Objective markers for sleep propensity: comparison between the Multiple Sleep Latency Test and the Vigilance Algorithm Leipzig

Transcription

1 J Sleep Res. (2015) 24, Markers for sleep propensity Objective markers for sleep propensity: comparison between the Multiple Sleep Latency Test and the Vigilance Algorithm Leipzig SEBASTIAN OLBRICH 1,2,3, *, MARIE M. FISCHER 1, *, CHRISTIAN SANDER 1, ULRICH HEGERL 1, HUBERT WIRTZ 4 and ANDREA BOSSE-HENCK 4 1 Department of Psychiatry and Psychotherapy, University of Leipzig, Leipzig, Germany; 2 Integrated Research and Treatment Centre for Adiposity Diseases, University of Leipzig, Leipzig, Germany; 3 Department of Psychiatry, Psychotherapy and Psychosomatics, Hospital of Psychiatry, University of Zurich, Zurich, Switzerland; and 4 Department for Internal Medicine, Sleep Laboratory, University of Leipzig, Leipzig, Germany Keywords electroencephalogram, sleep-wake regulation, neurophysiology Correspondence Sebastian Olbrich, Clinic and Polyclinic for Psychiatry and Psychotherapy, University Leipzig, Semmelweißstraße 10, Leipzig, Germany. Tel.: ; fax: ; sebastian.olbrich@medizin.uni-leipzig. de *S.O. and M.M.F. contributed equally to this manuscript. Accepted in revised form 29 January 2015; received 21 October 2014 DOI: /jsr SUMMARY The regulation of wakefulness is important for high-order organisms. Its dysregulation is involved in the pathomechanism of several psychiatric disorders. Thus, a tool for its objective but little time-consuming assessment would be of importance. The Vigilance Algorithm Leipzig allows the objective measurement of sleep propensity, based on a single resting state electroencephalogram. To compare the Vigilance Algorithm Leipzig with the standard for objective assessment of excessive daytime sleepiness, a four-trial Multiple Sleep Latency Test in 25 healthy subjects was conducted. Between the first two trials, a 15-min, 25-channel resting electroencephalogram was recorded, and Vigilance Algorithm Leipzig was used to classify the sleep propensity (i.e. type of vigilance regulation) of each subject. The results of both methods showed significant correlations with the Epworth Sleepiness Scale (q = 0.70; q = 0.45, respectively) and correlated with each other (q = 0.54). Subjects with a stable electroencephalogram-vigilance regulation yielded significant increased sleep latencies compared with an unstable regulation (multiple sleep latency s versus s; P = 0.03). Further, Vigilance Algorithm Leipzig classifications allowed the identification of subjects with average sleep latencies <6 min with a sensitivity of 100% and a specificity of 77%. Thus, Vigilance Algorithm Leipzig provides similar information on wakefulness regulation in comparison to the much more cost- and time-consuming Multiple Sleep Latency Test. Due to its high sensitivity and specificity for large sleep propensity, Vigilance Algorithm Leipzig could be an effective and reliable alternative to the Multiple Sleep Latency Test, for example for screening purposes in large cohorts, where objective information about wakefulness regulation is needed. INTRODUCTION The regulation of wakefulness from high alertness to relaxed quiet waking to drowsiness until sleep onset is of fundamental importance for all high-order organisms. This regulation is both a stable individual trait (Tucker et al., 2007; Van Dongen et al., 2005) and dependent on a variety of state factors, such as sleep deficits, recreational drugs and psychopharmacological treatment. Among the medical disorders affecting the 450 wakefulness regulation are sleep disorders, such as narcolepsy (Guilleminault and Brooks, 2001), but also affective disorders and attention deficit hyperactivity disorder (Hegerl and Hensch, 2014). The Vigilance Algorithm Leipzig (VIGALL) has been developed to get an objective parameter of this regulation of wakefulness. Based on a 15-min multichannel electroencephalogram (EEG)-activity recorded under resting conditions, VIGALL classifies each 1-s EEG epoch into one of seven

2 Comparison between the MSLT and VIGALL 451 EEG-vigilance stages (see below). The computer-based algorithm takes into account the information of several EEG frequency bands and their spatial cortical distribution for each analysed segment (for a detailed description of the algorithm, see Hegerl et al., 2012; Olbrich et al., 2011a, 2012a,b; and VIGALL Manual VIGALL has been validated in a couple of studies: the investigations of simultaneous EEG/functional magnetic resonance imaging showed a negative correlation between EEG-vigilance levels and blood oxygen level-dependent signals in vast cortical areas (Olbrich et al., 2009). It was also shown that decreases in vigilance go along with increased levels of the parasympathetic and decreased levels of the sympathetic autonomic nervous system activity (Olbrich et