- Email: [email protected]
The effect of temperature on cerebral metabolism and blood flow in adults during cardiopulmonary bypass
The effect of temperature on cerebral metabolism and blood flow in adults during cardiopulmonary bypass The effect of temperature on cerebral blood flow and metabolism was studied in 41 adult patients scheduled for operations requiring cardiopulmonary bypass. Plasma levels of midazolam and fentanyl were kept constant by a pharmacokinetic model-driven infusion system. Cerebral blood flow was measured by xenon 133 clearance (initial slope index) methods. Cerebral blood flow determinations were made at 27° C (hypothermia) and 37° C (normothermia) at constant cardiopulmonary bypass pump flows of 2 L/min/m2 • Blood gas management was conducted to maintain arterial carbon dioxide tension (not corrected for temperature) 35 to 40 mm Hg and arterial oxygen tension of 150 to 250 mm Hg. Blood gas samples were taken from the radial artery and the jugular bulb. With decreased temperature there was a significant (p < 0.0001) decrease in the arterial venous-oxygen content difference, suggesting brain flow in excess of metabolic need. For each patient, the cerebral metabolic rate of oxygen consumption at 37° C and 27° C was calculated from the two measured points at normothermia and hypothermia with the use of a linear relationship between the logarithm of cerebral metabolic rate of oxygen consumption and temperature. The temperature coefficient was then computed as the ratio of cerebral metabolic rate of oxygen consumption at 37° C to that at 27° C. The median temperature coefficient for man on nonpulsatile cardiopulmonary bypass is 2.8. Thus reducing the temperature from 37° to 27° C reduces cerebral metabolic rate of oxygen consumption by 64 %. (J THORAC CARDIOVASC SURG 1992;103:549-54)
N. Croughwell, CRNA,b L. R. Smith, PhD,a Timothy Quill, MD,b Mark Newman, MD,b William Greeley, MD,b, c Frank Kern, MD,b, c Joe Lu, MD,b and J. G. Reves, MD,b Durham, N. C.
Hypothermia is an integral part of cardiopulmonary bypass (CPB) management; yet despite the widespread use of CPB the effect of temperature on average cerebral metabolic rate of oxygen consumption (CMR02) in human beings has not been well defined. The relationship of temperature to metabolism can be characterized by the temperature coefficient (QIO), which is the ratio of metabolic rates at two temperatures separated by 10° C. Knowledge of the QIO in human beings is obviously From the Departments of Community and Family Medicine," Anesthesiology,"and Pediatrics," The Heart Center at Duke University Medical Center, Durham, N.C. Received for publication July 6, 1990. Accepted for publication Feb. 4, 199 I. Address for reprints: J. G. Reves, MD, Box 3094, Duke University Medical Center, Durham, NC 27710. 12/1/28465
important if cerebral protection is accounted for by metabolic suppression. We have examined the cerebral metabolic and blood flow effects of hypothermia and calculated the QIO during general anesthesia in patients undergoing CPB.
Methods Population and anesthetic management. After institutional reviewboard approvaland informedpatient consent had been obtained, 41 patients electively scheduledfor cardiac operations with nonpulsatile CPR were entered into this study. Patients with a history of diabetes, hypertension,stroke, or other neurologic diseases were excluded from the study. All patients were premedicated with diazepam (0.1 mg/kg) and methadone (0.1 rug/kg) orally 90 minutes before induction of anesthesia. Catheters were placed in the radial artery and retrogradely into the right jugular bulb via the right internal jugular vein before inductionofanesthesia.Anesthesiawasinducedand maintained with midazolam and fentanyl administered by computer-assisted continuousinfusion': 2 based on a pharmacokinetic model of
549
The .Journal ot Thoracic and Cardiovascular Surgery
5 5 0 Crough well et al.
10r;:::::::::;:;==::;::::=;==;--------i
..........
o •
Hypothermia Normothermia
i::
oa: ';'E".
()
65 ± 10
Age Male Female Crossdamp time (min) CPB time (min) Time between measurements (min) Operations Coronary artery bypass grafting
N
:E
Table I. Demographic variables
g
,...
0
I
26
15 51 ± 27 132 ± 74 52 ± 22 31
xl o 26
28
30
32
34
36
38
40
Temperature
ee)
Fig. 1. CMROz measurements expressed logarithmically at stablehypothermia(27 ± 2.0) and normothermia (37° ± 0.7° C). Slope = 0.1131 ± 0.33 and intercept = -3.83 ± 1.3. There is a significant difference in CMROz between the two temperatures. Both slope and intercept are significant at the p < 0.0001 level.
the drugs being infused. During CPB, midazolam was infused continuously to maintain a level of 75 ng/rnl and fentanyl to maintain a predicted level of 6 ng/rnl throughout the time of cerebral blood flow (CBF) measurements. Vecuronium was given as needed to maintain complete neuromuscular blockade. CPB management. During CPB, extracorporeal circulation was maintained by means of nonpulsatile pump flow with a Sarns 7000 MDX pump (Sarns Inc./3M, Ann Arbor, Mich.) and Cobe CML membrane oxygenator (Cobe Laboratories, Inc., Lakewood, Colo.). An asanguineous prime was used and hematocrit value maintained at 20% or more. Pump flow was controlled at 2 L/min/m z during all CBF determinations by xenon 133 clearance. Carbon dioxide tension in arterial blood (Paco-) was maintained at 35 to 40 mm Hg (uncorrected for body temperature-alpha-stat method) and oxygen tension in arterial blood (Pao-) at 150 to 250 mm Hg. Arterial blood pressure was maintained between 50 and 80 mm Hg. Sodium nitroprusside and phenylephrine, which have no effect on CBF,3.4 were used to control arterial blood pressure. CBF methodology. CBF was measured by xenon 133 clearance methods.>? Xenon 133 (1.5 to 3 mCi dissolved in 2 ml sterile saline) was injected into an injection port of the arterial perfusate circuit of the pump-oxygenator. Two extracranial, cadmium telluride detectors with wide-angle collimators were placed over the right and left temporal lobes, and the average of the two values was used to determine CBF. CBF was calculated from data by a modification of the initial slope index methods described by Olesen, Paulson, and Lassen.' CBF was calculated by the following formula: CBF (rnl/ lOO gm/rnin) = (-) slope· A . 100 where slope is the slopeofthe natural logarithm of the count rate during the first I minute of the xenon 133 washout curve obtained by linear regression and A is the temperature- and hematocrit-corrected blood-brain partition coefficient." Baseline xenon 133 was measured before the normothermic mea-
2 2
x2 x3 x4 x5
18 7 2
AVR MVR CABG and AICD Aortic valvuloplasty, ascending aortic aneurysm repair, CABG xz AVR. CABG AVR, MVR, tricuspid valvuloplasty
3 3 1 I
xz
Mean values ± standard deviation. CPB, Cardiopulmonary bypass; AYR, aortic valve replacement; MYR, mitral valve replacement; CABG, coronary artery bypass grafting; AICD, automatic implantable cardioverter-defibrillator.
