Zhong nan da xue xue bao. Yi xue ban = Journal of Central South University. Medical sciences · 2025
[A multimodal disease-specific cohort for melanoma research: Construction, governance, and preliminary report].
Аннотация на английскомСтатьяНазвание 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.
Xie Dongcheng, Yuan Yongxiang, Zhao Shuang, Zheng Wei, Lu Hui, Huang Kai
Аннотация
OBJECTIVES: Given the high incidence, poor prognosis, and scarcity of high-quality multimodal research data for acral melanoma in the Chinese population, this study aims to build a representative multimodal disease-specific cohort to provide a high-quality data foundation for characterizing disease features and exploring mechanisms of prognosis.
METHODS: This single-center prospective cohort was constructed by consecutively enrolling melanoma patients with pathological confirmation at Xiangya Hospital of Central South University between April 2016 and September 2024, while excluding minors and pregnant or lactating women. A standardized dataset comprising 16 internationally aligned core modules was built following global ontologies and clinical data standards. Data governance integrated a hybrid automated extract-transform-load (ETL) and artificial intelligence (AI)-assisted architecture. Pathology narratives were processed using Transformer-based natural language processing (NLP) models to extract textual and morphologic features, while raw imaging data were handled with deep learning pipelines for pixel-level modeling. Data quality was ensured through dual clinical-engineering validation, and outcomes were adjudicated using RECIST 1.1 and CTCAE v5.0 criteria.
RESULTS: The cohort included 1 036 melanoma patients, consisting of 514 males (49.614%) and 522 females (50.386%), with a mean age of (60.2±13.7) years. Among 898 patients with a known clinical subtype, 606 (67.483%) were acral melanoma, the dominant subtype. In 678 patients with complete Breslow thickness data, the median Breslow thickness was 3.25 mm. Of 612 patients with recorded Clark levels, 277 (45.261%) were Clark level IV, the most frequent category. Among 880 patients with available clinical TNM (cTNM) staging, 329 (37.386%) were stage II, representing the largest proportion; Driver mutation profiling showed BRAFV600E positivity of 21.786% (61/280) and c-KIT mutation rate of 9.459% (7/74). Regarding treatment patterns, 217 patients (20.946%) received immunotherapy and 59 (5.695%) received targeted therapy. In 372 patients with complete systematic follow-up, there were 129 deaths (34.677%) and 86 recurrence/metastasis events (23.118%). Acral subtype, Clark level IV, and stage III-IV cTNM patients represented the major at-risk groups for death and recurrence/metastasis. The cohort integrated 8 536 digital pathology whole-slide images (WSI), 5 084 raw imaging data complying with Digital Imaging and Communications in Medicine (DICOM) imaging encounters, over 350 000 structured clinical-molecular data points, and a total data volume of 20.5 TB.
CONCLUSIONS: This melanoma-specific multimodal cohort captures population-specific clinical, imaging, digital pathology, and molecular features for the Chinese population. It establishes a scalable, governed real-world research resource to enable comparative-effectiveness studies, prognostic biomarker discovery, and future development of AI-assisted diagnostic and prognostic models.
METHODS: This single-center prospective cohort was constructed by consecutively enrolling melanoma patients with pathological confirmation at Xiangya Hospital of Central South University between April 2016 and September 2024, while excluding minors and pregnant or lactating women. A standardized dataset comprising 16 internationally aligned core modules was built following global ontologies and clinical data standards. Data governance integrated a hybrid automated extract-transform-load (ETL) and artificial intelligence (AI)-assisted architecture. Pathology narratives were processed using Transformer-based natural language processing (NLP) models to extract textual and morphologic features, while raw imaging data were handled with deep learning pipelines for pixel-level modeling. Data quality was ensured through dual clinical-engineering validation, and outcomes were adjudicated using RECIST 1.1 and CTCAE v5.0 criteria.
RESULTS: The cohort included 1 036 melanoma patients, consisting of 514 males (49.614%) and 522 females (50.386%), with a mean age of (60.2±13.7) years. Among 898 patients with a known clinical subtype, 606 (67.483%) were acral melanoma, the dominant subtype. In 678 patients with complete Breslow thickness data, the median Breslow thickness was 3.25 mm. Of 612 patients with recorded Clark levels, 277 (45.261%) were Clark level IV, the most frequent category. Among 880 patients with available clinical TNM (cTNM) staging, 329 (37.386%) were stage II, representing the largest proportion; Driver mutation profiling showed BRAFV600E positivity of 21.786% (61/280) and c-KIT mutation rate of 9.459% (7/74). Regarding treatment patterns, 217 patients (20.946%) received immunotherapy and 59 (5.695%) received targeted therapy. In 372 patients with complete systematic follow-up, there were 129 deaths (34.677%) and 86 recurrence/metastasis events (23.118%). Acral subtype, Clark level IV, and stage III-IV cTNM patients represented the major at-risk groups for death and recurrence/metastasis. The cohort integrated 8 536 digital pathology whole-slide images (WSI), 5 084 raw imaging data complying with Digital Imaging and Communications in Medicine (DICOM) imaging encounters, over 350 000 structured clinical-molecular data points, and a total data volume of 20.5 TB.
CONCLUSIONS: This melanoma-specific multimodal cohort captures population-specific clinical, imaging, digital pathology, and molecular features for the Chinese population. It establishes a scalable, governed real-world research resource to enable comparative-effectiveness studies, prognostic biomarker discovery, and future development of AI-assisted diagnostic and prognostic models.