Comparison of pH-state and Alpha-stat Cardiopulmonary Bypass on Cerebral Oxygenation and Blood Flow in Relation to Hypothermic Circulatory Arrest in Piglets 

We studied 16 newborn pigs ages 2–5 days that weighed 1.55 to 2.45 kg. Studies were approved by the Institutional Animal Care and Use Committee at the Joseph Stokes Research Institute, the Children's Hospital of Philadelphia. After anesthetic induction with 33 mg/kg intramuscular ketamine and 3.3 mg/kg acepromazine, the tracheas were cannulated and lungs mechanically ventilated. Catheters were inserted into the external jugular vein to administer drugs and into the femoral artery to monitor pressure (23PXL, Statham), blood gases, and pH (179, Corning, Corning, NY). End-expired carbon dioxide (Normocap, Puritan-Bennett) and arterial pressure were recorded (2000, Gould, Cleveland, OH). Anesthesia was maintained with intravenous fentanyl (a 25 [micro sign]g/kg load and then 10 [micro sign]g [middle dot] kg^-1[middle dot] h^-1), droperidol (0.25 mg [middle dot] kg^-1[middle dot] h^-1), and pancuronium (0.2 mg [middle dot] kg^-1[middle dot] h^-1). During surgical preparation, a circulating water blanket and overhead heating lamp were used to maintain normal body temperature.

After the scalp was resected, three small holes were made in the skull over the right frontoparietal cerebrum to accommodate temperature and laser Doppler flow probes. A thermistor (522, Yellow Springs Instruments, Yellow Springs, OH) was placed subdurally to monitor cortical temperature. Laser Doppler flowmetry optical fibers (0.8 mm diameter, BMP2, Vasamedics, St. Paul, MN) were placed on the dura and 1.5 cm deep to the dura to monitor superficial and deep CBF, respectively. Near infrared spectroscopy (NIRS) optical fibers (CW2000, NIM Incorporated, Philadelphia, PA) were placed on the skull over the left frontoparietal cerebrum to monitor cortical oxygen saturation. Emitter and detector fibers were placed 3 cm apart. Thermistors (401; Yellow Springs Instruments) were also inserted in the rectum and esophagus to monitor core body temperature.

The heart and ascending aorta were exposed through a median sternotomy. After intravenous heparin (200 units/kg) was administered, 8-French arterial and 16-French venous cannulas were inserted in the proximal ascending aorta and right atrial appendage for CPB. The CPB circuit used a bubble oxygenator (bio-2, Baxter Healthcare, Irvine, CA) receiving oxygen at 500 ml/min and a nonpulsatile roller pump (RS 7800, Renal Systems, Minneapolis, MN) flowing at 150 ml [middle dot] kg^-1[middle dot] min (^-1). For the pH-stat group, the oxygenator received 5% carbon dioxide in a balance of oxygen; for the alpha-stat group, the oxygenator received only oxygen. [5,6,8,9]The pump prime (approximately 400 ml) contained pig whole blood, 1,500 units heparin, 25 [micro sign]g/kg fentanyl, 2 mg pancuronium, 500 mg calcium chloride, and 25 mEq sodium bicarbonate. Plasma-lyte A (Baxter Healthcare, Deerfield, IL) was added to yield a hematocrit of 20–25% during CPB.

Cardiopulmonary bypass perfusate temperature was controlled by a water bath heater-cooler system (1141, VWR Scientific, Philadelphia, PA). During CPB cooling, the water bath temperature was adjusted to keep the perfusate at approximately 10 [degree sign]C less than body temperatures, and arterial blood was sampled every few minutes to confirm pH-stat or alpha-stat management. When the cortex became 19 [degree sign]C, the CPB pump was turned off and the heart was bathed in ice water until it was asystolic. Cortical temperature was maintained as close as possible to 19 [degree sign]C during arrest by positioning ice bags around the head. After 60 min of arrest, CPB was resumed at 150 ml [middle dot] kg^-1[middle dot] min^-1. The water bath temperature was adjusted to keep perfusate 5–10 [degree sign]C greater than body temperatures. The maximum perfusate temperature was 38 [degree sign]C. Rewarming was facilitated with a circulating warm water blanket and overhead heating lamp. After 20 min of reperfusion, mechanical ventilation was resumed, carbon dioxide in the pH-stat group was discontinued, and the heart was defibrillated if needed. After 25 min of reperfusion, CPB was discontinued, the cannulas were removed, and 4 mg/kg protamine was administered intravenously. Pig whole blood was transfused intravenously as needed to maintain mean arterial pressure >50 mmHg. The inspired oxygen concentration and minute ventilation were adjusted to maintain arterial P^CO(2) at 35–40 mmHg and arterial P^O(2) at >75 mmHg.

