Dynamics of slow wave sleep in advanced Parkinson's disease: An entropy-based study

Parkinson's disease (PD) can progress to an advanced stage characterized by motor fluctuations and or levodopa-induced dyskinesia (LID). Although both motor fluctuations and LID reflect altered dopaminergic reactivity, LID has been specifically linked to disrupted synaptic plasticity. Slow wave sleep (N3) reflects homeostatic processes and may offer neurophysiological markers related to the motor phenotype in PD. This study tested whether overnight dynamics of N3 slow-wave activity (SWA) and cortical complexity, i.e., sample entropy (SampEn) differ between PD patients with and without dyskinesia, and whether these dynamics relate to dyskinesia severity. Polysomnography (PSG) was analyzed in seven healthy volunteers and two PD cohorts: non-dyskinetic (PD^dys-, n = 8), and dyskinetic (PD^dys+, n = 10). NREM sleep was divided into 10 segments; SWA and SampEn were mapped across the scalp and summarized in Early/Late N3 windows. Within-group Late-Early changes were assessed using paired permutation tests with TFCE correction. Late-Early changes were then compared between groups, and associations with the functional impact of dyskinesia (MDS-UPDRS-Part IV- item4.2) were examined in PD^dys+. In CTL, SWA decreased and SampEn increased overnight (p_TFCE < 0.001). PD^dys- showed a similar but more restricted pattern, whereas PD^dys+ showed no significant Late-Early change in either metric. In PD^dys+, Late-Early SampEn change was inversely related to dyskinesia severity (r = -0.75, pHolm = 0.068) and remained strong after adjustment for SWA (r = -0.90, pHolm = 0.018). These findings suggest that entropy-based measures capture dyskinesia-related alterations in nocturnal dynamics beyond SWA, providing a quantitative link between sleep neurophysiology and daytime motor complications in PD.