JHEP reports : innovation in hepatology · 2026
Machine Learning-Guided Personalized Immunochemotherapy Strategies in Intrahepatic Cholangiocarcinoma.
СтатьяНазвание и аннотация: Перевод в очереди
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.
Mei Jun-Hao, Zhang Kai, Zhang Ying, Li Zhen, Ge Nai-Jian, Han Xue, Zhong Sui-Xing, Zhang Xiu-Ping, Wu Ping-Ping, Lv Wei-Fu, Wu Jun, Wu Jian-Xiong, Wang Chao, Yan Hui, Huang Tian, Liu Jie, Jia Kai-Zhi, Liu Yue, Guo Jin-He, Liu Rong, Teng Gao-Jun, Lu Jian
Аннотация
BACKGROUND & AIMS: Immunochemotherapy (IO-chemo) has become standard care for patients with unresectable intrahepatic cholangiocarcinoma (iCCA), but benefit of adding IO varies greatly among individuals. We sought to develop a system to identify patients most likely to benefit from this treatment based on individualized treatment effect (ITE) estimation.
METHODS: This study included patients with unresectable iCCA who underwent either IO-chemo or chemotherapy alone (chemo). Patients were enrolled from three discovery and seven external validation centers. Target trial emulation was employed to obtain unbiased average treatment effect estimation, and a causal machine-learning model was used to estimate heterogeneous treatment effects for IO-chemo. Based on predicted ITE, patients were stratified into high-benefit, no- to moderate-benefit, and negative-benefit groups, with overall survival used as the primary end point.
RESULTS: The discovery cohort included 1485 patients; the external validation cohort included 562 patients. In the high-benefit group, compared with chemo, the hazard ratio (HR) for IO-chemo was 0.39 (95% CI, 0.30-0.52; p < 0.001), with mortality reduced by 24.1%, 31.2%, and 28.5% at 12, 24, and 36 months, respectively. In the no- to moderate-benefit group, IO-chemo did not differ from chemo (HR, 0.91; 95% CI, 0.70-1.18; p = 0.488). In the group with negative predicted ITEs, IO-chemo was associated with shorter OS than chemo (HR, 1.91; 95% CI, 1.47-2.48; p < 0.001). In the external validation cohort, the corresponding HRs were 0.45 (95% CI, 0.30-0.68; p < 0.001), 0.62 (95% CI, 0.42-0.93; p = 0.018), and 1.93 (95% CI, 1.28-2.91; p = 0.002), respectively.
CONCLUSIONS: A causal machine learning-based model can estimate individualized treatment effects of immunochemotherapy in patients with unresectable intrahepatic cholangiocarcinoma, enabling clinically meaningful benefit stratification.
IMPACT AND IMPLICATIONS: The overarching goal of intrahepatic cholangiocarcinoma (iCCA) management is to deploy biomarkers that identify patients most likely to benefit from immunochemotherapy and thus enable personalized therapy, but little research has examined how clinicians choose between chemotherapy and immunochemotherapy beyond head-to-head efficacy comparisons or efforts to define subgroups suited to a single regimen. Built on a causal machine-learning (ML) framework, the Causal ML-Guided Personalized Immunochemotherapy Strategies in Intrahepatic Cholangiocarcinoma model described here offers a validated, clinically practical tool to quantify the benefit of immunochemotherapy in iCCA. In this study, 17 clinical variables were used to reliably identify patients most likely to benefit from immunochemotherapy and to predict long-term survival. By focusing on individualized treatment effects rather than average treatment effects, the model provides a strong foundation for future trials that seek to operationalize intelligent treatment selection and pushes precision immunochemotherapy for iCCA closer to routine practice.
CLINICAL TRIAL NUMBER: NCT06849193.
METHODS: This study included patients with unresectable iCCA who underwent either IO-chemo or chemotherapy alone (chemo). Patients were enrolled from three discovery and seven external validation centers. Target trial emulation was employed to obtain unbiased average treatment effect estimation, and a causal machine-learning model was used to estimate heterogeneous treatment effects for IO-chemo. Based on predicted ITE, patients were stratified into high-benefit, no- to moderate-benefit, and negative-benefit groups, with overall survival used as the primary end point.
RESULTS: The discovery cohort included 1485 patients; the external validation cohort included 562 patients. In the high-benefit group, compared with chemo, the hazard ratio (HR) for IO-chemo was 0.39 (95% CI, 0.30-0.52; p < 0.001), with mortality reduced by 24.1%, 31.2%, and 28.5% at 12, 24, and 36 months, respectively. In the no- to moderate-benefit group, IO-chemo did not differ from chemo (HR, 0.91; 95% CI, 0.70-1.18; p = 0.488). In the group with negative predicted ITEs, IO-chemo was associated with shorter OS than chemo (HR, 1.91; 95% CI, 1.47-2.48; p < 0.001). In the external validation cohort, the corresponding HRs were 0.45 (95% CI, 0.30-0.68; p < 0.001), 0.62 (95% CI, 0.42-0.93; p = 0.018), and 1.93 (95% CI, 1.28-2.91; p = 0.002), respectively.
CONCLUSIONS: A causal machine learning-based model can estimate individualized treatment effects of immunochemotherapy in patients with unresectable intrahepatic cholangiocarcinoma, enabling clinically meaningful benefit stratification.
IMPACT AND IMPLICATIONS: The overarching goal of intrahepatic cholangiocarcinoma (iCCA) management is to deploy biomarkers that identify patients most likely to benefit from immunochemotherapy and thus enable personalized therapy, but little research has examined how clinicians choose between chemotherapy and immunochemotherapy beyond head-to-head efficacy comparisons or efforts to define subgroups suited to a single regimen. Built on a causal machine-learning (ML) framework, the Causal ML-Guided Personalized Immunochemotherapy Strategies in Intrahepatic Cholangiocarcinoma model described here offers a validated, clinically practical tool to quantify the benefit of immunochemotherapy in iCCA. In this study, 17 clinical variables were used to reliably identify patients most likely to benefit from immunochemotherapy and to predict long-term survival. By focusing on individualized treatment effects rather than average treatment effects, the model provides a strong foundation for future trials that seek to operationalize intelligent treatment selection and pushes precision immunochemotherapy for iCCA closer to routine practice.
CLINICAL TRIAL NUMBER: NCT06849193.