Deem, Steven A., Michael K. Alberts, Michael J. Bishop, Akhil Bidani, and Erik R. Swenson.CO2 transport in normovolemic anemia: complete compensation and stability of blood CO2 tensions. J. Appl. Physiol. 83(1): 240–246, 1997.—Isovolemic hemodilution does not appear to impair CO2 elimination nor cause CO2 retention despite the important role of red blood cells in blood CO2 transport. We studied this phenomenon and its physiological basis in eight New Zealand White rabbits that were anesthetized, paralyzed, and mechanically ventilated at a fixed minute ventilation. Isovolemic anemia was induced by simultaneous blood withdrawal and infusion of 6% hetastarch in sequential stages; exchange transfusions ranged from 15–30 ml in volume. Variables measured after each hemodilution included hematocrit (Hct), arterial and venous blood gases, mixed expired and , and blood pressure; also, O2 consumption, CO2 production, cardiac output (Q˙), and physiological dead space were calculated. Data were analyzed by comparison of changes in variables with changes in Hct and by using the model of capillary gas exchange described by Bidani (J. Appl. Physiol. 70: 1686–1699, 1991). There was complete compensation for anemia with stability of venous and arterial between Hct values of 36 ± 3 and 12 ± 1%, which was predicted by the mathematical model. Over this range of hemodilution,Q˙ rose 50%, and the O2 extraction ratio increased 61% without a decline in CO2production or a rise in alveolar ventilation. The dominant compensations maintaining CO2transport in normovolemic anemia include an increasedQ˙ and an augmented Haldane effect arising from the accompanying greater O2extraction.
- carbon dioxide
- pulmonary gas exchange
the quantitative importance of red blood cells in blood O2 transport is a function of the direct binding of O2 to hemoglobin. The nearly equivalent role of red blood cells in blood CO2 transport is not generally appreciated, since CO2 is roughly twenty times more soluble than O2in plasma and only 5% of total blood CO2 is bound to hemoglobin. However, physically dissolved CO2accounts for only another 5% of CO2 content; the remainder of CO2 carriage in blood occurs because of the contributions of hemoglobin and two other red cell proteins, carbonic anhydrase and band 3 protein (a membrane-bound Cl−/ exchanger). Although direct CO2binding to hemoglobin as carbamate accounts for only 5% of blood CO2 content, it contributes roughly 15% to CO2 exchange because of the oxylabile characteristics of hemoglobin (Haldane effect). More importantly, 80% of CO2 is carried in blood in the form of , the formation of which is catalyzed within the red blood cell by carbonic anhydrase (7, 24,43). A high rate of formation within the red blood cell is promoted by the large oxylabile buffering capacity of hemoglobin and rapid extrusion of into plasma across the red cell membrane in exchange for Cl−(Cl− shift). These processes forestall a rate-limiting accumulation of H+ and , thereby facilitating greater formation from CO2. The net result is the transport of large amounts of CO2at the expense of only small tissue-to-blood and blood-to-alveolar differences (7, 24, 43).
The effect of a diminished red blood cell concentration on CO2 homeostasis can be predicted by mathematical modeling. With a sophisticated model of human blood-gas transport, Bidani and Crandall (6) predicted that moderate anemia [hematocrit (Hct) ∼30%] will result in a 25% reduction in CO2 elimination (V˙co 2), whereas severe anemia (Hct ∼15%) will result in a 50% reduction in the absence of any compensatory changes in metabolism, ventilation-perfusion ratio (V˙a/Q˙) relationships, ventilation, and cardiac output (Q˙). Similar results were obtained in an earlier model simulation by Hill et al. (21). In vivo, a reduction inV˙co 2 would necessarily lead to CO2 retention and ultimately bring CO2 elimination back into a steady state with its production, albeit at the expense of higher venoarterial gradients. Other evidence that anemia may affect CO2 homeostasis comes from a model of brain gas exchange (40). This model predicts that without changes in metabolism or blood flow, and assuming a constant arterial ( ), brain tissue would rise ∼6 Torr with a fall in Hct from 42 to 15%. In the absence of compensations, therefore, anemia should result in either hypercarbia or a reduction in CO2 output.
