Biomarkers in amyotrophic lateral sclerosis: combining metabolomic and clinical parameters to define disease progression.

Background and purposeThe objectives of this study were to define the metabolomic profile of cerebrospinal fluid in amyotrophic lateral sclerosis (ALS) patients, to model outcome through combined clinical and metabolomic parameters and independently to validate predictive models. MethodsIn all, 74 consecutive newly diagnosed patients were enrolled into training (Tr, n = 49) and test (Te, n = 25) cohorts. Investigators recorded clinical data and the metabalomic profile of cerebrospinal fluid at baseline was analyzed with H-1 nuclear magnetic resonance spectroscopy. Markers of disease progression, collected in 1-year prospective follow-up, included change in ALS Functional Rating Scale (var_ALSFRS), change in weight (var_weight) and survival time. Stepwise multiple regression selected from metabolomic and clinical parameters to model rate of progression in the Tr cohort. Best fit models were validated independently in the Te cohort. ResultsThe best-fit statistical models, using both metabolomic and clinical covariates, predicted outcome with 70.8% (var_weight), 72% (var_ALSFRS) and 76% (survival) accuracy in the Te cohort. Models that used metabolomics or clinical data alone predicted outcome less well. Highlighted metabolites are involved in pathophysiological pathways previously described in ALS. ConclusionCerebrospinal fluid metabolomics can aid in predicting the clinical course of ALS and tap into pathophysiological processes. The precision of predictive models, independently reproduced in this study, is enhanced through inclusion of both metabolomic and clinical parameters. The findings bring the field closer to a clinically meaningful disease marker.