Metabolic markers of MRI-confirmed lacunar stroke: observational and Mendelian randomization analyses

Abstract

Background and Objectives: Lacunar stroke results from cerebral small vessel disease (SVD) and is a major cause of vascular dementia and cognitive decline. We aimed to evaluate the associations of metabolites with lacunar stroke, neuroimaging markers of SVD, and cognition to help elucidate SVD pathogenesis. Methods: In a clinical cohort of MRI-confirmed lacunar stroke cases and unrelated hospital controls recruited from stroke centers across the UK, we assayed 250 serum metabolites using NMR spectroscopy. We investigated the association of metabolites with MRI-confirmed lacunar stroke, neuroimaging markers, and cognitive impairment. We also conducted a genome-wide association analysis for each metabolite and applied bidirectional Mendelian randomization (MR) to investigate causality. Results: Among 2,408 participants, there were 1,456 lacunar stroke cases (mean age, 62 [SD 12]; 34% female) and 952 controls (mean age, 57 [SD 6]; 38% female). Observational analyses identified 211 metabolites that were significantly associated with lacunar stroke. MR analyses identified significant evidence to support causal associations of two of these metabolites, glycine and the proportion of cholesterol to total lipids within medium LDL (%C/Total in M-LDL), with two distinct lacunar stroke subtypes—isolated lacunar infarcts (ILI, 1 lacunar infarct with Fazekas grade <2) and multiple lacunar infarcts and/or leukoaraiosis (MLI/LA, >1 lacunar infarct and/or Fazekas grade ≥2). Glycine was inversely associated with ILI (OR: 0.704, 95% CI 0.585-0.848), and %C/Total in M-LDL was inversely associated with MLI/LA (OR: 0.541, 95% CI 0.401-0.730). Five metabolites (Total S-HDL, S-HDL concentration, PL in S-HDL, %CE/Total in L-LDL, and lactate) were significantly associated with executive functioning and processing speed in observational analyses. MR analyses also implicated genetically determined levels of ω-3/Total FA with two diffusion tensor imaging markers of SVD. Overall, the most robust and consistent associations across both observational and MR analyses were identified for glycine and %C/Total in M-LDL with lacunar stroke subtypes. Discussion: We identified inverse associations of glycine and the proportion of cholesterol to total lipids within medium LDL with distinct subtypes of MRI-confirmed lacunar stroke. These findings reveal metabolomics signatures associated with lacunar stroke subtypes, which may inform future studies of SVD pathogenesis and biomarker development.