Abstract:
Objective To assess the value of dual-layer spectral detector CT in distinguishing between lung minimally invasive adenocarcinoma (MIA) and invasive adenocarcinoma (IA) manifesting as a pure ground-glass nodule (pGGN).
Method Clinical and imaging data of 101 patients (103 pGGN) with pGGN on preoperative chest enhanced dual-layer spectral detector CT and pathologically confirmed MIA and IA from August 2019 to June 2021 at Tianjin Medical University Cancer Institute & Hospital were retrospectively analyzed. pGGN cases were assigned into the MIA group (n=47) and IA group (n=56). Independent sample t-test, Mann-Whitney U test, and χ2 test were used to compare differences in CT features and quantitative parameters between the two groups. Multivariate Logistic regression was used to identify independent predictors of IA, and a predictive model for IA was constructed. The diagnostic performance of each independent predictors and the model was evaluated using ROC curve analysis, and DeLong test was used to compare the diagnostic efficacy of the model and independent predictors.
Results The IA group had higher frequencies of irregular shape (P=0.034), pleural retraction (P=0.005), air bronchogram (P=0.001), and vascular abnormalities (P<0.001) than the MIA group. Nodule size (P<0.001), CT value of traditional 120 kVp images (CT120 kVp, P=0.001), and 40–200 keV monochromatic images (CT40-200 keV, P≤0.003) were significantly higher in the IA group than in the MIA group, while the effective atomic number (Eff-Z) value was significantly lower in the IA group than in the MIA group (P=0.018). Logistic regression analysis showed that nodule size (OR=1.435, 95%CI:1.204-1.709, P<0.001) and CT200 keV (OR=1.011, 95%CI:1.005-1.016, P<0.001) were independent factors for predicting IA. ROC curve analysis showed that the area under the curve (AUC), sensitivity, and specificity of the predictive model were 0.855, 0.661, and 0.957, respectively, and AUC was significantly higher than nodule size (P=0.046) and CT200 keV (P=0.002).
Conclusions Dual-layer spectral detector CT is of great value in differentiating between MIA and IA manifesting as pGGNs and the predictive model established by Logistic regression analysis has good accuracy.