Plasma proteomics identify biomarkers predicting Parkinson's disease up to 7 years before symptom onset

Parkinson's disease (PD) is a prevalent neurodegenerative disorder characterized by progressive motor and non-motor symptoms resulting from α-synuclein aggregation in dopaminergic cells. The lack of objective biomarkers for early disease stages hinders the development of effective neuroprotective strategies. Current methods like α-synuclein seed amplification assays (SAA) rely on invasive cerebrospinal fluid (CSF) sampling. Peripheral fluid biomarkers offer a less invasive alternative for repeated monitoring, crucial for population-based screenings in prevention trials. While neurofilament light chain (NFL) shows promise, it lacks disease specificity. Growing evidence suggests PD pathology in the peripheral system, with inflammatory events potentially preceding α-synuclein aggregation. Inflammatory blood markers are associated with motor deterioration and cognitive decline in PD and are even detectable in individuals with isolated REM sleep behavior disorder (iRBD), a strong predictor of PD. This study aims to identify a panel of blood biomarkers for early PD using mass spectrometry-based proteomics, validated across independent cohorts, including longitudinal iRBD subjects.

The literature extensively documents the challenges in early diagnosis and treatment of Parkinson's disease due to the lack of reliable biomarkers. Existing biomarkers, such as neurofilament light chain (NFL), while showing promise in tracking disease progression, lack the specificity needed for early detection. Cerebrospinal fluid (CSF) analysis, while sensitive, is invasive. The focus has shifted towards peripheral biomarkers due to their accessibility and ease of repeated measurement. Studies have hinted at the involvement of inflammation in the early stages of Parkinson's disease, and certain inflammatory markers in blood have shown association with disease progression. The use of mass spectrometry-based proteomics in this area is gaining traction as it allows for unbiased identification of differentially expressed proteins, providing valuable insights into disease mechanisms. This review of relevant literature highlighted the need for a less invasive, more specific biomarker panel for early detection of Parkinson's Disease.

This study employed a three-phase approach. Phase 0 (discovery) involved untargeted mass spectrometry analysis of plasma samples from 10 de novo PD patients and 10 healthy controls (HC) after depleting major blood proteins. This identified 47 differentially expressed proteins, suggesting early inflammatory involvement. Phase I (validation) used a targeted multiplexed mass spectrometry assay on an independent cohort of 99 de novo PD patients, 36 HC, 41 patients with other neurological disorders (OND), and 18 iRBD patients. This validated 23 differentially expressed proteins between PD and HC. Key pathways involved included acute phase response signaling, coagulation system, complement system, and ER stress. Multivariate analysis (PCA and OPLS-DA) showed distinct clustering of PD and HC groups. Phase II (clinical evaluation) involved a refined assay focusing on the most reliably measured proteins (n=32) applied to 146 longitudinal samples from 54 iRBD individuals with up to 7 years of follow-up. Machine learning (OPLS-DA and SVM) models were used for classification and prediction. Correlations between protein expression and clinical data (UPDRS, MMSE) were also assessed using linear mixed-effects models. Plasma and serum samples were collected under fasting conditions, processed, and analyzed using mass spectrometry. The Synuclein Seed Amplification Assay (SAA) was used on CSF samples to confirm α-synuclein pathology in a subset of participants.

The study identified a panel of eight blood proteins (Granulin precursor, Mannan-binding-lectin-serine-peptidase-2, Endoplasmatic-reticulum-chaperone-BiP, Prostaglaindin-H2-D-isomaerase, Intercellular-adhesion-molecule-1, Complement C3, Dickkopf-WNT-signalling pathway-inhibitor-3, and Plasma-protease-C1-inhibitor) that accurately differentiated between PD and HC (100% specificity) and predicted 79% of iRBD subjects up to 7 years before motor PD onset using machine learning models. Many of these biomarkers correlated with symptom severity. The discovery phase, using untargeted mass spectrometry, initially identified a larger number of proteins, but the targeted proteomic analysis focused on a smaller, more reliable set for validation and clinical translation. The OPLS-DA model, utilizing all 32 detected proteins, accurately classified 70% of iRBD samples as PD-like. The SVM model, which incorporated a subset of eight proteins, achieved a 79% classification rate in the iRBD group. The earliest correct classification of iRBD samples was up to 7.3 years before phenoconversion to PD/DLB. Several pathways showed enrichment with these proteins, including those related to inflammation, coagulation, complement activation, and unfolded protein response. Negative correlations were observed between Granulin, DKK3, and clinical measures of severity, indicating a potential loss of neuroprotection.

The study successfully identified a blood-based biomarker panel for predicting Parkinson's disease years before symptom onset, addressing the critical need for early detection and intervention. The high specificity and predictive power of the eight-protein panel, validated in an independent longitudinal iRBD cohort, is a significant advancement. The findings underscore the importance of inflammation in the early pathophysiology of PD and highlight potential therapeutic targets. The correlation of several markers with clinical symptom severity further strengthens the clinical relevance of the biomarker panel. The methodology used, with its multi-phased approach and use of machine learning, enhances the robustness and translational potential of the findings. The limitations concerning the lack of perfect correlation between peripheral blood proteome and CSF and potential issues with phenoconversion prediction are discussed. Future research should focus on replicating these results in larger, independent cohorts and exploring potential causal relationships between the biomarkers and PD pathophysiology.

This study provides a novel and robust blood-based biomarker panel for early prediction of Parkinson's disease. The panel's ability to identify individuals at risk years before symptom onset has significant implications for clinical trials focused on disease prevention. Future research should focus on validating the panel in diverse populations and exploring its utility in guiding treatment decisions. Further investigation is needed into the specific roles of these proteins in the disease mechanism and their suitability as outcome measures in clinical trials.

While the study demonstrates strong predictive power, some limitations exist. The sample size, although substantial, could be further expanded for increased statistical power. The study focused primarily on the similarities between PD and iRBD, and further investigation is needed to determine the generalizability of these findings across other at-risk populations. Not all iRBD subjects phenoconverted to PD/DLB, highlighting the need for additional markers to accurately predict conversion risk. The limited correlation between blood and CSF protein expression warrants further investigation into the underlying mechanisms.