Metabolomic biomarkers as strong correlates of Parkinson disease progression.

OBJECTIVE:To determine whether a Parkinson disease (PD)-specific biochemical signature might be found in the total body metabolic milieu or in the CSF compartment, especially since this disorder has systemic manifestations beyond the progressive loss of dopaminergic nigrostriatal neurons. METHODS:Our goal was to discover biomarkers of PD progression. Using ultra-high-performance liquid chromatography linked to gas chromatography and tandem mass spectrometry, we measured concentrations of small-molecule (≤1.5 kDa) constituents of plasma and CSF from 49 unmedicated, mildly affected patients with PD (mean age 61.4 years; mean duration of PD 11.4 months). Specimens were collected twice (baseline and final) at intervals up to 24 months. During this time, mean Unified Parkinson's Disease Rating Scale (UPDRS) parts 2 + 3 scores increased 47% (from 28.8 to 42.2). Measured compounds underwent unbiased univariate and multivariate analyses, including fitting data into multiple linear regression with variable selection using least absolute shrinkage and selection operator (LASSO). RESULTS:Of 575 identified plasma and 383 CSF biochemicals, LASSO led to selection of 15 baseline plasma constituents with high positive correlation (0.87, p = 2.2e-16) to baseline-to-final change in UPDRS parts 2 + 3 scores. Three of the compounds had xanthine structures, and 4 were either medium- or long-chain fatty acids. For the 15 LASSO-selected biomarkers, pathway enrichment software found no overrepresentation among metabolic pathways. CSF concentrations of the dopamine metabolite homovanillate showed little change between baseline and final collections and minimal correlation with worsening UPDRS parts 2 + 3 scores (0.29, p = 0.041). CONCLUSIONS:Metabolomic profiling of plasma yielded strong prediction of PD progression and offered biomarkers that may provide new insights into PD pathogenesis.