Acute bleeding is the most common etiology of acute circulatory failure in surgical patients. Monitoring of organ function during an acute reduction in oxygen delivery (DO2) remains inaccurate. A reduction in arterial pressure or in cardiac index (CI), or an increase in heart rate (HR) are neither sensitive nor specific for inadequate DO2, and furthermore, do not accurately reflect cellular function. During acute bleeding, cardiac output (CO) decreases earlier than blood pressure , but the mixed venous oxygen saturation (SVO2) simultaneously decreases, reflecting an increase in oxygen extraction (O2 ER)  that can maintain oxygen uptake (VO2). Monitoring of SVO2 may therefore represent an early warning sign of decreased DO2, but does not accurately reflect the development of cellular hypoxia. A reduction in base excess or an increase in blood lactate levels are better indicators of cellular hypoxia, but these variables lack specificity . In addition, blood lactate levels are influenced not only by lactate production but also by elimination, which may be decreased in some patients. Therefore, there is a need for other indicators of cellular hypoxia during severe hemorrhage.
In experimental conditions, progressive hemorrhage has been used as a model to study the relationship between VO2 and DO2. In the early stages of hemorrhage, the decrease in DO2 is compensated by an increase in O2 ER so that VO2 is maintained and remains DO2 independent. It is only when DO2 decreases below a critical value that VO2 can no longer be maintained, and becomes DO2-dependent. The decrease in VO2 is associated with the development of anaerobic metabolism, as reflected by an abrupt increase in blood lactate level [4-6].
Several authors have reported a widening in arteriovenous gradient for PCO2 (AV PCO2) and for pH (AV pH) in acute circulatory failure related to low flow states [7-11]. This increase in AV PCO2 and AV pH, proportional to the severity of shock , has been related not only to a reduced clearance of CO2 by the decreased pulmonary vascular flow [7-10,13] but also to an increased generation of CO2 related to the buffering of increased intracellular hydrogen ion production in the hypoxic cell [7-9,12]. Therefore, the hypoxic threshold of the tissues during bleeding may be reflected not only by an increase in blood lactate levels but also by an abrupt widening in AV PCO2 and pH. The present study tested this hypothesis in anesthetized dogs submitted to progressive hemorrhage. To separate the effects of the decreased blood flow from the effects of the reduction in hemoglobin concentration, we maintained the hematocrit constant in these experiments.
After approval of our institutional Animal Care Committee, 24 mongrel dogs (weight, 27.6 +/- 2.7 kg) were studied. After intravenous administration of 20 mg/kg of thiopental, endotracheal intubation was performed and mechanical ventilation was started on control mode (Elema 900B; Siemens, Solna, Sweden) with air (FIO2 = 0.21). The respiratory rate was set at 12 breaths/min and the tidal volume was adjusted to obtain a PaCO2 between 35 and 40 mm Hg. The respiratory conditions then remained unchanged until the end of the study. Muscle relaxation was obtained with vecuronium bromide given as an initial bolus dose of 0.1 mg/kg, followed by a continuous infusion of 0.1 mg centered dot kg-1 centered dot h-1. Anesthesia was maintained with isoflurane, at the end-tidal concentration of 1.4%, corresponding to 1 minimum alveolar anesthetic concentration for the dog .
Electrodes were attached to the four limbs for HR monitoring. Two catheters (16-gauge; Becton Dickinson, Rutherford, NJ) were inserted percutaneously, one into a peripheral vein for fluids and drug infusion, and one from a femoral artery into the distal aorta for arterial pressure monitoring and arterial blood sampling. A pulmonary artery catheter (Swan-Ganz catheter model 93A-131-7F; Baxter, Irvine, CA) was inserted by surgical cutdown of the right jugular vein into a pulmonary artery. A gas analyzer (Capnomac AGM-103; Datex, Helsinki, Finland) was inserted in the respiratory circuit for PETCO2 and isoflurane monitoring. Exhaled gases were directed through a mixing chamber for sampling. Expired oxygen fraction (FEO2) was continuously measured by a semirapid gas analyzer (model 500D PK; Morgan Co., Chatham, England).
