Untargeted 1H-NMR metabolomics in CSF: toward a diagnostic biomarker for motor neuron disease

Abstract

Objectives: To develop a CSF metabolomics signature for motor neuron disease (MND) using (1)H-NMR spectroscopy and to evaluate the predictive value of the profile in a separate cohort.

Methods: We collected CSF from patients with MND and controls and analyzed the samples using (1)H-NMR spectroscopy. We divided the total patient sample in a 4:1 ratio into a training cohort and a test cohort. First, a metabolomics signature was created by statistical modeling in the training cohort, and then the analyses tested the predictive value of the signature in the test cohort. We conducted 10 independent trials for each step. Finally, we identified the compounds that contributed most consistently to the metabolome profile.

Results: Analysis of CSF from 95 patients and 86 controls identified a diagnostic profile for MND (R(2)X > 22%, R(2)Y > 93%, Q(2) > 66%). The best model selected the correct diagnosis with mean probability of 99.31% in the training cohort. The profile discriminated between diagnostic groups with 78.9% sensitivity and 76.5% specificity in the test cohort. Metabolites linked to pathophysiologic pathways in MND (i.e., threonine, histidine, and molecules related to the metabolism of branched amino acids) were among the discriminant compounds.

Conclusion: CSF metabolomics using (1)H-NMR spectroscopy can detect a reproducible metabolic signature for MND with reasonable performance. To our knowledge, this is the first metabolomics study that shows that a validation in separate cohorts is feasible. These data should be considered in future biomarker studies.

Classification of evidence: This study provides Class III evidence that CSF metabolomics accurately distinguishes MNDs from other neurologic diseases.