Polysomnographic study of nocturnal sleep in idiopathic hypersomnia without long sleep time



Giuseppe Plazzi, MD, Department of Neurological Sciences, University of Bologna, via Ugo Foscolo, 7, Bologna, Italy. Tel.: +39-51-2092926;
fax: +39-51-2092963; e-mail: giuseppe.plazzi@unibo.it


We investigated nocturnal sleep abnormalities in 19 patients with idiopathic hypersomnia without long sleep time (IH) in comparison with two age- and sex- matched control groups of 13 normal subjects (C) and of 17 patients with narcolepsy with cataplexy (NC), the latter considered as the extreme of excessive daytime sleepiness (EDS). Sleep macro- and micro- (i.e. cyclic alternating pattern, CAP) structure as well as quantitative analysis of EEG, of periodic leg movements during sleep (PLMS), and of muscle tone during REM sleep were compared across groups. IH and NC patients slept more than C subjects, but IH showed the highest levels of sleep fragmentation (e.g. awakenings), associated with a CAP rate higher than NC during lighter sleep stages and lower than C during slow wave sleep respectively, and with the highest relative amount of A3 and the lowest of A1 subtypes. IH showed a delta power in between C and NC groups, whereas muscle tone and PLMS had normal characteristics. A peculiar profile of microstructural sleep abnormalities may contribute to sleep fragmentation and, possibly, EDS in IH.


Idiopathic hypersomnia (IH) is classified among the hypersomnias of central origin (CNS hypersomnias), a group of syndromes lumped together by the presence of excessive daytime sleepiness (EDS), in the absence of disturbed nocturnal sleep or circadian rhythm sleep disorders (AASM, 2005). CNS hypersomnias encompass conditions in which an altered functioning of the central nervous system, including unhealthy behavior (sleep deprivation), causes EDS. However, while all the different CNS hypersomnias are characterized by specific clinical features or can be objectively confirmed by neurophysiological or laboratory investigations, IH is defined by the exclusion of any other (known) cause of EDS (AASM, 2005; Billiard, 2007). Despite the current distinction into two subtypes based on the length of the nocturnal sleep period, patients with IH can be regarded as ‘normal’ subjects suffering from EDS.

The early descriptions of IH by Bedrich Roth highlighted as a clinical telltale the occurrence of ‘sleep drunkenness’ together with a ‘deep and prolonged sleep’, confirmed by polysomnographic evidence of normal sleep features, with the frequent impossibility to objectively replicate the abnormal prolonged duration of the major sleep period in the laboratory setting (Roth, 1981; Roth et al., 1972). Accordingly, Roth stated that the ‘exact borderline between clinical hypersomnia and normal people who ‘like to sleep’ may be difficult to determine’ and considered IH as ‘an extension and ‘intensification’ of normal sleep’ (Roth et al., 1972).

In the Multiple Sleep Latency Test (MSLT) era, a short sleep latency became the diagnostic criterion for EDS in general, and the occurrence of sleep onset REM sleep periods the diagnostic hallmark of narcolepsy (AASM, 2001). Afterwards, IH was repeatedly investigated focusing on its clinical and neurophysiological differences from narcolepsy, the occurrence of REM sleep-related phenomena was described in a subgroup of IH patients (Bassetti and Aldrich, 1997), thus merging this clinical population in a continuum ranging from a ‘slow wave’ (Berti Ceroni et al., 1967) to a ‘REM’ (Dement et al., 1966) sleep type of ‘narcolepsy’.

The increasing technological advances together with the description of other clinical conditions (e.g. behaviorally induced insufficient sleep syndrome, upper airway resistance syndrome) has progressively curtailed the proportion of patients in which the EDS remains of ‘idiopathic’ nature. This has been mirrored by the subsequent definitions of IH in the consecutive international classifications of sleep disorders (AASM, 2001, 2005), as well as by the different diagnostic criteria applied to the study of its clinical and neurophysiological borders (Anderson et al., 2007; Baker et al., 1986; Bassetti and Aldrich, 1997; Billiard et al., 1998; Lavault et al., 2011; Sforza et al., 2000; Takei et al., 2012; Vanková et al., 2001; Vernet and Arnulf, 2009). Nevertheless, to date we ignore not only the pathophysiological origin but also the neurophysiological features of IH.

