Targeted Metabolomic Biomarkers for Stroke Subtyping

Background And Purpose: Ischemic stroke is a heterogeneous disease with various etiologies. The current subtyping process is complicated, time-consuming, and costly. Metabolite-based biomarkers have the potential to improve classification and deliver optimal treatments. We here aimed to identify novel, targeted metabolomics-based biomarkers to discriminate between large-artery atherosclerosis (LAA) and cardioembolic (CE) stroke.

Methods: We acquired serum samples and clinical data from a hospital-based acute stroke registry (ischemic stroke within 3 days from symptom onset). We included 346 participants (169 LAA, 147 CE, and 30 healthy older adults) and divided them into training and test sets. Targeted metabolomic analysis was performed using quantitative and quality-controlled liquid chromatography with tandem mass spectrometry. A multivariate regression model using metabolomic signatures was created that could independently distinguish between LAA and CE strokes.

Results: The training set (n = 193) identified metabolomic signatures that were different in patients with LAA and CE strokes. Six metabolomic biomarkers, i.e., lysine, serine, threonine, kynurenine, putrescine, and lysophosphatidylcholine acyl C16:0, could discriminate between LAA and CE stroke after adjusting for sex, age, body mass index, stroke severity, and comorbidities. The enhanced diagnostic power of key metabolite combinations for discriminating between LAA and CE stroke was validated using the test set (n = 123).

Conclusions: We observed significant differences in metabolite profiles in LAA and CE strokes. Targeted metabolomics may provide enhanced diagnostic yield for stroke subtypes. The pathophysiological pathways of the identified metabolites should be explored in future studies.