Precapillary pulmonary gas exchange is similar for oxygen and inert gases.

Abstract

Some pulmonary gas exchange is known to occur proximal to the pulmonary capillary, but the magnitude of this gas exchange is uncertain, and it is unclear whether oxygen and inert gases are similarly affected. This has implications for measuring shunt and associated gas exchange consequences. By measuring respiratory and inert gas levels in the proximal pulmonary artery (P), a distal pulmonary artery 1cm proximal to the wedge position (using a 5-F catheter) (D), and a systemic artery (A), we evaluated precapillary gas exchange in 27 paired samples from seven anaesthetized, ventilated canines. Fractional precapillary gas exchange (F) was quantified for each gas as F =(P–D)/(P–A). The lowest solubility inert gases, sulfur hexafluoride (SF₆) and ethane were used, because with higher solubility gases, the P-A difference is sufficiently small that experimental error prevents accurate assessment of F. Distal samples (n=12) with oxygen (O₂) saturation values that were, within experimental error, equal to or above systemic arterial values, suggestive of retrograde capillary blood aspiration, were discarded, leaving 15 for analysis. D was significantly lower than P for SF₆ (D/P=88.6±18.1%, p=0.03) and ethane (D/P=90.6±16.0%, p=0.04), indicating partial excretion of inert gas across small pulmonary arteries. Distal pulmonary arterial O₂ saturation was significantly higher than proximal (73.7±7.0% vs. 69.1±5.0%, p=0.03). Fractional precapillary gas exchange was similar for SF₆, ethane and O₂ (0.12±0.19, 0.12±0.20 and 0.19±0.16, respectively, p=0.54). Under these experimental conditions, 12–19% of pulmonary gas exchange occurs within the small pulmonary arteries and the extent is similar between oxygen and inert gases.

INTRODUCTION

The conventional view of pulmonary gas exchange is that it occurs in the pulmonary capillary bed. However, the exchange of both O₂ and N₂ (an inert gas) in vessels proximal to the pulmonary capillaries was demonstrated many years ago in humans (Jameson, 1963; Sobol et al., 1963) and animals (Sobol et al., 1963; Conhaim & Staub, 1980). Work using intravital microscopy to estimate the magnitude of precapillary O₂ exchange in a murine model demonstrated that as much as 50% of oxygen exchange may occur in the smallest precapillary vessels (Tabuchi et al., 2013). Precapillary gas exchange, if present, has implications for measuring shunt, and the associated effects of shunt on pulmonary gas exchange, since blood in an intrapulmonary vessel undergoing precapillary gas exchange will exchange some gas even when there is a downstream shunt preventing any further gas exchange. This has the potential to reduce the estimate of shunt using gas exchange methods.

The issue of precapillary gas exchange has received recent attention (Hopkins et al., 2009a, b; Lovering et al., 2009b, a; Stickland et al., 2013) because of a considerable body of work showing the presence of intrapulmonary arteriovenous anastomoses detected with agitated saline contrast echocardiography in human subjects. Approximately 28% of subjects have evidence for intrapulmonary arteriovenous anastomoses at rest (Elliott et al., 2013) and most show evidence of these connections being functional during exercise (Eldridge et al., 2004; Stickland et al., 2004; Lovering et al., 2006). In addition to exercise, interventions such as hypoxia, and inotrope infusion (Lovering et al., 2006; Stickland et al., 2006; Bryan et al., 2012; Laurie et al., 2012; Elliott et al., 2014) have been shown to increase the transpulmonary passage of agitated saline contrast, interpreted as recruiting these pathways. Intrapulmonary arteriovenous connections have also been shown in canines during exercise using microspheres (Stickland et al., 2007). These anatomical arteriovenous pathways potentially shunt blood, bypassing the pulmonary capillaries, and on this basis could contribute deoxygenated blood into the systemic circulation, lowering the PO₂ and increasing the AaDO₂.

However, over many decades, gas exchange studies using the Multiple Inert Gas Elimination Technique (MIGET) have never shown significant shunts in normal young human subjects at rest or during exercise (Hammond et al., 1986; Wagner et al., 1986; Hopkins et al., 1994; Podolsky et al., 1996; Hopkins et al., 1998; Rice et al., 1999; Olfert et al., 2004; Jonk et al., 2007). Similarly, significant shunts at rest or during exercise have not been shown in animal models such as canines (Hsia et al., 1990), horses (Wagner et al., 1989) and pigs (Hopkins et al., 1999).

It has been proposed that the reason MIGET fails to detect shunts despite pulmonary transmission of small bubbles is that the inert gases used undergo precapillary diffusive gas exchange (Stickland et al., 2004; Stickland & Lovering, 2006; Lovering et al., 2009a; Stickland et al., 2013). Thus, even with a downstream shunt from anatomical arteriovenous pathways, inert gases may diffuse sufficiently from precapillary blood into alveolar gas to obscure detection of the downstream shunt. If O₂ does not undergo the same degree of precapillary exchange, this could then explain why anatomical arteriovenous pathways might affect arterial oxygenation without MIGET detecting them as shunts.

The notion that O₂ would not undergo as rapid diffusive exchange as inert gases is based on fundamental principles: The completeness of diffusive exchange depends on the ratio of lung diffusive to perfusive conductance for the gas in question (Piiper & Scheid, 1981). As detailed below, this ratio is determined (in a given lung) mostly by ratio of the solubility of the gas in the hemoglobin-free tissue separating gas and blood to its solubility in blood. This solubility ratio is about 30 times greater for all inert gases than for O₂. Consequently, inert gases may differ from respiratory gases in the extent of precapillary gas exchange because of a much faster rate of diffusion equilibration (Wagner, 1977). In the normal human (and canine) lung, there is usually sufficient time for both inert gases and O₂ to come to diffusion equilibration over the transit of a red cell through the capillaries so that the different rates of equilibration are of no consequence to the composition of arterial blood (Wagner, 1977). In this case, MIGET will then correctly predict O₂ exchange. Similarly, if there were sufficient time for both O₂ and inert gas diffusive exchange to reach equilibration in the precapillary exchange vessels, one would not expect different degrees of precapillary exchange. In this scenario, MIGET would again be expected to appropriately predict the gas exchanges consequences of precapillary exchange for O₂. Conversely if precapillary exchange was adequate for inert gas exchange, but insufficient for O₂ equilibration, MIGET would fail to recognize anatomical arteriovenous pathways as causing a gas exchange impairment for O₂. The consequence of this would be that MIGET would underestimate the effects of shunt on the alveolar-arterial difference for oxygen (AaDO₂) and arterial oxygenation. However, the extent of precapillary gas exchange for inert gases compared to oxygen has not been previously determined.

