Preoperative diagnostic accuracy of dominant lesion in multifocal pulmonary nodules

Due to the widespread implementation of computed tomography (CT) in lung cancer screening, multifocal pulmonary nodules (MPNs) are increasingly detected. Given the importance of selecting preoperative dominant lesions (DLs) in the management of MPNs, this study evaluates the diagnostic accuracy of commonly used approaches for assessment of preoperative DLs.

Patients who underwent surgical resection and CT for MPNs from May 2019 to September 2023 were retrospectively collected from a single center. The postoperative DLs were determined based on the pathology results. Four methods were employed to identify preoperative DLs including diameter, Mayo model, Brock model, and Peking University multiple pulmonary nodules malignancy prediction model (PKU-M model) and the predictive results of these methods were compared with postoperative DLs. Subgroup analysis was conducted based on the type of nodules.

A total of 999 patients with 2,285 nodules were included in this study. The accuracy of the proposed methods including diameter, Mayo model, Brock model, and PKU-M model for the assessment of preoperative DLs were 81.09%, 78.69%, 81.73%, and 78.77%, respectively. Compared to the pathology results, the Kappa values for the four methods were 0.62, 0.51, 0.59, and 0.51, respectively. Among the three subgroups, four methods applied in subsolid nodule group demonstrated the best performance with accuracy of 83.07%, 77.88%, 81.90% and 72.59%, respectively.

Current assessment approaches for identifying preoperative DLs still have room for improvement and further studies are warranted to develop a more effective approach for the assessment of preoperative DLs.