Hemodilution Elevates Cerebral Blood Flow and Oxygen Metabolism During Cardiopulmonary Bypass in Piglets

Transcription

1 CARDIOVASCULAR Hemodilution Elevates Cerebral Blood Flow and Oxygen Metabolism During Cardiopulmonary Bypass in Piglets Takahiko Sakamoto, MD, Georg D. A. Nollert, MD, David Zurakowski, PhD, Janet Soul, MD, Lennart F. Duebener, MD, Jason Sperling, MD, Mitsugi Nagashima, MD, George Taylor, MD, Adre J. duplessis, MD, and Richard A. Jonas, MD Departments of Cardiovascular Surgery, Biostatistics, Neurology, and Radiology, Children s Hospital and Harvard Medical School, Boston, Massachusetts Background. Hemodilution continues to be widely used during cardiopulmonary bypass (CPB) for both adults and children. Previous studies with nonbypass models have suggested that an increase in cerebral blood flow (CBF) compensates for the reduced oxygen-carrying capacity; however, this increased CBF is achieved by an increase in cardiac output. We hypothesized that even with the fixed-flow perfusion of CPB, CBF would be increased during hemodilution. Methods. Two experiments were conducted and analyzed separately. In each experiment, 10 piglets were randomized to two different groups, one with a total blood prime yielding a high hematocrit (25% or 30%), and the other with a crystalloid prime resulting in a low hematocrit (10% or 15%). Animals were cooled with ph-stat strategy at full flow (100 or 150 ml kg 1 min 1 ) to a nasopharyngeal temperature of 15 C, a period of low flow (50 ml kg 1 min 1 ) preceding deep hypothermic circulatory arrest (45 or 60 minutes), and a period of Hemodilution has been used during cardiopulmonary bypass (CPB) for decades. The practice was introduced in the 1960s to decrease exposure to homologous blood [1]. Another rationale for hemodilution was the suggestion that brain damage could occur secondary to the microcirculatory disturbances resulting from the increase in viscosity and deleterious rheologic effects of deep hypothermic bypass [2]. It was speculated, although not demonstrated, that these microcirculatory disturbances would be ameliorated by use of a lower hematocrit. However, recent studies in our laboratory using a piglet model demonstrated that use of a low hematocrit is associated with an important decline in cerebral highenergy phosphates and a decrease in intracellular cerebral ph as determined by magnetic resonance spectroscopy [3]. Near-infrared spectroscopy in our studies also suggested that cerebral oxygenation is compromised Accepted for publication Oct 10, Address reprint requests to Dr Jonas, Department of Cardiovascular Surgery, Children s Hospital, 300 Longwood Ave, Boston, MA 02115; richard.jonas@tch.harvard.edu. rewarming at full flow. Cerebral blood flow was measured at the beginning of CPB, at the end of cooling, at the end of low flow, 5 minutes after the start of rewarming, and at the end of rewarming by injection of radioactive microspheres. Results. Mean arterial pressure was significantly greater with higher hematocrit at each time point (p< 0.05). Cerebral blood flow and the cerebral metabolic rate of oxygen decreased during cooling and further during low flow bypass but were significantly greater with lower hematocrit during mild hypothermia and at the end of rewarming (p< 0.05). Conclusions. Hemodilution is associated with decreased perfusion pressure, increased CBF and increased the cerebral metabolic rate of oxygen during hypothermic CPB. (Ann Thorac Surg 2004;77: ) 2004 by The Society of Thoracic Surgeons with a low hematocrit. In addition, functional and histologic endpoints also were consistent with neurologic injury occurring secondary to low hematocrit [3]. However, our previous studies did not include measurement of cerebral blood flow (CBF) and oxygen metabolism. Furthermore, a potential explanation for the findings of cerebral hypoxia in our previous studies was that CBF was decreased with hemodilution because of the decrease in perfusion pressure that occurred secondary to the decreased viscosity. An alternative explanation, however, which forms the hypothesis for the current study, is that even in settings of a fixed cardiac output provided by a CPB pump, CBF is increased under hemodilution conditions. However, at severe degrees of hemodilution the reduced oxygencarrying capacity of blood may result in inadequate cerebral oxygen delivery. Two sets of experiments were undertaken. On analysis of the data from the first set of animals, we were concerned that an artifact in measurement of CBF or the cerebral metabolic rate of oxygen (CMRo 2 ) had resulted in the unexpected findings. The second set of experi by The Society of Thoracic Surgeons /04/$30.