al., 2011b). Further, the discriminative power of VIGALL has been proven in investigations of patients with depression (Hegerl et al., 2012) and obsessive compulsive disorder (Olbrich et al., 2012a), that both show a more stable regulation of EEG vigilance compared with healthy controls as well as in patients with cancer-related fatigue (Olbrich et al., 2011a) that in contrast show decreased levels of EEG-vigilance. As an established tool for clinical diagnostics, the Multiple Sleep Latency Test (MSLT) is the standard for the assessment of daytime sleepiness as suggested by the American Association of Sleep Medicine (AASM; Littner et al., 2005). The main objective of the MSLT is to measure the time span from lights off to the occurrence of sleep patterns, indicating sleep onset, derived through several psychophysiological recordings comprising EEG, electrooculogram (EOG), electromyogram (EMG) and electrocardiogram (ECG). The MSLT further comprises 1 night of sleep evaluation via polysomnography to exclude other sleep disorders before the MSLT measurements take place with four five attempts to fall asleep within 20 min every 2 h, starting 1 2 h after waking up (Carskadon et al., 1986). Although a matter of debate due to missing normative databases, an averaged sleep latency of less than 5 8 min during the day and/or the occurrence of at least two sleep-onset rapid eye movement periods (SOREMPs) is considered as a sign of a pathological lability of wake sleep regulation (Carskadon et al., 1986; Pizza et al., 2013; Thorpy, 1992). Due to the time- and costintensive procedure, the MSLT is not suitable to be used, for example, for screening purposes of large cohorts. Further, critics of the MSLT argue that the manual rating of the sleeping stages is prone to laboratory- and rater-dependent results (Pataka et al., 2013). Moreover, the visual rating of the MSLT relies on a very limited number of EEG-channels. Therefore, the MSLT comprises several limiting facts, for example, the low temporal resolution of 30-s segments, the neglect of spatial information of the EEG activity and the coarse stage classification (for a review of the critics, see Himanen and Hasan, 2000). An objective test for sleepiness and sleep propensity that takes into account the fast changing characteristics of neurophysiological activity during the transition from wakefulness to sleep onset, such as VIGALL does, might yield similar results to the MSLT with less need for extensive data from multiple testing. Therefore, the aim of this study was to compare the results of a four-trial MSLT with those of the vigilance classification derived through VIGALL during one session. MATERIALS AND METHODS Subjects Twenty-five healthy subjects were recruited between 2010 and The sample size was chosen to obtain >5 subjects of each EEG-vigilance regulation type (stable, slowly declining and unstable), based on the finding that each type can be found in 25 40% of subjects (Olbrich et al., 2012b). Participants had to be in the age range of years and give written informed consent for their participation. Exclusion criteria were assessed by a senior psychiatrist, and included a history of psychiatric or neurological disorders according to criteria of ICD-10, sleep disorders, usage of illegal drugs, psychopharmacological medication that impacts the EEG or wakefulness regulation, and a current physical illness that required treatment. The study was approved by the local ethics committee. All participants had to give written informed consent to the study prior to inclusion. Polysomnography and the MSLT Sleep the week prior to the MSLT and VIGALL measurement was assessed using continuous actometer recordings and sleep diaries. Subjects had to spend 1 night in the sleep laboratory, where a polysomnography following the manual of the AASM was implemented (Iber et al., 2007). Two hours after polysomnography and a light breakfast, the first MSLT trial started at 08:00 hours, and was repeated at 10:00, 12:00 and 14:00 hours. Between consecutive trials participants had to get out of their bed and were not allowed to sleep. Smoking had to be refrained from for at least 30 min before the beginning of the next trial. Electroencephalogram (with C3-A2, C4-A1, O1-A2 and O2-A1), EOG (placed besides the left and right eyes), a mental/submental EMG and ECG leads were placed following the recommendations of the AASM (Littner et al., 2005). Participants received a standard instruction to lie in bed quietly, close their eyes and try to fall asleep. The test was stopped after 20 min if participants did not fall asleep during this time. Recordings were prolonged for 15 min if subjects fell asleep starting from the first epoch measured as such to look for the occurrence of SOREMPs. Assessment of EEG-vigilance regulation via the VIGALL Between the first and the second trial of the MSLT (between 09:00 and 10:00 hours), a 15-min resting-eeg was recorded for assessment of vigilance regulation via VIGALL. Using a QuickAmp amplifier (Brain Products, Gilching, Germany), 25