The aim of this study was to perform an in depth investigation of nocturnal sleep recordings of patients with IH in comparison to patients with narcolepsy with cataplexy (NC) and to normal subjects, in order to fully characterize the nocturnal sleep by means of the application of modern polysomnographic analyses of subjects fulfilling the latest nosographic criteria (AASM, 2005).

Subjects and Methods

This study was approved by the Local Ethical Committee and all subjects signed a written informed consent.

Patients and control subjects

Nineteen consecutive patients (six females and 13 males, mean age 46.0 years, 12.75 SD) with IH attending the Department of Neurological Sciences University of Bologna, Bologna, Italy and the Department of Neurology, Oasi Institute IRCCS, Troina, Italy were recruited for this study.

Our diagnostic procedure included a clinical evaluation by a physician expert in sleep medicine with systematic assessment of sleep symptoms and habits (including the Italian version of the Epworth Sleepiness Scale - ESS) (Vignatelli et al., 2003) in the setting of our Outpatients Clinic for Narcolepsy. In this context, a careful history of habitual sleep pattern during weekdays and weekends (or holidays) was collected. All subjects with a clinical suspicion of inadequate sleep hygiene (in the absence of symptoms suggesting other sleep disorders) were asked to extend nocturnal sleep for 3 weeks and monitored by means of sleep log or actigraphy before being re-evaluated and admitted to hospital only if the daytime sleepiness complaint (i.e. ESS score ≥ 10) persisted after sleep extension. Similarly, patients under antidepressant/psychoactive treatment were considered as having psychiatric hypersomnia, otherwise drug treatment was discontinued after a psychiatric evaluation excluding significant psychiatric comorbidity and the patient was subsequently re-evaluated.

During hospitalization the examinations were performed in drug-free condition (drug-naïve patients or after drug discontinuation for at least 3 weeks), including two nocturnal polysomnographic recordings for two consecutive nights (the first for adaptation, and the second for diagnostic purposes), followed by a clinical MSLT with five nap opportunities in the third day (Littner et al., 2005).

Per ICSD-2 criteria, all patients with sleep-disordered breathing (apnea-hypopnea index, AHI, > 5) and periodic limb movements during sleep (periodic limb movement index, PLMI, > 15) were excluded, except for those in whom a diagnosis of NC was confirmed by low/absent CSF hypocretin-1 despite high PLMI (AASM, 2005).

Selected patients fulfilled the International Classification of Sleep Disorders – 2nd Edition (AASM, 2005) for IH without long sleep time. In particular, these patients had a daily complaint of EDS lasting since a long time (>3 months) with 47% of them reporting sleep drunkenness at morning awakening, their reported nocturnal sleep length was 6–10 h, confirmed by routine clinical nocturnal polysomnography that also did not show abnormalities that could cause daytime sleepiness (in particular, apnea/hypopnea index <5 per h and periodic leg movements during sleep (PLMS) index <15 per h), and the MSLT showed a mean sleep latency <8 min in all patients. Additionally, 16% and 11% of the patients reported history of sleep paralysis and hypnagogic hallucinations, respectively. Table 1 reports the clinical details of these patients.

Table 1. Clinical features of patients with idiopathic hypersomnia (n = 19)
  1. BMI, Body mass index; EDS, excessive daytime sleepiness, MSL, multiple sleep latency test, SOREMP, sleep-onset REM periods.

Age, years46.012.75
Weight, Kg75.111.44
Height, m1.70.09
BMI (Kg m−2)25.83.51
1st symptomEDS = 18Pavor = 1
Age at EDS onset, years28.713.84
MSLT latency, min5.61.37
SOREMPs1 subject = 118 subjects = 0
HLA DQB1*06/02Yes = 2No = 6
Hypocretin (n = 9), pg ml−1329.1105.12

For the normal control group, 13 subjects were recruited (six females and seven males, mean age 44.9 years, 14.39 SD). Control subjects exclusion criteria were the presence of subjective sleep complaints (insomnia, daytime sleepiness confirmed by an ESS score <10, restless legs syndrome, REM sleep behavior disorder symptoms, sleep paralysis or hallucinations, snoring, or witnessed apnea) or the presence of other important neurological or general clinical conditions. None of the controls was under drug treatment.