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg SAKAMOTO ET AL 2004;77: HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM Abbreviations and Acronyms ANOVA analysis of variance AVDo 2 arteriovenous oxygen difference Cao 2 arterial oxygen content CBF cerebral blood flow CDo 2 cerebral oxygen delivery CMRo 2 cerebral metabolic rate of oxygen CPB cardiopulmonary bypass Cvo 2 venous oxygen content DHCA deep hypothermic circulatory arrest Hct hematocrit MAP mean arterial pressure OER oxygen extraction ratio Po 2 arterial oxygen pressure Pco 2 arterial carbon dioxide pressure ments used a modified method for great vessel cannulation and microsphere measurement of CBF as well as slightly different flows, hematocrit levels, and temperatures. Material and Methods Experimental Preparation Twenty Yorkshire piglets weighing 6.4 to 11.0 kg (n 10, kg and n 10, kg for experiments 1 and 2, respectively) were anesthetized with intramuscular ketamine (20 mg/kg) and xylazine (4 mg/kg) and were intubated with a 5-mm cuffed endotracheal tube. The animals were ventilated with a pressure-controlled ventilator (Healthdyne model 105, Healthdyne Technologies, Marietta, GA) at a peak inspiratory pressure of 20 cmh 2 O, and fraction of inspired oxygen of 0.21 at a rate of 12 to 20 minutes 1. After an intravenous bolus of fentanyl (50 g/kg) and pancuronium (0.5 mg/kg), anesthesia was maintained by continuous infusion of fentanyl (25 g kg 1 h 1 ), midazolam (0.2 mg kg 1 h 1 ), and pancuronium (0.2 mg kg 1 h 1 ) throughout the entire experiment except during circulatory arrest. Esophageal and rectal temperatures were recorded continuously. For intraoperative monitoring and blood sampling, arterial and venous lines were placed in the right femoral artery and vein, respectively. Another arterial line was placed in left femoral artery to draw the reference blood during radioactive microsphere injection. The chest was opened by a median sternotomy. After systemic heparinization (300 IU/kg), an 8F arterial cannula (DLP, Medtronic Inc, Grand Rapids, MI) and 28F venous cannula (Bard, Inc, USCI Division, Billerica, MA) were inserted into the right femoral artery (experiment 1), ascending aorta (experiment 2) and the right atrium, respectively. A venous line was placed as high as possible in the jugular vein through the superior vena cava to measure the jugular venous saturation during the study. All animals received humane care in compliance with the Principles of Laboratory Animal Care formulated by the National Society for Medical Research and the Guide for the Care and the Use of Laboratory Animals 1657 prepared by the Institute of Laboratory Animal Resources and published by the National Institute of Health (National Institutes of Health Publication No , revised 1985). Cardiopulmonary Bypass Technique The CPB circuit consisted of a roller pump, membrane oxygenator (Minimax, Medtronic Inc, Anaheim, CA), and sterile tubing with a 40- m arterial filter (Olson Medical Sales Inc, Ashland, MA). The pump prime consisted of a balanced electrolyte solution Plasmalyte A (Baxter, Deerfield, IL) and freshly drawn heparinized blood to achieve the desired hematocrit (see study groups below). Fresh whole blood or crystalloid solution was transfused into the pump as required to keep the hematocrit at the target level. Methylprednisolone (30 mg/kg), furosemide (0.25 mg/kg), sodium bicarbonate (10 ml), fentanyl (50 g/kg), and pancuronium (0.5 mg/kg) were added to the prime. Full bypass flow was set at 150 ml kg 1 min 1 in experiment 1 and 100 ml kg 1 min 1 in experiment 2. Ventilation was stopped after the onset of CPB. A venting cannula was inserted into the left ventricle and the ascending aorta was clamped to prevent residual forward blood flow from the heart. In experiment 1, after reaching mild hypothermia and maintaining esophageal temperature at 34 C for 10 minutes, experimental conditions were kept constant to allow measurement of CBF. These and the following measurements took 15 minutes each. The piglet was then cooled to an esophageal temperature of 15 C for more than 30 minutes applying the ph-stat strategy, ie, addition of CO 2 to maintain the temperature corrected ph at 7.40 [4, 5] followed by a second measurement period. Low flow bypass was then initiated at a flow rate of 50 ml kg 1 min 1. After 15 minutes of low flow, the third measurement period was started. Following 45 minutes of deep hypothermic circulatory arrest (DHCA), furosemide (0.25 mg/kg), sodium bicarbonate (10 ml), and mannitol (0.5 g/kg) were administered into the pump and reperfusion was initiated. Within 5 minutes, the temperature of the piglet had increased to 20 C and was controlled at this level for the fourth measurement period. Subsequently, the piglets were fully rewarmed more than 40 minutes and then the last measurements were taken. In experiment 2, animals were perfused for 10 minutes to achieve steady state at normothermia (37 C), and the experimental conditions were kept constant to allow for measurement of CBF. These and the following measurements took 10 minutes each. Animals were then cooled to an esophageal temperature of 15 C for more than 30 minutes applying the ph-stat strategy followed by a second measurement period. Low flow bypass was then initiated at a flow rate of 30 ml kg 1 min 1. After 10 minutes of low flow, the third measurement period was started. Then animals underwent DHCA for 60 minutes. Before reperfusion, furosemide (0.25 mg/kg), sodium bicarbonate (10 ml), and mannitol (0.5 g/kg) were administered into the pump. Reperfusion was begun at 100 ml kg 1 min 1. Within 5 minutes, the temperature of CARDIOVASCULAR

