For many years, it was generally believed that the pulmonary system did not pose a significant barrier to exercise performance and, in particular, that pulmonary gas exchange did not deteriorate with exercise. Gender differences in the pulmonary system’s response to exercise, if any, were unknown. A reduction in the arterial partial pressure of oxygen (PaO2) during exercise was reported in the early twentieth century in a healthy normal subject, and sporadically thereafter. However, the topic did not attract widespread attention until Dempsey et al. (1) reported arterial blood gases in 16 male athletes during treadmill running. In this study, 12 subjects experienced exercise-induced arterial hypoxemia (EIAH) characterized by a decrease in PaO2 sufficient to impair O2 transport, a marked increase in the alveolar-arterial oxygen pressure difference (AaDO2), with little accompanying alveolar hyperventilation. Since that time, EIAH has been the focus of increasing scientific interest and debate. The reader is referred to an excellent review of the topic (2), which reviews the potential causes of EIAH in detail and provides the framework for classification of the severity of hypoxemia, which is used here to discuss gender differences in pulmonary gas exchange during exercise. The use of these criteria allows the identification of key components of EIAH, namely ventilation-perfusion (a/) inequality, diffusion limitation, and insufficient hyperventilatory response that individually may not result in a decrement in PaO2. Evidence suggests that a consequence of EIAH is that even small amounts of EIAH have a significant detrimental effect on limiting O2 transport and use during maximal exercise (3,13).
There has been considerable interest recently in defining any gender-based differences in the pulmonary system’s response to exercise. However, the results of investigations have been confusing, controversial, or even nonexistent. Specific to pulmonary gas exchange, pulmonary structural and morphologic differences between genders may render women more susceptible to impairments in pulmonary gas exchange compared with men. However, this may be mitigated in part by a lower maximal oxygen consumption and cardiac output in women, resulting in a lower demand placed on the pulmonary system.
THEORETICAL BASIS FOR A GENDER DIFFERENCE
It is generally believed that, with few exceptions, the lung in physically trained humans is not appreciably different from the sedentary individual. Also, physical training sufficient to increase maximal aerobic capacity has no measurable effect on lung function or structure. However, there are important gender differences in pulmonary structure and resting pulmonary function that may have an effect on gas exchange and the integrated ventilatory response during exercise. Figure 1 represents a schematic illustration of potential differences between genders. It has been known for some time that adult women consistently have lower resting lung diffusing capacity ([hemoglobin] corrected), smaller lung volumes, and lower maximal expiratory flow rates even when corrected for age and standing height relative to men (10,11,15). Some of the gender differences in resting lung function are partially explained by differences in sitting height, which may serve as a surrogate for chest volume (15). In support, more recent reports found that women had significantly smaller lung volumes and lower maximal expiratory flow rates compared with predicted values for men at the same age and standing height (4,10). It has been suggested that gender differences in lung diffusing capacity can be explained by fewer total number of alveoli (smaller surface area) and smaller airway diameter relative to lung size in women (lower maximum flow rates), and these differences probably become significant relatively late in the growth period of the lung. Therefore, given these pulmonary structural differences that exist between men and women, it could be predicted that women would be more likely to have gas exchange disturbances and ventilatory limitations during exercise compared with men with similar metabolic demand.
MECHANISMS OF EXERCISE-INDUCED ARTERIAL HYPOXEMIA
Gas Exchange Efficiency
Although an increase in AaDO2, inadequate hyperventilation, and temperature- and pH-induced shifts in the oxygen hemoglobin (Hb) dissociation curve are factors that contribute to EIAH, the focus of this discussion is on how the pulmonary system during exercise is affected by gender. Thus, issues related to oxygen delivery are not discussed further.