HR, arterial pressure, and pulmonary artery pressure were continuously monitored (Sirecust; Siemens AG, Erlangen, Germany) and recorded together with PETCO (2) and FEO2 (model 8000S; Gould Electronique, Ballainvilliers, France); the zero pressure was set at the midchest level of the animal. Body temperature was kept constant throughout the study, using warming blankets and humidified heated gases.
A splenectomy was performed through a midline laparotomy to prevent blood autotransfusion during hypovolemia. To compensate for insensible losses, the dogs received an intravenous infusion of 5% dextrose in 0.9% saline at a rate of 10 mL/kg during splenectomy and 1 mL centered dot kg-1 centered dot h-1 thereafter, until the end of the experiment.
HR and intravascular pressures, including mean arterial pressure (MAP), mean pulmonary artery pressure (MPAP), right atrial pressure (RAP), and pulmonary artery balloon occlusion pressure (PAOP) were measured at end-expiration. PETCO2 and FEO2 were measured from the paper trace over 1 min. Expired volume was measured with a spirometer (model Magtrak II; London, England) over the same period. CI was measured by the thermodilution technique (module E 2261/A; Siemens AG, Erlangen, Germany) by repeated injections of 5 mL of cold (<5 degrees C) 5% dextrose in water. Each injection was started at the end of inspiration. Three to five measurements, with a variability of less than 10%, were averaged for each CO determination and divided by the dog's weight. Immediately thereafter, arterial and mixed venous blood samples were withdrawn for measurement of blood gas tensions (ABL 2; Radiometer, Copenhagen, Denmark). Hemoglobin concentration was determined using a spectrophotometric method (Coulter JT; Coulter Electronics, Hialeah, FL), and its saturation was calculated using the Rossing and Cain nomogram , as direct measurement was not possible. Serum lactate concentration was measured enzymatically by an automated analyzer (Kontron, Basel, Switzerland). Calibration with standard solutions, containing 0, 5, and 10 mM/L of lactate was performed before each experiment.
Stroke index (SI), left ventricular stroke work index (LVSWI), systemic vascular resistance (SVR), and DO2 were calculated by the following formulas: Equation 1 where CaO2 and CvO2 represent the O2 contents of the arterial blood and the mixed venous blood, respectively. VO2 was calculated from the measured FEO2 and minute ventilation, using the appropriate mass balance Equation indexedfor weight .
Baseline measurements were performed 30 min after the end of the splenectomy. The dogs were then slowly hemorrhaged by repeated withdrawals of 3-mL/kg aliquots of arterial blood to reduce cardiac output and DO2. The blood withdrawn was collected into citrate phosphate dextrose bags and then centrifuged to obtain packed cells. After each blood withdrawal, a 10-min period was allowed to achieve a new steady state defined by stable MAP, HR, PETCO2, and FEO2. Then a new set of measurements was performed. Blood withdrawals were repeated to obtain a stepwise reduction in DO2 until the animal could no longer maintain a stable blood pressure, and death occurred shortly thereafter. In each dog, hematocrit was controlled at each step and maintained by retransfusing autologous erythrocytes if necessary.
In each animal, the determination of the critical DO2 (DO2 crit) was obtained from a plot of VO2 versus DO2 using the dual lines regression method described by Samsel and Schumacker . DO2 crit was defined as the point of intersection of the two best-fit regression lines as determined by a least sum of squares technique. The critical O2 ER (O2 ERcrit) was calculated as the ratio of VO2/DO2 at DO2 crit. The value of DO2 crit was also determined in each animal using the same dual regression lines analysis from a plot of AV PCO2, AV pH, and blood lactate versus DO2. These values were called DO2 (AV PCO2)crit, DO2 (AV pH)crit, and DO2 (Lac)crit, respectively. Figure 1 shows data from one illustrative dog.
Hemodynamic and gasometric variables obtained at baseline, just before the critical point, and at the final stage of the experiment were compared using a two-way analysis of variance for repeated measurements, followed by a Tukey's test when it was statistically significant. Critical values of DO2 were compared using a one-way analysis of variance. For all tests, a P value < 0.05 was considered as significant. All values are expressed as mean +/- SD.