Accordingly, the purpose of the present study was to assess precapillary gas exchange for oxygen and inert gases. We compared O₂ saturation and inert gas concentrations measured in the proximal pulmonary artery to values obtained simultaneously from a precapillary sampling site as close to the alveoli as possible by sampling from the distal port of a 5-F gauge Swan-Ganz catheter positioned 1cm proximal to the wedge position, in an intact canine animal model under experimental conditions that have been previously shown to recruit intrapulmonary arteriovenous anatomic pathways and increase venous admixture/shunt fraction (Bryan et al., 2012; Elliott et al., 2014). We hypothesized that precapillary gas exchange for oxygen would be observed, as it has been before, and therefore relative to blood in the main pulmonary artery, the distal blood sample would show an increase in O₂ saturation, while inert gas concentrations in the distal sample would be lower and demonstrate partial elimination. The second, and more important aim was to determine whether the extent of precapillary gas exchange for inert gases was different than that of O₂. We hypothesized that precapillary gas exchange would be greater for inert gases than for O₂ due to their more complete diffusive equilibration as reasoned above. A companion paper (Stickland et al., Submitted) reports the relationship between blood flow through arteriovenous anatomic pathways measured by microspheres, and shunt measured by MIGET as well as a qualitative assessment of these pathways with agitated saline contrast in the same animals.

METHODS

Ethical approval

All surgical and experimental procedures were approved by the University of Alberta Research Ethics Office (Protocol #AUP00001296), and studies were conducted in accordance with the American Physiological Society’s “Guiding Principles in the Care and Use of Animals”.

Theoretical considerations

1. Factors affecting diffusional exchange of gases in the lung

Diffusional transport across the blood gas barrier depends on several factors, some that are the same for all gases, (such as the time available for gas exchange to occur and the physical properties of the barrier, which include thickness and surface area available for diffusion), as well as some that are dependent on the unique physical properties of each gas (Johnson et al., 1996; Stickland et al., 2013). The latter include a) solubility of the gas in the blood gas barrier, (because the rate of transfer of a gas across the blood gas barrier is directly proportional to its solubility in the barrier), and b) the effective solubility of the gas in blood (which dictates how much the partial pressure of the transferred gas rises (gas uptake) or falls (gas elimination) for a given amount of gas transferred into or out of the blood. The ratio of these two solubilities (barrier and blood) in the blood defines in large part, for any gas, the rate of diffusion equilibration (Piiper & Scheid, 1981). Because hemoglobin has little effect on inert gas solubility, the barrier:blood solubility ratio for all inert gases is ~1. However, this is not true of oxygen. Oxygen has a low physical solubility, of about 0.003 ml ˑ dl⁻¹ ˑ mmHg⁻¹ in the Hb-free blood-gas barrier, but much higher effective solubility (i.e. the average slope of the dissociation curve) in blood because of Hb. This average slope is normally about 0.09 ml ˑ dl⁻¹ ˑ mmHg⁻¹, making the barrier:blood ratio for O₂ some 30-fold lower (i.e. 0.003 / 0.09 = 0.03) than for inert gases (Wagner, 1977). This renders O₂ more vulnerable to diffusion limitation than inert gases by more than an order of magnitude (Hopkins & Wagner, 2017). Therefore, any precapillary diffusive gas exchange would be expected to favor inert gases over O₂. Consequently, greater vulnerability of O₂ to diffusion limitation in small precapillary pulmonary arteries might explain why vascular pathways identified by bubble transmission could give rise to a gas exchange impediment for O₂ that could be underestimated by methods using inert gases.

The same argument holds for CO₂, with the ratio of solubility in the barrier to solubility in the blood of 0.08. This is ~2–3 fold greater than for O₂, but still much less than for inert gases. Thus, CO₂ may undergo greater precapillary gas exchange than O₂, but less than for inert gases. However as outlined below, precapillary gas exchange will be difficult to detect for CO₂.

2. Choice of inert gases against which to compare precapillary O₂ exchange

Since all inert gases have a ratio of barrier:blood solubility ~1, this would initially suggest that any of the six inert gases used in MIGET could be used to compare precapillary exchange between inert gases and O₂. However, when the systemic arterial inert gas tension is only minimally to moderately lower than that in the pulmonary artery – which is the expectation for gases of moderate to high solubility – attempting to measure how much of that difference occurs prior to the pulmonary capillaries may be problematic. This is because the random experimental error of inert gas measurement may exceed the expected proximal to distal pulmonary arterial inert gas tension difference. As an example, consider a gas such as ether, with blood-gas partition coefficient of about 10, being eliminated by a homogeneous lung with normal ventilation-perfusion $({\dot{V}}_{\text{A}}/\dot{Q})$ ratio of 1.0. If inflowing proximal pulmonary arterial (i.e. mixed venous) inert gas tension, P $\overline{v}$, is normalized to 100 units, systemic arterial tension, Pa, is given by:

$$\text{Pa}=\text{P}\overline{\text{v}}·(\lambda /(\lambda +{\dot{\text{V}}}_A/\dot{\text{Q}})$$ Equation 1

Where λ is the blood:gas partition coefficient. Solving for λ = 10, the expected arterial tension is 91 (i.e. 100 × 10/(10+1)).