3 452 S. Olbrich et al. electrodes (Fp1/2, F3/4, Fz, F7/8, FC1/2, FC5/6, Cz, C3/4, T7/ 8, CP5/6, Pz, P3/4, P7/8, O1/2) were fixed on the head of the subjects following an extended international system. Also, two electrodes for horizontal and two for vertical eye movements were applied. Impedances were kept below 10 ko. Participants were laid down in a half-reclined position, instructed to close their eyes and relax. The light was dimmed, temperature kept at C and sounds were attenuated. Questionnaires For exclusion of depressive or anxiety symptoms, the Beck Depression Inventory (BDI; Beck et al., 1961), the Hospital Anxiety and Depression Scale (HADS; Zigmond and Snaith, 1983) and the State-Trait Anxiety Inventory (STAI; Spielberger and Sydeman, 1994) were handed out to the participating subjects. Further, the Epworth Sleepiness Scale (ESS) as a self-rating questionnaire was used to assess the subjective general daytime sleepiness (Johns, 1991). Processing Multiple Sleep Latency Test Polysomnographic data of the night before the MSLT were analysed by two experienced raters following the standard procedures of the AASM. No pathological patterns were found in the healthy subjects. Also, the MSLT data were evaluated by two experienced raters. When there was disagreement on the results of segments, the respective data set was revised and a combined solution was figured out. According to the guidelines (Carskadon et al., 1986; Littner et al., 2005), sleep latency was defined as the time from lights turned off to the first stages classified as sleep. This was considered to be the case if the first occurrence of sleep patterns amounted to at least 50% out of a 30-s scoring epoch. Mean sleep latency (MSL) was calculated from the sleep latency of the four consecutive trials. While stage 0 is characterized by a low-voltage EEG without slow horizontal eye movements, A-stages are accompanied by a dominant alpha rhythm (with different grades of shifts to central and frontal cortical areas from stage A1 to A3), B-stages are characterized by a low-amplitude non-alpha EEG with slow horizontal eye movements (substage B1), or by high delta and theta power (substages B2/3). Stage C is defined by the occurrence of sleep spindles and K-complexes. To further individualize the classification, VIGALL automatically detects the individual alpha peak frequency and the average alpha power to adapt the algorithm scheme. For a more detailed description of the parameters used, please refer to the VIGALL manual ( ~vigall/) or validation studies (Hegerl et al., 2012; Olbrich et al., 2009, 2011b). The EEG-vigilance stage time series were transformed into a measure ranging from 7 (highest vigilance stage 0) to 1 (lowest vigilance stage C). No data set contained more than 10% of artefacts (Fig. 1). For assessment of the EEG-vigilance regulation type, two different approaches were applied. At first, each individual EEG-vigilance time series (consisting of 900 consecutive EEG-vigilance stages as classified via VIGALL) was compared with three different types of EEG-vigilance regulation derived from a k-means cluster solution described elsewhere (Olbrich et al., 2012a,b). Artefact segments were treated with the last observation carried forward approach for the cluster analysis. Each subject was classified as a cluster type with stable, medium or unstable EEG-vigilance regulation following its decline to low vigilance stages during the resting state. Secondly, each individual EEG-vigilance time series was classified using a lability index (LI). The LI is an ordinal scaled measure that spans between 1 and 11 points, with higher VIGALL and its classification The recorded EEG data were evaluated using the computerbased VIGALL. Ocular and muscle artefacts were removed from the raw EEG data using an independent component analysis approach. Remaining artefacts were marked after visual inspection. A certain vigilance stage was attributed to each 1-s EEG segment. VIGALL takes into account the vigilance stages as described by authors such as Bente (1976), Roth (1961) and Loomis et al. (1937), and more recently by others (Broughton and Hasan, 1995; De Gennaro et al., 2005; Tsuno et al., 2004). In short, vigilance stages range from stage 0 (corresponding to high alertness) to stage A (with substages A1, A2 and A3; corresponding to relaxed wakefulness) and further to stage B (with substages B1 and B2/3; corresponding to increasing drowsiness) and finally to stage C (sleep onset). Figure 1. Scheme for vigilance stage classification for 25 electroencephalogram (EEG)-channel 1-s segments by the Vigilance Algorithm Leipzig (VIGALL).

