Establishment and validation of a predictive nomogram for severe pleural effusion in liver cancer patients after hepatectomy

Abstract

This study aims to develop and validate a predictive nomogram for severe postoperative pleural effusion (SPOPE) in patients undergoing hepatectomy for liver cancer. A total of 536 liver cancer patients who underwent hepatectomy at the Department of Hepatobiliary Surgery I of the Affiliated Hospital of North Sichuan Medical College from January 1, 2018, to December 31, 2022, were enrolled in a retrospective observational study and comprised the training dataset. Lasso regression and logistic regression analyses were employed to construct a predictive nomogram. The nomogram was internally validated using Bootstrapping and externally validated with a dataset of 203 patients who underwent liver cancer resection at the Department of General Surgery III of the same hospital from January 1, 2020, to December 31, 2022. We evaluated the nomogram using the receiver operating characteristic curve, calibration curve, and decision curve analysis. Variables such as drinking history, postoperative serum albumin, postoperative total bilirubin, right hepatectomy, diaphragm incision, and intraoperative blood loss were observed to be associated with SPOPE. These factors were integrated into our nomogram. The C-index of the nomogram was 0.736 (95% CI: 0.692-0.781) in the training set and 0.916 (95% CI: 0.872-0.961) in the validation set. The nomogram was then evaluated using sensitivity, specificity, positive predictive value, negative predictive value, calibration curve, and decision curve analysis. The nomogram demonstrates good discriminative ability, calibration, and clinical utility.

Figure 1.

Flowchart of this study.

Figure 2.

Variable selection by the LASSO regression. (A) LASSO lambda graph, it is used to describe the relationship between coefficients and regularization parameters in Lasso regression. (B) lambda.pdf, 10 variables with nonzero coefficients were selected by deriving the optimal lambda.

Figure 3.

Multivariate logistic regression analysis and forest plot based on training dateSet.

Figure 4.

Nomogram to estimate the probability of severe postoperative pleural effusion. POST-ALB = postoperative serum albumin, POST-TBIL = postoperative total bilirubin.

Figure 5.

Receiver operating characteristic (ROC) curves of nomograms in the training dataset and validation dataset, respectively. (A) The AUC value of the training dataset is 0.736, 95% CI: 0.692–0.781 (P < .05). (B) The AUC value of the validation dataset is 0.916, 95% CI: 0.872–0.961 (P < .05).

Figure 6.

The calibration curve of the nomogram for predicting severe postoperative pleural effusion in the training dataset (A) and validation dataset (B), respectively.

Figure 7.

Decision curve analysis for the severe postoperative pleural effusion predictive nomogram. (A) Training dataset. (B) Validation dataset.