Differential responses of cerebral and renal oxygenation to altered perfusion conditions during experimental cardiopulmonary bypass in sheep

We tested whether the brain and kidney respond differently to cardiopulmonary bypass (CPB) and to changes in perfusion conditions during CPB. Therefore, in ovine CPB, we assessed regional cerebral oxygen saturation (rSO2) by near‐infrared spectroscopy and renal cortical and medullary tissue oxygen tension (PO2), and, in some protocols, brain tissue PO2, by phosphorescence lifetime oximetry. During CPB, rSO2 correlated with mixed venous SO2 (r = 0.78) and brain tissue PO2 (r = 0.49) when arterial PO2 was varied. During the first 30 min of CPB, brain tissue PO2, rSO2 and renal cortical tissue PO2 did not fall, but renal medullary tissue PO2 did. Nevertheless, compared with stable anaesthesia, during stable CPB, rSO2 (66.8 decreasing to 61.3%) and both renal cortical (90.8 decreasing to 43.5 mm Hg) and medullary (44.3 decreasing to 19.2 mm Hg) tissue PO2 were lower. Both rSO2 and renal PO2 increased when pump flow was increased from 60 to 100 mL kg−1 min−1 at a target arterial pressure of 70 mm Hg. They also both increased when pump flow and arterial pressure were increased simultaneously. Neither was significantly altered by partially pulsatile flow. The vasopressor, metaraminol, dose‐dependently decreased rSO2, but increased renal cortical and medullary PO2. Increasing blood haemoglobin concentration increased rSO2, but not renal PO2. We conclude that both the brain and kidney are susceptible to hypoxia during CPB, which can be alleviated by increasing pump flow, even without increasing arterial pressure. However, increasing blood haemoglobin concentration increases brain, but not kidney oxygenation, whereas vasopressor support with metaraminol increases kidney, but not brain oxygenation.

pulsatile flow.The vasopressor, metaraminol, dose-dependently decreased rSO 2 , but increased renal cortical and medullary PO 2 .Increasing blood haemoglobin concentration increased rSO 2 , but not renal PO 2 .We conclude that both the brain and kidney are susceptible to hypoxia during CPB, which can be alleviated by increasing pump flow, even without increasing arterial pressure.However, increasing blood haemoglobin concentration increases brain, but not kidney oxygenation, whereas vasopressor support with metaraminol increases kidney, but not brain oxygenation.
acute kidney injury, brain injury, cardiac surgery, cerebral hypoxia, renal hypoxia

