Background Why pulmonary gas exchange deteriorates after administration of epinephrine during cardiopulmonary resuscitation (CPR) is unclear.
Methods and Results Forty-four anesthetized swine received an infusion of six inert gases. Animals underwent ventricular fibrillation with CPR and intravenous administration of saline (control), epinephrine (15 μg/kg), or methoxamine (150 μg/kg). Cardiac output, aortic blood pressure, pH, and arterial oxygen saturation were recorded. Distributions of V̇a and Q̇ were determined by the multiple inert gas elimination technique. Ventricular fibrillation and CPR caused significant decreases in cardiac output, aortic blood pressure, and arterial pH. With epinephrine (versus saline), diastolic blood pressure was significantly higher (23±7 versus 8±4 mm Hg), but the increase in shunt (from 7±4% to 29±17%) and the reduction in Sao2 (from 99.7% to 76.8%) were significantly larger. Also, the increase in dead space was greater and elimination of CO2 less. There were no differences between animals given methoxamine or saline, except for increased diastolic blood pressure.
Conclusions During experimental ventricular fibrillation and CPR, epinephrine increased intrapulmonary shunt ≈300% more than saline or methoxamine and significantly reduced arterial oxygen saturation. We suspect that the β-adrenergic receptor activity of epinephrine attenuated hypoxic pulmonary vasoconstriction. Methoxamine is as effective a pressor as epinephrine for CPR and devoid of β-adrenergic activity. We recommend that such an agent be considered, instead of epinephrine, for CPR.
Studies have shown that during cardiopulmonary resuscitation (CPR), pulmonary gas exchange becomes less efficient after administration of epinephrine.1 2 Unfortunately, these investigations have used only determinations of the difference between the alveolar and arterial partial pressures of oxygen and measurements of the partial pressure of end-tidal carbon dioxide. Such calculations do not allow one to differentiate and quantify the various abnormalities that cause mismatching of alveolar ventilation (the volume of alveolar gas in the tidal volume per minute, V̇a) and perfusion (the volume of blood passing through the lungs per minute, Q̇).
For example, use of indices derived from oxygen tension would not reveal whether the cause of impaired gas exchange was right-to-left intrapulmonary shunting of blood, perfusion of lung units having a low value for the ventilation-to-perfusion ratio (Q̇), a defect in oxygen diffusion, or venous admixture from the thebesian (cardiac) and bronchial circulations. Likewise, use of indices derived from carbon dioxide tension would not reveal whether the cause was true dead space ventilation or ventilation of lung units having a high but finite value for V̇a. Moreover, these indices of gas exchange do not permit determination of whether impairment of the process resulted from a change in the distribution of alveolar ventilation or blood flow or both.
Cardiac output and pulmonary blood flow decrease significantly during ventricular fibrillation and CPR. The marked reduction in cardiac output in turn causes a much greater distribution of blood flow to gravity-dependent lung units than to nondependent lung units.3 Positive-pressure breathing, as occurs during CPR, asymmetrically distributes alveolar ventilation to nondependent versus dependent lung units.4 Therefore, significant mismatching of ventilation and perfusion will probably occur during CPR.
Theoretically, the β-adrenergic effects of epinephrine would produce bronchodilation that could affect the distribution of ventilation. However, because epinephrine activates both α-and β-receptors, both pulmonary vasoconstriction (α) and vasodilation (β) are possible. Therefore, to determine the mechanism of epinephrine-induced deterioration of pulmonary gas exchange, we examined the distribution of ventilation and perfusion, relative to regional Q̇ ratios, that were determined by the multiple inert gas elimination technique5 during ventricular fibrillation and CPR in swine.
After approval of the Institutional Animal Care and Use Committee was obtained, swine of either sex were given an injection of ketamine hydrochloride (25 mg/kg IM) followed by sodium pentobarbital (25 mg/kg and 8 mg · kg−1 · h−1 IV). Subcutaneous needle electrodes were placed for monitoring of the ECG. We positioned the animals in the supine position and intubated the trachea (internal diameter of tube, 7.5 mm) through a tracheostomy. For airway sampling of respired gas, a pneumotachograph (Bicore) was attached to the tracheal tube and was connected to a pulmonary mechanics computer (Bicore). Between the pneumotachograph and breathing circuit, we positioned a gas analyzer (Ultima, Datex) calibrated with room air and a gas mixture of 95% O2 and 5% CO2 (Air Products) and the aspiration port of the sample tubing.
