Counterpoint: Exercise-induced intrapulmonary shunting is real

The conventional pulmonary circulatory route begins with the pulmonary artery that travels in parallel with the airway, dividing with the airway, until finally reaching the capillary bed within the acinus (4; Fig. 1A). The capillary bed consists of vessels 7 to 10 μm in diameter, never exceeding 13 μm even under very high, non-physiological perfusion pressures (8). The conventional veins then collect blood from capillaries, combining to form progressively larger vessels. Despite this traditional view of the pulmonary vascular circuit, there is substantial anatomic evidence of large-diameter arteriovenous anastomoses in the lung that bypass the traditional blood flow circuit (Fig. 1B).

Fig. 1.

Conventional and alternative views of the pulmonary circulation and gas exchange during exercise. A) Conventional view with the primary gas exchange site restricted to capillary bed. B: alternative view under normoxic conditions (Pa_O2 ≈ 120 Torr), note precapillary O₂ exchange is minimal, but sufficient for the excretion of low solubility multiple inert gas elimination technique (MIGET) gases while precapillary arteriovenous anastomoses allow for deoxygenated blood to bypass capillary beds and allow for some large diameter bubbles and microspheres to escape entrapment by the capillary bed. C: alternative view under hyperoxic conditions (Pa_O2 ≈ 700 Torr), note precapillary O₂ exchange is maximal with elimination of MIGET gases while precapillary arteriovenous anastomoses are closed preventing or reducing arteriovenous blood flow and large diameter bubbles and/or microsphere escape.

A shunt can be defined as “a vascular passage by which blood is diverted from its usual or normal path (arteriovenous)(1).” Arteriovenous anastomoses (i.e., shunts) were first described 129 years ago (19) and these pathways allow for arterial blood to bypass the capillary beds and join up with postcapillary venous blood. Large diameter intrapulmonary arteriovenous pathways (or shunts) are known to exist in many species including humans (25, 27), dogs (16), cats (17), and rabbits (17). A critique of previous anatomic work is that the methods used were not physiological. Recently we documented intrapulmonary arteriovenous pathways using 50 and 25 μm solid microspheres in healthy human, baboon, and dog lungs, which were isolated, ventilated, and perfused at physiological pressures (14, 22). These studies established the patency and functional diameter of some of these intrapulmonary arteriovenous shunt vessels under conditions that more closely replicate physiological conditions.

Using all anatomic based approaches, there is a significant amount of evidence that intrapulmonary arteriovenous shunting during exercise is indeed real. In healthy humans we have demonstrated transpulmonary passage of saline contrast bubbles during submaximal through maximal exercise, but not during upright normoxic rest (6, 12, 13, 23, 24). With the use of saline contrast echocardiography, intrapulmonary shunt is defined as the presence of saline contrast bubbles in the left heart three or more cardiac cycles after appearance of contrast bubbles in the right heart (6, 9, 12, 23). Because saline contrast bubbles small enough to travel through even the largest pulmonary capillaries (<13 μm) have a life span less than three cardiac cycles (even at maximal exercise), transpulmonary passage of these bubbles must occur via large diameter intrapulmonary arteriovenous shunt pathways (2, 15, 18, 28, 29). Of note, saline contrast bubbles can be forced through the normal pulmonary microcirculation using a firmly wedged pulmonary artery catheter with a perfusion pressure of 300 Torr. However, these extreme pulmonary driving pressures do not occur in healthy exercising humans making this an unlikely explanation for the transpulmonary passage of saline contrast bubbles (15).

Consistent with the human, intrapulmonary arteriovenous shunting occurs in dogs. Intravenously injected 25 μm microspheres were found in the tissue and arterial blood of the systemic circulation during exercise but not at rest (22). Dogs were confirmed not to have intracardiac shunts and with an established diameter of 25 μm, these microspheres bypassed the pulmonary capillaries via arteriovenous vessels at least 25 μm in diameter.

