The Interaction of Pulmonary Ventilation and the Right–Left Shunt on Arterial Oxygen Levels

In most fish as well as in adult mammals and birds, arterial blood gas levels closely reflect the gas composition at the gas-exchange organ, and systemic arterial blood gases can, therefore, be effectively regulated through changes in ventilation. This is not the case among reptiles, amphibians, many air-breathing fishes and embryonic mammals and birds. In these animals, the systemic and pulmonary circulations are not completely separated, resulting in the possibility of central vascular shunts. Central vascular shunts have traditionally been characterized by their direction as right-to-left shunts and left-to-right shunts (R–L shunt and L–R shunt, respectively), where an R–L shunt represents recirculation of systemic venous blood (pulmonary bypass), while an L–R shunt represents recirculation of O₂-rich blood within the pulmonary circulation (systemic bypass). In the presence of an R–L shunt, arterial systemic blood O₂ levels (⁠⁠, O₂ content and haemoglobin O₂-saturation) are depressed relative to the O₂ levels in blood returning from the lungs (chorioallantoic blood in the case of embryonic mammals and birds). Similarly, an L–R shunt results in an increase in the oxygen levels in the pulmonary arterial blood relative to that in the systemic venous blood. Arterial blood gas composition in animals possessing central vascular shunts is, accordingly, determined not only by ventilation but also by the magnitude of admixture as well as the blood gas composition of systemic venous blood. Changes in the shunt pattern, therefore, represent an alternative mechanism for arterial blood gas regulation and a means of altering systemic O₂ delivery (the product of blood flow and blood O₂ content) that is independent of ventilation and absolute levels of blood flows (Burggren, 1988; Hicks and Wood, 1989; Burggren et al. 1989).

It is the purpose of this paper to quantify and discuss the effects of the R–L shunt on arterial blood O₂ levels with the specific aim of comparing the effects of altered cardiac admixture with that of changes in pulmonary ventilation.

We have used the two-compartment model (Fig. 1) to analyze the effects of cardiac shunts on arterial O₂ levels. Pulmonary ventilation convectively transports O₂ from the environment to a single-compartment lung, where O₂ diffuses to the blood perfusing the lung. The circulatory system is characterized by the possibility of an R–L shunt, and O₂-poor blood returning from the tissues can flow to the lungs (Q̇_pul) or may directly re-enter the systemic circulation and bypass the lung altogether (Q̇_R–L). Oxygen uptake from the environment is assumed to occur exclusively at the lungs, and all O₂ consumption is assumed to occur in the tissue. At steady state, O₂ uptake in the lung is equal to tissue O₂ consumption. All calculations are based on standard equations for mass transport, which have all been included in Fig. 1. Briefly, at a given inspired ⁠, lung is determined by ventilation relative to O₂ uptake, and pulmonary venous blood is assumed to achieve a equal to minus a diffusion deficit which is considered constant at 10 mmHg. Hemoglobin (Hb) O₂-saturation of pulmonary venous blood is determined by blood O₂-binding characteristics (P₅₀ and Hill’s n_H) and pulmonary venous ⁠. In turn, pulmonary venous blood O₂ content is given as the product of blood O₂-carrying capacity ([O₂]_cap) and Hb O₂-saturation. In the presence of an R–L shunt, arterial blood is a mixture of venous systemic and pulmonary venous blood, and arterial O₂ content ([O₂]_a) is given by the weighed mean of the O₂ content from these two circuits (Berggren, 1942). Under these conditions, arterial depends on the resulting arterial Hb O₂-saturation and the O₂-binding characteristics of the blood. Finally, venous O₂ levels are determined by arterial O₂ delivery ([O₂]_a×Q̇_sys) relative to oxygen uptake ⁠.

The predictions of the present model are inherently biased by the choice of values for any of the parameters included in Fig. 1. The applied values (listed in Table 1) are based on published values for freshwater turtles (see, for example, Milsom and Chan, 1986; Wang and Hicks, 1996), but the general findings are applicable to all animals possessing a R–L shunt.

