Cerebral autoregulation after aneurysmal subarachnoid haemorrhage. A preliminary study comparing dexmedetomidine to propofol and/or midazolam

Cerebral autoregulation is often impaired after aneurysmal subarachnoid haemorrhage (aSAH). Dexmedetomidine is being increasingly used, but its effects on cerebral autoregulation in patients with aSAH have not been studied before. Dexmedetomidine could be a useful sedative in patients with aSAH as it enables neurological assessment during the infusion. The aim of this preliminary study was to compare the effects of dexmedetomidine on dynamic and static cerebral autoregulation with propofol and/or midazolam in patients with aSAH.


| INTRODUC TI ON
Aneurysmal subarachnoid haemorrhage (aSAH) is a detrimental disease often affecting a relatively young population and leading to severe disability. Cerebral autoregulation maintains the cerebral blood flow constant despite changes in mean arterial pressure by changing cerebrovascular resistance. Dynamic and static autoregulation reflect rapid responses to changes in pressure pulsation and slow responses to changes in mean arterial pressure, respectively. 1 Previous studies have reported that impaired cerebral autoregulation in patients with aSAH predicts the development of radiological vasospasm and delayed cerebral ischemia (DCI); and may correlate with poor outcome. [2][3][4] Dexmedetomidine is a selective α 2 -agonist which induces sedation, anxiolysis and analgesia with minimal respiratory depression.
Importantly, it allows patient awakening despite uninterrupted infusion enabling frequent neurological evaluation. 5 These properties and potential neuroprotective effects make dexmedetomidine desirable in neurocritical care. 6 While dexmedetomidine seems to attenuate dynamic cerebral autoregulation in healthy volunteers, 7 its effects on cerebral autoregulation in aSAH patients are unknown.
Thus, the primary aim of this preliminary study was to compare dynamic and static cerebral autoregulation in patients with aSAH under dexmedetomidine vs propofol and/or midazolam sedation.
We did not aim to study the effect of aSAH on autoregulation per se.

| ME THODS
The study was conducted in the Intensive Care Unit in Turku University Hospital, Turku, Finland. The study protocol (EudraCT 2012-000068-11, ClinicalTrials.gov identifier NCT01664520) conformed to the revised Declaration of Helsinki 8  Written informed consent was obtained from the next of kin.
Patients aged 18-80 years suffering from aSAH and requiring sedation and mechanical ventilation after aneurysmal treatment (coiling or clipping) were included. aSAH patients not requiring sedation or mechanical ventilation were excluded. Other exclusion criteria were pregnancy, nursing women, sick sinus syndrome, carotid stenosis, heart rate <50 beats/min, mean arterial pressure <55 mm Hg, baseline middle cerebral artery (MCA) flow velocity (V MCA ) ≥120 cm/s suggesting vasospasm and clinical signs of DCI.
All patients received intravenous nimodipine infusion and were treated according to Neurocritical Care Society recommendations with local modifications. 9 ICP was measured continuously using intraparenchymal sensors, either Neurovent-PTO (Raumedic AG, Helmbrechts, Germany) or Codman® ICP Monitoring system (DePuy Synthes, Wokingham, UK). Neurovent-PTO was used in seven patients whereas Codman was used in one patient. Of the eight patients with intraparenchymal ICP catheter, four patients also had an external ventricular drainage (EVD) due to hydrocephalus. Two patients had only EVD and in those patients, cerebrospinal fluid was continuously drained and ICP was measured periodically. Cerebral oxygenation was monitored continuously using bilateral near-infrared spectroscopy (NIRS) with INVOS TM cerebral oximetry. In patients with Neurovent-PTO, brain tissue oxygenation (PtiO 2 ) was also monitored. Transcranial Doppler (TCD) was used to measure bilateral V MCA using 2 MHz-probes (Atys, Soucieu-en-Jarrest, France) fixed with a head frame.

| Study protocol
First, static and dynamic autoregulation were assessed during a constant rate of propofol (2-5 mg/kg/h) and/or midazolam infusion (0.03-0.3 mg/kg/h) (baseline). Thereafter propofol and/or midazolam infusion was stopped and dexmedetomidine (Dexdor ® Orion Oyj, Helsinki, Finland) infusion was commenced at a dose of 0.7 μg/ Conclusions: Compared to propofol and/or midazolam, dexmedetomidine did not alter static cerebral autoregulation in aSAH patients, whereas a significant change was observed in dynamic SA. Further and larger studies with dexmedetomidine in aSAH patients are warranted.