Arteriovenous vessels would divert deoxygenated blood away from pulmonary capillaries. If a significant amount of cardiac output was diverted through these pathways when mixed venous partial pressure of oxygen is reduced, such as during exercise, then pulmonary gas exchange as evaluated by the alveolar to arterial oxygen difference (AaDO₂) would be impaired. With the use of the Bergman equation, only a 2% shunt of cardiac output would be required to increase AaDO₂ during exercise (11). Indeed, a 1.4 ± 0.8% shunt has been calculated in exercising dogs (22) and exercise-induced intrapulmonary arteriovenous shunting is correlated to AaDO₂ in healthy humans (23), suggesting these vessels may play an important role in pulmonary gas exchange impairment during exercise.

Based on the amount of morphological and functional anatomic-based data supporting the existence of inducible intrapulmonary shunts, it may be somewhat surprising that work using the 100% O₂ technique or the multiple inert gas elimination technique (MIGET) has not detected these pathways in healthy humans during exercise(see Ref. 5 for complete list of references), suggesting that shunts are imaginary. However, this discrepancy may be explained by precapillary gas exchange and the vasomotor effect of O₂ on the pulmonary circulation, both of which are critically dependent on concentration gradient and physical properties of the gas (Fig. 1, B and C).

Conhaim and Staub (3) demonstrated precapillary O₂ exchange in rapidly frozen cat lungs. In these studies, oxyhemoglobin saturation in 500 μm pulmonary arteries from lungs ventilated with room air were as high as 77% at the perimeter of the blood vessel, while blood at the core of the vessel was as low as 47% saturated with oxygen (3). The size of the perimeter of the blood vessel becoming oxygenated increased from 62 μm in normoxia to 401 μm in lungs ventilated with 100% O₂. The authors calculated that in normoxia, mixed venous blood may be as much as 15% oxygenated by the time it reaches the alveolar capillary, while blood would be fully oxygenated before reaching the capillary when breathing 100% O₂. Importantly, precapillary gas exchange of both O₂ and N₂ have also been demonstrated in humans (10, 20). These studies demonstrated that precapillary O₂ exchange occurs in normoxia, with a greater O₂ exchange occurring in larger vessels with an increased fraction of inspired oxygen. Accordingly, in subjects breathing 100% O₂ during exercise, O₂ exchange would occur proximal to the intrapulmonary arteriovenous pathways(3, 10, 20), and thus these vessels would not be “seen” as true shunt, as the calculated venous admixture (Qs/Qt) would be minimal (Fig. 1C).

Furthermore, a fundamental assumption of the 100% O₂ technique is that the elevated level of inspired oxygen does not have an effect on the pulmonary microcirculation. This does not appear correct, as we recently demonstrated that exercise-induced intrapulmonary arteriovenous shunting can be eliminated in subjects within 2 min of breathing 100% O₂ (13). These findings raise a concern for the use of the 100% O₂ technique as a valid method for assessing exercise-induced arteriovenous shunt in normoxia and may explain why venous admixture decreases from 3.5% to 0.5% of cardiac output when subjects breathe 100% O₂ during exercise (26).

With respect to the MIGET, even a small degree of precapillary gas exchange (i.e., restricted to the perimeter blood of a 500 μm vessel) would allow elimination of low-soluble inert gas within the arteries/arterioles. Therefore, if low-solubility gases are exiting the blood within the pulmonary artery upstream of the capillary beds, then these inert gases would never even reach smaller functional arteriovenous shunt vessels (>25 to 50 μm), and thus these anatomical shunts would appear imaginary to those using the MIGET. In addition, intrapulmonary arteriovenous pathways themselves may participate in limited gas exchange restricted to their perimeter blood, which would allow some deoxygenated core blood to bypass the pulmonary capillary bed in normoxia, but not be recorded as true mixed venous shunt for the same reasons detailed above (7, 21).

More than 100 years of anatomic data document large diameter arteriovenous pathways in the lung. Recent work has simply demonstrated that these vessels are not always open but become functional under specific conditions, such as during exercise. Is exercise-induced intrapulmonary shunting real? When using anatomic-based techniques (microbubbles and microspheres) they are indeed real.

GRANTS

This work was funded by AHA Scientist Development Grant (to A. T. Lovering), AHA Grant-In-Aid 0550176Z (to M. W. Eldridge) and Canadian Institutes of Health Research New Investigator Award (to M. K. Stickland).