A clinical, pharmacologic, and polysomnographic study of sleep benefit in Parkinson's disease

Parkinson's disease (PD) patients may experience fluent mobility upon awakening from a night's sleep. This is contrary to what would be expected after a night without drugs. Marsden¹ used the term sleep benefit (SB) for this phenomenon and defined it as "restoration of mobility on awakening from sleep prior to drug intake."

Several hypothesis were proposed about underlying mechanisms in SB. Basically they can be summarized into two principal lines: sleep related and pharmacologic. The sleep-related hypothesis assumes sleep effects on central accumulation and storage of dopamine,^1-4 dopamine brain levels,⁵ and dopamine receptors.⁴ The pharmacologic hypothesis implicates drug-induced motor fluctuations,⁶ transient worsening after the first morning levodopa dose,⁷ or competition of large neutral amino acids with levodopa for transport into the brain.⁸ An association with diurnal or circadian rhythms of motor performance including fatigue was also proposed.⁹ Data about the frequency of SB within the PD population range from 10 to 20%³ to 55%¹⁰ according to the literature.

Current concepts about SB in PD are based on questionnaire studies,^4,6,10,11 anecdotal patient reports,^2,5 and theoretic considerations.^7,9 Objective data are lacking, although there is evidence of the existence of SB under certain experimental conditions as reported by Nutt et al.¹² We conducted the present study first to investigate if SB can be objectively reproduced and to assess its magnitude and its effects on motor performance. Second, we analyzed peripheral pharmacokinetics to evaluate if levodopa handling differs between patients with and without SB (NSB). Third, we performed polysomnography to investigate if SB is associated with distinct sleep integrity measures or predominance of a specific sleep stage compared with NSB patients. In addition, type of circadian synchronization and degree of depression were assessed with questionnaires.

Methods. Design. Ten PD patients with SB, pairwisely matched to 10 NSB patients, were studied with serial motor evaluations before and after sleep, serial levodopa-plasma determinations, and polysomnography. The study was designed to resemble as close as possible the conditions under which the patients experienced SB. All patients were studied on their usual antiparkinsonian and sleep facilitating medication to avoid modification of sleep measures or motor function after selection. To balance for medications, a close pairwise matching procedure between patients with and without SB was performed.

Patients and selection criteria. Twenty outpatients with PD according to U.K. PD Society Brain Bank Criteria,¹³ a greater than 35% Unified Parkinson's Disease Rating Scale(UPDRS)¹⁴ motor score reduction in a previous acute levodopa test, and a normal brain CT or MRI for age participated in this study. Patients were selected and studied between November 1995 and November 1996. All participants gave written informed consent for the protocol previously approved by the local ethics committee. Patients with cognitive impairment (Mini-Mental State Examination¹⁵ score<27), former or present hallucinations, or pretreatment with typical and atypical neuroleptics were excluded.

SB was established with a structured questionnaire. The following criteria had to be fulfilled: clearcut subjective motor restoration after a night's sleep, self-perceived mobility in the morning before drug intake appreciated as better than during the rest of the day, subjective PD symptoms clearly reduced or even abolished, and minimum duration of SB ≥ 1 hour. Isolated improvement of tremor, absence of drug motor side effects, or absence of offdystonias were not considered sufficient to establish SB. Matched pairs without SB (control subjects) were required to experience worse motor function in the morning before drug intake than during the rest of the day. Two control subjects with short disease duration did not experience any diurnal variations of motor state.

Pairwise matching was performed for gender, age (±6 years), years since onset of symptoms (±2), and antiparkinsonian and sleep-facilitating medications. Bromocriptine (respectively, lisuride transformed in bromocriptine-equivalents 1:10) and Pergolide were treated separately because different effects of additional D1-receptor stimulation on sleep can be deduced from animal studies (reviewed in^{reference 16}) and hence possible distinct sleep effects on morning motor function had to be taken into account.

