Objective: To determine the clinical profile of cognitive subtypes in Parkinson’s disease (PD) based on their rate of change in global cognitive ability over time.

Background: A subset of people with PD develop dementia more rapidly than others. While some predictors of dementia in PD have been established, it is unclear whether the neurocognitive profile at baseline of a patient is related to their rate of cognitive decline.

Method: Data from the PPMI and ICICLE-PD cohorts was used. Latent class mixed modelling was performed to identify subtypes (latent classes) among 775 newly diagnosed PD patients. Subtypes were stratified based on their rate of change in global cognitive decline over 7 (PPMI) or 4.5 years (ICICLE-PD) as measured using the Montreal Cognitive Assessment. Exploratory-confirmatory analyses were performed, with the PPMI data being used to determine the optimal number of classes that was subsequently applied to the ICICLE-PD data. Baseline clinical data were compared between classes to determine their profiles.

Results: The final model revealed four latent classes (PPMI: BIC = 13635.65, ICICLE-PD: BIC = 3074.38). Two of the classes were younger and cognitively intact at baseline, with stable cognition over time. An older, cognitively impaired subtype with consistently poor cognition was also identified. The final class differed between the two datasets. PPMI data revealed a subtype that was cognitively intact at baseline but with rapid decline, while ICICLE-PD data instead revealed a subtype that was mildly impaired at baseline with gradual decline over time. Despite differences in cognitive measures used across datasets, baseline neurocognitive profiles revealed measures that could be predictive of subsequent cognitive decline. Poor judgement of line orientation, but not category fluency, was predictive of decline in the PPMI dataset, whereas deficits in category fluency, but not measures of visuospatial function, were predictive of decline in the ICICLE-PD dataset.

Conclusion: Our results suggest that PD patients with different cognitive trajectories present with unique clinical profiles, although these differed to some extent between cohorts. Identifying those at risk of dementia prior to severe cognitive decline will facilitate person-centred care and help to target clinical trials and interventions to the appropriate patients.

« Back to 2023 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/using-latent-class-mixed-modelling-to-profile-parkinsons-disease-patients-at-risk-of-cognitive-decline/