Near infrared spectroscopy is an optical technique to assess tissue oxygenation. It relies on the relative transparency of biologic tissues to near infrared light (700–900 nm), where oxygenated and deoxygenated hemoglobin have distinct absorption spectra. By measuring light attenuation at wavelengths where the extinctions of oxygenated and deoxygenated hemoglobin differ, it is possible to monitor Sc^O(2). [14–16]Near infrared spectroscopy monitors hemoglobin located mainly in small gas-exchanging vessels (capillaries, venules, arterioles) within a “banana-shaped” tissue volume between the optical fibers. [17–19]Recent work [15,17,20]indicates that a 3-cm optical fiber separation monitors tissue up to 2 cm deep to the optical fibers, which in piglets encompasses the superficial and deep neocortex. The scalp and skull do not interfere with NIRS in piglets. [15]Near infrared spectroscopy Sc^O(2) represents a mixed vascular oxygen saturation weighted toward capillary and venous saturation and reflects oxygen extraction by the tissue. Near infrared spectroscopy data are presented relative to baseline because uncertainties in optical path length and light scattering preclude determination of baseline Sc^O(2).

Laser doppler flowmetry is an optical method to assess tissue blood flow. It relies on the absorption and backscattering of near infrared light where oxygenated and deoxygenated hemoglobin are isosbestic ([approximately] 800 nm). Light is scattered by both stationary tissue particles (e.g., neurons, mitochondria) and moving erythrocytes. Light scattered by moving erythrocytes results in a Doppler frequency shift of the light, whereas light scattered from stationary tissue particles remains unshifted. The frequency-shifted backscattered light yields information about the erythrocyte velocity, whereas the attenuated backscattered light yields information about blood volume. Tissue blood flow is a function of erythrocyte velocity and volume. Laser Doppler flowmetry monitors flow in small blood vessels (capillaries, venules, arterioles) located in approximately 1 mm³of tissue beneath the probe. Blood flow in superficial neocortex can be monitored from the dural surface; blood flow in deeper regions can be monitored by insertion of an optical fiber into the brain. [21]Laser Doppler flowmetry data are presented relative to baseline because uncertainties in optical path length and light scattering preclude determination of baseline CBF.

In our study, Sc^O(2), CBF, and cortical temperature were archived to a computer every 15 s. The flow probes in the superficial and deep neocortex were placed to monitor the superficial and deep cortical domain monitored by NIRS in an attempt to examine blood flow and oxygenation in similar brain regions. The signal average time for NIRS and laser Doppler flowmetry instruments was 15 s.

Brain tissue adenosine triphosphate and total protein concentrations were measured by the lucifer-luciferase and Lowry methods. [22,23]Before the piglets were killed, a hole was created in the skull and dura over the left hemisphere between the NIRS optical fibers. A biopsy gun containing liquid nitrogen was directed through the hole and a plug of brain tissue approximately 1.5 cm long was aspirated and instantly frozen. Superficial neocortex (the top half of the plug) and deep neocortex (the bottom half of the plug) were stored at -70 [degree sign]C. After the piglets died, the skull over the right hemibrain was removed. The superficial and deep neocortex were separated, weighed, placed in an oven (60 [degree sign]C) for 72 h, and then reweighed. The water content (%) of the tissue was 100 x (desiccated weight/wet weight).

Piglets were randomly assigned to alpha-stat or pH-stat groups before surgical preparation. Arterial blood gases and pH were recorded before, during, and after CPB was discontinued. During circulatory arrest, the curvilinear decrease in Sc^O(2) can be described by an exponential function, [24]Equation 1where t is the time after the onset of arrest and k and a are constants. This function may be expressed in terms of Sc^O(2) half-life (t^1/2), given by Equation 2The Sc^O(2) versus the time curve during circulatory arrest in each animal was curve-fit to Equation 1by least squares regression and k was determined. The Sc^O(2) half-life was calculated according to Equation 2. Termination of cerebral oxygen metabolism during arrest was identified from the Sc^O(2) versus time curve when Sc^O(2) had decreased to a nadir and remained unchanged thereafter. At the end of the experiment, animals were killed with 100 mg/kg intravenous pentobarbital.

Data are presented as means +/- SD. Comparisons between groups or across time were made by analysis of variance. When significant F values existed, multiple means were compared by Turkey's test. Significance was defined as P < 0.05.

There were no significant differences in physiologic data between the pH-stat and alpha-stat groups during the study except, as intended, in arterial pH and P^CO(2) during CPB (Table 1).

(Figure 1) shows cortical temperature and Sc^O(2) during CPB cooling in the pH-stat and alpha-stat groups. In both groups, Sc^O(2) increased while cortical temperature decreased, although Sc^O(2) increased significantly more with pH-stat than with alpha-stat management (123 +/- 12% vs. 80 +/- 9%; P < 0.001).

Cortical temperature was less with pH-stat than with alpha-stat management at 5 and 10 min CPB (P < 0.05) but not thereafter. Cooling duration varied among the animals: with pH-stat, all eight piglets were cooled for 10 min, six for 15 min, and none for 20 min; with alpha-stat bypass, all eight piglets were cooled for 15 min and four were cooled for as long as 20 min. The CPB cooling duration was less with pH-stat compared with alpha-stat management (15 +/- 2 min vs. 18 +/- 2 min; P < 0.05). The CPB perfusate temperature was similar with the alpha-stat and pH-stat groups (14 +/- 1 [degree sign]C vs. 16 +/- 3 [degree sign]C at 5 min; 12 +/- 2 [degree sign]C vs. 13 +/- 3 [degree sign]C at 10 min; 10 +/- 1 [degree sign]C vs. 12 +/- 3 [degree sign]C at 15 min).

The CBF response to CPB cooling was significantly different in the pH-stat and alpha-stat groups (Figure 2). In the alpha-stat group, superficial and deep CBF decreased steadily during CPB cooling, whereas in the pH stat group, they increased initially and then decreased. At the end of cooling, superficial and deep CBF were -56 +/- 22% and -29 +/- 28% in the alpha-stat and 22 +/- 25% and 3 +/- 19% in the pH-stat groups (P < 0.001 alpha-stat vs. pH-stat; P = NS superficial vs. deep CBF).

During hypothermic arrest, Sc^O(2) decreased curvilinearly in both groups (Figure 3). The Sc^O(2) half-life was increased in pH-stat compared with the alpha-stat group (10 +/- 1 min vs. 7 +/- 2 min; P < 0.01), indicating that brain hemoglobin desaturated significantly faster in the alpha-stat group. Cortical deoxygenation was slower in the pH-stat compared with the alpha-stat group (35 +/- 6 min vs. 24 +/- 7 min; P < 0.01)). The average cortical temperature during arrest, calculated from all temperatures archived in each animal from the beginning to the end of arrest, were similar with alpha-stat and pH-stat bypass (18.9 +/- 0.4 [degree sign]C vs. 19.2 +/- 0.5 [degree sign]C; P = NS).

During reperfusion, both groups had similar cortical temperatures and Sc^O(2) values (Figure 4). The Sc^O(2) increased immediately with CPB reperfusion to above baseline and then decreased to baseline with rewarming and further reperfusion. The CPB time to 36 [degree sign]C cortex was similar in the pH-stat and alpha-stat groups (23 +/- 5 min vs. 25 +/- 3 min; P = NS).

The CBF response to reperfusion was significantly different between the pH-stat and alpha-stat groups (Figure 5). Although superficial and deep CBF increased immediately with CPB reperfusion in both groups, they were significantly greater in the pH-stat than in the alpha-stat group, from 15 to 25 min when the brain was warm (30–35 [degree sign]C). Superficial and deep CBF had returned to baseline by 15 min reperfusion in the pH-stat group, whereas in the alpha-stat group, superficial CBF remained significantly depressed until 80 min reperfusion, while deep CBF remained significantly depressed throughout reperfusion.

Cortical physiologic recovery 2 h after discontinuing CPB was improved in the pH-stat compared with the alpha-stat group (Table 2). Water content in the superficial and deep neocortex was significantly greater in the alpha-stat group, suggesting cerebral edema in the alpha-stat group. Deep CBF and adenosine triphosphate concentrations were significantly less in the alpha-stat group, suggesting cerebral hypoperfusion and impaired energy metabolism in the alpha-stat group. Superficial CBF and adenosine triphosphate concentrations and Sc^O(2) were similar for the two groups.

Deep hypothermic circulatory arrest is used widely to repair complex heart defects in neonates, and it involves CPB to cool the body before arrest and to rewarm it afterward. Arterial pH and P^CO(2) during CPB may be managed by pH-stat or alpha-stat strategies, which, in turn, may render differences in neurologic recovery. Purported advantages of pH-stat include (1) decreased brain metabolic rate, [13,25]which may slow substrate consumption during arrest;(2) increased CBF, [5–7]which may improve brain resuscitation;(3) increased brain cooling rate, [9]which may improve brain hypothermia during arrest; and (4) decreased oxyhemoglobin affinity, [26]which may enhance oxygen availability. Purported disadvantages of pH-stat include increased CBF, which may increase brain emboli, and decreased pH, which may increase free radical-mediated brain damage. [27]In support of the first three pH-stat advantages, the current study found slower cortical deoxygenation during arrest, increased CBF during reperfusion, and increased brain cooling rate with pH-stat compared with alpha-stat management.

Cortical deoxygenation during hypothermic circulatory arrest was assessed by NIRS. The NIRS hemoglobin desaturation curve during circulatory arrest reflects cortical oxygen metabolism during ischemia. Mathematical modeling of this curve (Equation 1) can be used to describe the kinetics of tissue deoxygenation. [24]The model contains an offset constant and a rate constant, corresponding to the initial substrate concentration and enzymatic rate constant in a first-order biochemical reaction. The substrate concentration, oxygen, is monitored by the disappearance of Sc^O(2). Prearrest Sc^O(2) indicates the initial substrate concentration and Sc (^O)(2) half-life the enzymatic rate constant and cerebral oxygen metabolic rate. [16,24]We found that pH-stat management increased both prearrest Sc (^O)(2) and Sc^O(2) half-lives, indicating pH-stat increased cortical oxygen concentration before arrest and decreased cortical oxygen metabolic rate during arrest. By increasing oxygen supply and decreasing oxygen metabolic rate, pH-stat slowed cortical deoxygenation during arrest. The change in supply and metabolic rate appeared to contribute similarly to slowing the deoxygenation, because the prearrest Sc^O(2) and Sc^O(2) half-lives were each about 50% greater with pH-stat compared with alpha-stat bypass (123 vs. 80%, 10 vs. 7 min).

The presence of cortical oxygen metabolism during arrest, however, does not necessarily imply oxygen sufficiency or protection against brain damage, because the NIRS desaturation curve reflects oxygen disappearance from the cerebrovasculature and not the status of parenchymal intracellular energetics or activity of injurious reactions. As the tissue oxygen concentration decreases, high-energy phosphate concentrations and the oxidative-reductive state of electron transport proteins change to sustain oxidative metabolism, until anoxia occurs, and then oxidative metabolism ceases. [28]During the transition from oxygen sufficiency to anoxia, the oxygen concentration becomes rate limiting and the oxygen metabolic rate decreases. In our study, cortical oxygen metabolism disappeared 24 min into arrest with alpha-stat and 36 min with pH-stat management. When intracellular energetics began to change is uncertain. As tissue oxygen concentration decreases, a host of injurious reactions occurs, including excitotoxity, calcium influx, glycolysis, and free radical generation, [29]which pH-stat management may delay by slowing deoxygenation. Further, the acidosis associated with pH-stat may inhibit many of these reactions independent of its effect on tissue oxygen concentration. [10–12,30,31]Thus it is possible for pH-stat to protect against ischemic damage by several mechanisms.

The rate of hemoglobin desaturation depends on many factors: It varies directly with tissue oxygen demand, oxygen diffusivity, oxygen solubility, the partial pressure of oxygen at 50% saturation (P⁵⁰), and the hemoglobin-oxygen dissociation rate constant; and indirectly with tissue hemoglobin concentration and erythrocyte surface and volume. [33]Of these factors, tissue hemoglobin concentration, P⁵⁰, and tissue oxygen demand depend on pH and P^CO(2). [25,26,33–35]Because the increase in P⁵⁰associated with pH-stat would increase the hemoglobin desaturation rate during ischemia [26](which is the opposite of what we observed), P⁵⁰cannot explain the increased Sc^O(2) half-life with pH-stat. Tissue hemoglobin concentration refers to hemoglobin in gas-exchanging vessels (e.g., capillaries). Hypercapnia can dilate capillaries to increase the tissue hemoglobin concentration. [35]pH-stat management during deep hypothermic CPB can also depress cerebral oxygen demand. [25,33]The increased Sc^O(2) half-life with pH-stat management thus may have resulted from decreased cortical oxygen demand and increased tissue hemoglobin concentration.