surement, and counts were subtracted from the subsequent CBF values to correct for residual xenon 133. The first of two CBF and metabolism (CMROz) measurements during CPB was made after aortic crossclamping when the nasopharyngeal temperature had been at stable hypothermia (nasopharyngeal temperature of 27° C) for 10 minutes. The second measurement was made at normothermia (nasopharyngeal temperature 2:36° C.) Cerebral metabolism and oxygen extraction calculated, Blood was sampled from the radial arterial and jugular bulb catheters for determinations of pH, Pao-, Paco-, oxygen saturation, and hematocrit values I minute after injection of xenon 133. From these values and CBF, CMROz and oxygen extraction were calculated by the following formulas:
CMROz (ml/100 grn/rnin)
Oxygen extraction
CBF (ml/lOO gm/rnin) . 1.39 . Hb ([Saoz - Svoz] + 0.003 [Paoz - Pvo z]) 100
1.39 . Hb . (Sao, - Svo-) 0.003 (Paoz - Pvoz) 1.39 . Hb . (Saoz) + 0.003(Paoz)
+ X 100
where Sao, and Svoz are the arterial and venous oxygen saturation of the radial arterial and jugular bulb venous blood, respectively, Hb is hemoglobin, and Pvo- is venous oxygen tension. Statistical analysis. For each patient the CMROz at 37° C and 27° C was calculated from the two measured points at normothermia and hypothermia by means of a linear relationship between the logarithm of CMRO z and temperature. The QIO for each patient was then computed as the ratio of the CMRO z at 37° C to the CMROz at 27° C. All variables are presented
Volume 103 Number 3 March 1992
Cerebral metabolism and blood flow during CPB
551
Table II. Physiologic variables Normothermia
Hypothermic Temperature (0C) CBF ml/IOO gm/rnin A-Vo 2 (mm Hg) CMR0 2 (ml/IOO gm/rnin) % Oxygen extraction CBF: CM R0 2
27 21 2.7 0.5 27 46
± 2.0 ± 6.8 ± 1.05 ± 0.2 ± 8.9 ± 26.0
37 35 4.3 1.4 42 24
± 0.7 ± 9.0 ± 1.03 ± 0.34 ± 9.3 ± 6.3
p
0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
CBF, Cerebral blood flow: A-Vo" arterial-venous oxygen difference: CMRO" cerebral metabolic rate of oxygen consumption.
as mean ± standard deviations. The Wilcoxon rank sum test was used to test for differences in variables between the two measurements, and backward stepwise linear regression was used to assessthe effecton CBF owingto time between the two measurements, and the physiologic variables, hematocrit, flow, mean arterial pressure, and hemoglobin. Results The mean demographic data for the 41 patients, including age, gender, operation, duration of crossclamping, duration of CPB, and time between CBF measurements, are displayed in Table I. The physiologic and derived variables obtained at hypothermia and normothermia are presented in Table II. Hypothermic observations were made at a nasopharyngeal temperature of 27° ± 2.0° C and normothermia at 37° ± 0.7° C. At hypothermia the CBF was significantly (p < 0.0001) lower than normothermia (21 ± 6.8 versus 35 ± 9.0, respectively). The CMR02 differed significantly between conditions; at hypothermia it was 0.5 ± 0.20 and at normothermia, 1.4 ± 0.34 (p:s 0.0001). Arterial-venous oxygen content differences were significantly (p:s 0.0001) different at hypothermia (2.7 ± 1.06)than at normothermia (4.3 ± 1.03). The percent oxygen extraction differed significantly (p :s 0.0001)-27 ± 8.9 during hypothermia and 42 ± 9.3 at normothermia. The ratio of CBF to CMR02, which is normally 16 ± 25 in the awake state in patients.l? was significantly different (p:s 0.0001) at each condition-46 ± 26.0 at hypothermia and 24 ± 6.3 at normothermia. The relationship of temperature to CMR0 2 is depicted in Fig. 1, with temperature on the X axis and the log of CMR02 on the Y axis. This exponential plot of the data can be fit to the equation CMR02 = 0.021e .
1147· T
where T is the nasopharyngeal temperature in degrees centigrade. The relationship of temperature and CMR02 was further analyzed to calculate the QIO for these 41 patients. Fig. 2 contains a plot of the individual and quartile values of the QIO. The median QIO value was 2.8.
The time difference between measurements was 51 ± 22 minutes. Changes in hematocrit, (0.58 ± 3.3), hemoglobin (0.28 ± 0.98), pump flow (0.02 ± 0.10), and mean arterial pressure values (-1.6 ± 11.5) were not significant. These variables, including time between measurements, did not influence the change in CBF, whereas the effect of temperature was significant (p < 0.001). Discussion The principal finding in this study is the direct effect of temperature on cerebral metabolism and CBF in adult patients during CPB. A convenient form of expression for the effect of temperature on oxidative metabolism is the ratio ofCMR02 separated by 10° C (QIO). Wefound the median QIO to be 2.8 in this adult cardiac surgical population. Our methods involved the use of xenon 133 washout as a measure of CBF and radial arterial-jugular bulb blood gas difference to calculate oxygen uptake and consumption. Xenon 133 washout has been used to measure CBF during CPB by several investigative groups.f- 7 We have recently validated this method under the conditions of nonpulsatile CPB in the canine laboratory. CBF by xenon 133 clearance was compared with CBF determined by the Kety-Schmidt method with xenon 133 as the tracer, and the correlation coefficient was r = 0.983 (p :s 0.0001) in the 12 animals examined. Immediately after the start of CPB, temperature was maintained at 32° C, as is customary in our institution. A stable hypothermic state was then achieved after aortic crossclamping. Measurements at normothermia were made before CPB was discontinued. The flow was nonpulsatile. At normothermia the heart was kept empty to assure no ejection. We have not used a time correction for xenon 133 measurements as advocated by Rogers and coworkers.' because in this study, as in our larger, previous report.l we have been unable to detect a significant effect of time on CBF measurement. The calculation of CMR02 with arterial-jugular bulb oxygen concentration differences used to calculate oxygen uptake and consumption data is valid over the
The Journal of
5 5 2 Croughwell et al.