Only a few studies have directly addressed the impact of anemia on CO2 transport and, in general, although documenting remarkable stability of in the presence of anemia, these studies have been incomplete in that they have not measured or controlled for all the variables involved in CO2 transport and elimination (4,12, 31). Our study was designed to measure and quantitate the relevant physiological parameters influencing CO2 homeostasis in the rabbit during isovolemic hemodilution. A second aim was to model the gas-exchange and hemodynamic data to determine whether the measured compensations to anemia in vivo are sufficient to predict our measured venous CO2 gas tension ( ) and or whether additional mechanisms of compensation need to be invoked.
The protocol was approved by the Animal Care Committee of the Seattle Veterans Affairs Medical Center. Eight New Zealand White rabbits weighing 3–4 kg were anesthetized with 15 mg/kg iv ketamine and 0.33 mg/kg iv xylazine, and their tracheas were intubated. Mechanical ventilation with a Harvard dual-phase ventilator was initiated at a tidal volume of ∼15 ml/kg and a respiratory rate of 25–30 breaths/min to achieve a baseline of 35–40 Torr. The inspired O2 fraction was set at 0.209. Bilateral femoral arterial catheters were placed surgically for blood pressure monitoring and blood withdrawal. An internal jugular venous catheter was surgically placed and advanced into the right ventricle for withdrawal of mixed venous blood gases and pressure measurements. Anesthesia was maintained with pentobarbital sodium, ∼10 mg ⋅ kg−1 ⋅ h−1by intravenous infusion, and animals were monitored closely for sudden increases in blood pressure and/or heart rate that might indicate insufficient anesthesia. Muscle relaxation was provided by intravenous pancuronium, 0.3–0.4 mg ⋅ kg−1 ⋅ h−1after tracheal intubation.
After stabilization, baseline measurements including mean arterial pressure, right ventricular pressure, Q˙, airway pressure, temperature, and Hct were recorded. Mixed venous and arterial samples were drawn for blood-gas analysis, and expired gas was sampled from a 250-ml mixing chamber attached to the expiratory limb of the ventilator circuit.
Isovolemic anemia was then induced by simultaneous blood withdrawal (via an arterial catheter) and infusion (via the jugular vein) of an equal volume 6% hetastarch in sequential stages; exchange transfusions ranged from 15 to 30 ml in volume. Repeat measurements were made 30 min after each hemodilution or when mixed expired had stabilized. Minute ventilation (V˙e) and inspired O2 fraction were maintained constant throughout each experiment, andV˙ewas measured by a timed collection of expired gas at the conclusion of the experiment. At the conclusion of the experiment (achievement of a Hct of ∼15% or severe hemodynamic derangement), the rabbits were euthanized with an overdose of pentobarbital sodium.
Venous and arterial blood gases and mixed expired CO2 and O2 were measured by an IL 1306 blood-gas machine (Instrumentation Laboratory, Lexington, MA). Mixed expired CO2 was also continuously measured by using a 1100 medical gas analyzer mass spectrometer (Perkin-Elmer, Norwalk, CT). Venous and arterial O2 content were measured using an IL 482 CO-oximeter (Instrumentation Laboratory).
O2 consumption (V˙o 2) andV˙co 2 were calculated from the volumes and concentrations of inspired and expired gases (using appropriate temperature and humidity corrections) and corrected for body weight. Q˙ was calculated by using the Fick equation. The fraction of physiological dead space (Vds/Vt)was calculated by using the Enghoff modification of the Bohr equation, and reported as a percentage. O2extraction ratio ( ) was calculated as the ratio of arteriovenous O2content difference to arterial O2 content, and reported as a fraction.
Data Analysis and Statistics
Two methods of data analysis were used.
Method 1. The slopes of the lines generated by plotting Hct vs. individual measured variables at baseline and after each hemodilution were calculated for individual rabbits by using ordinary least squares. Student’st-test (2-tailed) was then used to determine whether the mean of individual slopes was significantly different from zero. A P value of <0.05 was accepted as statistically significant.