Animals were ventilated by means of an oxygen-powered mechanical ventilator/precordial compressor (Thumper, Michigan Instruments) at the rate of 12 breaths per minute and with a tidal volume adjusted to keep the arterial partial pressure of carbon dioxide (Paco2) between 35 and 45 mm Hg.
A 7F thermistor-tipped catheter was inserted into the right internal jugular vein and was advanced into a main pulmonary artery. A polyethylene catheter was inserted into the right carotid artery and was advanced into the descending aorta. Each of these two catheters was hydraulically linked to a pressure transducer, and each had its final position confirmed by pressure tracing. A third catheter was inserted percutaneously into an ear vein. A solution of six inert gases representing a wide range of solubilities in blood (sulfur hexafluoride, ethane, cyclopropane, enflurane, diethyl ether, and acetone) was prepared as described elsewhere5 and was infused at the rate of 2.0 mL/min into the ear vein.
Measurements and Calculations
We used a timed recording of the ECG to determine heart rate and a trace of aortic blood pressure on a calibrated chart to determine systolic and diastolic blood pressures. Pulmonary blood flow was derived by the mass balance method (Fick’s principle) using values for the retention and excretion of sulfur hexafluoride, ethane, cyclopropane, and enflurane and the minute ventilation.5 Core body temperature was measured by means of the thermistor-tipped catheter in the pulmonary artery. The differential pressure produced by the pneumotachograph was processed by the computer to calculate tidal volume (Vt), respiratory rate, and expired minute volume (V̇e). We verified the accuracy of these three computations by simulating tidal ventilation with a giant syringe (Hamilton Medical), using injection-aspiration cycles of 300 mL at the rate of 12 per minute. Gas volume was converted to BTPS. We also measured the fraction of oxygen and carbon dioxide in respired gas.
Following the methods of Wagner et al,5 we obtained 10-mL samples of aortic and pulmonary artery blood simultaneously and in duplicate and used a gas chromatograph (HP 5890, Hewlett-Packard) to assay the samples immediately for their concentrations of inert gases. The blood-gas partition coefficient for each inert gas was determined for each pig as previously described.5 6 We also obtained separate 5-mL samples of aortic and pulmonary artery blood simultaneously and in duplicate and used the appropriate electrodes (IL 1301, Instrumentation Laboratories) to assay the samples immediately for respiratory gas tensions and pH. The hemoglobin content and oxyhemoglobin saturation of these samples were assayed spectrophotometrically (Co-oximeter IL 282, Instrumentation Laboratories). We then used a computer program developed by Ruiz et al7 to convert oxyhemoglobin saturation values to reflect the oxygen-combining characteristics of adult pig hemoglobin.6
Expired gas was collected with a heated corrugated hose and a 9-L box heated to 40°C to 42°C and was vented to the ambient atmosphere through a calibrated Wright respirometer (British Oxygen Co) for collaborative measurement of minute ventilation. We sampled mixed expired gas in duplicate from the collection box for assay of inert gas tensions (gas chromatography) and respiratory gas tensions (gas analyzer). Airway sampling of respired gas was suspended during the collection of expired gas and the determination of lung mechanics.
