BACKGROUND:
Parkinson’s disease (PD) is primarily characterised by motor symptoms, but dementia is a common outcome that signals poor prognosis and care home admission. Predicting who will develop dementia remains a clinical challenge. The NULISA TM platform is a multiplex proteomic assay that detects neurodegeneration-relevant proteins with high sensitivity.
METHODS:
We used the NULISAseq TM CSF Disease Panel 120 to quantify baseline serum proteins from 730 individuals with PD. Participants underwent serial cognitive assessments using the Montreal Cognitive Assessment (MoCA) over a mean of 3.9 ± 2.1 years, with an average of 3.4 ± 1.3 assessments per person. We used a Cox proportional hazards model to identify proteins associated with time-to-dementia, and a linear mixed model (LMM) to identify which proteins predict a decline in MoCA scores within the first 3 years of follow-up. Domain-specific cognitive decline was also evaluated. All models were adjusted for sex, age at serum sampling and batch. P-values were corrected using the Benjamini-Hochberg (BH) method.
RESULTS:
Age at baseline was 67.1 ± 8.9 years, with mean disease duration from diagnosis of 1.3 ± 0.9 years. The mean adjusted baseline MoCA score was 25.5 ± 3.8. Survival analysis revealed significant associations for phosphorylated Tau proteins (pTau-217, pTau-181 and pTau-231), NPTXR and FCN2. Adjusting for APOE4 status attenuated most signals, with only FCN2 surviving BH correction. Adding these proteins to a predictive model with age and sex improved the AUC from 0.70 to 0.75 (DeLong’s test, adjP = 0.041). The AUC further increased to 0.81 when adding the adjusted MoCA score at baseline (adjP = 0.020). The LMM revealed that higher levels of pTau-231, GFAP, pTau-181 and NFLH predicted faster MoCA decline within the first 3 years of follow-up. GFAP was linked to deterioration in attention, executive, and language domains, while CXCL10, pTau-181, and pTau-231 were associated with language decline over 5 years.
CONCLUSION:
We identified serum proteins associated with cognitive decline and time-to-dementia in PD. The attenuation of signals after APOE4 adjustment suggests that these markers may reflect Alzheimer’s co-pathology. Further studies in PD patients without AD co-pathology are needed to identify biomarkers specific to Lewy Body disease.