Finally, we also selected a second control group formed by patients affected by NC (four females and 13 males, mean age 45.0 years, 12.94 SD). All patients with NC met the corresponding International Classification of Sleep Disorders – 2nd Edition (AASM, 2005) criteria including (i) definite cataplexy triggered by typical emotions such laughing, joking or anger; (ii) persistent daytime sleepiness; (iii) at least 2 SOREMPs and mean sleep latency <8 min at MSLT. NC patients had a mean ESS score of 17 ± 3 and a mean MSLT sleep latency of 3.3 ± 2.5 min with 3.8 ± 1.1 SOREMPs. Additionally, 71% and 65% of them had history of sleep paralysis and hallucinations, respectively, whereas 23% complained of unrefreshing morning awakenings suggesting sleep drunkenness.

Sleep recordings

Subjects slept two nights in the sleep laboratory. The first night was used for adaptation and the second night for analysis. Patients were asked to refrain from drinking caffeinated and alcohol beverages during the day prior to the sleep investigation. Sleep recording was carried out during the usual sleep time.

The recording protocol included scalp EEG (at least three channels, F3 or F4, C3 or C4 and O1 or O2, referenced to the contralateral mastoid; electrodes placed following the 10–20 International System), two eye movement channels, chin and leg electromyogram, electrocardiogram, snoring, oronasal thermistor, thoracic and abdominal respiratory effort, body position and pulse oximetry. Recordings of the second night were transformed into European data format, with indications of lights off and lights on and coded for anonymity.

Conventional sleep parameters and CAP analysis

First, sleep stages were visually scored in all recordings according to standard criteria (Rechtschaffen and Kales, 1968) by a trained rater. Cyclic alternating pattern (CAP) was scored according the criteria published by Terzano et al. (Terzano et al., 2001). CAP is a periodic EEG activity of NREM sleep characterized by sequences of cycles composed by a phase A (transient electrocortical event) and a phase B (recurring EEG background activity). Phase A activities can be classified into three subtypes. This classification is based on the reciprocal proportion of high-voltage slow-waves (EEG synchrony) and low-amplitude fast rhythms (EEG desynchrony). Subtype A1 shows a predominance of synchronized EEG activity; if present, EEG desynchrony occupies <20% of the entire A phase duration. Subtype A1 specimens include delta bursts, K-complex sequences, vertex waves, polyphasic bursts with <20% of EEG desynchrony. Subtype A2 is scored in the presence of 20% to 50% of desynchronized EEG activity, with predominance of polyphasic bursts. Subtype A3 EEG activity is predominantly rapid low-voltage rhythms with more than 50% of phase A occupied by EEG desynchrony. Subtype A3 include EEG arousals, K-alpha and polyphasic bursts (with at least 50% of EEG desynchrony). CAP cycles are based on the presence of two successive phases A and B. CAP sequences are defined as three or more A phases separated from each other by no more than 60 s. CAP rate is defined as the percentage of total NREM time occupied by CAP sequences. The remaining NREM sleep is called non-CAP (NCAP).

The following CAP parameters were measured: CAP time (temporal sum of all sequences) in NREM; CAP rate (percentage of total NREM time occupied by CAP sequences); number and duration of CAP cycles; number and duration of CAP sequences; number, duration and percentages of A phases (including the phase A subtypes); A1% (percentage of A1 number from total A phases number); A2% (percentage of A2 number from total A phases number); A3% (percentage of A3 number from total A phases number) and B phases duration. All of these variables were visually detected and their parameters measured by means of the sleep analysis software Hypnolab 1.2 (SWS Soft, Italy; software made by Raffaele Ferri).

Selection of EEG mini-epochs and power spectra computation

The EEG channel used was C3/A2 or C4/A1 of all recordings; the signal, sampled at 128 Hz and digitally prefiltered at 0.1–35 Hz, was subdivided into 2-s mini-epochs; as an example, a sleep recording lasting for 8 h was subdivided into 14 400 mini-epochs (8 h × 60 min × 60 s/2 = 14 400). Each mini-epoch was assigned to a sleep stage, based on the sleep scoring previously performed; only mini-epochs from sleep stage 2, sleep stages 3 or 4 (SWS), and REM sleep were considered in this study. Mini-epochs containing muscle or other technical artifacts were carefully excluded from the analysis.