3 CARDIOVASCULAR 1658 SAKAMOTO ET AL Ann Thorac Surg HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM 2004;77: the piglet had increased to 20 C and was controlled at this level for the fourth measurement period. Subsequently, the piglets were fully rewarmed for more than 30 minutes and then the last measurement was performed at 37 C. Data Collection BLOOD FLOW MEASUREMENTS. Regional blood flow to the brain was measured by microspheres labeled with one of the following radioactive nuclides: 141 Ce, 51 Cr, 103 Ru, 95 Nb, or 46 Sc suspended in 2 ml of 0.9% saline followed by 2 ml of 0.9% saline injected into a side port on the arterial tubing. In experiment 1, the side port was upstream to the arterial filter (40- m mesh) and 80 cm from the tip of the arterial cannula to ensure complete mixing. In experiment 2, the side port was downstream to the filter and 1 m from the tip of the arterial cannula. Each microsphere injection consisted of approximately microspheres. A measured quantity of blood (approximately 5 ml) as reference was withdrawn at a constant rate by a syringe pump (model 55:1143, Harvard Apparatus) from the arterial line placed in left femoral artery. Arterial blood was withdrawn before, during, and after the microsphere injection for a total of 90 seconds. At the end of each experiment, the brain was removed and weighed. Three samples were taken from the left and right frontal cortex. Additional samples were taken from the left and right parietal and occipital cortex, the left and right cerebellum, pons, and medulla oblongata. Radioactivity was counted with a gamma counter (Cobra II Autogamma, Packard Instrument, IL) with a spillover correction between the nuclides. The reference blood was also analyzed for radioactivity. Regional blood flow calculated from the rate of withdrawal of the reference blood and the ratio of the radioactivity of the brain to the reference blood. Average CBF was derived by adding the average regional blood flows and dividing by number of regions. METABOLIC MEASUREMENTS. Blood gas tensions and ph, hematocrit, plasma glucose, and lactate levels were measured in arterial blood and blood withdrawn from the jugular bulb before CPB and before each microsphere injection. Measurements including hemoglobin, saturation, and oxygen content were performed by a blood gas analyzer (Stat Profile 9, Nova Biomedical, Waltham, MA). Cerebral oxygen delivery (CDo 2 ), cerebral metabolic rate for oxygen (CMRo 2 ), and oxygen extraction ratio (OER) were calculated by standard formulas [6] and listed in the Appendix. HEMATOCRIT STUDY GROUPS AND POWER ANALYSIS. In experiment 1, a hematocrit of 15% was specified for low hematocrit and 25% for high hematocrit during CPB. In experiment 2, a low hematocrit was specified as 10% and a high hematocrit was 30% or freshly drawn heparinized blood to achieve a high hematocrit of 30%. Because the protocols and hematocrit study groups were not identical between the two sets of experiments, we chose to analyze the experiments separately. The sample size of 10 piglets equally randomized to the two hematocrit study groups provided 80% statistical power (beta error 0.20) to detect differences based on a desired effect size of 1.75 in each of the CBF and metabolic variables in each experiment using analysis of variance (ANOVA) with repeated measures with a Bonferroni corrected significance level of 0.01 (0.05 divided by 5) to account for the multiple time points of the study (version 4.0, nquery Advisor, Statistical Solutions, Boston, MA) [7]. Statistical Analysis Data are represented as means SD. After verifying the assumption of normality by the Kolmogorov-Smirnov test [8], continuous variables including hematocrit level, arterial oxygen pressure (Po 2 ), arterial carbon dioxide pressure (Pco 2 ), jugular venous ph, esophageal and rectal temperature, and mean arterial pressure (MAP) were compared at each of five time points (mild hypothermia, end of cooling, after 15 minutes of low flow, 5 minutes after rewarming, end of rewarming) between the two hematocrit groups in each experiment using unpaired Student t tests. Two-way repeated-measures ANOVA was used to test the interaction between experiment and hematocrit level for the four main outcome variables (CBF, CDo 2, CMRo 2, and OER) and to assess changes across the time points and between hematocrit study groups [9]. Statistical significance