P1.11-07 Utility of the Maximum CT Value in Predicting Invasiveness of Pure GGNs

Abstract

In the current TNM classification of lung cancer, the pulmonary lesions presenting pure ground-glass nodules (GGNs) without solid component are classified as cTis tumor. However, some of them are pathologically diagnosed as invasive adenocarcinomas. This study aimed to predict the histological invasiveness using the computed tomography (CT) value in pure GGN lesions. 138 patients underwent resection of pure GGNs between 2011 and 2016. The maximum diameter and CT value were measured using a computer graphics support system. We selected the axial section which showed the densest component of each GGN. The CT value was measured separately in several areas excluding portions of apparent vessels and bronchi manually. We analyzed the correlation between the CT value of pure GGNs and the histological diagnosis, such as atypical adenomatous hyperplasia (AAH), adenocarcinoma in situ (AIS), minimally invasive adenocarcinoma (MIA) and invasive adenocarcinoma (Ad). The number of the patients with AAH, AIS, MIA and Ad was 6, 81, 45 and 6, respectively. 37% of the pure GGN lesions contained histologically invasive component and 4% of them were diagnosed as Ad. One tumor of Ad had lymphatic invasion, while there was no case with vascular invasion. 37 lobectomies, 38 segmentectomies and 63 wedge resections were performed and there was no recurrence. In comparison between the preinvasive lesions (AAH and AIS) and the invasive lesions (MIA and Ad), the latter was significantly correlated with the higher age of the patients (60 ± 9 years vs 67 ± 7 years), the larger total size (12 ± 5 mm vs 16 ± 5 mm), the higher maximum CT value (-388 ± 125 HU vs -208 ± 129 HU) and the presence of pleural indentation (odds ratio, 2.8). When the cut-off point of the maximum CT value in predicting histological invasiveness was set at -300 HU using the ROC curve analysis, the sensitivity, specificity, positive predictive value and negative predictive value were 80%, 77%, 67% and 87%, respectively. The 100% of Ad and the 78% of MIA were correctly estimated. Invasive adenocarcinoma and MIA accounted for 4% and 33% of the pure GGN lesions, respectively. The maximum CT value was correlated with the pathological diagnosis. It may be useful as a predictor of histological invasiveness. The threshold at -300 HU can be the basis of the computer-aided automatic diagnosis.