Thoracic and Cardiovascular Surgery
Quartiles Min
25% 50%
75%
Max
-ITJr--------
o
1
2
3
4
5
6
7
8
Fig. 2. Box-and-whisker plotillustrates the quartile Q 10 measurements. Each linerepresents a calculated QIO from an individual patient. Fifty percentof the data pointslie between 2.4 (lower quartile) and 3.4 (higherquartile);this data range is represented by the large rectangle. The median QIO on nonpulsatile cardiopulmonary bypass is 2.8.
2.5 Tr==::;::~===~---------' o Hypothermia • Normothermia ..... 2.0 ~
s:::
0""
.~
1.5
o :5 ,....
==
Cl
1.0
IX ';'
E 0.5 ~
°
0.0 +------.---.------.---.--.----.-,....---,---.------.--1 18 20 22 24 26 28 30 32 34 36 38 40
Temperature (0C)
Fig. 3. CMRO z is expressed in relation to temperaturebased on the equationIn(CMROz) = a +b*Twhere In (.) is the natural logarithmfunction, T is the temperature in degrees centigrade, a is the intercept, and b is the slope.
temperatures of this study1I and has been used in patients during CPB,?' 12 It has long been established that temperature directly affects cerebral metabolism with experiments conducted in vitro l 3- 15 and in vivo.16-19 These studies do not necessarily apply to clinical conditions because of species differences and the experimental conditions that depart significantly from standard CPB practice. Other investigations in a variety of species and experimental settings have reported QIO values in intact animal and human experiments. Reported QIO varies from 2.2 to 3.6 (Table III). Woodcock and coworkers'? measured CBF and CMR02 during CPB in 12 patients undergoing coronary artery bypass operations. These patients were anesthetized with fentanyl. The investigators found that CMR02wasOAl at 27° Cand 1.16at37° C. From these
data we have estimated that this produces a QIO of approximately 2.8, which is identical to the value we observed in the 41 patients anesthetized with midazolam and fentanyl. Greeley and coauthors/" have recently observed that children between the ages of 1 day and 12 years have a median QIO of 3.6 over the range of 18° to 37° C. This difference could reflect size-related changes in complete brain cooling and hence effect of cooling on cerebral metabolism. Hypothermia reduces CMR02 to a greater extent than CBF, resulting in an apparent relative luxuriant brain blood flow. At pump flow rates of 2 L/min/m 2, less oxygen extraction occurs and arterial-venous oxygen content difference narrows during moderate hypothermia. If CBF were reduced to a greater extent, one would anticipate increased oxygen extraction and widening of the arterialvenous oxygen content difference. This physiologic compensatory mechanism of assuring adequate tissue oxygen delivery has been shown in two laboratory investigations. During profound (20° C) hypothermia when pump flow was reduced to 0.5 L/min/m 2 in monkeys" and to 30 ml/kg/rnin in dogs,22 the oxygen extraction increased to maintain a constant CMR02. In the present study during moderate hypothermia, oxygen extraction decreased and there was relative hyperperfusion represented by the greater ratio of CBF/CMR02 than seen at normothermia. With moderate hypothermia there is a decrease in CBF that is consistent with preservation of CBF autoregulation, but there is a poorer association of CBF and CMR02 at hypothermia. CBF tends to be higher relative to CMR02 during hypothermia. The relationship between CMR02 and temperature has been the subject of some dispute. In our study, two observations, one at hypothermia and one at normothermia, were made on each patient. These data do not permit the determination of the exact nature of the CMR02-
Volume 103 Number 3
Cerebral metabolism and blood flow during CPR
March 1992
553
Table III. QlO values in other investigations Reference No.
18 17 24 20 12 Present
Species
Temperature range (0C)
26-35 19-37 28-38 37-18 37-27 37-27
Dog Monkey Dog Infants, children Man Man
QIO
CBF method
Anesthetic
3.0 3.5 2.2 3.6 2.8* 2.8
Arterial isolation Kety-Schmidt, N 20 Sagittal sinus flow 133Xe clearance 133Xe clearance 133Xe clearance
Pentobarbital Pentobarbital Halothane Fentanyl Fentanyl Midazolam, fentanyl
Brain
lJJXe. Xenon 133. 'Estimated.
temperature relationship over the 10° C temperature range of customary adult moderate hypothermia. Three models, however, can be proposed: CMROz = a + b*T 1n(CMROz) = a + b*T 1n(CMROz) = a + b/(273
(1) (2) (3)
+ T)
where In (.) is the natural logarithm function, T is the temperature in degrees centigrade, a is the intercept, and be is the slope. Equation 1 is purely a linear relationship between CMROz and temperature. Equation 2, based on work by Michenfelder and Theye.l? is a linear relationship of log (CMROz) and temperature. This can be rewritten as CMROz = a' e(b*Tl. Equation 3, based on work by Bering.!? is a linear relationship between In(CMROz) and the inverse of temperature in kelvins. This can be rewritten as CMROz = a' e(b/(Z73 + Tl.Equation 1 is not a viable candidate to model the data because for our data CMROz = -2.23 + 0.10*T would predict a CMROz= 0 when T = 22° C. Clearly this relationship can be dismissed on purely physiologic grounds. The choice is then between equations 2 and 3. We have chosen equation 2, based on work by Greeley and coworkers.P Their study found that the slope of the In(CMROz)-temperature line was 0.1171 for patients cooled from 37° C to 28 ° C, and the slope was 0.1170 for patients cooled from 37° C to 18° C. In our data the slope is 0.1131. The similarity among the slopes indicates the log-linear relationship holds over the range of temperatures used in cardiac operations (18° to 37° C). This relationship does not hold for equation 1 or 3. Both equations 1 and 3 would yield different slopes over the two temperature intervals. Use of equation 2 results in a plot of our data shown in Fig. 3. One important implication of equation 2 is that the QIO will be constant over the entire temperature range, independent of which two temperatureswereused (as long as they were separated by 10° C). QIO can be calculated as exp(b*(T z - T 1 Both equations 1 and 3 would yield a different QIO if measured at different points on the temperature curve. Our data and
».
