Cardiovascular variability as a function of sleep–wake behaviour in narcolepsy with cataplexy

Authors

  • Alessandro Silvani,

    Corresponding author
    • PRISM Laboratory, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Daniela Grimaldi,

    1. IRCCS Bologna Institute of Neurological Sciences, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Bologna, Italy
    Search for more papers by this author
    • These authors contributed equally to this work.
  • Giorgio Barletta,

    1. IRCCS Bologna Institute of Neurological Sciences, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Bologna, Italy
    Search for more papers by this author
  • Stefano Bastianini,

    1. PRISM Laboratory, Department of Biomedical and Neuromotor Sciences, University of Bologna, Bologna, Italy
    Search for more papers by this author
  • Stefano Vandi,

    1. IRCCS Bologna Institute of Neurological Sciences, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Bologna, Italy
    Search for more papers by this author
  • Giulia Pierangeli,

    1. IRCCS Bologna Institute of Neurological Sciences, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Bologna, Italy
    Search for more papers by this author
  • Giuseppe Plazzi,

    1. IRCCS Bologna Institute of Neurological Sciences, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Bologna, Italy
    Search for more papers by this author
  • Pietro Cortelli

    1. IRCCS Bologna Institute of Neurological Sciences, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Bologna, Italy
    Search for more papers by this author

  • This study was part of the PhD research programme of Dr Daniela Grimaldi.

Correspondence

Dr Alessandro Silvani MD PhD, Department of Biomedical and Neuromotor Sciences, Alma Mater Studiorum, Università di Bologna, Piazza di Porta San Donato 2, 40126 Bologna, Italy.

Tel.: +39 051 2091739;

fax: +39 051 2091737;

e-mail: alessandro.silvani3@unibo.it

Summary

Hypocretin/orexin signalling varies among sleep–wake behaviours, impacts upon cardiovascular autonomic control and is impaired in patients with narcolepsy with cataplexy (NC). However, evidence concerning disturbed cardiovascular autonomic control in NC patients is contrasting, and limited mainly to waking behaviour. We thus investigated whether control of cardiovascular variability is altered in NC patients during wakefulness preceding sleep, light (1–2) and deep (3–4) stages of non-rapid eye movement (NREM) sleep and rapid eye movement (REM) sleep. Polysomnographic recordings and finger blood pressure measurements were performed on nine drug-free male NC patients and nine matched healthy control subjects during spontaneous sleep–wake behaviour in a standardized laboratory environment. Indices of autonomic function were computed based on spontaneous fluctuations of systolic blood pressure (SBP) and heart period (HP). During wakefulness before sleep, NC patients showed significant decreases in indices of vagal HP modulation, cardiac baroreflex sensitivity and amplitude of central autonomic (feed-forward) cardiac control compared with control subjects. During NREM sleep, the negative correlation between HP and subsequent SBP values was greater in NC patients than in control subjects, suggesting a greater contribution of central autonomic commands to cardiac control. Collectively, these results provide preliminary evidence that autonomic control of cardiac variability by baroreflex and central autonomic (feed-forward) mechanisms is altered in NC patients during spontaneous sleep–wake behaviour, and particularly during wakefulness before sleep.

Introduction

Narcolepsy with cataplexy (NC) is a chronic disorder of sleep–wake behaviour (Dauvilliers et al., 2007) associated with loss of neurones releasing hypocretin/orexin peptides (Peyron et al., 2000). Whether NC also entails derangements in the control of cardiovascular variability is still a matter of debate (Plazzi et al., 2011). Impairments in some (i.e. heart rate changes with handgrip, deep breathing and Valsalva manoeuvre) but not all (i.e. heart rate changes with head-up tilt and cold-face tests) classic cardiac function tests were reported in NC patients (Sachs and Kaijser, 1982) but were not replicated by two subsequent studies (Grimaldi et al., 2010; Hublin et al., 1994). Studies employing computer analysis of cardiovascular variability have also shown derangements in cardiac autonomic control in NC patients, but have not been fully consistent. In particular, a shift in sympathovagal balance away from vagal dominance was reported in NC patients during wakefulness either before sleep (Ferini-Strambi et al., 1997) or in the morning after sleep (Grimaldi et al., 2010), whereas an increase in heart rate variability without changes in sympathovagal balance was reported by another study in NC patients during morning wakefulness (Fronczek et al., 2008). Blunted heart rate fluctuations have been reported recently in NC patients compared with control subjects during non-rapid eye movement (NREM) sleep in association with periodic leg movements (PLM) (Dauvilliers et al., 2011).