The hemorrhage protocol resulted in progressive decreases in aortic and pulmonary artery pressures and CO Table 1. HR increased, but only in the late stage of the protocol. Core temperature and hemoglobin levels remained stable throughout the study.
During the first part of the protocol, VO2 remained stable as the progressive reduction in DO2 was associated with a corresponding increase in O2 ER. Lactate increased slightly but not significantly. PaCO2 remained stable, but mixed venous PCO2 (PvCO2) increased, resulting in a widening of the AV PCO2 gradient. pHv decreased more than pHa, so that AV pH increased significantly. PETCO2 decreased significantly Table 1.
When DO2 decreased below a critical value, the increase in O2 ER became insufficient to maintain VO2, which began to decrease in parallel to DO2. Blood lactate increased markedly Table 1, Figure 2 and Figure 3. There was also an abrupt widening in AV PCO2 and AV pH gradients Figure 2 and Figure 3. The response of AV CO2 to the decrease in CO appeared more biphasic than that of SVO2Figure 4. Moreover, to define in each animal a critical value of DO2 from a plot of SVO2 versus DO2 was generally impossible. PETCO2 also decreased abruptly Figure 5. These changes were associated with an increase in alveolar dead-space, reflected by an increase of the ratio (PaCO2 - PETCO2)/PaCO2Table 1. The critical values of DO2, obtained from blood lactate, AV PCO2, and AV pH, were similar to those obtained from VO2Table 2. A good correlation was found between blood lactate and AV PCO (2) or AV pH Figure 6.
The present study demonstrated that during hemorrhage in anesthetized dogs, an abrupt widening in AV PCO2 and AV pH may delineate the hypoxic threshold of the tissues as accurately as do blood lactate levels. The present study was conducted in anesthetized, paralyzed, and mechanically ventilated dogs, so as to reproduce the conditions encountered when bleeding occurs during surgery. To maintain a stable oxygen demand, we also kept the body temperature constant during the study. As previously reported for this model, VO2 started to decline when DO2 decreased below a critical value of approximately 9 mL centered dot min-1 centered dot kg-1. The DO2 crit values obtained from AV PCO2 and AV pH gradients were similar to those obtained from VO2 or blood lactate measurements.
As expected from the Fick Equation forPCO2, hemorrhage resulted in a progressive increase in PvCO2 with a corresponding widening of the AV PCO (2) and AV pH gradients, reflecting the decrease in blood flow at any given VO (2) and CO2 uptake [7-10,13]. However, the abrupt increase in AV PCO2 and AV pH below DO2 crit cannot be explained solely by this mechanism. One must therefore incriminate a relative overproduction of PCO2 by the ischemic tissues.
During cellular hypoxia, the hydrogen ions generated by the hydrolysis of adenosine triphosphate [8,9] are buffered by the bicarbonate present in the cells, so that CO2 is generated. The increase in tissue PCO2 observed during hypoxia can be predicted from the solubility of CO2 and the amount of hydrogen ion generated . Approximately one-third of the increase in PCO2 may be produced from the hydrolysis of adenosine triphosphate to adenosine diphosphate . These observations could explain the significant correlation found between blood lactate and AV PCO2 and AV pH gradients in our study.
Previous studies have reported an increase in PvCO2 during an acute reduction in blood flow during tamponade  or cardiopulmonary resuscitation  but did not relate this mixed venous hypercarbia to the development of cellular hypoxia. Some other studies focused on increases in PvCO2 in various tissues such as the myocardium [19,20], the brain , the kidney , and the intestine . We focused on global systemic changes, as these only are monitored easily during anesthesia.