analyses do not conclusively indicate which model is correct. A careful experiment in man needs to be undertaken where CMROz is measured at small temperature increments (both during cooling and warming) in a number of patients under similar anesthetic conditions to be certain of the exact shape of the temperature-CMROz curve. Equation 2 can be used to predict the CMROz at any given temperature in the range of moderate hypothermia in our study. From the equation and the QIO, predictions regarding safe periods of anoxia may be made. Assuming the QIO is 2.8 for adults, and if normothermic brain (37° C) can withstand up to 6 minutes of circulatory arrest,Z3 then during CPB at 27° C, 17 minutes of total circulatory arrest can be tolerated. This explains the central nervous system's toleration of infrequent and usually brief, but sometimes necessary, cessations of CPB to complete surgical repair of the heart or great vessels, or both, when moderate hypothermia is used. Of note is the fact that cerebral protection is afforded by at least two confounding variables, temperature and general anesthesia. Of these, hypothermia produces greater protection than anesthesia even when both produce the same decrease in CMROz. Michenfelder and Theye'" have shown in dogs that it is temperature that provides the most significant cerebral protection from anoxia. They did this by reducing the CMROz equally in animals with temperature at 30° C and with high-dose thiopental (46 mg/kg) at normothermia. Each condition reduced CMROz the same amount from the control level, but hypothermia significantly enhanced cerebral protection as measured by adenosine triphosphate loss and lactate production 4 minutes after decapitation. Thus, despite the additive effects to hypothermia of general anesthesia on CMROz reduction, IZ, Z4 it is primarily the temperature effect that confers cerebral protection. In summary.. we have demonstrated that moderate hypothermia has significant effects on CBF and metabolism. The relationship of temperature to cerebral metabolism can be used to postulate cerebral protection from
5 5 4 Croughwell et al.
ischemic insult, although data supporting this hypothesis in adult surgical patients have not been provided. We are grateful to the cardiac surgeons atDuke University, whose support and interest in the conduct of this investigation made this work possible. We are also grateful to Laraine Tuck for her handling of the manuscript throughout its various stages. REFERENCES I. Jacobs JR. Algorithm for optimal linear model-based control with application to pharmacokinetic model-driven drug delivery. IEEE Trans Biomed Eng 1990;37:107-9. 2. Reves JG, Glass P, Jacobs JR. Aifentanil and midazolam: new anesthetic drugs for continuous infusion and an automated method of administration. Mount Sinai J Med 1989;56:99-107. 3. Rogers AT, Stump DA, Gravlee GP, et aI. Response of cerebral blood flow to phenylephrine infusion during hypothermic cardiopulmonary bypass: influence of Pac02 management. Anesthesiology 1988;69:547-51. 4. Rogers AT, Prough OS, Stump DA, et aI. Cerebral blood flow does not change following sodium nitroprusside infusion during hypothermic cardiopulmonary bypass. Anesth Analg 1989;68:122-6. 5. Govier AV, Reves JG, McKay RD, et aI. Factors and their influence on regional cerebral blood flow during nonpulsatile cardiopulmonary bypass. Ann Thorac Surg 1984; 38:592-600. 6. Prough OS, Stump DA, Roy RD, et aI. Response of cerebral blood flow to changes in carbon dioxide and tension during hypothermic cardiopulmonary bypass. Anesthesiology 1986;64:576-81. 7. Murkin JM, Farrar JK, Tweed WA, McKenzie FN, Guiraudon G. Cerebral autoregulation and flow/metabolismcoupling during cardiopulmonary bypass: the influence of Paco-, Anesth Analg 1987;66:825-32. 8. Olesen J, Paulson OB, Lassen NA. Regional cerebral blood flowin man determined by the initial slope of the clearance of intraarterially injected 133Xe. Stroke 1971;2:519-40. 9. Chen RYZ, Foun-Chung F, Syngcuk K, Kung-Ming J, Scunichi U, Shu C. Tissue-blood partition coefficient for xenon: temperature and hematocrit dependence. J App1 Physiol 1980;49:I78-83. 10. Michenfelder JD. Anesthesia and the brain. New York: Churchill Livingstone, 1988. II. Steen PA, Newberg L, Milde JH, Michenfelder JD. Hypothermia and barbiturates: individual and combined effects on canine cerebral oxygen consumption. Anesthesiology 1983;58:527-32.
The Journal of Thoracic and Cardiovascular Surgery
12. Woodcock TE, Murkin JM, Farrar JK, Tweed WA, Guiraudon GM, McKenzie FN. Pharmacologic EEG suppression during cardiopulmonary bypass: cerebral hemodynamic and metabolic effects of thiopental or isoflurane during hypothermia and normothermia. Anesthesiology 1987;67:218-24. 13. Field J II, Fuhrman FA, Martin AW. Effect of temperature on the oxygen consumption of brain tissue. J Neurophysiol 1944;7:117-26. 14. Norwood WI, Norwood CR, Ingwall JS, Castaneda AR, Fosse1 ET. Hypothermic circulatory arrest. J THORAC CARDIOVASC SURG 1979;78:823-30. 15. Stocker F, Herschkowitz N, Bossi E, et aI. Cerebral metabolic studies in situ by 31P-nuclear magnetic resonance after hypothermic circulatory arrest. Pediatr Res 1986; 20:867-71. 16. Wollman H, Stephen GW, Clement AJ, Danielson GK. Cerebral blood flow in man during extracorporeal circulation. J THORAC CARDIOVASC SURG 1966;52:558-64. 17. Bering EA Jr. Effect of body temperature change on cerebral oxygen consumption of the intact monkey. Am J PhysioI1961;200:417-9. 18. Rosomoff HL, Holaday DA. Cerebral blood flow and cerebral oxygen consumption during hypothermia. Am J Physiol 1954;179:85-8. 19. Michenfelder JD, Theye RA. Hypothermia: effect on canine brain and whole-body metabolism. Anesthesiology 1968;29:1107-12. 20. Greeley WJ, Kern FH, Ungerieider RM, et aI. The effect of hypothermic cardiopulmonary bypass and total circulatory arrest on cerebral metabolism in neonates, infants, and children. J THORAC CARDIOVASC SURG 1991;101:783-94. 21. Fox LS, Blackstone EH, Kirklin JW, Bishop SP, Bergdahl LAL, Bradley EL. Relationship of brain blood flow and oxygen consumption to perfusion flow rate during profoundly hypothermic cardiopulmonary bypass. J THORAC CARDIOVASC SURG 1984;87:658-64. 22. Miyamoto K, Kawashima Y, Matsuda H, Okuda A, Maeda S, Hirose H. Optimal perfusion flow rate for the brain during deep hypothermic cardiopulmonary bypass at 20° C. J THORAC CARDIOVASC SURG 1986;92:I065-70. 23. Weinberger LM, Gibbon MH, Gibbon JH Jr. Temporary arrest of the circulation to the central nervous system. Arch Neurol Psychiatry 1940;43:615-34. 24. Michenfelder JD, Theye RA. The effects of anesthesia and hypothermia on canine cerebral ATP and lactate during anoxia produced by decapitation. Anesthesiology 1970; 33:430-9.