| INTRODUCTION
Cardiac surgery requiring cardiopulmonary bypass (CPB) is associated with perioperative organ dysfunction. 1 Globally $22% of the 2 million cardiac surgical procedures performed each year result in acute kidney injury. 2 Brain injury (which encompasses a variety of diagnoses including postoperative cognitive dysfunction, delirium, seizure and symptomatic stroke) is experienced by 30% to 70% of patients after onpump cardiac surgery. 3The complex aetiologies of brain and kidney injury likely differ.Nevertheless, there is strong evidence implicating tissue hypoxia in both forms of cardiac surgery-associated organ dysfunction. 1,3,4sceptibility to hypoxia is organ-dependent.For example, in anaesthetised rats, the kidney is more susceptible to anaemia than the heart or intestine 5 or the brain. 6The brain also appears to be more tolerant of mild hypotension than the kidney. 7During CPB, oxygenation of vital organs could be influenced by perfusion conditions, including pump flow, arterial pressure, pulsatility of flow, blood haemoglobin concentration and body temperature. 8However, little is known regarding the relative impacts of perfusion conditions on oxygenation of the brain and kidney during CPB. 3,4 recently reported a series of studies, in an ovine model of CPB, of the effects on renal perfusion and oxygenation of: (i) altered pump flow at a set target arterial pressure, 9 (ii) the vasopressor metaraminol, 9 (iii) simultaneously increased pump flow and arterial pressure, 10 (iv) partially pulsatile flow, 10 and (v) variations in blood haemoglobin concentration. 11Regional cerebral oxygen saturation (rSO 2 ) was also measured in these studies using near-infrared spectroscopy (NIRS).However, these data were not included in those reports because of uncertainty regarding the validity of NIRS measurement of rSO 2 in sheep.
Here, we first report a comparison of rSO 2 with direct measurement of brain tissue PO 2 indicating relatively good qualitative agreement between these measures.Therefore, we are now able to present a direct comparison of the effects on cerebral rSO 2 and renal tissue PO 2 of transition to CPB and variations in perfusion conditions.We tested the hypothesis that cerebral and renal oxygenation respond differently to CPB and to changes in perfusion conditions during CPB.0][11][12][13] All studies fulfilled the Animal Research: Reporting of In Vivo Experiments (ARRIVE) 2.0 criteria. 14In brief, the 56 female Merino ewes (mean, 40.5 kg; range, 29.0-53.7 kg) included in the various protocols first underwent a preliminary operation under isoflurane anaesthesia (3.0%-4.0%,Isoflo, Zoetis) after induction with sodium thiopental (15 mg/kg i.v., Jurox) during which they were instrumented with a transit-time ultrasound flow probe around their renal artery for measurement of renal blood flow (RBF), a catheter in the carotid artery for measurement of mean arterial pressure (MAP), and fibre optic probes in the renal cortex and medulla for measurement of tissue oxygen tension (PO 2 ) and perfusion (laser Doppler flux).A total of 21 of these sheep were also instrumented with fibre optic probes in the frontoparietal lobe of the cerebral cortex, as previously described, 15 for measurement of brain tissue PO 2 and perfusion.
Five days later, sheep were again anaesthetised with either (a) isoflurane (3.0-4.0%v/v after induction with 15 mg/kg sodium thiopental intravenously (i.v.) (volatile anaesthesia, n = 37); [9][10][11]13 (b) a mixture of propofol (4 mg/kg h À1 , i.v.; AFT Pharmaceuticals) and fentanyl (5 μg/kg h À1 , i.v.; Hameln Pharmaceuticals) after induction with propofol (5 mg/kg, i.v.) and fentanyl (3 μg/kg, i.v.) (total intravenous anaesthesia, n = 5); 12 or (c) a combination of inhalation and intravenous anaesthesia (n = 14) comprising sevoflurane (4.0% v/v), propofol (4 mg/kg h À1 , i.v.) and fentanyl (3 μg/kg h À1 , i.v.) after induction with propofol (4 mg/kg, i.v.) and fentanyl (3 μg/kg, i.v.). Thse differences in the protocol for anaesthesia reflect the refinement of our methods over the course of multiple studies, to more closely reflect clinical protocols.After intubation, all sheep were equipped for cerebral oximetry by placing standard adult sensors, connected to an INVOS5100C Monitor (Medtronic), over each cerebral hemisphere, for measurement of rSO 2 (Figure 1).To avoid the sagittal sinus, the sensors were positioned lateral to the midline, and, so they were above the cerebral cortex, the plug-end of the sensor was in line with the most caudal part of the sheep's ear (Figure 1).All animals received 2 mL kg À1 h À1 compound sodium lactate (Baxter) as a maintenance fluid for the remainder of the experiment.During anaesthesia before CPB, the target arterial PCO 2 of $30 mm Hg was achieved by varying tidal volume and ventilation rate with a gas mixture of 60% O 2 and 40% N 2 .
0][11][12][13] In brief, we used a Stockert SIII heart-lung machine and a hollow fibre membrane oxygenator with integrated arterial filters (Sorin 6F, LivaNova, Dandenong), along with circuit tubing specifically designed for ovine CPB (cat no.06634900; LivaNova).The circuit was primed with 1 g cefazolin (AFT Pharmaceuticals), 50 mL mannitol (20% wt/vol Osmitrol, Baxter), 10 000 IU unfractionated heparin, and in some cases (n = 30) 300 to 500 mL of donor blood (stored for no more than 1 week), with the volume made up to 1.3 L with either compound sodium lactate (Baxter; n = 25) or PlasmaLyte (Baxter; n = 31).During the surgery for CPB, which was performed through a right thoracotomy, the fourth rib was removed and the sheep was heparinized (300 IU/kg, heparin injection; Pfizer) to permit insertion of an arterial cannula into the ascending aorta (18 Fr, EOPA77424; Medtronic) and a single stage venous cannula into the right atrium (32 Fr, DPL66132; Medtronic), along with an aortic root vent.Initial conditions for CPB varied slightly between the various protocols, with non-pulsatile pump flow varying from 60 to 80 mL min À1 kg À1 , within the range of expected cardiac output for an anaesthetised sheep of $65 mL min À1 kg À1 . 16As we have argued in detail previously, 9 this range of pump flow and other parameters we deploy for CPB closely reflects the clinical situation.
Target MAP of 55 to 70 mm Hg was achieved by administration of metaraminol (0.1 mg boluses).Target body temperature was set at 36 C using a water bath (Hemotherm, Cincinatti Subzero).Once target pump flow was achieved and the surgical procedures complete, the heart was either fibrillated with a 9-V DC pulse across the ventricle (n = 42) or asystole was achieved with anterograde (del Nido) cardioplegia (n = 14).An arterial blood sample was then taken so that gas flow through the oxygenator (50:50 air and oxygen) could be varied between 800 and 1200 mL min À1 to achieve a target arterial PCO 2 of $50 mm Hg.Once set at this level, total gas flow through the oxygenator was not further varied for the remainder of the experiment.This protocol was conducted in a series of nine sheep equipped with probes for measurement of cerebral perfusion and PO 2 , in which the effects of various vasopressors (metaraminol, noradrenaline, phenylephrine and vasopressin) had been tested, so the protocol described here commenced $3 h after commencement of CPB under propofol/ fentanyl/sevoflurane anaesthesia.The effects of the vasopressors will be described in a separate publication.
In the current protocol, the oxygen content of gas delivered to the membrane oxygenator was varied between 21%, 40%, 60% and 80%.Each gas mixture was maintained for 15 min and the order of presentation of the gases was randomized.Data were averaged over the final 5 min of presentation of the gas mixture, during which all variables were relatively stable and arterial and mixed venous blood oximetry had stabilized.