4 Comparison between the MSLT and VIGALL 453 scores representing higher vigilance instability. Scoring is done using specific decision criteria: a score of 9 11 points is given if C-stages are reached during the recording period minutes 1 5 (LI = 11), minutes 6 10 (LI = 10) or minutes (LI = 9). A score of 6 8 points is given if within a 1-min period at least 1/3 of the respective 60 segments have been classified as B2/3. A score of 3 5 points is given if within a 1-min period at least 1/3 of the respective 60 segments have been classified as any kind of B-stage. Finally, 1 2 points are given if throughout the complete 15-min resting EEG, there is no 1-min period during which less than 2/3 of the 60 segments are classified as either 0/A-stages (LI = 2) or 0/A1-stages (LI = 1). If several of these criteria are fulfilled in an individual EEG, the highest score is used. Statistics STATA software Version 10.1 (StataCorp LP, College Station, TX, USA) and SPSS 20.0 (IBM SPSS Statistics for Windows; IBM Corp, Armonk, NY, USA) were used for statistical analysis. For correlation between EEG-vigilance regulation type, MSLT results and ESS, a Spearman rank correlation was performed. Significance level was set to P < The MSLT results of the subjects with different EEG-vigilance regulation types were analysed using one-way ANOVA (MSLT as dependent variable and vigilance regulation type as independent variable) with post hoc Scheffe tests. The sensitivity and specificity of the EEG-vigilance regulation types for identification of subjects with an increased propensity to fall asleep were calculated using receiver-operating characteristics (ROC curves). These were computed for MSLT cut-off values at 6, 7, 8, 9, 10 and 12 min. To rule out an effect of age on the results, a split-half comparison of the MSLT and VIGALL results was performed for the younger and older participants using Fisher s exact test (vigilance regulation type and LI) and two-sided t-test (MSLT). RESULTS Sociodemographic data Twenty-five healthy subjects (15 females, years, mean 38.44, SD 10.54; Table 1) were included. Participants showed neither depressive symptoms with a BDI mean of 3.40 (3.66) and HADS mean of 1.84 (2.20), nor symptoms of anxiety with a HADS mean of 3.68 (3.15) and STAI mean of (8.80). No difference between old and young subjects was found for the MSL (P = 0.12), the vigilance regulation cluster (P = 0.17) and the LI (P = 0.91). Multiple Sleep Latency Test No subject showed a pathological aspect in the actometer recordings or during the initial night of polysomnography. The MSL averaged over all four trials was 732 s (SD 277 s; Table 1), with a minimum MSL of 308 s and maximum MSL of 1178 s. No subject had an MSL below 5 min (300 s), and no subject showed SOREMPs. Vigilance analysis The average amounts of the EEG-vigilance stages 0, A, B and C (from wakefulness to sleep onset) can be found in Table 1. Ten probands were assigned to EEG-vigilance regulation cluster 1, indicating a high stability of EEGvigilance; seven probands were assigned to cluster 2, indicating a moderate vigilance decline; and eight probands were assigned to cluster 3, with a fast declining vigilance (Table 1). For results of the LI, refer to Table 1. Questionnaires Both results of the MSLT (MSL) and of the VIGALL classification, i.e. EEG-vigilance cluster and LI, showed a significant correlation with the ESS with a medium effect size (MSL: q = 0.70, P < 0.001; EEG-vigilance cluster: q = 0.45, P = 0.026; LI: q = 0.49, P = 0.012). Comparison between MSLT and VIGALL A negative correlation of q = 0.54 was found between the MSL and the assigned EEG-vigilance regulation cluster (P = 0.006), and q = 0.48 (P = 0.015) between the MSL and the LI, indicating that subjects with a stable EEGvigilance regulation during rest show longer sleep latencies than subjects