Time-dependent variation in cerebral and autonomic activity during periodic leg movements in sleep: implications for arousal mechanisms

Clinical Neurophysiology 113 (2002) 883–891 www.elsevier.com/locate/clinph

Time-dependent variation in cerebral and autonomic activity during periodic leg movements in sleep: implications for arousal mechanisms Emilia Sforza*, Christophe Juony, Vincent Ibanez Sleep Laboratory, Department of Psychiatry, Geneva University Hospital, 2 Chemin du Petit Bel Air, 1225 Chene Bourg, Geneva, Switzerland Accepted 2 March 2002

Abstract Objectives: A hierarchy in arousal response has been proposed for spontaneous arousal by analyzing the temporal changes in heart rate (HR) and electroencephalographic (EEG) activity. To address the question as to whether the same continuum may be proposed in sleep disorders, we performed temporal spectral EEG and HR analyses during periodic leg movements (PLM) associated or not with microarousal (MA). Methods: Data were obtained in 12 patients with restless leg syndrome and/or PLM syndrome. PLMs were classified into 3 types including PLM associated with MA, PLM without MA, and PLM associated with delta or K-complex bursts. HR and EEG spectral analyses were done for 10 s before and 10 s after the PLM onset. Results: Each type of PLM was associated with a typical EEG and autonomic pattern consisting of an increase in HR and delta band activity before the PLM, regardless of the presence or absence of MA. Thereafter, a rise in delta, alpha and beta2 activity was noted associated with tachycardia. This was greater when MA or bursts of slow wave activity were present. In the period following the PLM, HR, delta and alpha power showed a long-lasting decrease with values significantly below the baseline. Conclusions: From these data, we can conclude that: (1) cardiac and cerebral changes occur in association with PLM even when MA cannot be detected; (2) the combined increase in delta activity and HR before the onset of PLM suggests that these changes are part of the arousal response during PLM; (3) the graded arousal response during PLM confirms that the human arousal response involves a progression of central nervous system activation from brainstem to cortical level. q 2002 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Periodic leg movement; Arousal; Cardiac activation; Electroencephalographic spectral activity

1. Introduction An arousal from sleep is a physiological event resulting from two competing forces, the stimulus that disrupts ongoing sleep and the cortical and subcortical mechanisms that attempt to minimize its impact and to preserve the continuity of sleep. To investigate the processing of arousal response and the effect of sleep fragmentation, many studies have focused on microarousal (MA), corresponding to a brief change in fast electroencephalographic (EEG) activity occurring simultaneously with or after the pathological phenomenon (ASDA, 1992). It has been proposed for years that the frequency of apnea (Roehrs et al., 1989), periodic leg movement (PLM) (Rosenthal et al., 1984; Saskin et al., 1985) and the associated arousals are responsible for the complaint of non-restorative sleep, daytime sleepiness and fatigue reported by patients. However, * Corresponding author. Tel.: 141-22-3055329; fax: 141-22-3055343. E-mail address: [email protected] (E. Sforza).

when we consider more recent studies we must confront the fact that a cause–effect relationship between the number of arousals and somnolence is not yet well established. The index of PLM associated with MA did not differentiate between sleepy and non-sleepy PLM patients (Coleman et al., 1982; Mendelson, 1996) and apnea recurrence and concomitant sleep fragmentation (Kingshott et al., 1998) do not interfere significantly with reduced alertness in patients with obstructive sleep apnea (OSA). Perhaps this discrepancy stems in part from a lack of appropriate visual methods to detect arousal. It has now been suggested that there is a hierarchy of arousal phenomena (Halasz, 1998) generated from subcortical and cortical areas. Arousal can vary from brief increases in myographic activity (McNamara et al., 1999), respiratory (Carley et al., 1996) and cardiac activity (Sforza et al., 1999) to sleep stage shifts and full awakening (Halasz, 1998). We recently (Sforza et al., 2000) tested the hypothesis of a categorization of the spontaneous arousal response starting with surges in sympathetic activity, i.e. autonomic arousal,

1388-2457/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1388-245 7(02)00066-4

