Assessment of high-grade pattern in lung invasive nonmucinous adenocarcinoma based on multiparametric features from dual-layer detector spectral computed tomography combined with extracellular volume fraction

Background: The International Association for the Study of Lung Cancer (IASLC) introduced a new grading system for invasive nonmucinous adenocarcinoma (INMA), in which complex glandular, micropapillary, and solid subtypes are classified as high-grade patterns (HGPs). Since a higher proportion of HGP is associated with poorer prognosis, accurate assessment is essential for individualized treatment. The aim of this study was to assess the value of conventional computed tomography (CT) features and spectral CT quantitative parameters, particularly extracellular volume (ECV) fraction, in predicting HGP ≥20% in INMA. Methods: test. Significant variables from univariate analysis were entered into multivariate logistic regression to identify independent predictors of HGP ≥20%. Subsequently, three predictive models (conventional, spectral, and combined) were developed. Model performance was quantified via the area under the receiver operating characteristic (ROC) curve (AUC) and compared via the DeLong test, and clinical utility was assessed through decision curve analysis (DCA). Results: (P=0.004), lobulation sign (P=0.007), and spiculation sign (P=0.006). The AUCs of the conventional CT model, spectral CT quantitative parameter model, and combined model for predicting HGP ≥20% were 0.825, 0.829, and 0.909, respectively. The combined model demonstrated significantly superior performance compared with the conventional CT model (Z=3.350; P=0.001) and the spectral CT model (Z=2.817; P=0.005). The DCA indicated that the combined model provides more clinical benefits than do the other models. Conclusions: The combined model can aid in the prediction of the HGP in INMA. This method involves the use of EVC, allowing for a noninvasive diagnosis of HGP, which is critical for patient care.