with an unstable vigilance regulation pattern. ANOVA testing revealed a significant difference of MSLT results for subjects with different EEG-vigilance regulation types (F = 4.38/df 2; P = 0.025). Post hoc Scheffe tests showed that subjects assigned to the stable vigilance regulation cluster 1 had a significantly larger MSLT than subjects assigned to unstable vigilance regulation cluster 3 (MSL: 899 s; SD 242 s versus MSL: 550 s; SD 239 s; P = 0.026). No differences were found for comparisons of MSLT results between vigilance clusters 1 and 2 (MSL: 705 s; SD: 217 s), or 2 and 3, respectively (Fig. 2). Taking the MSL at different cut-off values as the goldstandard for an increased propensity for sleep, the ROCs for the EEG-vigilance regulation clusters (stable vigilance regulation = 1; slowly declining vigilance = 2; instable vigilance regulation = 3) were computed. The largest number of correctly classified subjects (high sleep propensity versus low sleep propensity) were found for MSL cut-off values below 7 min (420 s; 80% correct classifications) and 6 min (360 s; 80% correct classifications), with a sensitivity of 80% and a specificity of 80; 100 and 77%, respectively, for subjects classified with EEG-vigilance regulation cluster 3 (Fig. 3). DISCUSSION The results show that the assessment of EEG-vigilance regulation during the resting state in a single 15-min EEG

5 454 S. Olbrich et al. Table 1 Sociodemographic data, MSLT times and results of the EEG-vigilance regulation test as well as results of self-rating questionnaires for all participating subjects Subject Sociodemography Questionnaire results VIGALL MSLT Sex Age (years) BDI HADS-A HADS-D STAI ESS Cluster LI MSL (s) 1 F F M M M M F F M F F F M F F M F F M F M F F F M Mean SD BDI, Beck Depression Inventory Total Score; ESS, Epworth Sleepiness Scale Total Score; HADS, Hospital Anxiety and Depression Scale (A = Anxiety Score; D = Depression Score); LI, lability index; MSL, multiple sleep latency; MSLT, Multiple Sleep Latency Sleep Test; STAI, State Trait Anxiety Inventory Score; VIGALL, Vigilance Algorithm Leipzig. Figure 2. Boxplot for Multiple Sleep Latency Test (MSLT) times of 25 subjects assigned to one of three different electroencephalogram (EEG)-vigilance regulation types (stable cluster; slowly declining cluster; unstable cluster) that have been derived from one 15-min resting EEG. ANOVA test confirmed significant differences, post hoc Scheffe test showed a significantly higher sleep latency for subjects of the stable vigilance regulation type compared with the unstable type. yields comparable classifications of a subjects sleep propensity during daytime as compared with the results of the more time- and cost-intensive MSLT. Subjects that have been attributed with an unstable EEG-vigilance regulation showed a significantly shorter MSL than subjects with a stable EEG-vigilance regulation. The sensitivity and specificity for identification of subjects with a MSL below 6 min (360 s) was high, with 100 and 77.27%, respectively, for subjects classified with an unstable EEG-vigilance regulation. This suggests the possible usage of the used VIGALL algorithm, for example, for screening purposes or as entry for the diagnostic process. Further, both measures (the MSLT and the EEG-vigilance regulation) showed significant but medium effect-sized correlations with the subjective estimation of the general probability to fall asleep, assessed by the ESS. The average MSL in this study of 732 s (SD 277 s) was slightly larger than revealed from a pooled data analysis by Littner et al. (2005) of 624 s (SD 258 s). This might be explained by the smaller age range in this study, while the findings from Littner integrated several studies and results