The increase in the AaDO2 with exercise is of primary importance in the development of EIAH. The AaDO2 increase in response to different exercise types varies even among the same subjects exercising at the same absolute and relative oxygen consumption, and the AaDO2 is greater during running than cycling exercise (5). This can complicate comparisons between studies that use differing exercise methods and testing protocols. However, there is little doubt that gas exchange limitations are more common in those individuals (and animals) capable of a high level of aerobic work. Notably, the pulmonary gas exchange response to heavy and maximal exercise is highly variable even among populations of similar aerobic fitness. Figure 2 and the data in Figures 3 and 4 that follow represent data collated from previously published studies (including (1,4,5,7,12,14)) in 57 women (O2 max 32–70 mL·kg−1·min−1) and 135 men (O2max 30–83 mL·kg−1·min−1). Figure 2 shows AaDO2 as a function of oxygen consumption (O2) during very heavy exercise at 90% to 100% of maximum in 118 male subjects using data obtained during both cycling and running exercise. The variability in individual response is striking, and the AaDO2 varies almost 10-fold (∼5–50 Torr) at any given level of O2. An AaDO2 of greater than 25 Torr can be thought of as representing a mild gas exchange limitation, with values greater than 35 Torr representing a severe gas exchange limitation (2). It can be appreciated that although most of the subjects who have an AaDO2 greater than 25 Torr are exercising at a O2 of more than 60 mL·kg−1·min−1, there are still some who experience an AaDO2 of more than 25 Torr at a O2 of less than 50 mL·kg−1·min−1. It should also be noted that although the AaDO2 is often most increased during very heavy to maximal exercise, there is a clear trend among some subjects toward developing a gas exchange impairment even during moderate exercise (4). To date, this last observation has received limited attention.
Mechanisms of the Increase in Alveolar-Arterial Oxygen Pressure Difference
Potential mechanisms of an increased AaDO2 with exercise include intrapulmonary or extrapulmonary shunting, a/ inequality, and failure of end-capillary diffusion equilibrium. Small extrapulmonary shunts from bronchial and Thebesian veins in the range of 1% of cardiac output could account for almost a 10-Torr reduction in PaO2 during exercise (7) and have not been well investigated as a possibly contributor to the AaDO2. Although theoretical arguments have been advanced suggesting that they are not of primary importance, extrapulmonary noncardiac shunts cannot be ruled out as a cause of an increase in the AaDO2.
The contribution of a/ inequality to the AaDO2 observed during heavy exercise varies with study population, and in some, it may account for most of the AaDO2 (7). Figure 3A shows the LogSD, an index of a/ inequality, derived using the multiple inert gas elimination technique, in subjects divided based on aerobic fitness exercising close to O2max. There is no difference in a/ inequality between highly fit subjects (O2max > 60 mL·kg−1·min−1) and those of low to average aerobic fitness (O2max < 50 mL·kg−1·min−1). Thus, although the amount of a/ inequality within an individual generally increases with increasing exercise intensity, there does not seem to be a significant relationship between the development of a/ inequality and aerobic fitness. In fact, very fit cyclists selected for presence or absence of EIAH (14) had no significant effect of exercise intensity on a/ matching, and the contribution of a/ inequality to the AaDO2 was less than 40% of the total in these subjects. Although a/ inequality contributes to the AaDO2, it is perhaps unlikely to be the primary factor in the development of EIAH in either gender. In keeping with this idea, the LogSD is not different between subjects who do and do not have evidence of gas exchange impairment as previously defined (data not shown). Ventilation-perfusion inequality includes variations in PaO2 relative to blood flow and the effects of this in-homogeneity can be in part mitigated by increasing alveolar ventilation, which acts to shift the overall a/ distribution to the right. This represents overventilation of area with limited perfusion (increased ventilation of area of high a/ ratio) rather than perfusion of area with low ventilation.
The amount of pulmonary–end capillary diffusion limitation also varies markedly between individuals (6) and is increased with increasing exercise intensity, and in subjects with greater aerobic fitness. Figure 3B shows the amount of the AaDO2 estimated using multiple inert gas elimination technique modeling to be attributable to diffusion limitation in subjects of varying fitness. The AaDO2 contribution from diffusion limitation as estimated from inert gas measurements is markedly increased in highly fit subjects (P < 0.0001) and is also markedly increased in those subject with mild or severe pulmonary gas exchange impairment (Fig. 3C; P < 0.0001). Thus, pulmonary diffusion limitation is likely an important feature of EIAH. However, the cause of the pulmonary diffusion impairment of oxygen transport is unknown. It is likely because of rapid pulmonary capillary transit resulting in a failure of end-capillary diffusion equilibration.
