Objective: To apply growth mixture models (GMM) to elucidate heterogeneity in cognitive change trajectories in Parkinson’s Disease (PD) among idiopathic PD (iPD), GBA1, and LRRK2 variant carriers with PD.

Background: Determining subgroups of PD has the potential to inform clinical trial design as well as clinical care, and to further understand underlying pathophysiology.

Method: Using longitudinal data from MJFF PPMI, LCC, PDBP, and Mount Sinai Beth Israel, GMMs were constructed to model trajectories of cognitive decline as assessed with Montreal Cognitive Assessment (MoCA) up to a 20-year period. A quadratic trend was applied to model trajectories controlling for age and PD duration at baseline. 1-5 classes were run, with BIC and class size proportions utilized to determine the optimal number of classes. Analyses were performed in R 4.4.0 using the package lcmm.

Results: 2586 PD were evaluated, including 1632 iPD (37% women), 337 GBA-PD (41% women), 574 LRRK2-PD (51% women) with mean ages of 65.1±10.1 years at baseline and 60.6±10.4 years at diagnosis. A 3-class model optimally described progression groups. 1055 (40.8%) patients were classified as slow decliners (ClassS), 1222 (47.3%) as medium decliners (ClassM), and 309 (11.9%) as fast decliners (ClassF). ClassF were older at baseline (70.3±9.5 vs. ClassS 63.8±9.7 years, p<0.001), with longer disease duration (5.7±5.1 vs. 4.3±4.5 years, p<0.001), and were less likely to be women (34% vs. ClassS 45%, p<0.001). ClassF were also more likely to be hyposmic (88% vs. ClassS 74%, p<0.001). Within genetic groups, ClassF had higher proportions of GBA-PD men (14% vs. ClassS 6%, p<0.001), while ClassS had overrepresentation of iPD women (25% vs. 17%, p=0.002).

Conclusion: Our findings support distinct cognitive progression profiles in PD, with faster decline associated with older age, male sex, and GBA-PD, raising the question as to whether these factors should be considered in clinical trial design. The observed variations in hyposmia across latent classes further suggest distinct phenotypic profiles linked to Lewy body pathology, highlighting potential biological underpinnings of heterogeneity in PD. However, predicting individual patient prognosis remains challenging, as some GBA-PD men exhibited slower decline. At the same time, this paradoxically well performing subgroup likely harbors protective factors which might warrant further investigation.

Table 1: Clinical summary by MoCA latent class

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/applying-growth-mixture-models-to-elucidate-heterogeneous-trajectories-of-cognitive-change-in-parkinsons-disease/