Power spectra were calculated for each mini-epoch using the sleep analysis software Hypnolab 1.2, after Welch windowing, in order to minimize the truncation error and reduce spectral leakage by suppressing sidelobes (Press et al., 1989), by means of the Fast Fourier Transform. The power spectrum was calculated for frequencies between 0.5 and 32 Hz with a frequency step of 0.5 Hz. The average absolute and relative power for the different sleep EEG bands (delta 0.5–4 Hz; theta 4.5–7.5 Hz; alpha 8–11 Hz; sigma 11.5–15.5 Hz; beta 16–32 Hz) was obtained for sleep stage 2, SWS, and REM for each subject.

Finally, the first 6 h of sleep were subdivided into four periods of 90 min each and the same spectral analysis was performed for each period; in this way, we obtained four new averages for each sleep stage, for each subject. We choose to use fixed-length sleep periods because the definition of sleep cycles or REM cycles is not unanimous in the literature and because of their important interindividual variability.

Quantification of submentalis muscle EMG amplitude

For the computer quantitative analysis of the submentalis muscle EMG, we used an already validated automatic method (Ferri et al., 2008a, 2010). The submentalis muscle EMG signal was digitally band-pass filtered at 10–100 Hz, with a notch filter at 50 Hz and rectified. Subsequently, each REM sleep epoch included in the analysis was divided into 30 1-s mini-epochs. The average amplitude of the rectified submentalis muscle EMG signal was then obtained for each miniepoch. The values of the submentalis muscle EMG signal amplitude in each miniepoch were used to compute the percentage of values in the following 20 amplitude (amp) classes (expressed in μV): amp ≤ 1, 1 < amp ≤ 2, ….., 18 < amp ≤ 19, amp > 19. Muscle atonia is expected to be reflected by high values of the first class (amp ≤ 1) while phasic and tonic activations are expected to increase the value of the other classes. As proposed in the previous studies, an index summarizing in a single value the degree of preponderance of the first class was used in REM sleep:

REM Sleep Atonia Index = amp  1/(100−1 < amp ≤ 2).

Mathematically, this index can vary from 0 (absence of mini-epochs with amp ≤ 1), i.e. complete absence of EMG atonia, to 1 (all mini-epochs with amp ≤ 1) or stable EMG atonia in the epoch. Normal values for this index lay above 0.9, while clearly abnormal values, usually observed in a relatively large proportion of patients with REM sleep behavior disorder, do not reach 0.8 (Ferri et al., 2010).

Analysis of leg movements during sleep

Leg movements during sleep were first detected by the software Hypnolab 1.2 (SWS Soft, Italy) which allows their computer-assisted detection. With this software, the detection is performed on the rectified EMG signal from the tibialis anterior muscles, using a human-supervised automatic approach controlled by the scorer that used the international criteria. The performance of this system has been evaluated and validated (Ferri et al., 2006a), but in this study one scorer, blind to group assignment, visually edited the detections proposed by the automatic analysis, before computing the final results. In particular, the PLMS index was calculated as the number of LM included in a series of four of more, separated by more than five and <90 s, per hour of sleep.

Statistical analysis

For the statistical analysis, all comparisons between patients and controls were performed first by means of the non parametric Kruskal–Wallis anova for unpaired datasets; when a statistical significance was found at this test, it was followed by the Mann–Whitney test for unpaired datasets for the comparison of each different pair of groups (i.e., IH versus normal controls, IH versus NC, and normal controls versus NC). For the intragroup comparisons, the Friedman anova for paired datasets was used. The commercially available Statistica software package (StatSoft, Inc., 2001. Statistical data analysis software system, version 6. www.statsoft.com) was used. Differences were considered significant when they reached the P < 0.05 level.


Sleep architecture and CAP

Table 2 shows the comparison between the sleep architecture parameters found in the three groups of subjects included in this study. Patients with IH show statistically significant differences from both normal controls and NC subjects. As expected, they have increased time in bed, sleep period time and total sleep time and decreased sleep latency with respect to normal subjects, but not different from NC. Indexes of sleep fragmentation, such as number of stage shifts, number of awakenings and percentage of sleep stage 1, however, are significantly increased in patients with IH with respect to both groups of normal controls and NC patients. Finally, sleep stage 2 is similar to that of normal controls but higher than that of NC patients.

