Application of 18F-deoxyglucose PET and HRCT combined prediction model in the diagnosis of early invasive lung adenocarcinoma with consolidation-to-tumor ratio ≤ 0.5 / 中华放射学杂志

ABSTRACT

Objective:To explore the value of 18F-deoxyglucose (FDG) PET and high resolution CT (HRCT) combined prediction model in the identification of invasiveness of early lung adenocarcinoma with consolidation-to-tumor ratio (CTR)≤0.5. Methods:A retrospective analysis was performed on 91 patients with early lung adenocarcinoma with CTR≤0.5 who underwent PET/CT and HRCT before surgery in the Third Affiliated Hospital of Soochow University from October 2011 to October 2019, including 110 ground-glass nodules (GGNs). According to the pathological subtypes, they were divided into preinvasive-minimally invasive adenocarcinoma (MIA) group ( n=22) and invasive adenocarcinoma (IAC) group ( n=88). The image feature parameters of GGNs of the two groups were compared, and the HRCT model and PET-HRCT combined model were constructed using Logistic regression analysis. Receiver operating characteristic (ROC) curve analysis was used to compare the diagnostic efficacy of different models. The Bootstrap resampling (times = 500) method was used for internal verification of the model, and we also performed interaction and hierarchical analysis on the model. Results:The proportions of mixed GGN, irregular shape, lobulation sign, dilated/distorted/cutoff bronchial sign, pleural indentation and vascular convergence in IAC group were significantly higher than those in preinvasive-MIA group (all P<0.05). Nodule diameter, solid component diameter, solid component ratio, CT value of ground glass attenuation component (CT GGO), and SUVindex of the IAC group were larger than those of the preinvasive-MIA group, and the differences were statistically significant ( P<0.001). Among the quantitative parameters of HRCT, CT GGO had the best diagnostic efficacy (AUC=0.775), with a sensitivity of 0.580 and a specificity of 0.909. The diagnostic efficacy of HRCT model and PET-HRCT combined model were better than CT GGO (AUC 0.907 vs. 0.775, 0.931 vs. 0.775; P=0.027, 0.002, respectively), but the diagnostic efficacy of the former two was not statistically different ( P=0.210).When the specificity was 0.909, the sensitivity of the HRCT model and the PET-HRCT model (0.784 and 0.875, respectively) were significantly higher than that of the CT GGO (0.580), and the combined PET-HRCT model had a more significant increase in sensitivity. The PET-HRCT combined model showed no significant interaction between different nodule types, between groups with or without pleural indentation, and among nodule diameter subgroups (all P>0.05). Conclusion:PET-HRCT combined model has a good predictive value for the invasiveness of early lung adenocarcinoma with CTR≤0.5, and it can be used for GGN risk stratification to guide clinical decision-making.