| Protocol 5: Effects of variations in pump flow and arterial pressure or of partially pulsatile flow
In 8 sheep anaesthetised with isoflurane, 10 target pump flow and arterial pressure were varied from 80 mL kg À1 min À1 /65 mm Hg to 104 mL kg À1 min À1 /80 mm Hg.In a separate experimental protocol in these sheep, flow was varied from continuous to partially pulsatile ($15 mm Hg pulse pressure). 10The order of these protocols was alternated between sheep.Observations were made over 30-min periods of stable MAP and RBF.

| Protocol 6: Effects of variations in blood haemoglobin concentration
In 10 sheep, 11 target blood haemoglobin concentration was varied from $70 to $90 g/L by addition of donor blood or crystalloid to the circuit.For this protocol, donor blood was used within 18 hours of collection.The order of the target blood haemoglobin concentrations was randomized.Observations were made over 60-min periods of stable MAP and RBF.

| Statistical analysis
Data are presented as between-sheep mean (95% confidence intervals) unless otherwise stated.Data were subjected to repeated measures analysis of variance, with the Greenhouse-Geisser correction applied to within subject factors. 17Specific comparisons between time-periods within each protocol were made using Student's paired t test (with Dunn-Sidak correction for multiple comparisons), Dunnett's test or Tukey's test, as identified in the footnotes of the tables and the figure legends. 18Associations between continuous variables were assessed using Pearson's product-moment correlation coefficient (r). 19| RESULTS

| Protocol 1: Comparison of cerebral rSO 2 with direct measurement of cerebral tissue PO 2
Progressively greater oxygen content of gas delivered to the membrane oxygenator during CPB was associated with progressively greater arterial blood PO 2 , SO 2 and oxygen content, but little change in other aspects of arterial blood chemistry (Table 1, Figure 2).Mixed venous PO 2 , SO 2 and oxygen content also progressively increased.
Systemic oxygen delivery (DO 2 ) increased, but systemic oxygen consumption (VO 2 ) did not change significantly, so fractional oxygen extraction progressively decreased.
Both brain tissue PO 2 and cerebral rSO 2 progressively increased with increased oxygen content of the gas delivered to the oxygenator (Figure 2).Both measures of cerebral oxygenation were significantly correlated with mixed venous SO 2 and with each other (Figure 3).Renal medullary perfusion was highly variable at the onset of CPB, with no significant change being detected (Figure 5).From the minute before commencing CPB to the 10 min after CPB commenced, renal medullary tissue PO 2 fell from 32.1 (20.4-43.7) to 22.3 (13.6-30.1)mm Hg (P = 0.007, paired t test).In contrast, total renal blood flow and renal cortical tissue perfusion and PO 2 were relatively stable, other than a transient reduction in cortical perfusion coinciding with the transient fall in MAP at commencement of CPB.