Method 2. The experimental data were analyzed by using a previously described model of capillary gas exchange (5). Several simplifying assumptions are incorporated in the mathematical model: alveolar gas is assumed to be well mixed and uniform throughout the lung, and alveolar gas tensions are assumed to be time invariant. Blood is considered to consist of two well-mixed compartments (plasma and red blood cell) that have the same residence time in the pulmonary capillaries. Blood flow in the capillaries is assumed to be constant and uniform, and axial and radial diffusions are considered negligible. A kinetic mathematical description of pulmonary capillary O2 and CO2 exchange is obtained by deriving mass balance equations for each of the chemical species included. These include CO2, O2, water (or volume), hemoglobin-carbamate (oxygenated and reduced), , Cl−, and H+. Because the latter three exist at different concentrations in the plasma and red cell compartments, the behavior of eleven variables must be described as a function of time (from the time blood enters the pulmonary capillaries). The mass balances describing the time rates of change of the eleven variables involve: 1) rate of consumption or production of that species by chemical reaction within its compartment, and/or2) net transport of the species into or out of its compartment.
Processes included in the quantitative analysis are1) CO2-H2CO3hydration-dehydration reactions in plasma and red blood cells;2) CO2 reactions with hemoglobin;3) O2 binding to hemoglobin and the release of Bohr protons from hemoglobin on oxygenation;4) buffering of H+ intra- and extracellularly by hemoglobin and plasma proteins, respectively;5) /Cl−exchange across the red cell membrane mediated by band 3 protein;6) transcellular movement of water in responses to changes in osmolality; and7) diffusion of O2 and CO2 between alveolar gas and pulmonary capillary blood. Bohr and Haldane effects are included. Ion and water fluxes are described assuming passive diffusion down their respective electrochemical potential and osmotic gradients. Red cell anion exchange is described in terms of a “phenomenological permeability coefficient (PHC )” (5). To account for the availability of pulmonary vascular carbonic anhydrase activity to plasma during capillary transit, an acceleration factor of 200 was used (8).
The mathematical model consists of a set of eleven simultaneous nonlinear ordinary differential equations. The only difference between capillary and postcapillary equations is that net O2 and CO2 movement can take place into or out of the blood only while blood is in the pulmonary capillaries; however, after blood enters the postcapillary vessels, it becomes a closed system, and total O2 and CO2 contents remain constant.
Numerical integration of the model differential equations was implemented by using the Gear algorithm (15), ideally suited for stiff differential systems. Input parameters for solution included kinetic parameters (Table 1; Ref. 5), estimated alveolar gas tensions, central venous blood-gas parameters, and Q˙ corresponding to each experimental condition. To provide uniform groups and to include all rabbits in this analysis, plots were constructed for each variable vs. Hct and values interpolated at percentages of baseline Hct of 100, 84, 76, 52, and 44% (some rabbits did not survive <44% of baseline Hct). Interpolated values of Hct, , venous ( ), Q˙, and venous pH were entered into the model, which subsequently generated predicted values for ,V˙co 2, andV˙o 2.
Six to seven hemodilutions were performed per rabbit, with only five rabbits surviving seven hemodilutions. Data from measurements taken at baseline and with each hemodilution are reported in Table 2. Starting Hct was 36 ± 3% and fell to 12 ± 1%, or 32% of baseline, at its lowest point. There were no significant changes in , , venoarterial difference,V˙co 2, and arterial ( ) with hemodilution, whereas Vds/Vtincreased minimally (Fig. 1 and Table 2).Q˙ and both increased significantly with hemodilution (Fig.2). Despite the rise in Q˙, both systemic and right ventricular pressures remained unchanged. ,V˙o 2, and arterial pH all fell with hemodilution.