Few studies have attempted to measure pulmonary transit times in exercising athletes and link the results to gas exchange. A significant fall in whole lung pulmonary transit time with exercise has been shown (6), but unequivocal evidence to show that capillary transit times approach the minimum time for oxygen equilibration is lacking. Thus, rapid pulmonary capillary transit time has not been established as a cause of pulmonary diffusion limitation. However, studies measuring capillary transit time have measured only mean transit time, and thus some portion of the red cells in the pulmonary capillaries will have a shorter transit time and may contribute to diffusion limitation of oxygen transport even when the mean transit time is adequate. Subjects in whom inert gas evidence of diffusion impairment develops often experience this at submaximal levels of exercise (14), where presumably capillary recruitment is not maximal and transit time is not minimal, which also may argue against reduced transit times as being the cause. However it is theorized that the ratio of diffusional (D) to perfusional () conductance in the factor that determines the completeness of end capillary equilibration (D/β). Subjects with EIAH have been shown to have a lower D/β compared with those in whom EIAH does not develop (14). Thus, it is possible that the perfusional conductance, for example, blood flow, is recruited more quickly than capillary volume, causing transient diffusion impairment even at relatively low levels of exercise.
Although the excessively widened AaDO2 is commonly viewed as the primary cause for EIAH among most individuals, an inadequate hyperventilatory response (defined as an arterial PCO2 of > 35 Torr during heavy exercise) also has been shown to play a role. Also, it seems that those individuals who experience the most severe EIAH have equal contributions from an inadequate hyperventilation and widened AaDO2. The ventilatory response to exercise varies widely among individuals. This is in part the result of the response to circulating chemical stimuli (e.g., [H+], [K+], O2, and catecholamines), the mechanical influences of airway diameter and respiratory muscle force production, pressure or force developed by inspiratory muscles, and the ventilatory responsiveness or receptor sensitivity. The contribution of each to EIAH is complex and poorly understood.
GENDER DIFFERENCES IN GAS EXCHANGE
Gas Exchange Efficiency
Exercise-induced arterial hypoxemia and arterial partial pressure of oxygen
The previous discussion focused on the contributions of a/ inequality and diffusion limitations to EIAH. There are few published data directly comparing pulmonary gas exchange between genders. However, again referring to compiled data, it can be appreciated that the slope of the relationship of the AaDO2 and O2 relationship (Fig. 4A) during heavy to maximal exercise is greater in women than men. From these data, 12% of the women with a O2max of less than 50 mL·kg−1·min−1 had evidence of gas exchange impairment. For men of the same fitness level, less than 2% have evidence of gas exchange impairment. These differences in gas exchange are reflected in the PaO2 data, and women have a greater negative slope of the PaO2/O2 relationship (Fig. 4B) than do men. Approximately 10% of women with a O2max less than 50 mL·kg−1·min−1 had a PaO2 of less than 90 Torr during heavy and maximal exercise compared with less than 2% of the men.
The specific components of the AaDO2 are poorly investigated in women. However, a recent study (12) compared men and women during normoxic and hypoxic exercise of varying intensities. By matching the subjects for height and o2max, comparisons could be made between genders independent of these confounding variables. Coincidentally, matching on these two variables resulted in similar lung volumes in the two genders because the women were, on average, 110% of predicted for lung size (forced vital capacity) for normal women. The LogSD was less in these women than in men by an average of 17% over all exercise levels, suggesting that a/ inequality may be a less important contributor to the AaDO2 in women than men. A mechanism by which a/ inequality is less in women compared with men is unclear, and because of the very limited data in women, it is difficult to know whether it is a representative characteristic of women in general, or whether they reflect an unusual group: athletic females larger than average lungs. A previous study in male athletes, documented an increase in a/ inequality in individuals with smaller lung size and the observation that the incidence of high altitude pulmonary edema may also be related to lung size (along with ascent rate and exercise intensity) supports the contention that absolute lung size is important in the susceptibility to pulmonary gas exchange impairment during periods of heavy physical activity.
