September - December 2004: 
Volume 17, Issue 3

Click on the image to download the Issue in PDF format.


Alveolar-Arterial Oxygen Difference Increase during Exercise
ABSTRACT. The difference between the alveolar and arterial oxygen pressure is known as alveolar – arterial difference (PA-aO2) or efficiency for gas exchange. Gas exchange within the lung is not perfect even at rest. During exercise, the gas exchange progressively worsens. The mechanisms that contribute to A-a DO2 are not well known. Ventilation-perfusion mismatch (VA/Q) is considered to be the main reason for A-aDO2 widening. VA/Q mismatch is due to the effect of gravity - structural differences in airways – blood vessels, bronchoconstriction, secretions from airways irritated by high flows or dry-cold air and mild interstitial edema. Diffusion limitation is a second contributing factor to A-aDO2 widening. It can be attributed to several reasons, such as: the surface area for diffusion, the distance required for diffusion from the alveolar membrane to the red blood cell, the transit time and the rate of equilibration of mixed venous blood with alveolar gas. Diffusion limitation may occur to well-trained athletes at high exercise intensity, while it is unlikely to occur in untrained subjects during low and moderate exercise. The last contributing factor to A-aDO2 widening is considered to be the mixing of shunted blood (intra-extrapulmonary shunt) with arterial blood. Even though there is not enough published data to support the previous statement, it seems that not only the A-aDO2 is negatively affected but simultaneously may play a positive role by protecting the alveolar – capillary membrane from liquid accumulation and disruption in athletes. Pneumon 2004, 17(3):265-271.
Full text


The efficiency of pulmonary gas exchange for O2 is defined and quantified as the difference between the alveolar (PAO2) and the arterial (PaO2) oxygen pressure and referred to as alveolar - arterial difference (PA-aO2). Gas exchange within the lung is not perfect even at rest (A-aDO2= 5-10mmHg). During exercise, the gas exchange efficiency progressively worsens in an intensity dependent manner. Therefore, at maximal exercise A-aDO2 can reach values of 35-50 Torr in elite athletes. A-aDO2 starts widening at moderate exercise and continues to increase further at maximal exercise. According to Wetter et al.1 endurance exercise does not result in a time dependent decline of gas exchange. This means that the magnitude of A-aDO2 is determined by the metabolic rate, rather than the duration of exercise.

Mechanisms involved in A-aDO2 widening

The mechanisms that contribute to A-aDO2 widening during exercise are complex and not well known. It seems that during exercise, gas exchange worsening is due to exaggeration of mechanisms present at rest.

1. Ventilation - Perfusion mismatch

Ventilation - perfusion mismatch (VA/Q) is considered to be the major contributor to A-aDO2. In a normal upright position lung inspired ventilation is distributed more toward the lung bases rather than the lung apex, since prior to the start of inspiration, the alveoli are more expanded at the top than at the bottom. We also know that in larger lung volumes, the lung is becoming stiffer. Therefore, the alveoli at the bottom can expand more than those at the apex during inspiration, since they do not have their largest volume. Furthermore, gravity contributes to interregional differences. Approximately, twice as much of the inspired ventilation is delivered to the bottom versus top regions of the upright lung. In addition, there are differences in ventilation in isogravitational regions of the lung, due to variations in their mechanical characteristics. Furthermore, it is also known that there is heterogeneity in pulmonary flow. The blood flow distribution in the normal upright lung is five times greater at the bottom than at the apex, due to effect of gravity. Also, there are differences in blood flow distribution within a given horizontal plane of the lung. Within - region non-uniformities of perfusion are due to structural differences in the diameter length and branching angles of the vessels. So, the (VA/Q) ratio is also non - uniform at rest. At rest VA/Q mismatching could explain 50% of the observed A-aDO22.

During exercise interregional VA/Q uniformity improves up to oxygen consumption (VO2) values of 1-2 L/min. This occurs due to the incremental rise in pulmonary artery pressure, which causes distention and recruitment of the pulmonary vascular bed3. Even if the ratio becomes more uniform during moderate exercise increases, it still cannot be completely uniform, due to intra-regional differences. During high exercise intensity the ratio VA/Q becomes more non - uniform due to intra - regional differences and also due to disproportional increase of VA compared to Q. Even if the distribution for VA/Q becomes more non - uniform, the higher overall VA/Q assures that a small portion, if any, of the lung will be considerably under ventilated. The VA/Q abnormality generally disappears within twenty minutes after completion of the exercise4. It should be noted that the increase in VA/Q non - uniformity during exercise could not explain the dramatic widening of A-aDO2.

