A novel molecular-clinicopathologic nomogram to improve prognosis prediction of hepatocellular carcinoma

Background: Emerging evidence suggests that long non-coding RNA (lncRNA) plays a crucial part in the development and progress of hepatocellular carcinoma (HCC). The objective was to develop novel molecular-clinicopathological prediction methods for overall survival (OS) and recurrence of HCC. Results: An 8-lncRNA-based classifier for OS and a 14-lncRNA-based classifier for recurrence were developed by LASSO COX regression analysis, both of which had high accuracy. The tdROC of OS-nomogram and recurrence-nomogram indicates the satisfactory accuracy and predictive power. The classifiers and nomograms for predicting OS and recurrence of HCC were validated in the Test and GEO cohorts. Conclusions: These two lncRNA-based classifiers could be independent prognostic factors for OS and recurrence. The molecule-clinicopathological nomograms based on the classifiers could increase the prognostic value. Methods: HCC lncRNA expression profiles from the cancer genome atlas (TCGA) were randomly divided into 1:1 training and test cohorts. Based on least absolute shrinkage and selection operator method (LASSO) COX regression model, lncRNA-based classifiers were established to predict OS and recurrence, respectively. OS-nomogram and recurrence-nomogram were developed by combining lncRNA-based classifiers and clinicopathological characterization to predict OS and recurrence, respectively. The prognostic value was accessed by the time-dependent receiver operating characteristic (tdROC) and the concordance index (C-index).