INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Carbon monoxide (CO) inhalation is a leading cause of illness and fatal poisoning in the United States (1), United Kingdom (2), Korea (3), France (4), and much of the industrialized world. CO is rapidly absorbed because of its 200 to 250 times greater affinity than O₂ for the O₂ binding site on hemoglobin (5). CO displaces O₂ from hemoglobin and impairs the release to the tissues of the remaining bound O₂ (6). The most effective therapy is administration of 100% O₂ at increased pressures (hyperbaric O₂). Unfortunately, the vast majority of patients who require hyperbaric O₂ treatment (7) are subject to delays associated with transfer to suitable facilities, if available. During this interval, 100% O₂ at ambient pressures can be administered, but is much less effective (2, 8).

In considering other therapeutic approaches to more effective early treatment, we noted that mathematical models of CO kinetics (9) suggest that increased minute ventilation enhances the rate of CO elimination, a hypothesis confirmed in animal studies (13). However, it is not a recommended therapeutic option for CO poisoning (14). The use of hyperventilation in treating CO poisoning is constrained by concerns about the hypocapnia-induced reduction of O₂ delivery to the brain owing to both a decrease in cerebral blood flow (15) and a shift of the oxyhemoglobin dissociation curve further to the left.

Using a simple passive nonrebreathing circuit that allows hyperpnea without hypocapnia (16), we tested the hypothesis that isocapnic hyperpnea as an adjunct to 100% O₂ accelerates elimination of CO from the blood of anesthetized dogs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

After obtaining institutional board approval, studies were performed on five pentobarbital-anesthetized, intubated, ventilated mongrel dogs of either sex weighing 20 to 22 kg. Adequacy of anesthetic depth was deduced from the eyelash reflex response, lack of spontaneous movements, and the stability of heart rate and blood pressure. Anesthesia was supplemented as necessary to prevent spontaneous respiratory efforts. Catheters were placed in the femoral and pulmonary arteries for monitoring blood pressure, measuring cardiac output by thermodilution (Model 9520A; American Edwards Laboratories, Irvine, CA), and periodic sampling of blood for analysis. Percentage carboxyhemoglobin (COHb) was measured photometrically (Radiometer OSM3; Radiometer A/S, Copenhagen, Denmark) and corrected for canine blood. CO₂ was sampled from the proximal end of the endotracheal tube and analyzed continuously (Ametek; Thermox Instruments Division, Pittsburgh, PA). Flow was measured with a pneumotachograph (Vertek series 47303A; Hewlett-Packard, Palo Alto, CA) and the signal integrated to obtain volume. Analog signals were digitized at 17 samples per second and recorded using commercial electronic data acquisition software (WINDAQ/200; DATAQ Instruments, Inc., Akron, OH).

Protocol

The dogs were ventilated (Model 618; Harvard Apparatus, South Natick, MA) with 0.28% CO in room air for 70 to 90 min (resulting in COHb range of 59 to 71%) and then with room air. Control ventilator settings were set at a tidal volume (VT) of 16 ml · kg¹ and a frequency (f) of 11 min¹ (duty cycle 0.5). All dogs were ventilated to maintain Pa_CO2 at 40 mm Hg and to just below their apneic threshold (by adjusting VT ± 50 ml). These initial ventilator settings were defined as "normal ventilation" (Figures 1 and 2). The dogs were then ventilated with 100% O₂ at normal ventilation followed by normocapnic hyperpnea (VT = 50 ml · kg¹, f = 24 min¹). All ventilator settings were maintained for at least 42 min; measurements of blood gases, COHb, blood pressure, and cardiac output were made approximately every 7 min. A circuit connected to the gas inlet of the ventilator (16) was used to prevent hypocapnia. It consists of a nonrebreathing valve attached to two gas sources. The first provides a flow of fresh gas (fraction of inspired oxygen [FI_O2] = 1.0, FI_CO and FI_CO2 = 0) equal to the control minute ventilation and contributes to elimination of both CO and CO₂. At higher ventilations, additional inhaled gas is provided via a demand valve from the second source. This reserve gas has a PCO = 0 mm Hg and a PCO₂ approximating that of the oxygenated mixed venous CO₂ pressure (Pv_CO2) and thus provides a gradient for elimination of CO but not CO₂. With hyperpnea, therefore, the circuit allows CO elimination at a rate commensurate with the minute ventilation, but limits CO₂ elimination to that during control ventilation. We refer to this as "isocapnic hyperpnea" because changes in Pa_CO2 are minimized despite large increases in minute ventilation (16). Note that at high minute ventilation, reserve gas makes up most of the inspired gas and the FI_O2 approaches that of the reserve gas.

View larger version (14K): [in this window] [in a new window]   Figure 1.   The effect on COHb levels (as percent) in a dog sequentially exposed to CO (Poisoning), ventilated at control levels with room air (NVRA) and 100% O2 (NVO2), and then six times control with 100% O2 (HO2). At high minute ventilations with the reserve gas supplying approximately 5.5% CO2, FIO2 approached 0.95. COHb is plotted on a logarithmic scale to linearize its exponential decrease with time.