6 Comparison between the MSLT and VIGALL 455 Figure 3. Plotted receiver-operator characteristics (ROC) curves for different Multiple Sleep Latency Test (MSLT) cut-off values (as indicator for high sleep propensity) and the different EEG-vigilance regulation types. were collapsed over all age spans. Another possible explanation is the fact that the subjects of the presented study had one additional resting period (for EEG-vigilance regulation assessment) between the first two MSLT trials. Because the MSLT is sensitive to multiple factors that impact sleep behaviour (Arand et al., 2005; Thorpy, 1992), this might have decreased the overall propensity to fall asleep in the last three MSLT trials with a resulting larger averaged MSL. However, the findings of the presented study with 25% of subjects showing a MSL below 8 min are in line with the reported 30% of subjects that fell below this cut-off value in other studies (Mignot et al., 2006; Singh et al., 2006). The findings of the significant negative correlation of the MSL with ESS scores (q = 0.70) are in line with reports from the literature, although most other studies found lower values with, for example, q = 0.42, n = 44 (Johns, 1994) or q = 0.27, n = 522 (Sangal et al., 1999). Neither study reported on patient groups with increased sleep propensity, which might impact the comparability between the presented results and their findings. However, comparing the EEGvigilance regulation type and the ESS scores also resulted in a significant correlation with a medium effect size (i.e. between the general subjective probability to fall asleep in everyday situations and the objectively assessed vigilance regulation) comparable to what has been reported on the correlation between the ESS and the MSLT (Chervin et al., 1997; Sangal et al., 1999). The clinical usage of the MSLT has been discussed critically (Arand et al., 2005; Bonnet, 2006), and also the association between neurophysiological measures of sleep propensity and the ESS is subject to debate (Johns, 2000). The presented results underline that both the MSLT and the EEG-vigilance regulation as measured by VIGALL capture similar features of the sleep wake regulation system. As a time cut-off for analysing the ROC curves, we used a range of 6 12 min. In the face of the ongoing discussion about the pathological threshold of the MSLT that ranges from 5 min, as suggested from Richardson et al. (1978) and Carskadon et al. (1986) to 8 min, as proposed by others (Pizza et al., 2013), the choice of the right threshold was beyond the scope for this study. Still, the sensitivity of 100% and specificity of 77% for the EEG-vigilance algorithm to discriminate subjects with a MSL below 6 min (no smaller MSLs were found in this study) indicates that the VIGALL results are comparable to those of the MSLT. Limitations of the presented study are that the investigated cohort comprised no pathological conditions. To further facilitate the possible usage of EEG-vigilance regulation

7 456 S. Olbrich et al. markers as assessed by the VIGALL for measuring sleep propensity, studies with patients, for example, suffering from narcolepsy or idiopathic insomnia, have to be conducted to validate the results of the algorithm using data comprising larger variance in sleep latencies and sleep propensities. Further, it has to be noted that several psychopharmacological drugs impact on the amplitude and frequency spectrum of the resting EEG. While neuroleptic drugs such as clozapine and the mood stabilizer lithium lead to increased delta- and theta-band activity (Hyun et al., 2011; Schulz et al., 2000), sedative drugs such as benzodiazepines and barbiturates and also antidepressants can lead to a remarkable increase of beta activity (Joy et al., 1971), especially at frontal brain areas. Because the VIGALL algorithm mainly depends on spatiotemporal features of the EEG, the interpretation of results from subjects that