Editorial Comment
There may be concern for effects of potent vasodilatory drugs on cerebral vascular tone and responsiveness in patients with intracranial injury. The authors report here that cerebral blood flow autoregulation was unchanged in postaneurysmal subarachnoid haemorrhage patients who were sedated with dexmedetomidine compared to sedation with midazolam or propofol.
F I G U R E 1 Study protocol kg/h according to the total body weight. After two hours of dexmedetomidine infusion at constant rate (in order to reach a steady state), static and dynamic autoregulation were assessed again. While obtaining the steady dexmedetomidine state, all patients received 1-3 mg intravenous doses of oxycodone for analgesia at the discretion of the ICU nurse treating the patient. Next, dexmedetomidine infusion was increased to 1 μg/kg/h for two hours following which, static and dynamic autoregulation were assessed. Finally, dexmedetomidine dose was increased to 1.4 μg/kg/h for another two hours and autoregulation tests were performed once again (see Figure 1). After each period of two hours of dexmedetomidine infusion, arterial blood was sampled to measure the plasma dexmedetomidine concentration (details in supplementary material). During the autoregulation tests, no opioids or other drugs were administered.
No blinding was performed. To avoid the effects of temperature, haemoglobin (Hb) and ventilation on cerebral blood flow velocity, these variables were kept constant throughout the study. Transient hyperaemic response ratio (THRR) was calculated as previously described. 11 The baseline FV was systolic FV before compression. Due to overshoot phenomenon, the first cardiac cycle was ignored and the systolic FV hyperaemia was the average of the systolic FV of the second and third cardiac cycles after the compression. 11 THRR was repeated three times on both hemispheres and there was at least 60 seconds interval between the measurements. The data were averaged from both hemispheres 13 and considered normal if THRR was ≥1.09 and impaired if THRR was <1.09. 10 Strength of dynamic autoregulation (SA) was calculated as where systolic FV hyperaemia is the first flow velocity after the release of carotid compression and MAP of 60 mm Hg represents the lower limit of autoregulation. MAP is the mean arterial pressure immediately before the THRT and FV baseline is the flow velocity before carotid compression. 14 SA value <1 suggests under regulation and >1 hyperregulation. 15 In the underregulated situation, cerebral blood vessels are already dilated and a decrease in blood pressure does not result in further dilation, whereas in hyperregulation the cerebral blood vessels dilate more than should be necessary. 15 However, in those patients who had only EVD, we used MAP (as the CSF was continuously drained) in the formula and sROR% was calculated:

| Static autoregulation test
sRoR of 100% or more indicates that autoregulation is completely intact, meaning that cerebral blood flow velocity is independent of cerebral perfusion pressure. 17 sROR of 0% indicates that cerebral autoregulation is completely absent, and cerebral blood flow is linearly related to cerebral perfusion pressure. sROR of 50% is regarded as the cut-off for failure of autoregulation. Accordingly,

| Statistical analysis
Data are presented as mean ± SD. All physiological and TCD parameters including autoregulation indices at various dexmedetomidine doses were compared to their respective baseline values recorded with propofol/midazolam. Power analysis was not performed, and our sample size was based on previous similar autoregulation studies using TCD measurements. [19][20][21] The initial aim was to recruit 15 patients, but the recruitment was unexpectedly difficult (due to refusals of next of kin or suspicion of vasospasms in TCD) and very slow

