Cao Jie, Chen Nan, Zhou Lingyu, Yi Le, Peng Zhiyu, Qiu Lin, Wu Haokun, Tan Xiyue, Wu Kunhao, Lin Huahang, Huang Zhaokang, Liu Zetao, Guo Chenglin, Xu Xiuyuan, Yi Zhang, Mei Jiandong
Department of Thoracic Surgery, West China Hospital, Sichuan University, Chengdu, China.
Western China Collaborative Innovation Center for Early Diagnosis and Multidisciplinary Therapy of Lung Cancer, Sichuan University, Chengdu, China.
Transl Lung Cancer Res. 2025 May 30;14(5):1516-1530. doi: 10.21037/tlcr-24-890. Epub 2025 May 27.
The preoperative prediction of spread through air spaces (STAS) in patients with early-stage lung adenocarcinoma (LUAD) is crucial for selecting the appropriate surgical approach and improving patient outcomes. Previous research has confirmed that there is a significant correlation between consolidation-to-tumor ratio (CTR) and STAS. This study aimed to develop a Bayesian deep learning (DL) model based on the CTR prior to predict STAS in patients with stage IA LUAD.
This large-scale diagnostic study included patients with solitary primary invasive LUAD who underwent complete resection between November 2017 and October 2023. Enrolled patients were randomly assigned to training, validation, and test cohorts in a 7:2:1 ratio. Using a variational Bayesian inference framework, we developed a DL model based on the CTR prior (STAS-DL). The performance of STAS-DL was compared with another DL model without the CTR prior (STAS-DL) using the receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA), and clinical impact curve (CIC).
A total of 1,374 patients were included, with 961 in the training cohort, 275 in the validation cohort, and 138 in the test cohort. The results showed that CTR in the STAS-positive group was significantly higher than that in the STAS-negative group [0.63 (interquartile range, 0.36, 0.98) . 0.35 (interquartile range, 0.19, 0.60), P<0.001]. Compared to STAS-DL, the area under the ROC curve (AUC) tends to be higher for STAS-DL (0.831 . 0.731, P=0.06) in the validation cohort, and STAS-DL demonstrated a significantly higher AUC (0.858 . 0.637, P=0.008) in the test cohort. Additionally, the calibration curve suggested better calibration for STAS-DL. DCA and CIC also indicated that STAS-DL conferred higher clinical net benefit.
The proposed model based on the CTR prior offers significant advantages in predicting STAS in patients with stage IA LUAD, and incorporating doctors' knowledge as priors can effectively guide the development of DL models.
术前预测早期肺腺癌（LUAD）患者的气腔播散（STAS）对于选择合适的手术方式和改善患者预后至关重要。既往研究证实实变与肿瘤比值（CTR）与STAS之间存在显著相关性。本研究旨在建立基于CTR先验的贝叶斯深度学习（DL）模型，以预测IA期LUAD患者的STAS。
这项大规模诊断研究纳入了2017年11月至2023年10月期间接受完整切除的孤立性原发性浸润性LUAD患者。入组患者按7:2:1的比例随机分配到训练、验证和测试队列。使用变分贝叶斯推理框架，我们建立了基于CTR先验的DL模型（STAS-DL）。使用受试者操作特征（ROC）曲线、校准曲线、决策曲线分析（DCA）和临床影响曲线（CIC），将STAS-DL的性能与另一个无CTR先验的DL模型（STAS-DL）进行比较。
共纳入1374例患者，其中训练队列961例，验证队列275例，测试队列138例。结果显示，STAS阳性组的CTR显著高于STAS阴性组[0.63（四分位数间距，0.36，0.98）对0.35（四分位数间距，0.19，0.60），P<0.001]。在验证队列中，与STAS-DL相比，STAS-DL的ROC曲线下面积（AUC）更高（0.831对0.731，P=0.06），在测试队列中，STAS-DL的AUC显著更高（0.858对0.637，P=0.008）。此外，校准曲线表明STAS-DL的校准更好。DCA和CIC也表明STAS-DL具有更高的临床净效益。
所提出的基于CTR先验的模型在预测IA期LUAD患者的STAS方面具有显著优势，将医生的知识作为先验纳入可以有效指导DL模型的开发。
