Delta sleep-inducing peptide (DSIP) is an endogenous sleep-associated hormone, first characterized in 1977 [1–3], that enhances delta wave activity in the electroencephalogram (EEG). DSIP is present in blood, crosses the blood–brain barrier, and has been used therapeutically in the treatment of insomnia [4,5].
Direct administration of DSIP to the brainstem of anaesthetized cats was reported to modulate heart rate variability (HRV) . This was likely to be a direct effect on the dorsal vagal nucleus, nucleus ambiguus and/or the nucleus tractus solitarius, which together constitute the vagal complex of the medulla oblongata and are responsible for the control of HRV. It is known from a functional imaging study (positron emission tomography) that volatile anaesthesia affects glucose metabolism in this vagal complex of the brainstem in humans in an asymmetric (right side of the brainstem), dose-dependent manner  when vagal tone was assessed using HRV.
Bispectral index (BIS) is a proprietary, weighted composite of processed EEG derivatives presented as a number from 0 to 100 , widely used to indicate the depth of anaesthesia. Certain multichannel BIS monitors enable the recording of bilateral EEG derivatives, in addition to BIS.
Based on the literature, we considered that DSIP was a putative natural hypnotic. The principal objective of our studies was to determine whether DSIP has a therapeutic role as an adjunct to hypnotic anaesthesia. Our hypothesis was that DSIP would increase the overall depth of anaesthesia as measured using BIS and HRV. We are not aware of any prior literature by other groups reporting the effects on EEG or HRV of co-administration of DSIP with a volatile anaesthetic in humans.
The design of this study was a randomized, controlled pilot to characterize the effect of DSIP on bilateral components of the EEG and HRV in patients undergoing routine elective surgery. No power calculations were possible, as DSIP has been given in conjunction with anaesthesia in neither animals nor humans. Twenty-four ASA I or II, right-handed female patients scheduled for minor elective surgery gave written, informed consent to a protocol approved by the Central Manchester local research ethics committee, which entailed contact and distribution of information sheets at least 24 h before surgery. Patients were tested and had to be negative to a screen for drugs.
Clinical anaesthesia was standardized to propofol induction and isoflurane maintenance. Two anaesthetists attended. One anaesthetist made up the independently randomized test pack of DSIP (Clinalfa) provided by our hospital pharmacy, and administered the agent. The other anaesthetist administered a routine propofol induction/isoflurane maintenance surgical anaesthetic and was blinded to whether the agent was active or placebo.
The experimental procedure is shown in Fig. 1. Doses were chosen on the basis of prior therapeutic administration of DSIP reported in the literature for the treatment of insomnia . One of three doses of DSIP (25 nmol kg−1 = 21.2 μg kg−1; 50 nmol kg−1 = 42.4 μg kg−1 or 100 nmol kg−1 = 84.8 μg kg−1) was randomly administered intravenously to each patient in the active group using the contralateral arm to that used for injection of anaesthetic agents. The same dose of DSIP was administered once while awake and once during isoflurane anaesthesia. The control group received an equivalent volume of saline. In order to screen for adverse events, the first administration of DSIP after a 10 min baseline was in the form of a tolerance test, and was conducted while the patient was lying awake in the reception area of the operating rooms and monitored with standard ECG and noninvasive blood pressure equipment. A delay of at least 30 min allowed for washout of DSIP after the awake tolerance test, and before induction of anaesthesia.
In addition to standard monitoring (SpO2, noninvasive blood pressure, and ECG), a series of independent monitors were attached to the patient for continuous, automated high-resolution logging to a PC. These comprised EEG (Aspect Medical Systems A-1000 with BIS v3.3, bilateral two-channel referential montage 10/20 locations Fp1–F7, Fp2–F8 with ZipPrep electrodes and matched impedances, updating every 15 s); ECG at 1 kHz (CED 1401 with Digitimer Neurolog amplifier); and end-tidal and inspired gas concentrations (Datex Ultima-1). Bilateral EEG electrode-derived data included the following parameters (as described by the manual for the A-1000 monitor): unfiltered BIS; absolute delta power (0.5–3.75 Hz measured in decibel with respect to 0.001 μV2); relative delta power (% of delta power within the total power); total power (0.5–30.0 Hz measured in decibel with respect to 0.001 μV2); electromyogram (EMG) band (70–110 Hz measured in decibel with respect to 0.001 μV2); burst suppression (percentage of epochs in the past 63 s in which the EEG signal is suppressed, i.e. free of any waveform ); signal quality index (percentage of artefact-free epochs in the last minute that could be used for BIS calculations); spectral edge frequency (SEF; the frequency derived from spectral analysis of the EEG waveform below which 95% of the power is present); median frequency [the frequency derived from spectral analysis of the EEG waveform that is at the median (50%) of the power distribution]. Ten-minute epochs of EEG, HRV and gas data were identified after administration of DSIP or saline. Data were analysed off-line using standard software (CED Spike2 version 4.0 and Microsoft Excel). Tachygrams of instantaneous ECG frequency were calculated from R–R wave intervals resampled at 4 Hz. A 1024-point fast Fourier transform-based power spectral analysis was performed on these tachygrams in order to determine low-frequency (LF; 0.032–0.138 Hz; sympathetic and parasympathetic activity) and high-frequency (0.15–0.5 Hz; predominately vagal parasympathetic activity) HRV from 5 min epochs of ECG, as previously described . Data were statistically analysed using standard software (SPSS version 15). All EEG data were marked as ‘missing’ when the signal quality index was less that 95.
Normality of data was rejected using the Kolmogorov–Smirnov test, and the Mann–Whitney U nonparametric test was subsequently used to compare the randomized active (DSIP) versus control (saline) groups. The effect of DSIP in the awake patients was assessed by calculating in each patient the difference between BIS before and after DSIP administration. Coincident measurements of paired, processed bilateral unsmoothed raw EEG (Aspect A-1000) collected in the same patients every 15 s were studied using a Wilcoxon signed ranked test. Statistical significance was assigned with P less than 0.001 for EEG data and P less than 0.05 for HRV data.
DSIP was well tolerated in participants while awake; there were no adverse reactions. There was no significant change in heart rate (HR) or HRV when DSIP was administered to patients while awake (data not shown). The Wilcoxon signed rank test showed that asymmetry in left and right paired BIS values was significant with 25 nmol kg−1 and 100 nmol kg−1 DSIP (Fig. 2). This asymmetry was confirmed when changes in BIS relative to pre-DSIP baseline were calculated (Fig. 3) in the same patients. The relative BIS changes observed while awake were significantly different between 50 and 100 nmol kg−1 DSIP (Fig. 3).
There was no significant difference in the level of end-tidal isoflurane administered between each of the experimental or control groups (data not shown). When anaesthetized, administration of 25 nmol kg−1 DSIP resulted in a significant increase in mean HR (from 73.6 to 92.4 beats min−1; P < 0.05), whereas HRV significantly decreased; the low-frequency area under the curve (LFAUC) fell significantly from 1.4 × 10−6 to 2.6 × 10−7 Hz2 (P < 0.05) and the high-frequency area under the curve (HFAUC) fell significantly from 3.010−8 to 4.7 × 10−9 Hz2 (P < 0.05). No significant changes were observed in HR or HRV at the other DSIP doses (50 nmol kg−1 DSIP: mean HR 62.8 beats min−1, LFAUC 7.2 × 10−7 Hz2, HFAUC 5.71 × 10−9 Hz2; 100 nmol kg−1 DSIP: mean HR 65.4 beats min−1, LFAUC 1.3 × 10−6 Hz2, HFAUC 2.2 × 10−8 Hz2).
Figure 4 shows the significant increase in bilateral BIS observed with 25 nmol kg−1 DSIP. Most notable was that no BIS values below 30 were recorded when 25 nmol kg−1 DSIP was administered, suggesting that a component of processed EEG within the BIS algorithm, and essential for coding deep anaesthesia, had been attenuated. No BIS level above 65 was observed when 100 nmol kg−1 DSIP was administered, at which level the significant asymmetry in BIS values observed at lower levels of DSIP was lost.
Figure 5 shows that total power of the EEG was significantly reduced with 25 and 100 nmol kg−1 DSIP compared with saline controls during isoflurane anaesthesia. Significant asymmetry in total power was observed at 0, 25 and 50 nmol kg−1 DSIP.
Figure 6 demonstrates that a significant reduction in the right relative delta rhythm was observed with 25 and 100 nmol kg−1 DSIP administered during isoflurane anaesthesia. Significant asymmetry in relative delta rhythm observed in the control groups was not seen on administration of DSIP.
Figure 7 demonstrates that the low levels of burst suppression (<2%) encountered during isoflurane anaesthesia were significantly reduced on the left side of the head with 25 nmol kg−1 DSIP, and eliminated with 50 nmol kg−1 DSIP. At 100 nmol kg−1 DSIP, bilateral levels of burst suppression significantly increased compared with saline controls.
Figure 8 demonstrates that significant asymmetry in the EMG frequency band (70–110 Hz) was present during anaesthesia. The magnitude of EMG asymmetry was most pronounced in non-DSIP controls and smallest with 100 nmol kg−1 DSIP. Significant increases in EMG power were observed with 25 nmol kg−1 DSIP compared with controls, with significant decreases in EMG power at 50 and 100 nmol kg−1 DSIP.
Figure 9 demonstrates that the unilateral level of SEF was not significantly altered by DSIP. Significant bilateral asymmetry of SEF was observed in anaesthetized controls and with 50 and 100 nmol kg−1 DSIP.
Median frequency was significantly reduced compared with controls in the left EEG montage during administration of 25 nmol kg−1 DSIP (Fig. 10). Paired analysis suggested significant asymmetry in median frequency in controls and those treated with 50 nmol kg−1 DSIP. The range of median frequency values was notably large in control and 50 nmol kg−1 DSIP groups.
DSIP did not induce hypnotic anaesthesia when administered as the sole agent. Our hypothesis was that DSIP could act as a hypnotic adjunct to isoflurane. Although we did observe some significant changes in HR, HRV, BIS and components of the EEG, these effects were relatively small and so the original hypothesis was not supported. It is possible that DSIP will have more clinically relevant synergistic actions with anaesthetic agents other than isoflurane. After data collection was completed, and initial analysis was performed, we were advised that a crossover rather than randomized, controlled experimental design would have been more appropriate for looking at the small changes in EEG we observed in this study. Such a design would have no controls, and analysis would be based on a comparison of EEG in each participant before and after intervention with DSIP. Although such retrospective analysis was possible on data collected from awake participants in this study (Fig. 3), the drawback with such a crossover design would have been an additional delay in surgery of 10 min while baseline EEG data were collected before administration of DSIP, which was impossible for most of the cases we studied (and not in the approved experimental protocol). Ideally, future research into DSIP should consider such a crossover design.
DSIP (25 nmol kg−1) rapidly, and significantly, reduced both LFAUC and HFAUC components of HRV in patients already anaesthetized with isoflurane, accompanied by a significant increase in HR. HFAUC is believed to be mostly parasympathetic in origin, whereas LFAUC is reported to be a combination of sympathetic and parasympathetic tone . A reduction in parasympathetic tone would be expected to increase HR, whereas an equal reduction in parasympathetic and sympathetic tone may not alter HR. Therefore, our results suggest that DSIP is effecting a small reduction in parasympathetic tone.
DSIP (25 nmol kg−1) caused a small, but statistically significant, rise in BIS during anaesthesia (Fig. 4). BIS is a complex analysis incorporating several different signal analyses of the EEG waveform, including several others apparently influenced by DSIP in this study, such as delta rhythm, burst suppression, SEF and median frequency; any combination of which may have been influenced by DSIP. The version of BIS studied here was 3.3, the most recent available for the A-1000 monitor. The A-1000 was used, rather than a more recent BIS monitor, as it had the capability to record asymmetry in BIS values during anaesthesia, and because the A-1000 is compatible with ZipPrep electrodes needed for a bilateral electrode montage, rather than BIS sensors designed for application to one side of the head. We noted a wide range of median frequency values in 0 and 50 nmol kg−1 groups that may be indicative of artefacts present within the median frequency data even with a signal quality index greater than 95, which was used as our automatic screen for artefacts.
The site(s) of action for the DSIP administered in this study is unknown. It is known that DSIP crosses the blood–brain barrier . DSIP is reported to increase (gamma-aminobutyric acid) GABAA-activated currents, and to inhibit N-methyl-D-aspartic acid activated currents [10,11]in vitro (cultured cerebellar and hippocampal neurones). Considering the small effect on parasympathetic tone that we have described and the effect of DSIP reported when administered directly to the brainstem of the cat , a putative site for such a central effect in vivo would be the vagal complex of the medulla oblongata, especially the dorsal vagal nucleus and nucleus ambiguus.
We were surprised by the degree of asymmetry in BIS observed in this study, even with no DSIP. This asymmetry in BIS was accompanied by asymmetry in the facial EMG and was also present in multiple EEG parameters. The magnitude of EMG asymmetry was most pronounced in the absence of DSIP, but the greatest EMG activity was observed with 25 nmol kg−1 DSIP. Facial EMG originates neurally in the brainstem . This supports the suggestion that DSIP has a direct effect on this region of the brainstem. Future studies using functional imaging (e.g. positron emission tomography) of the brainstem during anaesthesia in volunteers treated with DSIP may help to locate the sites of action for this agent.
Whether DSIP has any clinical utility during anaesthesia, rather than just insomnia, remains to be demonstrated. We found some evidence that DSIP at 25 and 50 nmol kg−1significantly reduced burst suppression of the EEG, i.e. periods of isoelectric EEG during deepening depths of anaesthesia. Burst suppression constitutes a large part of the BIS algorithm when BIS is less than 40 , which is the region of BIS implicated in cumulative deep hypnotic time (BIS < 45) as described by Monk et al.. Reducing the level of anaesthetic agent would be the first course of action to reduce the amount of burst suppression, but perhaps DSIP or an agent with a similar profile could serve some role in moderating the level of burst suppression in a preventive manner.
We found that DSIP appeared to asymmetrically influence the level of EMG activity (Fig. 8). BIS changed in a similar way (Fig. 4), especially when asymmetry was taken into account. It has already been suggested that the EMG affects BIS levels when ZipPrep electrodes are used with an A-1000 monitor in a bipolar frontal montage  and the A-2000 monitor using a four-electrode Quattro Sensor . It would be of interest to study bilateral responses from the EMG, especially with respect to their influence on BIS and other EEG-derived indices. Should EMG demonstrate measurable asymmetry, this may provide additional methods to extract it from the EEG signal.
Synthetic DSIP is currently prohibitively expensive for all but research use, although a more potent derivative of DSIP is commercially available in some countries (Deltaran). We observed a reduction in parasympathetic HRV that may be of interest in certain specialties, such as cardiac anaesthesia and intensive care medicine. The apparent lack of side-effects and rapid breakdown of DSIP make it suitable for further study.
1 Monnier M, Dudler L, Gachter R, Schoenenberger GA. Transport of the synthetic peptide DSIP through the blood-brain barrier in rabbit. Experientia 1977; 33:1609–1610.
2 Monnier M, Dudler L, Gachter R, et al
. The delta sleep inducing peptide (DSIP). Comparative properties of the original and synthetic nonapeptide. Experientia 1977; 33:548–552.
3 Schoenenberger GA, Monnier M. Characterization of a delta-electroencephalogram (-sleep)-inducing peptide. Proc Natl Acad Sci U S A 1977; 74:1282–1286.
4 Pollard BJ, Pomfrett CJD. Delta sleep-inducing peptide. Eur J Anaesthesiol 2001; 18:419–422.
5 Schneider-Helmert D, Graf M, Schoenenberger GA. Synthetic delta-sleep-inducing peptide improves sleep in insomniacs. Lancet 1981; 1:1256–1257.
6 Pokrovsky VM, Osadchiy OE. Regulatory peptides as modulators of vagal influence on cardiac rhythm. Can J Physiol Pharmacol 1995; 73:1235–1245.
7 Pomfrett CJD, Alkire MT. Respiratory sinus arrhythmia as an index of anaesthetic depth: evidence from functional imaging studies. J Physiol (Lond) 1999; 518P:180.
8 Rampil IJ. A primer for EEG signal processing in anesthesia. Anesthesiology 1998; 89:980–1002.
9 Pomfrett CJD. R-R intervals and the depth of anaesthesia
. In: Adams AP, Cashman JN, editors. Recent advances in anaesthesia
and analgesia. Edinburgh: Churchill Livingstone; 1995. pp. 89–105.
10 Grigorev VV, Ivanova TA, Kustova EA, et al
. Effects of delta sleep-inducing peptide on pre and postsynaptic glutamate and postsynaptic GABA receptors in neurons of the cortex, hippocampus, and cerebellum in rats. Bull Exp Biol Med 2006; 142:186–188.
11 Sudakov KV, Umriukhin PE, Rayevsky KS. Delta-sleep inducing peptide and neuronal activity after glutamate microiontophoresis: the role of NMDA-receptors. Pathophysiology 2004; 11:81–86.
12 Sethi NK, Sethi PK, Torgovnick J, et al
. EMG artifact in brain death electroencephalogram, is it a cry of ‘medullary death’? Clinic Neurol Neurosurg 2008; 110:729–731.
13 Morimoto Y, Hagihira S, Koizumi Y, et al
. The relationship between bispectral index and electroencephalographic parameters during isoflurane anesthesia. Anesth Analg 2004; 98:1336–1340.
14 Monk TG, Saini V, Weldon BC, Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg 2005; 100:4–10.
15 Bruhn J, Boullon TW, Shafer SL. Electromyographic activity falsely elevates the bispectral index. Anesthesiology 2000; 92:1485–1487.
16 Liu N, Chazot T, Huybrechts I, et al
. The influence of a muscle relaxant bolus on bispectral and Datex-Ohmeda entropy values during propofol-remifentanyl induced loss of consciousness. Anesth Analg 2005; 101:1713–1718.