TY - JOUR
T1 - Untargeted 1H-NMR metabolomics in CSF
T2 - Toward a diagnostic biomarker for motor neuron disease
AU - Blasco, Hélène
AU - Nadal-Desbarats, Lydie
AU - Pradat, Pierre François
AU - Gordon, Paul H.
AU - Antar, Catherine
AU - Veyrat-Durebex, Charlotte
AU - Moreau, Caroline
AU - Devos, David
AU - Mavel, Sylvie
AU - Emond, Patrick
AU - Andres, Christian R.
AU - Corcia, Philippe
PY - 2014/4/1
Y1 - 2014/4/1
N2 - Objectives: To develop a CSF metabolomics signature for motor neuron disease (MND) using 1H-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 1H-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 (R2X . 22%, R2Y . 93%, Q2 . 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 1H-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.
AB - Objectives: To develop a CSF metabolomics signature for motor neuron disease (MND) using 1H-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 1H-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 (R2X . 22%, R2Y . 93%, Q2 . 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 1H-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.
UR - http://www.scopus.com/inward/record.url?scp=84898728770&partnerID=8YFLogxK
U2 - 10.1212/WNL.0000000000000274
DO - 10.1212/WNL.0000000000000274
M3 - Article
C2 - 24587475
AN - SCOPUS:84898728770
SN - 0028-3878
VL - 82
SP - 1167
EP - 1174
JO - Neurology
JF - Neurology
IS - 13
ER -