We aimed at assessing whether NC entails derangements in autonomic control of cardiovascular variability not only during wakefulness, but also during NREM and rapid eye movement (REM) sleep, using a methodological approach differing from those employed in previous reports. In particular, we analysed the coupling between spontaneous fluctuations of heart period (HP) and systolic blood pressure (SBP) in order to assess the autonomic control of cardiovascular variability exerted by baroreflex and central autonomic (feed-forward) mechanisms (Silvani, 2008; Silvani et al., 2012).

Methods

Subjects

The analysis was performed on nine unrelated drug-free (two drug-naive, seven having withdrawn modafinil >2 weeks before the study, one of these also having withdrawn clomipramine 1 month before the study) male NC patients [age: 34 ± 5 years; body mass index (BMI): 28 ± 4] referred to the Bologna Sleep Disorders Center (Italy) and meeting the International Classification of Sleep Disorders (ICSD-II) diagnostic criteria. All patients were positive for haplotype human leucocyte antigen (HLA) DQB1*0602 and eight of nine had low/undetectable hypocretin-1 levels in the cerebrospinal fluid (≤110 pg ml−1; one patient refused lumbar puncture). NC patients were compared to nine healthy male control subjects (CS, age 38 ± 12 years, BMI: 26 ± 4) free of drugs and medications and not complaining for daytime sleepiness. In NC patients and CS, cardiac, endocrine, metabolic and renal diseases were excluded on the basis of history-taking, physical examination and routine laboratory tests, while obstructive sleep apnoea syndrome was excluded by a dynamic polysomnographic study [apnoea–hypopnoea index (AHI) <10]. The institutional review board of the Department of Neurology of the Bologna University approved the project. All subjects gave their written informed consent. The present study describes entirely novel results concerning cardiovascular variability in the male NC patients and in four male CS, who had been included previously in a study on the sleep–cardiovascular system interaction in NC (Grimaldi et al., 2012) and whose recordings satisfied the technical requirements for data transfer to the computing software environment of the present study.

Experimental protocol

During the study, subjects lived in a room with a controlled temperature (24 ± 1 °C), humidity (40–50%) and light–dark schedule (light off from 23:00 hours to 07:00 hours). Subjects were required to lie in bed, except for eating (1800 kcal day−1 divided into three meals and three snacks with a fixed time schedule), but were allowed to read, watch television and sleep ad libitum. Continuous non-invasive recordings of physiological variables were performed for 48 h. Finger blood pressure was measured with the volume-clamp method (Portapres Model-2; Finapres Medical Systems, Amsterdam, the Netherlands). Electrocardiogram, electroencephalogram (C3-A2 and C4-A1 leads), electro-oculogram, electromyogram (mylohyoideus muscle) and ventilation measurements were recorded with a Colleague recorder (Albert Grass Heritage, Model PSG16P-1; Astro-Med, Inc, West Warwick, RI, USA). All signals were digitized at 1 kHz.

Data analysis

Sleep states were scored visually in 30-s epochs according to standard Rechtschaffen and Kales criteria as light (Stages 1 and 2) NREM, deep (Stages 3 and 4) NREM and REM sleep. Sleep efficiency was computed as the percentage of time spent asleep with lights off. The PLM sleep index (PLMSI) was computed as the number of PLM per hour of nocturnal sleep. Cardiovascular data analysis was performed on all episodes of light NREM, deep NREM or REM sleep with duration ≥5 min. Episodes of wakefulness ≥5 min with subjects lying in bed in the hour before the onset of the night-time sleep period were also analysed. The number of PLM per hour during these episodes of wakefulness was computed as the index PLMWI.