N. Croughwell, CRNA,b L. R. Smith, PhD,a Timothy Quill, MD,b Mark Newman, MD,b William Greeley, MD,b, c Frank Kern, MD,b, c Joe Lu, MD,b and J. G. Reves, MD,b Durham, N. C.
Hypothermia is an integral part of cardiopulmonary bypass (CPB) management; yet despite the widespread use of CPB the effect of temperature on average cerebral metabolic rate of oxygen consumption (CMR02) in human beings has not been well defined. The relationship of temperature to metabolism can be characterized by the temperature coefficient (QIO), which is the ratio of metabolic rates at two temperatures separated by 10° C. Knowledge of the QIO in human beings is obviously From the Departments of Community and Family Medicine," Anesthesiology,"and Pediatrics," The Heart Center at Duke University Medical Center, Durham, N.C. Received for publication July 6, 1990. Accepted for publication Feb. 4, 199 I. Address for reprints: J. G. Reves, MD, Box 3094, Duke University Medical Center, Durham, NC 27710. 12/1/28465
important if cerebral protection is accounted for by metabolic suppression. We have examined the cerebral metabolic and blood flow effects of hypothermia and calculated the QIO during general anesthesia in patients undergoing CPB.
Methods Population and anesthetic management. After institutional reviewboard approvaland informedpatient consent had been obtained, 41 patients electively scheduledfor cardiac operations with nonpulsatile CPR were entered into this study. Patients with a history of diabetes, hypertension,stroke, or other neurologic diseases were excluded from the study. All patients were premedicated with diazepam (0.1 mg/kg) and methadone (0.1 rug/kg) orally 90 minutes before induction of anesthesia. Catheters were placed in the radial artery and retrogradely into the right jugular bulb via the right internal jugular vein before inductionofanesthesia.Anesthesiawasinducedand maintained with midazolam and fentanyl administered by computer-assisted continuousinfusion': 2 based on a pharmacokinetic model of
549
The .Journal ot Thoracic and Cardiovascular Surgery
5 5 0 Crough well et al.
10r;:::::::::;:;==::;::::=;==;--------i
..........
o •
Hypothermia Normothermia
i::
oa: ';'E".
()
65 ± 10
Age Male Female Crossdamp time (min) CPB time (min) Time between measurements (min) Operations Coronary artery bypass grafting
N
:E
Table I. Demographic variables
g
,...
0
I
26
15 51 ± 27 132 ± 74 52 ± 22 31
xl o 26
28
30
32
34
36
38
40
Temperature
ee)
Fig. 1. CMROz measurements expressed logarithmically at stablehypothermia(27 ± 2.0) and normothermia (37° ± 0.7° C). Slope = 0.1131 ± 0.33 and intercept = -3.83 ± 1.3. There is a significant difference in CMROz between the two temperatures. Both slope and intercept are significant at the p < 0.0001 level.
the drugs being infused. During CPB, midazolam was infused continuously to maintain a level of 75 ng/rnl and fentanyl to maintain a predicted level of 6 ng/rnl throughout the time of cerebral blood flow (CBF) measurements. Vecuronium was given as needed to maintain complete neuromuscular blockade. CPB management. During CPB, extracorporeal circulation was maintained by means of nonpulsatile pump flow with a Sarns 7000 MDX pump (Sarns Inc./3M, Ann Arbor, Mich.) and Cobe CML membrane oxygenator (Cobe Laboratories, Inc., Lakewood, Colo.). An asanguineous prime was used and hematocrit value maintained at 20% or more. Pump flow was controlled at 2 L/min/m z during all CBF determinations by xenon 133 clearance. Carbon dioxide tension in arterial blood (Paco-) was maintained at 35 to 40 mm Hg (uncorrected for body temperature-alpha-stat method) and oxygen tension in arterial blood (Pao-) at 150 to 250 mm Hg. Arterial blood pressure was maintained between 50 and 80 mm Hg. Sodium nitroprusside and phenylephrine, which have no effect on CBF,3.4 were used to control arterial blood pressure. CBF methodology. CBF was measured by xenon 133 clearance methods.>? Xenon 133 (1.5 to 3 mCi dissolved in 2 ml sterile saline) was injected into an injection port of the arterial perfusate circuit of the pump-oxygenator. Two extracranial, cadmium telluride detectors with wide-angle collimators were placed over the right and left temporal lobes, and the average of the two values was used to determine CBF. CBF was calculated from data by a modification of the initial slope index methods described by Olesen, Paulson, and Lassen.' CBF was calculated by the following formula: CBF (rnl/ lOO gm/rnin) = (-) slope· A . 100 where slope is the slopeofthe natural logarithm of the count rate during the first I minute of the xenon 133 washout curve obtained by linear regression and A is the temperature- and hematocrit-corrected blood-brain partition coefficient." Baseline xenon 133 was measured before the normothermic mea-
2 2
x2 x3 x4 x5
18 7 2
AVR MVR CABG and AICD Aortic valvuloplasty, ascending aortic aneurysm repair, CABG xz AVR. CABG AVR, MVR, tricuspid valvuloplasty
3 3 1 I
xz
Mean values ± standard deviation. CPB, Cardiopulmonary bypass; AYR, aortic valve replacement; MYR, mitral valve replacement; CABG, coronary artery bypass grafting; AICD, automatic implantable cardioverter-defibrillator.