View larger version (10K): [in this window] [in a new window]   Figure 2.   The t1/2 for elimination of COHb in all dogs. Bars represent means. Differences between categories are all highly significant (p < 0.001). Abbreviations as defined in Figure 1.

Data Analysis

The half-times for decrease of COHb (t_1/2 ) were calculated from exponential curves fit to the COHb values by the method of least squares. A minimum of five data points were used to fit each curve (17). Data from the first 10 min after poisoning and first 7 min after a change in FI_O2 or ventilation were excluded from the calculation in order to minimize the effect of equilibration of COHb between the various tissue compartments (17, 18) on the t_1/2 calculation. Results were compared using paired t tests with Bonferroni correction where appropriate; p < 0.05 was considered significant. Data are expressed as means ± standard deviation unless stated otherwise.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The changes in COHb with time in a dog treated with 100% O₂ are presented in Figure 1. Regressions for all phases of experiments in all dogs had r² > 0.95. The t_1/2 for all dogs are presented in Figure 2; during normal ventilation on room air (NVRA), t_1/2 was 212 ± 17 min and decreased to 42 ± 3 min during normal ventilation with 100% O₂ (NVO₂). Isocapnic hyperpnea on 100% O₂ (HO₂) led to a further decrease in t_1/2 to 18 ± 2 min (p < 0.0005).

Physiologic data for all dogs are presented in Table 1. CO poisoning did not decrease Pa_O2 or arterial pH. Cardiac output increased from 3.2 ± 0.4 L · min¹ to 4.4 ± 0.2 L · min¹ during exposure to CO. NVO₂ was associated with a decrease in cardiac output from 4.3 ± 0.3 L · min¹ during NVRA to 3.6 ± 0.5 L · min¹. Following HO₂, cardiac output fell further to 2.6 ± 0.5 L · min¹, which did not differ from that during control ventilation. The fall in mean arterial pressure from 96 ± 6 mm Hg during NVO₂ to 70 ± 5 mm Hg during "isocapnic" hyperpnea almost reached statistical significance (p = 0.053). Otherwise, the physiologic status of the dogs (arterial oxygenation and acid-base status) was the same during HO₂ as during NVO₂. During "isocapnic" hyperpnea, Pa_CO2 did not differ from that during NVRA despite a sixfold increase in minute ventilation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Isocapnic hyperpnea, when added to normal ventilation with 100% O₂, can more than double (t_1/2 = 18 min) the mean rate of elimination of COHb in anesthetized dogs. Comparable data from animals with controlled ventilation treated with hyperbaric O₂ are not available from the literature. We therefore calculated the t_1/2 in two similarly prepared normocapnic ventilated dogs administered a standard hyperbaric treatment (19) for CO poisoning; their t_1/2 values were 20 and 28 min.

Dogs have long been used to model CO poisoning in humans (13, 20). The affinity of their hemoglobin for CO and O₂ and their cellular O₂ transport and respiration are similar to those of humans (13). The t_1/2 for COHb elimination in our dogs were longer than those reported in dogs breathing 100% O₂ at ambient (20, 21) and increased (20) pressures. The likely explanation is that the spontaneously breathing dogs of previous studies, like CO-poisoned humans (23, 24), hyperventilated and thus had minute ventilations exceeding those of our mechanically ventilated dogs. Nevertheless, the t_1/2 for COHb elimination in our dogs ventilated at control levels are similar to those reported by Pace and coworkers (25) for human volunteers spontaneously breathing 100% O₂ at ambient (t_1/2 in males = 47 min) and increased pressures (t_1/2 = 22 min in males, at 2.5 Atm).

The effect of ventilation on CO elimination is explained by the Haldane equation (26): M · PCO/PO₂ = COHb/O₂Hb, where M is the relative affinity ratio of CO to O₂ for blood, and PCO and PO₂ are the partial pressures of CO and O₂, respectively, and COHb and O₂Hb are expressed as percent of total Hb concentration. The large M of about 200 predicts that even small (fractions of a mm Hg) reductions in PCO will result in major improvements in blood O₂-carrying capacity owing to a large decrease of COHb and equivalent increase in O₂Hb.

In our protocol we sequentially introduced progressively more efficacious treatments, with isocapnic hyperpnea last. This protocol has two major advantages. First, it follows the probable clinical sequence of CO exposure: poisoning, breathing room air, receiving treatment with 100% O₂, then possibly the test treatment. Second, had we randomized the animals to one test condition per animal, many more test animals would have been required to account for interanimal variability; our protocol allowed each animal to serve as its own control.

We believe that neither the sequence of treatment nor changes in physiologic state of the animals during each stage of the protocol decreased the t_1/2 during isocapnic hyperpnea for four reasons. First, the different COHb during each stage of the protocol should not affect the calculation of t_1/2 as this is independent of initial values.