Patient data are summarized in table 1. All patients had a Mini-Mental State of 28 or more. No patient was taking any other medication than specified in table 1 and none had alcohol intake exceeding two glasses of wine per week. Patients were on their medication regimens for at least 3 months and doses were kept unchanged for at least 3 weeks before the study.

Clinical and motor examinations. UPDRS motor examinations(Factor III)¹⁴ were performed at night before sleep, immediately after awakening in the morning and every 30 minutes throughout the baseline period (91 ± 23 minutes), and after intake of the usual levodopa dose until return to baseline motor state or worse. These examinations were done by a rater blind to the patient's condition with respect to SB (G.G.-A., O.S.). The best morning baseline state was taken as reference because we assumed that state most decisively influenced patients' subjective experience of morning motor function. Because SB patients usually took their medications a few hours after awakening and NSB patients did so after a few minutes, the morning baseline time had to be restricted to achieve comparable periods. Minimum duration was set to 1 hour and maximum to 2 hours. Patients continued to take their evening medication at usual hours and were free in deciding when to retire for sleep. Morning evaluations began after spontaneous awakening and verbal confirmation of the patient to feel ready to get up. During the study, a standardized dinner was served at 9:30 PM and a protein-restricted breakfast at 8:30 AM to avoid impairment of levodopa response by large neutral amino acids.⁸

To compare subjective and objective motor state, patients gave a qualitative self-assessment of their morning baseline state; to control for laboratory effects, they evaluated whether this resembled their usual state. Patients also self-rated their usual sleep quality on a scale from 1(very poor) to 10 (excellent). Additionally, all patients had a complete UPDRS examination (Factors I to VI)¹⁴ upon entering the study, a Montgomery-Asberg depression rating scale,¹⁷ and completed a Horne-Östberg morningness-eveningness questionnaire.¹⁸ Patients were asked to answer that questionnaire for general preferences and not to take into account actual needs imposed by PD such as adherence to a strict medication schedule or their known "off" times.

Pharmacologic examination. Plasma levodopa was determined by HPLC with electrochemical detection as described elsewhere.¹⁹ Five-milliliter samples of venous blood were drawn from a cannulized forearm vein into heparinized tubes, twice during basal morning state and after levodopa intake, during the first hour every 15 minutes, consecutively every 30 minutes until decline of motor score, and again every 15 minutes until return to baseline score or worse. Blood samples were centrifuged immediately and deep frozen at -70 °C until assay.

Polysomnography. A digital polygraph (Akonic Inc., Buenos Aires, Argentine) was used for recording of full EEG montage (16 electrodes, 10/20 system), electrooculography, nasal and buccal airflow (thermistor), laryngeal microphone, respiratory effort (abdominal strain gauge belt), and oxygen saturation (Criticare Systems Inc., Milwaukee, WI). Surface EMG was registered submentally from the tibialis anterior muscle and dorsal forearm of the more affected side. In nine patients, in variation of the standard recording, extremity EMG was obtained from hamstring flexors and extensor muscles of the more affected side to look for tonus inversion, as described in a case with juvenile dystonic parkinsonism.²⁰ Sensitivity of EMG channels was set according to basal tone and, if necessary, adjusted during the night. Video recording was performed throughout the night.

Sleep studies were scored from a referential montage in 20-second epochs according to standard criteria²¹ by a trained technician and supervised by one of us blind to the patient's condition(M.B.). To balance for poor spindle and K-complex formation frequently described in PD,²² light sleep stages 1 and 2 and deep sleep stages 3 and 4 were grouped together. Rapid eye movement (REM) sleep could be scored despite persistent tonic or phasic EMG activity as proposed for REM sleep without atonia.²³ Periodic leg movements during REM and isolated respiration-dependent increase of submental tone were not counted as REM sleep without atonia. Sleep latency was calculated as the time from lights out to the first of three consecutive epochs of at least stage 1 sleep. REM latency was calculated from sleep onset to the first REM. In two patients with severe sleep fragmentation, no REM sleep at all could be detected. REM latency in these patients was calculated as time from sleep onset to final awakening. Sleep fragmentation was quantified with standard awakenings²¹ and American Sleep Disorders Association(ASDA) arousals,²⁴ both reported as indices per hour of sleep. Respiratory disturbance index was considered relevant when more than 10 hypoapneas or apneas > 10 seconds occurred per hour of sleep. Periodic leg movements are reported, but the extremity electrode montage used in this study may have led to underestimation of their true incidence.

