Abstract
BACKGROUND: According to the World Health Organization (WHO) Classification of Thoracic Tumors published in 2021, lung adenocarcinoma (LUAD) includes minimally invasive adenocarcinoma (MIA) and invasive adenocarcinoma (IAC). The invasive component of MIA is ≤5 mm, while IAC is, on the contrary. This difference in the extent of invasion leads to distinct biological behaviors, which correspond to different treatment approaches and prognostic outcomes. Therefore, this study aimed to construct and validate a radiomics-based nomogram for preoperatively predicting the invasiveness of LUAD appearing as pulmonary nodules. METHODS: From January 2020 to June 2024, data of a total of 611 pulmonary nodule patients who underwent preoperative computed tomography (CT) examinations at three centers were retrospectively analyzed. All patients were pathologically diagnosed with LUAD, including IAC and MIA. Individuals from the primary center were randomly divided into a training cohort and an internal validation cohort at a ratio of 7:3, while patients from the other two centers were included in an external validation cohort. Firstly, univariable and multivariable logistic regression (LR) analyses were performed to develop a clinical model. Secondly, three representative machine learning classifiers were used to construct radiomics models, and their performances were compared to select the optimal one. Finally, with the integration of the clinical and radiomics signatures, a combined model was established, and a nomogram was generated to visualize the risk scores. The diagnostic performance was evaluated with the receiver operating characteristic (ROC) curves, the goodness-of-fit was verified through the calibration curves, and the clinical utility was assessed by the decision curve analysis (DCA). RESULTS: Six hundred and eleven patients were recruited, with 356 individuals in the training cohort (IAC, 227; MIA, 129), 153 in the validation cohort (IAC, 108; MIA, 45), and 102 in the test cohort (IAC, 70; MIA, 32). The combined model worked robustly and effectively, with areas under the curves (AUCs) of 0.968 [95% confidence interval (CI), 0.953-0.984], 0.902 (95% CI, 0.856-0.949), and 0.899 (95% CI, 0.839-0.959) in the training, validation, and test cohorts, respectively, demonstrating an excellent predictive power, a good model fit, and a significant clinical utility. CONCLUSIONS: This radiomics-based nomogram could serve as a clinical predictive model for preoperatively predicting the invasiveness of LUAD, which contributes to the future development of more individualized therapeutic strategies.