White matter connectivity networks predict levodopa-induced dyskinesia in Parkinson's disease

Although levodopa-induced dyskinesia-relevant white matter change has been evaluated, it is uncertain whether these changes may reflect the underlying predisposing conditions leading to the development of levodopa-induced dyskinesia.

To elucidate the role of white matter connectivity networks in the development of levodopa-induced dyskinesia in drug-naïve Parkinson's disease.

We recruited 30 patients who developed levodopa-induced dyskinesia within 5 years from MRI acquisition (vulnerable-group), 47 patients who had not developed levodopa-induced dyskinesia within 5 years (resistant-group), and 28 controls. We performed comparative analyses of whole-brain white matter integrity and connectivity using tract-based spatial and network- and degree-based statistics. We evaluated the predictability of levodopa-induced dyskinesia development and relationship with its latency, using the average connectivity strength as a predictor in Cox- and linear-regression, respectively.

Mean-diffusivity was lower mainly at the left frontal region in the vulnerable-group compared to the resistant-group. Network-based statistics identified a subnetwork consisting of the bilateral fronto-striato-pallido-thalamic and lateral parietal regions (subnetwork A) and degree-based statistics identified four subnetworks (hub-subnetwork) consisting of edges centered on the left superior frontal gyrus, left putamen, left insular, or left precentral gyrus, where the vulnerable-group had stronger connectivity compared to the resistant-group. Stronger connectivity within the subnetwork A and hub-subnetwork centered on the left superior frontal gyrus was a predictor of levodopa-induced dyskinesia development independent of known risk factors and had an inverse relationship with its latency.

Our data suggest that white matter connectivity subnetworks within corticostriatal regions play a pivotal role in the development of levodopa-induced dyskinesia.