Proteomic Signatures Associated with Genetic and Clinical Subtypes of Parkinson’s Disease (S42.006)

Abstract

<h3>Objective:</h3> Identify proteomic signatures associated with Parkinson9s disease (PD) patients9 genetic status, movement and cognitive tests, neuroimaging, and cerebrospinal fluid (CSF) phosphorylated-tau, total-tau, alpha-synuclein, and amyloid-beta levels. Prioritize protein signatures by integrating genome-wide genotyping and CSF proteomic data. <h3>Background:</h3> PD is a heterogeneous disorder with identifiable clinical-pathological subtypes based on symptom severity and predominance. An accurate molecular profile could reduce PD clinical heterogeneity. Few studies have applied a proteogenomic approach to samples from PD patients to identify proteins associated with clinical, neuroimaging, or neuropathological subtypes. <h3>Design/Methods:</h3> We used data from the Parkinson9s Progression Markers Initiative (PPMI) cohort (N = 1158); 683 individuals passed quality control for pQTL analysis. Differentially expressed proteins in LRRK2+, GBA1+, or sporadic PD (sPD) were curated using F-test, adjusting for sex, age, and principal components 1–4. Weighted correlation network analysis (WGCNA) and least absolute shrinkage and selection operator (LASSO) algorithm were used to explore modules of proteins associated with CSF biomarkers and clinical and neuroimaging data. We used the data from Posavi et al. (2019) as a replication cohort. <h3>Results:</h3> We identified subgroups of proteins associated with mutation carriers in LRRK2 (n=709) and GBA1 (n=505), or sPD (n=97). The WGCNA revealed two modules of proteins associated with higher levels of CSF biomarkers in PD patients. We found 117 proteins distinguishing genetically defined PD from sPD (AUC=0.94) and identified 158 proteins that differentiate PD from controls (AUC=0.86). Our model in the replication cohort (AUC=0.92/0.71) outperformed the model with four robust biomarkers from Posavi et al. (AUC=0.88/0.63) for PD prediction. We identified genetic modifiers for TMEM106A, ITGB2, ENTPD1, OLR1, GRN, GPNMB, GREM2, C1QTNF1, and HLA-DQA2 levels in the LRRK2 locus. <h3>Conclusions:</h3> Genetically defined PD subgroups are associated with distinct proteomic signatures. Protein modules correlated to CSF biomarkers and differentiated PD from controls. Our proteome-genome approach found that LRRK2 is a pleiotropic genetic modifier. <b>Disclosure:</b> Ms. Lai has nothing to disclose. Dr. Palanisamy has nothing to disclose. Bruno Benitez has nothing to disclose.