CT diagnosis of pleural and stromal invasion in malignant subpleural pure ground-glass nodules: an exploratory study.

<bold>Objectives: </bold>To assess the risk of visceral pleural invasion (VPI) and improve the diagnosis of invasive adenocarcinoma (IA) in pure ground-glass nodules (pGGNs) in contact with pleura, through a comprehensive analysis of the thin-section CT features of subpleural malignant pGGNs.<bold>Methods: </bold>CT findings and clinical information of 115 consecutive patients in our hospital between January 2012 and December 2015 who met the following criteria were retrospectively studied: (a) thin-section CT within 1 month before surgery proved pGGN in contact with pleura, and (b) the pGGN was confirmed as malignancy by surgery. Univariate analysis and a multivariate logistic regression analysis were conducted to identify the independent risk factors of IA and VPI.<bold>Results: </bold>No pleural invasion was observed microscopically in any of the pGGNs. Univariate analysis indicated that tumour shape (p = 0.004), relative density (p = 0.038) and the existence of pleural retraction (p < 0.001) were significantly different between the invasive group and pre- or minimally invasive group. Multivariate logistic regression analysis revealed that pleural retraction (OR, 5.663; p < 0.001), lobulated tumour shape (OR, 4.812; p = 0.016) and tumour relative density greater than 1.60 (OR, 4.449; p = 0.001) were independent risk factors of IA.<bold>Conclusions: </bold>Pulmonary adenocarcinoma manifesting as pGGN generally does not invade the pleura. A comprehensive consideration of tumour shape, relative density and tumour-pleural relationship can independently predict IA.<bold>Key Points: </bold>• This study showed that pGGN-like adenocarcinoma generally does not invade the pleura. • This study suggested that persistent pGGN with pleural retraction, lobulated shape and high relative density (> 1.60) may very likely be invasive adenocarcinoma. • Using "relative density" can reduce confounding of contrast agent and respiratory status in analysis of CT images. [ABSTRACT FROM AUTHOR]