| Major findings and conclusions
Our current findings indicate that cerebral and renal oxygenation does not, predictably, behave similarly when perfusion conditions are changed during CPB (Figure 12).Both cerebral rSO 2 and renal (cortical and medullary) PO 2 were reduced during CPB compared with stable anaesthesia, regardless of whether volatile or total intravenous anaesthesia was used.However, the brain appears more resistant to hypoxia during CPB than the kidney, particularly the renal medulla.Recently, we observed little or no impact of moderate hypothermia on cerebral rSO 2 or PO 2 , or renal cortical or medullary tissue PO 2 . 20 the current study, we found that both cerebral rSO 2 and renal PO 2 were improved by increasing pump flow, either when target MAP was maintained at mm Hg or when it was increased from 65 to 80 mm Hg.Both were also relatively unaffected by the use of partially pulsatile flow.However, the effects of vasopressor support with metaraminol and of variations in blood haemoglobin concentration differed in the brain compared with the kidney.Vasopressor support with metaraminol dose-dependently reduced cerebral rSO 2 , although, at least at lower doses, it increased renal tissue PO 2 .In contrast, increasing blood haemoglobin concentration improved cerebral rSO 2 but not renal PO 2 .Indeed, renal cortical tissue PO 2 fell with increased blood haemoglobin concentration.Our findings indicate that these two critical organs, which are susceptible to injury during CPB, respond differently to interventions that might be applied by perfusionists to avoid organ hypoxia.They also indicate that cerebral rSO 2 monitored using NIRS should not be interpreted as an index of the oxygenation of other organs, especially the kidney.

| Validity of NIRS in sheep
NIRS is now routinely used for monitoring cerebral oxygenation and autoregulation of cerebral blood flow 21 in many clinical centres during cardiac surgery, although its efficacy in guiding the deployment of interventions in routine surgery is still a matter of debate. 22Regardless, the devices used in operating theatres around the world, including the device used in the current study (INVOS5100C), were designed and optimized for use in human patients, not sheep. 23To assess the validity of NIRS in sheep, we performed an experiment in which sheep undergoing CPB were subjected to variations in the oxygen content of air delivered to the membrane oxygenator.We found that the NIRS signal (cerebral rSO 2 ) correlated well with both directly measured brain tissue PO 2 and mixed venous SO 2 , providing a degree of confidence that NIRS is a valid approach for assessing cerebral oxygenation in sheep.Nevertheless, interpretation of our current finding must be tempered by recognition that the relationship between brain tissue PO 2 and NIRS-derived cerebral rSO 2 is not linear.Therefore, although it is possible to compare responses of renal tissue PO 2 and NIRS-derived cerebral rSO 2 in a qualitative way (i.e.whether they change and in what direction), comparison of these changes from a quantitative perspective would not be valid.

| Susceptibility of the brain and kidney to hypoxia during CPB
Recently, we demonstrated profound renal medullary hypoxia as an early event during cardiac surgery, with reduced medullary tissue PO 2 in sheep and reduced bladder urinary PO 2 (a surrogate measure of renal medullary oxygenation) in human patients, commencing even before initiation of CPB. 13 We were able to replicate our previous findings in sheep in the current study in an additional cohort of 12 sheep in which we were also able to measure brain tissue PO 2 and cerebral rSO 2 (protocol 2).In contrast to the profound hypoxia in the renal medulla during transition to CPB, we found little evidence of cerebral hypoxia, as assessed by measurement of brain PO 2 or cerebral rSO 2 .5][26] Our findings add to the growing evidence that, during CPB, the renal medulla is especially sensitive to hypoxia. 1,4It is not clinically feasible to directly measure renal tissue PO 2 during cardiac surgery in human patients.
Nevertheless, indirect evidence for development of renal medullary hypoxia in humans undergoing cardiac surgery comes from the observation of reduced urinary PO 2 during and after CPB.observations of increased renal fractional oxygen extraction during CPB 33,34 and reduced renal rSO 2 as assessed by NIRS before and during CPB. 24Considering the findings in protocols 2 and 3 collectively, it appears that the renal medulla is more sensitive to hypoxia during CPB than either the renal cortex or the cerebral cortex.This susceptibility could be because of the relatively poor autoregulatory capacity of the renal medullary circulation. 35e choice of perfusion conditions for CPB remains controversial. 8The introduction of NIRS for assessment of cerebral rSO 2 has provided a method to evaluate cerebral oxygenation and to intervene in a patient-specific manner when it is inadequate. 36However, recently completed clinical trials have not provided unequivocal support for the use of such NIRS-based algorithms for avoidance of poor neurological 37,38 or renal 37 outcomes.Furthermore, the brain is not the only organ vulnerable to injury during CPB.Our current findings show that cerebral and renal oxygenation do not always respond similarly to manipulation of perfusion conditions.Therefore, changes in cerebral rSO 2 may not reflect changes in kidney oxygenation.

