Three-Dimensional Ground Glass Opacity Ratio in CT Images Can Predict Tumor Invasiveness of Stage IA Lung Cancer

ABSTRACT

PURPOSE: We investigated the relationship between various parameters, including volumetric parameters, and tumor invasiveness according to the International Association for the Study of Lung Cancer (IASLC)/American Thoracic Society (ATS)/European Respiratory Society (ERS) classification. MATERIALS AND METHODS: We retrospectively reviewed 99 patients with completely resected stage IA lung adenocarcinoma. The correlation between several parameters [one-dimensional ground glass opacity (1D GGO) ratio, two-dimensional (2D) GGO ratio, three-dimensional (3D) GGO ratio, 1D solid size, 2D solid size, and 3D solid size] and tumor invasiveness according to IASLC/ATS/ERS classification was investigated using receiver operating characteristic (ROC) analysis. Adenocarcinoma in situ and minimally invasive adenocarcinoma were referred to as noninvasive adenocarcinoma. RESULTS: The areas under the curve (AUC) to predict invasive adenocarcinoma for the 1D, 2D, and 3D GGO ratios were 0.962, 0.967, and 0.971, respectively. The optimal cut-off values for the 1D, 2D, and 3D GGO ratios were 38%, 62%, and 74%, respectively. The AUC values for 1D, 2D, and 3D solid sizes to predict invasive adenocarcinoma were 0.933, 0.944, and 0.903, respectively. The optimal cut-off values for 1D, 2D, and 3D solid sizes were 1.2 cm, 1.5 cm2, and 0.7 cm3, respectively. The difference in the ROC curves for 3D GGO ratio and 3D solid size was significant (p=0.01). CONCLUSION: Computed tomography image-related parameters based on GGO were well correlated with and predictive of invasiveness according to IASLC/ATS/ERS classification. 3D GGO ratio was more strongly correlated with pathologic invasiveness than 3D solid size.