have been prescribed psychopharmacological medication should be done with caution. CONCLUSION Vigilance Algorithm Leipzig has not been designed to replace the MSLT: a repeated measure will always be more reliable in a diagnostic process, and the MSLT has proven its power for diagnostic purposes, such as differential diagnosis of narcolepsy and idiopathic hypersomnia over several decades. Still, within times of economic considerations, in the face of large cohort studies or in the case where a sleep laboratory is not available, the assessment of EEG-vigilance regulation by VIGALL provides a diagnostic approach to screen for subjects with an increased sleep propensity to facilitate a further diagnostic process. Although studies with larger sample sizes are needed to prove the validity and reliability of VIGALL, the sensitivity of 100% for a MSL below 6 min suggests that the VIGALL might be capable of identifying all subjects with a high sleep propensity and then to advise further measurements such as the MSLT. ACKNOWLEDGEMENT This study was supported by the Integrated Research and Treatment Centre for Adiposity Diseases (IFB), University of Leipzig, Germany. REFERENCES Arand, D., Bonnet, M., Hurwitz, T., Mitler, M., Rosa, R. and Sangal, R. B. The clinical use of the MSLT and MWT. Sleep, 2005, 28: Beck, A. T., Ward, C. H., Mendelson, M., Mock, J. and Erbaugh, J. An inventory for measuring depression. Arch. Gen. Psychiatry, 1961, 4: Bente, D. EEG aspects of waking-sleep behaviour and the chronophysiology of endogenous depressions. Arzneimittelforschung, 1976, 26: Bonnet, M. H. ACNS clinical controversy: MSLT and MWT have limited clinical utility. J. Clin. Neurophysiol., 2006, 23: Broughton, R. and Hasan, J. Quantitative topographic electroencephalographic mapping during drowsiness and sleep onset. J. Clin. Neurophysiol., 1995, 12: Carskadon, M. A., Dement, W. C., Mitler, M. M., Roth, T., Westbrook, P. R. and Keenan, S. Guidelines for the multiple sleep latency test (MSLT): a standard measure of sleepiness. Sleep, 1986, 9: Chervin, R. D., Aldrich, M. S., Pickett, R. and Guilleminault, C. Comparison of the results of the Epworth Sleepiness Scale and the Multiple Sleep Latency Test. J. Psychosom. Res., 1997, 42: De Gennaro, L., Vecchio, F., Ferrara, M., Curcio, G., Rossini, P. M. and Babiloni, C. Antero-posterior functional coupling at sleep onset: changes as a function of increased sleep pressure. Brain Res. Bull., 2005, 65: Guilleminault, C. and Brooks, S. N. Excessive daytime sleepiness: a challenge for the practising neurologist. Brain, 2001, 124: Hegerl, U. and Hensch, T. The vigilance regulation model of affective disorders and ADHD. Neurosci. Biobehav. Rev., 2014, 44: Hegerl, U., Wilk, K., Olbrich, S., Schoenknecht, P. and Sander, C. Hyperstable regulation of vigilance in patients with major depressive disorder. World J. Biol. Psychiatry, 2012, 13: Himanen, S. and Hasan, J. Limitations of Rechtschaffen and Kales. Sleep Med. Rev., 2000, 4: Hyun, J., Baik, M. J. and Kang, U. G. Effects of psychotropic drugs on quantitative EEG among patients with schizophrenia-spectrum disorders. Clin. Psychopharmacol. Neurosci., 2011, 9: Iber, C., Ancoli-Israel, S. and Chesson, A. (Eds) The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology, and Technical Specification. American Academy of Sleep Medicine, Westchester, IL, Johns, M. W. A new method for measuring daytime sleepiness: the Epworth sleepiness scale. Sleep, 1991, 14: Johns, M. W. Sleepiness in different situations measured by the Epworth Sleepiness Scale. Sleep, 1994, 17: Johns, M. W. Sensitivity and specificity of the multiple sleep latency test (MSLT), the maintenance of wakefulness test and the Epworth sleepiness scale: failure of the MSLT as a gold standard. J. Sleep Res., 2000, 9: Joy, R. M., Hance, A. J. and Killam, K. F. A quantitative