| Anaesthesia
In our ovine model of CPB, both cerebral rSO 2 and renal tissue PO 2 were similar in sheep under total intravenous anaesthesia (propofol/ fentanyl) compared with sheep under volatile anaesthesia (isoflurane).
These observations are in accord with the apparent equivalence of these two anaesthetic regimens in human cardiac surgery in terms of mortality or acute kidney injury, 39,40 and conflicting evidence regarding, which is best for neuroprotection. 41

| Arterial pressure and vasopressor therapy
Increasing arterial pressure with the vasopressor metaraminol dosedependently reduced cerebral rSO 2 , albeit only marginally from 63.8% (56.8%-70.8%)under control conditions to 56.8% (50.0%-63.6%)at the highest dose administered (protocol 4).4][45] The mechanisms underlying the apparent cerebral constrictor response to systemically administrated vasopressor agents during CPB remain unknown.Vasopressor agents do not cross the blood-brain barrier under normal physiological conditions. 46Therefore, they are unable to induce direct cerebral vasoconstriction.One possibility is reflexively increased sympathetic vasoconstriction in response to increased MAP, which has been demonstrated in lambs. 47Moreover, contamination of the NIRS cerebral rSO 2 signal with extracranial tissue has been demonstrated. 48Therefore, it remains possible that our current finding of reduced cerebral rSO 2 T A B L E 3 Systemic oxygenation (protocols 3-6).Abbreviation: CI, confidence intervals; NA, data not available before cardiopulmonary bypass.
with higher doses of metaraminol reflect vasoconstriction in extracranial tissue.
In contrast to cerebral rSO 2 , renal tissue PO 2 , both in the cortex and medulla, increased when MAP was increased, regardless of whether it was induced by metaraminol (protocol 4) or by increased pump flow (protocol 5).Our observations likely reflect the poorer autoregulatory capacity of the kidney relative to the brain during normothermic CPB, because RBF was increased by both metaraminol and increased pump flow when MAP >70 mm Hg was targeted.In contrast, the lower limit of cerebral autoregulation in patients during normothermic CPB, although variable, was found to be $65 mm Hg. 49 Therefore, targeting a MAP greater than this should not be expected arm. 52Moreover, some benefit of higher perfusion pressure (presumably achieved by use of vasopressor agents) with regard to delirium was demonstrated in another (single-centre) trial. 53 appears to be the case with cerebral injury, available evidence from observational studies indicates that relative hypotension during CPB is associated with increased risk of acute kidney injury (reviewed in Jufar et al, Evans et al and Lankadeva et al). 1,4,54However, randomized trials have failed to detect a beneficial effect on postoperative renal function of increasing arterial pressure with phenylephrine (21 patients) 55 or noradrenaline (300 patients). 56

| Pump flow
Increased pump flow, whether accompanied by increased arterial pressure (protocol 5) or not (protocol 4), led to improved oxygenation of both the brain and kidney.The literature regarding the impact of pump flow on risk of cerebral and renal injury is contradictory. 8Particularly in the era of routine hypothermic CPB, some observational studies failed to detect associations between pump flow and postoperative brain or kidney dysfunction. 57Similarly, cerebral perfusion appears to vary little with pump flow during hypothermic CPB, at least in the absence of changes in MAP. 58However, the consensus from more recent studies in which CPB has more commonly been performed under normothermic or mildly hypothermic conditions, is that low pump flow during CPB, and in particular the consequent low systemic DO 2 , is associated with both cerebral and renal dysfunction postoperatively. 1,4For example, Magruder and colleagues 59   despite variable changes in MAP, in normothermic porcine CPB. 60In contrast, in a recent small (n = 25) clinical trial, a pump flow 20% greater than normal was not found to increase cerebral rSO 2 or improve cerebral autoregulation during rewarming on bypass. 61

| Pulsatility of flow
The potential for pulsatile flow to increase organ perfusion during CPB, and therefore, reduce postoperative organ dysfunction, remains controversial. 1,4Our current findings (protocol 5) do not support the hypothesis that pulsatile flow increases cerebral or renal oxygenation.
However, it must be acknowledged that the wave form generated by the heart-lung machine used in the current study differs from that induced by cardiac contraction.Therefore, it should be considered, at best, only 'partially pulsatile'.