Two recent studies also reported a significant widening in AV PCO2 and AV pH during an acute reduction in DO2 below its critical value. The present study extends these observations. Bowles et al.  also used a dog model of acute hemorrhage, but there are three differences between their study and ours. First, they did not maintain a constant hematocrit. Although it may seem logical to allow the hematocrit to decrease during acute bleeding, we preferred to avoid this interference because we have recently shown that hematocrit changes can influence the oxygen extraction capabilities during acute hemorrhage . Second, they gave bicarbonate to compensate for the decrease in pH and this intervention can clearly influence PCO2. Third, they used another anesthetic technique, i.e., a continuous infusion of pentobarbital, that may differently alter the cardiovascular response of the dog to hemorrhage [25,26]. Zhang and Vincent , using a model of cardiac tamponade, found similar results in pentobarbital-anesthetized dogs. However, the conditions of their study are less directly applicable to the surgical patient in the operating room. Nevertheless, these two reports concur to indicate that AV PCO2 and AV pH are reliable indexes of tissue hypoxia regardless of the mechanism accounting for the acute reduction in blood flow.
As in other studies, the reduction in DO2 below its critical value was also associated with an abrupt increase in blood lactate levels [4-6]. We believe that AV PCO2 and AV pH have three attractive properties over blood lactate levels. First, the elimination of PCO2 and pH can be more rapid than the lactate which is particularly prolonged in the presence of altered liver function. Second, lactate levels may increase in conditions other than hypoxia. Conditions such as recent seizures or decompensated diabetes may complicate the interpretation of blood lactate levels . Also the use of lactate levels as an index of tissue perfusion in the presence of sepsis has been recently debated [28,29]. Third, lactate analyzers may not be readily available in the immediate vicinity of many operating rooms, whereas blood gas analyzers are. In any case, the analysis of AV PCO2 and AV pH is not meant to replace measurements of lactate levels but rather to complement them. The present data show that simultaneous assessment of arterial and mixed venous blood gases can represent a simple and yet useful monitoring system of cellular function during severe hemorrhage. Clinical studies should confirm the usefulness of this monitoring system in these conditions. Recent studies have indicated that PETCO2 can monitor tissue perfusion in extremely low flow states, and during cardiopulmonary resuscitation [30,31]. We also observed that PETCO2 decreased abruptly when DO2 decreased below DO2 crit. Since alveolar ventilation was maintained constant, this reduction in PETCO2 could only reflect a reduction in pulmonary blood flow or a reduction in CO2 production, and the two occurred together during hypoxia. The increase in alveolar dead space was also documented by an increase in the (PaCO2 - PETCO2)/PaCO2 ratio. Hence, the present study also illustrates that PETCO2 monitoring can reliably detect a critical reduction in blood flow when ventilation is constant. These observations are of particular interest, since capnography is now used routinely during anesthesia.
1. Guyton AC. Circulatory shock and physiology of its treatment. In: Guyton AC, ed. Textbook of medical physiology. Philadelphia: WB Saunders, 1986;326-35.
2. Smith EE, Crowell JW. Effect of hemorrhagic hypotension on oxygen consumption of dogs. Am J Physiol 1964;207:647-9.
3. Mizock BA, Falk JL. Lactic acidosis in critical illness. Crit Care Med 1992;20:80-93.
4. Van der Linden P, Gilbart E, Engelman D, et al. Effects of anesthetic agents on systemic critical O2
delivery. J Appl Physiol 1991;71(1):83-93.
5. Nelson DP, Beyer C, Samsel RW, et al. Pathological supply dependence of O2
uptake during bacteremia in dogs. J Appl Physiol 1987;63:1487-92.
6. Heusser F, Fahey JT, Lister G. Effect of hemoglobin concentration on critical cardiac output and oxygen transport. Am J Physiol 1989;256:H527-32.
7. Grundler W, Weil MH, Rackow EC. Arteriovenous carbon dioxide and pH gradients during cardiac arrest. Circulation 1986;74:1071-4.
8. Weil MH, Rackow EC, Trevino R, et al. Difference in acid-base state between venous and arterial blood during cardiopulmonary resuscitation. N Engl J Med 1986;315:153-6.
9. Adrogue HJ, Rashad MN, Gorin AB, et al. Assessing acid-base status in circulatory failure. N Engl J Med 1989;320:1312-6.
10. Mathias DW, Clifford PS, Klopfenstein HS. Mixed venous blood gases are superior to arterial blood gases in assessing acid-base status and oxygenation during acute cardiac tamponade in dogs. J Clin Invest 1988;82:833-8.
