Hallucinations, REM sleep, and Parkinson’s disease

Patient 1.

The initial symptoms of PD in a 69-year-old man were akinesia and rigidity. Levodopa therapy was effective from the onset. Seven years later, at the time of the study, the patient was taking 100 mg of levodopa four times a day. His parkinsonian motor disability was stage 3 on the Hoehn and Yahr scale.13 Motor fluctuations, abnormal involuntary movements, mood disorders, and cognitive disturbances (Mini-Mental State Examination [MMSE] score: 28/30)14 were absent. Occasional falls, sometimes related to freezing, were observed. After 3 years of treatment, he experienced frequent visual, auditory, tactile, and olfactory hallucinations. He saw his children sitting beside him on the sofa, tiny people moving on the wallpaper, and a doll making faces at him. He heard his neighbors whisper insults, felt someone hit his shoulder, and smelled unpleasant odors in his bed. Although the patient was critical of his visual, auditory, tactile, and olfactory hallucinations and illusions, he was convinced that his wife was unfaithful and that his neighbors were conspiring against him. His spouse reported aggressive motor behavior during the night and severe daytime sleepiness. She thought he had erotic dreams, since he had erections, moved his body, and moaned while asleep.

The results of overnight polysomnography followed by a multiple sleep latency test showed that total sleep time during the night was reduced to 155 minutes with poor (37%) sleep efficiency and a long period of wakefulness lasting 201 minutes. REM sleep lasted 42 minutes and was not always accompanied by atonia. Violent motor behavior was observed during this stage. After one episode, the patient was sure that an aggressor was hiding in the corner of the room (figure 1A). During the day, the mean sleep latency was 2.4 minutes, confirming excessive daytime sleepiness. REM periods occurred at the onset of sleep during the first and fifth tests and a brief episode of atonia was noted during the fourth test. His Class II human leukocyte antigen (HLA) was DQB1-03-011, not DQB1-602, which is a marker of narcolepsy.15

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Hypnograms in two patients with abnormal REM sleep and hallucinations. (A) Patient 1. The x axis shows hours, the y axis shows non-REM sleep stages 1–4 (open bars) and REM sleep (filled bars). A delusional episode, indicated by the arrow, followed the second REM sleep episode during the night. During the day, the patient could take five 20-minute naps at 2-hour intervals (multiple sleep latency tests). (B) Patient 2. Hallucinations reported by the patient are indicated with arrows and occurred during daytime REM sleep episodes.

The patient died of pulmonary embolism 4 months after the sleep recordings. Macroscopic examination of the brain showed depigmentation in the subtantia nigra and locus coeruleus but no cortical, subcortical, or brainstem atrophy. Lewy bodies in the substantia nigra and in the pedunculopontine nucleus stained with hematoxylin-eosin or with antibodies against ubiquitin and synuclein were few. They were numerous, however, in both melanized and nonmelanized neurons in the caudal part of locus coeruleus. A few ubiquitin-positive neurons were scattered throughout the cerebral cortex. An antibody against Aβ protein revealed large numbers of senile plaques, vascular deposits, and diffuse labeling in cortical samples. Tau-positive tangles were confined to the entorhinal cortex and pyramidal cells of the hippocampus. The lesions were classed as Alzheimer stage 1.16 Neither Bodian silver impregnation nor ubiquitin immunoreactivity detected the characteristic inclusions of multiple system atrophy or Pick’s disease.

Patient 2.

A 66-year-old woman had a mild form of PD (Hoehn and Yahr score on medication: 3) for 15 years. Her response to levodopa was good, but disabling axial dyskinesia was observed. Treatment at the time of the study included levodopa (500 mg), bromocriptine (30 mg), piribedil (150 mg), and selegiline (2 mg). Over a 3-month period, she experienced frequent silent visual hallucinations. She saw her husband, who had died 4 years earlier, sitting on the sofa, but his head and expression were not exactly those of her husband. Sometimes, the image was so frightening, with bleeding cuts on the arms, that she was afraid to go into the living room. One evening, she saw a complex image where moving children were mixed with tree leaves. She called the doctor and refused to sleep in her house, because she thought that someone was deliberately sending these images in order to harm her. Images of children were recurrent and less frightening, but she often called her daughter, whispering: “They are here again; do you think I should prepare dinner for them?” Her family reported that she was agitated during the night, moved her hands, and shouted, and was sleepy during the day. Daytime hallucinations were frequent after very short naps.

