Perfusion of Intrapulmonary Arteriovenous Anastomoses Is Not Related to VO 2max in Hypoxia and Is Unchanged by Oral Sildenafil

Introduction

The existence of intrapulmonary arteriovenous anastomoses (IPAVA), blood vessels that bypass the pulmonary microcirculation, has been well described in the literature (Tobin, 1966). More recently, researchers have attempted to quantify the perfusion of these IPAVAs (Q_IPAVA) using injected microspheres in dog (Stickland et al., 2007) and human (Lovering et al., 2007) lungs, and transthoracic contrast echocardiography (Eldridge et al., 2004), among other techniques (Duke et al., 2015). While some humans display evidence of Q_IPAVA at rest (Elliott et al., 2013; Duke et al., 2015), the majority of participants experienced increased Q_IPAVA during increasing levels of exercise (Eldridge et al., 2004; Lovering et al., 2009b) and normobaric hypoxia (Lovering et al., 2008a; Laurie et al., 2010), which can be almost completely abolished with hyperoxia (Lovering et al., 2008b).

The impact of IPAVA on gas exchange has not been demonstrated experimentally. Research using labeled aggregated albumin has estimated that IPAVAs are responsible for a mean 1.2 ± 0.8% anatomic shunt in humans (Lovering et al., 2009b) and 1.42 ± 0.75% shunt in dogs using microspheres (Stickland et al., 2007). The impact of this anatomic shunt on gas exchange is unclear. Research using the multiple inert gas elimination technique (MIGET) and 100% oxygen methods has not demonstrated an impact of intrapulmonary shunt on the alveolar-arterial oxygen difference (A-aDO₂) (Wagner et al., 1986; Hopkins et al., 1994, 1996), suggesting that the observed A-aDO₂ during exercise is the result of pulmonary ventilation/perfusion heterogeneity and diffusion limitation (Dempsey and Wagner, 1999; Hopkins, 2006).

While the hemodynamic significance of increased Q_IPAVA is unclear, its mechanism has also been the subject of extensive experimentation. Increases in Q_IPAVA during exercise and hypoxia were initially attributed to be the result of increases in pulmonary artery pressure (PAP) and cardiac output (CO) (Stickland et al., 2004); however, more recent evidence that changes in Q_IPAVA occur independent of PAP and CO manipulations suggests that direct sensing of hypoxia itself is a major controller (Lovering et al., 2015; Tremblay et al., 2015; Duke et al., 2016). Although increased Q_IPAVA is correlated with a reduction in S_PO₂ (Laurie et al., 2010), it is unclear the extent to which Q_IPAVA is affected by redistribution in pulmonary blood flow, the hypoxic (HYP) stimulus, an unknown stimulus, or some combination.

While the effect of exercise as a mediator of Q_IPAVA is well described in the literature (Lovering et al., 2009b), less is known about the relationship between exercise capacity and Q_IPAVA. La Gerche et al. (2010) found that subjects with low bubble scores (0–2) had a lower mean VO_2max in normoxia compared to those with higher bubble scores (3–5), with no effect on S_PO₂. While Norris et al. found age (<50 years) to be associated with greater Q_IPAVA, they did not find any significant difference in VO_2max between high and low bubble scores. If both exercise performance in hypoxia (Chapman, 2013) and Q_IPAVA are modulated by variation in S_PO₂, we would expect to see some relationship between the two. Alternatively, if increased Q_IPAVA is in fact associated with increased VO_2max in normoxia, but not negatively affecting the A-aDO₂, it is possible that increased Q_IPAVA could be associated with improved performance in hypoxia.

Several studies have hypothesized that vasoactive drugs may increase Q_IPAVA directly through their vasodilatory properties. Nitroglycerine, norepinephrine, and aminophylline had no effect on Q_IPAVA in resting participants in normoxia (Lozo et al., 2014). Sildenafil, nifedipine, and acetazolamide had no effect on Q_IPAVA in resting participants inspiring 100% O₂ (Elliott et al., 2014). In both of these studies, participants had normal values of resting sea level S_pO₂, which may account for the lack of effect on Q_IPAVA. It is possible the chance of observing an effect using some or all of these medications may be greater in HYP conditions.

