Resting state connectivity changes in rat model of per-motor stage Parkinson’s disease

Resting state (RS) connectivity is a sensitive measure of brain network alterations, and has been shown to change during age and in several disorders including Alzheimer’s Disease, Depression, ADHD, and Multiple Sclerosis [1]. This study analyzed the seed based RS connectivity changes in our rat model of pre-motor deficit stage Parkinson’s disease as compared to the age matched controls. The seeds were chosen based on the ICA components, generated using the GIFT toolbox in MATLAB (Mathworks, Natick, MA), which matched the regions of interest (ROI). Only pixels with within component z-scores > 1.65 were used to generate the averaged seed time-series which was used to generate the connectivity map for that seed. Even though, a seed only comprised of a few pixels within the ROI, the final connectivity map had a good representation of the entire ROI. The echo-planar imaging technique, used to collect the RS fMRI scans during this study, can suffer from image distortions [2, 3]. Anatomical atlas based seeds can have effects of surrounding regions in its final connectivity map. Therefore, choosing high z-score ICA components as seeds is a better technique of seed selection than pure anatomical atlas based seeds.

Two seeds that exhibited major differences between the PD and the control groups were the motor cortex and the amygdala. The positive correlation between motor cortex and the majority of the brain had a significant increase in the PD rats as compared to controls.  This parallels the increased motor cortex connectivity reported in un-medicated early stage PD patients [4]. The amygdala in controls had significant negative correlation with PFCs, cingulate, motor and somatosensory cortices, as well as, caudate putamen and thalamus similar to the reports in normal human controls [5]. The amygdala in PD rats lost these negative correlations. Similar observations have been noted in addicts and is thought to be linked to impulse control [6].