Beat-to-beat values of HP and SBP were computed and artefacts were dealt with as described in detail previously (Silvani et al., 2008). The total variability of SBP and HP was estimated as the standard deviation of the respective beat-to-beat values over 5-min windows, yielding the indices standard deviation of systolic blood pressure (SDSBP) and standard deviation of the average NN interval (SDNN), respectively. The index RMSSD was computed as the square root of the mean squared differences of successive HP values. The interpretation of RMSSD as an index of cardiac vagal modulation is supported by pharmacological experiments on human subjects (Van Den Berg et al., 1997). The low-frequency/high frequency (LF/HF) index was computed with fast Fourier transforms as the ratio between HP spectral power in the low-frequency (0.04–0.15 Hz) and high frequency (0.15–0.4 Hz) bands (ESC/NASPE Task Force, 1996). The interpretation of LF/HF as an index of the prominence of sympathetic over parasympathetic modulation of HP (i.e. sympathovagal balance) is supported by pharmacological experiments on dogs and human subjects (Pagani et al., 1986). Cardiac baroreflex sensitivity (BRS, i.e. the estimated baroreflex change in HP driven by a unit change in SBP) was estimated with two different algorithms. The seq-BRS index was computed in the time domain with the sequence technique (Bertinieri et al., 1988), as described previously (Silvani et al., 2008). The dependency of seq-BRS estimates on the baroreceptor reflex is supported by experimental sino-aortic denervation in cats (Bertinieri et al., 1988). The LF-BRS index was computed in the frequency domain as the gain of the transfer function between SBP and HP in the low-frequency band (Robbe et al., 1987). LF-BRS and pharmacological BRS estimates are correlated highly in humans (Robbe et al., 1987).

The coupling between spontaneous fluctuations of HP and SBP was analysed with two complementary approaches. The first approach estimated the amplitude of HP fluctuations that were associated temporally with spontaneous events (SBP surges), in which SBP increased transiently ≥20 mm Hg above baseline (Silvani et al., 2012). The time–series were synchronized at the peak SBP increase and averaged after subtraction of their respective baseline values. Data on animal models indicate that in each sleep–wake state, HP decreases before the peak of spontaneous SBP surges and returns towards or above baseline levels thereafter (Silvani et al., 2012). Mathematical modelling supports the interpretation of these HP trough and peak as the magnitudes of HP changes driven by central autonomic and baroreflex cardiac control, respectively (Silvani et al., 2011). The other approach analysed the overall coupling between HP and SBP fluctuations at frequencies below the breathing rate (<0.15 Hz) by means of a cross-correlation function (CCF) analysis (Silvani et al., 2008, 2012). The CCF yields the correlation coefficient between HP and SBP as a function of the time shift between these variables. Negative values of the time shift indicate that changes in HP follow those in SBP. In human subjects (Silvani et al., 2008) and animal models (Silvani et al., 2012), the CCF between HP and SBP shows a negative trough at positive time shifts and/or a positive peak at negative time shifts. The first CCF pattern represents a negative correlation between HP and subsequent SBP values, indicating, e.g. cardiac acceleration followed by an increase in SBP. The second CCF pattern represents a positive correlation between HP and previous SBP values, indicating, e.g. hypertension followed by cardiac slowing. Mathematical modelling indicates that these CCF troughs and peaks reflect the relative contributions of central autonomic commands and the baroreflex to cardiac control, respectively (Silvani et al., 2011).

The indices SDSBP, SDNN and seq-BRS were computed on beat-to-beat values of HP and SBP. The other analyses were performed after resampling HP and SBP at 20 Hz with piecewise cubic Hermite interpolation. Cardiovascular data analysis was performed in MatLab (Mathworks, Natick, MA, USA).

Statistics

Statistical tests were performed with spss (SPSS, Chicago, IL, USA). Data were analysed by analysis of variance (anova) [general linear model (GLM) procedure with mixed-model design and Huynh-Feldt correction] to test the effects of sleep–wake state, group (NC patients versus CS) and the state × group interaction effect with significance at < 0.05. Differences between NC patients and CS in each sleep–wake state were tested by t-tests. The significance of these t-tests was set at the conventional < 0.05 level in case of significant state × group interaction effect at anova or with the false discovery rate procedure (Curran-Everett, 2000) otherwise. Data are reported as means ± standard error of the mean (SEM) in the text, table and figures consistently with the use of parametric tests, with nine NC patients and nine CS patients.

Results

In CS, total sleep time was 392 ± 13 min each night with sleep efficiency of 82 ± 3%. Corresponding figures in NC patients were 368 ± 2 min and 77 ± 0.4%. CS spent 50 ± 3%, 27 ± 2% and 24 ± 2% of their sleep time in light NREM, deep NREM and REM sleep, respectively. Corresponding figures in NC patients were 56 ± 4%, 24 ± 1% and 19 ± 3% respectively. As reported previously (Grimaldi et al., 2012), the PLMSI was significantly higher (= 0.03) in NC patients (26 ± 7) than in CS (7 ± 4). The PLMWI before nocturnal sleep onset did not differ significantly between groups (NC, 17 ± 7; CS, 27 ± 14).