Patients were asked not to nap during the day before the study. An adaptation night was not performed, because our study aimed at the difference between SB and NSB patients and not at absolute sleep values.

Statistical analysis. StatView statistics software was used(version 4.5, Abacus Concepts, Berkeley, CA). Two-tailed paired t-tests were used for analyzing sleep and pharmacokinetic data; UPDRS scores and questionnaire scores were compared with Wilcoxon tests. Significance was determined when p ≤ 0.05.

Results. UPDRS night and morning motor examinations. The UPDRS motor scores (Factor III) before and after sleep are shown intable 2. Although the morning baseline score was slightly lower in the SB group, the difference failed to reach significance level. Comparison of score changes between night and morning showed an improvement of motor function in SB patients and a deterioration in NSB patients. The difference between both proved to be significant.Figure 1 illustrates the night and morning basal scores and their changes for each patient.

The difference of score changes between SB and NSB patients also remained significant, when calculations were done using the first morning score immediately after awakening instead of the best morning baseline score (-4.5± 7.6 in SB, 6.4 ± 6.7 in NSB, p ≤ 0.03). However, because the score immediately after awakening may be influenced by sleep inertia, it was not used for further calculations.

Levodopa response. The different characteristics of the response to the usual levodopa morning dose are illustrated infigure 2. Patients with and without SB had a clearcut motor improvement of similar magnitude. Transient deterioration also formed part of the levodopa response. It ocurred before and after clinical full "on" and was especially marked in SB patients after "on." The magnitude of this deterioration was significantly greater in patients with SB (seefigure 2). We also examined how many patients had a deterioration to a state worse than baseline that was clinically relevant. This was determined as a deterioration that exceeded 30% of baseline score. Such a deterioration occurred before "on" in four SB and in one NSB patient. After "on," it occurred in eight SB and two NSB patients. The latencies to the transient deterioration before "on" were 24 ± 13 (SB) and 27± 7 (NSB) minutes (mean ± SD). Best "on" occurred at 90± 46 (SB) and 75 ± 39 minutes (NSB), and deterioration after"on," when observed, occurred at 267 ± 33 (SB) and 229 ± 18 minutes (NSB).

Global UPDRS evaluation. Factor I (mentation, behavior, and mood) was 1.8 ± 1.4 in SB and 2.4 ± 1.2 in NSB patients (means± SD). Factor II (activities of daily living, mean between "on" and"off") was 11.3 ± 4.8 and 12.1 ± 6.4 (for Factor III, see above), Factor IV (complications of therapy) was 5.0 ± 2.7 and 6.4± 5.0, Factor V (modified Hoehn and Yahr staging) was 2.6 ± 0.4 and 2.7 ± 0.6, and Factor VI (Schwab and England activities of daily living percent) during best "on" was 90 ± 8.1 and 91 ± 8.8 and during worst "off" was 67 ± 14.2 and 64 ± 22.2. Dyskinesias were present in eight SB and nine NSB patients, mostly biphasic or mixed. Dyskinesia incidence or severity did not differ significantly between patient groups. Patients in both groups were not distinguished by their UPDRS factor scores (Wilcoxon tests).

Pharmacokinetic data. Results of plasma levodopa determinations are given in table 3. In none of the patients could residual levodopa be detected in plasma under morning baseline conditions. The morning levodopa dose was 150 ± 65 mg in SB and 157 ± 66 mg in NSB (corresponding to 2.2 and 2.3 mg/kg). Maximum levodopa concentrations and the time to reach this maximum were similar in both conditions. Although the area under the curve was slightly greater in patients without SB, this difference did not prove to be significant. Two patients (and their pairs) had to be excluded for technical reasons, and results are based on eight pairs.