Data from the mathematical model are presented in Table3. Because of the previously described derivation by interpolation of data used in the model, the numbers of data points after baseline differ slightly from those reported in Table2. Using our input data, the model predicts that should remain virtually unchanged with each subsequent hemodilution, whereasV˙co 2 andV˙o 2 are predicted to fall only slightly. There is close agreement (<5% difference) between measured (interpolated) values and predicted values using modeling for all variables except V˙o 2. MeasuredV˙o 2 falls more than predictedV˙o 2, particularly at the last analysis point (44% of baseline Hct) where there is a 20% difference between measured and predicted values. For this last data set, if it is assumed that hemodilution and acute anemia result in a small fall in 2,3-diphosphoglycerate, with a corresponding left shift of the oxyhemoglobin dissociation curve (change in at 50% saturation of hemoglobin, ∼5 Torr), there is closer agreement between measured and predicted V˙o 2 (3% difference in values). The difference between measured and predicted , however, becomes slightly greater when using this correction, although the difference is still only 7%.
Our main findings are that acute normovolemic hemodilution to Hct values <40% of normal in anesthetized and mechanically ventilated rabbits does not alter and mixed venous or the venoarterial difference. Stability of values occurs despite an unchanged total and slightly lower alveolar ventilation and no decline in V˙co 2. AlthoughQ˙ rises with anemia, it is not fully compensatory for O2 transport, since we found the classic progressive widening of the arteriovenous difference (i.e., greater ) with decreasing Hct. Modeling of our hemodynamic and respiratory data predicts the observed stability in values.
Critique of Methods
We studied the effects of acute anemia on CO2elimination and sought to improve on previous studies by measuring all variables influencing CO2 production, transport, and elimination, including Q˙,V˙co 2, , pH, and Vds/Vt. Use of anesthesia permitted control and stability of metabolic rate, temperature, and ventilation. Intravascular volume was maintained by performing isovolemic hemodilution with hetastarch to induce acute anemia, thus allowing the expected compensatory increase in Q˙ to occur.
Comparison with Previous Studies
Well over 100 studies of anemia have documented the critical role of increased Q˙ and O2 extraction that compensate for reduced blood O2-carrying capacity in vivo. In only about one-third were measurements of made, and the great majority either failed to measure or lacked explicit description regarding ventilation and V˙co 2, two critical determinants of blood . Those whole animal and human studies that made two or more consecutive hemodilutions to achieve >50% hemodilution found no change in blood values (10, 20, 32, 33, 35, 37,41, 45, 47-49) or a nonstatistically small 1–2 Torr fall (14,22, 27). Only three studies concurrently measured ,V˙co 2, and ventilation but lacked Q˙ measurements, since these were studies in rats. Bartlett and Tenney (4) studied unanesthetized rats hemodiluted with plasma and found that over a wide Hct range (25–45%) there were no changes in tissue (as measured in a subcutaneous gas pocket),V˙e, orV˙co 2. Djordjevich et al. (12) hemodiluted anesthetized, but spontaneously breathing, rats to Hct values of <10% using 7% albumin and found that decreased with anemia from 57.7 to 39.1 Torr and decreased from 67 to 47 Torr, but the venoarterial difference was unchanged. The reason for the marked drop in values in their study may have been a combined result of a nonstatistically significant increase in respiratory rate (80–87 breaths/min) and fall inV˙co 2 (from 18.9 to 16.4 ml ⋅ min−1 ⋅ 100 g−1). Matsuoka et al. (31) found that acute hemodilution with lactated Ringer to an Hct of 50–60% of baseline resulted in ∼20% decreases in both V˙co 2 andV˙o 2, with a concomitant 30% increase in V˙e. Despite these changes, decreased only 3 Torr, implying a possible impairment in CO2 transport, although was not measured. This protocol was complicated by use of lactated Ringer in a one-to-one exchange with blood, which may have decreased intravascular volume and limited the circulatory response, as suggested by the fall in arterial pressure. Interestingly, when studied again 3 days after hemodilution, the anemic rats had similar mean arterial pressures, heart rates, respiratory rates, V˙co 2,V˙o 2, and compared with a group of sham-hemodiluted rats.
Analysis of Factors Preventing CO2Retention with Anemia
The important factors yielding complete compensation for anemia in maintaining CO2 elimination include an increased Q˙ and obligatory greater O2 extraction engendered by anemia that generates a larger Haldane effect. Other factors that may play a role in compensation for anemia include increases in ventilation, changes in metabolism (V˙co 2), and microcirculatory changes.