CLINPH 2001721

884

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891

progressing to synchronized EEG-sleep patterns, i.e. slow delta and K-complex bursts, and finally to MA. Using the temporal analysis of cerebral and autonomic variations across arousals, we found that irrespective the type of arousal considered, there was a stereotyped pattern in EEG and autonomic response consisting of a rise in heart rate (HR) and fast EEG frequencies, that was greater when a cortical MA was present, and starting 1 s before the arousal onset. Similar data were obtained by experimental studies in animals (Quattrocchi et al., 2000; BuSha et al., 2001) and in humans (Carley et al., 1996; Davies et al., 1993), showing that during spontaneous or evoked arousal increases in ventilation, HR and blood pressure may occur even in the absence of MA, reflecting the activation of the brainstem arousing systems. In this study, we examined the evolution over-time of EEG and HR activities during PLM, a pathological arousing stimulus providing a naturally occurring condition in which to test the hierarchical hypothesis of arousal response. The PLM was chosen for its ability to control for confounding variables such as hypoxia, hypercapnia and intrathoracic effort present in obstructive apnea. Moreover, during PLM, the effect of our categorization may be easily assessed by comparing the temporal variation in EEG and HR taking place when MA or bursts in synchronized activity occurs, with those taking place in an equivalent time period when MA is not present. The main purpose of the study was to ascertain whether even in PLM a specific and stereotyped pattern of EEG and autonomic variation occurs before and during the PLM, independent of the electrophysiological signs of cortical activation, and confirming the existence of a common arousal system governing the autonomic and cerebral arousal response.

2. Methods 2.1. Patient population Twelve patients (7 men and 5 women, mean age: 47.0 ^ 2.5 years, range: 31–63), diagnosed with primary restless leg syndrome (RLS) (n ¼ 8) or PLM (n ¼ 4), were included in the study. The patients fulfilled the clinical criteria for the diagnosis of RLS (Walters, 1995) or PLM (Coleman, 1982) according to standard criteria. Conditions known to be associated with PLM such as diabetes, renal failure, peripheral neuropathy, myelopathy, and the presence of any other neurological or sleep disorder such as narcolepsy and sleep apnea syndrome were excluded by clinical interview and laboratory findings. No patient had been taking medication that might affect the central nervous system or the cardiovascular system, and none was treated for cardiopathy, RLS or PLM syndrome. The most frequent complaints were unrefreshing nocturnal sleep, fatigue and insomnia. The mean Epworth Sleepiness Scale (ESS) score was 10.4 ^ 1.5, range: 4–19, with subjective pathological sleepiness (ESS . 11) in 4 patients.

All patients underwent two consecutive nights of recording. The first night was used to rule out other sleep disorders and the second night was the polygraphic recording used for analyses. On both nights, recordings took place between 22:00 and 07:00 hrs. Each patient had more than 10 PLMs per hour of sleep and none had an index of apneas and hypopneas .10 per hour of sleep (mean AHI: 5.7 ^ 2.1). Written informed consent was obtained from all patients before they participated in the study. 2.2. Nocturnal sleep studies Sleep was monitored using 3 EEG leads (C3–A2, C4–A1, PZ–O2), right and left electro-oculogram and chin electromyogram. In order to assess apneas and hypopneas, nasal and oral airflows were recorded with thermistors, and thoracic and abdominal respiratory movements with strain gauges. Oxygen saturation (SaO2) was measured continuously with a finger oxymeter. Tibialis electromyographic activity (EMG) was monitored using surface electrodes placed on the lower third of the right and left legs. The EMG signal was recorded at a time-constant of 0.3 s and a high band-pass filter setting of 90 Hz. A 50 mV sinusoidal calibration signal of approximately 1 min duration was obtained in all subjects at the start of monitoring. Electrode impedance was below 5000 ohms at the beginning of the recording. The quality of the EMG recording was ascertained by asking the patient to flex his knees and feet. An electrocardiogram (ECG) was recorded from a standard D2 lead. Sleep was scored according to the standard criteria using 20 s epochs (Rechtschaffen and Kales, 1968). Microarousal was defined according to ASDA criteria (ASDA, 1992) as a return to alpha or fast frequency, well differentiated from the background EEG activity. The duration was, however, extended to include MA lasting $1.5 s and ,3 s (Martin et al., 1997). MA detection criteria for rapid eye movement (REM) sleep included an increase in submental EMG amplitude, in addition to a shift in EEG activity. Recordings were analyzed with the ERA q software package (Phitools w, Grenoble, France) for polysomnography and spectral analysis. 2.3. Data analysis The procedure employed for the time-dependent analysis of EEG and ECG activity has been previously described (Sforza et al., 1999, 2000) and is outlined below. 2.4. EMG activity analysis The PLM was scored using Coleman’s criteria (Coleman, 1982) i.e. movement lasting 0.5–5 s with inter-movement intervals of 4–90 s occurring in series of at least 4 consecutive movements, and associated with the stage in which they occurred. PLM was categorized into 3 types (Fig. 1) accord-