This analysis neglects the L–R shunt and, thus, assumes that the L–R shunt does not affect arterial blood gas levels. If the lung functions as a perfect gas exchanger, pulmonary venous blood gas levels would be equal to those of the lung, and recirculation of pulmonary venous blood to the lung (L–R shunt) would indeed have no effect. However, the lung is not a perfect gas exchanger and, if the difference between lung gas and capillary blood is large, recirculation of blood to the lung may elevate pulmonary venous ⁠. It has also been proposed that the L–R shunt and the associated increase in Q̇_pul improve ventilation–perfusion matching within the reptilian lung; recent experimental evidence supports this view (Hopkins et al. 1996).

In all analyses, we assume a constant deficit (equivalent of the mammalian alveolar–arterial difference) of 10 mmHg, which is similar to values determined experimentally in turtles (e.g. Burggren and Shelton, 1979). This difference is the combined result of diffusion limitation, pulmonary shunts and ventilation–perfusion inequalities (for a review on reptiles, see Wang et al. 1996) and is, therefore, unlikely to remain constant in vivo. Furthermore, our predictions are based on the condition that gas exchange is in steady state. Although this assumption is rarely (if ever) fulfilled in lower vertebrates, the predicted importance of the R–L shunt as well as the predicted trends are valid even in the absence of a steady state. Finally, several studies on reptiles have documented higher O₂ levels in the right aortic arch than in the left aortic arch (Burggren and Shelton, 1979; Ishimatsu et al. 1988) but, for simplicity, the O₂ levels in the systemic circuit are assumed to be uniform in this study. Again, this assumption does not affect the qualitative predictions of the present analysis and could be modelled using the conceptually similar three-vessel model described by Tazawa and Johansen (1987) and Ishimatsu et al. (1988).

As depicted in Fig. 1, arterial O₂ levels in the presence of an R–L shunt depend not only on the degree of the R–L shunt but also on the O₂ levels in the systemic and pulmonary venous blood. Since several of these parameters are mutually dependent, we performed several different analyses in which different parameters were varied. A summary of the four analyses (I–IV) is presented in Table 1. In all these analyses, we have assumed a fixed oxygen dissociation curve (P₅₀=25 mmHg; n_H=2.5).

Fig. 2 shows the effects of increasing the R–L shunt at absolute levels of Q̇_pul and Q̇_sys during normoxia and at a fixed level of ventilation. At any value of Q̇_sys, an increase in the R–L shunt (by a reduction in Q̇_pul) reduces both arterial and venous Hb O₂-saturation and the accompanying values. As the rate of oxygen uptake was constant in this analysis, the O₂ content difference between arterial and venous blood increases in proportion to the reduction in systemic blood flow. In all the following analyses, Q̇_sys was kept constant at 60 ml kg⁻¹ min⁻¹. At least in turtles, this premise seems to be valid, whereas other animals may alter Q̇_sys (e.g. Lillywhite and Donald, 1989).

At constant and Q̇_sys, a reduction in blood O₂-binding capacity lowers venous Hb O₂-saturation. In the absence of an R–L shunt, this reduction does not affect arterial blood gas levels, which in this case are entirely determined by lung gas composition. However, in the presence of an R–L shunt, a reduction in venous Hb O₂-saturation proportionally reduce arterial Hb O₂-saturation and, thus, arterial ⁠. These relationships are presented in Fig. 3 in the absence of an R–L shunt (circles), at a Q̇_R–L of 20 ml kg⁻¹ min⁻¹ (triangles) and at a Q̇_R–L of 40 ml kg⁻¹ min⁻¹ (squares) and at two rates of ventilation (20 and 100 ml kg⁻¹ min⁻¹, represented by filled and open symbols, respectively). Note that an increased rate of ventilation has only a small effect on arterial Hb O₂-saturation and that only in the absence of an R–L shunt is the effect on arterial sizable.

In this analysis, the effect of increasing metabolic rate was assessed at two rates of ventilation and three levels of R–L shunt (Table 1; Fig. 4). At constant ventilation, increasing linearly decreased lung ⁠, which, in the absence of an R–L shunt, resulted in a corresponding decrease in (circles in Fig. 4A). In the presence of an R–L shunt, increasing decreased arterial O₂ levels because of the reduction in venous O₂ content. This relationship is depicted for two levels of R–L shunt (triangles and squares) and at two rates of ventilation (open and filled symbols) in Fig. 4A,B. Note that the highest levels of can only be sustained if ventilation is high and the R–L shunt is eliminated.