Now assume 20% of the gas exchange for ether is occurring prior to the pulmonary capillary. If the proximal gas tension is 100, ether tension in the distal pulmonary artery (i.e. precapillary) would be 20% of (100–91) lower than 100, or 98.2 units. The measurement task is to distinguish the distal pulmonary artery value of 98.2 from the proximal pulmonary arterial value of 100, a difference of just 1.8%, in the face of a typical measurement coefficient of variation for inert gas of 3% (i.e. 3 units on the current scale) (Hopkins & Wagner, 2017). Note that both the proximal and distal values will be independently subject to this 3% coefficient of variation. Thus, the high solubility gases lack sensitivity to detect the difference between 100 +/− 3 and 98.2 +/− 3 resulting from 20% precapillary gas exchange. If only 10% of gas exchange occurs in the small pulmonary arteries, the task would be correspondingly harder. In contrast, the low solubility gas sulfur hexafluoride (SF₆) has a blood-gas partition coefficient of about 0.005 (Hopkins & Wagner, 2017) and in a homogeneous lung with a normal V. _A/Q. ratio of 1.0, less than 1% of SF₆ is retained in the systemic arterial circulation. Thus, if precapillary gas exchange accounted for 20% of the total pulmonary gas exchange, the concentration of SF₆ in the distal pulmonary artery would be 19.9% lower than that of the proximal pulmonary artery (i.e. 20% of (1 – 0.01)), which would be easily detectable, even with the 5% coefficient of variation for this gas.

Accordingly, only the least soluble inert gases can be used to test the hypothesis that precapillary inert gas exchange for inert gases is relatively greater than that of O₂. Based on the above theoretical analysis, only SF₆ (λ ~ 0.005), ethane (λ ~ 0.1) and cyclopropane (λ ∼ 0.5) potentially offer sufficient sensitivity to detect 20% precapillary gas exchange in the face of experimental error. Since the magnitude of any precapillary gas exchange is unknown, we compared concentrations of all 6 gases used in MIGET in the proximal and distal pulmonary arteries for a statistically significant difference. Only SF₆ and ethane were significantly different between proximal and distal concentration and thus these two inert gases were used to compare the magnitude of precapillary inert gas to that of O₂.

3. Identification of precapillary O₂ and CO₂ exchange

The principles applied above to inert gases are equally important for both O₂ and CO₂ in order to evaluate the extent of precapillary gas exchange. Mixed venous (i.e. proximal pulmonary artery) O₂ saturation is about 75% while systemic arterial O₂ saturation is close to 100%, making a difference of about 25%. In the presence of 20% precapillary gas exchange, precapillary O₂ saturation would be expected to be higher than in the distal pulmonary artery by about 5 percentage points (20% of (100–75)), in the face of a measurement coefficient of variation of PO₂ <2% in our laboratory. Thus, the expected effect size for O₂ should be readily identifiable if present. This is consistent with its effective solubility equating to that of a medium to low solubility gas (i.e. slope = (CaO₂ – C$\overline{\text{v}}$O₂)/(PaO₂ – P$\overline{\text{v}}$O₂) = (20–15)/(95–40) = 0.09 ml/dl/mmHg, corresponding to an effective blood:gas partition coefficient, $\lambda$ ~ 0.64) (Wagner, 1977).

For CO₂, the same analysis shows that the same degree of precapillary exchange as modeled above (i.e. 20%) will not be detectable, since arterial CO₂ content is typically only about 4 ml/dl lower than mixed venous, with corresponding PCO₂ values of about 40 and 45 mm Hg respectively. This is also consistent with the higher effective solubility of CO₂, with λ ~ 4.3 (Wagner, 1977), similar to that of a moderately high solubility gas such as ether as described above.

Study Overview

As noted above, this work was part of a larger project that also examined pulmonary gas exchange and arteriovenous shunt (Stickland et al., Submitted). For completeness of the current manuscript, there is some text duplication with the companion manuscript (Stickland et al., Submitted) and this text is indicated by italics. A total of seven mixed-breed canines weighing between 19 and 25 kg were studied (Marshall BioResources). For logistical reasons, the first three animal investigations were conducted several months earlier than the last four.

The experiments were conducted in a temperature-controlled (21–23°C), ventilated room. Once anaesthetized, ventilated, and instrumented, animals were studied under baseline and while receiving either 2 μg ˑ kg⁻¹ ˑ min⁻¹ dopamine or 10 μg ˑ kg⁻¹ ˑ min⁻¹ dobutamine. Data were obtained 15 minutes after the start of each 30-minute pharmacological infusion. Steady-state was confirmed within each condition by stable oxygen consumption and heart rate for at least five minutes prior to data collection. Under each condition, blood was simultaneously drawn from: 1) femoral artery (i.e. systemic arterial), 2) the distal port of a Swan-Ganz catheter where the end of the catheter was placed 1–2 cm distal to the right ventricular/pulmonary artery interface (i.e. proximal pulmonary artery), 3) the distal port of a 5-French Swan-Ganz catheter, where the end of the catheter was placed 1 cm proximal to the wedge position (i.e. a distal small pulmonary artery). Data collected included blood gases and concentrations of inert gas, pulmonary arterial, pulmonary arterial wedge, and systemic arterial pressures; metabolic rate, ventilation and cardiac output. Shunt determined by microspheres and transpulmonary bubble transmission by contrast echocardiography were also measured and are reported elsewhere (see Stickland et al. Submitted).

Anaesthesia

Animals were premedicated with acepromazine (0.05 mg ˑ kg⁻¹) and hydromorphone (0.1 mg ˑ kg⁻¹) injected intramuscularly. In the first three animals, surgical plane anaesthesia was maintained by I.V. propofol (1 mg ˑ kg⁻¹ I.V. bolus, then maintenance doses of 0.2 mg ˑ kg⁻¹ ˑ min⁻¹). In the last four animals, surgical plane anaesthesia was maintained by pentobarbital (20–30 mg ˑ kg⁻¹ I.V. bolus, then maintenance doses of 1–5 mg ˑ kg⁻¹ ˑ hour⁻¹). Veterinarian staff continuously monitored anaesthesia level, and following completion of data collection, animals were euthanized with 120 mg ˑ kg⁻¹ I.V. pentobarbital.