Table 2. Comparison between the sleep architecture parameters found in the groups of subjects included in this study
 a-Normal controls (n = 13)b-Idiopathic hypersomnia (n = 19)c-Narcolepsy with cataplexy (n = 17)Kruskal–Wallis anovaMann–Whitney P
MeanSDMeanSDMeanSD P a versus bb versus ca versus c
  1. TIB, Time in bed; SPT, sleep period time; TST, total sleep time; SOL, sleep onset latency; FRL, first REM latency; SS, stage shifts; AWN, awakenings number; SE, sleep efficiency; WASO, wakefulness after sleep onset; S1, S2, sleep stages 1 and 2; SWS, slow-wave sleep; REM, rapid eye movement sleep; AHI, apnea – hypopnea index; ODI, oxygen desaturation index.

TIB, min445.543.11498.863.78516.069.010.01850.03NS0.005
SPT, min418.146.20485.963.50499.963.230.0060.008NS0.002
TST, min352.848.87431.569.16414.
SOL, min19.816.558.46.419.
FRL, min58.036.8368.733.0123.942.600.0013NS0.00090.007
AHI, n/h1.151.142.831.983.592.03NS   
ODI, n/h0.920.911.671.692.871.63NS   

Also the comparison of CAP parameters, reported in Table 3, shows several significant differences with a significant decrease in total CAP rate only in NC while IH patients have higher CAP rate than NC during the lighter stages of sleep (stages 1 and 2) and lower CAP rate than normal subjects during slow-wave sleep. The relative amount of the different CAP A phase subtypes is peculiar in IH with a lower amount of A1 (characterized by prevalent slow EEG activity) and higher amounts of A2 and A3 subtypes (characterized by fast EEG activities), with respect to both normal subjects and NC patients. The same peculiarity can be seen for the duration of CAP A subtypes that are generally shorter in IH than in the other two groups of subjects.

Table 3. Comparison between the CAP parameters found in the three groups of subjects included in this study
 a-Normal controls (n = 13)b-Idiopathic hypersomnia (n = 19)c-Narcolepsy with cataplexy (n = 17)Kruskal–Wallis anovaMann–Whitney P
MeanSDMeanSDMeanSD P a versus bb versus ca versus c
  1. TIB, Time in bed; SPT, sleep period time; TST, total sleep time; SOL, sleep onset latency; FRL, first REM latency; SS, stage shifts; AWN, awakenings number; SE, sleep efficiency; WASO, wakefulness after sleep onset; S1, S2, sleep stages 1 and 2; SWS, slow-wave sleep; REM, rapid eye movement sleep.

Total CAP Rate,%34.613.8725.99.1921.713.010.04NSNS0.023
In sleep stage S124.624.6818.011.308.48.350.045NS0.016NS
In sleep stage S226.515.8826.510.5215.412.630.017NS0.0060.05
In slow-wave sleep65.624.4432.720.7237.618.420.00430.002NS0.007
A1 subtypes,%68.020.0145.514.2972.410.500.00 0010.0010.000 009NS
A2 subtypes,%15.29.3619.06.3712.06.660.028NS0.007NS
A3 subtypes,%16.812.9335.512.6315.68.220.00 0010.00030.000 018NS
A1 mean duration, s7.50.744. 0010.000 0070.000 001NS
A2 mean duration, s10. 0010.000 0080.000 0010.027
A3 mean duration, s11.72.839.31.6013.92.750.00 0010.010.000 0160.048
A1 index, number/h34.116.9216.48.0222.214.050.0090.001NSNS
A2 index, number/h7.98.537.53.883.43.170.008NS0.0030.04
A3 index, number/h7.47.5011. 0010.030.000 0050.025
B phase mean duration, s21.64.1926.72.3225.73.760.0040.0016NS0.02
CAP Cycle (A + B) duration, s30.24.5033.32.2035.14.040.025NSNS0.013
CAP sequence mean duration, s199.264.15159.246.79169.651.58NS   
Number of CAP sequences28.17.0532.49.1719.99.230.002NS0.00070.029

Sleep EEG power spectra

In agreement with the above mentioned higher amount of slow wave-containing A1 CAP subtypes, in NREM sleep (especially SWS) also the delta band of normal subjects showed a power higher than that of both groups of patients; this did not reach statistical significance in any of the comparisons shown in Fig. 1 reporting the absolute power of the bands, but was clearly significant when relative power values (percentage) were considered (Fig. 2), in particular during the first two time blocks analyzed. In Fig. 1 it can also be noticed that the other bands tended to show higher power values in both groups of patients than in normal subjects during REM sleep occurring in the second part of the night; however, statistical significance was reached only of the 3rd 90-min block. The different significant differences reported in Fig. 2 regarding the alpha, sigma and beta bands during NREM sleep essentially represent the counterbalance of the changes observed above for the delta band, while those during REM sleep confirm the trends observed in Fig. 1.