| RE SULTS
Five male and five female patients with mean age of 58.4 ± 10.5 years were recruited. One patient was excluded from dynamic and static autoregulation tests as she had no temporal acoustic window for the TCD, leaving nine patients for the analysis. The demographic data for the patients is shown in Table 1. There was no significant difference between the Hb, PaO 2 and PaCO 2 values at each concentration of dexmedetomidine compared to baseline (Table 2). MAP and CPP were controlled with a noradrenaline-infusion.
Overall, the SA was significantly lower after the 1.  There was no difference in cerebral oxygenation (PtiO 2 and NIRS, Table 4) between baseline and different doses of dexmedetomidine. Figure 5 shows a Bland-Altman plot to compare the agreement of the sROR% measurements using MAP or CPP.
We performed a subgroup analysis of the data for the five patients who had TCD measurements from both sides. We found no statistical differences in the autoregulation indices at any dexmedetomidine dose compared to baseline (data not shown).
The cumulative dose of propofol before the commencement of dexmedetomidine was 6004 ± 4165 mg (N = 7) and that of midazolam was 125.7 ± 79 mg (N = 3). The mean dexmedetomidine concentrations were 0.7 ± 0.22 ng/mL (after 0.7 μg/kg/h for 2 hours),  Moreover, some anaesthetic agents impair dynamic but not static cerebral autoregulation. 24,25 There are multiple methods for investigating cerebral autoregulation. These include examining the recovery of CBFV after a rapid decrease of arterial blood pressure by using thigh cuff test [26][27][28] and examining spontaneous fluctuations in CBFV in relationship with MAP, ICP and brain tissue oxygenation (Cambridge Enterprise Ltd, University of Cambridge, UK). 29 We used the THRT for assessing dynamic autoregulation because it allows brief period of reduction in CBFV and is deemed safe in aSAH. Furthermore, we calculated SA, which is not affected by hemodynamic factors. 15 Studying static cerebral autoregulation is traditionally performed by increasing MAP with a vasopressor and simultaneously measuring MAP and MCA flow velocities. 27 Modern techniques use PET scanning, which images CBF. 30 However, we preferred TCD to be able to perform the testing on the patient's bedside and to avoid the need to transport intubated and sedated patients for imaging.

| D ISCUSS I ON
In healthy, anaesthetised patients, there is a correlation be-  dynamic strength of autoregulation) and static autoregulation when static rate of autoregulation is used. 27,31 However, various methods are probably not interchangeable 32,33 and there is no general consensus on how autoregulation should be monitored. 29 Dynamic autoregulation has previously been shown to be impaired in healthy volunteers after administration of dexmedetomidine. 7 In this study, we found that the SA was significantly lower after the 1.0 μg/kg/h dose of dexmedetomidine compared to baseline, but not after the initial 0.7 μg/kg/h or the highest dose 1.4 μg/ kg/h. It is possible that the effect of dexmedetomidine on dynamic autoregulation may be dose dependent in this population. Also, the ICP was not elevated in our patients and hence, they may have been less susceptible to any autoregulatory impairment with dexmedetomidine. Elevated ICP is known to correlate with impaired cerebral autoregulation. 34 In previous volunteer studies, dexmedetomidine administration was associated with decrease in MAP, cardiac output and CBF. 7,35 In contrast, we kept the blood pressure constant during dexmedetomidine infusion and THRT which may explain to some extent why there were no changes in most of the autoregulation indices. Dexmedetomidine induced reduction in CBF has been shown to be less in patients with traumatic brain injury if blood pressure is maintained at the pre-sedation level. 36 We used noradrenaline infusion in maintaining and increasing MAP and CPP levels. 37 Consequently, it is possible that autoregulatory impairment was not detected. It is possible that autoregulation was intact within the blood pressure range studied but may have been impaired outside of that range.
During the first days after aSAH, many patients have low CBF regardless of vasospasm or DCI. 38  Ogawa et al found that midazolam seems to improve whereas propofol seems to have no effect on dynamic cerebral autoregulation in healthy volunteers. 39 In patients with traumatic brain injury, large, doses of propofol (more than 4 mg/kg/h) have been shown to impair static autoregulation. 30 We discontinued propofol 2 hours before the first and almost 6.5 hours before the last autoregulation assessments. After > 12 hours of infusion of propofol, it takes 3.5 hours to reduce the plasma concentration to 20%. 40 Therefore, the likelihood of residual propofol affecting cerebral autoregulation is small. Midazolam, on the other hand, may have had an effect on the dynamic autoregulation in our patients but only three patients were administered midazolam. Next, TCD only provides an estimate of CBF and it is assumed that the diameter of basal cerebral arteries varies only minimally with changes in MAP. 41,42 This presumption has been recently challenged and TCD observations should be carefully interpreted for CBF. 43 Fourth, we performed the study during the first days after aSAH, before the DCI usually appears and it is possible that autoregulatory status may change over time.
In conclusion, our preliminary findings indicate that in patients

ACK N OWLED G EM ENTS
We thank Mikko Kuoppamäki and Peter Makary from Orion Corporation for providing the analysis of dexmedetomidine concentrations. We also thank the nursing staff at the Intensive Care Unit in Turku University Hospital for collaboration and patience during the study.