Second, the effect of redistribution of CO from blood to tissues can decrease the calculated t_1/2. However, after an equilibrium is established between the PCO of blood and tissues (as during prolonged poisoning), initiating rapid CO elimination, such as occurs during hyperpnea with O₂, results in CO moving into the blood down its partial pressure gradient. This replenishment of the "blood pool" of CO would tend to prolong t_1/2. In preliminary experiments and as can be seen in Figure 1, we noted that for about 10 min after initiating a more effective method of CO elimination, the rate of decrease in COHb transiently increased above the immediately succeeding values, perhaps because of a lag in the establishment of an equilibrium for CO transfer from the tissues. We therefore excluded the first sample after a change in ventilation in order that our calculated t_1/2 reflect more accurately the true rate. The regression of the remaining values of COHb versus time exhibited r² values > 0.95, indicating a high probability that a single factor, minute ventilation, determined t_1/2.

Third, changes in Pa_O2, which can affect the CO distribution between blood and tissues, and which differed with each stage, did not account for the accelerated CO elimination during isocapnic hyperpnea. Coburn and Forman stated, "Under conditions where tissue PO₂ falls and CO/O₂ increases, CO binding to proteins should increase. This has been observed for [myoglobin] during hypoxic hypoxia with Pa_O2 < 40 mm Hg. . . . " (emphasis ours) (5). As can be seen from Table 1, mixed venous oxygen pressure (Pv_O2) increased with NVO₂ and remained high during HO₂, and there was no acidosis. The higher Pv_O2 after NVRA implies better tissue oxygenation indicating that, if anything, there was a redistribution of CO into the blood, prolonging the t_1/2.

Finally, the decrease in cardiac output with HO₂ compared with that with NVO₂ (presumably due to increased pleural pressures associated with higher stroke volumes of the ventilator) would, if anything, prolong the t_1/2 (27) and thus bias against our hypothesis.

In previous studies in animals, CO₂ was added to inspired gas during treatment of CO poisoning in an attempt to augment the release of O₂ to the tissues (Bohr effect), decrease hemoglobin affinity for CO (6, 26), and stimulate breathing to overcome respiratory depression accompanying severe CO toxicity (6, 13, 28). The conclusion drawn from these studies was that the addition of a fixed concentration of CO₂ to inspired gas confers little benefit beyond that afforded by 100% O₂ at ambient (13, 29) or increased (22, 30) pressures and that CO₂ does not increase CO elimination except through stimulation of ventilation in animals with respiratory depression (13).

In humans, the predicted effect of minute ventilation on the rate of decrease of COHb is described by a decreasing hyperbolic curve that is relatively flat above 10 to 15 L · min¹ (9), suggesting that only moderate increases in minute ventilation (with minimal cardiovascular compromise) are required to accelerate CO elimination. Clinically, however, it is mandatory to avoid hypocapnia and the resulting decrease in O₂ delivery to the brain. Although using 5 to 10% CO₂ as the sole inspirate would prevent hypocapnia and stimulate ventilation, the rise in Pa_CO2 causes discomfort in most conscious patients and is unlikely to be sustained voluntarily for the duration required to be of therapeutic benefit. In fact, Leathart (14) pointed out in 1962 that addition of CO₂ to the inspirate "would be harmful," as it would exacerbate the metabolic acidosis present in those patients with severe poisoning. A nonrebreathing method of maintaining Pa_CO2 within an acceptable range during hyperpnea (16) is required to prevent problems associated with hypo- and hypercapnia. The use of rebreathing methods to prevent hypocapnia with hyperpnea, as with the Mapleson D anesthetic circuit, will not accelerate CO elimination as it will cause the CO to be rebreathed along with the CO₂.

The major advantage of our nonrebreathing circuit, which passively maintains isocapnia regardless of the level of hyperpnea, is that, being portable, it can be used both at the scene of poisoning and during transport of a patient. Early initiation of an effective treatment for CO poisoning may substantially improve long-term outcome (31). It stands to reason that even a modest early decrease in COHb and increased O₂ delivery may salvage, or prevent damage to, ischemic tissues. Early treatment may also limit the accumulation of CO in extravascular tissues (17). The half-time of equilibration of CO may be up to 1 h (9) and equilibration of CO between the blood of mother and fetus may require several hours (32). Once CO has entered the fetal tissues, its elimination from this compartment may be even slower (32). Early and rapid elimination of CO from the blood prior to such equilibration would minimize the fetal CO burden, and may reduce the need for subsequent treatment with hyperbaric O₂. However, if hyperbaric O₂ is still necessary after initial treatment (3, 8, 31, 33), complications from barotrauma and hyperoxia may be reduced by employing isocapnic hyperpnea to maintain high rates of CO elimination at lower pressures, O₂ concentrations, or both.

In conclusion, isocapnic hyperpnea with 100% O₂ in dogs more than doubled the rate of CO elimination compared with normal ventilation with 100% O₂. The method and equipment are simple and may provide an early, effective, and practical adjunct to standard therapy with 100% O₂ or even hyperbaric O₂.