Sleep data. Results of polysomnographic studies are reported intable 4. Patients in both conditions had low sleep efficiencies, high amounts of wakefulness after sleep onset, and low amounts of REM sleep when compared with published values for age-matched normal subjects.²⁵ REM sleep tended to be fragmented in some patients. In addition, there was a high frequency of awakenings and arousals compared with normal subjects.²⁶ Although most abnormalities were more pronounced in SB patients, none of the differences proved to be significant between groups (paired t-tests).

Five SB and six NSB patients had continuous or transient preserved tonic submental EMG activity or excessive phasic limb or submental muscle twitching during REM sleep (REM without atonia). A history of violent behavior could be obtained in one SB and four NSB patients and could be traced back to before PD onset in two patients. Whereas electrophysiologic REM sleep without atonia was evenly distributed among sexes (five women and six men), violent behavior related to men in all cases. Vocalizations during REM or non-REM occurred in four SB patients and three NSB patients. Three patients of either group had altered dream content at present. No sleep cycle-related muscular tonus inversion as described in a case of juvenile dystonic parkinsonism²⁰ could be detected. Proximal muscle tone showed fluctuations that were spontaneous or associated with arousals, awakenings, stage shifts, and position changes. Three patients (two SB, one NSB) had a mild obstructive sleep apnea syndrome. Periodic leg movements in sleep were present in five patients (two SB, three NSB). Median overall self-reported sleep quality was equal in both groups: 8.5 on a scale from 1(worst) to 10 (best). amytryptiline and benzodiazepines probably contributed to subjective sleep quality in these patients.

Montgomery-Asberg Rating for depression was 10.5 ± 7.1 for the SB group and 13.6 ± 6.7 for the NSB group (not significant). Horne-Östberg morningness-eveningness questionnaire ratings did not prove to be significantly different between patients with (61.9 ± 7.1) and without SB (58.8 ± 9.7). Median getup hours were 6:20 AM (range, 5:23 to 7:30 AM) in SB patients and 6:31 AM (range, 5:52 to 7:14 AM) in NSB patients (not significant).

Discussion. This study showed that PD patients with SB have better morning motor function compared with the night and that they have a different clinical response profile to levodopa than NSB control subjects. It also demonstrated that SB is not associated with different pharmacokinetics or with a distinct polysomnographic sleep pattern. Further, no self-reported circadian type predominance was found.

Patients with SB had a small improvement between night and morning, whereas NSB patients, on the contrary, had a slight deterioration. Two different factors could account for this finding. First, the favorable night-morning difference in SB patients might be attributed to better restoration of endogenous dopaminergic function during the night, proposed earlier.^2,4 Second, a pharmacodynamic factor could be involved. The response to levodopa consists in a short- and a long-duration component. The short-duration response closely parallels plasma levodopa concentrations and the long-duration response (LDR) is independent of levodopa plasma levels and persists for days after levodopa withdrawal.⁹ LDR can account for up to 50% of disability reduction in treated versus untreated patients and influences the morning motor baseline state.⁹ Our data cannot provide direct evidence for the involvement of LDR, because this could only be assessed by levodopa withdrawal for several days. However, the tendency for lower baseline motor score in SB patients would be compatible with greater magnitude of the LDR in this group.

Deterioration of motor performance to a state worse than baseline after levodopa intake was more frequent and severe in SB patients. This deterioration corresponds to the so-called "super-off" phenomenon or interdose "off." "Super-off" can occur before^7,27 and after^7,28 levodopa-induced full "on" and can be very disabling. It is associated with low plasma levodopa and probably low striatal dopamine concentrations and has been attributed to both a negative feedback via presynaptic autoreceptors or predominantly inhibitory postsynaptic receptors (reviewed in^{reference 29}). In animal studies, our group found evidences that support the involvement of postsynaptic inhibitory mechanisms in the "super-off" phenomenon.²⁹ In the absence of peripheral pharmacokinetic differences, the results of the present study would suggest that patients with SB differ from those without in the degree of sensitivity or preferential stimulation of this inhibitory receptor subtype.