Instrumentation

Following induction of anaesthesia, an endotracheal tube was inserted and animals were ventilated on room air (PEEP = 5 cm H₂0, target PaCO₂ ~ 35 mmHg). A positive end-expiratory pressure of 5 cm H2O was used throughout, and frequent sighs were conducted between experimental conditions in an attempt to prevent atelectasis. Four venous catheters were inserted, one in each hind limb for 1) infusion of anaesthetic, 2) infusion of saline containing inert gases, 3) infusion of dobutamine, 4) back-up catheter. The femoral artery was cannulated with the catheter (A) advanced into the abdominal aorta for peripheral arterial blood sampling. Two Swan-Ganz catheters were introduced, one through each jugular vein. The first Swan-Ganz catheter (7-F, 2.3 mm) was positioned approximated 1–2 cm distal to the right ventricular/pulmonary artery interface as confirmed by direct pressure monitoring (proximal pulmonary artery, P). The second Swan-Ganz catheter (5-F, 1.7 mm) was positioned approximately 1 cm proximal to the wedge position (distal pulmonary artery, D). The initial wedge position was confirmed by direct pressure monitoring, the catheter was then retracted 1 cm and balloon deflated for 5 minutes prior to any sampling. Positions of both Swan-Ganz catheters were confirmed prior to every data collection period.

Cardiopulmonary data

Expired gases were passed through heated tubing to a mixing chamber set to maintain temperature > 37° C in order to avoid condensation of water vapor. Expired O₂ and CO₂ were measured (Analyzers 17625/17630; Vacumed, Ventura, CA USA), ventilation was determined by pneumotachometer (Series 3700A, Hans Rudolph Inc., Shawnee, KS USA) and oxygen consumption and carbon dioxide production were calculated (Powerlab, AD Instruments, Colorado Springs, CO USA). Pressure transducers (Surgivet Advisor, Smiths Medical, Dublin OH USA) were zeroed to the level of the right atrium. Mean arterial, pulmonary arterial, and pulmonary arterial wedge pressures were recorded immediately before each set of inert-gas measurements. Cardiac output was calculated from the systemic arterial - pulmonary arterial O₂ concentration difference and O₂ consumption using the Fick equation.

Multiple Inert Gas Elimination Technique

The multiple inert-gas elimination technique (MIGET) was applied as previously described (Wagner et al., 1974a; Wagner et al., 1974b; Wagner et al., 1975; Dueck et al., 1978). The inert-gas solution was prepared in normal saline and infused for 20 min before collection of baseline samples. The infusion rate of MIGET solution was set to 0.25 mL ˑ min⁻¹ per L ˑ min⁻¹ of ventilation for an average infusion rate of ~1 mL ˑ min⁻¹ in these animals. The total volume of fluid infused during the study from all sources was 1 L over a period of 3–4 hours.

After confirmation of steady state conditions, quadruplicate 15 mL samples of mixed expired gas (used in shunt calculations, reported in (Stickland et al., Submitted)) as well as duplicate 6 mL samples of: 1) systemic arterial blood, 2) proximal pulmonary artery blood, and 3) distal pulmonary artery blood were obtained simultaneously in gas-tight glass syringes at each condition for measurement of the steady-state concentrations of the six inert gases by gas chromatography (model 5890A; Hewlett-Packard, Wilmington, DE). All blood samples were obtained using a slow withdrawal (~0.5 mL ˑ sec⁻¹) to minimize chance of retrograde flow from the pulmonary veins, which would contaminate the sample with capillary blood that had undergone gas exchange. Ventilation-perfusion distributions as well as right-to-left shunt were determined by MIGET as described (Hopkins & Wagner, 2017) and are reported elsewhere (Stickland et al., Submitted).

Blood Gas Measurements

Immediately following each blood sample for MIGET, 3 mL samples of: 1) systemic arterial blood, 2) proximal pulmonary artery blood, and 3) distal pulmonary artery blood were simultaneously obtained. Samples were immediately cleared of any bubbles and were maintained on ice until analyzed for PO₂, PCO₂, pH, and hemoglobin concentration using a blood-gas analyzer (ABL80 FLEX, Radiometer, Copenhagen, Denmark). For the present study, the coefficient of variation of measurement for PO₂ was 1.6% and 2.2% for CO₂. For each sample, O₂ saturation was calculated based on PO₂, PCO₂, pH, and rectal temperature (Kelman, 1966, 1967, 1968), and an assumed partial pressure of O₂ at 50% saturation of hemoglobin of 30 mmHg, as has been previously reported for canine blood (Cambier et al., 2004; Zaldivar-Lopez et al., 2011).

4. Effect Size and Number of Samples Required.

Power calculations were conducted with G*Power (Faul et al., 2007). In the present study we identified 20% precapillary gas exchange as being a meaningful difference in concentrations of gases between proximal and distal sampling sites. This figure was a conservative estimate based on the previous work of Tabuchi et al., (Tabuchi et al., 2013) showing 50% precapillary gas exchange. Given a coefficient of variation between repeated measures of oxygen concentration of 2% and an assumed modest correlation of 0.5 between proximal and distal samples, a sample size of 5 gives a power of >0.95 to detect a statistically significant difference between proximal and distal samples. To detect a difference between precapillary exchange between oxygen and inert gases, we again assumed that 20% precapillary gas exchange of inert gases was meaningful and that precapillary gas exchange for oxygen was half of that (10%), with a coefficient of variation of 50% (also very conservative) for both measures. A sample size of 7 provided a power of >0.8 to detect a statistically significant difference, p = 0.05, 2 tailed. Since withdrawal of blood from a distal catheter has the potential to create retrograde flow by aspiration across the alveolar-capillary interface, thus reducing gas concentrations below arterial and thus introducing erroneous data (see below), we sought to withdraw at least two paired samples from each animal.

Data and Statistical Analysis

All statistical analyses were performed using SPSS Statistical software v. 24.0.0.0 (IBM Corp™, Armonk, NY, USA). For all inferential analyses, the probability of Type I error was set at 0.05. The highest concentration of inert gas is contained within the blood leaving the right ventricle and entering the proximal pulmonary artery (i.e. prior to any excretion while traversing the lungs) because MIGET gases are dissolved and infused intravenously. Systemic arterial blood contains inert gas that has not been excreted after passage through the pulmonary circulation and the gas exchanging portions of lungs, and thus has the lowest concentrations among the blood samples. As the measurement of the inert gas concentrations for MIGET by gas chromatography is not made in absolute terms, but rather in arbitrary units that are expressed as ratios, distal pulmonary artery and systemic arterial inert gas concentrations were expressed, for each gas, as percentages of its concentration in the proximal pulmonary artery. Gas concentration between proximal and distal were evaluated by repeated measures ANOVA with Greenhouse-Geisser correction for violation of sphericity applied when applicable.

