The causal relationship between sarcopenia-related traits and ischemic stroke: Insights from univariable and multivariable Mendelian randomization analyses

Abstract

Aims: The causal relationship between sarcopenia-related traits and ischemic stroke (IS) remains poorly understood. This study aimed to explore the causal impact of sarcopenia-related traits on IS and to identify key mediators of this association.

Methods: We conducted univariable, multivariable two-sample, and two-step Mendelian randomization (MR) analyses using genome-wide association study (GWAS) data. This included data for appendicular lean mass (ALM), hand grip strength (HGS), and usual walking pace (UWP) from the UK Biobank, and IS data from the MEGASTROKE consortium. Additionally, 21 candidate mediators were analyzed based on their respective GWAS data sets.

Results: Each 1-SD increase in genetically proxied ALM was associated with a 7.5% reduction in the risk of IS (95% CI: 0.879-0.974), and this correlation remained after controlling for levels of physical activity and adiposity-related indices. Two-step MR identified that six mediators partially mediated the protective effect of higher ALM on IS, with the most significant being coronary heart disease (CHD, mediating proportion: 39.94%), followed by systolic blood pressure (36.51%), hypertension (23.87%), diastolic blood pressure (15.39%), type-2 diabetes mellitus (T2DM, 12.71%), and low-density lipoprotein cholesterol (7.97%).

Conclusion: Our study revealed a causal protective effect of higher ALM on IS, independent of physical activity and adiposity-related indices. Moreover, we found that higher ALM could reduce susceptibility to IS partially by lowering the risk of vascular risk factors, including CHD, hypertension, T2DM, and hyperlipidemia. In brief, we elucidated another modifiable factor for IS and implied that maintaining sufficient muscle mass may reduce the risk of such disease.

Keywords: Mendelian randomization; appendicular lean mass; ischemic stroke; mediators; sarcopenia.

FIGURE 1

Overview of the MR study. This MR study comprised two analytical phases. In phase 1, UVMR estimated the causal associations between sarcopenia‐related traits and IS, suggesting that higher ALM may causally reduce the risk of IS. In phase 2, two‐step MR was performed to screen for 21 candidate mediators potentially underlying the pathways between ALM and IS, and the mediation effects and proportions for qualified mediators were quantified. ALM, appendicular lean mass; MR, Mendelian randomization; MVMR, multivariable Mendelian randomization; MV‐IVW, multivariable inverse variance weighted model; NAFLD, nonalcoholic fatty liver disease; SNPs, single nucleotide polymorphisms; UVMR, univariable Mendelian randomization (figure was created with BioRender.com).

FIGURE 2

Forest plots of the association between sarcopenia‐related traits and IS. (A) Forest plot presenting the causal estimates between sarcopenia‐related traits and IS as determined by UVMR. (B) Forest plot showing the causal estimates between ALM and IS, before and after adjustments for BMI, WBFM, BF%, WHR, levels of PA, and BMR using MVMR. ALM, appendicular lean mass; BMI, body mass index; BF%, body fat percentage; BMR, basal metabolic rate; IS, ischemic stroke; MVMR, multivariable Mendelian randomization; OR, odds ratio; PA, physical activity; SNPs, single nucleotide polymorphisms; UVMR, univariable Mendelian randomization; WBFM, whole‐body fat mass; WHR, waist‐to‐hip ratio.

FIGURE 3

Screening process for mediators in the causal correlation of ALM with IS. Five analytical criteria were performed to select mediators in the causal pathway from higher ALM to IS: (i) causal association of ALM with mediators; (ii) mediator's causal effect on the outcome via UVMR; (iii) direct causal effect of mediators on IS independent of ALM from MVMR; (iv) absence of reverse causation with ALM; and (v) alignment of total and mediating effects of ALM on IS. Eventually, six mediators meeting the above criteria were included in further analyses to quantify their individual mediating contributions. ALM, appendicular lean mass; IS, ischemic stroke; MVMR, multivariable Mendelian randomization (figure was created with BioRender.com).

FIGURE 4

Overall MR estimates for the causal impacts. (A) Causal effects of ALM on potential mediators determined by UVMR; (B) causal effects of remaining candidate mediators on IS determined by UVMR; (C) independent causal effects of remaining candidate mediators on IS after controlling for ALM by MVMR; (D) causal effects of potential mediators on ALM determined by UVMR. Different colors in the boxes indicate the magnitude of the p‐value. AF, atrial fibrillation; ALM, appendicular lean mass; CHD, coronary heart disease; CKD, chronic kidney disease; CRP, C‐reactive protein; DBP, diastolic blood pressure; FG, fasting glucose; FI, fasting insulin; HDL‐C, high‐density lipoprotein cholesterol; IS, ischemic stroke; IVW, inverse‐variance weighted; LDL‐C, low‐density lipoprotein cholesterol; ML, maximum likelihood; MR‐PRESSO^c, outlier‐corrected MR‐PRESSO; MR‐PRESSO^r, raw MR‐PRESSO; NAFLD, nonalcoholic fatty liver disease; MVMR, multivariable Mendelian randomization; SBP, systolic blood pressure; T2DM, type‐2 diabetes mellitus; TC, total cholesterol; TG, triglyceride; THG, two‐hour postprandial glucose; UVMR, univariable Mendelian randomization; WM, weighted median.

FIGURE 5

Two‐step MR estimates for the causal influence of ALM on IS via each mediator. (A) Two‐step MR analysis framework; (B) MR estimates for the causal effects of higher ALM on mediators (left) and the causal effects of mediators on IS with adjustment for ALM (right); (C) MR estimates for the mediating effect of each mediator (left) and the proportion mediated by each mediator (right). “Total effect (β0)” indicates the effect of higher ALM on IS, “direct effect A (β1)” indicates the effect of higher ALM on each mediator, “direct effect B (β2)” indicates the effect of each mediator on IS with adjustment for ALM, and “mediation effect” indicates the effect of higher ALM on IS via each mediator. Total effect (β0) and direct effect A (β1) were derived by the IVW method in UVMR (Table S9); direct effect B (β2) was from the IVW method in MVMR (Table S13). Standard errors for both mediating effects (β1*β2) and proportions (β1*β2/β0) were determined using the delta method. ALM, appendicular lean mass; CHD, coronary heart disease; CI, confidence interval; DBP, diastolic blood pressure; IS, ischemic stroke; IVs, instrumental variables; IVW, inverse variance weighted; LDL‐C, low‐density lipoprotein cholesterol; MR, Mendelian randomization; MVMR, multivariable Mendelian randomization; SBP, systolic blood pressure; T2DM, type‐2 diabetes mellitus; UVMR, univariable Mendelian randomization.