Sildenafil citrate inhibits phosphodiesterase type-5 from degrading cyclic guanosine monophosphate (cGMP), promoting pulmonary smooth muscle relaxation through reduced intracellular calcium levels as part of the nitric oxide pathway. Several studies have shown an improvement in endurance performance in hypoxia as a result of orally administered sildenafil (Ghofrani et al., 2004; Richalet et al., 2005; Snyder et al., 2008). These reports suggest that improved endurance performance is related to an increase in S_PO₂ caused by a decrease in PAP, which reduces right ventricular afterload, and increases CO (Ghofrani et al., 2004; Faoro et al., 2007); however, the mechanism(s) for improved performance remain unclear (Carter et al., 2019a).

As sildenafil has been shown to increase S_PO₂, and Q_IPAVA appears to be at least partially regulated by S_PO₂, it is possible that Q_IPAVA could be stimulated by sildenafil, either directly through its vasodilatory properties as hypothesized by Elliott et al. (2014) or indirectly by sildenafil-induced increases in S_PO₂. The purpose of this study was to determine if a therapeutic dose (50 mg) of the pulmonary vasodilator sildenafil could be used to reduce Q_IPAVA during hypoxia at rest. The secondary purpose was to explore the relationship between Q_IPAVA at rest and the decrement in power output at VO_2max in hypoxia.

Materials and Methods

Sixteen male competitive cyclists were recruited, and after screening for patent foramen ovale, 13 participated in the study. All participants resided below 150 m above sea level and had not travelled above 2000 m (excluding commercial air travel) in the preceding 6 months. The nature of the terrain around the South Coast of British Columbia is such that occasional short exposures (e.g., driving over mountain passes) to altitudes of 1000–1800 m occur, but are brief and uncommon, so the participants were considered residents of sea level and altitude naive. Participants maintained a consistent volume of endurance training (five sessions per week for more than 1 year) and competed regularly in endurance cycling or triathlon (≥3 competitions per year). The study was approved by the Clinical Research Ethics Board at the University of British Columbia and all procedures were in accordance with the ethical standards of the 1964 Helsinki Declaration. Participants were informed of the procedures and possible risks involved in the study and informed consent was obtained.

Participants visited the University of British Columbia Environmental Physiology Laboratory on three occasions. During the first visit, participants completed two graded maximal exercise tests on a magnetically braked cycle ergometer (Velotron, Racermate, SD) under a normoxic (NOR; F_IO₂ = 0.209) and normobaric HYP condition (F_IO₂ = 0.147, equivalent to a simulated altitude of 3000 m).

The HYP exercise condition was produced using a custom HYP enclosure (Colorado Altitude Training, CO) that reduces the O₂ concentration of room air and delivers it to the participant using large bore tubing and a one-way non-rebreathing valve with breathing mask (Hans Rudolph, KS). The HYP enclosure was programmed to deliver an F_IO₂ of 0.14; however, mean F_IO₂, as measured by fast response O₂ gas analyzer (Vacumed, CA), was 0.147. The order of the tests in the HYP and NOR condition was randomized. Workload started at 100 W and increased in a ramp manner by 1 Watt every 2 seconds until exhaustion. During the test procedure, respiratory gases were collected using fast response O₂ and CO₂ gas analyzers and a pneumotachometer (Vacumed). Participants' S_PO₂ and heart rate (HR) were collected with a Nonin 9600 ear clip pulse oximeter (Nonin Medical, MN). The signals from these devices were stored digitally on a PowerLab data acquisition system (ADInstruments, Dunedin, New Zealand) and averaged every 10 seconds.

During the two remaining visits, participants completed experimental trials with either the placebo or study medication, sildenafil, using a double-blinded, randomized crossover design. Sildenafil was administered as a single oral dose of 50 mg. This dose is used most commonly in studies of athletic performance in hypoxia and was previously shown to elicit a decrease in PAP (Carter et al., 2019a). Participants and researchers were blinded to the condition and medication. The placebo and sildenafil capsules were compounded locally (McDonalds Pharmacy, British Columbia, Canada) and were indistinguishable. Study sessions were completed a minimum of 7 days following maximal exercise testing and with 7 days between each session.

