A Practical Nomogram and Risk Stratification System Predicting Cancer-Specific Survival for Hepatocellular Carcinoma Patients With Severe Liver Fibrosis

Abstract

Objective: Hepatocellular carcinoma (HCC) is the second leading cause of cancer-related deaths worldwide. This study aims to construct a novel practical nomogram and risk stratification system to predict cancer-specific survival (CSS) in HCC patients with severe liver fibrosis.

Methods: Data on 1,878 HCC patients with severe liver fibrosis in the period 1975 to 2017 were extracted from the Surveillance, Epidemiology, and End Results database (SEER). Patients were block-randomized (1,316 training cohort, 562 validation cohort) by setting random seed. Univariate and multivariate COX regression analyses were employed to select variables for the nomogram. The consistency index (C-index), the area under time-dependent receiver operating characteristic curve (time-dependent AUC), and calibration curves were used to evaluate the performance of the nomogram. Decision curve analysis (DCA), the C-index, the net reclassification index (NRI), and integrated discrimination improvement (IDI) were used to compare the nomogram with the AJCC tumor staging system. We also compared the risk stratification of the nomogram with the American Joint Committee on Cancer (AJCC) staging system.

Results: Seven variables were selected to establish the nomogram. The C-index (training cohort: 0.781, 95%CI: 0.767-0.793; validation cohort: 0.793, 95%CI = 95%CI: 0.779-0.798) and the time-dependent AUCs (the training cohort: the values of 1-, 3-, and 5 years were 0.845, 0.835, and 0.842, respectively; the validation cohort: the values of 1-, 3-, and 5 years were 0.861, 0.870, and 0.876, respectively) showed satisfactory discrimination. The calibration plots also revealed that the nomogram was consistent with the actual observations. NRI (training cohort: 1-, 2-, and 3-year CSS: 0.42, 0.61, and 0.67; validation cohort: 1-, 2-, and 3-year CSS: 0.26, 0.52, and 0.72) and IDI (training cohort: 1-, 3-, and 5-year CSS:0.16, 0.20, and 0.22; validation cohort: 1-, 3-, and 5-year CSS: 0.17, 0.26, and 0.30) indicated that the established nomogram significantly outperformed the AJCC staging system (P < 0.001). Moreover, DCA also showed that the nomogram was more practical and had better recognition.

Conclusion: A nomogram for predicting CSS for HCC patients with severe liver fibrosis was established and validated, which provided a new system of risk stratification as a practical tool for individualized treatment and management.

Keywords: cancer-specific survival; hepatocellular carcinoma; nomogram; risk stratification system; severe liver fibrosis.

Figure 1

Flow diagram of the hepatocellular carcinoma patients with severe liver fibrosis with training and validation cohorts.

Figure 2

A nomogram for hepatocellular carcinoma patients with severe liver fibrosis.

Figure 3

ROC of the nomogram for 1-, 3-, and 5-year prediction. (A) Training cohorts based on the nomogram; (B) Validation cohorts based on the nomogram.

Figure 4

Calibration plots of 1-year, 3-year, and 5-year CSS for hepatocellular carcinoma patients with severe liver fibrosis. (A,C,E) Calibration plot of 1-year, 3-year, and 5-year CSS in the training cohort; (B,C,F) Calibration plot of 1-year, 3-year, and 5-year CSS in the training cohort; CSS, cancer-specific survival.

Figure 5

Decision curve analysis. (A,C,E) DCA curve of 1-year, 3-year, and 5-year CSS in the training cohort; (B,D,F) DCA curve of 1-year, 3-year, and 5-year CSS in the validation cohort. DCA, decision curve analysis; CSS, cancer-specific survival.

Figure 6

C-index analysis. (A) The nomogram related C-index; (B) AJCC staging criteria related C-index.

Figure 7

Cut-off point for risk stratification selected using X-tile.

Figure 8

Kaplan–Meier CSS curves of hepatocellular carcinoma patients with severe liver fibrosis based on different criteria. (A,B) Kaplan–Meier CSS curves of training and validation cohorts based on the new risk stratification system; (C,D) Kaplan–Meier CSS curves of training and validation cohorts based on AJCC staging criteria.