This study describes the variations of the PTA index in anaesthetized horses according to haemodynamic variations and to their physical status. The main findings revealed significant variations of the PTA index during hypotension and administration of dobutamine, but no significant variation of the PTA following cutaneous incision. Horses of the Colic group demonstrated lower PTA values for several predefined time-points in comparison with those of the elective group, whereas MAP did not differ between groups. Finally, the PTA index showed a fair performance to predict MAP changes, in particular with high specificity of a 25% increase in PTA in 1 min to predict a 10% decrease in MAP with 5 min. However, although statistically significant, the threshold of 1% decrease in ∆PTA to predict a 10% increase in MAP is not clinically relevant.
The analysis of heart rate variability (HRV) is a non-invasive method, which can detect the fluctuations in the autonomic input to the sinoatrial node and the activity of the individual components of the ANS . To evaluate the ANS, HRV uses a frequency domain-based analysis . Classically, three spectral domains are considered: the very low frequency (VLF) reflecting the peripheral vasomotor tone, thermoregulation and the rennin–angiotensin-system; the low frequency (LF) related to sympathetic and parasympathetic tones modulations; and the high frequency (HF) mainly associated with the parasympathetic tone and influenced by efferent vagal activity and respiratory sinus arrhythmia . Compared to other methods such as stress hormone dosage, HRV is simpler, non-invasive and can provide continuous recording . During anaesthesia, HRV has been used in humans for various purposes, in particular for the prediction of blood pressure variation , quality of recovery and the evaluation of analgesia nociception balance  . However, data remain conflicting with some studies reporting a failure to detect sudden shifts in ANS activity before hypotension and during recovery . The PTA index is similar to the ANI, validated in human medicine to detect perioperative nociception . The ANI index has also been used to predict intraoperative haemodynamic reactions  and hypotension caused by spinal anaesthesia , as well as a tool to investigate the process of emotional regulation in humans .
Hitherto, the use of HRV analysis in anaesthetized animal has been sparsely reported. Recently, the dynamic variation of the PTA index (∆PTA) has been evaluated in anaesthetized dogs, with a correct performance to predict haemodynamic reactivity associated with intraoperative nociceptive stimuli  , but to our knowledge, this index has not been evaluated in anaesthetized horse. In horses, HRV power spectrum has been reported to the power spectrum of humans, rats and pigs  with two main frequency ranges: a HF range set at 0.07–0.6 Hz and a LF range set at 0.01–0.07 Hz . The normal resting horse is considered as having a prevailing parasympathetic tone, which was confirmed by HRV analysis . A recent study in horses has reported that HRV is a sensitive and non-invasive method to detect sympathovagal stimulation during ocular surgery . It has also been used as a prognostic information for postoperative horses with severe gastrointestinal disease, with a significantly better performance for predicting poor outcome than HR at admission . However, HRV failed to demonstrate a correlation between preanaesthetic ECG variables and the recovery scores after general anaesthesia in horses .
Maintaining a stable cardiovascular function in the anaesthetized animal is often challenging particularly in large animals. In addition to the physical status and the recumbency, many drugs used in anaesthesia interfere with the cardiovascular system either directly or through their action on the ANS . Inhaled anaesthetics can induce hypotension through peripheral vasodilation and depression of baroreceptor responsiveness, α2-agonists may promote hypotension through a decrease of cardiac output and heart rate following inhibition of the sympathetic pathway . Conversely, intraoperative noxious stimulation may cause a shift toward sympathetic dominance .
In the present study, predefined time-points were chosen to allow a comparison between animals of different physical status undergoing different surgical procedures. The time-point TSS was designed to evaluate the stability of the signal without any surgical stimulation, whereas TCut was designed to evaluate the potential influence of a nociceptive stimulation on PTA. The time-points TPre−Hypo, TDobut and TPost−dobut were designed to assess the influence of hypotension and administration of inotropes on the index, as these events may influence the sympathovagal balance .
At steady-state (TSS), the absence of significant difference within and between groups was expected, as no surgical or pharmacological stimulus was carried out during this time-point.
