miR-1293, miR-139, miR-99a, and miR-130a) related to HCC recurrence within 3 years following LT were identified. The five miRNA-based risk score displayed a good predictive efficacy for the 3-year recurrence risk of HCC post-LT [area under the curve (AUC) =0.901, P<0.001]. Furthermore, a nomogram integrating the miRNA-based model with Milan criteria (MC) was developed and exhibited a high predictive accuracy (AUC =0.926, P<0.001) and clinical applicability, which was remarkably better than that of MC (AUC =0.629, P<0.001).

CONCLUSIONS

This model may contribute to the reasonable selection of LT candidates in HCC patients and the personalized postoperative management.