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891

ing to the following criteria: 1. A leg movement was considered to be associated with MA (PLM with MA) (Fig. 1, upper panel) if the latter

885

occurred simultaneously or within 1 s after the onset of tibialis muscle EMG activity (Mendelson, 1996). 2. PLM not associated with any change in the EEG activity in all EEG leads and not associated with MA was classi-

Fig. 1. Polygraphic recordings of a leg movement ending in a visible MA (upper panel), one without MA (middle panel), and one with concomitant slow wave activity (bottom panel).

886

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891

fied as PLM without MA (PLM without MA) (Fig. 1, middle panel). 3. PLM was defined as PLM associated with synchronized EEG activity (PLM with slow activity) (Fig. 1, bottom panel) when a burst of delta activity or K-complex occurred simultaneously or within 1 s following the leg movement onset. Delta bursts were defined as a sequence of delta waves, exceeding by at least one-third the amplitude of background activity detectable on all EEG derivations (Parrino et al., 1998). K-bursts were defined as a sequence of two or more K-complexes without alpha activity, detectable on all EEG derivations. All types of PLM were determined with all polygraphic data to define as accurately as possible the start of the event. To optimize the detection of the PLM onset, any increase in the amplitude of the EMG signal (even when the amplitude was less than 20%), was considered as the beginning of the motor phenomenon. For each type of PLM, the number, index (number of PLMs per hour of sleep), and mean duration were computed. 2.5. EEG spectral analysis To analyze the temporal EEG changes occurring in association with PLM, we measured EEG spectra bands for 10 s before the PLM onset and for 10 s after the onset of the motor event. Fast Fourier transform (FFT) was performed on the PZ–O2 lead in all patients. The EEG power spectra were computed for 1 s non-overlapping windows using a Hanning window. Seven frequency bands were defined: delta (1–4 Hz), theta (4.5–7.5 Hz), alpha (8–11 Hz), sigma (11.5–15 Hz), beta1 (15.5–18 Hz), beta2 (18.5–35 Hz), and gamma (35.5–45 Hz). For each band, the power within each 1 s window was then normalized by expressing its value as percent change from the mean value before the PLM onset. This method was used in order to evaluate the magnitude of potential changes within a specific frequency band over time, and to compare, albeit individual differences, the changes in activity to that of other frequency bands. Analysis was done in stage 2 non-rapid eye movement (NREM) sleep for PLM separated for a minimum of 10 s after rejection of PLM showing EEG and/or ECG artifacts and discarding PLM followed by awakenings. After rejection, 85% of the total scored PLM was examined. 2.6. Heart rate analysis The cardiac effect of PLM was quantified for each PLM type throughout the sleep study according to a method previously described (Sforza et al., 2000). Briefly, HR values were measured during 10 s prior to PLM onset, and the event-related HR fluctuations were calculated during 10 s after PLM onset. QRS peaks were detected, and then the HR was calculated directly from the R-R interval. The over-time pattern of HR response was measured by examining the HR response before and after PLM onset.

Measurements of HR were normalized by subtracting from each HR value before and after the onset of PLM the mean value obtained over the beats preceding the onset of the motor event. Tachycardia was defined as an increase of at least 30% of the baseline values. 2.7. Statistical analyses To assess any differences in the EEG and ECG activities among the 3 types of PLM, a two-way analysis of variance (ANOVA) for repeated measures was computed with PLM type as the between factor and time (10 s before and 10 s after PLM onset) as the repeated measure. For comparisons that reached significance, post-hoc analysis was performed using the Student–Newman–Keuls method. Bonferroni’s corrections were made for multiple comparison and differences were considered significant if they had values of P , 0:01 after the correction. All statistical analyses were performed with the SPSS statistical software package (SPSS for Windows, 9.0, SPSS Inc, Chicago). Results in the text and in Tables are presented as means ^ SEM.

