Related Experiment Video
Updated: Aug 16, 2025
MALDI Imaging Mass Spectrometry of Neuropeptides in Parkinson's Disease
Published on: February 14, 2012
Diagnostic metabolomic profiling of Parkinson's disease biospecimens
Peter A LeWitt1, Jia Li2, Kuan-Han Wu2
1Departments of Neurology, Henry Ford Hospital, West Bloomfield, MI, USA; Wayne State University School of Medicine, West Bloomfield, MI, USA.
Background:
Reliable and sensitive biomarkers are needed for enhancing and predicting Parkinson's disease (PD) diagnosis.
Objective:
To investigate comprehensive metabolomic profiling of biochemicals in CSF and serum for determining diagnostic biomarkers of PD.
Methods:
Fifty subjects, symptomatic with PD for ≥5 years, were matched to 50 healthy controls (HCs). We used ultrahigh-performance liquid chromatography linked to tandem mass spectrometry (UHPLC-MS/MS) for measuring relative concentrations of ≤1.5 kDalton biochemicals. A reference library created from authentic standards facilitated chemical identifications. Analytes underwent univariate analysis for PD association, with false discovery rate-adjusted p-value (≤0.05) determinations. Multivariate analysis (for identifying a panel of biochemicals discriminating PD from HCs) used several biostatistical methods, including logistic LASSO regression.
Results:
Comparing PD and HCs, strong differentiation was achieved from CSF but not serum specimens. With univariate analysis, 21 CSF compounds exhibited significant differential concentrations. Logistic LASSO regression led to selection of 23 biochemicals (11 shared with those determined by the univariate analysis). The selected compounds, as a group, distinguished PD from HCs, with Area-Under-the-Receiver-Operating-Characteristic (ROC) curve of 0.897. With optimal cutoff, logistic LASSO achieved 100% sensitivity and 96% specificity (and positive and negative predictive values of 96% and 100%). Ten-fold cross-validation gave 84% sensitivity and 82% specificity (and 82% positive and 84% negative predictive values). From the logistic LASSO-chosen regression model, 2 polyamine metabolites (N-acetylcadaverine and N-acetylputrescine) were chosen and had the highest fold-changes in comparing PD to HCs. Another chosen biochemical, acisoga (N-(3-acetamidopropyl)pyrrolidine-2-one), also is a polyamine metabolism derivative.
Conclusions:
UHPLC-MS/MS assays provided a metabolomic signature highly predictive of PD. These findings provide further evidence for involvement of polyamine pathways in the neurodegeneration of PD.
