By Lucy Piper, medwireNews Reporter
medwireNews: Researchers have used a machine learning approach to create a panel of eight blood biomarkers that can identify people at risk for developing Parkinson’s disease (PD) up to 7 years before the first signs or symptoms.
“Our identified panel of blood biomarkers significantly advances NSD [neuronal synuclein disease] research by providing potential screening and detection markers for use in the earliest stages of NSD”, say the investigators, adding that this has potential for identifying and stratifying individuals for trials aimed at preventing motor PD.
The study involved an initial discovery phase involving 10 drug-naïve patients with PD and 10 matched healthy individuals randomly selected from the de novo PD cohort, which identified 47 differentially expressed proteins.
These proteins, along with known PD-related proteins and others identified in previous studies, then formed the basis of a mass spectrometric targeted proteomic assay, which was developed and refined using an independent cohort, comprising 99 individuals with de novo PD, 36 healthy controls, 41 people with other neurological disorders and 18 with pre-motor isolated rapid eye movement sleep behaviour disorder (iRBD).
In all, 32 proteins were reliably detected in plasma, of which 23 were significantly and differentially expressed between the PD patients and healthy controls, and six were differentially expressed between iRBD patients and those with other neurological disorders and healthy controls. When the impact of these proteins was evaluated using pathway analysis, the researchers found three major pathways with the highest levels of protein enrichment.
These pathways were: the expression of serine protease inhibitors or serpine and complement and coagulation components; endoplasmic reticulum stress/heat shock-related proteins; and the expression of vascular-cell adhesion molecule (VCAM)-1, selectin E (SELE), and protein phosphatase 3 catalytic subunit beta (PPP3CB).
“These are all pathways involved in inflammatory responses”, note Jenny Hällqvist (University College London, UK) and colleagues. “We also identified pathways related to the unfolded protein response and neuroinflammation, although with lower enrichment scores.”
The researchers then applied machine learning to determine the most discriminating proteins and identified a panel of eight protein predictors, which together classified and separated 30 of the de novo PD patients from 11 of the healthy controls with 100% accuracy. These proteins were: granulin precursor (GRN), mannan-binding lectin-serine peptidase-2 (MASP2), endoplasmic reticulum chaperone BiP (HSPA5), prostaglandin-H2 D-isomerase (PTGDS), intercellular adhesion molecule (ICAM)-1, complement C3, dickkopf-WNT-signalling pathway-inhibitor (DKK)-3 and plasma protease C1 inhibitor (SERPING1).
The accuracy of the individual biomarkers for predicting PD ranged from 53% to 92%.
The prediction rate for PD using this panel was 79% for independently validated testing of 146 serum samples from 54 high-risk individuals with confirmed iRBD who were followed up for 10 years, during which time 16 phenoconverted to either PD (n=11) or dementia with Lewy bodies (n=5).
Hällqvist et al point out that this rate of prediction is “high” and that “the earliest correct classification was 7.3 years prior to phenoconversion and the latest was 0.9 years prior to diagnosis.”
The team notes in Nature Communications that “[t]he main shortcoming with many previously explored PD biomarkers is weak or no correlation with clinical progression data.”
But among the biomarkers in their assay, GRN, DKK3, PP3CB and SELE all negatively correlated with Hoehn & Yahr stage and Unified Parkinson’s Disease Rating Scale (UPDRS) parts II, III and total score, synonymous with a more severe clinical impairment with lower protein expression in the Wnt-signalling pathways. And complement C3 levels were associated with higher scores for symptom severity on UPDRS part III and total score, as well as with lower scores in cognitive performance on the Mini-Mental State Examination.
The researchers conclude that their biomarker panel provides “a distinct signature of protective and detrimental mechanisms, finally triggering oxidative stress and neuroinflammation, leading to α-synuclein aggregation and [Lewy body] formation.”
News stories are provided by medwireNews, which is an independent medical news service provided by Springer Healthcare Ltd. © 2024 Springer Healthcare Ltd, part of the Springer Nature Group
Nat Commun 2024; 15: 4759