The effects of increasing ventilation, eliminating the R–L shunt and the combination of both are presented in Fig. 4C,D, using a ventilation of 20 ml kg⁻¹ min⁻¹ and a Q̇_pul/Q̇_sys of 40/60 as reference (open triangles in Fig. 4A,B). The increases in and Hb O₂-saturation obtained by eliminating the R–L shunt, but maintaining ventilation constant, are depicted with open triangles, while the effects of increasing ventilation to 100 ml kg⁻¹ min⁻¹, but maintaining the R–L shunt constant, are depicted with open circles (Fig. 4C,D). The filled squares represent the effect of both increasing ventilation and eliminating the R–L shunt. At low ⁠, increasing ventilation only modestly affected and Hb O₂-saturation, while elimination of the R–L shunt had a pronounced effect on ⁠. In contrast, at higher ⁠, an increase in ventilation had a larger effect on arterial oxygen levels than elimination of the R–L shunt. For any given condition, the combined effect of increasing ventilation and reducing the R–L shunt is larger than the effect of changing only ventilation or only the R–L shunt.

The effects of decreasing the inspired O₂ fraction were determined at two rates of ventilation and three levels of R–L shunt (Fig. 5). At constant ventilation, a reduction in linearly reduces of pulmonary venous blood which, in the absence of an R–L shunt, is mirrored by a similar decrease in (open and filled circles, respectively, in Fig. 5A,B). In the presence of an R–L shunt, this reduction in is less pronounced (triangles and squares in Fig. 5A,B).

Fig. 5C,D shows the effects of increasing ventilation (open circles), eliminating the R–L shunt (open triangles) or a combination of both (filled squares) using a ventilation of 20 ml kg⁻¹ min⁻¹ and a Q̇_pul/Q̇_sys of 40/60 as reference (open triangles in Fig. 5A,B). At normoxia ⁠, the effect of increasing ventilation is virtually absent while the effect of eliminating the R–L shunt is maximal. As is lowered, the effect of increasing ventilation increases while the effect of eliminating the R–L shunt becomes progressively less pronounced. As in the previous analysis, the combined effect of changing both ventilation and the R–L shunt is always larger than that achieved by changing only one parameter.

Our analyses emphasize that both ventilation and the R–L shunt are important determinants of arterial O₂ levels. This observation is by no means novel, and the importance of the R–L shunt in determining arterial blood gas levels has long been recognized within both clinical and comparative physiology (see, for example, Rossoff et al. 1980; Wood, 1982, 1984). Nevertheless, most existing models for gas exchange in amphibians and reptiles have either ignored intracardiac shunts or assumed the R–L shunt to be fixed at a constant value (e.g. Boutilier and Shelton, 1986; Withers and Hillman, 1988). In addition, the principal focus of the present study was to quantify and compare the effects of altering ventilation and the R–L shunt within physiologically realistic limits. Such quantification has, at least to our knowledge, not been conducted prior to this study.

This study demonstrates that the effect of changing ventilation and the R–L shunt on arterial O₂ levels varies with parameters such as inspired ⁠, the rate of O₂ uptake and blood O₂-carrying capacity. These changes in the relative importance of altering ventilation and the cardiac R–L shunt in determining arterial blood gas levels are largely explained by the sigmoidal shape of the O₂ dissociation curve (ODC). If lung is high relative to blood O₂ affinity, the O₂ levels of blood leaving the lung are positioned on the flat portion of the ODC and even a large increase in lung (by increasing ventilation) has little effect on pulmonary venous O₂ content. If an R–L shunt is taking place, this implies that arterial O₂ content is affected only marginally, with a correspondingly small effect on arterial ⁠. In contrast, if the O₂ levels of pulmonary venous blood are positioned on the step portion of the ODC, a small change in lung affects pulmonary venous O₂ content markedly, and the resulting impact on arterial O₂ levels is accordingly more pronounced. This explains why, in the presence of an R–L shunt, an increase in ventilation from 20 to 100 ml kg⁻¹ min⁻¹ has only small effects on arterial O₂ levels at normoxia (Figs 3–5), whereas the impact of increasing ventilation is large when lung is low. The converse argument explains the effect of the R–L shunt on arterial O₂ levels. As presented in Fig. 5D, the impact of eliminating the R–L shunt on arterial Hb O₂-saturation is independent of the level of hypoxia. Nevertheless, the corresponding change in due to elimination of the R–L shunt is large (Fig. 5C) because the change in Hb O₂-saturation occurs on an increasingly steep portion of the ODC. The same arguments hold true in the analysis where the rate of oxygen uptake is altered (Fig. 4), although this situation is more complicated because the change in arterial Hb O₂-saturation for a given level of R–L shunt depends on the rate of oxygen uptake.