3. Results 3.1. Polygraphic data Table 1 summarizes the sleep and polygraphic variables among patients. The average number of PLM was 320.2 ^ 41.3 and the total number of PLM detected was 5054. Overall, 59.6% of movements were associated with MA, 35.6% were characTable 1 Sleep and polygraphic parameters in the study group (mean ^ SEM) a

Total sleep time (min) Sleep efficiency (%) WASO (min) Awakenings (n) Stage 1 (%) Stage 2 (%) Stages 3–4 (%) Stage REM (%) PLM index (n/h) PLM with MA index (n/h) PLM with MA duration (s) PLM without MA index (n/h) PLM without MA duration (s) PLM with slow activity index (n/h) PLM with slow activity duration (s)

Mean

SEM

Range

420.9 78.7 117.2 109.7 19.0 51.3 7.6 22.0 45.6 27.3 2.9 17.3 2.4 2.64 2.8

^ 35.6 ^ 5.1 ^ 25.6 ^ 25.1 ^ 1.6 ^ 2.6 ^ 1.5 ^ 1.9 ^ 5.0 ^ 0.99 ^ 0.02 ^ 0.77 ^ 0.03 ^ 0.48 ^ 0.06

213–509 40–96 22–333 22–324 11–28 34–63 1–16 6.7 20–87 14.1–55.6 2–4 4.4–26.4 2–3 0.4–9.4 2–4

a WASO, wake after sleep onset; PLM index, periodic leg movements index (number of periodic leg movements/hour of sleep); PLM with MA, PLM associated with microarousal; PLM without MA, PLM not associated with microarousal; PLM with slow activity, PLM associated with burst of delta waves and/or K-complexes.

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891 Table 2 Polygraphic characteristics of PLMs during different sleep stages a Stage 1 Stage 2 Stages 3–4 Stage REM 26.5 PLM with MA (%) PLM with MA duration (s) 2.9 PLM without MA (%) 3.4 PLM without MA duration (s) 2.1 – PLM with slow activity (%) PLM with slow activity duration (s) – a

64.6 2.9 71.8 2.4 90 2.6

0.46 4.0 18.3 2.9 10 2.5

8.5 2.3 6.4 2.0 – –

See Table 1.

terized by no changes in EEG activity, and 4.8% were defined as PLM with slow activity. In terms of movement duration, there were no significant differences between types, PLM with MA lasting 2.9 ^ 0.02 s, PLM without MA 2.4 ^ 0.03 s and PLM with slow activity 2.8 ^ 0.06 s. More PLMs were associated with MA during stages 1 (26.5%) and 2 (64.6%) of NREM sleep and fewer during stages 3 and 4 of NREM sleep and during REM sleep (Table 2). PLM without MA occurred more frequently in stage 2 and deep sleep. PLM with MA and without MA tended to be longer in stages 3–4 of NREM sleep than during light NREM sleep and REM sleep (Table 2). 3.2. Spectral EEG analysis Data in Fig. 2 show the mean profiles of EEG spectra in delta, theta, and alpha activities 10 s before and 10 s after PLM onset for the 3 types of PLM. For all types of PLM, the pattern of EEG activity was remarkably stable up to window 1 before the PLM onset, at which point all PLMs displayed a clear EEG change. This was greater for delta band, showing a substantially significant rise 1 s before the PLM onset. At the time of PLM onset, i.e. windows 1–3 s after the PLM onset, delta, theta, alpha and beta2 activities showed a significant increase in power peaking in the window 2 s after the onset. In the post-PLM period, all frequency bands tended progressively to return to baseline values, but there was a sustained reduction in power compared to baseline, particularly for alpha and delta power. ANOVA revealed a significant mean effect of time for the delta (F ¼ 28:1, P , 0:0001), theta (F ¼ 11:8, P , 0:001), alpha (F ¼ 11:4, P , 0:0001), sigma (F ¼ 7:3, P , 0:0001), and beta2 (F ¼ 9:03, P , 0:0001) activities. Tests of simple main effect showed that in the delta band more activation started in a window 1 s before the onset of PLM (P ¼ 0:003) and remained until the third window after the onset (P , 0:0001). Theta activity rose significantly in the window 1 s after PLM onset (P ¼ 0:0001) associated with a concomitant rise in alpha activity (P ¼ 0:0001), persisting in the window 3 s after PLM onset. For fast frequencies, sigma and beta2, the EEG power rose in the window 1 s after PLM onset, and the increase persisted for a further 2–3 s.

