Development and validation of a nomogram for predicting visceral pleural invasion in patients with early-stage non-small cell lung cancer.

Abstract

Background: Visceral pleural invasion (VPI) is associated with a poor outcome in early-stage non-small cell lung cancer (NSCLC). Preoperative prediction of VPI could have an impact on surgical planning. The aim of this study was to establish a nomogram model based on computed tomography (CT) features to predict VPI in early-stage NSCLC.

Methods: This study is a retrospective review of patients enrolled with surgically pathologically confirmed NSCLC between December 2019 and June 2022. Patients were divided into training and testing cohorts at a ratio of 7:3. Clinicopathologic and radiologic characteristics such as types of tumor pleura relationships (types I-V) were recorded. Multivariable logistic regression analysis was used to identify independent risk factors for VPI, and the optimized variables were used to build a nomogram model. Model performance was evaluated with receiver operating characteristic (ROC) curves and calibration curves. The clinical utility of the nomogram was determined using decision curve analysis (DCA).

Results: Of the 766 patients [56.9% female patients; median age, 59 years; interquartile range (IQR): 53, 66] with early-stage NSCLC, VPI was confirmed in 250 patients (32.6%). There were 536 individuals in the training cohort (172 with VPI and 364 without VPI), and 230 individuals in the testing cohort (78 with VPI and 152 without VPI). The preoperative CT features related to VPI were tumor pleura relationship of type I and type III, solid, maximum diameter of tumor, lobulation, and lymphadenopathy. There was good discriminative power in the nomogram that included these six features. The training and testing cohorts' areas under the ROC curve (AUCs) were 0.815 and 0.825, respectively, with well-fitting calibration curves. DCA demonstrated that the nomogram was clinically useful.

Conclusions: The nomogram established with the identified CT features has the potential to assist with the prediction of VPI preoperatively in early-stage NSCLC and facilitate the selection of a rational treatment strategy.