Exploring and validating key genetic biomarkers for diagnosis of Parkinson's disease

Abstract

Background: Parkinson's disease (PD) is a neurological condition characterized by complex genetic basic, and the reliable diagnosis of PD remained limited. Objective: To identify genes crucial to PD and assess their potential as diagnostic markers. Methods: Differentially expressed genes (DEGs) were screened from the PD tissue dataset and blood dataset. Two machine learning methods were used to identify key PD-related genes. The genes were validated in an independent dataset. Further validation using 120 peripheral blood mononuclear cells (PBMCs) from PD patients. The clinical significance and the diagnostic value of the genes was determined. The function of genes was analyzed and verified by cells experiments. Results: Thirteen common upregulated genes were identified between PD tissue dataset and blood dataset. Two machine learning methods identify three key PD-related genes (GPX2, CR1, ZNF556). An independent dataset and PBMCs samples results showed increased expression in PD patients. Clinical analysis showed that GPX2 and CR1 expression correlated with early-stage PD. The validated dataset of blood samples revealed each three gene showed moderate diagnostic potential for PD, with combined analysis outperforming individual gene analysis (AUC:0.701). The PBMCs samples showed similar diagnostic value of each gene, and the combination of the three genes presented better diagnostic value (AUC:0.801). Functional studies highlighted the involvement of these genes in key pathways in PD pathology. The results of SH-SY5Y cells showed that these three genes increased from PD cell model. Conclusions: GPX2, CR1, ZNF556 were critical to the development of PD and might serve as diagnostic markers for PD.

Keywords

• Parkinson's disease
• Genetic markers
• Diagnosis
• Machine learning methods
• Blood