The following is a summary of “High-Throughput CSF Proteomics and Machine Learning to Identify Proteomic Signatures for Parkinson Disease Development and Progression,” published in the August 2023 issue of Neurology by Tsukita et al.

Researchers conducted a retrospective study to find unique CSF proteomic signatures for Parkinson’s disease (PD) and gauge their clinical relevance.

They included PD patients, healthy controls (HCs), with GBA1, LRRK2, and/or SNCA mutations (genetic-prodromal), from Parkinson’s Progression Markers Initiative (PPMI). Subjects categorized as non-genetic-PD, genetic-PD, genetic-prodromal, or HCs using aptamer-based CSF proteomic data. Applying differentially expressed protein (DEP) analysis and least absolute shrinkage and selection operator (LASSO), data from non-genetic-PD and HCs were processed. Non-genetic-PD signatures were quantified as PD proteomic score (PD-ProS), internally and externally validated using 1,556 CSF proteins from LRRK2 Cohort Consortium (LCC). PD-ProS studied in genetic-PD and prodromal, linking to clinical progression.

The results analyzed data from 279 non-genetic-PD patients (mean ± standard deviation, age, 62.0 ± 9.6 years; M, 67.7%), 141 HCs (age, 60.5 ± 11.9 years; M, 64.5%) for deriving PD-ProS. Using 23 DEPs, LASSO assigned weights to 14 DEPs for PD-ProS (area under the curve [AUC] = 0.83 [95% CI, 0.78–0.87]), validated in an independent cohort of 71 non-genetic-PD patients and 35 HCs (AUC = 0.81 [95% CI, 0.73–0.90]). In LCC, 5 of 14 DEPs differentiated 31 non-genetic-PD patients from 34 HCs (AUC = 0.75 [95% CI, 0.63–0.87]) using consistent weights. The PD-ProS distinguished 258 genetic-PD patients from 365 genetic-prodromals. Irrespective of genetic status, the PD-ProS autonomously predicted cognitive and motor decline in PD (dementia, adjusted hazard ratio in the highest quintile [aHR-Q5] = 2.8 [95% CI, 1.6–5.0]; Hoehn and Yahr stage IV, aHR-Q5 = 2.1 [95% CI, 1.1–4.0]).

They concluded Crucial PD-associated CSF proteomic signatures were revealed using proteomics and machine learning.

Source: n.neurology.org/content/early/2023/08/14/WNL.0000000000207725