| Blood transfusion
Cerebral rSO 2 , but not renal tissue PO 2 , was increased at greater blood haemoglobin concentration (protocol 6).3][64] Consistent with these observations, pre-operative anaemia is associated with increased risk of stroke in patients undergoing cardiac surgery. 65Yet, the recent Transfusion Requirements in Cardiac Surgery (TRICS) III trial failed to detect significant benefit, in terms of neurological or renal failure, of a liberal over a restrictive threshold for transfusion in patients undergoing on-pump cardiac surgery. 66As previously discussed in detail in the original report, 11 the observed reduction in renal cortical tissue PO 2 at higher blood haemoglobin concentration occurred despite reduced fractional renal oxygen extraction.Therefore, it may reflect reduced efficiency of oxygen extraction from blood to renal cortical tissue. 11However, this phenomenon is unlikely to be attributable to variation in the affinity of haemoglobin for oxygen, because the estimated P 50 of venous haemoglobin was similar in the low (33.8 ± 3.6 [mean ± 95% confidence interval] mm Hg) and high (35.1 ± 3.7 mm Hg) haemoglobin state. 11

| Limitations
Our study had a number of limitations.First, we used different methods for assessing renal and cerebral oxygenation.The phosphorescence-lifetime probes we used for assessing renal and (in some protocols) brain tissue oxygenation provides a direct measure of tissue PO 2 .However, they sample a relatively small ($1 mm 3 ) area of tissue 67 and their chronic placement is associated with a small amount of scarring around the tip of the probe in the kidney, 68 although not in the brain. 15In contrast, NIRS is a non-invasive method, but provides a measure of the saturation of haemoglobin with oxygen within a relatively large volume of tissue.However, the measurement it provides is heavily weighted to venous blood, 69 and varies both between the various commercially available devices 70 and according to sensor location. 25Moreover, these devices are not designed for use in sheep and their performance can be influenced by factors such as scalp-cortex distance and blood haemoglobin concentration. 70,71Nevertheless, we were able to show relatively strong correlations between rSO 2 and both brain tissue PO 2 and mixed venous blood SO 2 (protocol 1).2][73] Regardless, interpretation of our current findings should be restricted to a qualitative rather than quantitative basis.Furthermore, it would be informative to replicate these experiments in sheep equipped with both NIRS probes and phosphorescence-lifetime probes for monitoring cerebral and renal oxygen saturation and tissue oxygen tension, respectively.Interpretation of our findings is also complicated by the fact that we used a catecholamine vasopressor agent to maintain target MAP during CPB.
Choice of this agent was based on the fact that (a) it is commonly used during CPB in Australia; and (b) that it increases renal vascular resistance less than systemic vascular resistance in sheep during CPB. 9 Nevertheless, we cannot discount the possibility that our findings may have differed if we had used a different vasopressor agent.
We must also acknowledge the fact that multiple aspects of our methods varied across the various protocols, including mode of anaesthesia, the composition of the priming solution in the CPB circuit and the method used to achieve cardiac arrest (as detailed in the Methods section).These methodological differences, and in particular the use of different forms of anaesthesia, may have influenced our findings.
Nevertheless, our current findings provide evidence for contrasting responses of cerebral and renal oxygenation to two manoeuvres that might be applied by perfusionists to improve organ oxygen delivery, vasopressor support with metaraminol and blood transfusion.
Procedures These were approved by the Animal Ethics Committee of the Florey Institute of Neuroscience and Mental Health (16-051-FINMH, approved 16 September 2016; 18-119-FINMH, approved 21 January 2019) under guidelines of the National Health and Medical Research