Fractional precapillary gas exchange (F) for both inert gas and O₂ was determined by the following equation:

$$\text{F = }\left(\text{P – D}\right)\text{/}\left(\text{P-A}\right)$$ Equation2

where for each inert gas, P = concentration obtained from the proximal pulmonary artery sample, D = concentration obtained from the distal pulmonary artery sample, and A = concentration obtained from the systemic arterial sample. For O₂, P, D and A were represented by O₂ saturation values because of the nonlinear O₂ hemoglobin dissociation curve. The comparison of whether fractional precapillary gas exchange for SF₆, ethane and O₂ was significantly different from zero was evaluated by Welch’s unequal variance t-test. Comparison of the extent of fractional precapillary gas exchange between SF₆, ethane and O₂ was evaluated by repeated measures ANOVA. Pearson product moment correlation was used to evaluate the relationships between fractional precapillary gas exchange and cardiac output.

RESULTS

A total of 27 paired proximal and distal pulmonary artery samples were collected. Twelve of the distal pulmonary artery samples were found to have O₂ saturation values that were, within experimental error, above or equal to the corresponding systemic arterial blood sample. These entire samples (i.e. O₂ and inert gas data) were removed from analysis, as they reflect inadvertent retrograde flow by aspiration of blood from the alveolar-capillary interface, leaving 15 samples for analysis (7 baseline, 2 during dopamine, and 6 during dobutamine). Figure 1 displays the oxygen saturation from the distal samples, plotted against the corresponding arterial samples showing a clear demarcation between the retained samples and the discarded samples. There were no significant differences in fractional precapillary gas exchange across the three conditions (baseline, dopamine, dobutamine) for SF₆ (0.04 ± 0.11 vs. 0.11 ± 0.07 vs. 0.20 ± 0.25, p = 0.32), ethane (0.07 ± 0.13 vs. 0.08 ± 0.08 vs. 0.12 ± 0.20, p = 0.57) or O₂ saturation (0.16 ± 0.21 vs. 0.02 ± 0.18 vs. 0.30 ± 0.33, p = 0.41), and thus the samples from all the conditions were combined. Therefore, data reported are from 15 samples from seven animals.

Figure 1.

Distal pulmonary arterial saturation (SO₂) and systemic arterial saturation for samples that were retained vs. discarded. Discarded distal samples had SO₂ values that were, within experimental error, equal to or HIGHER than the corresponding systemic arterial SO₂ value, suggesting retrograde sampling of alveolar blood. Dotted line is the line of identity.

Hemodynamic Data and overall gas exchange data

At baseline (n = 7), mean (±SD) heart rate, stroke volume and cardiac output were 95 ± 19 beats ˑ min⁻¹, 16 ± 3 mL, and 1.5 ± 0.4 L ˑ min⁻¹ respectively. Mean systemic arterial blood pressure, pulmonary arterial blood pressure and pulmonary arterial wedge pressure were 74 ± 17 mmHg, 13 ± 3 mmHg and 10 ± 3 mmHg respectively. Oxygen consumption was 65 ± 7 mL ˑ min⁻¹, and CO₂ production was 51 ± 7 mL ˑ min^-1. With dopamine (n=2), mean (±SD) heart rate, stroke volume and cardiac output were 123 ± 22 beats ˑ min⁻¹, 12 ± 1 mL, and 1.4 ± 0.1 L ˑ min⁻¹ respectively. Mean systemic arterial blood pressure, pulmonary arterial blood pressure and pulmonary arterial wedge pressure were 80 ± 20 mmHg, 13 ± 1 mmHg and 7 ± 3 mmHg respectively. Oxygen consumption was 65 ± 1 mL ˑ min⁻¹, and CO₂ production was 43 ± 1 mL ˑ min^-1. With dobutamine (n=6), mean (±SD) heart rate, stroke volume and cardiac output were 194 ± 14 beats ˑ min⁻¹, 17 ± 5 mL, and 3.2 ± 1.0 L ˑ min⁻¹ respectively. Mean systemic arterial blood pressure, pulmonary arterial blood pressure and pulmonary arterial wedge pressure were 84 ± 29 mmHg, 21 ± 2 mmHg and 10 ± 2 mmHg respectively. Oxygen consumption was 99 ± 16 mL ˑ min⁻¹, and CO₂ production was 66 ± 11 mL ˑ min^–1.

As expected, inert gas concentrations sampled from the systemic artery (A) were significantly lower than proximal pulmonary arterial values (P) for SF₆ (A/P = 2.5% ± 1.5%, p<0.001), ethane (A/P = 19.1% ± 6.8%, p<0.001), cyclopropane (A/P = 56.3% ± 11.2%, p<0.001), isoflurane (A/P = 76.0% ± 9.6%, p<0.001), and ether (A/P = 96.1% ± 6.4%, p = 0.03). Acetone was not significantly different between proximal and arterial (A/P = 101.1% ± 6.3%, p = 0.43), consistent with the very high solubility of acetone and experimental noise. Systemic arterial values for PO₂ (89.8 ± 9.9 mmHg, p<0.001), O₂ saturation (94.7 ± 2.1%, p<0.001), and PCO₂ (33.7 ± 5.0 mmHg, p<0.001) were also all significantly different compared to proximal values.

