Development and validation of a circulating serum-based six- microRNA signature for predicting hepatocellular carcinoma

Abstract

Abstract Background: Hepatocellular carcinoma (HCC) is one of the leading causes of cancer-related death, and early detection and treatment play an important role in improving prognosis.. Methods: The prediction model establishment and validation were conducted in GSE113740. Herein, we focused on the differentially expressed miRNAs, highly detected miRNAs and accurately diagnostic capability to identify miRNA candidates. Lasso regression, univariate, multivariate logistic analysis was used to construct a novel diagnostic model based on circulating miRNAs, hoslem.test was used to perform goodness of fit tests. The performance of the model was assessed with respect to its area under curve (AUC), sensitivity and specificity. Finally, we added AFP to the model and applied AUC and the integrated discrimination improvement (IDI) to compare whether the new model had an improvement in two cohorts. Results：The diagnostic model characterized by six miRNAs was established in the training set. hoslem.test p-value = 0.997. The AUC of the model in the training set was 0.995 (95%CI: 0.987-0.999), the sensitivity and specificity were 95.98% and 97.22%, respectively. In the validation set, the AUC was 0.977(95%CI: 0.964-0.987), the sensitivity was 94.74%, and the specificity was 93.29%. The new model with AFP showed an improvement in both the training and validation sets. Conclusions：This study presents a diagnostic model that incorporates six-microRNA signature, which can be used to facilitate the prediction of patients with HCC.