Using a Bohr integration technique, it is possible to calculate the diffusing capacity of the lung for oxygen (DLO2), from inert gas data, when there is evidence of diffusion limitation. Because not all subjects experience diffusion impairment during normoxic exercise, hypoxia is used to induce diffusion impairment so that the DLO2 can be calculated. In the study previously cited (12), the DLO2 was lower in females than males. However, the mean DLO2/β determining the completeness of end capillary oxygen diffusion was not different between genders, largely because of lower Hb concentration, and thus β, in the women compared with men. Consequently, despite a lower DLO2 in the women compared with men, the contribution of diffusion limitation to the AaDO2 was not different between genders. It should be emphasized that results of the study indicate effects of gender independent of lung size, body size, and aerobic fitness, because these factors were controlled. Despite this, the DLO2 for oxygen was lower in women than men, suggesting that women may be more susceptible to EIAH in situations of high demand (high ), or at an altitude where β may be increased. Also, individuals with small lungs and a small surface area available for diffusion may be at risk regardless of gender. However intriguing this possibility is, it remains to be established.
Evidence suggests that the contribution of alveolar ventilation to gender differences in EIAH is likely the result of mechanical limits to expiratory flow and to hormonal factors. Because of smaller lungs, women tend to show greater mechanical limits to expiratory flow, creating a smaller maximal flow–to-volume envelope compared with men (10). Figure 5 shows ensemble averaged tidal flow-to-volume loops for rest through maximal exercise in highly fit and less fit women. This greater expiratory flow limitation in more fit women leads to greater relative hyperinflation during heavy exercise compared with men and a greater respiratory muscle demand. Thus, women would be expected to show significant expiratory flow limitation sooner (i.e., at a lower e and at a much lower o2). At a comparable emax, more of the tidal volume would be flow limited compared with men. From this, it would be expected that women would experience a lack of substantial hyperventilation at a O2 (and CO2) that men would not. Although women tend to show greater mechanical constraints to ventilation compared with men, the significance of the variability in ventilation between genders may be minimal. For example, Figure 4C shows that the intercept for the PaCO2–O2 relationship is 2 Torr lower in women than in men. Additionally, 10% of men with a O2 less than 50 mL·kg−1·min−1 have evidence of limited hyperventilation, as previously defined, compared with 5% of women. This suggests that although there may be additional ventilatory constraints imposed by small lung size on women compared with men, these are unlikely to be of sufficient magnitude to affect the ventilatory response to exercise. The specific contribution to the increased expiratory flow limitation commonly observed in women to EIAH is unclear. With regard to hormonal contribution, progesterone in a known ventilatory stimulant that peaks during the luteal phase of the menstrual cycle. How changes in the circulation of progesterone throughout the menstrual cycle affect the ventilatory response to exercise and to EIAH are also unclear.
What is the consequence of the mechanical constraints of ventilation to EIAH? Alveolar ventilation at O2max would have to be well above estimates of ventilatory capacity during exercise to offset the widened AaDO2 and EIAH reported in women. However, we cannot be sure that, even if the mechanical constraints on flow limitation were removed, subjects would increase ventilation sufficiently to offset the EIAH. For example, an increase in ventilation with a helium–oxygen gas mixture (which increases the size of the flow-to-volume envelope) at O2max is only approximately one third of that required to completely offset the arterial hypoxemia during high intensity exercise (9,10). Direct measurements of the effect of heliox on PaO2 were performed by Dempsey et al. in 1984 (1). These authors showed that subjects increased Pao2 during the hyperventilation with heliox inspirate during heavy exercise and PaO2 increased in proportion to PAO2, demonstrating that mechanical constraints contribute at least in part to EIAH. Overall then, the smaller lung volumes and lower maximal flow rates that creates expiratory flow limitation in a fit, exercising women seem to constrain the capacity of a highly fit women to compensate for an inadequate alveolar to arterial O2 exchange, thereby exacerbating the EIAH.
CONSEQUENCES OF EIAH IN WOMEN
Does EIAH affect oxygen transport? During submaximal work (where EIAH sometimes occurs), O2 is not compromised, probably because an increased O2 extraction by the working muscle is possible, and this compensates for any hypoxemia-induced deficiencies in O2 transport. The reduced ventilatory response in those with EIAH is certainly more economical and may be beneficial under these circumstances. However, at maximal exercise, no room remains for compensation, and the occurrence of EIAH means that the maximal (available) arterial-mixed venous oxygen difference will be compromised. Accordingly, preventing EIAH via a mildly hyperoxic inspirate, which maintains SaO2 at resting levels, led to significantly higher O2max in men (13) and women (3).