Figure 1.

Comparison between the absolute power of the different EEG spectral bands, during sleep stage 2, SWS and REM, in the three groups of subjects included in this study; values are shown as mean and SEM (whiskers).

Figure 2.

Comparison between the relative power of the different EEG spectral bands, during sleep stage 2, SWS and REM, in the three groups of subjects included in this study; values are shown as mean and SEM (whiskers).

The comparison between the total power of the delta EEG band obtained during the four 90-min blocks in the three groups of subjects included in this study is reported in Fig. 3. All groups showed a progressive and significant decline of the delta power across the night. However, this band tended to be higher in normal subjects than in the patient groups, with NC showing the smallest values; however, the difference between the groups at any time block did not reach statistical significance.

Figure 3.

Comparison between the total power of the delta EEG bands, during the four 90-min blocks, in the three groups of subjects included in this study; values are shown as mean and SEM (whiskers).

Chin EMG tone analysis

The detailed comparison between the REM sleep Atonia Index computed in the three groups of subjects included in this study is shown in Fig. 4. As already reported earlier (Ferri et al., 2008a), a significant proportion of patients with NC show clearly reduced values of the REM sleep Atonia Index, with a mean value significantly lower than both normal subjects and IH patients. No difference was found for this parameter between normal subjects and patients with IH.

Figure 4.

Comparison between the REM sleep Atonia Index computed in the three groups of subjects included in this study; individual values are shown as well as their mean (black-filled circles) and SEM (whiskers).

Analysis of the leg motor activity during sleep

Table 4 reports the comparison between the leg movement parameters found in the three groups of subjects included in this study. As expected, due to the inclusion criteria used for IH and the already known prevalence of PLMS in NC (Ferri et al., 2006b), our patients with NC show an increase in the values of most of the leg movement parameters considered in this study, with respect to both normal subjects and patients with IH (see also Fig. 5). It is interesting to note, however, that patients with IH show longer durations of PLMS and isolated leg movements than normal subjects. Finally, Fig. 6 shows the comparison between the distribution of number of PLMS per hour of sleep (first 8 h shown) in the three groups of subjects included in this study. Besides the obvious higher values found in NC patients, it is interesting to note, in this figure, that the few PLMS found in IH and normal subjects show a similar gradual decrease during the night, which recalls the gradual decrease of PLMS in restless legs syndrome (Ferri et al., 2006a); on the contrary, NC subjects show a bell-shaped distribution during NREM sleep, as already reported earlier (Ferri et al., 2006b), and gradually increasing, during REM sleep.

Figure 5.

Comparison between the distribution of inter-LM in the three groups of subjects included in this study; values are shown as mean and SEM (whiskers). The grey-shaded areas indicate the points for which a significant difference was found between patients with Narcolepsy with Cataplexy and the other two groups of subjects.

Figure 6.

Comparison between the distribution of number of PLMS per hour of sleep (first 8 h shown) in the three groups of subjects included in this study; values are shown as mean and SEM (whiskers). Asterisks indicate the points for which a significant difference was found between patients with Narcolepsy with Cataplexy and those with Idiopathic Hypersomnia.

Table 4. Comparison between the leg movement parameters found in the three groups of subjects included in this study
 a-Normal controls (n = 13)b-Idiopathic hypersomnia (n = 19)c-Narcolepsy with cataplexy (n = 17)Kruskal–Wallis anovaMann–Whitney P
MeanSDMeanSDMeanSD P a versus bb versus ca versus c
NREM sleep
Total LM, index12.813.2513.010.7736.624.600.00 070NS0.00040.0047
PLMS, index7.811.956.39.9728.626.360.00 370NS0.00120.027
Isolated LM, index5.13.706.72.878.03.22NS   
REM sleep
Total LM, index13.310.058.86.4928.814.340.00 010NS0.00 0050.0078
PLMS, index3.77.621.22.9713.911.960.00 020NS0.00020.015
Isolated LM, index9.75.537.64.1414.94.970.00 050NS0.00020.015
Total sleep
Total LM, index13.511.7012.08.1633.717.090.00 010NS0.00 0070.0023
PLMS, index7.310.325.07.2023.017.610.00 060NS0.00020.009
Isolated LM, index6.23.596.92.5110.72.350.00 070NS0.00050.0047
PLMS sequence, number4.95.593.33.1212.97.080.00 020NS0.00 0060.005
PLMS sequence duration, s27.058.7437.591.2344.366.44NS   
PLMS duration (REM), s1.41.982.91.702.50.930.04 5000.05NSNS
PLMS duration (NREM), s1.71.423.00.972.40.600.04 5000.03NSNS
Isolated LM duration (REM), s2.71.443.   
Isolated LM duration (NREM), s2.41.823.30.782.60.700.01 6000.030.01NS
Periodicity Index, total0.3970.2890.4350.3090.4440.233NS   