887

Although the pattern of EEG changes was similar, the delta (F ¼ 3:34, P , 0:0001), alpha (F ¼ 2:61, P , 0:0001) and beta2 powers (F ¼ 1:84, P ¼ 0:002) showed significant time £ type interaction, indicating a greater increase in slow activity when PLM was associated with a burst of slow wave activity, and a still greater rise in fast frequencies when MA occurred. No event type or time £ event type interaction was present for theta, beta1 and gamma activities. After 4 s in the post-PLM period, the values of all EEG bands tended to decrease to near baseline, with values significantly below the baseline levels for the delta (P ¼ 0:0001) and alpha powers (P ¼ 0:0001). 3.3. Heart rate analysis HR variation for the 3 PLM types during the 20 s analysis period is illustrated in Fig. 3. A two-way ANOVA with time and type factors revealed a significant effect for the factor time (F ¼ 96:1, P , 0:0001) and type (F ¼ 6:18, P ¼ 0:005). For each PLM, the HR showed a typical pattern of tachycardia–bradycardia, indicating an activation across the motor event followed by an inhibitory effect. This pattern started 1 s before the PLM onset and reached significance during the motor phenomenon. Although not significant, in the pre-PLM onset period, a trend to progressive rise was noted 3–1 s before the onset of the PLM, greater at 1 s. Thereafter, a significant tachycardia was recorded from the first to the sixth beat, peaking at 3 and 4 s. This cardioacceleration was greater when MA (P , 0:001) and slow burst (P , 0:01) were associated with the PLM. Subsequently, the HR values decreased to below baseline and remained significantly (P , 0:0001) lower compared to pre-PLM values, without difference between PLM types. 4. Discussion The main goal of this study was to characterize the pattern of temporal changes in EEG and ECG activities during PLM, and two major findings were obtained. First, PLM without MA and with slow activity is associated with EEG and cardiac changes similar to those in the presence of MA. Second, irrespective of the type of PLM, there is a continuum in EEG and cardiac responses consisting of an initial increase in delta activity and HR recorded before the onset of the PLM, and followed during the PLM by a gradual increase first in theta and alpha activity and, thereafter, in beta activity. The combined changes in cardiac tone and EEG activity before the onset of the motor phenomena may be considered as the first level of transient activation from sleep in the continuum of the human arousal response. Similar to previous reports (Sforza et al., 2000; BuSha et al., 2001), the most important observation of this study was that despite the amplitude of the arousal response was greater when PLM was associated with MA, all types of PLM elicited a complex and stereotyped variation in cortical and cardiac activity which was independent of the

888

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891

Fig. 2. Overtime spectral EEG variations before (b) and after (a) the PLM onset. The lines show the percentage changes in mV 2 (mean ^ SEM) in 3 spectral EEG bands for PLM with MA, PLM without MA, and PLM with slow wave activity. The onset of PLM (indicated by the arrow) is associated with an increase in delta power just before the onset of the motor phenomena, followed thereafter by an increase in theta and alpha bands.

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891

889

Fig. 3. Distribution of changes in mean (^SEM) HR for each heart beat before (b) and during (a) the PLM for PLM with MA, PLM without MA and PLM with slow wave activity. Each point represents the average value of each HR minus the mean value of the pre-movement HR values considered as the reference point. A similar trend in HR rise was observed during all types of the PLM, with difference between types achieving significance for the third, fourth and fifth heart beats (P , 0:0001). The rise in HR started before the PLM onset (arrow) with greater values one beat before the onset.

