A Multimodel Biomarker Approach to Evaluating and Predicting Cognitive Decline in Lewy Body Diseases

Alice Chen-Plotkin, MD Alice Chen-Plotkin, MD
Project Leader

Parkinson's disease studies at Penn

Project Description

This project will work to replicate previously-reported candidate biomarkers of cognitive impairment (CI) in a training cohort of LBD patients. We and others have reported promising candidate biomarkers of CI in PD and other LBD.  These candidate markers encompass multiple modalities: (1) clinical features, (2) genetic markers, (3) biochemical markers, and (4) imaging markers. We propose to evaluate a set of 20 candidate markers that have been previously reported in the literature for association with cognitive performance in a training cohort (n=375) of LBD patients. The goal of this aim is to replicate previously-reported findings in our cohort, thereby demonstrating the generalizability of the markers and the relevance of our cohort to other LBD populations.

The project will also define relationships among candidate biomarkers in Aim 1 and identify potential pathophysiological subtypes of CI in LBD. There is ongoing controversy regarding the pathophysiological substrate of CI in LBD. We propose to evaluate relationships among markers in two distinct ways. First, in an extension of our prior work, we will conduct a hypothesis-driven analysis to determine whether markers associated with Alzheimer’s disease (AD) correlate with each other, defining a subgroup of patients in whom CI is substantially due to co-existing AD pathology. Second, we will use unsupervised classification methods to unmask latent subtypes of CI in LBD distinguished by specific patterns of clinical and biological markers. 

Penn Udall Center for Parkinson's Research Public Health StatementLastly, the project will develop a multimodal predictive algorithm for cognitive decline in LBD and apply it to an independent test cohort of PD patients. We will use data from Aims 1 and 2 to develop three types of multimodal models for assessing risk of significant cognitive decline in individual PD patients. We will then apply these models to a separate, independent test cohort (n=225) of PD patients. In this cohort, we will assess the ability of each type of model to identify those individuals most at risk for cognitive decline in a 2-year window. Finally, we will construct a user-friendly web-based clinical tool for stratifying near-term dementia risk in patients with PD.