Overnight polysomnography showed that total sleep time was only 296 minutes. Sleep efficiency was reduced (65%) although sleep latency was short (7 minutes). During four daytime sleep latency tests, she fell asleep within 2.5 ± 1.9 minutes, and REM sleep was present during each test. After two tests with REM onset, she recalled having seen a person under the TV set. After the other tests, she remembered dreaming (figure 1B). Piribedil, selegiline, and bromocriptine were withdrawn, and the hallucinations disappeared within 2 months. FIGURE

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Daytime sleep latencies in hallucinating (▴) and nonhallucinating (▪) patients with PD. Mean ± SEM. *The groups are significantly different at 8 am; p < 0.01.

Methods.

Following these two observations, we studied nighttime and daytime sleep in eight more consecutive levodopa-responsive PD patients17 referred for hallucinations and 10 control PD patients without hallucinations. The patients were recruited between November 1998 and September 1999 at the movement disorders outpatient clinic, a tertiary care center of a large university hospital, where approximately 1500 patients with PD are seen in consultation quarterly. Eighteen of them were seen regularly in the clinic and lived in the Paris area. Two were new patients referred by neurologists in private practice for advice about levodopa-induced psychoses. The inclusion criteria for the hallucinators group were: 1) clinical symptoms of idiopathic PD17 with a motor response to levodopa greater than 50%; 2) presence of daily hallucinations; 3) absence of dementia (MMSE score14 greater than 24) or confusional states. Two patients declined to participate, arguing they would not sleep in the hospital.

The 10 patients with PD (two women and eight men, 59 to 80 years of age, with disease duration of 5 to 15 years) had Hoehn and Yahr scores of 2.5 (n = 2), 3 (n = 3), and 4 (n = 5) when treated.13 None was demented (MMSE scores: 26 to 30). Motor fluctuations and dyskinesia were observed in four patients. Occasional falls were reported in six. Some falls were associated with stressful stimuli, like hearing the telephone or crossing a road. All had been treated long-term with levodopa (400 to 850 mg/day); other antiparkinsonian drugs were associated in three patients (bromocriptine 5 mg/day, selegiline 10 mg/day, and apomorphine 12 mg/day; ropinirole 15 mg/day; bromocriptine 30 mg/day and piribedil 150 mg/day). In the remaining seven patients, dopamine agonists, tricyclics, and other antidepressant drugs had been withdrawn for at least 6 months because of severe hallucinations. None of the patients received neuroleptics or anticholinergic drugs. Five patients took benzodiazepines at bedtime (table 1). All patients reported having experienced daily visual hallucinations for a mean of 12 months (range 6 to 48 months). Most visions consisted of people seen in color (a little girl with a red hairband and umbrella, Playboy-type images of topless women, an African man raking). One patient regularly saw small unknown persons with aggressive faces at the end of his garden. He tried to photograph them but did not see them on the prints. The hallucinations were felt to be real by all. Delusions were present in seven. They feared danger (n = 6; one attempted to strangle his wife), were convinced of being followed (n = 1), or believed that their spouses were impostors (n = 3) or unfaithful (n = 2). One patient had dysmorphophobia.