presence of MA. Irrespective of the type of PLM, the pattern of arousal response was consistently similar, consisting of an abrupt increase in delta activity and HR near the onset of the PLM, followed by a progressive rise in fast EEG frequencies and a tachycardia 1–3 s after the onset. Thereafter, there was a decline in HR and spectral EEG amplitude with values below the baseline level. The magnitude and temporal variation of cerebral and autonomic responses before and during the PLM support the hypothesis of a continuum in arousal response present not only at visual analysis but also at spectral level. When we consider visual scoring, the continuum would be translated by the progression of the arousal response from no changes in the EEG activity, to delta and K-complex bursts, and finally to MA and full awakening (Halasz, 1998; Sforza et al., 2000). Spectrally the same continuum is present, starting with an increase in delta activity (Sforza et al., 2000) and heart and respiratory rates (Quattrocchi et al., 2000; BuSha et al., 2001), and followed by a transition to fast EEG activities and tachycardia, whether cortical desynchronization occurs or not. Other arousing stimuli related to sleep disorders caused fluctuations in HR and cerebral activity similar to those described herein. In the upper airway resistance syndrome (Black et al., 2000), the esophageal

pressure (Pes) reversal is associated with an increase in delta and thereafter in fast activities occurring 2 s before the Pes reversal and present even when MA did not occur. Similar results were obtained in patients with SAS in whom delta amplitude progressively rose throughout the apnea and continued to increase simultaneously with the appearance of alpha and fast activities just before the start of the hyperventilation period (Svanborg and Guilleminault, 1996). Taken together, these observations suggest that irrespective of the type of arousing stimulus, either spontaneous or related to respiratory or motor phenomena, there is a stereotyped pattern of EEG and cardiac activation starting before the onset of the arousing stimulus and characterized by an initial rise in slow EEG frequency band and HR. The close link between cardiac and EEG changes during arousals indicates that the generation of all variation in cerebral and cardiac activity shares a similar pattern of neural activation, implicating a common generator located at the brainstem level. According to the theory of a common brainstem network (Lambertz and Langhorst, 1998; Langhorst et al., 1975), the respiratory, vasomotor, and cardiac systems receive synaptic inputs from a common lower brainstem circuit, with a discharge pattern synchronized to slow EEG activity. Whether induced by internal or external

890

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891

stimuli, there is an increase in the firing rate of this system that can disinhibit arousal and autonomic systems yielding, as a survival defensive response, abrupt changes in cardiac and cerebral activities. The clinical significance of these results is not certain, but variations in EEG and ECG activities during PLM lead us to two further conclusions. First, the analysis of ‘cortical arousal’ as proposed by ASDA criteria may not allow the proper insight into the real sleep fragmentation occurring in patients with PLM, explaining some of the poor correlations seen between measures of sleep fragmentation (e.g. PLM with MA and arousal index) and daytime sleepiness (Coleman et al., 1982; Mendelson, 1996). Recurrent arousals requiring only cardiac or spectral EEG activation may be equally disruptive to sleep architecture, resulting in abnormalities of attention, alertness and performance. Therefore, studies investigating the consequences of sleep fragmentation on daytime function should include more sophisticated EEG and autonomic analyses in patients with sleep disorders. Second, since a stereotyped pattern of variation in HR and EEG power was present before and during the PLM, it is tempting to propose that there exist oscillatory processes in autonomic and EEG activities that adjust in anticipation to the motor phenomenon. Although we cannot confirm this hypothesis on the basis of our data, several investigations indirectly support this view. We know that sleep exerts tight control on motor and autonomic activity, and therefore if the sleep state is unstable, motor pattern and autonomic activity will also be unstable. When analyzed in the EEG frequency domain, these fluctuations correspond to the two components of the oscillatory sleep process called cyclic alternating pattern (CAP) (Terzano et al., 1985; Terzano and Parrino, 1993) during which oscillations in HR may also be found (Ferri et al., 2000). Moreover, previous studies have shown that PLM tended to occur most often during phase A of the CAP (Parrino et al., 1996), and periodic EEG changes occur at a rate similar to that of the PLM both during wakefulness (Montplaisir et al., 1994) and sleep (El-Ad and Chervin, 2000). According to this hypothesis, the PLM would be considered not simply a motor phenomenon occurring periodically and inducing sleep fragmentation (Karadeniz et al., 2000), but as the expression of an underlying arousal disorder related to dysfunction of the oscillatory neural networks regulating the cyclic arousability of the sleepy brain. Even though the hypothesis of the CAP may unify the sleep oscillatory processes and the hierarchy of the arousal response, a still unanswered question is whether the arousal response in slow activity, i.e. phase A1 of CAP, translates a mechanism maintaining sleep (Parrino et al., 2001) or whether it indicates a defensive response to arousing stimuli. Since all arousing stimuli either spontaneous (Sforza et al., 2000) or induced by respiratory phenomena (Black et al., 2000) always induce an initial increase in delta power and HR followed by an inhibition in the EEG and HR activities, we favor the hypothesis that the arousal response in slow wave activity may represent the

