Résumé
Although normal subjects do not move during REM sleep, patients with Parkinson's disease may experience REM sleep behaviour disorder (RBD). The characteristics of the abnormal REM sleep movements in RBD have, however, not been studied. We interviewed one hundred consecutive non-demented patients with Parkinson's disease and their bed partners using a structured questionnaire assessing the presence of RBD. They rated the quality of movements, voice and facial expression during RBD as being better, equal or worse than in awake ON levodopa condition. Night-time sleep and movements were video-monitored during polysomnography in 51 patients to evaluate the presence of bradykinesia, tremor and hypophonia during REM sleep. Fifty-nine patients had clinical RBD with 53/59 bed partners able to evaluate them. All 53 (100%) reported an improvement of at least one component of motor control during RBD. By history, movements were improved in 87% patients (faster, 87%; stronger, 87%; smoother, 51%), speech was better in 77% patients (more intelligible, 77%; louder, 38%; better articulated, 57%) and facial expression was normalized in 47% patients. Thirty-eight per cent of bed partners reported that movements were 'much better', even in the most disabled patients. The video-monitored purposeful movements in REM sleep were also surprisingly fast, ample, coordinated and symmetrical, without obvious sign of parkinsonism. The movements were, however, jerky, violent and often repetitive. While all patients had asymmetrical parkinsonism when awake, most of the time they used the more disabled arm, hand and leg during the RBD (P = 0.04). Movements involved six times as often the upper limbs and the face as the lower limbs (OR: 5.9, P = 0.004). The percentage of time containing tremor EMG activity decreased with sleep stages from 34.9 +/- 15.5% during wakefulness, to 3.6 +/- 5.7% during non-REM sleep stages 1-2, 1.4 +/- 3.0% during non-REM sleep stages 3-4, and 0.06 +/- 0.2% during REM sleep (in this last case, it was subclinical tremor). The restored motor control during REM sleep suggests a transient 'levodopa-like' reestablishment of the basal ganglia loop. Alternatively, parkinsonism may disappear by REM sleep-related disjunction between pyramidal and extrapyramidal systems. We suggest the following model: the movements during the RBD would be generated by the motor cortex and would follow the pyramidal tract bypassing the extrapyramidal system. These movements would eventually be transmitted to lower motor neurons because of brainstem lesions interrupting the pontomedullary pathways which mediate the REM sleep atonia.