Background: Frozen sections (FS) deferral sometimes occurs in the intraoperative pathological classification of early lung adenocarcinoma, which is not conducive to the decision-making of surgical treatment. Here, we compared the predictive performance of the combined nomogram based on the computer tomography (CT) features with FS to investigate whether the nomogram could be used as a complementary method for FS when FS deferral occurs to predict invasive adenocarcinoma (IAC) manifesting as ground-glass nodules (GGNs) during surgery.
Methods: In this study, 205 early lung adenocarcinomas manifesting as GGNs from 178 patients who had undergone surgical treatment were included and divided into a training set (n=123) and a validation set (n=82). The training set defined a hybrid nomogram incorporating CT features and intraoperative measured tumor size based on multivariate logistic regression to predict IAC, and the validation set was used to verified the predictive performance. We also collected the diagnostic results of FS and compared the predictive performance of the established nomogram with FS.
Results: The accuracy of combined nomogram in predicting IAC in the training and validation sets was 91.1% and 89.0%, respectively, and the predictive accuracy of FS in the training set and validation set was 87.0% and 86.6%, respectively. The predictive accuracy between the combined nomogram and FS have no significant difference.
Conclusions: Compared with FS, the performance of the combined nomogram in predicting the lung IAC manifesting as GGNs is satisfactory, which has the potential to be used as a complementary method for FS when FS deferrals during surgery.
Keywords: Nomogram; forecasting; frozen section (FS); lung neoplasms; spiral computed tomography.
2020 Journal of Thoracic Disease. All rights reserved.