Modeling the Heterogeneity and Trajectories of Cognitive Dysfunction in Parkinson’s Disease Using Partially Ordered Set Models

IMPORTANCE

Cognitive dysfunction is a common and burdensome non-motor feature of Parkinson’s disease (PD). Since the heterogeneity and trajectories are not fully captured by current binary classification schemes of mild cognitive impairment (MCI) and dementia, there is a critical need to apply novel approaches for studying cognitive dysfunction in PD.

OBJECTIVES

1. To review MCI and dementia classification schemes and describe the Partially Ordered Set (POSET) model for studying cognitive dysfunction in PD. 2. To illustrate application of POSET modeling for classifying cognitive dysfunction in PD using the Parkinson’s Progression Marker Initiative (PPMI) data.

DESIGN

Based on neuropsychologists’ expert opinion, we define and map five domains of cognitive performance ( Attention, Visuospatial Judgement, Executive Functioning, Working Memory, Episodic Memory ) onto nine neuropsychological measures available in the PPMI data and then use POSET modeling to calculate corresponding Cognitive Performance Scores (CPS) as posterior probabilities of a patient’s level of performance. We then compare differences and trajectories of median value of CPS in PD patients with or without cognitive dysfunction at year 3 in a study cohort of 264 patients (selected from original cohort of 423 patients, after removing patients with MCI or dementia at baseline or missing relevant data at any of four annual visits from baseline to year 3).

EXPOSURE

Relatively lower value of CPS at baseline.

MAIN OUTCOME MEASURE

Cognitive dysfunction at year 3 (yes or no), which is defined as either MCI or dementia per assessment by the site investigator in the PPMI study.

RESULTS

The mean age was 61.16 years and there were 67% males in the study cohort. PD patients with cognitive dysfunction at year 3 (n=56, 21.2%) had significantly lower (p < 0.05) median values for all CPS at all four visits (except Episodic Memory at baseline) and exhibited distinct median trajectories over all four years. Logistic regression demonstrated that two baseline CPS (Attention and Visuospatial Judgement) were significant predictors for development of cognitive dysfunction at year 3.

CONCLUSIONS

It is feasible to apply POSET models-based approach for delineating the heterogeneity and trajectories of cognitive dysfunction in PD.

Key Points