On the other hand, the autoreceptor hypothesis of "super-off" would allow to speculate on the occurrence of better morning baseline function and greater "super-off" at the same time in SB patients. If SB patients had more remaining dopaminergic nerve terminals, they also would probably have more autoreceptors on them. The latter would account for better baseline function or better recovery of function during sleep, and for the more pronounced"super-off." The "super-off" phenomenon could also be involved in the tendentially worse evening scores found in SB patients. Clinically a patient, who experiences through this phenomenon, a state worse than baseline as part of his levodopa response, will perceive his morning baseline function as quite good, even if he is "off." Nutt and coworkers have speculated on this hypothesis earlier,⁷ and our results appear to support it.

Surprisingly, despite the strict selection criteria for SB patients used in this study, the motor differences between groups were quite subtle (albeit significant). Furthermore, the levodopa response in both patient groups showed a clear improvement of similar magnitude, apart from the aforementioned transient deterioration. SB patients were clearly in the "off" state during baseline motor examination. This is in contrast to the patients' overwhelmingly positive rating of their SB as the "best moment of the day, fluent mobility, no PD symptoms at all." Patients stated that their motor state at the time of morning baseline examination largely or exactly resembled the situation at home and constituted their SB, so a relevant laboratory effect does not seem probable.

How could this apparent discrepancy be explained? It is indeed possible that subjective mechanisms may play a role in SB. Several SB patients, during the protocol examinations, praised their baseline mobility as unrestrained, and free compared to the state experienced after levodopa, although their overall appearance and objective evaluation by UPDRS were indicative of the patients being in the "off" state. Possibly, very subtle, subclinical motor side effects of levodopa could lead to impairment of subjective movement quality. Another possible explanation is that nondopaminergic mechanisms could influence subjective movement quality. In one particular instance, a 73-year-old man with 11 years of disease duration, belonging to the SB group, had claimed that he had excellent mobility, agile and long-stepped gait with ample arm swing, that lasted for hours after awakening. On observation, however, he was brady- and hypokinetic, his gait was short-stepped, and without arm swing. We confronted him with the fact that after levodopa he had doubled his velocity in timed tests (results not reported) and that his motor disability score was improved by more than 50% compared to baseline. After insistent questioning by the examiners to explain this discrepancy, he admitted: "it may be true that I am slightly slower, but this is not relevant." Time perception, therefore, might play a role in SB. Pastor and co-workers showed that PD patients underestimated time intervals verbally and reproduced given time samples as longer than they really were. Both abnormalities were significantly reduced after levodopa intake.³⁰ Based on their findings, the authors suggested an abnormally slow inner clock in PD patients. In view of SB this could mean that abnormally slow inner clocks during baseline preclude patients from noticing, how slow they really are. Another recent study found reduced kinesthesia, or movement perception in PD patients and proposed a model of a smaller set sensorimotor apparatus to explain why this could pass unnoticed by the subject.³¹ This might lead to a false perception of movement amplitude in hypokinetic patients. For this interpretation to be valid, one might assume that time perception and kinesthesia are more affected in baseline conditions in SB patients.