The mechanisms underlying the greater VA/Q mismatch during exercise are not well established. However, certain speculations can be made regarding the factors contributing to VA/Q mismatch:

  1. The effect of gravity -structural differences in airways- blood vessels,
  2. bronchoconstriction may alter ventilation distribution,
  3. secretions from airways irritated by high flows or dry / cold air -resulting in alterations in ventilation distribution, and
  4. mild interstitial edema leading to changes in compliance or resistance, affecting ventilation and blood flow5.

2. Diffusion limitation

A second contributing factor to the A-aDO2 is diffusion limitation. Diffusion through tissues is proportional to the tissue area and the difference in gas partial pressure between two sides and inversely related to tissue thickness. The major determinants of alveolar - capillary diffusion are:

  1. Surface area for diffusion,
  2. distance required for diffusion from alveolar membrane to red blood cells (RBC),
  3. transit time and
  4. rate of equilibration of mixed venous blood with alveolar gas6.

During exercise, hydrostatic pressures increase in the pulmonary artery and this sometimes may cause low grade edema, which would result in increment in diffusion distance from the alveolar membrane to RBC7.

Piiper et al8 explain the degree of diffusion limitation in terms of their compound variable D/βQ, where D (ml· min-1 ·Torr-1) is lung diffusing capacity and β (ml O2·l blood-1·Torr1) is the mean slope of the conceptually linear O2 - Hb dissociation curve in the physiological range. With exercise, as D, β, and Q increase, the ratio D/βQ falls. This fall may be explained from the fact that β and Q increase more proportionally to D. As we mentioned earlier, athletes show high cardiac output (Q) and high oxygen extraction from blood perfusing muscles. This results in a higher increase of the β component, causing greater reduction in D/βQ in athletes than in inactive subjects.

Furthermore, the increase in cardiac output with increasing exercise intensity causes a reduction in the RBC transit time. At rest, the mean time is 0.75 seconds, which will decrease with exercise, as pulmonary blood flow increases. As we know, with exercise pulmonary vascular and left atrium pressures cause an expansion and recruitment of pulmonary capillary surface area, which minimizes any reduction in RBC transit time. Even if we do not know the distribution of transit times through the lungs, in highly trained athletes during high intensity exer­cise, RBC mean transit time does not decrease by more than 0.45 seconds9. Diffusion limitation could occur if pulmonary blood flow continues to rise after pulmonary capillary blood volume has reached its maximal morphologic limit (220~ml). That would probably occur in elite athletes with high oxygen consumption and pulmonary blood flow (>35 L/min). Hammond et al10, examined gas exchange during exercise in healthy subjects at sea level. They noticed that steady - state exercise increases VA/Q inequality, which was not reversible with breathing one hundred percent (100%) oxygen and also persisted transiently after exercise. Additionally, they mentioned that there was evidence of alveolar - end capillary diffusion limitation in normal individuals at high VO2 >3 L/min. The previous study was in agreement with the study of Wagner et al3. Both studies used the Multiple Inert Gas Elimination Technique (MIGET). MIGET estimates the amount of VA/Q inequality and assumes complete alveolar - end capillary diffusion equilibration and negligible postpulmonary shunts. According to the previous assumptions, any difference between the actual A-aDO2 and that predicted from the measured amount of VA/Q inequality by MIGET can be attributed to diffusion limitation. Up to two-thirds (2/3) of the total A-aDO2 can be explained with diffusion limitation. On the other hand, it should be noted that MIGET does not directly measure diffusion limitation and also uses assumptions that may not be correct.

In conclusion, diffusion limitation may contribute significantly to A-aDO2 in healthy trained athletes at high exercise intensity, where is unlikely to occur in untrained subjects during low or moderate exercise.

Exercise - induced pulmonary hemorrhage in athletes

Hopkins et al11, examined the integrity of pulmonary blood gas - barrier in elite athletes after intense exercise. They found that the number of RBC and proteins in the BAL fluid of exercising athletes were higher compared with the sedentary subjects, supporting the hypothesis that the integrity of the blood - gas barrier in athletes is impaired by high intensity and short duration exercise. They attributed the blood - gas barrier impairment to mechanical stress rather than to an inflammatory mechanism since they found no significant elevation in cytokines. In addition, West et al12 suggested that some elite athletes during heavy exercise reach capillary transmular pressures >40mmHg and capillary wedge pressures of 25 mmHg. This would result in "stress failure", meaning that a primary hydrostatic edema could lead to permeability edema. On the contrary, Croix et al13, reported that repeated maximum exercise reduces the A-aDO2.