We determined arterial (Cao2) and mixed venous (C¯vo2) oxygen content, respectively, from arterial and mixed venous oxygen tension and arterial and mixed venous oxyhemoglobin saturation from the following equation: oxygen content= (oxyhemoglobin saturation×hemoglobin content×1.32)+(partial pressure of oxygen×0.0031). Systemic oxygen delivery was computed as the product of arterial blood oxygen content and cardiac output. We calculated the uptake of oxygen and the elimination of carbon dioxide as the product of minute ventilation and the difference between the percentages of inspired and mixed expired oxygen or carbon dioxide, respectively. Intrapulmonary shunting of blood was estimated from the equation Q̇ps/Q̇t=(Cćo2−Cao2)/(Cćo2−C¯vo2), where Q̇ps is physiological shunt, Q̇t is cardiac output, and Cćo2 is the oxygen content of alveolar capillary blood, as determined by assuming a capillary oxygen tension calculated according to the alveolar air equation.8 Extrapulmonary venous admixture was assessed by comparing the shunt value derived from the classic equation with the value derived by the inert gas technique. We assessed the presence of an oxygen diffusion defect by comparing the measured partial pressure of oxygen in arterial blood (Pao2) and the Pao2 predicted by the multiple inert gas elimination technique.9
Inert Gas Analysis
The details for determining the distribution of blood flow and ventilation as a function of the V̇a/Q̇ ratio appear elsewhere.5 6 Briefly, for each inert gas, the retention ratio (the ratio of the partial pressure of the gas in arterial blood to that in mixed venous blood) and the excretion ratio (the ratio of the partial pressure of the gas in mixed expired gas to that in mixed venous blood) were plotted against solubility of the gas. Then, by use of a computer program written in FORTRAN that incorporates the mathematical approach of Evans and Wagner,10 values for retention and excretion were transformed into a plot of the percentage of total ventilation and blood flow as a function of V̇a/Q̇ by a linear least-squares regression method with enforced smoothing (smoothing coefficient, Z=40).
We classified the distributions of V̇a and Q̇ relative to the value for V̇a/Q̇ as follows: (1) perfusion of unventilated lung regions (right-to-left intrapulmonary shunting of venous blood), ie, areas having a value of almost zero (≤0.005) for V̇a/Q̇; (2) perfusion of poorly ventilated lung units, ie, lung regions having a low value (>0.005 through 0.1) for V̇a/Q̇; (3) perfusion of lung regions having a normal value (>0.1 through 10) for V̇a/Q̇; (4) ventilation of poorly perfused lung regions, ie, those having a high value (>10 through 100) for V̇a/Q̇; and (5) ventilation of unperfused lung regions (dead space), ie, those having a very high value (>100) for V̇a/Q̇. Perfusion and ventilation as functions of V̇a/Q̇ are expressed as percentages of pulmonary blood flow and minute ventilation, respectively. The SD of the ventilation distribution (LSD|$$˙) was computed and reflects changes in ventilation heterogeneity.
We collected baseline data at least 60 minutes after initiating infusion of inert gases and when animals were stable: that is, when minute ventilation, end-tidal carbon dioxide (Petco2), heart rate, and mean arterial blood pressure had not changed more than 10% over three consecutive observation periods, each period being 5 minutes. Then, ventricular fibrillation was induced with transthoracic electrocution (400 W/s), and animals underwent CPR by means of mechanical precordial compression (60 compressions per minute). Ventilation (12 breaths per minute) was provided at a compression-to-ventilation ratio of 5:1 and a compression-to-relaxation ratio of 1:1. Compression force was adjusted to produce sternal displacement of 25% of the anteroposterior thoracic diameter.11
After the start of CPR, animals were randomly assigned to receive an injection of 5.0 mL IV of physiological saline alone or with epinephrine (15 μg/kg) or methoxamine (150 μg/kg) immediately and every 5 minutes for 15 minutes. After 15 minutes of CPR, we again collected data. Animals were euthanized without attempting defibrillation.
Data are summarized as mean±SD. We used the Levine median test to assess the variance for response variables. When the variance of response variables during intragroup or intergroup comparisons was equal (P>.05) or not equal (P<.05), statistical analysis was performed with, respectively, a parametric or nonparametric statistic. Intragroup statistical comparisons were performed with Student’s t test for paired observations when variance was equal or the Mann-Whitney rank sum test when variance was not equal. When the variance of the data was equal, intergroup comparisons were performed with an ANOVA for repeated measures; when P<.05, differences were distinguished with Tukey’s test. Intergroup data with unequal variance were compared by the Kruskal-Wallis ANOVA on ranks; when P<.05, differences were distinguished with Dunnett’s method. Statistical analyses were performed with SigmaStat computer software (version 2.0, Jandel Scientific).
Forty-four swine (23±2 kg) underwent positive-pressure ventilation with comparable minute ventilation (5.4±1.2 L/min) and inspired concentrations of oxygen (0.95±0.03) throughout the study, and those data were pooled for summary. Data are summarized in the Tables and Figs 1⇓ and 2⇓. There were no between-group differences in cardiopulmonary function before induction of ventricular fibrillation and initiation of CPR. The results of ventricular fibrillation and CPR by themselves and with epinephrine or methoxamine are as follows.