Besides decreasing cortical oxygen demand, pH-stat management also increased cortical oxygen supply before circulatory arrest. Tissue oxygen supply consists of dissolved oxygen and hemoglobin-bound oxygen, the latter the product of saturation and concentration. pH-stat bypass increased saturation and may have increased the tissue hemoglobin concentration (hypercapnic vasodilation). [35]At normothermia, dissolved oxygen contributes minimally to the total tissue oxygen concentration, whereas at deep hypothermia it contributes substantially more. [34,36]The dissolved oxygen concentration is the product of tissue P^O(2) and the oxygen solubility constant. Because pH and P^CO(2) do not alter the oxygen solubility constant, [34]any difference in the dissolved oxygen concentration between the pH-stat and alpha-stat groups would have resulted from differences in tissue P^O(2). Because Sc^O(2) was greater in the pH-stat group before arrest (Figure 1), and acidosis shifts the oxyhemoglobin curve to the right, [26]brain tissue P^O(2) was greater in the pH-stat group, and therefore the dissolved oxygen concentration was greater as well. However, we cannot quantify the dissolved oxygen concentration difference between the groups, given the uncertainties in tissue pH and other factors that affect the oxyhemoglobin dissociation curves.

Near infrared spectroscopy Sc^O(2) represents a mixed vascular saturation of capillary, venular, and arteriolar blood, weighted toward capillary and venous saturation. [14,16,18,20]Because arterial saturation was constant (100%) during CPB cooling in our study, the increase in Sc^O(2) indicated that saturation increased in capillaries and venules, the increase being greater with pH-stat management. This finding is consistent with less cerebral oxygen extraction by pH-stat CPB cooling compared with alpha-stat CPB cooling [33]and may have resulted from the greater CBF during pH-stat CPB.

Improved brain hypothermia has been proposed as an advantage of pH-stat management. In our piglet model, deep hypothermia in the cortex (19 [degree sign]C) is achieved a few minutes faster with pH-stat than with alpha-stat CPB cooling (14 +/- 2 min vs. 17 +/- 3 min), [9]as confirmed in the current study (15 +/- 2 min vs. 18 +/- 2 min). [9]Our current study conducted CPB cooling to a target cortical temperature regardless of cooling duration to control for cooling rate differences between the groups. Cortical temperature heterogeneity did not exist, because alpha-stat and pH-stat yield minimal temperature variation (<1.5 [degree sign]C) across the cortex after cooling with the CPB technique of this study. [9]Thus our findings did not originate from cortical temperature differences between pH-stat and alpha-stat groups.

Another purported advantage of the pH-stat strategy has been the increased CBF during reperfusion. [5–7]Although this increased CBF might improve brain resuscitation, it might also indicate hypercapnic vasodilation or less ischemic damage after hypothermic arrest. Several observations suggest the latter. Cerebral hypoperfusion is a sign of ischemic injury. [29]During recirculation, cerebral hypoperfusion did not last as long in the pH-stat group (Figure 5). Greater reperfusion CBF in the pH-stat group was not a result of hypercapnia, because hypercapnic cerebral vasodilation is lost after hypothermic arrest. [37] 

Potential disadvantages of pH-stat management include the risk of embolic and free radical-mediated brain damage. [5,6,27,38]Increased CBF during CPB may increase delivery of gaseous and particulate matter to the brain. Acidosis enhances oxygen free radical generation in vitro, although hypercapnic acidosis enhances it less than metabolic acidosis. [27]Whether pH-stat management increases oxygen free radical damage after hypothermic arrest remains unknown.

In a retrospective study, Jonas et al. [4]observed a strong association between higher arterial P^CO(2) before deep hypothermic arrest and improved developmental outcome in children after neonatal heart surgery, suggesting cerebral protection with the pH-stat strategy. These investigators also found improved cerebral physiologic recovery with pH-stat management in a piglet model of deep hypothermic circulatory arrest. [5,6]However, brain temperature was not measured in these studies, and thus outcome may have resulted from temperature rather than pH management. As we did, Kuluz et al. [7]observed improved cerebral physiologic recovery with pH-stat management in a piglet model controlling for brain temperature, indicating an effect of pH strategy. In a randomized clinical trial of neonates undergoing cardiac surgery using deep hypothermic arrest, duPlessis et al. [8]found faster electroencephalographic recovery and fewer postoperative seizures with pH-stat compared with alpha-stat bypass. Among the limitations of this clinical study and the animal studies is the lack of long-term neurologic outcome assessment. Despite this limitation, together the studies provide evidence for cerebral protection by pH-stat management in neonates undergoing deep hypothermic circulatory arrest. Our results indicate that the kinetics of cerebral deoxygenation might contribute to the mechanism of protection.

The authors thank Barbara Simon and John McCann for technical assistance.