surement, and counts were subtracted from the subsequent CBF values to correct for residual xenon 133. The first of two CBF and metabolism (CMROz) measurements during CPB was made after aortic crossclamping when the nasopharyngeal temperature had been at stable hypothermia (nasopharyngeal temperature of 27° C) for 10 minutes. The second measurement was made at normothermia (nasopharyngeal temperature 2:36° C.) Cerebral metabolism and oxygen extraction calculated, Blood was sampled from the radial arterial and jugular bulb catheters for determinations of pH, Pao-, Paco-, oxygen saturation, and hematocrit values I minute after injection of xenon 133. From these values and CBF, CMROz and oxygen extraction were calculated by the following formulas:
CMROz (ml/100 grn/rnin)
Oxygen extraction
CBF (ml/lOO gm/rnin) . 1.39 . Hb ([Saoz - Svoz] + 0.003 [Paoz - Pvo z]) 100
1.39 . Hb . (Sao, - Svo-) 0.003 (Paoz - Pvoz) 1.39 . Hb . (Saoz) + 0.003(Paoz)
+ X 100
where Sao, and Svoz are the arterial and venous oxygen saturation of the radial arterial and jugular bulb venous blood, respectively, Hb is hemoglobin, and Pvo- is venous oxygen tension. Statistical analysis. For each patient the CMROz at 37° C and 27° C was calculated from the two measured points at normothermia and hypothermia by means of a linear relationship between the logarithm of CMRO z and temperature. The QIO for each patient was then computed as the ratio of the CMRO z at 37° C to the CMROz at 27° C. All variables are presented
Volume 103 Number 3 March 1992
Cerebral metabolism and blood flow during CPB
551
Table II. Physiologic variables Normothermia
Hypothermic Temperature (0C) CBF ml/IOO gm/rnin A-Vo 2 (mm Hg) CMR0 2 (ml/IOO gm/rnin) % Oxygen extraction CBF: CM R0 2
27 21 2.7 0.5 27 46
± 2.0 ± 6.8 ± 1.05 ± 0.2 ± 8.9 ± 26.0
37 35 4.3 1.4 42 24
± 0.7 ± 9.0 ± 1.03 ± 0.34 ± 9.3 ± 6.3
p
0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
CBF, Cerebral blood flow: A-Vo" arterial-venous oxygen difference: CMRO" cerebral metabolic rate of oxygen consumption.
as mean ± standard deviations. The Wilcoxon rank sum test was used to test for differences in variables between the two measurements, and backward stepwise linear regression was used to assessthe effecton CBF owingto time between the two measurements, and the physiologic variables, hematocrit, flow, mean arterial pressure, and hemoglobin. Results The mean demographic data for the 41 patients, including age, gender, operation, duration of crossclamping, duration of CPB, and time between CBF measurements, are displayed in Table I. The physiologic and derived variables obtained at hypothermia and normothermia are presented in Table II. Hypothermic observations were made at a nasopharyngeal temperature of 27° ± 2.0° C and normothermia at 37° ± 0.7° C. At hypothermia the CBF was significantly (p < 0.0001) lower than normothermia (21 ± 6.8 versus 35 ± 9.0, respectively). The CMR02 differed significantly between conditions; at hypothermia it was 0.5 ± 0.20 and at normothermia, 1.4 ± 0.34 (p:s 0.0001). Arterial-venous oxygen content differences were significantly (p:s 0.0001) different at hypothermia (2.7 ± 1.06)than at normothermia (4.3 ± 1.03). The percent oxygen extraction differed significantly (p :s 0.0001)-27 ± 8.9 during hypothermia and 42 ± 9.3 at normothermia. The ratio of CBF to CMR02, which is normally 16 ± 25 in the awake state in patients.l? was significantly different (p:s 0.0001) at each condition-46 ± 26.0 at hypothermia and 24 ± 6.3 at normothermia. The relationship of temperature to CMR0 2 is depicted in Fig. 1, with temperature on the X axis and the log of CMR02 on the Y axis. This exponential plot of the data can be fit to the equation CMR02 = 0.021e .
1147· T
where T is the nasopharyngeal temperature in degrees centigrade. The relationship of temperature and CMR02 was further analyzed to calculate the QIO for these 41 patients. Fig. 2 contains a plot of the individual and quartile values of the QIO. The median QIO value was 2.8.
The time difference between measurements was 51 ± 22 minutes. Changes in hematocrit, (0.58 ± 3.3), hemoglobin (0.28 ± 0.98), pump flow (0.02 ± 0.10), and mean arterial pressure values (-1.6 ± 11.5) were not significant. These variables, including time between measurements, did not influence the change in CBF, whereas the effect of temperature was significant (p < 0.001). Discussion The principal finding in this study is the direct effect of temperature on cerebral metabolism and CBF in adult patients during CPB. A convenient form of expression for the effect of temperature on oxidative metabolism is the ratio ofCMR02 separated by 10° C (QIO). Wefound the median QIO to be 2.8 in this adult cardiac surgical population. Our methods involved the use of xenon 133 washout as a measure of CBF and radial arterial-jugular bulb blood gas difference to calculate oxygen uptake and consumption. Xenon 133 washout has been used to measure CBF during CPB by several investigative groups.f- 7 We have recently validated this method under the conditions of nonpulsatile CPB in the canine laboratory. CBF by xenon 133 clearance was compared with CBF determined by the Kety-Schmidt method with xenon 133 as the tracer, and the correlation coefficient was r = 0.983 (p :s 0.0001) in the 12 animals examined. Immediately after the start of CPB, temperature was maintained at 32° C, as is customary in our institution. A stable hypothermic state was then achieved after aortic crossclamping. Measurements at normothermia were made before CPB was discontinued. The flow was nonpulsatile. At normothermia the heart was kept empty to assure no ejection. We have not used a time correction for xenon 133 measurements as advocated by Rogers and coworkers.' because in this study, as in our larger, previous report.l we have been unable to detect a significant effect of time on CBF measurement. The calculation of CMR02 with arterial-jugular bulb oxygen concentration differences used to calculate oxygen uptake and consumption data is valid over the
The Journal of
5 5 2 Croughwell et al.
Thoracic and Cardiovascular Surgery
Quartiles Min
25% 50%
75%
Max
-ITJr--------
o
1
2
3
4
5
6
7
8
Fig. 2. Box-and-whisker plotillustrates the quartile Q 10 measurements. Each linerepresents a calculated QIO from an individual patient. Fifty percentof the data pointslie between 2.4 (lower quartile) and 3.4 (higherquartile);this data range is represented by the large rectangle. The median QIO on nonpulsatile cardiopulmonary bypass is 2.8.
