Lungs evolved to increase diffusing capacity by compartmentalizing and reducing the size of individual gas exchange units. This increased the potential for gas exchange limitations from ventilation-perfusion heterogeneity. However, comparative studies on reptiles, birds, and mammals show that heterogeneity is independent of lung complexity.
In the animal kingdom, evolution of the respiratory system has produced an amazing variety of structures to satisfy a wide range of oxygen demands. Vertebrate lungs include simple, unicameral, sack-like structures in some amphibians and reptiles as well as complex branching structures like the alveolar lungs of mammals and parabronchial lungs of birds. However, pulmonary gas exchange in all vertebrates can be represented by a general model with O2 diffusion across a blood-gas barrier in series between the two convective transport processes of ventilation (V̇ with air or water) and blood flow (Q̇, perfusion). Animals with higher metabolic rates and greater O2 demand have more complex lungs with smaller gas exchange units. Smaller gas exchange units have greater surface-to-volume ratios, increasing the diffusing capacity for O2.
One potential problem with subdividing the lung is maintaining efficient matching of ventilation and perfusion in hundreds to millions of functional gas exchange units. It is well known that ventilation-perfusion (V̇/Q̇) heterogeneity is the most important factor preventing O2 exchange from occurring at ideal levels in resting mammals (19). Hence, the benefits of subdividing the lungs to increase diffusing capacity must be balanced against the potential costs of decreased gas exchange efficiency from V̇/Q̇ heterogeneity. Recent measurements of V̇/Q̇ heterogeneity in a wide variety of animals with very different lung structures provides some insights into these cost-benefit relations (8). The results suggest that there is a general plan for bronchial and vascular branching that results in similar V̇/Q̇ heterogeneity in lungs with a variety of designs.
Lung structure-function in birds and mammals
Both mammals and birds have the highest metabolic rates and the most complex lungs of the vertebrates. However, mammals and birds have evolved very different respiratory systems (Fig. 1⇓). The mammalian lung is homogenously partitioned, and it is structurally uniform compared with the lungs of other vertebrates (2). In mammals, the functions of ventilation and gas exchange are shared by common structures in the respiratory bronchioles, alveolar ducts, and hundreds of millions of alveoli. Less than 20% of the total lung volume is dedicated to ventilation alone in the conducting airways. In contrast, the avian respiratory system is heterogeneously partitioned and completely separates the functions of ventilation and gas exchange (2). In birds, nine air sacs act as bellows to ventilate the small, constant volume. The air sacs occupy ~90% of the total respiratory system volume in a bird; the remaining 10% is comprised of the lung, containing hundred of gas exchange units called parabronchi.
These structural differences between alveolar and parabronchial lungs result in different models of gas exchange in birds and mammals. The gas-exchanging parabronchi in avian lungs are arranged in parallel and are connected at both ends to secondary bronchi, which act as conducting airways that ventilate the parabronchi with air from the trachea or air sacs. The parabronchi are perfused along their entire length by pulmonary mixed-venous blood, so ventilation and perfusion can be thought of as occurring at right angles to one another and gas exchange in a bird lung is described by a cross-current model (14). The theoretical efficiency of cross-current gas exchange is greater than alveolar exchange, and under ideal conditions arterial po2 (Pao2) exceeds end-parabronchial (or expired) po2. In contrast, ideal alveolar gas exchange in mammals can only result in Pao2 equaling expired (i.e., alveolar) values.
V̇/Q̇ heterogeneity in birds and mammals
V̇/Q̇ heterogeneity can be quantified with V̇/Q̇ distributions measured by a method known as the multiple inert gas elimination technique, or MIGET (19). The details of the MIGET are beyond the scope of this brief review, but the V̇/Q̇ distributions determined by this method provide important insights into comparative gas exchange. The method considers the lung as if it were comprised of 50 individual gas exchange units with different V̇/Q̇ ratios equally spaced on a logarithmic scale (Fig. 2⇓). V̇/Q̇ ratios of the individual compartments range from 0, representing blood flow without ventilation (shunt), to >100, representing ventilation without significant blood flow (dead space ventilation).
