Differences in cognitive performance in nondemented Parkinson's disease: A latent profile analysis of cognitive subtypes.

Introduction: Cognitive impairments are common in Parkinson's disease (PD) patients without dementia. These deficits are quite heterogeneous, which makes it difficult to recognize and treat them. For this reason, many authors have attempted to classify patients into more homogeneous groups with diverse results. The present study was designed to analyze the cognitive heterogeneity in PD patients using a novel data-driven approach, latent profile analysis (LPA), to classify patients according to cognitive characteristics. This methodology, which has been used in previous studies focused on motor and psychiatric symptomatology, seems to be better than traditional cluster analysis for the establishment and comparison between different subgroups because it does not require prior decision making about some theoretical or methodological aspects.

Method: LPA was applied to 71 PD patients evaluated with a broad neuropsychological battery including different memory and executive function tests. The clusters obtained from the analysis were described by making comparisons with a control group of 51 healthy subjects matched in age, sex, and educational level.

Results: The LPA resulted in a four-cluster solution, which could be described as: (a) executive dysfunction (32.4%), (b) memory and executive dysfunction (28.2%), (c) memory dysfunction (23.9%), and (d) noncognitive dysfunction (15.5%). These four PD cluster differ in age and Mini-Mental State Examination (MMSE) score. However, there were no differences between clusters in disease duration, clinical impression of severity index, depression, and cognitive reserve.

Conclusions: LPA is a very interesting method for the establishment of more homogeneous groups of PD patients based on their neuropsychological characteristics. Moreover, the distinction between different cognitive profiles will allow us to design interventions better adapted to each patient.