血浆蛋白与静脉血栓栓塞症:前瞻性队列与孟德尔随机化分析
在11803名基线无静脉血栓栓塞症(VTE)参与者中开展队列分析,中位随访6.6年诊断353例新发VTE,鉴定出21种与VTE风险相关的蛋白;机器学习模型显示体重指数与血管性血友病因子(vWF)对模型预测贡献相同且最大;孟德尔随机化分析发现遗传预测的vWF、SERPINE1、EPHB4、TYRO3、TNFRSF11A和BOC水平与VTE风险存在因果关联。
要点
- 队列分析鉴定出21种蛋白与VTE风险相关(错误发现率校正后)。
- 机器学习模型中体重指数与vWF对整体模型预测的贡献相同且最大。
- 孟德尔随机化发现vWF、SERPINE1、EPHB4、TYRO3、TNFRSF11A和BOC与VTE风险有因果关联。
核心结论
- 队列与孟德尔随机化分析为血浆蛋白与VTE风险的关联及因果性提供了证据,可为疾病预防和药物开发提供依据。
研究设计与样本
- 队列分析纳入11803名基线无VTE的参与者,采用Cox回归估计257种蛋白与VTE风险的关联。
- 构建机器学习模型比较所鉴定蛋白与传统风险因素的重要性。
- 遗传关联数据来自全基因组荟萃分析(26066例病例和624053例对照)和FinnGen(14454例病例和294700例对照)。
主要发现
- 队列分析在6.6年随访期间诊断353例新发VTE,经错误发现率校正后鉴定出21种蛋白与VTE风险相关。
- 机器学习模型显示体重指数和血管性血友病因子(vWF)对整体模型预测的贡献相同且最大。
- 孟德尔随机化分析发现遗传预测的vWF、SERPINE1(纤溶酶原激活物抑制剂1,即PAI-1)、EPHB4(ephrin B型受体4)、TYRO3(酪氨酸蛋白激酶受体TYRO3)、TNFRSF11A(肿瘤坏死因子受体超家族成员11A)和BOC(CDO的兄弟蛋白)水平与VTE风险存在因果关联。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
We conducted cohort and Mendelian randomisation (MR) analyses to examine the associations of circulating proteins with risk of venous thromboembolism (VTE) to provide evidence basis for disease prevention and drug development. Cohort analysis was performed in 11 803 participants without baseline VTE. Cox regression was used to estimate the associations between 257 proteins and VTE risk. A machine-learning model was constructed to compare the importance of identified proteins and traditional risk factors. Genetic association data on VTE were obtained from a genome-wide meta-analysis (26 066 cases and 624 053 controls) and FinnGen (14 454 cases and 294 700 controls). The cohort analysis, including 353 incident VTE cases diagnosed during a 6.6-year follow-up, identified 21 proteins associated with VTE risk after false discovery rate correction. The machine-learning model indicated that body mass index and von Willebrand factor (vWF) made the same as well as most of the contributions to the overall model prediction. MR analysis found that genetically predicted levels of vWF, SERPINE1 (plasminogen activator inhibitor 1, known as PAI-1), EPHB4 (ephrin type-B receptor 4), TYRO3 (tyrosine-protein kinase receptor TYRO3), TNFRSF11A (tumour necrosis factor receptor superfamily member 11A), and BOC (brother of CDO) were causally associated with VTE risk.
引用
- Yuan S, Titova OE, Zhang K, et al. Plasma protein and venous thromboembolism: prospective cohort and mendelian randomisation analyses. Br J Haematol. 2023. PMID: 36734038. doi:10.1111/bjh.18679
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。