Computer algorithms are used to partition total ventilation and cardiac output between the 50 V̇/Q̇ compartments to minimize error between data predicted from the modeled V̇/Q̇ distributions and experimentally measured data (19). Experimental input data include ventilation and expired, arterial, and mixed-venous levels of inert gases that are infused in a peripheral vein in trace amounts. Hence, the method can be used to measure V̇/Q̇ distributions under a variety of conditions, including exercise. Although the method was designed originally to measure V̇/Q̇ distributions in alveolar lungs, it can be modified to measure V̇/Q̇ distributions in avian lungs by incorporating the cross-current model of gas exchange in the algorithms. It can be used to measure V̇/Q̇ distributions in reptiles also, with the same algorithms developed for alveolar gas exchange in mammals (5). Heterogeneity can be quantified from these V̇/Q̇ distributions as the log standard deviation of the Q̇ distribution among the 50 units (log SDQ̇), where a higher log SDQ̇ indicates greater V̇/Q̇ heterogeneity.
Representative V̇/Q̇ distributions obtained with this method are shown in Fig. 2⇑. The V̇/Q̇ distributions in resting birds and mammals are remarkably similar despite large differences in lung structure. Furthermore, these V̇/Q̇ distributions are similar to those in most reptiles as discussed in the next section. This is in contrast to, for example, an alveolar lung with chronic obstructive pulmonary disease, which will have bimodal distributions of V̇ or Q̇ vs. V̇/Q̇ and a log SDQ̇ of 2.0 (8). Hence, it appears that the most complex vertebrate lungs have evolved with amounts of V̇/Q̇ heterogeneity despite very different structure-function relationships.
V̇/Q̇ heterogeneity in reptiles
Reptiles form an interesting intermediate between birds and mammals in terms of lung structure. Reptilian lung types can be divided into single-chambered (unicameral), few-chambered (paucicameral), and the more complex many-chambered (multicameral) lungs (2) (Fig. 3⇓). Reptiles typically have much lower metabolic rates (~10-fold) compared with birds and mammals of the same size, but they show wide variability in O2 demand and their lung structure has evolved accordingly. For example, the varanid lizards, which are known for their relatively high metabolic scopes and active behavior patterns in the wild, have relatively complex multicameral lungs (12). By contrast, the tegu lizard has lower O2 demands and its unicameral lung has a large central lumen lined with gas-exchanging falveoli resembling a layer of honeycomb.
Despite this variation in lung structure and complexity among this diverse order, Fig. 2⇑ shows that V̇/Q̇ heterogeneity is similar in all of the reptiles studied to date. Surprisingly, the amount of heterogeneity tends to be greater in the simpler lungs, as reflected in the larger log SDV̇ for the tegu and turtle. Apparently, nonuniform blood flow in tegu lungs leads to V̇/Q̇ heterogeneity because gas mixing is not perfect in the unicameral lung. New results from awake alligators (Fig. 3⇑) show less V̇/Q̇ heterogeneity than previously published results from anesthetized animals (8), presumably because cardiac output was decreased by anesthesia. Increasing Q̇ is reported to decrease V̇/Q̇ heterogeneity in turtles (10).
Summarizing, reptilian V̇/Q̇ distributions are, in general, unimodal and relatively narrow, for example, compared with human lungs with disease as described above. Hence, it appears that all healthy vertebrate lungs studied to date have evolved to have a similar amount of V̇/Q̇ heterogeneity independent of the degree of subdivision and complexity.