In this study we applied strict criteria to define the diagnosis of IH in order to fully characterize its nocturnal sleep features, in comparison to normal subjects from one side and to patients with NC on the other, as the two ideal extremes of daytime sleepiness. The combined application of advanced analyses of sleep disclosed the following main characteristics: (i) in both IH and NC patients, EDS is reflected by the need to sleep longer during the nocturnal period, but sleep quality is negatively influenced by high sleep fragmentation, especially in IH; (ii) both EDS disorders show an overall reduced CAP rate, with a peculiar profile of higher sleep instability in lighter sleep in IH than in NC and lower instability in SWS in IH than in normal subjects, and the lowest amount of A1 subtypes in IH compared to both NC and normal subjects; (iii) the progressive reduction of the CAP rate from normal subjects to IH and NC is mirrored by the declining delta power during NREM sleep in the different groups; (iv) motor control during sleep is normal in IH when evaluated by means of quantitative analyses of PLMS and muscle tone, thus differing significantly from NC. Our findings suggest that a subtle alteration of sleep microstructure may blunt the restorative function of sleep in IH, possibly underlying the occurrence of EDS.

As mentioned above, the definition of IH has significantly changed over time, across the different diagnostic criteria, and in the few studies available in the literature (Anderson et al., 2007; Baker et al., 1986; Bassetti and Aldrich, 1997; Billiard et al., 1998; Lavault et al., 2011; Sforza et al., 2000; Takei et al., 2012; Vanková et al., 2001; Vernet and Arnulf, 2009); thus a direct comparison with the findings of other authors is difficult. However, our data confirm the most commonly reported finding of increased sleep duration in IH, as compared to normal subjects (Vanková et al., 2001; Vernet and Arnulf, 2009). Considering nocturnal sleep fragmentation, we did not confirm the significant differences in terms of higher sleep efficiency than NC reported by other authors (Bassetti and Aldrich, 1997; Anderson et al., 2007; Lavault et al., 2011), but we found higher sleep stage shifts and time spent in sleep stage 1 in both hypersomniac groups that was more pronounced in IH. Concerning SWS, our study did not show significant differences between groups, whereas other authors reported that IH could have a percentage of SWS comparable to that of controls (Sforza et al., 2000), within the normal range (Roth, 1981; Roth et al., 1972), or even higher (Vernet and Arnulf, 2009). Conversely, when compared to NC, the amount of ‘deep’ NREM sleep was higher in some studies (Anderson et al., 2007; Baker et al., 1986), especially at the end of the nocturnal sleep (Anderson et al., 2007), and comparable in others (Lavault et al., 2011; Takei et al., 2012). The only work that contrasted IH with and without long sleep time disclosed that the latter patients had lower total sleep time and sleep efficiency in the absence of further sleep architecture differences, thus being closer to the control group and similar to our current results (Vernet and Arnulf, 2009).