Because the effects of changing ventilation and/or the R–L shunt on arterial oxygen levels are determined by lung relative to blood oxygen affinity, the most ‘beneficial’ cardiorespiratory response (in terms of improving O₂ transport) depends on the conditions of oxygen loading at the lungs. If lung is high relative to blood O₂ affinity, a change in ventilation will have virtually no effect, while a decrease in the R–L shunt is relatively more powerful. Conversely, if lung is low relative to blood O₂ affinity, the effect of changing ventilation is larger. Importantly, simultaneously increasing the ventilation and eliminating the R–L shunt always results in the largest increases in blood O₂ levels.

The R–L shunt invariably results in a decrease in arterial oxygen levels and, at a given Q̇_sys, the R–L shunt therefore reduces the O₂ transport capacity of the cardiovascular system. As a result of this reduction, the presence of an R–L shunt reduces the maximum sustainable rate of O₂ consumption (Fig. 4) and reduces the tolerance to hypoxia (Fig. 5) and to reductions in blood O₂-carrying capacity (Fig. 3). Teleologically, it may therefore be argued that it is beneficial for gas exchange for the R–L shunt to be eliminated under circumstances where O₂ transport is challenged (hypoxia and hypoxaemia) or when metabolic rate is increased (during digestion and muscular exercise).

Detailed descriptions of cardiovascular responses under conditions other than rest are, unfortunately, scarce. In fact, not a single study has determined total systemic and pulmonary blood flows simultaneously during hypoxia, hypoxaemia or exercise in any animal possessing the possibility of cardiac shunting. This lack of simultaneous determinations of blood flows makes it difficult to assess the changes in intracardiac shunt in detail. Nevertheless, some studies have determined Q̇_pul in addition to blood flows in one or more systemic vessel, and the results from these studies seem to support our suggestion that the R–L shunt is eliminated, or at least reduced, during activity and hypoxia.

In the green sea turtle Chelonia mydas, Q̇_pul increases proportionately more than left aortic blood flow during swimming (West et al. 1992), which strongly suggests a reduction in the R–L shunt. Large increases in Q̇_pul compared with Q̇_sys and the development of a net L–R shunt during terrestrial exercise have also been noted by Burggren and Shelton (1979) for Testudo graeca and Pseudemys scripta, and large increases in Q̇_pul have been reported during movements under water in the turtle Chelonia mydas (Johansen et al. 1970). Detailed determinations of cardiac shunt patterns during different forms of activity are clearly needed, and studies on non-chelonian reptiles would be most welcome.

Compared with the considerable information on ventilatory responses to reductions in inspired oxygen levels, surprisingly few studies have investigated the effects of hypoxic hypoxia on central vascular blood flows. Nevertheless, in the aforementioned study on the green sea turtle, West et al. (1992) found a twofold increase in Q̇_pul at moderate hypoxia (10 % O₂), while left aortic blood flow increased by only 30 %. Similarly, Burggren et al. (1977) reported large increases in Q̇_pul during hypoxia in two other species of chelonians, but did not determine Q̇_sys. In the Australian lungfish Neoceratodus forsteri, Q̇_pul increases two-to threefold during aquatic hypoxia (Fritsche et al. 1993) and, finally, hypoxia elicits increases in Q̇_pul in anaesthetized toads Bufo marinus (West and Burggren, 1984).