3. Shunt

The third contributing factor to the A-aDO2 is the mixing of shunted blood with arterial blood. Shunt can be intracardiac (between atria-ventricles), postpulmonary (consists of Thebesian venous drainage from the coronary circulation, which goes directly into the left ventricle)14, as well as anastomoses between the bronchial veins and pulmonary veins15 and intrapulmonary (anatomic connections between arterial and venous vessels). Intracardiac and postpulmonary shunts are also called extrapulmonary shunt and account for half of the increase in A-aDO2 from rest to moderate exercise (1%-2% of cardiac output)16. Bjork et al17, showed that postpulmonary shunting would influence the decrease of oxygen tension in arterial blood in resting subjects.

Furthermore, bronchial circulation may contribute to postpulmonary shunting. Contrary to the previous studies, Gledhill et al.16 suggested that during exercise it is unlikely that bronchial circulation increases in proportion to cardiac output, because this requires a reduction in bronchial vessels resistance. In addition, Hammond et al10, reported that shunt would not exceed 0,6% of the cardiac output during exercise. This study is in agreement with Dempsey et al18. By subjecting healthy individuals to breath one hundred percent (100%) oxygen, as conducted by the previous researchers, it may underestimate the shunted blood, since significant pre- pulmonary capillary oxygen exchange occurs in the lung and pre- capillary oxygenation is dependent on the diffusion gradient. Moreover, inspiration of high FiO2 may not be accurate to distinguish small shunt19.

Intrapulmonary shunt bypasses the pulmonary capillary bed and is not blood from low VA/Q inequality. Tobin et al20, showed that intrapulmonary shunt exists in perfused human lungs and probably affects the PaO2 during exercise of increasing intensity, since PvO2 falls. Intrapulmonary shunts act like postpulmonary and are undetectable to MIGET. There are other techniques, which show and quantify arteriovenous intrapulmonary shunt such as transesophageal echocardiogram21, cinefluorography in the lungs of normal dogs22, casts made from post mortem human lungs23, infusion of glass spheres (10-750μ in diameter) into pulmonary artery23, 100% oxygen breathing method, technetium 99m albumin microsphere method24,25 and saline contrast bubbles in the pulmonary vein26. The methods of saline contrast bubbles, in combination with technetium 99m, are more preferable than the others, since they are more comfortable, quicker, requiring small radiation dose, less invasive and can be applied to exercising humans, depending on exercise modality. In addition, bubbles are strong reflectors of ultrasound because the gas within the bubbles has significantly different acoustic impedance in relation to that of blood. However, agitate saline solution should consist of bubbles with size larger than 20 microns, otherwise, they will rapidly collapse (<1 second), due to surface tension effects and vascular pressures27. As a result, they may not enter the pulmonary microcirculation. Also, small bubbles could be also formed during agitation, but are unstable and dissolve rapidly by diffusion in 190ms28.

Another advantage of the saline contrast technique is that it is able to make a distinction between postpulmonary and intrapulmonary shunt. Atrial septal defect with right to left shunting contrast is usually recorded in the left atrium within 1-2 cardiac cycles after its appearance in the right atrium. Late arrival of contrast in the left atrium, after peripheral injection (4-8 cardiac cycles after right atrial appearance), denotes the presence of pulmonary arteriovenous shunt, especially when the contrast is seen entering from the pulmonary veins. This delay in arrival to the left atrium shows the time required for passage through the pulmonary circulation26.

In conclusion, shunt may contribute significantly to the A-aDO2 during exercise while further investigations are needed to show and quantify the total shunt fraction during exercise and it's contribution to A-aDO2.

Speculations about the role of the shunt

Dr Eldridge et al (2003), (unpublished data) using the bubble technique showed that in most subjects intrapulmonary shunt exists during exercise. While it was not seen at rest, this may be explained by the fact that shunt is a small percentage of cardiac output and with this technique is not easy to be visualized. Probably a quantification technique may elucidate this problem. Also, shunt size is very small at rest and bubbles >20 microns can not pass through the shunt vessels, entering directly to the circulation and getting trapped in the lung29. On the contrary, shunt can be seen during incremental exercise. This can be explained by the increase in pulmonary artery pressure due to the increase in cardiac output. When pulmonary artery pressure reaches a critical point, shunt may open and this "critical pressure" may differ between subjects. The distention of pulmonary capillaries may also cause distention of shunt vessels. If this speculation is correct, shunts contribute to the increase in A-aDO2 during exercise (reduce PaO2 and CaO2) and influence the performance negatively. In order to understand the physiological meaning of shunt, we may use a hypothetical example. A young athlete exercises on a cycle ergometer at 280 W. His Q=20 L/min, CVO2= 6,73 m/100ml, VO2= 3L/min, CC΄O2= 20,61 ml/100ml blood. We calculate PaO2 under different shunt percentages (1%, 5% and 10%) by using the following equation:

QT ΄ CaO2 =(Qs ΄ CVO2) + (Qt - Qs) ΄ CC΄O2

Solving for CaO2 and using the data from Table 1, located in the appendix, we calculate PaO2

Shunt 1% ® CaO2 =20.6 ml/100ml blood ® PaO2= normal

Shunt 5% ® CaO2 =19.91 ml/100ml blood ® PaO2= 70mmHg.

Shunt 10% ® CaO2 =19.22 ml/100ml blood ® PaO2= ~60mmHg.

The results from the previous example indicate that shunt percentages of 5% and 10% contribute significantly to PaO2 reduction. In this hypothetical example we did not take into consideration pH and the shift in the Hb-O2- dissociation curve, which may alter the results.

The appearance of shunt during exercise may also play a positive role, since shunted blood may cause a reduction in pulmonary pressure (Ppa). According to the following equations and the assumption that blood flow stays constant (ex. 20 l/min).

R=pressure/flow (1), Ppa=ax+b (2)30 upright position, a=1.01 and b=6.7

Shunt 1% ® Ppa=26.67mmHg ® R=1.33mmHg/l/min.

Shunt 5% ® Ppa=25.89mmHg ® R=1.29mmHg/l/min.

Shunt 10% ® Ppa=24.88mmHg ® R=1.24mmHg/l/min.

The above example shows that an increase of shunted blood percentage will cause reduction in Ppa and in pulmonary resistance, since less blood will pass through the pulmonary capillary bed. As a result alveolar capillary membrane may be protected from liquid accumulation and disruption in elite athletes. This speculation is in agreement with the study of Croix et al13 where no structural damage after exercise is reported. It is also speculated that there is a connection between cardiac output and shunted blood. As it has been already mentioned, exercise will increase cardiac output, which leads to an increase in Ppa and as a result shunt will open.

Table 1

Table 2


This will cause a reduction in PaO2 and CaO2 but may moderate the increase in Ppa, even if cardiac output still rises (shunts reduce PaO2 and probably increase the stimulus for ventilation - perfusion). It is not known if shunted blood is a stimulus for cardiac output to increase and further investigation is needed.

Conclusively, shunt contributes to the increase in A-aDO2 during exercise. The shunt serving vessels open due to vasodilation of pulmonary vascular bed and the increase in Ppa. Their existence may protect the integrity of blood gas barrier and may interact with cardiac output to modify the increment of Ppa.

In summary, A-aDO2 widens with exercise in healthy subjects. This widening can be attributed to VA/Q mismatch at low-moderate - high exercise intensity, to diffusion limitation at least at high exercise intensity and to intra -extrapulmonary shunt, which seem to play a potential role. Further research should be done to elucidate this question.


We appreciate the help of Dr Jerome Dempsey, University of Wisconsin- Madison, Department of Population Health Sciences, John Rankin Laboratory of Pulmonary Medicine, 504 N. Walnut Street, Madison WI 53726.



1. Wetter TJ, St Croix CM, Pegelow DF, et al. Effects of exhaustive endurance exercise on pulmonary gas exchange and airway function in women. J Appl Physiol 2001; 91:847-858.

2. Dempsey J. Pulmonary Physiology lectures, 2002.

3. Wagner P, Gales G, Moon R, et al. Pulmonary gas exchange in humans exercising at sea level and simulated altitude. J Appl Physiol 1986; 61:260-270.

4. Caillaud C, Anselme F, Mercier J, et al. Pulmonary gas exchange and breathing pattern during and after exercise in highly trained athletes. Eur J Physiol 1993; 65: 37-42.

5. Dempsey JA. Wagner PD. Exercise - induced arterial hypoxemia. J Appl Physiol 1999; 87:1997-2006.

6. Powers SK, D Martin, Dodd S. Exercise - induced hypoxemia in elite endurance athletes. Incidence, causes, and impact on VO2max. Sports Medicine 1993; 16: 14-22.