2.5 Tr==::;::~===~---------' o Hypothermia • Normothermia ..... 2.0 ~
s:::
0""
.~
1.5
o :5 ,....
==
Cl
1.0
IX ';'
E 0.5 ~
°
0.0 +------.---.------.---.--.----.-,....---,---.------.--1 18 20 22 24 26 28 30 32 34 36 38 40
Temperature (0C)
Fig. 3. CMRO z is expressed in relation to temperaturebased on the equationIn(CMROz) = a +b*Twhere In (.) is the natural logarithmfunction, T is the temperature in degrees centigrade, a is the intercept, and b is the slope.
temperatures of this study1I and has been used in patients during CPB,?' 12 It has long been established that temperature directly affects cerebral metabolism with experiments conducted in vitro l 3- 15 and in vivo.16-19 These studies do not necessarily apply to clinical conditions because of species differences and the experimental conditions that depart significantly from standard CPB practice. Other investigations in a variety of species and experimental settings have reported QIO values in intact animal and human experiments. Reported QIO varies from 2.2 to 3.6 (Table III). Woodcock and coworkers'? measured CBF and CMR02 during CPB in 12 patients undergoing coronary artery bypass operations. These patients were anesthetized with fentanyl. The investigators found that CMR02wasOAl at 27° Cand 1.16at37° C. From these
data we have estimated that this produces a QIO of approximately 2.8, which is identical to the value we observed in the 41 patients anesthetized with midazolam and fentanyl. Greeley and coauthors/" have recently observed that children between the ages of 1 day and 12 years have a median QIO of 3.6 over the range of 18° to 37° C. This difference could reflect size-related changes in complete brain cooling and hence effect of cooling on cerebral metabolism. Hypothermia reduces CMR02 to a greater extent than CBF, resulting in an apparent relative luxuriant brain blood flow. At pump flow rates of 2 L/min/m 2, less oxygen extraction occurs and arterial-venous oxygen content difference narrows during moderate hypothermia. If CBF were reduced to a greater extent, one would anticipate increased oxygen extraction and widening of the arterialvenous oxygen content difference. This physiologic compensatory mechanism of assuring adequate tissue oxygen delivery has been shown in two laboratory investigations. During profound (20° C) hypothermia when pump flow was reduced to 0.5 L/min/m 2 in monkeys" and to 30 ml/kg/rnin in dogs,22 the oxygen extraction increased to maintain a constant CMR02. In the present study during moderate hypothermia, oxygen extraction decreased and there was relative hyperperfusion represented by the greater ratio of CBF/CMR02 than seen at normothermia. With moderate hypothermia there is a decrease in CBF that is consistent with preservation of CBF autoregulation, but there is a poorer association of CBF and CMR02 at hypothermia. CBF tends to be higher relative to CMR02 during hypothermia. The relationship between CMR02 and temperature has been the subject of some dispute. In our study, two observations, one at hypothermia and one at normothermia, were made on each patient. These data do not permit the determination of the exact nature of the CMR02-
Volume 103 Number 3
Cerebral metabolism and blood flow during CPR
March 1992
553
Table III. QlO values in other investigations Reference No.
18 17 24 20 12 Present
Species
Temperature range (0C)
26-35 19-37 28-38 37-18 37-27 37-27
Dog Monkey Dog Infants, children Man Man
QIO
CBF method
Anesthetic
3.0 3.5 2.2 3.6 2.8* 2.8
Arterial isolation Kety-Schmidt, N 20 Sagittal sinus flow 133Xe clearance 133Xe clearance 133Xe clearance
Pentobarbital Pentobarbital Halothane Fentanyl Fentanyl Midazolam, fentanyl
Brain
lJJXe. Xenon 133. 'Estimated.
temperature relationship over the 10° C temperature range of customary adult moderate hypothermia. Three models, however, can be proposed: CMROz = a + b*T 1n(CMROz) = a + b*T 1n(CMROz) = a + b/(273
(1) (2) (3)
+ T)
where In (.) is the natural logarithm function, T is the temperature in degrees centigrade, a is the intercept, and be is the slope. Equation 1 is purely a linear relationship between CMROz and temperature. Equation 2, based on work by Michenfelder and Theye.l? is a linear relationship of log (CMROz) and temperature. This can be rewritten as CMROz = a' e(b*Tl. Equation 3, based on work by Bering.!? is a linear relationship between In(CMROz) and the inverse of temperature in kelvins. This can be rewritten as CMROz = a' e(b/(Z73 + Tl.Equation 1 is not a viable candidate to model the data because for our data CMROz = -2.23 + 0.10*T would predict a CMROz= 0 when T = 22° C. Clearly this relationship can be dismissed on purely physiologic grounds. The choice is then between equations 2 and 3. We have chosen equation 2, based on work by Greeley and coworkers.P Their study found that the slope of the In(CMROz)-temperature line was 0.1171 for patients cooled from 37° C to 28 ° C, and the slope was 0.1170 for patients cooled from 37° C to 18° C. In our data the slope is 0.1131. The similarity among the slopes indicates the log-linear relationship holds over the range of temperatures used in cardiac operations (18° to 37° C). This relationship does not hold for equation 1 or 3. Both equations 1 and 3 would yield different slopes over the two temperature intervals. Use of equation 2 results in a plot of our data shown in Fig. 3. One important implication of equation 2 is that the QIO will be constant over the entire temperature range, independent of which two temperatureswereused (as long as they were separated by 10° C). QIO can be calculated as exp(b*(T z - T 1 Both equations 1 and 3 would yield a different QIO if measured at different points on the temperature curve. Our data and
».