In comparison with previous results in dogs, no significant variation was registered at TCut, . However, as no concomitant haemodynamic reaction was registered at this time-point, which is likely due to an adequate level of analgesia. This can be explained by the use of xylazine and morphine as premedicants. Xylazine mediates a sympatholytic action with a reported duration of action of 20 to 30 minutes , morphine has a reported plasma half-life of elimination of 1.6 hours . This association most probably provided an appropriate analgesia at the time of cutaneous incision. No other nociceptive time-points were selected as the main objective of the study was to compare the influence of the haemodynamic and health status on the index.
The variations of PTA observed during blood pressure variations appeared to be inversely related to those of arterial pressure, and thus, seem to follow modifications of the sympathovagal balance. During hypotension, the increase in PTA reflects a shift toward a parasympathetic predominance or a decrease of the sympathetic tone. Similar results have been reported in human medicine with the analgesia nociception index  . This shift was blunted by dobutamine initiation (TDobut) with a decrease in the PTA and a concomitant increase in blood pressure values noticed during this time-point. We assume that dobutamine administration, because of its sympathomimetic activity, caused a shift toward sympathetic predominance, as described after cardiac β1-adrenegic receptors stimulation .
In general, lower PTA values were found in the horses of the Colic group, in comparison with those of the Elective group. This is in accordance with a presumed predominance of the sympathetic tone in Colic horses, associated with the stress response due to the critical condition. Moreover, autonomic dysfunctions are common during septic shock or associated disease. Similar findings have been reported in an experimental model of sepsis  and in human patients presented with endotoxaemia; the patients presented an uncoupling of autonomic nervous system and cardiovascular function leading to an impaired sympathetic modulation and maintenance of blood pressure . Indeed, patients with autonomic dysfunction are described as presenting an ineffective baroreflex and impaired regulation of blood pressure, failing to compensate the anaesthetics-induced hypotension . This dysfunction may potentially be explained in the septic patients by the endotoxins which contribute to a vasodilatory effect, a downregulation of the sympathetic myocardial responsiveness and an alteration of autonomic reflexes  . Our findings confirm thereby a previous report where horses with gastrointestinal disease, especially with ischemic gastrointestinal lesions, had an increased sympathetic tone and a reduced HRV .
The variations of MAP at the different time-points did not differ horses of the colic group and those from the Elective group. These results are probably related to the blood-pressure directed therapy that was guided to optimize MAP and maintain a systemic haemodynamic pressure above 60 mmHg. An additional potential explanation could be attributed to the sympathetic activation associated with the early stages of sepsis .
The ROC analysis revealed a fair performance of the dynamic variation of PTA to predict a MAP variation, in particular a decrease in MAP following an increase in PTA. This result is, to some extent, in agreement with several human studies that reported a good performance of ANI to predict intraoperative haemodynamic reactivity and hypotension in human patients  . However, other studies failed to show such a similar performance for the ANI .
We acknowledge several limitations for this study. Because of the small number of horses, this study was probably underpowered to show significant differences in PTA variations in Colic and Elective groups. There was a lack of homogeneity in the inclusion criteria, leading to different surgical stimulations, and different intestinal lesions in the Colic group, which could have biased the homogeneity of PTA measurements. However, one main objective of the study was to assess the influence of health status on the PTA index, which had to be performed in a clinical setting. Thus, it was particularly challenging to obtain a homogeneity of the cases. In addition, several other factors may have influence the HRV analysis, including posture, medication, preoperative stress, noxious stimulation and different ages and breeds   . Even though surgical procedures with the same recumbency were chosen to limit the influence of the posture, the anaesthetic protocols were slightly different between group, with intravenous lidocaine that was added in the Colic group and acepromazine that was used for horses of the Elective group. As this study was performed in a clinical setting, it was first difficult to change the anaesthetic protocol routinely used by the clinicians, and we were unable to provide a reference technique in order to analyse HRV and validate our results. Further studies should be considered in a more standardized condition to evaluate the performance of the PTA index to anticipate nociception in horses.