The clinical value of the logistic regression model with a combination of three-dimensional CT quantitative and qualitative parameters in predicting the invasiveness of pure ground glass nodules / 中华放射学杂志

ABSTRACT

Objective:To explore the value of the logistic regression model with three-dimensional CT quantitative parameters in combination with qualitative parameters in predicting the invasiveness of pure ground glass nodules (pGGN).Methods:The clinical data and image features of 191 patients (196 lesions) with pGGN on CT confirmed as lung adenocarcinoma by pathology from April 2019 to December 2019 in Anhui Chest Hospital were retrospectively analyzed. Totally, 196 lesions were divided into atypical adenomatous hyperplasia (AAH)+adenocarcinoma in situ (AIS)+minimally invasive adenocarcinoma (MIA) group ( n=128) and invasive adenocarcinoma (IAC) group ( n=68). CT quantitative parameters included the maximum diameter, volume, average CT value and quality of pGGN, and the qualitative parameters included the intrinsic vascular changes, abnormal air-bronchogram, lobulated signs, bubble-like sign, pleura-traction sign, and fuzzy tumor-lung interface sign. The Mann-Whitney U test was used to compare the quantitative parameters and the Pearson χ 2 test was used to compare the qualitative parameters between two groups. The independent predictors of IAC and predictive probability value were screened by univariate analysis in combination with multivariate logistic regression analysis. The ROC curves were drawn to compare the diagnostic probability of logistic regression prediction probability and each single parameter. Results:There were statistically significant differences in maximum diameter [0.92(0.77,1.14) cm vs. 1.41(1.12, 1.93) cm, Z=-7.366, P<0.001], volume[0.31(0.20, 0.53) cm 3 vs. 0.88(0.41, 2.00) cm 3, Z=-6.254, P<0.001], average CT value[-571.5(-637.2, -477.0) HU vs. -418.1(-532.4, -338.5) HU, Z=-5.882, P<0.001], quality[0.14(0.09, 0.25) g vs. 0.42(0.26, 1.21) g, Z=-7.438, P<0.001], intrinsic vascular changes(23 vs. 49, χ2=55.906, P<0.001), abnormal air-bronchogram(13 vs. 30, χ2=29.908, P<0.001) and pleura-traction sign(39 vs. 32, χ2=5.291, P=0.021) between the two groups. The logistic regression analysis showed that the maximum diameter, average CT value, intrinsic vascular changes and abnormal air-bronchogram were the independent risk factors of IAC, and the odds ratio value(95%CI) were 10.624(1.275-88.522), 1.004(1.000-1.008), 3.424(1.458-8.043) and 2.993(1.114-8.043), respectively. The ROC curve demonstrated that the area under the curve, sensitivity and specificity of the logistic regression model were 0.899, 0.912, and 0.711 respectively, which were better than separate analysis results from each single parameter. Conclusion:The logistic regression model with a combination of three-dimensional CT quantitative and qualitative parameters can predict the invasiveness of pGGN better.