analyses do not conclusively indicate which model is correct. A careful experiment in man needs to be undertaken where CMROz is measured at small temperature increments (both during cooling and warming) in a number of patients under similar anesthetic conditions to be certain of the exact shape of the temperature-CMROz curve. Equation 2 can be used to predict the CMROz at any given temperature in the range of moderate hypothermia in our study. From the equation and the QIO, predictions regarding safe periods of anoxia may be made. Assuming the QIO is 2.8 for adults, and if normothermic brain (37° C) can withstand up to 6 minutes of circulatory arrest,Z3 then during CPB at 27° C, 17 minutes of total circulatory arrest can be tolerated. This explains the central nervous system's toleration of infrequent and usually brief, but sometimes necessary, cessations of CPB to complete surgical repair of the heart or great vessels, or both, when moderate hypothermia is used. Of note is the fact that cerebral protection is afforded by at least two confounding variables, temperature and general anesthesia. Of these, hypothermia produces greater protection than anesthesia even when both produce the same decrease in CMROz. Michenfelder and Theye'" have shown in dogs that it is temperature that provides the most significant cerebral protection from anoxia. They did this by reducing the CMROz equally in animals with temperature at 30° C and with high-dose thiopental (46 mg/kg) at normothermia. Each condition reduced CMROz the same amount from the control level, but hypothermia significantly enhanced cerebral protection as measured by adenosine triphosphate loss and lactate production 4 minutes after decapitation. Thus, despite the additive effects to hypothermia of general anesthesia on CMROz reduction, IZ, Z4 it is primarily the temperature effect that confers cerebral protection. In summary.. we have demonstrated that moderate hypothermia has significant effects on CBF and metabolism. The relationship of temperature to cerebral metabolism can be used to postulate cerebral protection from
5 5 4 Croughwell et al.
ischemic insult, although data supporting this hypothesis in adult surgical patients have not been provided. We are grateful to the cardiac surgeons atDuke University, whose support and interest in the conduct of this investigation made this work possible. We are also grateful to Laraine Tuck for her handling of the manuscript throughout its various stages. REFERENCES I. Jacobs JR. Algorithm for optimal linear model-based control with application to pharmacokinetic model-driven drug delivery. IEEE Trans Biomed Eng 1990;37:107-9. 2. Reves JG, Glass P, Jacobs JR. Aifentanil and midazolam: new anesthetic drugs for continuous infusion and an automated method of administration. Mount Sinai J Med 1989;56:99-107. 3. Rogers AT, Stump DA, Gravlee GP, et aI. Response of cerebral blood flow to phenylephrine infusion during hypothermic cardiopulmonary bypass: influence of Pac02 management. Anesthesiology 1988;69:547-51. 4. Rogers AT, Prough OS, Stump DA, et aI. Cerebral blood flow does not change following sodium nitroprusside infusion during hypothermic cardiopulmonary bypass. Anesth Analg 1989;68:122-6. 5. Govier AV, Reves JG, McKay RD, et aI. Factors and their influence on regional cerebral blood flow during nonpulsatile cardiopulmonary bypass. Ann Thorac Surg 1984; 38:592-600. 6. Prough OS, Stump DA, Roy RD, et aI. Response of cerebral blood flow to changes in carbon dioxide and tension during hypothermic cardiopulmonary bypass. Anesthesiology 1986;64:576-81. 7. Murkin JM, Farrar JK, Tweed WA, McKenzie FN, Guiraudon G. Cerebral autoregulation and flow/metabolismcoupling during cardiopulmonary bypass: the influence of Paco-, Anesth Analg 1987;66:825-32. 8. Olesen J, Paulson OB, Lassen NA. Regional cerebral blood flowin man determined by the initial slope of the clearance of intraarterially injected 133Xe. Stroke 1971;2:519-40. 9. Chen RYZ, Foun-Chung F, Syngcuk K, Kung-Ming J, Scunichi U, Shu C. Tissue-blood partition coefficient for xenon: temperature and hematocrit dependence. J App1 Physiol 1980;49:I78-83. 10. Michenfelder JD. Anesthesia and the brain. New York: Churchill Livingstone, 1988. II. Steen PA, Newberg L, Milde JH, Michenfelder JD. Hypothermia and barbiturates: individual and combined effects on canine cerebral oxygen consumption. Anesthesiology 1983;58:527-32.
The Journal of Thoracic and Cardiovascular Surgery
12. Woodcock TE, Murkin JM, Farrar JK, Tweed WA, Guiraudon GM, McKenzie FN. Pharmacologic EEG suppression during cardiopulmonary bypass: cerebral hemodynamic and metabolic effects of thiopental or isoflurane during hypothermia and normothermia. Anesthesiology 1987;67:218-24. 13. Field J II, Fuhrman FA, Martin AW. Effect of temperature on the oxygen consumption of brain tissue. J Neurophysiol 1944;7:117-26. 14. Norwood WI, Norwood CR, Ingwall JS, Castaneda AR, Fosse1 ET. Hypothermic circulatory arrest. J THORAC CARDIOVASC SURG 1979;78:823-30. 15. Stocker F, Herschkowitz N, Bossi E, et aI. Cerebral metabolic studies in situ by 31P-nuclear magnetic resonance after hypothermic circulatory arrest. Pediatr Res 1986; 20:867-71. 16. Wollman H, Stephen GW, Clement AJ, Danielson GK. Cerebral blood flow in man during extracorporeal circulation. J THORAC CARDIOVASC SURG 1966;52:558-64. 17. Bering EA Jr. Effect of body temperature change on cerebral oxygen consumption of the intact monkey. Am J PhysioI1961;200:417-9. 18. Rosomoff HL, Holaday DA. Cerebral blood flow and cerebral oxygen consumption during hypothermia. Am J Physiol 1954;179:85-8. 19. Michenfelder JD, Theye RA. Hypothermia: effect on canine brain and whole-body metabolism. Anesthesiology 1968;29:1107-12. 20. Greeley WJ, Kern FH, Ungerieider RM, et aI. The effect of hypothermic cardiopulmonary bypass and total circulatory arrest on cerebral metabolism in neonates, infants, and children. J THORAC CARDIOVASC SURG 1991;101:783-94. 21. Fox LS, Blackstone EH, Kirklin JW, Bishop SP, Bergdahl LAL, Bradley EL. Relationship of brain blood flow and oxygen consumption to perfusion flow rate during profoundly hypothermic cardiopulmonary bypass. J THORAC CARDIOVASC SURG 1984;87:658-64. 22. Miyamoto K, Kawashima Y, Matsuda H, Okuda A, Maeda S, Hirose H. Optimal perfusion flow rate for the brain during deep hypothermic cardiopulmonary bypass at 20° C. J THORAC CARDIOVASC SURG 1986;92:I065-70. 23. Weinberger LM, Gibbon MH, Gibbon JH Jr. Temporary arrest of the circulation to the central nervous system. Arch Neurol Psychiatry 1940;43:615-34. 24. Michenfelder JD, Theye RA. The effects of anesthesia and hypothermia on canine cerebral ATP and lactate during anoxia produced by decapitation. Anesthesiology 1970; 33:430-9.