With Saline (Control Animals, n=14)
By themselves, ventricular fibrillation and CPR decreased pulmonary blood flow and blood pressure; arterial blood pH, oxygenation, and Pco2; oxygen uptake and delivery; and elimination of carbon dioxide. Dead space and shunting of inert gases increased. Data are summarized in the Tables and Figs 1⇑ and 2⇑.
With Epinephrine (n=15)
Although the consequences of CPR with epinephrine versus saline were similar, diastolic blood pressure was higher with epinephrine. Intrapulmonary shunting and dead space ventilation also were greater, conditions that significantly decreased Pao2 and arterial blood oxygen saturation (Sao2) and increased Paco2, respectively. The latter caused arterial pH to be significantly lower with epinephrine than saline or methoxamine. Dead space ventilation was increased by epinephrine compared with saline and methoxamine. Data are summarized in the Tables and Figs 1⇑ and 2⇑.
With Methoxamine (n=14)
When right-to-left intrapulmonary shunting of blood (percentage of cardiac output), as determined by the classic two-compartment model, was compared with shunting of inert gases, similar values resulted for the saline (10.3±5.8% versus 9.8±5.9%), epinephrine (30.8±23.9% versus 29.3±16.9%), and methoxamine (17.9±11.4% versus 16.5±10.7%) groups. This similarity indicated that a comparable small degree of extrapulmonary venous admixture occurred in the animals of all three groups. There was no evidence of an oxygen diffusion defect, because no significant difference existed between the measured Pao2 and the predicted Pao2 after administration of saline (265±142 versus 283±167 mm Hg), epinephrine (143±120 versus 159±137 mm Hg), or methoxamine (216±185 versus 231±176 mm Hg).
We were able to define the mechanism by which epinephrine impaired gas exchange during ventricular fibrillation and CPR in swine. Animals given epinephrine had more right-to-left intrapulmonary shunting of blood and more ventilation of dead space during CPR than those given saline or methoxamine. We found no evidence of increased extrapulmonary venous admixture, and we believe that epinephrine increased blood flow to unventilated alveoli, probably by attenuating hypoxic pulmonary vasoconstriction that most likely results from activation of β-adrenergic receptors.
Martin and colleagues1 observed a decrease in Petco2 after administration of epinephrine in dogs, despite constant minute ventilation during ventricular fibrillation and CPR. The authors speculated that the pulmonary vasoconstrictive effects of epinephrine may have increased shunting and thus impaired elimination of carbon dioxide. As an alternative mechanism, they also suggested that epinephrine may have increased right ventricular afterload, thereby reducing pulmonary blood flow and elimination of carbon dioxide. In view of our findings, it is likely that their observation was secondary to an increase in alveolar dead space.
Tang and associates2 reported a decrease in Pao2 and Petco2 after administration of epinephrine but not methoxamine or saline during ventricular fibrillation and CPR in rats. Also, Paco2 was higher and pH lower with epinephrine than with saline or methoxamine. Rats treated with epinephrine needed significantly more electrical countershocks for conversion of cardiac rhythm from ventricular fibrillation to a functional rhythm and for restoration of spontaneous circulation. The authors speculated that the adverse effect of epinephrine on pulmonary gas exchange might have been caused by diversion of blood from well-ventilated to poorly ventilated alveoli (via α-adrenergic–mediated vasoconstriction) and release of hypoxic vasoconstriction (via β-adrenergic–mediated vasodilation).
Previous investigations have not quantified pulmonary blood flow and have used measurements unsuitable to distinguish and quantify the elements contributing to inefficient elimination of carbon dioxide1 2 and deficient oxygenation2 induced by epinephrine during CPR. Moreover, results of these studies do not permit assessment of the presence of anatomic arteriovenous shunting within the lung12 or pulmonary edema,13 both of which have been observed after administration of epinephrine in dogs.
