Objective To investigate the influencing factors for pleural effusion after laparos-copic radical resection of colorectal cancer and to construct and validate a nomogram prediction model.
Methods The retrospective cohort study was conducted. The clinical data of 713 patients who underwent laparoscopic radical resection of colorectal cancer at Tongji Hospital, Huazhong University of Science and Technology, from January 2022 to January 2023 were collected. There were 301 males and 412 females, aged (65±8) years. Patients were randomly divided into a training set of 500 cases and a validation set of 213 cases in a 7:3 ratio using a random number table. The training set was used to construct the prediction model, while the validation set was used to validate it. Observation indicators: (1) analysis of influencing factors for pleural effusion after laparoscopic radical resection of colorectal cancer; (2) construction and validation of a prediction model for pleural effusion after laparoscopic radical resection of colorectal cancer. Comparison of measurement data with normal distribution between groups was conducted using the independent samples t test. Comparison of measurement data with skewed distribution between groups was conducted using the Mann-Whitney U test. Comparison of count data between groups was conducted using the chi-square test. LASSO regression and Logistic regression models were employed to select predictors. Receiver operating characteristic (ROC) curves and area under the curve (AUC), the Hosmer-Lemeshow test, calibration curves and decision curve were performed to evaluate the efficiency of prediction model.
Results (1) Analysis of influencing factors for pleural effusion after laparoscopic radical resection of colorectal cancer: among the 713 patients, 175 cases developed postoperative pleural effusion, including 123 cases in the training set and 52 cases in the validation set. Forty-eight clinical factors from the training set were included in LASSO regression. When λ was set to 0.035, 7 non-zero coefficient indicators significantly associated with postoperative pleural effusion were identified. Binary Logistic regression analysis showed that operation time, percentage of predicted forced expiratory volume in one second (FEV1% pred) <60%, white blood cell count, preoperative serum albumin, serum sodium, procalcitonin, and D-dimer were independent factors influencing pleural effusion in patients of the training set after laparoscopic radical resection of colorectal cancer odds ratio=1.016, 8.306, 1.150, 0.911, 1.227, 1.580, 2.167, 95% confidence interval (CI) as 1.012-1.021, 4.199-17.015, 1.062-1.252, 0.861-0.961, 1.114-1.356, 1.343-1.884, 1.286-3.647, P<0.05. (2) Construction and validation of a prediction model for pleural effusion after laparoscopic radical resection of colorectal cancer: based on multivariate analysis results, a nomogram prediction model for pleural effusion after laparoscopic radical resection of colorectal cancer was constructed. ROC analysis showed that the AUC for the training set was 0.923 (95%CI as 0.893-0.952), with a sensitivity of 88.6% and specificity of 86.5%. For the validation set, these values were 0.901 (95%CI as 0.855-0.947), 86.5% and 83.2%, respectively. Calibration curves for both the training and validation sets showed high consistency with actual outcomes, indicating good model fit. The Hosmer-Lemeshow test results showed that the training set and the validation set both failed to reject the null hypo-thesis, indicating no significant deviation between the predicted values and the actual incidence, demonstrating good calibration ability (χ²=11.204, 6.897, P>0.05). Decision curve analysis demons-trated the high clinical utility of the model.
Conclusions Operation time, FEV1% pred <60%, white blood cell count, preoperative serum albumin, serum sodium, procalcitonin, and D-dimer are independent factors influencing pleural effusion in patients after laparoscopic radical resection of colorectal cancer. The nomogram prediction model constructed based on these factors exhibits excellent predictive performance.
DownLoad: