Operative repair of aortic aneurysms or aortic dissections extending into the aortic arch requires temporary interruption of antegrade blood flow through the cerebral vessels. The primary technique used to protect against cerebral injury during the period of circulatory arrest has been deep hypothermia [1,2] [3-7]
Delivery of metabolic substrate to prevent cerebral ischemia and infarction during the temporary interruption of antegrade cerebral perfusion has been one rationale for using RCP. However, existing experimental and clinical studies have not consistently demonstrated that RCP provided sufficient nutrient flow to support cerebral metabolism and prevent cerebral ischemia during HCA. Some experimental models have demonstrated cerebral tissue flow via RCP as high as 33%-50% of that achieved with antegrade perfusion flow rates that were sufficient to support cerebral metabolic requirements [8,9] [10] [11-13] [14] [15] [15] [16]
The purpose of this study was to examine the time course of cerebral metabolic activity during deep hypothermia after interruption of antegrade cerebral perfusion when RCP was used by measuring serial concentrations of oxygen, carbon dioxide, and pH in RCP inflow and outflow specimens. The time course for changes in the oxygen extraction ratio was compared with the known functional changes in brain electrical activity during HCA demonstrated by somatosensory evoked potential (SEP) monitoring [16]
Methods
Intraoperative electroencephalogram (EEG) and SEP monitoring together with serial arterial (outflow) and venous (inflow) blood sampling was performed in 28 consecutive patients undergoing aortic reconstruction. Each patient had a formal neurological evaluation with a focused neurological examination before the operation and a complete neurological examination on the first or second day after the operation. Patients with abnormalities on either the initial or immediate postoperative neurological examination were followed with serial examinations. The circulation management strategy used for all patients consisted of: 1) cooling to deep hypothermia for 3 min after an isoelectric EEG and disappearance of the N20-P22 cortical SEP had been achieved; 2) partial exsanguination with interruption of antegrade cerebral perfusion; and 3) RCP (Table 1 ). Clinical data used for study purposes were acquired according to a protocol approved by the University of Pennsylvania Committee on Studies Involving Human Beings.
Table 1: Patient Characteristics
All patients were anesthetized with fentanyl (25-75 [micro sign]g/kg), midazolam (0.03-0.15 mg/kg), isoflurane (0.0-0.5 vol%), and pancuronium (0.25-0.5 mg/kg). No anesthetics were administered into the retrograde cerebral perfusate.
Cardiopulmonary bypass was instituted using standard bicaval cannulation and cannulation of either the left femoral artery, ascending aorta, or aortic arch. A 26F superior vena cava cannula was used to allow greater cephalad flow during RCP and positioned in the superior vena cava so that its tip was just cephalad to the insertion of the azygous vein. The left ventricle was vented via the right superior pulmonary vein. Patients were cooled on cardiopulmonary bypass for a minimum of 30 min. Antegrade cerebral perfusion was not interrupted until the EEG became isoelectric (<2 [micro sign]V for 3 min) and the N20-P22 SEP disappeared (amplitude <0.05 [micro sign]V) bilaterally. At that time, the patient was partially exsanguinated, the superior vena cava was snared between the right atrium and azygous vein, and the cardiopulmonary bypass circuit was converted to a system for delivering RCP [4]
Serial blood samples were obtained during HCA at different times after the interruption of antegrade cerebral perfusion. At each sampling time, blood was collected simultaneously from the RCP inflow (oxygenated blood entering the superior vena cava) and RCP outflow (blood trickling out of the left carotid or distal innominate arteries). Measurements of PO2 , PCO2 , pH, and co-oximetry were performed immediately on the blood samples and measured at 37[degree sign]C. Alpha-stat analysis in which measured values were not temperature-corrected to the patient's actual body temperature was used. Oxygen content was calculated using the formula: Equation 1 where SO2 is oxygen saturation (%) of the sample measured by co-oximetry, Hb is the concentration of hemoglobin (g/dL) of the sample measured by co-oximetry, and PO2 is the oxygen tension (mm Hg) in the sample. The oxygen extraction ratio (OER) was calculated from the formula: Equation 2 where O2 contentRCP inflow is the calculated oxygen content (vol %) of blood in the RCP perfusate entering the superior vena cava and O2 contentRCP outflow is the oxygen content (vol %) of blood collected from the left carotid or innominate artery.
All measured variables were treated as continuous variables. Repeated-measures analysis of variance was used to test for changes in measured variables during HCA. Spearman rank correlation analysis was used to determine whether patient-specific characteristics such as age, hemoglobin concentration, temperature at the time of HCA, and fraction of no-flow time were related to the changes in PO2 , PCO2 , or pH of the retrograde outflow. Multiple linear regression analysis was used to test whether the following factors-time after circulatory arrest, preoperative stroke, or intraoperative stroke-affected the OER or the RCP inflow-outflow gradients for SO2 , O2 content, PCO2 , or pH. Nonlinear regression analysis was performed to quantify further the change in OER over time during HCA with RCP.
Results
Nineteen patients (9 male and 10 female) did not have preoperative or postoperative strokes. Five patients had preoperative strokes, three patients had isolated intraoperative strokes, and one of the five patients with preoperative strokes also had an intraoperative stroke (Table 1 ). All 19 of the patients without strokes survived the operation. The mean (+/- SD) age of the patients without stroke was 66 +/- 12 yr, duration of HCA was 39 +/- 15 min, duration of RCP was 34 +/- 13 min, duration of no-flow was 5 +/- 4 min, fraction of no-flow time was 0.12 +/- 0.07, RCP flow rate was 220 +/- 83 mL/min, Hb in the RCP perfusate was 8.0 +/- 1.5 g/dL, and nasopharygeal temperature during HCA was 14.3 +/- 1.1[degree sign]C (Table 1 ). The mean nasopharyngeal temperature at the onset of HCA was 14.3[degree sign]C, and at the end of HCA was 14.2[degree sign]C, a change that was not statistically significant (P = 0.36).
Changes in the PO2 and PCO2 of blood sampled from the RCP inflow and outflow over time after HCA was analyzed from individual patients and from pooled data that included all patients. In 12 of 15 patients (80%) who had blood samples from at least two time points after HCA, the PCO2 in the RCP outflow increased between the first and second measurements (Figure 1 A). In four of five patients (80%) who had blood samples from at least three time points after HCA, the PCO2 in the RCP outflow also increased over time. The mean rate of increase in PCO2 in individual patients was 0.24 mm Hg/min with a range of -0.16 to 0.64 mm Hg/min. The mean rate of increase estimated from pooled data that included all patients was 0.43 mm Hg/min.
Figure 1: Changes in PCO2 (A) and PO2 (B) in retrograde cerebral perfusion (RCP) outflow blood specimens measured at two different time points in individual patients who had two or more serial measurements available for analysis (n = 15). The mean (+/- SD) time at which the first sample was obtained was 7 +/- 4 min after the initiation of hypothermic circulatory arrest (HCA). The second sample was obtained 26 +/- 11 min after the initiation of HCA.
In 13 of 15 patients (87%) who had blood samples from at least two time points after HCA, the PO2 in the RCP outflow decreased between the first and second measurements (Figure 1 B). In four of five patients (80%) who had blood samples from at least three time points after HCA, the PO2 in the RCP outflow also decreased over time. The mean rate of decrease in PO2 over time in individual subjects was 1.5 mm Hg/min, with a range of -0.63 to 4.5 mm Hg/min. The mean rate of decrease estimated on the basis of pooled data from all patients was 1.0 mm Hg/min.
Patient-specific factors did not affect the rate of change in any of the measured variables during HCA with RCP in the group of patients without strokes. The Spearman rank correlation among patient age, nasopharyngeal temperature, hemoglobin concentration, and fraction of no-flow time during HCA with RCP showed no relation (P > 0.2) to the rate of decrease in PO2 , the rate of increase in PCO2 , and the rate of decrease in pH.
Linear regression analysis performed on the pooled data from patients without stroke demonstrated a significant increase in OER (slope = 0.01 min-1 , r = 0.84, P < 0.001) and a decrease in the O2 saturation of the RCP outflow as a function of time after initiation of HCA (Table 2 ). In addition, the O2 content difference, PCO2 difference, and pH difference between the RCP inflow and outflow samples increased significantly over time after initiation of HCA with RCP in the group of patients without stroke (Table 2 , Figure 2 ). The minimal oxygen content difference measured was 0.6 vol% 4 min after interruption of antegrade cerebral perfusion, and the maximal oxygen content difference measured was 8.8 vol% 53 min after interruption of antegrade cerebral perfusion in the entire group of patients studied (Figure 2 ). The minimal OER measured was 0.07 at 1 min after interruption of antegrade cerebral perfusion, and the maximal OER measured was 0.66 at 53 min after interruption of antegrade cerebral perfusion in the entire group of patients studied.
Table 2: Effects of Time After Initiation of Hypothermic Circulatory Arrest and Neurologic Condition on Metabolic Variables Measured in the Retrograde Cerebral Perfusate
Figure 2: Changes in the retrograde cerebral perfusion (RCP) inflow-outflow O2 content difference (A) and RCP inflow-outflow PCO2 difference (B) as a function of time after the initiation of hypothermic circulatory arrest (HCA) for patients without strokes ([circle, open]; n = 19), with preoperative stroke ([up triangle, filled]; n = 4), with intraoperative stroke ([black square]; n = 3), and with both preoperative and intraoperative stroke (*; n = 1). Linear regression analysis (fitted line) indicated a change in RCP inflow-outflow O2 content difference of 0.11 vol%/min (r = 0.82, P < 0.001) and a change in the RCP inflow-outflow PCO2 difference of -0.73 mm Hg/min (r = 0.82, P < 0.001) over time after HCA in patients without stroke.
Nonlinear regression analysis on the effect of time after HCA on the OER in patients without strokes using a standard bounded monotonic response function yielded the Equation OER = 0.82 - 0.71e-t/40 .3 as the best fit with a correlation coefficient of r2 = 0.71 (Figure 3 ). According to this model, the OER reaches 0.5 of its maximal value of 0.66 at 15 min after HCA.
Figure 3: Changes in brainstem (N18) somatosensory evoked potential amplitudes ([center dot]) over time after initiation of hypothermic circulatory arrest (HCA) superimposed on the change in oxygen extraction ratio (OER) after initiation of HCA in patients without strokes ([large circle]; n = 19), with preoperative stroke ([up triangle, filled]; n = 4), with intraoperative stroke ([black square]; n = 3), and with both preoperative and intraoperative stroke (*; n = 1). Serial brainstem (N18) somatosensory evoked potentials were recorded from seven sites from a group of 15 patients undergoing deep hypothermic circulatory arrest using RCP in a previous study
[16] . The amplitude of the N18 somatosensory potentials was normalized to the amplitude just before the interruption of antegrade cerebral perfusion. The normalized amplitude of the N18 somatosensory evoked potentials were fitted to a standard decay function: A = A
o/ [1 + (t/t
50 )
P ], where A
0 = the amplitude of the N18 somatosensory evoked potential at the start of HCA, t = the elapsed time in minutes after HCA, t
50 = the elapsed time in minutes at which the N18 somatosensory evoked potential amplitude decreased to half of A
0 , and P = a variable that reflected the steepness of the curve. The normalized N18 somatosensory evoked potential decayed to half its initial value 16 min after interruption of antegrade cerebral perfusion. The change in OER over time in patients without stroke fitted the function OER = 0.82-0.71e (
-t/40 ).3 (r = 0.84) where t = the time in minutes after the initiation of HCA. The OER decreased to half its maximal detected value of 0.66 at 17 and 15 min after the initiation of HCA based on the linear and nonlinear regression equations, respectively.
Patients with either preoperative or intraoperative strokes exhibited less of an O2 content difference and PCO2 difference in the RCP inflow and outflow specimens over time after initiation of HCA with RCP (Figure 2 ). Multiple linear regression analysis was performed to quantify the effect of stroke on the changes in O2 saturation, O2 content difference, OER, PCO2 difference, and pH difference in the RCP inflow and outflow samples as a function of time after initiation of HCA. The measured variables were fitted to the equation Equation 3 where Y (the dependent variable) was the measured variable, t was the time in minutes after the initiation of HCA with HCA, postoperative stroke = 1 if the patient had a preoperative stroke and 0 if the patient did not have a preoperative stroke, and intraoperative = 1 if the patient had an intraoperative stroke and 0 if the patient did not have an intraoperative stroke. Although the number of patients with strokes was small, multiple regression analysis suggested a significant effect of stroke that tended to decrease the magnitude of changes in O2 saturation, O2 content difference, OER, PCO2 difference, and pH difference of the RCP inflow and outflow samples over time after the initiation of HCA (Table 2 ).