Impaired oxygenation during CPR frequently is attributed to pulmonary edema,14 15 16 the mechanism of which is unclear. Intracardiac and intrathoracic vascular pressures generated during effective cardiac compression are nearly equal to and often exceed 60 mm Hg.17 The combination of low cardiac output and high pulmonary venous pressure may increase permeability of the endothelium, thereby increasing extravascular lung water. In turn, pulmonary edema may impair gas exchange by compressing airways and alveoli (resulting in shunting and/or V̇a/Q̇ mismatch) and may limit oxygen diffusion. During CPR, an elevated concentration of inspired oxygen will mask the effects of a mismatch of ventilation and perfusion and a defect in oxygen diffusion but will not mask the effects of atelectasis or arteriovenous shunting. Therefore, we used the multiple inert gas elimination technique to quantify the contribution of each mechanism to the defect in gas exchange during CPR and to compare the effects of epinephrine and methoxamine.
Despite having similar mechanical ventilatory and cardiac support, inspired concentrations of oxygen, pulmonary blood flow, and mixed venous oxygen tensions, animals given epinephrine had more right-to-left shunting of blood and ventilation of dead space than those given either saline or the pure α-adrenergic agent methoxamine (these two groups were similar in results).
Distribution of blood flow and ventilation is normally such that gas exchange in alveolar blood is optimized. However, when alveolar hypoxemia occurs, compensatory pulmonary vasoconstriction diverts blood to better-ventilated alveoli. Hypoxic vasoconstriction is enhanced when the oxygen tension of mixed venous blood is lower than normal.18 Thus, pulmonary vasoregulation limits the effect of poor ventilation of the alveoli on arterial oxygenation. Hypoxic pulmonary vasoconstriction is particularly important during hypercapnia, acidemia, and extremely low oxygen tension in mixed venous blood. The increase in blood flow to poorly ventilated alveoli and the concomitant reduction in blood flow to well-ventilated alveoli with epinephrine suggest that the drug attenuates hypoxic pulmonary vasoconstriction. The epinephrine-induced increase in inert gas shunt was reflected by significant arterial oxyhemoglobin desaturation (see Fig 1⇑).
Shunting and dead space ventilation induced by epinephrine was responsible for significant arterial oxyhemoglobin desaturation, increased Paco2, and acidemia. Arterial oxyhemoglobin saturation decreased more with epinephrine than with saline or methoxamine. Arterial blood oxygen content did not decrease significantly, only because the concentration of hemoglobin increased in animals who received epinephrine. The reported mobilization of blood from the spleen and liver of dogs and cats19 20 may account for the increase in hemoglobin content seen in our animals. Alternatively, animals who received epinephrine may have had greater extravasation of intravascular fluid, causing a relative concentration of the hemoglobin. Such increases in hemoglobin content have not been reported to occur in humans undergoing CPR. Therefore, a comparable reduction in oxyhemoglobin saturation would markedly reduce oxygen content and delivery. Oxygen delivery is already critically low during CPR, and further decreases in oxygen saturation and content are undesirable. Animals who received epinephrine had elevation of Paco2. Animals given saline or methoxamine had lower Paco2 values, despite equivalent minute ventilation. These differences were of physiological significance. The increase in dead space and shunt was associated with a reduction in carbon dioxide elimination and respiratory acidosis. This would necessitate an increase in ventilation, which would decrease venous return and oxygen delivery further. This also might explain why Paco2 increases after the administration of sodium bicarbonate during cardiopulmonary resuscitation, ie, increased production of carbon dioxide coincident with decreased carbon dioxide elimination. Thus, methoxamine may render sodium bicarbonate more efficacious than previously appreciated. Moreover, although cardiac output produced with methoxamine and epinephrine is similar, because blood flow to unventilated alveoli is not oxygenated and CO2 is not removed, the cardiac output produced by epinephrine is functionally 20% less effective than that produced by methoxamine.
The goal of CPR is to restore cardiopulmonary function and perfusion of vital organs so that disruption of vital organ function is minimal. The success of resuscitation depends on the maintenance of adequate systemic blood flow and oxygenation of arterial blood and on the minimization of the detrimental effects of CPR. Epinephrine has been the preferred adrenergic agent for the management of cardiac arrest for more than three decades. Animal studies and anecdotal reports involving patients have provided the rationale for its use.21 22 23 24 In contrast, some authors argue that β-adrenergic stimulation might be deleterious during and after resuscitation because of the increase in myocardial oxygen demand.25 The ability of epinephrine to restore spontaneous circulation from ventricular fibrillation and asystole has been shown to be a result of its α1-adrenergic effect. Although other agents having selective α1-adrenergic effects (such as phenylephrine and methoxamine) have been as effective as epinephrine in restoring spontaneous circulation after cardiac arrest,12 13 15 26 27 they are not currently recommended for CPR. We did not attempt defibrillation in the present study.