The amount of EIAH needed to reduce O2max in women and the magnitude of this effect seems to differ in several respects from studies in men. Research has shown that either an exercise-induced reduction in SaO2 of more than 5% below rest to more than 92% was required to cause a measurable effect on O2max in trained men, with the effect approximating a 1% decrement of O2max for each 1% decrement in SaO2 less than 92% (13). More recent data in women suggest that the threshold of arterial O2 desaturation required to cause a measurable change in O2max is slightly less than 3% below resting levels, or in the 94% to 95% SaO2 range (3). Figure 6 shows a significant relationship between the change in O2max with FiO2 0.26 and SaO2. That is, those who showed the most desaturation under normoxic conditions tended to show the greatest improvement in O2max if the desaturation was prevented, whereas those subjects who desaturated the least during normoxia showed the least change in O2max. Whether this is truly a gender difference or a methodologic difference between studies is unknown at this time.
It is true that the most commonly encountered levels of EIAH among fit men and women are usually modest and will cause less than a 10% reduction in O2max. However, a smaller but still significant fraction of these subjects experienced greater levels of EIAH in maximal exercise, and the detrimental effect on O2max was in the 8% to 15% range. These effects of EIAH on O2max (at sea level) are comparable with those normally achieved by exposure to moderately high altitudes of 8000 to 10,000 feet in normal untrained subjects, and by moderate levels of anemia and loss of blood volume. We would predict that travel to even very modest elevations in altitude (to less than 5000 feet) would exacerbate the severity of EIAH and further limit O2max in some individuals. This effect would probably be predictable from tendencies toward EIAH measured at sea level. Finally, given the deleterious aging effects on lung elastic recoil, airway closure, and diffusion capacity even in the highly fit aged subject (8), it does not seem unlikely that the fit aged female may be especially susceptible to EIAH and its associated curtailment of O2max.
A related and untested question is the effect of EIAH on exercise performance. Does EIAH occur and is it sustained in heavy submaximal endurance exercise in the highly trained? Because the relationship between o2max and endurance exercise performance is debatable, would prevention of this sustained EIAH affect performance time as it has affected o2max? If, in fact, there is a high prevalence of EIAH in active women, then active women may be especially susceptible to EIAH-induced decrements in exercise performance.
CONCLUSIONS AND FUTURE DIRECTIONS
Although it is tempting to propose that women are more susceptible to EIAH than men, we must emphasize the need for more descriptive data in women. For example, sufficient studies in young adult men have been conducted to document clearly that untrained subjects normally widen their AaDO2 two- to threefold from rest to maximal exercise, and that they also hyperventilate, which raises alveolar PO2 sufficiently during strenuous exercise to prevent Pao2 from falling below resting levels. To date, there are few studies that have measured temperature-corrected arterial blood gases during exercise in women, of any activity level. Data obtained using pulse oximetry to monitor oxygen saturation during exercise is not sufficiently accurate to establish the prevalence of EIAH. Clearly, much more testing is needed to determine the prevalence of EIAH among the normal population of young women. Also, we need to establish the reference standard of response for these fundamental indices of lung function during exercise. Additionally, the need to document how hormonal cycling affects EIAH to avoid any confounding effects of changes in ovarian hormones is unclear. Clearly, these hormonal changes do affect a variety of ventilatory, cardiovascular, volume regulatory, and hematologic functions, which may influence EIAH. Finally, given the deterioration in lung elastic recoil, the increase in airway closing volume, and the reduced diffusion surface area coincident with the normal ageing process, we also may expect even greater gender effects on the prevalence of EIAH in older women across the entire fitness spectrum. However, it should be kept in mind that when lung size and fitness level are controlled for, many of the gas exchange differences between genders seem to be lost. Pulmonary limitations to exercise are found in individuals of varying fitness levers and both genders.
Supported by the National Institutes of Health (grant nos. HL17731 and MO1RR00827 [SRH]) and the American Lung Association (grant no. RG-039-N [CAH]). The authors thank Mark Olfert, Peter Wagner, Jerry Dempsey, Bruce Johnson, and Gerry Zavorsky for generously sharing their data.