Q˙. Although several compensations exist to ensure maintenance of normalV˙o 2 andV˙co 2 during acute isovolemic anemia, the most important is an increase in Q˙.Q˙ increases in anemia predominantly because of a fall in blood viscosity, which results in a reduction in peripheral vascular resistance (9). An increased Q˙ will permit greater CO2 transfer for the same CO2 driving gradients as long as capillary recruitment prevents a fall in capillary transit time, since CO2 equilibration between blood and tissue or alveolar gas requires almost the entire normal transit time of 0.5–1.0 s for completion (6, 24, 43). Q˙increased in our animals by ∼50% with a reduction in Hct of 66%, similar to the findings of other studies in this area referenced above (average Q˙ increase over the same Hct range was 40–70%). As expected, this increase was not sufficient to prevent a progressive increase in O2extraction with greater anemia and would not be sufficient to prevent CO2 retention in the absence of other compensations. When using the human gas-exchange model of Bidani (5), it would be necessary to increase Q˙ 170% for the same Hct reduction if no other compensations, including an increase in O2 extraction, are permitted (Fig.3).
O2 extraction and the Haldane effect.
The increase in Q˙ in anemia is insufficient to prevent greater O2 extraction by the tissues. This important factor largely helps to explain the lack of impairment in CO2 elimination in our model, since the Haldane effect, i.e., that amount of CO2 taken up or released by blood with a change in hemoglobin saturation at constant ,accounts for a large fraction (40–50%) of the total CO2 exchange with normal Hct and average and differences (16, 21, 25). Because greater changes in hemoglobin saturation occur during gas exchange in anemia, there will be greater linkage of CO2 transport with O2 exchange. An augmented Haldane effect in vivo counteracts the intrinsic flattening of the CO2 dissociation curve that occurs with anemia (2, 11, 18, 36, 44) and allows greater CO2 transfer across smaller gradients.
Whether anemia and greater differences in hemoglobin saturation per se increase the absolute Haldane coefficient has never been studied, although a theoretical treatment by Grønlund and Garby (17) predicts that the Haldane coefficient would be increased by 16 and 9%, respectively, over the Hct and the hemoglobin saturation changes of our study. Increased O2 extraction alone, however, is limited in preventing CO2 retention with anemia without the accompanying increase in Q˙, and even an of 100% will not prevent a small rise in the venoarterial difference (Fig.4). This limitation of O2 extraction is also suggested by the marked CO2 retention and widening of the venoarterial difference that accompanies a decrease in Q˙ despite large increases in O2 extraction (1).
V˙ co 2 and metabolism. Although our study showed no fall inV˙co 2 with progressive anemia, two previously mentioned studies did document such a fall (12, 31). In both, there was a marked reduction in O2 delivery, which likely resulted in a depression ofV˙o 2 and a linked reduction in V˙co 2. We also observed a definite fall inV˙o 2 with the last hemodilution to a Hct of 12% but no change inV˙co 2. Without blood and tissue lactate measurements or more sophisticated measurement of tissue metabolism (nuclear magnetic resonance or near-infrared spectrocopy), we can only reasonably speculate that at the lowest Hct values an obligatory anaerobic lactic acidosis developed, as suggested by the declining pH at constant (Table2). Although it is possible that anemia could be compensated in part by a reduction in metabolism andV˙co 2, either by a parallel reduction in V˙o 2 or respiratory quotient, our study and others (50) establish that this is rare and may only occur with the most severe degree of anemia. It is conceivable that anesthesia may have modified the metabolic response to anemia in our study; however, Matsuoka et al. (31) found that metabolic rate was not depressed in anemic unanesthetized rats 3 days after hemodilution, arguing against a confounding effect of anesthesia.