11. Benjamin ETJ, Paluch TA, Berger SR, et al. Venous hypercarbia in the canine hemorrhagic shock. Crit Care Med 1987;15:516-20.
12. Johnson BA, Weil MH. Redefining ischemia due to circulatory failure as dual defects of oxygen deficits and of carbon dioxide excesses. Crit Care Med 1991;19:1432-8.
13. Groeneveld ABJ, Vermeij CG, Thijs LG. Arterial and mixed venous blood acid-base balance during hypoperfusion with incremental positive end-expiratory pressure in the pig. Anesth Analg 1991;73:576-82.
14. Steffey EP, Howland DJ. Potency of enflurane in dogs: comparison with halothane and isoflurane. Am J Vet Res 1978;39:573-7.
15. Rossing RG, Cain SM. A nomogram relating PO2
, pH, temperature and hemoglobin saturation in the dog. J Appl Physiol 1966;215:195-201.
16. Hill RW. Determination of oxygen consumption by use of the paramagnetic oxygen analyzer. J Appl Physiol 1972;33:261-3.
17. Samsel RW, Schumacker PT. Determination of the critical O2
delivery from experimental data: sensitivity to error. J Appl Physiol 1988;64:2074-82.
18. von Hanwher R, Smith M, Siesjo BK. Extra- and intracellular pH during near-complete forebrain ischemia in the rat. J Neurochem 1986;43:331-8.
19. Khuri SF, Flaherty JT, O'Riordan JB, et al. Changes in intramyocardial ST segment voltage and gas tensions with regional myocardial ischemia in the dog. Circ Res 1975;37:455-63.
20. Khuri SF, Kloner RA, Karaffa SA, et al. The significance of the late fall in myocardial PCO2
and its relationship to myocardial pH after regional coronary occlusion in the dog. Circ Res 1985;56:537-47.
21. Nelimarkka O, Niinikoski J. Oxygen and carbon dioxide tension in the canine kidney during arterial occlusion and hemorragic hypotension. Surg Gynecol Obstet 1984;158:27-32.
22. Hartmann M, Montgomery A, Jonsson K, Haglund U. Tissue oxygenation in hemorrhagic shock measured as trans cutaneous oxygen tension, subcutaneous oxygen tension, and gastrointestinal intramucosal pH in pigs. Crit Care Med 1991;19:205-10.
23. Bowles SA, Schlichtig R, Kramer DJ, Klions HA. Arteriovenous pH and partial pressure of carbon dioxide detect critical oxygen delivery during progressive hemorrhage in dogs. J Crit Care 1992;7:95-105.
24. Van der Linden P, Gilbart E, Paques P, et al. Influence of hematocrit on tissue O2
extraction capabilities in anesthetized dogs during acute hemorrhage. Am J Physiol 1993;264:H1942-7.
25. Weiskopf RB, Townsley MI, Riordan KK, et al. Comparison of cardiopulmonary responses to graded hemorrhage during enflurane, halothane, isoflurane and ketamine anesthesia. Anesth Analg 1981;60:481-90.
26. Zimpfer M, Mander WT, Barger AC, Vatner SF. Pentobarbital alters compensatory neural and humoral mechanisms in response to hemorrhage. Am J Physiol 1982;243:713-21.
27. Zhang H, Vincent J-L. Arteriovenous differences in PCO2
and pH are good indicators of critical hypoperfusion. Am Rev Respir Dis 1993;148:867-71.
28. Vary TC, Siegel JH, Nakatani T, et al. Effect of sepsis on activity of pyruvate dehydrogenase complex in skeletal muscle and liver. Am J Physiol 1986;250:E634-40.
29. Hurtado FJ, Gutierrez AM, Silva R, et al. Role of tissue hypoxia as the mechanism of lactic acidosis during E. coli endotoxemia. J Appl Physiol 1992;72:1895-901.
30. Weil MH, Bisera J, Trevino RP, Rackow EC. Cardiac output and end-tidal carbon dioxide. Crit Care Med 1985;13:907-9.
31. Falk JL, Rackow EC, Weil MH. End-tidal carbon dioxide concentration during cardiopulmonary resuscitation. N Engl J Med 1988;318:607-11.