Imaging features of TSCT predict the classification of pulmonary preinvasive lesion, minimally and invasive adenocarcinoma presented as ground glass nodules

Objectives: To comprehensively investigate the role of thin section computed tomography(TSCT) features to distinguish invasive adenocarcinoma(IA) from preinvasive or minimally invasive adenocarcinoma(MIA) appearing as pure or mixed ground glass nodules (pGGNs, mGGNs), and to distinguish adenocarcinoma in situ (AIS) from atypical adenomatous hyperplasia (AAH) in pGGNs. Materials and methods: Three hundred thirteen patients with 334 pathologically diagnosed GGNs according to the 2011 IASLC/ATS/ERS classification were included into this study. The TSCT features of the AAH-MIAs and IAs were compared and analyzed respectively in pGGNs (158 cases) and mGGNs (176 cases). Additionally, AIS (30 cases) and AAH (33 cases) were further analyzed in pGGNs. Receiver operating characteristic(ROC) analysis were performed to determine the cutoff values for the qualitative variables and their diagnostic performances. Results: In pGGNs, significant differences were found in the tumor volume(p = 0.017, OR = 4.98, 95%CI 1.33-18.62) and tumor mass(p = 0.03, OR = 5.04, 95%CI 1.17-21.59) between AAH-MIAs (AAH, AIS, MIA) group and IAs group, and tumor mass(p = 0.037, OR= 4.32, 95%CI 1.09-17.10) and standard deviation(SD) (p = 0.019, OR= 13.92, 95%CI 1.53-126.57) could distinguish AIS from AAH. In mGGNs, significant differences were found in consolidation size (p = 0.006, OR= 21.98, 95%Cl 2.46-196.67) and consolidation mean CT value (p = 0.011, OR = 18.20, 95%CI1.96-168.88) between AAH-MIAs group and IAs group. Multivariate and ROC analyses revealed that in pGGNs, tumor size (>= 1125 mm) and mass (>386) were significantly associated with IAs. SD (>= 68) and mass (>= 70) were significant in distinguishing AIS from AAH. Larger consolidation of nodules (>= 8.1 mm) and higher CT values of the solid components (>=-222 HU) in mGGNs were significantly associated with lAs. Conclusion: TSCT features can help distinguish lAs from AAH-MIAs both in pGGNs and mGGNs, and identify AIS from AAH in pGGNs, which indicated that imaging features may be helpful to guide the therapeutic choice for patients with GGNs which were considered as high risk of malignant diseases. (C) 2017 Elsevier B.V. All rights reserved.