Ventilation. Although hyperventilation might present a potent compensation to anemia, our data and those of Bartlett and Tenney (4), and, indirectly, those of Matsuoka et al. (31), demonstrate that hyperventilation is neither necessary nor evoked if hypovolemia and hypotension are avoided. The lack of hyperventilation with anemia is likely the effect of increasedQ˙ and O2extraction, which prevent any arterial hypercapnia and tissue CO2 retention from developing and which would stimulate peripheral or central chemoreceptors. Another factor locally acting to prevent CO2 retention in the central nervous system and any subsequent ventilatory response is the proportionately greater increase with anemia in blood flow to the brain than in the rest of the body, with the exception of the heart (50). These findings help to explain the lack of hyperventilation in anemia and any effect of anemia on ventilation and the hypercapnic ventilatory response (3, 19, 38) or the ventilatory response to exercise (28, 42). Indeed, Bartlett and Tenney (3) raise the interesting point that a hyperventilatory response to anemia would be ultimately counterproductive in that any slight increase in in the normoxic range would minimally affect O2 content, but that the ensuing respiratory alkalosis would hinder O2 unloading in the tissues.
Tissue and pulmonary gas-exchange efficiency. The question of whether tissue and pulmonary gas-exchange efficiency might improve with anemia has not been investigated to any extent. Improved gas-exchange efficiency could result simply from increases in Q˙and/or reduced viscosity acting to improve the distribution of perfusion through capillary recruitment or by reducing the intraorgan perfusion distribution heterogeneity (26, 30). In the tissues, these factors would improve local perfusion metabolism matching and in the lung would improve ventilation-perfusion matching.
Although the effect of anemia onV˙a/Q˙relationships has never been studied, both lower (50–75% reductions) and greater (100–300% increases) blood flows have been shown to cause moreV˙a/Q˙heterogeneity (13, 34). Neither study examined changes in Hct, and, in the case of increased perfusion, the complicating factor of pulmonary artery hypertension may have offset any potential benefit of increased flow per se, since lobe wet-to-dry weight ratios increased as well. The increases in Q˙ with anemia in the present study were considerably smaller than those studied by Domino et al. (13), and, in contrast, right ventricular pressures (as a marker of pulmonary artery pressure) did not rise. However, using the Bohr Vds/Vtas a crude index ofV˙a/Q˙mismatch and alveolar ventilation, we could show no improvement with anemia (in fact, alveolar ventilation decreased at the most severe degree of hemodilution). Direct studies with the multiple inert-gas-elimination technique and microsphere distributions will be necessary to address this hypothesis and could be complicated by associated changes in Q˙.
Several other effects of hemodilution at the microcirculatory level, not measurable by our traditional systemic analysis, could also act to improve gas exchange and mitigate the negative consequences of a reduced O2- and CO2-carrying capacity of anemia. First, faster flows in the microcirculation with anemia reduce the degree of precapillary diffusive O2 loss (26), and, in an analogous sense, this could also reduce an equivalent precapillary diffusive CO2 retention. Second, hemodilution reduces the normal two- to threefold difference between capillary and large-vessel Hct (26, 29, 39, 50) and the heterogeneity of Hct distribution in the capillary vasculature (26, 30). In vitro, King and Mazal (23) have shown that unevenness of Hct values leads to greater arteriovenous differences for CO2 and O2 when blood undergoes gas exchange than when bloods of equal Hct are mixed.
In conclusion, we have shown that CO2 elimination is well preserved in anemia, as long as compensatory responses (mainly an increase in Q˙ and attendant O2 extraction) are mobilized. A mathematical model incorporating our experimental hemodynamic and respiratory data predicts blood CO2 values comparable to those measured. Interestingly, our experimental data and modeling do not measure and incorporate potentially important effects of anemia in the microcirculation, which could theoretically mitigate the detrimental consequences of anemia. Whether these microcirculatory changes play a role in CO2 homeostasis during anemia, particularly when other systemic compensations are limited, is uncertain. In addition, the capacity of the organism to maintain CO2 homeostasis during anemia in the presence of cardiovascular and/or respiratory disease requires further investigation.
This project was supported by grants from the American Heart Association–Washington Affiliate (95-WA-503 to S. Deem) and the National Heart, Lung, and Blood Institute (2R01-HL-45571 to E.R. Swenson).
Address for reprint requests: S. Deem, Univ. of Washington, Dept. of Anesthesiology, Box 356540, Seattle, WA 98195-6540.
- Copyright © 1997 the American Physiological Society