关键词: 腺癌, 体层摄影术,X线计算机, 诊断显像, 磨玻璃样结节

Abstract:

Objective Identify the value of multi-dimensional CT sings in predicting the tumor invasion degree and pathological subtype of lung adenocarcinoma characterized with ground-glass nodule(GGN).Methods Retrospective analysis of CT imaging data of 230 patients with pulmonary ground-glass nodules confirmed by pathology from January 2013 to December 2017 in Zhongshan Hospital affiliated to Dalian University, 12 patients had 2 lung ground-glass nodules. According to the latest classification criteria for lung adenocarcinoma in 2011, there were 110 preinvasive lesions, 32 microinvasive and 100 invasive adenocarcinomas. Analysis of CT signs, including general dimensions, marginal dimensions, internal dimensions, and lumen dimensions (including vascular and bronchial CT signs)Results (1) In general dimension signs, the proportion of round/quasi-round lesions in the preinvasive lesion group, microinvasive adenocarcinoma group and invasive adenocarcinoma group were 59.1% (65/110), 68.8% (22/32), 55.0% (55/100) respectively, the difference was not statistically significant (χ ²=1.904,P>0.05).However, the diameters of GGN lesions in three groups were (10.0±4.7)mm, (11.0±5.1)mm and (17.4±7.0)mm respectively,the difference was significant (U=68.312,P<0.05). ROC curve analysis showed that the preinvasive lesions and invasive adenocarcinoma lesions (including microinvasive and invasive lesions) had a critical value of 1.29 cm, and the diagnostic sensitivity is 70.0%, the specificity is 80.7%, and the AUC (area under ROC curve) is 0.802.(2)In marginal dimension signs, the proportion of lobulated signs were 21.8% (24/110), 56.3% (18/32) and 82.0% (82/100) in preinvasive lesion group, microinvasive adenocarcinoma group and invasive adenocarcinoma group respectively. The difference was statistically significant (χ ²=76.304,P<0.05).(3) In internal dimension signs, the proportion of vacuole signs in three groups were 10.9% (12/110), 18.8% (6/32), 55.0% (55/100) respectively. The difference was statistically significant(χ ²=50.620,P<0.05). (4) In lumen dimension signs, Vascular signs were obvious in type Ⅲ (57.0%, 57/100) and Ⅳ (40.0%, 40/100) invasive adenocarcinoma, type Ⅲ (62.5%, 20/32) microinvasive adenocarcinoma and in type Ⅱ (65.5%, 72/110) preinvasive lesions group. Bronchial signs were obvious in invasive adenocarcinoma type Ⅱ (34.0%, 34/100) and Ⅲ (37.0%, 37/100), microinvasive adenocarcinoma type Ⅲ (40.6%, 13/32) and preinvasive lesions type Ⅳ (50.9%, 56/110). There were significant differences among the three groups(χ ² values are 141.749、134.268,P<0.05).The accuracy of predicting probability of preinvasive lesions and invasive adenocarcinoma through vascular signs and bronchial signs alone or in combination were 86.4%(209/242)、82.2%(199/242)、89.3%(216/242)respectively. Conclusion Comprehensive analysis of multi-dimensional CT signs is helpful for preoperative prediction of GGN-like lung adenocarcinoma infiltration and pathological subtypes, especially the simultaneous identification of vascular signs and bronchial signs and combined diagnosis were more effective than the diagnosis of each sign. Multi-dimensional CT signs provides new information and direction for the preoperative prediction of GGN-like lung adenocarcinoma infiltration.

Key words: Adenocarcinoma, Tomography, X-ray computed, Diagnostic imaging, Ground-glass nodule (GGN)