Abstract
Background
Parkinson's disease (PD) is the second most common neurodegenerative disorder after Alzheimer's disease (AD), effecting ∼ 6 million individuals world-wide. Although clinically characterised by motor dysfunction, psychiatric symptoms and cognitive impairment are observed in 30-50% of PD patients. These non-motor symptoms have significant impact on patient well-being and are associated with a faster increase in disease severity over time. PD is itself significantly heritable and polygenic, with 90 independent risk loci identified in genome-wide analysis. As such, genetically informed approaches such as polygenic risk scores (PRS) may help us better understand the underlying biological contribution to psychiatric and cognitive outcomes in PD. Transcriptional risk scores (TRS) – the sum of an individuals observed gene expression weighted by their Z-score from transcriptome-wide association analysis (TWAS) – offer a complementary approach to PRS, having shown predictive utility in previous studies of AD, amyotrophic lateral sclerosis (ALS) and Crohn's disease.
Methods
Following TWAS, we calculated PRS and TRS for AD, depression (DEP), PD and schizophrenia (SCZ) in the Parkinson's Progression Markers Initiative (PPMI) (Ncases = 431; Ncontrols = 161) and tested their association with PD-related (case/control status, age of onset, overall PD severity), psychiatric (depression, anxiety, apathy and hallucinations) and cognitive outcomes (overall cognitive impairment, working memory, processing speed, verbal learning, visuospatial ability). Associations with these outcomes were assessed at baseline and longitudinally, extracting individual random slopes for each outcome across up to ten years of follow-up using mixed-effects models. Analyses were restricted to cases-only except for case/control status. An FDR correction was applied to results correcting for 176 tests (pFDR ≤ 0.05).
Results
Both PRS and TRS for PD were significantly associated with PD case/control status, although the association was stronger for the PRS (R2 = 0.17 vs. 0.015, pFDR = 7.40e-16 vs. 0.038). Higher DEP-PRS was significantly associated with higher anxiety (pFDR = 0.031) and depression (pFDR = 0.046) in PD cases at baseline, while higher PD and AD TRS were associated with hallucinations (pFDR = 0.046 and 0.031 respectively). Higher SCZ-PRS was associated with a faster increase in PD severity over time (pFDR = 0.031). No associations were observed for the longitudinal slopes of psychiatric outcomes. However, significant associations were identified between higher levels of all TRS and faster cognitive decline across multiple domains, including overall cognitive impairment (pFDR range: 0.033-0.001). No such associations were observed for the PRS, highlighting the utility of TRS in risk prediction.
Discussion
Findings suggest that transcriptional risk scores can be applied to help us better understand the way disease-related biological risk influences psychiatric and cognitive outcomes in PD. Forthcoming analyses will seek to disaggregate transcriptional scores to identify significant gene-sets driving associations, with a particular focus on TRS associated with cognitive decline.