Diabetes, obesity & metabolism · 2026
Data-Driven Cluster Subtyping and Its Association With Cardiometabolic Disease Progression: Integrating Multi-State Modelling, Metabolomics and Proteomics.
СтатьяНазвание 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.
Huang Yixing, Liu Wenhao, Wu Maoxiong, Gao Qingyuan, Gao Jingwei, Chen Qian, Wang Jingfeng, Chen Yangxin, Zhang Haifeng, Mai Jingting, Yuan Woliang
Аннотация
AIMS: Cardiometabolic diseases (CMDs) often manifest in the form of cardiometabolic multimorbidity. Traditional risk assessment methods that rely on single biomarkers fail to adequately characterise the heterogeneity among patients across various subtypes. This study seeks to utilise routine biomarkers to identify clinical subtypes with unique characteristics, thereby improving risk assessment.
MATERIALS AND METHODS: We included 337 334 UK Biobank participants without type 2 diabetes (T2D), ischaemic heart disease (IHD) or stroke at baseline. Multi-task Deep LASSO selected 12 variables for K-means clustering. Multi-state Cox models assessed transitions to first cardiometabolic disease (FCMD), cardiometabolic multimorbidity (CMM) and death. Analyses of 251 plasma metabolic measures and circulating proteins characterised molecular correlates; metabolite panels were evaluated in held-out testing.
RESULTS: Four risk-informed phenotypes were identified: metabolically healthy (MH), older cardio-renal-metabolic (OCRM), lean hypercholesterolaemia (LHC) and early-onset metabolic syndrome (EO-MetS). During a median follow-up of approximately 13 years, OCRM showed the highest FCMD and FCMD-to-CMM risks. In primary models, OCRM and EO-MetS showed strong T2D associations, whereas LHC showed stronger relative associations with IHD and stroke than with T2D. Age adjustment attenuated several OCRM and LHC associations, particularly with IHD and stroke; T2D associations remained largely unchanged. Phenotypes exhibited differing metabolic profiles, and exploratory decomposition showed statistical overlap with phenotype-outcome associations without establishing causality. Protein enrichment highlighted complement/coagulation and vascular-remodelling processes. Metabolite panels modestly improved discrimination beyond full clinical models, increasing the 10-year CMM area under the receiver operating characteristic curve from 0.834 to 0.848; phenotype membership added little further improvement.
CONCLUSIONS: Routine biomarker-derived subtypes showed distinct cardiometabolic trajectories and multi-omic profiles. These risk-informed phenotypes support exploratory description of cardiometabolic heterogeneity, pending external validation.
MATERIALS AND METHODS: We included 337 334 UK Biobank participants without type 2 diabetes (T2D), ischaemic heart disease (IHD) or stroke at baseline. Multi-task Deep LASSO selected 12 variables for K-means clustering. Multi-state Cox models assessed transitions to first cardiometabolic disease (FCMD), cardiometabolic multimorbidity (CMM) and death. Analyses of 251 plasma metabolic measures and circulating proteins characterised molecular correlates; metabolite panels were evaluated in held-out testing.
RESULTS: Four risk-informed phenotypes were identified: metabolically healthy (MH), older cardio-renal-metabolic (OCRM), lean hypercholesterolaemia (LHC) and early-onset metabolic syndrome (EO-MetS). During a median follow-up of approximately 13 years, OCRM showed the highest FCMD and FCMD-to-CMM risks. In primary models, OCRM and EO-MetS showed strong T2D associations, whereas LHC showed stronger relative associations with IHD and stroke than with T2D. Age adjustment attenuated several OCRM and LHC associations, particularly with IHD and stroke; T2D associations remained largely unchanged. Phenotypes exhibited differing metabolic profiles, and exploratory decomposition showed statistical overlap with phenotype-outcome associations without establishing causality. Protein enrichment highlighted complement/coagulation and vascular-remodelling processes. Metabolite panels modestly improved discrimination beyond full clinical models, increasing the 10-year CMM area under the receiver operating characteristic curve from 0.834 to 0.848; phenotype membership added little further improvement.
CONCLUSIONS: Routine biomarker-derived subtypes showed distinct cardiometabolic trajectories and multi-omic profiles. These risk-informed phenotypes support exploratory description of cardiometabolic heterogeneity, pending external validation.