GLM7代谢指数各成分与脑卒中因果关系的 孟德尔随机化研究进展

Progress in Mendelian Randomization Studies on the Causal Association of GLM7 Index Components with Stroke

摘要: 脑卒中是全球范围内致残和致死的主要原因之一，糖脂代谢异常是其重要危险因素。传统观察性研究易受混杂偏倚和反向因果干扰，孟德尔随机化(MR)利用遗传变异作为工具变量，可提供更可靠的因果推断。GLM7指数是新近提出的一个综合代谢–衰老指标，整合了年龄、BMI、空腹血糖、胰岛素、甘油三酯、LDL-C和HDL-C。由于年龄不具备遗传基础，本综述聚焦于GLM7中的六个可遗传代谢成分，系统梳理其与脑卒中的MR研究证据。通过检索PubMed、Web of Science、中国知网和万方数据库，纳入符合条件的原始MR研究。结果发现：LDL-C (尤其是小密亚型)与缺血性卒中(特别是大动脉亚型)的因果关系证据最为充分(Grade 1)；BMI和甘油三酯升高增加缺血性卒中风险(Grade 2)；HDL-C对缺血性卒中具有保护作用但对脑出血风险增加(Grade 2/3)；血糖(以HbA1c为代表)和胰岛素抵抗与卒中风险正相关，但空腹血糖和空腹胰岛素的直接证据尚不一致(Grade 3)。各成分对大动脉、小血管和心源性卒中的效应存在显著异质性。本文还探讨了GLM7复合指数进行MR分析的方法学框架，包括遗传工具构建、水平多效性校正及与观察性研究的差异。未来需开展跨种族验证、多变量MR及GLM7全基因组关联研究，以进一步明确该指数的因果效应。

Abstract: Stroke is one of the leading causes of disability and mortality worldwide, and abnormalities in glycolipid metabolism are important risk factors. Traditional observational studies are susceptible to confounding bias and reverse causation, while Mendelian randomization (MR) utilizes genetic variants as instrumental variables, providing more reliable causal inference. The GLM7 index is a recently proposed composite metabolic-aging indicator integrating age, BMI, fasting glucose, insulin, triglycerides, LDL-C, and HDL-C. Since age lacks a genetic basis, this review focuses on the six heritable metabolic components of GLM7 and systematically summarizes MR evidence on their associations with stroke. Eligible original MR studies were identified through searches in PubMed, Web of Science, CNKI, and Wanfang databases. Results showed that LDL-C (especially small dense subtype) has the most robust causal association with ischemic stroke (particularly large artery stroke) (Grade 1). Elevated BMI and triglycerides increase the risk of ischemic stroke (Grade 2). HDL-C shows a protective effect against ischemic stroke but increases the risk of intracerebral hemorrhage (Grade 2/3). Glycemic traits (represented by HbA1c) and insulin resistance are positively associated with stroke risk, while direct evidence for fasting glucose and fasting insulin remains inconsistent (Grade 3). Significant heterogeneity exists in the effects of different components on large artery, small vessel, and cardioembolic stroke. This review also discusses methodological frameworks for conducting MR analysis on the composite GLM7 index, including genetic instrument construction, correction for horizontal pleiotropy, and expected differences from observational studies. Future studies should focus on cross-ethnic validation, multivariable MR, and genome-wide association studies of GLM7 to further clarify the causal effects of this index.