Development and validation of nomogram to predict overall survival and disease-free survival after surgical resection in elderly patients with hepatocellular carcinoma

Abstract

Background: Hepatocellular carcinoma (HCC) is one of the common causes of tumor death in elderly patients. However, there is a lack of individualized prognostic predictors for elderly patients with HCC after surgery.

Method: We retrospectively analyzed HCC patients over 65 years old who underwent hepatectomy from 2015 to 2018, and randomly divided them into training cohort and validation cohort in a ratio of 3:1. Univariate Cox regression was used to screen the risk factors related to prognosis. Prognostic variables were further selected by least absolute shrinkage and selection operator regression model (LASSO) and multivariate Cox regression to identify the predictors of overall survival (OS) and disease-free survival (DFS). These indicators were then used to construct a predictive nomogram. The receiver operating characteristic curve (ROC curve), calibration curve, consistency index (C-index) and decision analysis curve (DCA) were used to test the predictive value of these independent prognostic indicators.

Result: A total of 188 elderly HCC patients who underwent hepatectomy were enrolled in this study. The independent prognostic indicators of OS included albumin (ALB), cancer embolus, blood loss, viral hepatitis B, total bilirubin (TB), microvascular invasion, overweight, and major resection. The independent prognostic indicators of DFS included major resection, ALB, microvascular invasion, laparoscopic surgery, blood loss, TB, and pleural effusion. In the training cohort, the ROC curve showed that the predictive values of these indicators for OS and DFS were 0.827 and 0.739, respectively, while in the validation cohort, they were 0.798 and 0.694. The calibration curve nomogram exhibited good prediction for 1-year, 2-year, and 3-year OS and DFS. Moreover, the nomogram models exhibited superior performance compared to the T-staging suggested by C-index and DCA.

Conclusion: The nomogram established in this study demonstrate commendable predictive efficacy for OS and DFS in elderly patients with HCC after hepatectomy.Core Tip: The purpose of this retrospective study is to screen the risk factors of survival and recurrence in elderly patients with HCC after hepatectomy. The nomogram included cancer embolus, viral hepatitis B, overweight, major resection, ALB, microvascular invasion, laparoscopic surgery, blood loss, TB, and pleural effusion as predictors. The calibration curve of this nomogram was good, indicating credible predictive value and clinical feasibility.

Keywords: elderly; hepatocellular carcinoma; nomogram; prognosis; recurrence.

Figure 1

Flow chart of included and excluded patients in the study.

Figure 2

Screening of variables based on Lasso regression. (A) OS, (C) DFS: The selection process of the optimum value of the parameter λ in the Lasso regression model by cross-validation method. (B) OS, (D) DFS: The variation characteristics of the coefficient of variables.

Figure 3

(A) Nomogram prediction model and prediction OS curve; (B) AUC for predicting OS in training cohort; (C) Calibration curves of 1-, 2-, and 3-year OS in training cohort; (D) AUC for predicting OS in validation cohort; (E) Calibration curves of 1-, 2-, and 3-year OS in the validation cohort.

Figure 4

(A) Nomogram prediction model and prediction DFS curve; (B) AUC for predicting DFS in training cohort; (C) Calibration curves of 1-, 2-, and 3-year DFS in training cohort; (D) AUC for predicting DFS in validation cohort; (E) Calibration curves of 1-, 2-, and 3-year DFS in the validation cohort.

Figure 5

The C-index of the nomograms and T-staging. (A) The C-index of the OS nomogram and T-staging; (B) The C-index of the DFS nomogram and T-staging.

Figure 6

(A) DCA of OS in training cohort; (B) DCA of OS in validation cohort; (C) DCA of DFS in training cohort; (D) DCA of DFS in validation cohort.

Figure 7

Kaplan-Meier curves for OS in the low-risk and high-risk groups in training cohort (A) and validation cohort (B); Kaplan-Meier curves for DFS in the low-risk and high-risk groups in training cohort (C) and validation cohort (D).