Sleep. Polysomnography did not reveal a relation between SB and specific sleep measures, specifically sleep integrity, amount of slow wave sleep, and amount and timing of REM sleep. We cannot exclude that our study design with close pairwise matching of patients and medications may have masked distinct sleep patterns otherwise associated with SB, that might have been found when comparing groups not matched for medications. However, a different design would implicate a drug washout or drug-restriction period, and hence manipulate the conditions under which SB is experienced. When comparing the present results of polysomnography to previously published work in PD patients (see following) or age matched normals,²⁵ one should keep in mind that our patients were studied on full medication and without adaptation night. We do not think, however, that this interfered with our results as the aim of this study was to investigate the difference between SB and NSB patients and not absolute sleep values. Medication effects or a first night effect should be similar for both groups. Nevertheless, our findings are in line with much published work on sleep in PD patients, such as sleep fragmentation, increase of wakefulness after sleep onset, and reduced sleep efficiency (reviewed in^{references 22 and 32}). This is, to our knowledge, the first study to include arousal indices as an additional sleep fragmentation measure in PD patients. Compared to published values for normals they were slightly increased in both conditions. REM sleep percentages were low and REM sleep onset delayed in both conditions. Previous studies in PD patients found varying amounts or latencies of REM sleep which may in part be explained by different disease stages or medications (reviewed in ^{reference 22}). Low amounts of REM sleep and delayed REM sleep onset have been described in untreated PD patients.³⁴ In the present study, they may additionally be attributed to sleep fragmentation, levodopa, dopaminergic agonists,³⁵ selegiline, benzodiazepines, and amitriptyline(reviewed in ^{reference 36}). Furthermore, a first-night-effect, although questioned in PD,³⁷ and the local-type late evening meal³⁸ may have reduced REM sleep and delayed its appearance. On the other hand, such abnormalities may have been attenuated by medications. Sleep in PD has been shown to improve with adequate antiparkinsonian therapy.³⁴ Despite the ad-lib times for sleeping and getting up, total times in bed were similar in both conditions. There was a nonsignificant tendency for most sleep measures to be more abnormal in the SB group, specifically higher amounts of wakefulness and lower amounts of REM sleep. This is in line with questionnaire study results¹⁰ with SB patients reporting more awakenings. A possible interpretation of these findings is that SB is independent from sleep quality. On the other hand in a rat model of parkinsonism, REM sleep deprivation had a beneficial effect on motor state, possibly due to dopaminergic receptor supersensitivity.³⁹

A collateral finding of this study was the high relative frequency of REM sleep without muscle atonia that was asymptomatic or associated with violent behavior. Mahowald and coworkers have drawn attention to the complex relation between REM sleep behavior disorder (RBD) and PD, with asymptomatic REM sleep without atonia on the one hand⁴⁰ and RBD preceding PD on the other hand.²³ Medications possibly influenced our findings, as behavioral manifestations of RBD respond to clonazepam, and levodopa.⁴⁰ Moreover REM sleep without atonia was also reported with tricyclic antidepressants and monoamineoxidase-inhibitors and activation of REM sleep behavior disorder was described with selegiline.⁴⁰ It is not known if REM sleep without atonia relates to SB. However, the similar frequency of this phenomenon observed in both conditions and the complex nature of SB do not support a simple relation.

The influence of circadian rhythms on SB was briefly addressed in this study with a questionnaire to determine morningness-eveningness.¹⁸ No significant difference in type of circadian synchronization was found. However, longstanding need of regular medication intake and drug-induced motor fluctuations much greater in amplitude than circadian changes, may have masked differences.

The relation between SB and diurnal motor variations has not been specifically addressed yet. Nutt and coworkers demonstrated in a group of PD patients not specifically selected for explicit SB a decline of tapping speed during the day and no further decline or even improvement during the night.¹²

The results of the present study characterize SB as a distinct, although probably multifactorial entity. Additionally SB in PD may be embedded in diurnal or circadian fluctuations as those well-known in various extrapyramidal disorders such as hereditary progressive dystonia, and the common, probably multifactorial afternoon deterioration of PD symptoms.

Although SB has a temporal relation to the morning after a night's sleep, significant evidence linking it to a specific sleep variable was not found. The differences in motor performance found in patients with and without SB point to underlying pharmaco-dynamic mechanisms. Apart from pharmacologic factors influencing motor behavior, sensory factors may additionally contribute to SB.

Acknowledgments

We thank Pablo Orué for excellent technical execution of the polysomnographies, Claudia García-Bonelli for performing the HPLC, Dr. Norma Poisson for handling the blood samples, Dr. José Tessler for advice with study design and statistics, and Dr. KL Leenders (PSI Villigen, Switzerland) for helpful comments.