Effects of V̇/Q̇ heterogeneity on resting O2 exchange
As introduced above, ideal gas exchange in alveolar lungs results in O2 equilibration between alveolar gas and arterial blood. Hence, a useful index of gas exchange efficiency in mammals is the alveolar-arterial po2 difference (often referred to as AaDo2), which is 0 in the ideal case. However, alveolar-arterial po2 averages between 4 and 10 Torr in resting mammals, and this can be explained in large part by V̇/Q̇ heterogeneity (19). The effects of V̇/Q̇ heterogeneity and intra-pulmonary shunting on O2 exchange can be quantified by comparing measured Pao2 with the pao2 predicted for a measured V̇/Q̇ distribution (8). If the measured pao2 is less than that predicted for a measured V̇/Q̇ distribution, then another gas exchange limitation must be present, such as a diffusion limitation or postpulmonary shunt.
In humans, dogs, pigs, and horses under resting normoxic conditions, there is no significant difference between the measured Pao2 and the value predicted by the measured V̇/Q̇ distributions (8, 9). Hence, V̇/Q̇ heterogeneity is the most important factor limiting O2 exchange under such conditions. Diffusion and shunts are not significant. Diffusing capacity in humans is about twice as large as is necessary for O2 equilibration under resting normoxic conditions, and physiological shunting, or venous admixture, is very small in most mammalian species. Postpulmonary shunts, such as from the bronchial circulation and Thebesian veins, are <1% of the total cardiac output and decrease Pao2 only slightly; true intrapulmonary shunting from pulmonary arteries to pulmonary veins is so small that often it cannot be measured (8).
The anatomic basis of V̇/Q̇ heterogeneity in resting mammals includes both 1) interregional differences, for example between different heights in an upright lung from the effects of gravity on blood flow and ventilation, and 2) intraregional heterogeneity, which occurs between individual acini independent of gravity (20). The relative contributions of these two types of heterogeneity can be estimated by comparing V̇/Q̇ distributions measured with MIGET, which detects both types of heterogeneity, and earlier measurements of V̇/Q̇ distributions using radioactive gases, which detect relatively coarse interregional differences. Earlier measurements of V̇/Q̇ distributions made in upright human lungs with radioactive gases show about half as much heterogeneity as V̇/Q̇ distributions determined with MIGET (19). Hence, interregional and intraregional V̇/Q̇ heterogeneity are similar in magnitude and make approximately equal contributions to reducing measured Pao2 from ideal levels at rest.
In birds, the same general approach can be used to quantify the effects of V̇/Q̇ heterogeneity on O2 exchange, although there are some important differences from the mammalian case. First, birds have parabronchi, not alveoli, so it is not accurate to refer to an alveolar-arterial Po2 difference in birds. Hence, we will refer to the expired-arterial Po2 difference, which is applicable to all animals, including birds with parabronchi, reptiles with falveoli, and mammals with alveoli. Second, in cross-current gas exchange the ideal expired-arterial Po2 difference is not always zero and can achieve negative values, in contrast to the ideal alveolar-arterial Po2 (14). In fact, negative expired-arterial Po2 values can occur in cross-current gas exchange with nonideal conditions such as with V̇/Q̇ heterogeneity. For example, the expired-arterial Po2 difference is predicted to be −10 in anesthetized ducks and geese on the basis of measured V̇/Q̇ distributions such as those in Fig. 2⇑ (13).
Comparing measured Pao2 in ducks and geese with the values predicted from measured V̇/Q̇ distributions reveals much larger effects of V̇/Q̇ heterogeneity on O2 exchange in birds compared with mammals (13). Pao2 predicted for measured V̇/Q̇ distributions in water fowl is 15 Torr lower than the ideal (homogeneous) level for cross-current gas exchange. However, the measured Pao2 value is another 10 Torr lower, or 25 Torr lower than the ideal cross-current value. Hence, only 60% of the reduction in Pao2 from ideal levels in birds can be explained by V̇/Q̇ heterogeneity. The remainder is explained by postpulmonary shunts, which can average 5% of cardiac output in birds under these conditions. This is much larger than postpulmonary shunts in mammals and is apparently explained by vertebral-pulmonary venous circulatory connections (1). Physiological estimates of O2-diffusing capacity corrected for V̇/Q̇ heterogeneity agree well with morphometric estimates, and diffusion limitations are not predicted for resting normoxic birds (14).
