Table 3.

Linear discriminant analysis for RNA PD biomarkers tested in PDBP.

	PD vs. PSP	PD vs. HC	PSP vs. HC	CI vs. CN
Variables retained in the model	COPZ1 PTPN1	COPZ1 PTPN1	COPZ1 PTPN1 FAXDC2 SLC14A1s NAMPT	COPZ1 PTPN1 FAXDC2 EFTUD2 MLST8 Age
Variables removed from the model	EFTUD2 FAXDC2 SLC14A1s NAMPT PTBP1 MLST8 COPS7A Age Sex	EFTUD2 FAXDC2 SLC14A1s NAMPT PTBP1 MLST8 COPS7A Age Sex	EFTUD2 PTBP1 MLST8 COPS7A Age Sex	SLC14A1s NAMPT PTBP1 COPS7A Sex
Overall diagnostic accuracy (%)	62.5%	58%	69%	65.9%

A forward step-wise linear discriminant analysis was performed to identify the best group of variables for discriminating between groups. CI, cognitively impaired; CN, cognitively normal; HC, healthy controls; PSP, progressive supranuclear palsy; PD, Parkinson’s disease.