Cancer letters · 2026
Integration of longitudinal MRI and biopsy whole slide images for early prediction of neoadjuvant therapy response and personalized management in breast cancer: A multicenter retrospective and prospective study.
СтатьяМногоцентровое исследованиеНазвание 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.
Zhang Song, Shi Zhihui, Du Siyao, Zhu Teng, Li Pinxiong, Zhao Ruimeng, Zhu Xiaobo, Wu Shiwei, Tan Yuyi, Xu Wenzhuo, Li Bao, Li Zhenhui, Wang Kun, Zhang Lina, Tian Jie, Liu Zhenyu
Аннотация
PURPOSE: Early prediction of treatment response during neoadjuvant therapy (NAT) is crucial for timely therapeutic adjustments in breast cancer. Jointly capturing dynamic tumor evolution from longitudinal imaging and tumor microenvironmental cues from biopsy enables improved predictive performance and fosters the development of a personalized tool to assist clinicians in managing NAT.
METHODS: We proposed a multimodal framework, RePALM, which integrated clinical information, pre-NAT MRI, biopsy pathology, and post-2cycle MRI. It was developed and validated on 2604 patients from four retrospective cohorts and further evaluated in a prospective cohort of 72 patients (ChiCTR.org.cn, ChiCTR2400079698). The stepwise integration strategy enabled assessment of the incremental value of each modality. Model performance was benchmarked against radiologists in a reader study, and its predictions were further integrated with RECIST 1.1 for exploratory response stratification. Transcriptomic analysis was conducted to provide biological interpretability.
RESULTS: RePALM demonstrated consistent performance across retrospective and prospective validation cohorts, with AUROC of 0.8730-0.8970 and AUPRC of 0.7450-0.8140. After two NAT cycles, RePALM identified ∼90% of non-pCR patients, enabling early consideration of surgical intervention or trial enrollment. We further demonstrated that stepwise incorporation of multimodal information markedly enhances predictive accuracy. The model also improved assessment performance for junior radiologists. RePALM-derived scores were independently prognostic for event-free survival, while transcriptomic analysis revealed enrichment of treatment-sensitivity pathways, underscoring biological plausibility.
CONCLUSION: RePALM integrates longitudinal imaging and biopsy pathology to predict early NAT response, with demonstrated associations with survival and underlying biological processes, supporting its potential utility in individualized management of breast cancer.
METHODS: We proposed a multimodal framework, RePALM, which integrated clinical information, pre-NAT MRI, biopsy pathology, and post-2cycle MRI. It was developed and validated on 2604 patients from four retrospective cohorts and further evaluated in a prospective cohort of 72 patients (ChiCTR.org.cn, ChiCTR2400079698). The stepwise integration strategy enabled assessment of the incremental value of each modality. Model performance was benchmarked against radiologists in a reader study, and its predictions were further integrated with RECIST 1.1 for exploratory response stratification. Transcriptomic analysis was conducted to provide biological interpretability.
RESULTS: RePALM demonstrated consistent performance across retrospective and prospective validation cohorts, with AUROC of 0.8730-0.8970 and AUPRC of 0.7450-0.8140. After two NAT cycles, RePALM identified ∼90% of non-pCR patients, enabling early consideration of surgical intervention or trial enrollment. We further demonstrated that stepwise incorporation of multimodal information markedly enhances predictive accuracy. The model also improved assessment performance for junior radiologists. RePALM-derived scores were independently prognostic for event-free survival, while transcriptomic analysis revealed enrichment of treatment-sensitivity pathways, underscoring biological plausibility.
CONCLUSION: RePALM integrates longitudinal imaging and biopsy pathology to predict early NAT response, with demonstrated associations with survival and underlying biological processes, supporting its potential utility in individualized management of breast cancer.