2. 2 . 2 |
Protocol 2: Dynamic changes in cerebral and renal oxygenation during transition to CPB In 12 sheep equipped with probes for measurement of cerebral perfusion and PO 2 , anaesthetised with either isoflurane (n = 7) or propofol/fentanyl/sevoflurane (n = 5), we monitored MAP, cerebral tissue perfusion, PO 2 and cerebral rSO 2 , along with measures of renal perfusion and oxygenation, for the 16 min before and 30 min after initiation of CPB.2.2.3 | Protocol 3: Cerebral and renal oxygenation during stable CPB Once systemic hemodynamics had stabilized under CPB (usually 30 min after commencing CPB), cerebral SO 2 and renal oxygenation were recorded for 20-to 30-min periods.This protocol was F I G U R E 1 Placement of the INVOS sensors over the left and right hemispheres of an anaesthetised sheep.See text for details.L = left, R = right.The dash line represents the reference point for placement of the plug end of the sensor in line with the most caudal part of the sheep's ears.This figure was produced with Biorender.com.conducted in 30 sheep under isoflurane anaesthesia and five sheep under propofol/fentanyl anaesthesia.A series of experimental manipulations commenced in the 30 sheep anaesthetised with isoflurane (protocol 4, 5 or 6).These interventions have been described in detail previously in the relevant primary publications 9-11 and are described in brief below.2.2.4 | Protocol 4: Effects of variations in pump flow or arterial pressure In 12 sheep, 9 pump flow was randomly varied between 60, 80 and 100 mL kg À1 min À1 with target MAP maintained at 70 mm Hg by use of metaraminol.At a pump flow of 80 mL kg À1 min À1 , metaraminol was infused at progressively greater doses of 0, 0.2, 0.4 and 0.6 mg/ min.Observations were made over 20-to 30-min periods of stable MAP and RBF.

3. 2 |
Protocol 2: Dynamic changes in cerebral oxygenation during transition to CPB MAP progressively fell during the 16 min of observation before commencing CPB, from 92.2 (77.4-107.0)mm Hg to 82.5 (75.8-89.1)mm Hg (Figure 4).Commencing CPB was accompanied by a very transient further fall in MAP to 65.8 (56.0-75.7)mm Hg, but MAP recovered within $1 min (to 90.6 [76.7-104.5])mm Hg in response to administration of metaraminol by the attending perfusionist, before slowly falling over the next 10 min to its target level of 70 mm Hg.There was a transient fall in brain perfusion at the onset of CPB, but both cerebral PO 2 and rSO 2 were relatively well maintained.

3. 4 |
Protocol 4: Effects of variations in pump flow or arterial pressure 3.4.1 | Altered pump flow By design, MAP varied little with altered pump flow, but RBF progressively increased as pump flow was increased (Table

F
I G U R E 2 Arterial and mixed venous oximetry, mean arterial pressure and cerebral oxygenation during changes in the oxygen content of air supplied to the membrane oxygenator during cardiopulmonary bypass (CPB).Columns and error bars represent means and 95% confidence intervals.Red lines represent data for the 9 individual sheep (except for PvO 2 , SvO 2 , venous O 2 content, VO 2 , O 2 extraction and brain PO 2 , for which n = 7 because of equipment malfunction).
Specific post-hoc comparisons between the various gas mixtures are shown as letters if p ≤ 0.05 by Student's unpaired t test after application of the Dunn-Sidak correction 18: Different from 21% = a, 40% = b, 60% = c, 80% = d.3.6 | Protocol 6: Effects of variations in blood haemoglobin concentration Blood haemoglobin concentration was 68.3 (64.7-71.9)g L À1 during the 'low haemoglobin state' and 89.5 (80.0-98.8)g L À1 during the 'high haemoglobin state'.Increased blood haemoglobin concentration was associated with increased systemic DO 2 (by 29.2% [14.8%-43.6%]),but no significant changes in systemic VO 2 or fractional systemic oxygen extraction were observed (Table

F I G U R E 3 4
Relationships between mixed venous oxygen saturation, brain PO 2 , and cerebral rSO 2 .Black circles and error bars represent means and 95% confidence limits for data derived from each of the four gas mixtures presented to the membrane oxygenator (21%, 40%, 60% and 80%).Red lines represent data for individual sheep.P values were derived from Pearson's product-moment correlation coefficient (r).Cerebral rSO 2 , cerebral oxygen saturation determined by near-infrared spectroscopy; SvO 2 , oxygen saturation of mixed venous haemoglobin.Cerebral perfusion and oxygenation during transition to cardiopulmonary bypass.Thick lines show the mean and thin lines show 95% confidence limits of the mean (SEM) for 1 min averages (n = 12).MAP, mean arterial pressure; rSO 2 , regional oxygen saturation.