The advantage of cross-current gas exchange in a bird lung becomes apparent in hypoxia and is evidenced by a negative value for the measured expired-arterial Po2 (13, 14). V̇/Q̇ heterogeneity is the same in normoxia and hypoxia, but it affects expired-arterial Po2 values less in hypoxia because gas exchange takes place on the steep part of the O2-blood dissociation curve. Also, the postpulmonary shunt described above is not observed in hypoxia. In anesthetized ducks under hypoxic conditions, the difference between measured and ideal Pao2 is 4 Torr but only 1 Torr is explained by V̇/Q̇ heterogeneity. The remainder can be explained by a diffusion limitation in hypoxia, which is consistent with best estimates of O2-diffusing capacity in birds (3).
The same principles can be applied in reptiles, but the situation is complicated by intracardiac shunts with three-chambered hearts or connections between left and right aortas in the case of crocodilians with four-chambered hearts (4). In reptiles, these shunts are more important than pulmonary limitations in terms of reducing Pao2, and it is necessary to sample pulmonary venous blood in the left atrium instead of systemic arterial blood to assess pulmonary gas exchange. Hence, analysis of reptilian gas exchange requires consideration of the expired-left atrial Po2 difference.
In varanid lizards, the expired-left atrial Po2 difference is 14–28 Torr at rest and is explained almost completely by V̇/Q̇ heterogeneity and intrapulmonary shunting (8). In contrast to birds and mammals, intrapulmonary shunts can exceed 5% of total pulmonary blood flow in lizards. However, this is still less than postpulmonary right-to-left cardiac shunts that comprise ~9% of cardiac output in lizards. There is no evidence for O2 diffusion limitation in resting varanids, but these studies have not been conducted in hypoxia.
Turtles have a larger expired-arterial Po2 difference, which varies with ventilatory state and pulmonary blood flow. The difference is ~33 Torr in anaesthetized turtles, with about half of the difference explained by V̇/Q̇ heterogeneity and half by a 17% intrapulmonary shunt (10). Also, turtles may have some diffusion limitation for O2 at rest, but rigorous determination of this requires a more precise blood-O2 dissociation curve to accurately predict Pao2 from measured V̇/Q̇ distributions. The turtle has multiple hemoglobins (11), so it is difficult to model the dissociation curve with the available algorithms.
Summarizing, Pao2 is relatively similar in birds and mammals, despite birds having theoretically more efficient cross-current gas exchange. This is because a similar amount of V̇/Q̇ heterogeneity in birds and mammals has a relatively greater impact on O2 exchange in theoretically more efficient cross-current lungs (14). The effects of V̇/Q̇ heterogeneity on O2 exchange in reptiles is similar to that in birds and mammals, although Pao2 is generally lower in reptiles because of postpulmonary shunts, larger intrapulmonary shunts, and diffusion limitations. Hence, complexity of lung structure is not tightly correlated with Pao2, and this suggests that other factors have a dominant effect on the evolution of branching patterns and complexity in vertebrate lungs.
Effects of exercise on heterogeneity and O2 exchange
Exercise reduces the efficiency of O2 exchange in most mammals, as evidenced by increases in the alveolar-arterial Po2 difference and, in some cases, reductions in Pao2 (8). Figure 4⇓ shows V̇/Q̇ heterogeneity measured during rest and exercise in mammals, reptiles, and birds. Both V̇/Q̇ heterogeneity and diffusion limitation contribute to increased alveolar-arterial Po2, but there are species differences for the relative contribution of these two factors. For example, horses show very little increase in V̇/Q̇ heterogeneity with exercise, but pulmonary diffusion limitation, combined with mechanical constraints on ventilation, cause marked arterial hypoxemia (16). On the other hand, pigs significantly increase V̇/Q̇ heterogeneity with exercise but show no appreciable diffusion limitation (9). O2 exchange in exercising humans is impacted by both V̇/Q̇ heterogeneity and diffusion limitation, although the relative contribution varies between individuals and with aerobic capacity (7). In elite athletes, V̇/Q̇ heterogeneity may be responsible for 60% of the alveolar-arterial Po2 difference, which can become as large as 40 Torr (7)!