The 10 controls were levodopa-responsive nondemented PD patients without hallucinations. Although sleep disturbances were not a criterion for entering the study, the patients who agreed to participate might have been motivated by their own sleep disturbances. This is a potential bias. However, as more than two-thirds of patients with PD have sleep disturbances,18 excluding such patients would have biased the study even more. Hallucinators and nonhallucinators were matched for age, disease course and severity, and levodopa-equivalent dose (see table 1). Motor fluctuations and dyskinesias were observed in four of the control patients. Six were treated with dopamine receptor agonists (3 peribedil, 3 bromocriptine, 1 ropinirole, 1 apomorphine). Three used hypnotics (benzodiazepine) at bedtime. No visual defects or visual neglect were found by neurologic examination in patients of either group.

Each patient was studied for 24 hours. The protocol included: 1) an interview of patient and caregiver about psychiatric signs and sleep disorders; 2) assays of HLA markers of narcolepsy in blood sampling from 18 patients; 3) sleep recordings from 10 pm (lights off) to 6:30 am (videorecording in 18) in the presence of a nurse who reported all abnormal events; 4) sleep latency tests at 8 am, 10 am, noon, 2 pm, and 4 pm. The patients were given the opportunity to nap in the dark for 20 minutes at 2-hour intervals, after which they were asked about their dreams or hallucinations. Polysomnographic recordings included fronto–central and occipito–central EEG, vertical and horizontal electro-oculograms, chin and tibialis anterior electromyograms, thermistance measures of naso–oral flux, abdominal and thoracic belts to measure respiratory movements, electrocardiogram, and pulse oximetry. Sleep was scored visually by two experienced scorers according to standard criteria.19 As atonia may be absent during REM sleep, REM sleep was defined as periods with rapid eye movements and saw-tooth waves on EEG or theta activity (if it was absent during a quiet period of wakefulness before going to bed) or brief episodes of atonia. The occurrence of REM sleep might therefore be underestimated.

Results.

Discussion.

All patients included in this study met the criteria for idiopathic PD.17 The association of levodopa-responsive parkinsonism and hallucinations in patients with psychosis is most often indicative of Lewy body disease.21 In the absence of dementia, however, this diagnosis seemed unlikely, although it cannot be excluded, except in Patient 1 who was examined postmortem.

Perspectives.

Several key features of the relationship between REM sleep intrusions and hallucinations in patients with PD remain to be clarified. For example, hallucinations were reported by the patients after a 20-minute nap. We could not therefore determine if they were hypnagogic or hypnopompic. Because hallucinations are rarely reported spontaneously by the patient, continuous 24-hour EEG recordings in severe hallucinators, interrogated constantly by the investigator, would help to determine the precise timing of the hallucinations with respect to REM sleep intrusions, and to clarify whether the patients’ eyes are open during the event. Other REM sleep correlates such as penile erection, functional imaging, or spontaneous middle ear activity would be useful to characterize dissociated states. Larger series of patients are necessary to determine whether there is a continuum between isolated daytime REM sleep episodes in PD patients without hallucinations and REM storms observed in severe hallucinators. Determining whether sleep attacks described with the new dopaminergic drugs10 are abrupt REM sleep onset is urgent. The relationship between the sudden loss of muscle tone observed in primary narcoleptics and falls in PD patients should be investigated. Multiple sleep latency tests should be performed in patients with diffuse Lewy body disease to determine whether fluctuations in alertness and hallucinations correspond to REM sleep intrusions. Neuropathologic examination, coupled with staining of neuromediators, such as acetylcholine, of REM sleep executive mechanisms should focus on the locus subcoeruleus, a region that is not routinely examined in PD patients (D. Seilhean, personal communication, 1999), but is clearly involved in modulating REM sleep in animals.

This study suggests that hallucinations and delusions in nondemented patients with PD can result from abnormal REM sleep, and that psychosis may have a narcolepsy-like rather than a psychogenic origin. Because the known biologic substrates of REM sleep are subcortical, lesions in these pathways, rather in than cortical circuits, are probably responsible for a significant subset of hallucinatory experiences. These patients should therefore avoid the use of sedative drugs, particularly benzodiazepines, during the daytime, but might benefit from the arousal-enhancing drugs used to alleviate sleepiness and REM sleep attacks in narcolepsy.