Abstract
Parkinson's disease is the second most prevalent neurodegenerative disorder in the Western world. It is estimated that the neuronal loss related to Parkinson's disease precedes the clinical diagnosis by more than 10 years (prodromal phase) which leads to a subtle decline that translates into non-specific clinical signs and symptoms. By leveraging diffusion magnetic resonance imaging brain (MRI) data evaluated longitudinally, at least at two different time points, we have the opportunity of detecting and measuring brain changes early on in the neurodegenerative process, thereby allowing early detection and monitoring that can enable development and testing of disease modifying therapies. In this study, we were able to define a longitudinal degenerative Parkinson's disease progression pattern using diffusion magnetic resonance imaging connectivity information. Such pattern was discovered using a de novo early Parkinson's disease cohort (n = 21), and a cohort of Controls (n = 30). Afterward, it was tested in a cohort at high risk of being in the Parkinson's disease prodromal phase (n = 16). This progression pattern was numerically quantified with a longitudinal brain connectome progression score. This score is generated by an interpretable machine learning (ML) algorithm trained, with cross-validation, on the longitudinal connectivity information of Parkinson's disease and Control groups computed on a nigrostriatal pathway-specific parcellation atlas. Experiments indicated that the longitudinal brain connectome progression score was able to discriminate between the progression of Parkinson's disease and Control groups with an area under the receiver operating curve of 0.89 [confidence interval (CI): 0.81-0.96] and discriminate the progression of the High Risk Prodromal and Control groups with an area under the curve of 0.76 [CI: 0.66-0.92]. In these same subjects, common motor and cognitive clinical scores used in Parkinson's disease research showed little or no discriminative ability when evaluated longitudinally. Results suggest that it is possible to quantify neurodegenerative patterns of progression in the prodromal phase with longitudinal diffusion magnetic resonance imaging connectivity data and use these image-based patterns as progression markers for neurodegeneration.