Serum metabolic biomarkers for synucleinopathy conversion in isolated REM sleep behavior disorder

Abstract

Isolated rapid eye movement (REM) sleep behavior disorder (iRBD) is a prodromal stage of Lewy-type synucleinopathies (LTS), which can present either with an initial predominant parkinsonism (Parkinson's disease (PD)) or dementia (dementia with Lewy bodies (DLB)). To provide insights into the underlying pathogenic mechanisms, the lipoprotein and protein glycosylation profile of 82 iRBD patients, collected before and/or after their conversion to an overt LTS, and 29 matched control serum samples were assessed by nuclear magnetic resonance (NMR) spectroscopy. Data were statistically analyzed to identify altered metabolites and construct predictive models. Univariant analysis detected no differences between iRBD patients with an LTS compared to controls. However, significant differences were found when the analysis distinguished between iRBD patients that manifested initially predominant parkinsonism (pre-PD) or dementia (pre-DLB). Significant differences were also found in the analysis of paired iRBD samples pre- and post-LTS diagnosis. Predictive models were built and distinguished between controls and pre-DLB patients, and between pre-DLB and pre-PD patients. This allowed a prediction of the possible future clinical outcome of iRBD patients. We provide evidence of altered lipoprotein and glycosylation profiles in subgroups of iRBD patients. Our results indicate that metabolic alterations and inflammation are involved in iRBD pathophysiology, and suggest biological differences underlying the progression of LTS in iRBD patients. Our data also indicate that profiling of serum samples by NMR may be a useful tool for identifying short-term high-risk iRBD patients for conversion to parkinsonism or dementia.

Fig. 1. Differences between iRBD patients before and after LTS diagnosis.

Graphs of the a concentration of glycosylated protein B (Glyc-B), b concentration of medium sized LDL particles (medium LDL-P), and c concentration of triglycerides in LDL (LDL-TG) in the same iRBD patients, before (pre-) and after (post-) the diagnosis of LTS. Graphs of the d concentration of Glyc-B, e concentration of LDL-TG, f size of LDL (LDL-Z), g concentration of medium LDL-P, and h concentration of large HDL particles (large HDL-P) in the same IRBD patients, before (pre-) and after (post-) the diagnosis of DLB. Graphs of the i concentration of triglycerides in VLDL (VLDL-TG), j large VLDL particles (large VLDL-P), k medium VLDL particles (medium VLDL-P), and l small VLDL particles (small VLDL-P) in the same IRBD patients, before (pre-) and after (post-) the diagnosis of PD. Mean with interquartile range represented. Wilcoxon matched-pairs signed-rank test, combined with the Benjamini–Hochberg procedure to correct for false discovery rate (FDR). *FDR-adjusted p value < 0.05. ^#FDR-adjusted p value < 0.25.

Fig. 2. Schematic representation of model building workflow.

Scheme illustrating the input/output of each step done for building models to discriminate between groups, and the statistical methods used to perform them. BH Benjamini–Hochberg, ROC receiver operating characteristics.

Fig. 3. Discriminatory models.

Graphs showing the results of the equation (written underneath) that allows distinguishing between controls and pre-DLB patients (a), and pre-DLB and pre-PD patients (b), including the threshold with the highest likelihood ratio (gray dotted line), together with the pertinent ROC curve and area under the curve (AUC) values. The application of model (b) in iRBD-only samples (c) distinguishes between putative pre-DLB (blue dots) and pre-PD (orange dots, darker orange implies 100% sensitivity) patients.