Neither the total variability of SBP (SDSBP, Fig. 1a) nor that of HP (SDNN, Fig. 1b) differed significantly between NC patients and CS in any sleep–wake state, although SDNN tended to be lower in NC patients than in CS during wakefulness (= 0.10). Cardiac vagal modulation (RMSSD, Fig. 1c) was reduced significantly in NC patients compared with CS during wakefulness (= 0.04), whereas the difference in sympathovagal balance (LF/HF, Fig. 1d) during wakefulness fell just short of statistical significance (= 0.06). Both indices of cardiac baroreflex sensitivity were significantly lower (= 0.01) in NC patients than in CS during wakefulness (BRS-LF, Fig. 1e; BRS-seq, Fig. 1f).

Figure 1.

Indices of cardiovascular variability and cardiac baroreflex sensitivity as a function of sleep–wake behaviour. (a,b) The total variability of systolic blood pressure (SDSBP) and heart period (SDNN), respectively. (c,d) Indices of cardiac vagal modulation (RMSSD) and cardiac sympatho-vagal balance [low-frequency/high frequency (LF/HF)], respectively. (e,f) Results of two different algorithms for estimation of spontaneous cardiac baroreflex sensitivity in the frequency domain (BRS-LF) and in the time domain (BRS-seq), respectively. W: wakefulness before sleep; NREMS 1–2 and 3–4: Stages 1–4 of non-rapid eye movement sleep; REMS: rapid eye movement sleep. Data are means ± standard error of the mean (SEM) in patients with narcolepsy–cataplexy (NC) and control subjects (CS). Analysis of variance (anova) state,  0.01 (all variables); group, = 0.04 (BRS-seq),  0.12 (other variables); state × group, < 0.01 (BRS-LF), < 0.05 (SDNN, RMSSD), = 0.08 (LF/HF),  0.14 (SDSBP, BRS-seq). *< 0.05 versus CS (t-test).

HP decreased before the SBP surge peak, consistently with the activity of central autonomic commands on the heart, and returned towards or above baseline levels after the SBP surge peak, consistently with baroreflex cardiac control (Fig. 2). This was expected from experiments on animal models (Silvani et al., 2012) and mathematical modelling of cardiovascular control (Silvani et al., 2011). HP peaks did not differ significantly between groups in any wake sleep state. Conversely, the negative HP trough was significantly less pronounced in NC patients than in CS during wakefulness (< 0.01), indicating a reduced amplitude of HP fluctuations driven by central autonomic (feed-forward) control.

Figure 2.

Coherent averaging of spontaneous surges in systolic blood pressure (SBP). The panels show time–series of SBP (ΔSBP, top) and heart period (ΔHP, bottom) normalized by subtraction of their respective baseline values and synchronized at the peak SBP increase. Data are means ± standard error of the mean (SEM) in narcolepsy–cataplexy (NC) patients and control subjects (CS). Analysis of variance (anova), ΔHP trough: state, = 0.056; group, = 0.19; state × group, = 0.02; anova, ΔHP peak: state, = 0.06; group, = 0.35; state × group, = 0.09. *< 0.05 between groups (t-test).

The average CCF between HP and SBP showed a negative trough at positive time shifts, which was evident in each sleep–wake state, and a positive peak at negative time shifts, which was evident in NREM sleep only (Fig. 3). This general pattern was again expected from experiments on human subjects (Silvani et al., 2008) and mathematical modelling of cardiovascular control (Silvani et al., 2011). The positive CCF peak did not differ significantly in magnitude between groups in any sleep–wake state. However, the negative CCF trough, which is consistent with central autonomic cardiac control (Silvani et al., 2011), was significantly more prominent in NC patients than in CS during light and deep NREM sleep (< 0.01).

Figure 3.

Cross-correlation functions (CCF) between low-frequency fluctuations of heart period and systolic blood pressure. Data are means ± standard error of the mean (SEM) in narcolepsy–cataplexy (NC) patients and control subjects (CS). Analysis of variance (anova), CCF peak: state, < 0.001; group, = 0.92; state × group, = 0.09; anova, CCF trough: state, P = 0.06; group, = 0.02; state × group, = 0.07. *< 0.05 between groups (t-test).