Our data indicate that the epinephrine-induced deterioration in pulmonary gas exchange during CPR results from a diversion of blood flow from relatively well ventilated to unventilated alveoli. Distribution of blood flow to unventilated alveoli was similar in swine given saline or the selective α1-adrenergic receptor agonist methoxamine, suggesting that the β-agonistic activity of epinephrine blunts or prevents hypoxic pulmonary vasoconstriction. Such an event would increase blood flow to unventilated alveoli and the shunting of blood. Because shunt and dead space were increased by epinephrine, blood flow to ventilated alveoli was decreased. β-Adrenergic control of anatomic shunts has also been studied by Shirai et al.28 In cats, the internal diameter of small pulmonary arteries and veins (100 to 600 μm) was measured during global alveolar hypoxia. The reduction in arterial lumen diameter was greater if cats had received propranolol before experiencing alveolar hypoxia. The authors concluded that hypoxic pulmonary vasoconstriction was attenuated by a β-adrenergic receptor–mediated vasodilation caused by reflex release of catecholamine from the adrenal gland and sympathetic nerves. In our study, the only difference was that the catecholamine (epinephrine) was administered exogenously.
We conclude that epinephrine but not methoxamine exacerbates hypoxemia and acidosis and therefore may not be the drug of choice for CPR. Use of a selective α-adrenergic receptor agonist such as methoxamine would increase diastolic blood pressure without adversely affecting pulmonary gas exchange. Because such an agent would be as effective in resuscitation from experimental ventricular fibrillation, it may be a more appropriate agent for CPR than epinephrine.
Reprint requests to Editorial Office, Department of Anesthesiology, MDC 59, University of South Florida College of Medicine, 12901 Bruce B. Downs Blvd, Tampa, FL 33612.
- Received November 25, 1996.
- Revision received May 15, 1997.
- Accepted May 28, 1997.
- Copyright © 1997 by American Heart Association
Tang W, Weil MH, Gazmuri RJ, Sun S, Duggal D, Bisera J. Pulmonary ventilation/perfusion defects induced by epinephrine during cardiopulmonary resuscitation. Circulation. 1991;84:2101-2107.
Ohlsson J, Middaugh M, Hlastala MP. Reduction of lung perfusion increases V̇a/Q̇ heterogeneity. J Appl Physiol. 1989;66:2423-2430.
Wagner PD, Saltzman HA, West JB. Measurement of continuous distributions of ventilation-perfusion ratios: theory. J Appl Physiol. 1974;36:588-599.
Berggren SM. The oxygen deficit of arterial blood caused by nonventilating parts of the lung. Acta Physiol Scand. 1942;4(suppl XI):1-92.
Evans JW, Wagner PD. Limits on V̇a/Q̇ distributions from analysis of experimental inert gas elimination. J Appl Physiol. 1977;42:889-898.
Himmelhoch SR, Dekker A, Gazzaniga AB, Like AA. Closed-chest cardiac resuscitation: a prospective clinical and pathological study. N Engl J Med. 1964;270:118-122.
Rudikoff MT, Maughan WL, Effron M, Freund P, Weisfeldt ML. Mechanisms of blood flow during cardiopulmonary resuscitation. Circulation. 1980;61:345-352.
Risöe C, Tan W, Smiseth OA. Effect of carotid sinus baroreceptor reflex on hepatic and splenic vascular capacitance in vagotomized dogs. Am J Physiol. 1994;266:H1528-H1533.
Maass-Moreno R, Rothe CF. Carotid baroreceptor control of liver and spleen volume in cats. Am J Physiol. 1991;260:H254-H259.
Shirai M, Shindo T, Ninomiya I. β-Adrenergic mechanisms attenuate hypoxic pulmonary vasoconstriction during systemic hypoxia in cats. Am J Physiol. 1994;266:H1777-H1785.