6 7 8 9
Cerebral and renal oxygenation during stable cardiopulmonary bypass.Symbols and errors bars show mean ± 95% confidence interval.For isoflurane anaesthesia n = 24 for cerebral oxygen saturation (Cerebral rSO 2 ) and n = 29 for both renal cortical PO 2 (renal cortex) and renal medullary PO 2 (renal medulla).For total intravenous anaesthesia (propofol/fentanyl) n = 5 for all variables.*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 (Student's paired t test).Responses of cerebral and renal oxygenation to changes in pump flow at a target arterial pressure of 70 mm Hg.Symbols and errors bars show mean ± 95% confidence interval for p = 6 for cerebral oxygen saturation (cerebral rSO 2 ), p = 11 for renal cortical PO 2 (renal cortex) and P = 10 for renal medullary PO 2 (renal medulla).*p ≤ 0.05, **p ≤ 0.01 (Tukey's test) for comparison with 60 mL kg À1 min À1 .†p ≤ 0.05 (Tukey's test) for comparison with 80 mL kg À1 min À1 .Cerebral and renal oxygenation during infusion of metaraminol.Symbols and errors bars show mean ± 95% confidence interval for n = 6 for cerebral oxygen saturation (cerebral rSO 2 ), n = 11 for renal cortical PO 2 (renal cortex) and n = 9 for renal medullary PO 2 (renal medulla).*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 (Dunnett's test) for comparison with the control condition (0 mg/min).Responses of cerebral and renal oxygenation to simultaneous increases in pump flow and target arterial pressure.Symbols and errors bars show mean ± 95% confidence interval for n = 8 for all variables.rSO 2 , regional oxygen saturation.*p ≤ 0.05, ***p ≤ 0.001 (Student's paired t test).to greatly improve cerebral perfusion and oxygenation.Taken collectively, these observations indicate that greater MAP during normothermic CPB may be more beneficial for kidney oxygenation than brain oxygenation.Furthermore, they indicate that improved cerebral oxygenation during CPB is best achieved by increasing pump flow rather than administration of a vasopressor agent.These insights may at least partially explain the apparent contradiction between the strong evidence that hypotension (presumably because of multiple causes) during CPB is associated with stroke 50 and the lack of benefit with regard to cerebral injury in a single-centre trial of high target MAP (70-80 mm Hg vs 40-50 mm Hg in the usual care arm).Critically, in that trial higher target MAP was achieved by intermittent boluses of phenylephrine and/or noradrenaline 51 and was associated with lower cerebral rSO 2 than in patients in the usual care (lower MAP)

1 0 1
found that low systemic DO 2 during mildly hypothermic CPB was associated with increased levels of the biomarker of brain injury ubiquitin C-terminal hydrolase L1.Furthermore, in an experimental study, decreased pump flow was found to decrease cerebral perfusion, Responses of cerebral and renal oxygenation to partially pulsatile flow.Symbols and errors bars show mean ± 95% confidence interval for n = 8 for all variables.rSO 2 , regional oxygen saturation.Responses of cerebral and renal oxygenation to changes in blood haemoglobin concentration.Symbols and errors bars show mean ± 95% confidence interval for n = 10 for cerebral oxygen saturation (rSO 2 ), n = 9 for renal cortical PO 2 and n = 10 for renal medullary PO 2 .*p ≤ 0.05, **p ≤ 0.01 for comparison with low haemoglobin (Student's paired t test).

4. 10 |
Summary and conclusionsOur findings indicate differential changes in cerebral and renal oxygenation in response to changes in perfusion conditions during CPB in sheep.Both cerebral and renal oxygenation were improved by increasing pump flow, with or without associated increases in arterial pressure.However, renal but not cerebral oxygenation was improved by vasopressor support with metaraminol to increase MAP $25 mm Hg above its target level of 70 mm Hg.In contrast, cerebral, but not renal oxygenation, was improved by blood transfusion to increase blood haemoglobin concentration from $70 to $90 g/L.AUTHOR CONTRIBUTIONS Clive N. May, Yugeesh R. Lankadeva, Roger G. Evans, Andrew D. Cochrane and Rinaldo Bellomo conceived and designed the research.Yugeesh R. Lankadeva, Clive N. May, Roger G. Evans,