Study on influencing factors and predictive model construction of cardiopulmonary complication after thoracoscopic surgery in stage Ⅰ-Ⅱ non-small cell lung cancer

LI Jianlin; SUN Sijin; WANG Dali

doi:10.11958/20250702

Abstract：Objective To construct a multivariable prediction model for assessing the risk of cardiopulmonary complication after thoracoscopic lobectomy in patients with stage Ⅰ-Ⅱ non-small cell lung cancer(NSCLC).Methods Clinical data of 600 patients with stage Ⅰ-Ⅱ NSCLC who underwent thoracoscopic lobectomy were retrospectively analyzed.Patients were divided into the complication group(84 cases)and the non-complication group(516 cases)based on the occurrence of postoperative complication within 7 days,including atelectasis,pulmonary embolism,respiratory failure,chylothorax,massive pleural effusion,hypoxemia and atrial fibrillation.Demographic characteristics,preoperative pulmonary function assessment,pathological features and perioperative indicators were collected.Multivariate Logistic regression analysis was used to identify independent risk factors influencing postoperative cardiopulmonary complication in patients and construct a predictive model.Internal validation was performed using the Bootstrap resampling method(1 000 iterations)to evaluate the discrimination,calibration and clinical decision-making value of the model.Results Multivariate Logistic regression analysis identified the following independent risk factors of postoperative cardiopulmonary complication,including age(OR=1.832,95%CI:1.537-2.183),history of chronic obstructive pulmonary disease(COPD)(OR=6.782,95%CI:2.685-17.130),Karnofsky performance status(KPS)score(OR=0.926,95%CI:0.888-0.965),the percentage of forced expiratory volume in the first second to the predictive value(FEV1%pred)(OR=0.906,95%CI:0.845-0.972),the percentage of diffusing capacity for carbon monoxide to the expected value(DLCO%pred)(OR=0.901,95%CI:0.832-0.975),intraoperative blood loss(OR=1.025,95%CI:1.014-1.036)and one-lung ventilation time(OR=1.057,95%CI:1.034-1.080).The area under the curve(AUC)of the combined diagnosis was 0.977(95%CI:0.965-0.989),with 96.4%sensitivity and 87.6%specificity.The Hosmer-Lemeshow test indicated excellent calibration(χ²=1.285,P=0.994).Decision curve analysis demonstrated significant clinical net benefit when the risk threshold probability ranged between 20%and 98%.Conclusion The multivariable prediction model for cardiopulmonary complication after thoracoscopic lobectomy in stage Ⅰ-Ⅱ NSCLC patients exhibits strong predictive performance.