Comparisons of Proximal (P) and Distal (D) Pulmonary Artery Inert Gas and Oxygen Data

Distal pulmonary artery inert gas concentrations relative to those in the proximal pulmonary artery, as a function of inert gas solubility, are reported for all inert gases in Figure 2. Inert gas concentrations in the distal pulmonary artery were significantly lower than in the proximal pulmonary artery for SF₆ (D/P, % = 88.6 ± 18.1%, p = 0.03) and ethane (D/P, % = 90.6 ±16.0%, p=0.04). However, cyclopropane (93.7 ± 12.4%, p = 0.07), isoflurane (96.2 ± 11.0%, p = 0.21), ether (98.8 ± 6.2%, p = 0.48) and acetone (101.8 ± 6.9%, p = 0.34) distal pulmonary artery data were not significantly different from those in the proximal pulmonary artery, consistent with our theoretical analysis. When the mean inert gas concentration obtained from the distal pulmonary artery (expressed relative to that in the proximal pulmonary artery) was graphed as a function of the inert gas blood gas partition coefficient, a greater reduction in distal inert gas concentration occurs in the lower solubility gases (see Figure 2), consistent with principles of solubility-dependence of gas exchange.

Figure 2.

Box and whisker plot of distal pulmonary artery inert gas concentrations relative to those in the proximal pulmonary artery, as a function of inert gas solubility.

NOTE: Relative concentration = Distal concentration / Proximal concentration x 100. Dashed line = mean, solid line = median. The box represents observations between the 25–75^th percentile, the whiskers represent the 10–90^th percentile, and the dots the 5–95^th percentile. * indicates where distal values are significantly different from proximal values (p = <0.05).

Figure 3 reports individual and mean proximal vs. distal values for SF₆, ethane, O₂ saturation and O₂ partial pressure. Arterial O₂ saturation was significantly increased from proximal to distal pulmonary artery (proximal: 69.0 ± 4.9%, distal: 74.1 ± 6.8 %, p = 0.01). Similarly, PO₂ was significantly increased from proximal to distal pulmonary artery (proximal: 46.2 ± 6.4 mmHg to distal: 48.5 ± 7.9 mmHg, p = 0.02), see Figure 3. PCO₂ was not significantly different from proximal distal pulmonary artery (proximal: 36.8 ± 4.7 mmHg to distal: 35.9 ± 4.5 mmHg, p = 0.15), consistent with the higher effective solubility of CO₂ as discussed previously.

Figure 3.

Grouped mean (±SD) and paired individual proximal vs. distal values for sulfur hexafluoride (SF₆), ethane, O₂ saturation, O₂ partial pressure. NOTE: For SF₆ and ethane the proximal concentration was normalized to 100%. Reported p value is the comparison between proximal and distal values.

Fractional precapillary gas exchange (F)

Fractional precapillary gas exchange was significantly greater than zero for SF₆: 0.12 ± 0.19 (p = 0.03), ethane: 0.12 ± 0.20 (p = 0.04), and O₂: 0.19 ± 0.26 (p = 0.01). Values of F were similar in magnitude for SF₆, ethane and O₂ (P = 0.54, see Figure 4), suggesting that under these experimental conditions the amount of precapillary gas exchange is 12–19% of the total gas exchanged across the lung, and similar between the two inert gases and oxygen. Fractional precapillary gas exchange was not correlated to cardiac output for SF₆ (r = 0.16, p = 0.58), ethane (r = 0.03, p = 0.93), or O₂ (r = 0.38, p = 0.19).

Figure 4.

Box and whisker plot of fractional precapillary gas exchange for sulfur hexafluoride, ethane and O₂.

NOTE: No significant difference in fractional gas exchange was observed between the three gases (p=0.54). Dashed line = mean, solid line = median. The box represents observations between the 25–75^th percentile, the whiskers represent the 10–90^th percentile, and the dots the 5–95^th percentile.

DISCUSSION

The purpose of this study was to evaluate precapillary gas exchange for O₂, and inert gases by comparing concentrations measured in the proximal pulmonary artery to values obtained at the same time sampled from the distal port of a 5-F Swan-Ganz catheter placed 1cm proximal to the wedge position in the pulmonary artery. The concentrations of the low solubility gases SF₆ and ethane were 12% lower, and O₂ saturation 19% higher in a distal pulmonary artery compared to the proximal pulmonary artery. The extent of precapillary gas exchange was not significantly different between SF₆, ethane and O₂. These results suggest that there is both inert gas and O₂ exchange across small pulmonary arteries proximal to the pulmonary capillary bed, and that under the conditions of this study, the magnitude of precapillary gas exchange is similar between inert gases and O₂.

Evidence for the presence of precapillary gas exchange

There is considerable experimental evidence for precapillary gas exchange. As early as 1959, Weibel suggested that oxygen diffuses from the airways into all adjacent tissues (Weibel, 1959). Later, Jameson provided evidence of precapillary gas exchange in 27 patients by wedging a 6-F (2.5mm diameter) catheter containing a platinum electrode just proximal to the catheter tip within a branch of the right pulmonary artery (Jameson, 1963). When 100% oxygen was inspired, an increase in O₂ concentration was detected by the electrode in less than 1 second (i.e. faster than can be explained by recirculation) in most patients. Additional experiments were conducted while breathing 100% hydrogen transiently, where the catheter was pulled back 1–3 cm proximal to the wedge position, and in these experiments, there remained evidence of a rapid increase in hydrogen detected by the probe that could not be attributed to retrograde flow (Jameson, 1964), or to recirculation (Sobol et al., 1963; Jameson, 1964). The present study confirms the finding of pre-capillary gas exchange in canines, adding to the work in rodents (Tabuchi et al., 2013) and cats (Conhaim & Staub, 1980).

The magnitude of precapillary gas exchange for SF₆, ethane and oxygen

In rodents, approximately 50% of total O₂ uptake has been shown to occur before the red blood cell enters the pulmonary capillary (Tabuchi et al., 2013). In cat lung tissue samples, Conhaim and Staub (Conhaim & Staub, 1980) used microspectrophotometry to demonstrate oxygenation along the outer rims of cross-sectionally cut pulmonary arteries of ≥500 μm in diameter. From their data, the authors estimated that arterial blood may be as much as 15% oxygenated by precapillary gas exchange while breathing room air. However, the extent of precapillary gas exchange for inert gases and comparison with oxygen has not been previously evaluated.