electroencephalographic comparison of some benzodiazepines in the primate. Neuropharmacology, 1971, 10: Littner, M. R., Kushida, C., Wise, M. et al. Practice parameters for clinical use of the multiple sleep latency test and the maintenance of wakefulness test. Sleep, 2005, 28: Loomis, A. L., Harvey, E. N. and Hobart, G. A. Cerebral states during sleep, as studied by human brain potentials. J. Exp. Psychol., 1937, 21: Mignot, E., Lin, L., Finn, L. et al. Correlates of sleep-onset REM periods during the Multiple Sleep Latency Test in community adults. Brain, 2006, 6: Olbrich, S., Mulert, C., Karch, S. et al. EEG-vigilance and BOLD effect during simultaneous EEG/fMRI measurement. NeuroImage, 2009, 45: Olbrich, S., Sander, C., Jahn, I. et al. Unstable EEG-vigilance in patients with cancer-related fatigue (CRF) in comparison to healthy controls. World J. Biol. Psychiatry, 2011a, 13: Olbrich, S., Sander, C., Matschinger, H. et al. Brain and body: associations between EEG-vigilance and the autonomous nervous system activity during rest. Neuropsychobiology, 2011b, 25: Olbrich, S., Olbrich, H., Jahn, I. et al. EEG-vigilance regulation during the resting state in obsessive-compulsive disorder. Clin. Neurophysiol., 2012a, 124: Olbrich, S., Sander, C., Minkwitz, J. et al. EEG vigilance regulation patterns and their discriminative power to separate patients with

8 Comparison between the MSLT and VIGALL 457 major depression from healthy controls. Neuropsychobiology, 2012b, 65: Pataka, A., Yoon, C. H., Poddar, A. and Riha, R. L. Assessment of multiple sleep latency testing in adults in Europe. Sleep Med., 2013, 14: Pizza, F., Moghadam, K. K., Vandi, S. et al. Daytime continuous polysomnography predicts MSLT results in hypersomnias of central origin. J. Sleep Res., 2013, 22: Richardson, G. S., Carskadon, M. A., Flagg, W., Van den Hoed, J., Dement, W. C. and Mitler, M. M. Excessive daytime sleepiness in man: multiple sleep latency measurement in narcoleptic and control subjects. Electroencephalogr. Clin. Neurophysiol., 1978, 45: Roth, B. The clinical and theoretical importance of EEG rhythms corresponding to states of lowered vigilance. Electroencephalogr. Clin. Neurophysiol., 1961, 13: Sangal, R. B., Mitler, M. M. and Sangal, J. M. Subjective sleepiness ratings (Epworth sleepiness scale) do not reflect the same parameter of sleepiness as objective sleepiness (maintenance of wakefulness test) in patients with narcolepsy. Clin. Neurophysiol., 1999, 110: Schulz, C., Mavrogiorgou, P., Schr oter, A., Hegerl, U. and Juckel, G. Lithium-induced EEG changes in patients with affective disorders. Neuropsychobiology, 2000, 42(Suppl. 1): Singh, M., Drakem, C. L. and Roth, T. The prevalence of multiple sleep-onset REM periods in a population-based sample. Sleep, 2006, 29: Spielberger, C. D. and Sydeman, S. J. State-Trait Anxiety Inventory and State-Trait Anger Expression Inventory. The Use of Psychological Testing for Treatment Planning and Outcome Assessment. Lawrence Erlbaum, Hillsdale, NJ, 1994: Thorpy, M. J. The clinical use of the Multiple Sleep Latency Test. The Standards of Practice Committee of the American Sleep Disorders Association. Sleep, 1992, 15: Tsuno, N., Shigeta, M., Hyoki, K., Faber, P. L. and Lehmann, D. Fluctuations of source locations of EEG activity during transition from alertness to sleep in Alzheimer s disease and vascular dementia. Neuropsychobiology, 2004, 50: Tucker, A. M., Dinges, D. F. and Van Dongen, H. P. A. Trait interindividual differences in the sleep physiology of healthy young adults. J. Sleep Res., 2007, 16: Van Dongen, H. P. A., Vitellaro, K. M. and Dinges, D. F. Individual differences in adult human sleep and wakefulness: leitmotif for a research agenda. Sleep, 2005, 28: Zigmond, A. S. and Snaith, R. P. The Hospital Anxiety and Depression Scale. Acta Psychiatr. Scand., 1983, 67: