In the current study, based on its strong correlation with MAC, STD HR seemed to be the most useful HRV parameter to indicate insufficient depression of nociception or insufficient anesthetic depths. However, HRV is significantly influenced by drug choice, thereby limiting the utility of standard HRV parameters across different anaesthetic protocols.
MAC
The MAC concept is a standard method for evaluating the efficiency of inhalant anaesthetics [14], enabling the results to be reproducible and comparable. One MAC is defined as the end-tidal concentration of an anaesthetic agent that prevents gross muscular movement in response to a painful stimulus [15]. Several stimulation techniques exist for MAC determination, especially in veterinary anaesthesia, including the tail clamping method and electrical stimulation protocols. Both techniques have been validated and compared in dogs and rabbits, with similar results [16], but to our knowledge, these techniques have not been validated in cats. The constant voltage electrical stimulation protocol as described for dogs was used in our study after confirmation in pre-trials that maximal stimulation was achieved in cats as well. The individual MAC values of the cats in group I with a mean MAC of 1.83 vol%were close to those reported in the literature [17]. This implies comparability of the stimulation method.
Remifentanil is a potent opioid with a short mean half-life of 15.7 min in cats [18]. Due to its pharmacological features, including fast elimination via extrahepatic metabolism, a cumulative effect is unlikely. Studies concerning the isoflurane-sparing effect of remifentanil in cats show some variation in their results [19]. Generally, the effect seems to not be as high as that in other species, such as rats [20] and dogs [21]. One study did not reveal any MAC-sparing effect from several remifentanil CRI dosages in cats undergoing isoflurane anaesthesia [22], whereas other studies found a relatively constant MAC-sparing effect between 23 and 30% [23] or a reduction of 15.6% [24]. In our study, the MAC in group IR was slightly reduced (9.8%), but there was a large variation among individuals, resulting in the lack of statistical significance. The aforementioned MAC increase in two cats in group IR can be explained by the central stimulating effects of opioids in cats [19]. The dosage of 18 μg/kg/h was chosen in accordance with clinical experience as well as published doses. Previous experiments with cats did not reveal any beneficial effects on isoflurane requirements at higher CRI dosages, implying a possible ceiling effect [24]. Several studies have been performed to describe pharmacological variables as well as clinically useful dosages. Because these studies had different goals, their proposed CRI rates are not comparable. One study stated a constant infusion rate of 42 μg/kg/h as the median analgesic effective dosage [22], whereas in another study, a dosage of 13.8 μg/kg/h was sufficient for ovariohysterectomy, and 18 μg/kg/h prevented movement after supramaximal stimulation. However, a possible mismatch between anaesthetic immobility and analgesia must be considered [22]. Dexmedetomidine is an α2-adrenoceptor agonist with analgesic and muscle relaxing properties, and both features can influence the MAC. The MAC-sparing effect in group ID was 55.2% in our study. This finding corresponds to the findings from studies on other species. An epidural administration of dexmedetomidine was able to reduce the isoflurane requirements up to 33% in dogs [25], and a CRI of dexmedetomidine in dogs at the same rate as in the present study had a MAC-sparing effect between 41% [26] and 59% [27]. In cats, pharmacokinetic studies have shown a dose-dependent reduction in isoflurane requirements with an estimated maximal sparing effect of approximately 80% [28]. Because these were studies with target-controlled dexmedetomidine infusions [28, 29] or studies that included only a short-term infusion of 2 μg/kg/min over 5 min [30], their results cannot be directly compared to ours.
The equilibration phase in our study (the time between starting the remifentanil or dexmedetomidine CRI and the first noxious stimulus) was set to 60 min. The aforementioned short half-life of remifentanil indicates a time period of approximately 75 min to reach steady state (5 times the half-life) [31]. Therefore, steady state was not necessarily reached at the first stimulus in group IR but was likely achieved by the following stimuli and measurements. In contrast, because the half-life of dexmedetomidine is much longer, at a mean of 198 min [30], a steady state at the beginning was unlikely reached in any of the experiments in group ID. A bolus, on the other hand, might have led to undesired high concentrations. This should be considered a possible source of error in our results for group ID as well as group IR, especially because plasma concentrations of dexmedetomidine and remifentanil were not determined.
Another considerable limiting factor is the calculation of the different MAC levels. The amount of isoflurane was changed from 1.0 MAC to 0.75 MAC or 1.5 MAC, but the CRI remained the same, which might have led to different anaesthetic planes when comparing the groups.
Epoch lengths
The circumstances during surgical procedures can change rapidly, and anaesthesia often requires spontaneous adjustments. Therefore, if HRV analysis is considered for anaesthetic monitoring, short measurement epochs or time-varying analysis methods are desirable to enable the detection of changes as early as possible [5]. In contrast, the guidelines presented by The Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology [3] recommend longer intervals for adequate HRV analysis, including a minimum of 1 min for HF and 2 min for LF analysis, as well as a general recommendation of 5-min intervals. Newer studies did not show disadvantages in the use of even shorter epochs in the evaluation of some time-domain parameters such as SDNN [32, 33]. Two-minute epochs have been used in a recent study with beagle dogs [26]. In particular, cats tend to show movement or signs of intraoperative awareness very quickly and, sometimes, unexpectedly. Three epochs with a maximum of 3 minutes were analysed in this study. Because the HRV values of the different prestimulation epochs did not differ, the 1-min epochs were used for further statistical analysis. Additionally, nociceptive stimulation could be detected in only the 1-min poststimulation epoch measurements, which were used for further statistical analysis, while longer measurements usually revealed a re-adjustment to the prestimulation values.
HRV measurements are also affected by CO2 and respiratory rate [34]. Both were held in physiological ranges as close as possible. Still, a confounding effect cannot be excluded, especially in these ultra-short-term measurements.
Drug influences and HRV analysis
Three different anaesthetic, sedative or analgesic drugs were used in this study, each of which influence HRV parameters, as shown in different species [6, 10, 35,36,37]. Nevertheless, because all groups except group I were treated with combinations of anaesthetics and there was no awake control group, the exact impact of each drug in this study can only be assumed.
The consistent anaesthetic factor throughout all groups in our study was the inhalant anaesthetic isoflurane. Changes in HRV under isoflurane anaesthesia have been described in several studies with humans [6, 7], but few data are available concerning its special influence on HRV in other species. Data from a study with beagle dogs [26] show a dose-dependent effect of isoflurane on HRV parameters, including low HF and high LF values as well as continuously high heart rates. These findings are quite similar to those of the present study and are likely a result of an increase in sympathetic activity, which can occur as a reflex to isoflurane-induced low systemic vascular resistance [38, 39]. In contrast, even more than those in the beagle study, the measured HRs of the cats in group I were reduced at higher MAC levels, which might be explained by a reduction in sympathetic tone at higher isoflurane concentrations [39, 40]. Nevertheless, the examination of conscious cats compared to anaesthetized cats was not part of the present study, and therefore, the effect of isoflurane on HRV analysis cannot completely be quantified.
Generally, there are two ways HRV analysis can be influenced by the addition of drugs other than the given volatile anaesthetic. First, the additive affects the required amount of inhalant for an equipotent effect on the MAC, lowering the influence of the inhalant on HRV. Second, the potency of the additive in changing HRV must also be considered.
Remifentanil is a strong analgesic that acts mainly via μ1-agonism [41]. Its effects on cardiovascular variables, but not on HRV parameters, have been previously examined in cats [42]. In the same study, bradycardia and reduced blood pressure were commonly observed side effects. In contrast to that study as well as to the aforementioned beagle study [26], mean HRs in group IR were not significantly reduced compared with those in group I at the corresponding MAC level. Instead, even higher HRs were observed at 0.75 and 1.0 MAC. This observation might be explained by opioid-induced excitatory effects and the following increase in sympathetic tone in cats [23], whereas the lower HRs at 1.5 MAC are likely the result of the relatively increased and now predominant influence of isoflurane. However, as an overall impression, most of the HRV parameter values of group IR in the present study are close to those of group I, which indicates that either there were no pronounced effects of remifentanil on HRV parameters or the effects were similar to those of isoflurane.
In contrast to remifentanil, dexmedetomidine reduced the HR at all MAC levels. Additionally, in agreement with results reported from beagle dogs [26], the highest STD HR values, as well as the lowest LF n.u. and highest HF n.u. values, were found at 0.75 MAC, leading to the lowest LF/HF ratio at all MAC levels. Dexmedetomidine reduces sympathetic tone via central α2-adrenoceptor agonism and peripheral α2β-adrenoceptor activation, leading to increased vascular resistance and, as a reflex, to a decrease in heart rate [43, 44]. The prominent MAC-sparing effect, which was present in group ID, implies a great presence of a cardiovascular-suppressing α2 agonist and therefore explains the constantly low prestimulation HR in group ID throughout all MAC levels. Because minor changes of ETISO were enough to go from 0.75 MAC to 1.0 or 1.5 MAC, only slightly different anaesthetic depths were achieved; therefore, most HRV parameters failed to discriminate between those levels. For the same reason, nociceptive stimulation could be depicted by HR and other parameters throughout all MAC levels in group ID. However, in contrast to group I and IR, STD HR was not a reliable indicator of nociception at all MAC levels. The most likely explanation for the increase of STD HR after nociception is sympathetic activation, leading to more overall ANS activity.
HRV analysis can be performed by obtaining time domain and frequency domain parameters. Sometime domain parameters, such as the SDANN, SDNNI and HRV Triangular Index, are useful only for longer measurements. For other parameters, such as the SDNN, NN50, pNN50 or RMSSD, shorter epochs of 60–240 s have been studied [2]. SDNN, or STD RR, as named in the Kubios program, is believed to show more reliability in longer time periods up to 24 h. STD RR displays the standard deviation of NN intervals, whereas STD HR shows the standard deviation of the instantaneous HR. The latter does not necessarily show the same changes as STD RR because it is the mean of only very few heartbeats and does not use the exact RR interval time. Nevertheless, it is interesting that in our study, STD HR showed greater correlations. In the frequency domain, the choice of the studied frequency bands (HF, LF, VLF and ULF) also depends on the measured epoch lengths. ULF and VLF are not meaningful in the context of 1 min epochs and have therefore not been examined in the present study, and only the aforementioned shorter-term time domain parameters have been considered for statistical analysis. Both domains are influenced by both branches of the ANS in a complex manner [2], and none of those domains or parameter can be regarded separately.
Currently, there are no defined standards for HRV frequency bandwidth in cats. For frequency domain analysis, the HF, LF and VLF bands of the present study were chosen in accordance with previous HRV studies with cats [10, 45, 46]. In most studies with other species, the HF band is usually set at 0.15–0.4 Hz but can be expanded to lower than 0.15 Hz and up to 1 Hz. The HF band is thought to mainly reflect influences of respiratory sinus arrhythmia. Because cats tend to have slightly higher respiratory rates and according to the rates found in the present study, the HF band was set to 0.15–0.833 Hz (equivalent to a respiratory rate of approximately 10–50/min). These settings have been found to work best for consistent conditions. Despite artificial ventilation, as provided in the present study, most cats showed spontaneous breathing attempts, especially at low MAC levels, after nociceptive stimulation. Irrespective of the interference with our measurements as mentioned above, this limiting factor must be considered for possible anaesthetic monitoring via HRV analysis in clinical settings.
Generally, as seen in Additional file 1: Table S1, there was a large individual overlap, especially in frequency domain parameter values within treatment groups or MAC multiples, as already reported in dogs [26]. Therefore, the ratios of these spectral portions might be more useful for the interpretation and inter-individual comparison of anaesthetic depth than absolute numbers.
Of note, we did not screen the cats via echocardiography before the experiments. Although none of the cats showed any abnormality at the clinical examination, including auscultation and baseline ECGs, it is still possible that underlying heart diseases might have been overlooked.
Few correlations have been found in the present study. The strongest correlation with MAC was found in STD HR. The correlation coefficient value of r = − 0.76 is good, but not as high as expected. The regression analysis results of our study indicate a non-linear relationship of MAC and STD HR. This should be considered as a possible limitation if HRV parameters are used for real time anaesthetic depth monitoring in a clinical setting.
Changes in several HRV parameters within groups were observed with increasing MAC and after nociception, supporting our hypothesis. Nevertheless, these changes were not as big as expected. The comparison between the treatment groups at the same MAC levels resulted in a wide variation of HRV parameters with decreasing MAC, which might be explainable by the preponderance of isoflurane effects at higher MAC levels.
A limitation of this study was, that the number of animals in this orientation study has been chosen in accordance with previous data from dogs [26] and for ethical reasons. The high variability of HRV parameters across the different protocols might have contributed to limited detection of true differences and underpowered results.