In the present study in a large animal model, we show evidence of precapillary exchange of 12–19% for O₂ and inert gases when blood is sampled from pulmonary arteries of ~ 1.7 mm in diameter and smaller, consistent with the results of Conhaim and Staub (Conhaim & Staub, 1980). As mentioned previously, O₂ is susceptible to diffusion limitation as O₂ has relatively low solubility in the blood gas barrier, and a higher effective solubility (the average slope of the dissociation curve) in blood (Wagner, 1977). However, contrary to the original hypothesis, the data do not show a difference in the extent of precapillary gas exchange between inert gases and oxygen, and thus no evidence for diffusion limitation of precapillary gas exchange of O₂ compared to inert gases.

CO₂ and the inert gases cyclopropane and isofluorane

Unlike O₂ and low solubility inert gases, there was no difference in proximal vs. distal values for CO₂. As detailed in the methods section, the divergent response of CO₂ vs. O₂ can be explained by the relatively small difference in CO₂ levels between proximal and systemic arterial samples, reflecting that CO₂ behaves quantitatively like a moderately high solubility gas. In fact, the slope of the CO₂ dissociation curve corresponds to a partition coefficient similar to that of ether. Thus, CO₂ is unsuitable for measurement of fractional precapillary gas exchange because the measurement variability is greater than the expected amount of precapillary exchange. While precapillary exchange could not be statistically confirmed for CO₂, the magnitude of the changes between proximal and distal for CO₂, cyclopropane and isoflurane were similar to that for oxygen and the low solubility inert gases at 22 ± 89%, 15 ± 31%, and 11 ± 61%, respectively.

Implications of precapillary gas exchange on shunt estimation

In the absence of shunt, precapillary gas exchange has no consequence on the final concentration of systemic arterial blood, as full equilibration is expected within the pulmonary capillary of a normal healthy lung. However, this is not the case in the presence of intrapulmonary arteriovenous anastomoses, which have the potential to add deoxygenated blood to the systemic circulation. A companion paper documented the transpulmonary passage of 25 μm microspheres in these same animals, confirming the presence of small functional pulmonary arteriovenous connections (Stickland et al., Submitted). The present study suggests that the amount of blood flowing through functional arteriovenous connections would be underestimated similarly by both MIGET and O₂ methods to evaluate shunt, because similar relative amounts of inert gas and oxygen flowing through the intrapulmonary arteriovenous anastomoses would be lost by precapillary exchange. However, based on the results of this study, the degree of shunt underestimation would be small and likely below limits of detection. For example, a 1.0% shunt as measured by MIGET might actually be 12–19% higher – (i.e., it would actually be 1.1 – 1.2%) - due to precapillary gas exchange reducing the inert gas levels returning via shunt vessels to the arterial circulation by 12–19%. Similarly, a right-to-left shunt of 10% measured by MIGET would reflect an actual right-to-left shunt of 11–12%.

This conclusion is strengthened by the findings in the companion paper demonstrating that MIGET-determined right-to-left shunt was not measurably different to shunt quantified by transpulmonary passage of 25 μm microspheres (Stickland et al., Submitted) under the conditions of this study. These data suggest that precapillary gas exchange upstream of an arteriovenous connection cannot explain the finding that agitated saline contrast studies show transpulmonary passage of bubbles, suggesting the potential for deoxygenated blood to enter the systemic circulation, whereas gas exchange studies fail to detect a significant shunt. Importantly the companion study (Stickland et al., Submitted) indicates that the agitated saline contrast technique is highly sensitive to detect such shunts but non-specific, frequently returning positive contrast scores when the anatomical shunt (as evaluated by microspheres) or gas exchange shunt (as evaluated by MIGET) is very small.

In the presence of an intrapulmonary arteriovenous anastomosis, any upstream precapillary gas exchange means that blood passing through that anastomosis would no longer be that of mixed venous blood. If so, MIGET may not classify this as a shunt, but rather as coming from a region of low $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio (Stickland et al., 2004; Stickland & Lovering, 2006; Lovering et al., 2009a; Stickland et al., 2013). The question becomes: What would the effective $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio of the precapillary vessel/arteriovenous anastomosis complex have to be to be consistent with the degree of precapillary exchange measured? Equation 1 can be solved to estimate the $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio required to explain the 12% elimination of SF₆ observed in the present study. Using the partition coefficient of ~ 0.005 for SF6, and the measured fractional precapillary gas exchange of 0.12 (such that Pa/P $\overline{\text{v}}$ = 1 – 0.12 = 0.88), a $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio of ~0.0007 would be required to explain the 12% SF₆ precapillary gas exchange observed. Importantly, since the MIGET analysis includes all lung units with a $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio below 0.005 within the shunt compartment, such vessels would be categorized as regions of shunt by MIGET, and not as regions of low $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio. In keeping with this, we did not detect any regions of low $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio (i.e. 0.005–0.1) under any condition in our companion study (Stickland et al., Submitted). Consequently, even if precapillary gas exchange results in low $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ regions, in the present study, the $\dot{\text{V}}\text{A}/\dot{\text{Q}}$ ratio of such regions is so low as to be indistinguishable from shunt.

Limitations

Retrograde aspiration of blood from the alveolar capillary interface:

Every attempt was made to ensure sampling from the distal pulmonary artery was indeed blood from the associated small pulmonary artery, and not capillary blood sampled from retrograde flow. Prior to each sample being obtained, the catheter was advanced just enough to be wedged, and then carefully withdrawn 1 cm. Samples were drawn at a slow sampling rate (i.e. ~0.5 mL ˑ sec⁻¹) with the balloon on the Swan-Ganz catheter deflated, and any samples that yielded O₂ values higher than obtained from systemic arterial blood were discarded. As seen in Figure 1, there was a clear separation between the analyzed and discarded samples, supporting this approach. Of note, all discarded distal pulmonary arterial samples had an average SF₆ concentration of 2.3% ± 3.4% relative to proximal pulmonary arterial values, which is similar to the 2.1% ± 1.4 relative concentration contained within the systemic artery. In contrast, the distal pulmonary arterial samples analyzed had a relative SF₆ concentration of 88.6% ± 18.1% of that in the proximal pulmonary artery, showing a clear and substantial difference between the discarded capillary samples and the pulmonary artery samples analyzed. However, we cannot rule out the possibility of a small amount of admixture of blood from the capillary interface being sampled that would be sufficient to affect the results, but not so large as to lower the inert gas concentrations below, and the oxygen concentrations above, arterial levels, leading to their being retained for analysis. For this reason, the 12–19% estimate of precapillary gas exchange we report here is likely an upper boundary for this sampling site and these experimental conditions. In this case, the amount of shunt underestimation would be less than reported here, to the extent that the distal samples reflect admixture of alveolar blood in the sample rather than precapillary gas exchange.