To aid the interpretation of cardiovascular variability, Table 1 reports the mean values of SBP and HP, on which SBP and HP variability was superimposed during the sleep–wake episodes analysed. SBP did not differ significantly between NC patients and CS in any sleep–wake state. However, NC patients had significantly lower values of HP than CS in each sleep–wake state (= 0.01 in wakefulness and = 0.03 in the sleep states), as reported previously, taking into account the whole 24-h recordings (Grimaldi et al., 2012).

Table 1. Mean values of finger blood pressure and heart period in the data segments employed for the analysis of cardiovascular variability
VariableGroupState
WNREMS 1–2NREMS 3–4REMS
  1. SBP, systolic blood pressure; HP, heart period; W, wakefulness before sleep; NREMS 1–2 and 3–4, Stages 1–4 of non-rapid eye movement sleep; REMS, rapid eye movement sleep. Data are means ± standard error of the mean (SEM) over the artefact-free data segments employed for the analysis of cardiovascular variability in control subjects (CS) and patients with narcolepsy–cataplexy (NC). Analysis of variance (anova), SBP, state, < 0.001; group and state × group, > 0.52; anova, HP, state, < 0.001; group, = 0.02; state × group, = 0.03. *< 0.05 versus CS (t-test).

SBP (mm Hg)CS124 ± 7114 ± 6113 ± 5119 ± 6
 NC122 ± 5110 ± 5108 ± 6119 ± 6
HP (ms)CS1026 ± 671128 ± 661117 ± 641083 ± 62
 NC803 ± 31*944 ± 41*931 ± 39*916 ± 35*

Discussion

The present study yielded two main novel findings. First, during wakefulness before sleep, NC patients showed reductions in vagal HP modulation, BRS and amplitude of central autonomic (feed-forward) cardiac control compared with CS. Secondly, during light and deep NREMS, NC patients showed alterations in cardiovascular coupling suggestive of an increased contribution of central autonomic commands to cardiac control compared with CS.

We assessed autonomic control in NC patients by analysing spontaneous cardiovascular variability. This approach is well suited to application during undisturbed sleep–wake behaviour, but is inherently indirect and complicated by the complexity of closed-loop cardiovascular system controls. In these conditions, a combined analysis of multiple variables with computational tools targeting different and interrelated aspects of autonomic control, as we employed in this study, is a strategy used frequently to improve data reliability and interpretation. In particular, our finding that BRS was lower in NC patients than in CS was based on qualitatively consistent results obtained with two different and validated algorithms of BRS estimation. BRS-LF (Fig. 1a) was based on slow cardiovascular fluctuations in the frequency domain (Robbe et al., 1987), whereas BRS-seq (Fig. 1f) was based on short sequences of a few heartbeats in the time domain (Bertinieri et al., 1988). Although these findings may indicate an isolated defect of the cardiac baroreflex in NC patients, our results support an alternative interpretation. Reductions in spontaneous BRS may, in fact, reflect reductions in the amplitude of cardiac vagal modulation (Lipman et al., 2003), which is available for entrainment by the cardiac baroreflex as well as by other non-baroreflex control mechanisms, such as central autonomic (feed-forward) cardiac control. Accordingly, we found a decrease in vagally mediated HP fluctuations (RMSSD, Fig. 1c) in NC patients during wakefulness compared with CS. Moreover, coherent averaging of spontaneous SBP surges indicated that the amplitude of the HP decrease, which preceded the SBP surge peak and reflects central autonomic (feed-forward) cardiac control (Silvani et al., 2011), was blunted during wakefulness in NC patients compared with CS (Fig. 2). The combined impairment of both baroreflex and non-baroreflex cardiac vagal control in NC patents during wakefulness may explain why, during wakefulness, the CCF analysis did not reveal any difference in the relative contributions of the baroreflex and central commands to HP variability between NC patients and CS (Fig. 3). A decrease in vagal modulation of HP during wakefulness may be expected to reduce total HP variability (SDNN, Fig. 1b) and to increase sympathovagal balance (LF/HF, Fig. 1d), shifting it away from vagal dominance in NC patients. Either effect fell just short of statistical significance ( 0.1), which may have been a type II statistical error because of small sample size. Significant increases in LF/HF during wakefulness in NC patients, in fact, have been reported in some (Ferini-Strambi et al., 1997; Grimaldi et al., 2010) but not all (Fronczek et al., 2008) previous studies. Conversely, mean HP was lower in NC patients than in CS during wakefulness and sleep (Table 1). This may be relevant for the present study because lower mean values of HP may be associated with lower values of indices based on HP variability, such as the BRS (Wesseling, 2003). Experimental autonomic stimulation in dogs (Bailey et al., 1996) indicates elegantly that this association reflects a physiological rather than a simple mathematical link between autonomic tone, which affects the mean HP value, and autonomic modulation, which affects HP variability.