It has long been recognized that animals with low blood O₂-carrying capacity possesses high Q̇_sys (Lenfant et al. 1970), but the effects of reducing blood O₂-carrying capacity on cardiac shunting have received virtually no attention. Nevertheless, the possible role of arterial O₂ content as opposed to as a regulated variable for ventilation has been discussed repeatedly (Wang et al. 1994). While this is an interesting question, our present analyses clearly show that an increased ventilation does not improve or safeguard O₂ transport in anaemic animals. Rather, an elimination of the R–L shunt and an increase in Q̇_sys are beneficial under these circumstances and future studies on the role of reduced blood O₂-carrying capacity should consequently address both the cardiovascular and the ventilatory events.

The potential role of the cardiovascular system in regulating arterial blood gas levels through changes in R–L shunting is undisputed. The existence of such potential, nevertheless, does not warrant the conclusion that the cardiovascular system is involved in this regulation. Chemoreceptors located in the arterial circulation and on the pulmocutaneous arches have been identified and recorded from several species of amphibians and reptiles. These chemoreceptors, like the peripheral chemoreceptors in mammals, increase their firing frequency during hypoxia (Ishii et al. 1985; Van Vliet and West, 1992). The afferent input from the receptors has mostly been studied from the viewpoint of ventilatory control, and its possible effects on the cardiovascular system are largely unknown.

Increases in Q̇_pul and the concomitant reductions in the R–L shunt are often associated with changes in ventilation. It can therefore be difficult to establish whether the cardiovascular performance changes independently or whether it is merely a function of the normal cardiorespiratory interaction. In several studies, Burggren and coworkers observed that turtles, while submerged and inactive, periodically increase Q̇_pul (Burggren and Shelton, 1979; Burggren, 1988; Burggren et al. 1989). It was therefore suggested that these transient perfusions of the lung serve to regulate arterial oxygenation, but it is imperative to emphasize that the involvement of arterial chemoreceptors and the appropriate afferent control of pulmonary perfusion remain to be demonstrated.

As discussed above and originally by Wood (1982, 1984), arterial blood gas levels in the presence of an R–L shunt are dependent variables determined by all the parameters included in Fig. 1. As an example, increases with increased temperature in virtually all amphibians and reptiles, and this pattern is commonly interpreted as being primarily a result of decreased blood oxygen-affinity at elevated temperature (e.g. Glass et al. 1985). While these interpretations may provide an adequate explanation, alternatives exist, as illustrated by blood gas data from one of our own studies on toads (Branco et al. 1993). In these experiments, ventilation was stimulated by perfusing the central chemoreceptor with a mock cerebrospinal fluid solution of varying acidity while arterial blood gas levels were monitored. The increased ventilation resulted in a respiratory alkalosis (0.3 pH units at 35 °C) which, according to the interpretations above, should result in a decrease in because of the expected increase in blood O₂-affinity. In contrast to this ‘expectation’, increased significantly. This finding can be explained by a decrease in the R–L shunt, an increase in Q̇_sys and the resulting increase in mixed venous O₂ content, an increase in left atrial oxygenation level, a decrease in the rate of O₂ uptake or any possible combination of all these. Naturally, an evaluation of these possibilities requires that all, or at least most, of these parameters be determined. The point we would like stress is that, in the absence of detailed and simultaneous determinations of blood flows and blood gas levels from several central sites, interpretations of the causes of changing arterial blood gas levels are, at best, tenuous.

The modelling of arterial O₂ levels in the present study unequivocally demonstrates the importance of intracardiac shunts for the determination of blood gas levels in lower vertebrates (and all embryonic vertebrates). It is, therefore, erroneous to view the control of arterial blood gas levels solely as a function of ventilation. Rather, arterial blood gas composition is the combined result of ventilation and cardiac shunts, and both the cardiovascular and the ventilatory systems must accordingly be evaluated if we aim to understand the control of arterial blood gas levels. Currently, only limited experimental data exist on cardiovascular performance in lower vertebrates and detailed studies on cardiac shunts are very scarce. Thus, although it is premature to conclude that the regulation of cardiac shunts is directly involved in the control of arterial blood gas levels, no data indicate the opposite. Clearly, future studies on lower vertebrates must combine measurements of cardiac shunts and ventilation.

T.W. was supported by a postdoctoral fellowship from the Carlsberg Foundation (Denmark) during the course of this study and J.W.H. received financial support from an NSF operating grant (IBN-9218936). We thank Dr S. C. Wood for critical comments on a previous version of this manuscript and Ole Wang for arithmetic advice.