International journal of surgery (London, England) · 2025
Multimodal radiopathomics approach for predictions of prognosis and immunotherapy response in patients with gastric cancer: a multicohort retrospective study.
СтатьяМногоцентровое исследованиеНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Wang Wei, Huang Weicai, Yang Xianqi, Fang Cheng, Wei Hongkun, Li Zhe, Qiu Liang, Zhong Rou, Chen Chuanli, Yuan Qingyu, Zhou Kangneng, Wu Lin, Xue Zhicheng, Zhou Zhiwei, Li Yuanfang, Xu Yikai, Li Guoxin, Li Zhenhui, Yun Jingping, Jiang Yuming
Аннотация
BACKGROUND: Current TNM staging systems insufficiently predict individual prognosis and chemotherapy response in gastric cancer (GC). We aimed to develop and validate a radiopathomics signature (RPS) integrating computed tomography (CT) and pathological features to improve prognostic stratification and guide treatment decisions. A preliminary evaluation of the immunotherapy response was also conducted.
METHODS: This retrospective multicenter analysis included 1133 GC patients across three Chinese institutions. We integrated features extracted from CT images and H&E stain slides using ResNet-50 and HoverNet, encompassing radiomics, pathomics microenvironment, single-cell spatial distribution, and pathomics nucleus data, to develop the RPS. Prognostic performance was evaluated using the area under the time-dependent curve (AUC) and C-index. Chemotherapy and immunotherapy benefits were determined using Kaplan-Meier analysis.
RESULTS: The RPS demonstrated strong performance in predicting 5-year overall survival (OS), with AUCs of 0.928 (95% CI: 0.899-0.956) in the training cohort and 0.857-0.917 in internal and external validation cohorts, showing improved prognostic accuracy compared with single-modality radiomics and pathomics models. The model can identify patients who could benefit from postoperative chemotherapy (HR: 11.751, P < 0.0001). Moreover, the RPS showed a preliminary but significant association with treatment response in the immunotherapy cohort ( n = 64; HR: 3.651, P = 0.009). The nomogram combining RPS and TNM stage yielded good C-indexes of 0.79-0.84 for OS across cohorts.
CONCLUSIONS: The RPS robustly predicts prognosis and chemotherapy benefit in GC patients and provides preliminary insights into immunotherapy response prediction, complementing the TNM staging system and helping to guide patient-specific treatment strategies.
METHODS: This retrospective multicenter analysis included 1133 GC patients across three Chinese institutions. We integrated features extracted from CT images and H&E stain slides using ResNet-50 and HoverNet, encompassing radiomics, pathomics microenvironment, single-cell spatial distribution, and pathomics nucleus data, to develop the RPS. Prognostic performance was evaluated using the area under the time-dependent curve (AUC) and C-index. Chemotherapy and immunotherapy benefits were determined using Kaplan-Meier analysis.
RESULTS: The RPS demonstrated strong performance in predicting 5-year overall survival (OS), with AUCs of 0.928 (95% CI: 0.899-0.956) in the training cohort and 0.857-0.917 in internal and external validation cohorts, showing improved prognostic accuracy compared with single-modality radiomics and pathomics models. The model can identify patients who could benefit from postoperative chemotherapy (HR: 11.751, P < 0.0001). Moreover, the RPS showed a preliminary but significant association with treatment response in the immunotherapy cohort ( n = 64; HR: 3.651, P = 0.009). The nomogram combining RPS and TNM stage yielded good C-indexes of 0.79-0.84 for OS across cohorts.
CONCLUSIONS: The RPS robustly predicts prognosis and chemotherapy benefit in GC patients and provides preliminary insights into immunotherapy response prediction, complementing the TNM staging system and helping to guide patient-specific treatment strategies.