primary aspect of the brain arousability during sleep, induced by lower intensity stimuli and implicating the activation of brainstem areas. Some methodological aspects of the study require further explanation. The first and most important point is to ascertain if any muscular activity were present before the onset of PLM, inducing the initial change in EEG and ECG signals. In order to optimize the results of the spectral analysis, we consider as the onset of the movement any increase in the signal amplitude in both muscles. We adopted this method in preference to an increase of 20% of the signal baseline value (Coleman, 1982) to reduce the possibility that the described EEG changes would be related to initial contraction of the muscles. Thus we believe that the increase in delta activity and in HR before the onset of the PLM is a reasonable marker of change in arousal level during sleep. Second, the low rate of PLM associated with slow wave activity may suggest that the association of the leg movement with delta and Kcomplex bursts is largely coincidental. Against this possibility are previous observations (Droste et al., 1996; Parrino et al., 1996; El-Ad and Chervin, 2000) showing occurrence of PLM during periods of sleep characterized by greater EEG synchronization, i.e. K-complex and delta bursts. Therefore, we do not believe that the occurrence of PLM with slow wave activity consists simply of a random association. Finally, this study was performed in only 12 patients and thus the ability to detect type of arousal response was low. However, the main thrusts of the study were not to quantify differences in the frequency of the PLM type, but to identify whether a common response in EEG and in cardiac rate was present. In addition, although the arousal effect of the PLM type might be found to be more robust with a larger sample size, the synchronous variations in cardiac and cerebral activity occurred in every subject in each PLM type and were strikingly similar to those found during spontaneous arousal. In summary, we have shown that during PLM, increases in delta activity and HR occur synchronously before the onset of the motor event, followed by a rise in fast EEG activities and HR whether or not the PLM was associated with MA. These findings support the hypothesis that all arousals from sleep, regardless of whether they occurred spontaneously or were evoked by a pathological event, consistently produced a primary acute change in autonomic and slow EEG activities that reflects the transient activation of neural brainstem areas. Further studies, including autonomic arousals and arousals in slow wave activity, need to be performed to establish the relative contribution of this model to the diurnal symptoms of fatigue, tiredness and sleepiness in patients with sleep disorders.

References American Sleep Disorders Association. EEG arousal: scoring rules and examples. Sleep 1992;15:174–184. Black JE, Guilleminault C, Colrain IM, Carrillo O. Upper airway resistance

E. Sforza et al. / Clinical Neurophysiology 113 (2002) 883–891 syndrome. Central electroencephalographic power and changes in breathing effort. Am J Respir Crit Care Med 2000;162:406–411. BuSha B, Leiter JC, Curran AK, Li A, Nattie EE, Damall RA. Spontaneous arousals during quiet sleep in piglets: a visual and wavelet-based analysis. Sleep 2001;24:499–513. Carley DW, Applebaum R, Basner RC, Onal E, Lopata M. Respiratory and electrocortical responses to acoustic stimulation. Sleep 1996;19:S189– S192. Coleman RM. Periodic leg movements in sleep (nocturnal myoclonus) and restless legs syndrome. In: Guilleminault C, editor. Sleeping and waking disorders, Menlo Park, CA: Addison-Wesley, 1982. pp. 265– 295. Coleman RM, Bliwise DL, Sajben N, Boomkamp A, De Bruyn LM, Dement WC. Daytime sleepiness in patients with periodic movements in sleep. Sleep 1982;5(Suppl. 2):S191–S202. Davies RJO, Belt PJ, Roberts SI, Stradling JR. Arterial blood pressure responses to graded transient arousal from sleep in normal subjects. J Appl Physiol 1993;74:1123–1130. Droste DW, Krauss JK, Hagedorn G, Kaps M. Periodic leg movements are part of the B-wave rhythm and the cyclic alternating pattern. Acta Neurol Scand 1996;94:347–352. El-Ad B, Chervin RD. The case of a missing PLM. Sleep 2000;23:450–451. Ferri R, Parrino L, Smerieri A, Terzano MG, Elia M, Musumeci SA, Pettinato S. Cyclic alternating pattern and spectral analysis of heart rate variability during normal sleep. J Sleep Res 2000;9:13–18. Halasz P. Hierarchy of micro-arousals and the microstructure of sleep. Neurophysiol Clin 1998;28:461–475. Karadeniz D, Ondze B, Besset A, Billiard M. EEG arousals and awakenings in relation with periodic leg movements during sleep. J Sleep Res 2000;9:273–277. Kingshott RN, Engleman HM, Deary IJ, Douglas NJ. Does arousal frequency predict daytime function? Eur Respir J 1998;12:1264–1270. Lambertz M, Langhorst P. Simultaneous changes of rhythmic organization in brainstem neurons, respiration, cardiovascular system and EEG between 0.05 and 0.5 Hz. J Auton Nerv Syst 1998;68:58–77. Langhorst P, Stroh-Werz M, Dittmar K, Camerer H. Facultative coupling of reticular neuronal activity with peripheral cardiovascular and central cortical rhythms. Brain Res 1975;87:407–418. Martin SE, Engleman HM, Kingshott RN, Douglas NJ. Microarousals in patients with sleep apnoea/hypopnoea syndrome. J Sleep Res 1997;6:276–280. McNamara F, Wulbrand H, Thach BT. Habituation of the infant arousal response. Sleep 1999;22:320–326. Mendelson WB. Are periodic leg movements associated with clinical sleep disturbance? Sleep 1996;19:219–223.