7. Wagner P. Ventilation - perfusion matching during exercise. Chest 1992; 101:192S-198S.

8. Piiper J, Scheid P. Model for capillary - alveolar equilibration with special reference to O2 uptake in Hypoxia. Respir Physiol 1981; 46:193-208.

9. Warren GL, Cureton KJ, Middendort WF. Red blood cell pulmonary capillary transit time during exercise in athletes. Medicine and Science in Sports and Exercise 1991; 23:1353-1361.

10. Hammond MD, Gale GE, Kapitan KS, et al. Pulmonary gas exchange in humans exercising at sea level. J Appl Physiol 1986; 60:1590-1598.

11. Hopkins SR, Schoene RB, Henderson WR, et al. Intense exercise impairs the integrity of the pulmonary blood - gas barrier in elite athletes. Am Respir Crit Care Med 1997; 155:1090-1094.

12. West JB, Mathieu-Costello O. Stress failure of pulmonary capillaries as a limiting factor for maximal exercise. Eur J Appl Physiol 1995; 70:99-108.

13. St Croix CM, Harms CA, McClaran SR, et al. Effects of prior exercise on exercise - induced arterial hypoxemia in young women. J Appl Physiol 1998; 85:1556-1563.

14. Ravin MB, Epstein RM, Malm JR. Contribution of thebesian veins to the physiologic shunts in anesthetized man. J Appl Physiol 1965; 20:1148-1152.

15. Agostini P, Godden DJ, Baile EM. Measurement of bronchial blood flow in humans. In Buttler J, editor. Lung biology in health and disease: the bronchial circulation. Vol 57. New York: Markel Dekker; 1992; pp.181-196.

16. Gledhill N, Frosese AB, Dempsey JA, et al. Ventilation to perfusion distribution during exercise in health. In Dempsey JA, Reed CE, editors. Muscular exercise and the lung. Wisconsin: University of Wisconsin Press; 1977, pp. 325-344.

17. Bjork VO, Malmstrom G, Uggla LG. Comparison of the oxygen tension in blood from the left atrium and systemic artery. Am Heart J 1954; 48:8-12.

18. Dempsey JA, Hanson PG, Henderson KS. Exercise - induced arterial hypoxemia in healthy human subjects at sea level. Journal of Physiology 1984; 355:161-175.

19. Conhaim RL, Staub NC. Reflection spectrophotometric measurement of O2 uptake in pulmonary arterioles of cats. J Appl Physiol: Respirat. Environ. Exercise Physiol 1980; 48:848-856.

20. Tobin CE, Zariquiey M. Arteriovenous shunts in the human lung. P.S.E.B.M. 1950; 75:827-829.

21. Nemec JJ, Davison MB, Marwick TH, et al. Detection and evaluation of intrapulmonary vascular shunt with "contrast Doppler" transesophageal echocardiography. J Am Soc Echocardiography 1991; 4:79-83.

22. Rahn H, Stroud RC, Tobin CE. Visualization of arteriovenous shunts by cineflurography in the lungs of normal dogs. Proc Soc Exp Biol And Med 1952; 80:239-241.

23. Tobin C. Arteriovenous shunts in the peripheral pulmonary circulation in the human lung. Thorax 1966; 21:197-204.

24. Whyte MKB, Hughes JMB, Jackson JE, et al. Cardiopulmonary response to exercise in patients with intrapulmonary vascular shunts. J Appl Physiol 1992; 75:321-328.

25. Strauss WH, Hurley PJ, Rhodes BA, et al. Quantification to right to left transpulmonary shunts in man. J Lab Clin Med 1969; 74:597-607.

26. Anuradha G, Vijay G, Xucai C, et al. Intrapulmonary arteriovenous shunt; Diagnosis by saline contrast bubbles in the pulmonary veins. A Soc Echocardiogr 2002; 15:1012-1014.

27. Meerbaum S. Principles of echo contrast. In: Nada N, Schlief R, editors. Advances in echo imaging using contrast enhancement. Netherlands: Kluwer 1993; 1993, p.p. 9-42.

28. Weyman AE. Principles and practice of echocardiography. 2nd ed. Philadelfia: Lea & Febiger 1994; 1994:304.

29. Buttler BD, Hills BA. The lung as a filter for microbubbles. J Appl Physiol 1979; 47:537-543.

30. Reeves JT, Dempsey JA, Grover RF. Pulmonary circulation during exercise. In Weik EK, Reeves JT. (eds), Pulmonary vascular physiology and pathophysiology. Vol. 38. New York: Marcel Dekker; 1998, 107-133.