Our present findings contribute to the understanding of NC as a multi-faceted disease that affects not only sleep–wake behaviour, but also autonomic control. Structural or tonic defects in cardiac vagal modulation in NC patients are unlikely to explain our findings because these patients showed normal results of classic autonomic function tests (Grimaldi et al., 2010). Rodent models of hypocretin deficiency may show subtle differences in HP control during sleep, but have no deficit in cardiac baroreflex during wakefulness (Silvani et al., 2012), thus not supporting a direct causative role of hypocretin deficiency in our present findings. NC patients also did not show any significant difference in total SBP variability (Fig. 1a) with respect to CS, strengthening the view that physiological sleep-dependent changes in SBP variability are not hypocretin-driven (Silvani et al., 2012). PLM occur frequently in NC patients (Dauvilliers et al., 2011; Ferri et al., 2006; Grimaldi et al., 2012) and entail distinct changes in cardiovascular variability (Dauvilliers et al., 2011), potentially contributing to our present findings. Analysis of the PLMW and PLMS indices in our population lent support to this hypothesis during sleep, but not during wakefulness before sleep. The CCF analysis may be particularly sensitive to the cardiovascular correlates of PLM during sleep, which entail a feed-forward pattern of HP control (Silvani et al., 2011). Accordingly, the CCF analysis indicated that the extent to which HP variability was explained by feed-forward mechanisms was greater during NREM sleep in NC patients than in CS (Fig. 3). Conversely, decreases in cardiac vagal modulation and spontaneous BRS may occur in healthy subjects with prolonged sleep deprivation (Anders et al., 2010; Chua et al., 2012; Zhong et al., 2005), and similar derangements have been also reported in association with excessive daytime sleepiness in sleep apnoea patients (Lombardi et al., 2008). Excessive daytime sleepiness and the inability to maintain continuous sleep behaviour are often the main symptoms in NC patients (Dauvilliers et al., 2007). These considerations thus suggest targeted studies of the relationship between autonomic derangements and sleepiness in NC patients during wakefulness in different environmental contexts and/or with different activities, which our present study was not designed to assess. Conversely, analysis of spontaneous HP variability during wakefulness before sleep onset is a somewhat inexpensive procedure that may be applicable in multi-centre studies on large NC patient samples. Such studies would provide information on the prevalence of impaired cardiac vagal modulation among male and female NC patients. This may be of clinical interest, because decreases in cardiac vagal modulation and baroreflex sensitivity are linked to cardiovascular risk in specific clinical conditions, such as after myocardial infarction (Billman et al., 1982; La Rovere et al., 1998).

Our study had several limitations. Although substantial in light of the rarity of NC and the complexity of the recording protocol, the sample size was low in absolute terms and limited to male subjects, with slight differences in age and BMI between groups. We sought to limit type I statistical errors (i.e. spurious significant results) without excessively increasing type II statistical errors (i.e. inability to detect genuine differences) by combining analysis of anova state × group interaction effects with false discovery rate correction (Curran-Everett, 2000) of the conventional 0.05 significance level (cf. Methods). Sample size also precluded meaningful statistical comparison between the two NC patients who were drug-naive and the seven NC patients who were on advanced (>2 weeks) washout from modafinil treatment. Finally, the present analysis was limited to wakefulness shortly before night-time sleep onset and to night-time sleep episodes longer than 5 min and free of artefacts and Portapres calibrations, and thus may not capture fully the cardiovascular phenomena occurring during the 24 h in NC patients. The present study should thus be regarded as a hypothesis-generating work, and awaits confirmation on larger samples.

In conclusion, this study provides preliminary evidence that autonomic control of cardiac variability by baroreflex and central autonomic (feed-forward) mechanisms is altered in NC patients during spontaneous sleep–wake behaviour and particularly during wakefulness before sleep.

Disclosure Statement

The authors declare no conflicts of interest.

Ancillary