Multimodal Imaging Biomarkers of Parkinson’s Disease

点击翻译按钮获取中文摘要

Project Summary/Abstract: There is a clear need for well-validated biomarkers for Parkinson's disease (PD) to aid early detection, more precise diagnosis, and clinical management. Numerous candidate markers are emerging, for example, spurred by initiatives such as the Parkinson's Disease Biomarker Program (PDBP) launched by the National Institute of Neurological Disorders and Stroke. One promising path to biomarker discovery involves the use of multimodal neuroimaging to reveal neuropathophysiologic characteristics of PD. PD involves a severe loss of dopamine producing neurons, which is expected to lead to downstream changes in brain function and structure, some of which manifest through in vivo neuroimaging (Politis, 2014). Identifying robust neuroimaging alterations in symptomatic PD patients creates an opportunity to assess the role of such changes for tracking disease progression and eventually to investigate whether similar changes emerge during the prodromal period. Biomarker discovery from a massive set of multimodal neuroimaging features depends critically on the development and application of advanced analytic techniques. In previous research (U18 NS082143), we developed a suite of analytic tools for cross-sectional multimodal neuroimaging data to accurately dissociate patients with mild to moderate PD from healthy control subjects. In this highly successful discovery phase, we used large-scale magnetic resonance imaging (MRI), resting-state functional MRI (rs-fMRI), and diffusion tensor imaging (DTI), and we identified three parsimonious panels of strongly predictive multimodal imaging markers. The first panel consists of 24 functional and structural markers (MRI, DTI, and rs-fMRI), which collectively reflect thalamic and limbic system alterations (e.g. hippocampus, amygdala, orbitofrontal cortex, and cingulate gyrus). The second identifies 23 markers, resulting from an analysis that includes more detailed coverage of the basal ganglia. Lastly, we identified a 15- feature structural panel (MRI and DTI), which we expect to be less susceptible to effects from PD medications. Long-term, each panel may offer advantages in practice. We embedded in our selection processes methods to promote reproducibility and model parsimony, while targeting high accuracy. In this new project, we will further evaluate the markers discovered in our previous research for validation and possibly refinement. We also seek to understand changes in these markers in distinct sets of patients who are on and off of their usual PD medications, to investigate the ability of these cross-sectional and new longitudinal markers to forecast progression, and to determine associations between clinical symptoms and the emergent imaging markers. A major advantage of this project is that we have three independent data sets, two of which have longitudinal scans, enabling further discovery and validation. The data come from the Parkinson's Progression Markers Intiative (PPMI) and from two studies conducted under the PDBP.