was established using a two-tailed value of p 0.05 with post hoc Bonferroni adjustment. Multivariate linear regression analysis was applied to ascertain whether the differences between low and high hematocrit level was independent of differences in MAP. Analysis of the data were performed with the SPSS software package (version 11.0, SPSS Inc, Chicago, IL). Results No significant differences were observed for several of the variables measured including Po 2,Pco 2, arterial ph, jugular venous ph, and esophageal and rectal temperature, except for a higher Po 2 at the beginning of cooling (p 0.01) and a lower jugular venous ph at the end of rewarming (p 0.01) in the low hematocrit group of each experiment. Mean arterial pressure was significantly higher in the high hematocrit group of each experiment for all five time points studied (Table 1), reaching nadir levels during low flow for all four hematocrit subgroups (Fig 1). Results from ANOVA with correction for repeated measures on the four main outcome measures (CBF, CDo 2, CMRo 2, and OER) indicated significant F test interactions between experiment and hematocrit level, supporting our decision to analyze the two experiments separately: CBF, F 29.87, p 0.001; CDo 2, F 3.98, p 0.05 (marginally significant); CMRo 2, F 13.09, p 0.001; OER, F 5.03, p Hence the results in comparing low versus high hematocrit were conditionally dependent on the specific experiment, although the actual results do not vary much, with some exceptions. The comparisons between hematocrit groups within each experiment for CBF, CDo 2, CMRo 2, and OER are presented in Table 2. Measures of cerebral metabolism were significantly different between low and high hematocrit study groups within each experiment. Higher CBF

4 Ann Thorac Surg SAKAMOTO ET AL 2004;77: HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM Table 1. Effects of Hemodilution on Mean Arterial Pressure Experiment 1 Experiment 2 Hct 15% Hct 25% Hct 10% Hct 30% Variable (n 5) (n 5) p a (n 5) (n 5) 1659 p a CARDIOVASCULAR MAP (mm Hg) Mild hypothermia End of cooling Low flow Rewarming 5 min End of rewarming a All p values were statistically significant. Data are mean SD. Hct hematocrit; MAP mean arterial pressure. was observed in the low hematocrit group at the beginning of cooling (ie, mild hypothermia) and at the end of rewarming for experiments 1 and 2 (Fig 2). In addition, a higher CBF was demonstrated among the animals in the low hematocrit group at the end of cooling and during low flow bypass in experiment 2. In both sets of experiments, CBF decreased sharply between mild hypothermia and the end of cooling and continued to decline throughout low flow and increase during the rewarming phase. This time-related pattern for the change in CBF was significant for both experiments (p 0.01, F test for time effect). Cerebral oxygen delivery was greater in the low hematocrit group for each experiment at each time point (except 5 minutes after rewarming in experiment 1), although the mean differences between the low and high hematocrit groups did not reach statistical significance except at the end of rewarming in experiment 2 (p 0.01). The profiles for CDo 2 in both experiments showed an initial decline at the end of cooling with lowest levels observed during low flow, and then an increase during rewarming (Fig 3). Cerebral metabolic rate of oxygen was significantly higher for animals in the low hematocrit group during mild hypothermia and at the end of rewarming in both experiments (Fig 4). Results were more pronounced in experiment 2 and this finding could be due to the larger difference in hematocrit levels in that experiment. For example, at the end of rewarming, the low and high hematocrit groups had CMRo 2 mean SD levels of 1.55 and 0.78 ml/100 g/min in experiment 1 (p 0.05) and 6.68 and 0.98 ml/100 g/min in experiment 2 (p 0.01) (Table 2). In both sets of experiments, levels of CMRo 2 showed a sharp decline between mild hypothermia and the end of cooling, attained nadir levels during low flow, and increased steadily during rewarming, although they did not return to their baseline values. Oxygen extraction ratios were not significantly different between the hematocrit groups in experiment 1, although the low hematocrit group had greater OER mean values throughout the study (Table 2). This was also the case in experiment 2, although the differences in OER between the two hematocrit groups reached statistical significance based on ANOVA at each of the five time points (all p 0.05). Lack of a statistically significant difference in experiment 1 could be due potentially to the smaller difference in hematocrit levels compared with experiment 2 as well as the possibility of a type II beta error. The effect of hematocrit on OER based on the results of experiment 2 showed an approximate twofold greater value in the low hematocrit group at each time point (Fig 5). Multiple linear regression analysis was performed to evaluate whether the effect of hematocrit on CBF, CDo 2, CMRo 2, and OER was modified by MAP. The rationale for this being that significant differences in MAP were found between low and high hematocrit groups at each time point (Table 1). However, results confirmed that hematocrit was a significant independent predictor of each CBF and metabolism variable. Furthermore, the differences in CBF, CDo 2, CMRo 2, and OER between hematocrit groups were clearly not explained by MAP. Fig 1. Mean arterial pressure (MAP) throughout the time course according to hematocrit (Hct) group for each experiment. Significant differences were observed between low and high Hct groups in each experiment at all five time points (all p 0.05). Error bars denote SD. Comment The experimental data have confirmed our hypothesis that CBF is increased with a low hematocrit even in the setting of CPB with fixed-pump output. The increase in

5 CARDIOVASCULAR 1660 SAKAMOTO ET AL Ann Thorac Surg HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM 2004;77: Table 2. Effects of Hemodilution on Measures of Cerebral Blood Flow and Metabolism Variable Experiment 1 Experiment 2 Hct 15% Hct 25% Hct 10% Hct 30% (n 5) (n 5) p (n 5) (n 5) p CBF (ml/100 g/min) Mild hypothermia * a End of cooling a Low flow a Rewarming 5 min End of rewarming * a CDo 2 (ml/100 g/min) Mild hypothermia End of cooling Low flow Rewarming 5 min End of rewarming a CMRo 2 (ml/100 g/min) Mild hypothermia * a End of cooling a Low flow a Rewarming 5 min End of rewarming * a OER (%) Mild hypothermia a End of cooling a Low flow a Rewarming 5 min a End of rewarming a a Statistically significant. Data are mean SD. Hct hematocrit; CBF cerebral blood flow; CDo 2 cerebral oxygen delivery; CMRo 2 cerebral metabolic rate for oxygen; OER oxygen extraction ratio. CBF, which serves to compensate for the reduced oxygen-carrying capacity of dilute blood, occurs despite reduced perfusion pressure. However, the unexpected finding of this study was that CMRo 2 is increased in the setting of hemodilution. Consistent with the results of this study, several previous studies from our laboratory [3] and preliminary results of a prospective, randomized clinical trial in infants [10] have demonstrated beneficial effects of a high hematocrit (30%) during CPB with and without DHCA. In a piglet survival model described by Shin oka and asso- Fig 2. Cerebral blood flow (CBF) throughout the time course according to hematocrit (Hct) group for each experiment. Cerebral blood flow was elevated in the low Hct group during mild hypothermia and at the end of rewarming for both experiments, as well as at the end of cooling and during low flow for experiment 2 (all p 0.05). Error bars represent SD. Fig 3. Cerebral oxygen delivery (CDo 2 ) throughout the time course according to hematocrit (Hct) group for each experiment. Cerebral oxygen delivery was significantly elevated in the low Hct group at the end of rewarming in experiment 2 (p 0.01). Error bars represent SD.

6 Ann Thorac Surg SAKAMOTO ET AL 2004;77: HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM Fig 4. Cerebral metabolic rate for oxygen (CMRo 2 ) throughout the time course according to hematocrit (Hct) group for each experiment. Cerebral metabolic rate for oxygen was elevated in the low Hct group during mild hypothermia and at the end of rewarming for both experiments, as well as at the end of cooling and during low flow for experiment 2 (all p 0.05). Error bars represent SD. ciates [11], both functional and structural outcome measures were improved with a higher hematocrit. The mechanism appeared to be related, in part, to inadequate reduced oxygen availability with a low hematocrit as determined by near-infrared spectroscopy [11]. Magnetic resonance spectroscopy demonstrated a decrease in high-energy phosphate levels and a drop in intracellular ph, particularly in the early phase of cooling with a low hematocrit (10%). Fig 5. Oxygen extraction ratio (OER) throughout the time course according to hematocrit (Hct) group for each experiment. Oxygen extraction ratio was elevated by approximately twofold at each time point in experiment 2 (all p 0.05). No significant differences were noted in experiment 1. Error bars represent SD. Blood Pressure Versus Pump Flow as Determinants of Cerebral Blood Flow Other groups have demonstrated previously that with ph-stat management, as used in the current study (ie, addition of CO 2 to correct for the alkaline shift associated with hypothermia), CBF autoregulation is impaired leading to pressure and flow-dependent perfusion of the brain [12]. Therefore, the higher arterial pressure associated with higher hematocrit could per se improve neurologic outcome. Indeed clinical reports on adult cardiac surgery have indicated that higher arterial pressures (MAP 70 mm Hg) are beneficial for neurologic outcome [13], although