Resting State Functional Connectivity Among Cortical and Subcortical Grey Matter Structures Can Distinguish Fallers in Parkinson's Disease

The underlying neural mechanisms associated with falls in Parkinson's disease (PD) are not fully understood. We aimed to identify measures of resting-state functional connectivity (rs-FC) among cortical and subcortical regions that best discriminate fallers from nonfallers in people with PD. People with mild-to-moderate PD were classified as fallers (n = 23) or nonfallers (n = 35) based on falls in the previous 12 months. Following resting-state functional magnetic resonance imaging, measures of rs-FC were obtained for multiple subcortical and cortical regions of interest (ROIs) associated with the control of gait and balance. To investigate the combinations of (up to three) measures that best discriminate fallers from nonfallers, we applied logistic regression employing a 'best subsets selection strategy' with a five-fold cross validation and calculated the area under the curve (AUC). The most consistently selected measures in the top 10 ranked models were the rs-FC between the following ROIs: left dorsolateral prefrontal cortex (DLPFC)-hypothalamus, and somatosensory cortex-globus pallidus internus. Relative to nonfallers, fallers had reduced rs-FC between these two ROIs pairs (Cohen's d = 0.918 and 0.765, respectively). The model combining these pairs with the rs-FC between the right DLPFC and the thalamus achieved an AUC of 0.87 (95% CI: 0.77-0.97). The rs-FC between the left DLPFC and the hypothalamus, and between the somatosensory cortex and the globus pallidus internus, discriminate fallers from nonfallers in PD, representing potential targets for interventions. Connectivity deficits may limit the ability of fallers in processing information in a timely and accurate manner.