Our study is the first analyzing CAP in IH, disclosing a peculiar profile of reduced CAP rate, with higher values than NC in light NREM sleep and lower than normal subjects in SWS, and a striking reduction of the A1 subtype amount. Two previous studies have found a reduction of CAP rate with a prominent impairment of A1 CAP subtypes in NC versus normal subjects (Ferri et al., 2005c; Terzano et al., 2006), indicating an overall reduced arousability with a lower pressure to stabilize NREM sleep and thus expanding out of REM sleep the nocturnal abnormalities of NC. Our data confirm previous findings in NC and show a further enhancement of these abnormalities in IH. We may therefore speculate that IH nocturnal sleep, characterized by the lowest A1 representation, may reflect different thalamo-cortical processes in terms of both consolidating slow-wave sleep and of preventing awakening from sleep (Ferri et al., 2005b; Parrino et al., 2012). In other words, a reduced CAP rate may be the expression of an overall reduced capacity to maintain NREM sleep and promote deepening into SWS (Parrino et al., 2012). This consideration is clinically mirrored by the opposite evidence of increased CAP rate (of all A subtypes) in patients suffering from insomnia, a disease characterized by an ‘hyperarousal’ nyctemeral state and by difficulties in initiating or maintaining sleep during the night (Terzano and Parrino, 1992). On the other hand, a reduction of the A1 CAP subtype may also reflect an inner difference in NREM sleep dynamics, potentially linking the reduced slow frontal activity of A1 in NC and IH to altered dynamics of slow-wave oscillations during sleep (Ferri et al., 2005b, 2008b). Thus, our findings of increased sleep fragmentation and CAP rate during superficial NREM sleep and the highest relative amount of CAP subtypes A3 suggest that IH could be partly due to an intrinsic weakness of sleep continuity and stability.

In order to better characterize sleep neurophysiology of IH, we analyzed EEG spectral activity in the different sleep stages and across subsequent nocturnal 90-min time blocks. The power of the EEG bands during sleep is modulated by several factors the strongest of which are the circadian process and the homeostatic mechanisms (Borbely, 1982); thus, time is crucial. The definition of sleep cycle, essentially based on the occurrence of REM sleep episodes, does not ensure the comparison of corresponding time periods when different subjects are pooled together. For this reason we have preferred to use fixed-length time blocks that we have already successfully used in previous studies (Bruni et al., 2009; Ferri et al., 2005a). We found an intermediate delta power of IH, lower than normal subjects, but higher than NC. A single study, applying different definitions (sleep cycles versus fixed 90-min periods), evaluated spectral activity overnight in IH patients and disclosed reduced delta activity in IH versus normal subjects, but with a preserved overnight decay. According to Sforza et al. (2000), the homeostatic mechanisms were preserved in IH, but showed a lower pressure. We have also observed the same overnight decay of the delta band power, similar to that of normal subjects and our data seem to be in full agreement. NREM sleep homeostasis, although less evident than in controls, is detectable in NC (Nobili et al., 1995; Poryazova et al., 2011). Khatami et al. disclosed that the decay of the slow wave activity was steeper in NC versus normal subjects for the occurrence of a less consolidated nocturnal sleep especially after the first sleep cycle (Khatami et al., 2007). We have not formally measured the slope of the decay but the visual analysis of the graphs does not allow us to confirm the above findings (Khatami et al., 2007). Therefore, when compared to NC, IH seems to have a profile of slow wave activity decay overnight more similar to normal subjects, whereas, when compared to normal subjects, IH has a global reduction of the delta power. Despite the intrinsic connection between CAP A1 and slow waves, the former promoting the latter, our study cannot determine which sleep feature is primarily altered in IH in the absence of sleep manipulations in our design (e.g. sleep deprivation challenge). However, we acknowledge that using a 90 min fixed period instead of sleep cycles for power analysis may have affected our results and limits comparison with most of published data.

Finally, for the first time we analyzed quantitatively motor control during sleep in terms of PLMS and muscle tone during REM sleep. Our results are strictly in line with the definition of IH, and confirm its ‘normal’ nature also from a quantitative standpoint. IH patients showed a progressive reduction of their few PLMS across night and thus differed significantly from NC (Ferri et al., 2006a,b). This difference was clearly evident also in the analysis of the chin muscle tone which was altered only in NC patients (Ferri et al., 2008a).

To summarize, we performed the first polysomnographic study of IH using quantitative approaches and taking advantage of two different well-defined extreme groups (normal subjects and NC patients) in order to draw the neurophysiological borders of IH. We disclosed a peculiar profile of sleep microstructure abnormalities with reduced CAP rate and low amount of A1 subtypes in IH that may be relevant from a pathophysiological standpoint and warrant further interventional (e.g. sleep deprivation) studies in order to confirm their functional relevance.


Fabio Pizza: no conflict of interests.

Raffaele Ferri: has consulted for Merck & Co., Sanofi-Aventis, and Sapio Life.

Francesca Poli: no conflict of interests.

Stefano Vandi: no conflict of interests.

Filomena I.I. Casentino: no conflict of interests.

Giuseppe Plazzi has consulted for UCB Pharma and Cephalon.