this population likely had both carotid and cerebrovascular disease. However, in both of our experiments, despite the use of ph-stat, CBF was independent of MAP over a wide range of perfusion pressures, ie, autoregulation was intact, which has been well described for this range with -stat management [12, 14]. In contrast to blood pressure, pump flow in our study had a significant influence on CBF. Few previous studies have attempted to differentiate between the effects on CBF of CPB perfusion flow rate and blood pressure in the setting of ph-stat management. Previous researchers have used phenylephrine and sodium nitroprusside to vary MAP while maintaining a constant pump flow [15, 16]. An increase of CBF with greater perfusion pressure was demonstrated, although the specific pharmacologic effects of the vasoactive drugs used in that study on cerebral resistance vessels are unknown. In another study, investigators manipulated perfusion pressure in the carotid artery independent of the perfusion flow rate by obstructing the descending aorta [17]. They found a highly significant correlation between CBF and blood pressure but not flow, but conceded that constriction of the aorta during low flow in order to increase blood pressure in the carotid artery may redirect blood flow to the brain. The relevance to clinical practice remains unclear. In contrast, others have demonstrated in clinical studies a dependency of CBF on perfusion flow but not pressure [18, 19]. Hematocrit and Cerebral Blood Flow 1661 In the current study, hemodilution generally increased CBF though not significantly early rewarming after DHCA. Microcirculatory dysfunction of cerebral vessels after prolonged DHCA has been described previously by others and may account for the fact that CBF measurements after DHCA were lower than at baseline, although a hyperemic response after ischemia might be expected [20]. In our own studies using intravital microscopy, we have observed reduced flow in the early reperfusion phase with a low relative to high hematocrit, which we have attributed to hypoxic endothelial injury resulting in impaired NO synthesis [21]. The influence of hematocrit on CBF has been studied most commonly in the past in a non-cpb setting [22]. An increase in CBF with decreasing hematocrit has been found in all studies but cardiac output was not controlled and therefore almost certainly increased significantly in all such studies; however, in some studies, oxygen delivery declined because of the decreased oxygen-carrying capacity of dilute blood [22]. In an important study by Cook and colleagues [23] using a canine model of CPB it was found that cerebral oxygen delivery was maintained with hemodilution. However, a significant difference in Cook s study from our own was that pump flow rate was increased in their study to compensate for the decrease in perfusion pressure resulting from hemodilution. CARDIOVASCULAR

7 CARDIOVASCULAR 1662 SAKAMOTO ET AL Ann Thorac Surg HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM 2004;77: Hematocrit and Cerebral Metabolism Probably the most surprising finding of our study was that CMRo 2 was higher in the low hematocrit group at all temperatures. It was this finding that led us to completely repeat the study using aortic rather than femoral arterial cannulation. We were concerned that the microsphere assessment of CBF may have been unreliable with femoral arterial cannulation because the reference blood was drawn from the other femoral artery. Also in experiment 1, the microsphere injections were made upstream to an arterial filter. Although the mesh size of the filter should not have trapped microspheres, we speculated that this possibility could have artificially elevated our assessment of CBF and therefore CMRo 2. However, the results obtained in experiment 2 by a completely new laboratory team and with aortic rather than femoral cannulation were similar to those obtained in experiment 1. Previous piglet studies from our laboratory have demonstrated that during cooling lower hematocrit is associated with lower levels of cerebral high-energy phosphates (ATP and phosphocreatine) and intracellular acidosis [3]. In the current study, lower Sjo 2 and arterial ph were also noted. These data also suggested to us that cerebral oxygen demands were not being met during CPB. However, the present study indicated that part of the problem may be an increasing demand for oxygen with low arterial oxygen content despite maintenance of oxygen delivery by markedly increased CBF. Perhaps more oxygen is needed to counteract acidosis due to the lower ph-buffering capacity of the dilute blood. Other possible mechanisms could be an increase in substrate transport energy requirements with more dilute blood or perhaps some energy-requiring aspect of the markedly increased CBF itself. In theory, the onset of anaerobic metabolism should reduce CMRo 2 but this may be a late phenomenon. A recent clinical trial in infants has demonstrated a deleterious effect of hemodilution [10] and some clinical studies in adult cardiac surgery have correlated poor neurologic outcome with lower intraoperative hemoglobin levels [24, 25]. In addition to the risk of hypoxic injury suggested by this study and in our previous studies, excessive CBF due to hemodilution may possibly increase the embolic load to the brain similar to the effect of ph-stat management in adults. Another possible explanation for the neurologic damage observed in adults with hemodilution is increased edema formation. Limitations of the Study The lowest mean arterial blood pressure in our study was 20 mm Hg in the low hematocrit group during low-flow bypass. Cerebral venous blood was drawn from the jugular bulb and not from the sagittal sinus. Therefore, we cannot exclude the possibility that the venous sample was contaminated by extracerebral blood sources leading to an underestimation of cerebral oxygen metabolism. However, sampling from the jugular bulb to represent brain venous blood is the gold standard in humans [6]. There is no doubt that CBF autoregulation may be altered by age and coexistent disease such as diabetes mellitus, hypertension, or atherosclerotic cerebrovascular disease [6]. It may not be appropriate to extrapolate our results from 4-week old piglets to an adult pig population or to humans. Nevertheless, the lower perfusion pressure that results from hemodilution could only exacerbate a watershed injury from hypoperfusion secondary to carotid or cerebrovascular disease. Furthermore, the greater CBF with hemodilution could only increase the microembolic load if atheromatous debris is dislodged by the manipulations of CPB. In summary, the results of this study call into question the common clinical practice of moderate-to-severe hemodilution during CPB in adults and children. In children there is a risk of hypoxic injury because of the increase in CMRo 2, which appears to result from hemodilution. Adults are at an increased risk with hemodilution secondary to both hypoxic and microembolic injury. Perhaps both of these mechanisms may explain the disturbingly high incidence of permanent cognitive dysfunction observed in adults after CPB [26]. The study was supported by National Institutes of Health grant HL (RAJ) and a Habilitandenstipendium of the Deutsche Forschungegemeinschaft N344/1 (GN). We thank Patricia Dunning from the Department of Radiology and Gene Walter from the Department of Neurology, Children s Hospital, Boston, for their technical support, as well as Sheryl and Jason Balera. We thank Laura Young for preparation of the manuscript. References 1. Neptune WB, Bougas JA, Panico FG. Open heart surgery without the need for donor blood priming in the pump oxygenator. N Engl J Med 1960;262: Bjork VO, Hultquist G. Brain damage in children after deep hypothermia for open-heart surgery. Thorax 1960;15: Shin oka T, Shum-Tim D, Jonas RA, et al. Higher hematocrit improves cerebral outcome after deep hypothermic circulatory arrest. J Thorac Cardiovasc Surg 1996;112: Aoki M, Nomura F, Stromski ME, et al. Effects of ph on brain energetics after hypothermic circulatory arrest. Ann Thorac Surg 1993;55: du Plessis AJ, Jonas RA, Wypij D, et al. Perioperative effects of alpha-stat versus ph stat strategies for deep hypothermic cardiopulmonary bypass in children. J Thorac Cardiovasc Surg 1997;114: Schell RM, Kern FH, Greeley WJ, et al. Cerebral blood flow and metabolism during cardiopulmonary bypass. Anesth Analg 1993;76: Montgomery DC. Design and analysis of experiments. 5th ed. New York: John Wiley & Sons, 2001: Sokal RR, Rohlf FJ. Biometry. The principles and practice of statistics in biological research. 3rd ed. New York: W. H. Freeman, 1995: Wallenstein S, Zucker CL, Fleiss JL. Some statistical methods useful in circulation research. Circ Res 1980;47: Jonas RA, Wypij D, Roth SJ, et al. The influence of hemodilution on outcome after hypothermic cardiopulmonary bypass: results of a randomized trial in infants. J Thorac Cardiovasc Surg 2003;126: Shin oka T, Shum-Tim D, Laussen PC, et al. Relative roles of colloid oncotic pressure and hematocrit in determining outcome after deep hypothermic circulatory arrest. Ann Thorac Surg 1998;65: Murkin JM, Farrar JK, Tweed WA, McKenzie FN, Guiraudon

