Study on the prediction model of pathological poorly differentiated subtype of stage T1 lung adenocarcinoma based on CT signs
10.3969/j.issn.1002-1671.2025.06.009
- VernacularTitle:基于CT征象构建T1期肺腺癌病理低分化亚型预测模型研究
- Author:
Gang XIANG
1
;
Hongfei WANG
;
Guangyan SI
;
Bin YANG
;
Ping DAI
;
Lin REN
;
Dan YU
;
Xiang WANG
Author Information
1. 西南医科大学附属中医医院影像科,四川 泸州 646000
- Publication Type:Journal Article
- Keywords:
lung adenocarcinoma;
computed tomography;
differentiation;
prediction model
- From:
Journal of Practical Radiology
2025;41(6):933-937
- CountryChina
- Language:Chinese
-
Abstract:
Objective To predict the poorly differentiated subtype of stage T1 invasive lung adenocarcinoma based on the morphological and quantitative characteristics of preoperative CT images.Methods The CT images,clinical information and pathological report of 333 cases with stage T1 lung adenocarcinoma in the Affiliated Traditional Chinese Medicine Hospital of Southwest Medical University and the Second Affiliated Hospital of Naval Medical University were retrospectively analyzed.All data were divided into histological subtypes according to the WHO Classification of Chest Tumors(5th Edition).CT signs include the maximum diameter,minimum diameter,location,nodule type,CT value,proportion of solid components,burr sign,lobular sign,vacuole sign,air bronchial sign,halo sign,bronchial truncation sign,vascular cluster sign,pleural depression sign.Clinical factors included gender,age,smoking history,tumor markers,and spread through air spaces.Independent predictors of poorly differentiated subtypes of lung adenocarcinoma were obtained via univariate and multivariate logistic regression analysis,and the clinical model,CT model and combined model were constructed based on the analysis results.Results In univariate logistic regression analysis,gender,smoking history,tumor markers,spread through air spaces,maximum diameter,minimum diameter,location,nodule type,CT value,proportion of solid components,burr sign,vacuole sign and vascular cluster sign were significantly related to poorly differentiated subtypes.Multivariate logistic regression analysis showed that gender,tumor markers,vacuole sign,minimum diameter,and nodule type were independent influencing factors of poorly differentiated subtypes of lung adenocarcinoma.Clinical model,CT model and combined model were constructed based on the analysis results.Area under the curve(AUC)of the clinical model,CT model and combined model were 0.772[95％confidence interval(CI)0.712-0.831],0.776(95％CI 0.714-0.837)and 0.825(95％CI 0.776-0.874)for the poorly differentiated subtypes,respectively.The combined model had a higher AUC,with the better prediction.There was significant difference in predicting lung adenocarcinoma poorly differentiated subtypes between the clinical model and combined model(P=0.019).Conclusion Logistic regression model based on CT signs has good diagnostic value in predicting poorly differentiated lung adenocarcinoma in stage Ⅰ.