Discussion
The finding that OER across the brachiocephalic vascular bed increased over time after interruption of antegrade cerebral perfusion when RCP was used indicates continuing metabolic activity during HCA. Although samples used for analysis consisted of pooled blood from the brachiocephalic vessels, it is likely that the measurements made on RCP outflow blood specimens did, to some degree, reflect cerebral metabolic activity. Support for this conclusion comes, in part, from the observation that RCP outflow oxygen content, PCO2 , and pH differences were less in patients with strokes compared with patients without strokes. This finding was based on a limited number of measurements but was expected as a consequence of reduced cerebral metabolic activity from cerebral infarction. Additional support for the contention that the RCP outflow measurements provided a measure of cerebral metabolic activity during HCA was the consistent increase in OER that paralleled the decrease in amplitude of brainstem somatosensory evoked potentials over time during HCA that was observed in a previous study [16] [20,21] Figure 3 , the time for the brainstem somatosensory evoked potential to decrease to half of its initial amplitude was 16 min, which is similar to the 15 min it took for the OER to increase to half its maximal value.
The metabolic changes observed in the study are also consistent with the metabolic changes observed in animal studies in which it was possible to perform direct measurements of cerebral metabolism during HCA with RCP. Significant gradients in oxygen saturation, PO2 , PCO2 , and pH across the cerebral vascular bed were observed after the interruption of antegrade cerebral perfusion in baboons subjected to HCA with RCP [10] [11,17] [9]
Clinical studies verifying the ability to perfuse the brain via RCP provide further support for the assumption that blood specimens from the RCP outflow were a sampling of blood that had perfused regions of the brain. Methods that have been used to verify the presence of cerebral blood flow during HCA with RCP in humans include intraoperative brain perfusion imaging using technetium Tc 99m-labeled d,L-hexamethyl propylene amine oxime [18] [15] [19] [14]
Sampling blood from the RCP outflow to assess cerebral metabolic activity during HCA may provide a means to determine the safe duration for HCA with RCP by detecting the time point when relative cerebral ischemia becomes cerebral infarction signaled by an acute decrease in the OER. Studies in humans using positron emission tomography to measure regional OER in the brain have demonstrated that symptomatic cerebral ischemia associated with stroke was associated with acute increases in the OER of affected brain regions [20,21] [20,21] [20] [22] 2 ) would signal infarction. Further investigation with larger numbers of patients, a wider range of HCA duration, and the inclusion of patients with intraoperative strokes would be required to verify this hypothesis.
An alternative explanation for the time course of changes in oxygen extraction, PCO2 , and pH in the RCP outflow specimens was that the transit time of blood and retrograde cerebral perfusate across the cerebral vascular bed was slow. A slow transit time would mean that the composition of blood sampled from the RCP outflow shortly after the onset of HCA consisted mainly of blood that had been pooled in the arterial circulation that had not perfused capillary beds in the brain. Over time, the composition of blood from the RCP outflow would more closely resemble the composition of blood perfusing capillary beds in the central nervous system. Variable shunting of blood away from cerebral vascular beds toward other vascular beds with lower metabolic activity at the onset of RCP could also explain the lower OER at the beginning of HCA. Finally, changes in oxygen extraction, PCO2 , and pH in the RCP outflow samples could also reflect metabolic changes outside the central nervous system caused by nonselective perfusion of the upper body via RCP. This consideration prompted the attempt to obtain RCP outflow samples from the distal carotid arteries when possible. However, these reasons would not explain the differences observed in the group of patients who had strokes.
In summary, RCP inflow and outflow blood samples were measured to assess the time course of metabolic changes across the brachiocephalic circulation during HCA when antegrade cerebral perfusion was interrupted. The observation that oxygen extraction was near maximal after 40-60 min of HCA with RCP is consistent with other studies that suggest relative hypoperfusion during RCP. Reduced cerebral metabolism in infarcted regions may explain the lower OER observed in stroke patients and deserves further investigation.
REFERENCES
1. Svensson LG, Crawford ES, Hess KR, et al. Deep hypothermia with circulatory arrest. J Thorac Cardiovasc Surg 1993;106:19-31.
2. Griepp EB, Griepp RB. Cerebral consequences of hypothermic circulatory arrest in adults. J Card Surg 1992;7:134-54.
3. Ueda Y, Miki S, Kusuhara K, et al. Surgical treatment of the aneurysm of dissection involving the ascending aorta and aortic arch utilizing circulatory arrest and retrograde cerebral perfusion. J Jpn Assoc Thorac Surg 1988;36:161-6.
4. Bavaria JE, Woo YJ, Hall RA, et al. Retrograde cerebral and distal aortic perfusion during ascending and thoracoabdominal aortic operations. Ann Thorac Surg 1995;60:345-53.
5. Deeb GM, Jenkins E, Bolling SF, et al. Retrograde cerebral perfusion during hypothermic circulatory arrest reduces neurologic morbidity. J Thorac Cardiovasc Surg 1995;109:259-68.
6. Murase M, Maeda M, Koyama T, et al. Continuous retrograde cerebral perfusion for protection of the brain during aortic arch surgery. Eur J Cardiothorac Surg 1993;7:597-600.
7. Bavaria JE, Pochettino A. Retrograde cerebral perfusion (RCP) in aortic arch surgery: efficacy and possible mechanisms of brain protection. Semin Thorac Cardiovasc Surg 1997;9:222-32.
8. Usui A, Oohara K, Liu T, et al. Determination of optimum retrograde cerebral perfusion conditions. J Thorac Cardiovasc Surg 1994;107:300-8.
9. Watanabe T, Iijima Y, Abe K, et al. Retrograde brain perfusion beyond the venous valves: hemodynamics and intracellular pH mapping. J Thorac Cardiovasc Surg 1996;111:36-44.
10. Boeckxstaens CJ, Flameng WJ. Retrograde perfusion does not perfuse the brain in non-human primates. Ann Thorac Surg 1995;60:319-28.
11. Filgueiras CL, Ryner L, Ye J, et al. Cerebral protection during moderate hypothermic circulatory arrest: histopathology and magnetic resonance spectroscopy of brain energetics and intracellular pH in pigs. J Thorac Cardiovasc Surg 1996;112:1073-80.
12. Sakurada T, Kazui T, Tanaka H, Komatsu S. Comparative experimental study of cerebral protection during aortic arch reconstruction. Ann Thorac Surg 1996;61:1348-54.
13. Yoshimura N, Okada M, Ota T, Nohara H. Pharmacologic intervention for ischemic brain edema after retrograde cerebral perfusion. J Thorac Cardiovasc Surg 1995;109:1173-81.
14. Levy WJ, Levin SK, Bavaria JE. Cerebral oxygenation during retrograde perfusion. Ann Thorac Surg 1995;60:184-6.
15. Lin PJ, Chang CH, Tan PP, et al. Prolonged circulatory arrest in moderate hypothermia with retrograde cerebral perfusion: is the brain ischemic? Circulation 1996;94(Suppl II):II-169-72.
16. Cheung AT, Bavaria JE, Weiss SJ, et al. Neurophysiologic effects of retrograde cerebral perfusion used for aortic reconstruction. J Cardiothorac Vasc Anesth 1998;12:252-9.
17. Filgueiras CL, Winsborrow B, Ye J, et al. A 31-P-magnetic resonance study of antegrade and retrograde cerebral perfusion during aortic arch surgery in pigs. J Thorac Cardiovasc Surg 1995;110:55-62.
18. Pagano D, Boivin CM, Faroqui MH, Bonser RS. Retrograde perfusion through the superior vena cava perfuses the brain in human beings. J Thorac Cardiovasc Surg 1996;111:270-2.
19. Ganzel BL, Edmonds HL, Pank JR, Goldsmith LJ. Neurophysiologic monitoring to assure delivery of retrograde cerebral perfusion. J Thorac Cardiovasc Surg 1997;113:748-57.
20. Wise RJS, Bernardi S, Frackowiak RSJ, et al. Serial observations on the pathophysiology of acute stroke: the transition from ischaemia to infarction as reflected in regional oxygen extraction. Brain 1983;106:197-222.
21. Powers WJ, Grubb RL, Raichle ME. Physiological responses to focal cerebral ischemia in humans. Ann Neurol 1984;16:546-52.
22. Dominguez-Roldan JM, Murillo-Cabezas F, Santamaria-Mifsut JL, et al. Cerebral oxygen extraction ratio: a useful determination for the diagnosis of brain death. Transplant Proc 1995;27:2393-4.