8 Ann Thorac Surg SAKAMOTO ET AL 2004;77: HEMODILUTION ELEVATES CEREBRAL BLOOD FLOW AND METABOLISM G. Cerebral autoregulation and flow/metabolism coupling during cardiopulmonary bypass: the influence of Paco 2. Anesth Analg 1987;66: Gold JP, Charlson ME, Williams-Russo P, et al. Improvement of outcomes after coronary artery bypass. A randomized trial comparing intraoperative high versus low mean arterial pressure. J Thorac Cardiovasc Surg 1995;110: Govier AV, Reves JG, McKay RD, et al. Factors and their influence on regional cerebral blood flow during nonpulsatile cardiopulmonary bypass. Ann Thorac Surg 1984;38: Rogers AT, Stump DA, Gravlee GP, et al. Response of cerebral blood flow to phenylephrine infusion during hypothermic cardiopulmonary bypass: influence of Paco 2 management. Anesthesiology 1988;69: Rogers AT, Prough DS, Stump DA, et al. Cerebral blood flow does not change following sodium nitroprusside infusion during hypothermic cardiopulmonary bypass. Anesth Analg 1989;68: Schwartz AE, Sandhu AA, Kaplon RJ, et al. Cerebral blood flow is determined by arterial pressure and not cardiopulmonary bypass rate. Ann Thorac Surg 1995;60: Soma Y, Hirotani T, Yozu R, et al. A clinical study of cerebral circulation during extracorporeal circulation. J Thorac Cardiovasc Surg 1989;97: van der Linden J, Priddy R, Ekroth R, et al. Cerebral perfusion and metabolism during profound hypothermia in children. A study of middle cerebral artery ultrasonic variables and cerebral extraction of oxygen. J Thorac Cardiovasc Surg 1991;102: Greeley WJ, Ungerleider RM, Smith L, Reves JG. The effects of deep hypothermic cardiopulmonary bypass and total circulatory arrest on cerebral blood flow in infants and children. J Thorac Cardiovasc Surg 1989;97: Duebener LF, Sakamoto T, Hatsuoka S, et al. Effects of hematocrit on cerebral microcirculation and tissue oxygenation during deep hypothermic bypass. Circulation 2001; 104(suppl I): Todd MM, Tommasino C, Moore S. Cerebral effects of isovolemic hemodilution with a hypertonic saline solution. J Neurosurg 1989;63: Cook DJ, Orszulak TA, Daly RC. Minimum hematocrit at differing cardiopulmonary bypass temperatures in dogs. Circulation 1998;98(suppl 2):II Shaw PJ, Bates D, Cartlidge NE, et al. An analysis of factors predisposing to neurological injury in patients undergoing coronary bypass operations. Q J Med 1989;72: Martin WR, Hashimoto SA. Stroke in coronary bypass surgery. Can J Neurol Sci 1982;9: Newman MF, Kirchner JL, Phillips-Bute B, et al. Longitudinal assessment of neurocognitive function after coronaryartery bypass surgery. N Engl J Med 2001;344: Appendix Derivation of Formulas for the Cerebral Blood Flow and Metabolism Variables CMRo 2 (CBF * AVDo 2 ) [ml o 2 /g/min] AVDo 2 (Cao 2 Cvo 2 ) [ml/100 ml] Cao * Hb (Sao Pao 2 ) Cvo * Hb (Svo Pvo 2 ) CDo 2 CBF*Cao 2 OER (Cao 2 Cvo 2 )/Cao 2 CMRo 2 cerebral metabolism rate of oxygen AVDo 2 arteriovenous oxygen difference Cao 2 arterial oxygen content Cvo 2 venous oxygen content CDo 2 cerebral oxygen delivery OER oxygen extraction ratio 1663 CARDIOVASCULAR INVITED COMMENTARY The concept of higher hematocrit maintained during cardiopulmonary bypass and circulatory arrest is becoming generally accepted, due in part to the innovative basic and clinical research from the group at Boston Children s Hospital, some of which is presented here. The authors of this manuscript easily convinced me that animals in the low hematocrit group have vasodilatation induced increased cerebral blood flow and pressure with evidence of decreased venous ph, which should signal the onset of cerebral anaerobic metabolism. Conversely, they give data that indicate an increased cerebral metabolic rate for oxygen-a counterintuitive result. They explain this unexpected finding as, Perhaps more oxygen is needed to counteract acidosis due to the lower ph buffering capacity of the dilute blood. It is my opinion that the experimental preparation is probably inadequate to obtain an accurate cerebral venous sample. In the pig, the brain is relatively small and has a blood supply of lesser magnitude than the large neck and skull-based muscles when compared to the human. Since the sagittal sinus was not sampled, the jugular venous cannula must be inserted well into the temporal bone to avoid contamination from the multiple systemic venous connections, which enter the jugular vein just below the exit from the skull. The authors do not explain how they verified this position of the catheter, thus leading me to believe that jugular vein cannula position must be at least one factor in explaining this odd result. While it is possible that CMRO 2 is elevated in this experimental situation, it is more likely that despite higher flow, O 2 delivery is reduced, resulting in cellular acidosis. This assertion has been confirmed by a number of previous studies, albeit using different methodology, as the authors mention in their discussion. Nevertheless, the main point of the manuscript is that higher hematocrits than have been traditionally accepted are generally safer, particularly in pediatric cardiac surgery. Their point is well taken and adds to the growing body of literature describing best practices in cardiopulmonary bypass. John W. Hammon, Jr, MD Department of Cardiothoracic Surgery Wake Forest University School of Medicine Medical Center Boulevard Winston-Salem, NC jhammon@wfubmc.edu 2004 by The Society of Thoracic Surgeons /04/$30.00 Published by Elsevier Inc doi: /j.athoracsur