Site of measurement and gas exchange:

Our measurement of precapillary gas exchange was in vessels that are relatively proximal to the pulmonary capillaries, and potentially much larger than the size of the anatomical arteriovenous pathways documented with agitated saline contrast. Given that the distal pulmonary artery catheter used to sample distal blood was ~1.7 mm in diameter suggests that it is sampling blood at or distal to an 8th generation pulmonary artery (Weibel & Gomez, 1962). While our sampling location is proximal to the pulmonary capillary, we have no way to know the precise location, as blood from vessels subtended by this vessel may be sampled in a non-uniform way. It is expected that, had we been able to sample still closer to the alveoli, fractional precapillary exchange of all gases would have been greater as pulmonary artery diameter and wall thickness decrease.

Comparison with exercise studies.

Since our data were acquired in animals that were under anaesthesia and with a relatively low cardiac output (mean 1.4 L ˑ min⁻¹), one might reason that the results of the present study might not apply during exercise or other conditions when the cardiac output is higher. During exercise, pulmonary blood flow increases dramatically, and whole lung (Hopkins et al., 1996) and pulmonary capillary transit times are reduced. Based on principles of diffusion, we would expect the amount of precapillary gas exchange to be lower during exercise than what was observed within the current experimental conditions, and therefore any underestimation of shunt would be less with exercise. Since oxygen is more vulnerable to diffusion disequilibrium than inert gases, it is possible that during exercise precapillary exchange of inert gases might exceed that of oxygen. However, in the present study, there was no significant correlation between the amount of precapillary gas exchange of O₂ or inert gas and cardiac output.

Blood gas analysis, estimation of P50 and calculation of oxygen saturation:

As a result of equipment failure within the animal surgical suite, it was not possible to immediately analyze blood gas samples. All blood gases were stored on ice and analyzed in random order. It is possible that some samples may have contained small micro bubbles within the samples which may have altered PO₂/SO₂; however, this would have theoretically affected all samples equally, and therefore represent random error, which would not explain the systematic difference in O₂ between proximal vs. distal samples. In addition, the coefficient of variations for PO₂ (1.6%) and PCO₂ (2.2%) were small, and similar to values seen following immediate analysis using the same blood gas analyzer (Tedjasaputra et al., 2015), suggesting no additional variance that would have been expected from inadvertent bubble contamination.

As inert gas retention and excretion are based on mass conservation principles, O₂ concentration (as reflected by O₂ saturation) was used to evaluate whether inert gas and O₂ exchange were similar or not. Use of O₂ partial pressure would have violated mass conservation rules due to the non-linear nature of the O₂Hb dissociation curve, and thus PO₂ was not appropriate for this comparison. The arterial blood gas analyzer (ABL80 FLEX, Radiometer, Copenhagen, Denmark), being configured for the human Hb spectrum, did not allow for co-oximetry of canine blood for direct O₂ saturation measurement. As such, O₂ saturation was calculated based on PO₂, PCO₂, pH, temperature (Kelman, 1966, 1967, 1968), and an assumed P50 (partial pressure of oxygen at 50% saturation) of hemoglobin of 30 mmHg (Cambier et al., 2004; Zaldivar-Lopez et al., 2011). It is unlikely that the difference in proximal vs. distal pulmonary artery O₂ saturation can be explained by inaccuracies of its determination, because the same P50 was used in estimating saturation from all blood samples, and thus they should all be affected similarly by any uncertainty.

Conclusion

The purpose of this study was to compare O₂ and inert gas concentrations measured in the proximal and distal pulmonary arteries to quantify precapillary gas exchange, and to determine whether O₂ and inert gases undergo precapillary exchange similarly. The main hypothesis was that precapillary inert gas exchange would be greater than for O₂ because of greater susceptibility of O₂ than inert gases to diffusion limitation. If supported, this could help explain why ultrasound studies show transpulmonary bubble passage suggesting a shunt, yet inert gas studies fail to show significant right-to-left shunt. However, the data demonstrated that the amount of precapillary gas exchange is small, and similar between oxygen and inert gases, such that 12–19% of gas exchange occurs across pulmonary arteries of ~1.7 mm in diameter. These findings suggest that a shunt may be slightly underestimated by both O₂-based and MIGET approaches. However, the similar degree of precapillary gas exchange between O₂ and inert gases suggests that if transpulmonary bubble transmission through arterio-venous anastomoses detected by ultrasound reflected a gas exchange shunt for O₂, MIGET would detect this shunt even in the presence of precapillary gas exchange, albeit with a small underestimation. Under the conditions of the present study (i.e. at rest and when the pulmonary circulation and cardiac output are modulated pharmacologically to increase transpulmonary bubble transmission), precapillary inert gas exchange is therefore not a tenable explanation for absence of shunt detected by MIGET when transpulmonary bubble passage suggests a shunt.

Key Points.

• Precapillary gas exchange for oxygen has been documented in both humans and animals.

• It has been suggested that if precapillary gas exchange occurs to a greater extent for inert gases than for oxygen, shunt and its effects on arterial oxygenation may be underestimated by the multiple inert gas elimination technique (MIGET).

• We evaluated fractional precapillary gas exchange in canines for O₂ and two inert gases, sulfur hexafluoride and ethane, by measuring these gases in the proximal pulmonary artery, distal pulmonary artery (1cm proximal to the wedge position), and systemic artery

• 12–19% of pulmonary gas exchange occurred within small (1.7 mm diameter or larger) pulmonary arteries, and was quantitatively similar for oxygen, sulfur hexafluoride and ethane

• Under these experimental conditions, this suggests only minor effects of precapillary gas exchange on the magnitude of calculated shunt and the associated effect on pulmonary gas exchange estimated by MIGET