Exercise also increases V̇/Q̇ heterogeneity and diffusion limitations in reptiles. In varanid lizards, log SDV̇ increases significantly from 0.39 at rest to 0.78 during maximal exercise (6). V̇/Q̇ distributions are very similar in mammals and these relatively aerobic reptiles, with the main difference being a modest intrapulmonary shunt (2% of cardiac output) in the varanids. Varanids are also similar to very fit mammals (e.g., Olympic runners and race horses) by having a significant diffusion limitation during normoxic exercise, which explains one-third of the 30-Torr measured expired-arterial Po2.
In contrast to mammals and reptiles, exercise did not increase V̇/Q̇ heterogeneity in the only study to date measuring V̇/Q̇ distributions in exercising birds (15). V̇/Q̇ distributions were similar in emus resting or running on a treadmill, in normoxia or hypoxia (log SDV̇ = 0.60–0.68). It may be premature to generalize this conclusion to birds because the emus were not exercising near maximal V̇o2, in contrast to the mammalian studies reviewed above. However, there are other reasons to expect that measured V̇/Q̇ heterogeneity is more stable during exercise in birds than in mammals. For example, physiological estimates of V̇/Q̇ heterogeneity are decreased in anesthetized ducks by elevated metabolism or cardiac output (3). Specifics of the pulmonary circulation in the parabronchial lung may also prevent V̇/Q̇ heterogeneity, as discussed below.
We cannot quantify the effects of V̇/Q̇ heterogeneity vs. diffusion limitation on O2 exchange during exercise in birds yet, because both O2-blood equilibrium curves and measured V̇/Q̇ are not available for a single species. However, Pao2 does not appear to decrease in birds approaching maximum V̇o2 as it does for mammals and reptiles (8). Physiological estimates of lung diffusing capacity for O2 in birds are increased by exercise (14), and this would tend to maintain Pao2 during exercise. Also, birds hyperventilate during exercise, which is predicted to increase the ideal Pao2 and make the ideal expired-arterial Po2 difference more negative. Hence, the rise in measured Pao2 needs to be compared with the rise in Pao2 predicted for a given V̇/Q̇ distribution, to quantify gas exchange limitations properly.
Physiological basis for V̇/Q̇ heterogeneity and diffusion limitation
The physiological basis for increased V̇/Q̇ heterogeneity with exercise is not certain, but a likely cause is interstitial pulmonary edema (8). Increased pulmonary vascular pressure with exercise reduces gravity-dependent differences in blood flow at different heights in a lung, but it also increases capillary filtration. Interstitial edema fluid would be expected to distort the surrounding architecture of the alveoli and capillary network. Altered airway and blood vessel diameter resulting from the presence of cuffing would affect distribution of blood flow, and perhaps air flow, in the lung. There is less evidence for increased heterogeneity of ventilation than for perfusion during exercise. However, a reduction of gas mixing in large airways during exercise would increase apparent V̇/Q̇ heterogeneity measured by MIGET (17).
Several lines of evidence implicate interstitial edema as the cause of increased V̇/Q̇ heterogeneity in mammals (8). In humans, V̇/Q̇ heterogeneity is increased by hypoxia and decreased by breathing 100% O2, which would tend to increase and decrease, respectively, capillary pressure and filtration. Increased pulmonary artery pressure during exercise would also increase capillary filtration pressure and may have similar effects on the distribution of blood flow. Subjects who have a history of high-altitude pulmonary edema (HAPE) also show larger increases in exercise-induced V̇/Q̇ heterogeneity compared with subjects without a history of HAPE. This has led to speculation that the underlying mechanism of increased V̇/Q̇ heterogeneity with exercise and HAPE may be the same.
