ISSN 1004-4140
CN 11-3017/P

基于CT影像特征的早期肺腺癌脏层胸膜侵犯风险预测模型构建研究

Risk-Prediction Model Construction of Visceral Pleural Invasion in Early Lung Adenocarcinoma Based on Computed Tomography Imaging Features

  • 摘要: 目的:构建基于CT影像特征早期肺腺癌脏层胸膜侵犯风险预测模型。方法:回顾性纳入2016年1月至2022年12月行手术治疗并确诊为IA期肺腺癌患者170例,根据有无脏层胸膜侵犯分为侵犯组(84例)和未侵犯组(86例);采用单因素和多因素法评价早期胸膜下肺腺癌脏层胸膜侵犯CT影像学特征相关独立危险因素,早期胸膜下肺腺癌脏层胸膜侵犯CT影像学特征预测模型效能分析。结果:单因素分析结果显示,实性成分占比、实性成分最大径、桥征、血管集束征及病灶与胸膜毗邻关系均可能与早期胸膜下肺腺癌患者脏层胸膜侵犯有关。多因素分析结果显示,桥征、血管集束征及病灶与胸膜毗邻关系Ⅱ型/Ⅲ 型均是早期胸膜下肺腺癌脏层胸膜侵犯独立危险因素。利用桥征、血管集束征、病灶与胸膜毗邻关系、Logistic模型预测概率对早期胸膜下肺腺癌脏层胸膜侵犯情况预测,最佳截断值分别为0.50、0.50、0.50和56.25%,约登指数分别为24.00%、23.95%、48.70% 和55.04%。结论:基于桥征、血管集束征及病灶与胸膜毗邻关系构建CT影像学特征模型可用于早期胸膜下肺腺癌脏层胸膜侵犯高危人群识别。

     

    Abstract: Objective: To construct a structural risk prediction model for visceral pleural invasion in early lung adenocarcinoma based on computed tomography (CT) features. Methods: 170 patients with early lung adenocarcinoma treated surgically in our hospital were retrospectively selected between January 2016 and December 2022 and grouped according into invasion (84 cases) and non-invasion (86 cases) groups. Independent risk factors related to CT imaging features of visceral pleural invasion in early lung adenocarcinoma were evaluated using univariate and multivariate factor methods. The effectiveness of the CT imaging feature prediction model for visceral pleural invasion in early subpleural lung adenocarcinoma was analyzed. Results: Univariate analysis showed that the proportion and maximum diameter of solid components, bridging sign, vascular bundle sign, and relationship between the lesion and pleura may be related to visceral pleural invasion in patients with early subpleural lung adenocarcinoma. Multivariate analysis showed that the bridging sign, vascular bundle sign, and type II/III relationship between the lesion and pleura were independent risk factors for visceral pleural invasion in early subpleural lung adenocarcinoma. The best cutoff values for predicting visceral pleural invasion in early subpleural lung adenocarcinoma using the bridge sign, vascular bundle sign, adjacent relationship between the lesion and pleura, and logistic model prediction probability were 0.50, 0.50, 0.50, and 56.25%, respectively. The Jordan index for each of these was 24.00%, 23.95%, 48.70%, and 55.04%, respectively. Conclusion: Based on the bridge sign, vascular cluster sign, and the relationship between tumor and pleural adjacency, the CT imaging feature model can be used to identify high-risk groups for visceral pleural invasion in early lung adenocarcinoma.

     

/

返回文章
返回