891

Montplaisir J, Lapierre O, Lavigne G. Le syndrome d’impatiences musculaires: une maladie associe´ e au ralentissement pe´ riodique ou ape´ riodique de l’EEG. Neurophysiol Clin 1994;24:131–140. Parrino L, Boselli M, Buccino GP, Spaggiari MC, Di Giovanni G, Terzano MG. The cyclic alternating pattern plays a gate-control on peripheral limb movements during non-rapid eye movement sleep. J Clin Neurophysiol 1996;13:314–323. Parrino L, Boselli M, Spaggiari MC, Smerieri A, Terzano MG. Cyclic alternating pattern (CAP) in normal sleep: polysomnographic parameters in different age groups. Electroenceph clin Neurophysiol 1998;107:439–450. Parrino L, Smerieri A, Rossi M, Terzano MG. Relationship of slow and rapid EEG components of CAP to ASDA arousals in normal sleep. Sleep 2001;24:881–885. Quattrocchi JJ, Shapiro J, Verrier RL, Hobson JA. Transient cardiorespiratory events during NREM sleep: a feline model for human microarousals. J Sleep Res 2000;9:185–191. Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring for sleep stages in human adults, Los Angeles, CA: Brain Information Service, Brain Research Institute, University of California, 1968. Roehrs T, Zorick F, Wittig R, Conway W, Roth T. Predictors of objective level of daytime sleepiness in patients with sleep-related breathing disorders. Chest 1989;95:1202–1206. Rosenthal L, Roehrs T, Sicklesteel J, Zorick F, Wittig R, Roth T. Periodic movements during sleep, sleep fragmentation and sleep–wake complaints. Sleep 1984;7:326–330. Saskin P, Moldofsky H, Lue FA. Periodic movements in sleep and sleep– wake complaint. Sleep 1985;8:319–324. Sforza E, Nicolas A, Lavigne G, Gosselin A, Petit D, Montplaisir J. EEG and cardiac activation during periodic leg movements in sleep. Support for a hierarchy of arousal responses. Neurology 1999;52:786–791. Sforza E, Juony C, Ibanez V. Cardiac activation during arousals in humans: further evidence for hierarchy in the arousal response. Clin Neurophysiol 2000;111:1611–1619. Svanborg E, Guilleminault C. EEG frequency changes during sleep apneas. Sleep 1996;19:248–254. Terzano MG, Mancia D, Salati MR, Costani G, Decembrino A, Parrino L. The cyclic alternating pattern as a physiological component of normal NREM sleep. Sleep 1985;8:137–145. Terzano MG, Parrino L. Clinical application of cyclic alternating pattern. Physiol Behav 1993;54:807–813. Walters AS, The International Restless Legs Syndrome Study Group. Toward a better definition of the restless legs syndrome. Mov Disord 1995;10:634–642.