It is possible that similar changes in the distribution of blood flow occur in the avian lung during exercise but without being detected as increased V̇/Q̇ heterogeneity by MIGET. This technique measures parallel V̇/Q̇ heterogeneity between different parabronchi (interparabronchial V̇/Q̇ heterogeneity) but not serial inequality of blood flow along a single parabronchus (intraparabronchial V̇/Q̇ heterogeneity) (14). Interstitial edema may not affect the larger interparabronchial arteries that control the distribution of blood flow between parabronchi. Interstitial edema would be more likely to impact blood flow in the smaller intraparabronchial arteries and alter the serial distribution of blood flow along a parabronchus, which does impair O2 exchange (14). Additional studies on the distribution of blood flow during rest vs. exercise (e.g., with microspheres) could test this hypothesis in birds.
Interstitial edema during exercise could also increase diffusion impairment for O2. Diffusion limitation during heavy exercise is likely on the basis of rapid pulmonary capillary transit of red blood cells (7). However, diffusion limitation from thickening of the blood-gas barrier is unlikely. After prolonged, severe exercise in humans, the diffusing capacity for carbon monoxide (DlCO) decreases, but there is no decrease in estimates of the O2 diffusing capacity estimated from the difference between measured Pao2 and Pao2 predicted for measured V̇/Q̇ distributions (8). The decrease in DlCO after exercise is best explained by a decrease in pulmonary capillary volume with a redistribution of central blood flow. There may also be a change in the matching of blood flow and diffusing capacity, which impacts gas exchange similar to V̇/Q̇ heterogeneity (3).
There is some evidence that diffusive-perfusive heterogeneity actually decreases with exercise in birds. Physiological estimates of O2 diffusing capacity in ducks are increased by either treadmill exercise or pharmacological increases in metabolic rate (8). However, increases in diffusing capacity from recruitment and distension of blood capillaries are expected to be minimal in the constant-volume parabronchial lung (3). Hence, decreases in diffusive-perfusive heterogeneity are the simplest explanation for the apparent decrease in diffusion impairment in exercising birds.
V̇/Q̇ heterogeneity is remarkably constant across a wide variety of lung designs in vertebrates. It is similar in mammals and reptiles despite the fact that O2 transport efficiency is dominated by cardiac shuts under most circumstances in reptiles and by pulmonary factors in mammals. Also, V̇/Q̇ heterogeneity is similar in birds and mammals with entirely different respiratory system designs and models of gas exchange. This suggests that a general vertebrate plan for vascular branching in lungs may be responsible for similar degrees of V̇/Q̇ heterogeneity in different animals (8). This idea has been formalized recently in a mathematical model of transportation networks in living organisms such as vascular and bronchial trees (18). The model minimizes the energy of transport in a fractal, branching network of tubes that fill the space of an organ and successfully predicts the 0.75 power scaling for physiological functions vs. body mass that is observed in nature.
In contrast to the similarity of V̇/Q̇ distributions between animals, the effects of V̇/Q̇ heterogeneity on O2 exchange varies. It has a large impact on O2 exchange in birds but much less of an effect in mammals and some reptiles. Furthermore, some animals show increased heterogeneity with increased O2 demand during exercise and some do not. However, increased diffusion limitation for O2 is a more uniform finding, and this occurs despite increases in O2 diffusing capacity with exercise in all animals. Diffusion limitations for O2 remain to be quantified in birds during exercise, but available results suggest that diffusing capacity increases in exercising birds, as would be expected to overcome potential diffusion limitations. These results suggest that the evolutionary pressures to increase diffusing capacity by partitioning the lung into smaller gas exchange units outweighed the potential cost of increased heterogeneity between gas exchange units in healthy lungs.
- © 2004 Int. Union Physiol. Sci./Am.Physiol. Soc.