Experimental visits began with a 4 mL venous blood draw followed by ingestion of a placebo or sildenafil capsule with a glass of water. At 46 ± 3 minutes after ingestion, participants underwent a second blood draw and a venous catheter was inserted in the right forearm. Participants were placed in a recumbent semi-left lateral decubitus position on an examination table and outfitted with the same one-way non-rebreathing valve and mask (Hans Rudolph) used for the exercise component, with gas supplied through large bore tubing from a mixing chamber connected to a pressurized gas canister containing varying concentrations of O₂ (F_IO₂ = 0.21, 0.14, and 0.12, equivalent to a P_IO₂ of 148, 98, and 84 mmHg, respectively). These exposure levels were chosen to correspond to altitudes of sea level, 3000 m, and 4500 m, common moderate altitudes at which athletes compete. Participants were fitted with the pulse oximeter and expired gasses were collected with the same analyzers used in the exercise component.

At 62 ± 3 minutes after ingestion of the study capsule, participants began breathing gas supplied by the mask, while lying quietly, and recording equipment was started. An apical view of the heart was obtained using a Vivid I ultrasound and 3S cardiac probe (General Electric, MA). Ventilatory data were recorded throughout the entire testing protocol, but only minutes 10–14 were included in the analysis. At 15 ± 1 minutes of breathing, a bolus of agitated saline contrast (1 mL air in 4 mL saline) was forcefully injected into the venous catheter. The passage of the contrast solution through the pulmonary circulation was recorded using the ultrasound from the moment of injection through ∼20 cardiac cycles and saved for posttest analysis by two blinded reviewers. As soon as recording was complete for the F_IO₂ = 0.21 condition, the procedure was repeated using the remaining levels of HYP gas (F_IO₂: 0.14 and 0.12) in order.

Ultrasound videos were stripped of identifying data and coded using a simple number system in a randomized order (1–81), and then distributed digitally to the reviewers who scored the presence of IPAVA according to the bubble score grading system described by Lovering and Goodman (2012). Bubble score uses a scale of 0–5 with the score approximating Q_IPAVA, depending on the amount of contrast that bypasses the pulmonary circulation and re-enters the left heart. The number of cardiac cycles between appearance of saline contrast in the right heart and left was counted. Three participants were excluded from participation after contrast appeared in the left heart in ≤3 beats on suspicion of patent foramen ovale. Two blinded raters scored all 81 video files, resulting in identical scores for 70 videos and a one-point disagreement on the remaining 11 videos. These 11 were rescored by a third blinded rater and the mode value of all three raters was used for the analysis. An enzyme-linked immunosorbent assay (BioScientific) was used to quantify serum cGMP in the predrug and postdrug blood samples.

Results

Participants' age, height, and body weight were 32 ± 11 years, 180 ± 7 cm, and 78.8 ± 11.6 kg, respectively. Participants' mean VO_2max at sea level was 60.0 ± 11.0 mL/kg/min. Mean VO_2max in hypoxia was 44.6 ± 10.1 mL/kg/min. Serum cGMP was significantly greater in the sildenafil group (p < 0.001) compared to pretest values and placebo (Table 1).

There was a significant increase in bubble score in hypoxia (p < 0.001; Fig. 1), but there was no effect of sildenafil on bubble score (p = 0.763; median difference: −0.17 [95% confidence interval, CI: −0.67 to 0.33]) or S_PO₂ (p = 0.580; median difference: −0.06 [95% CI: −1.22 to 1.10]). Sildenafil significantly increased HR (p < 0.01) and showed a trend toward increasing VO₂ (p = 0.127), especially at lower F_IO₂ (0.14 and 0.12). Sildenafil had no effect on systemic BP (Table 1).

OPEN IN VIEWER

FIG. 1.

Bubble score in varying levels of hypoxia with placebo and sildenafil. Brackets indicated median and first and third quartile. F_IO₂, fraction of inspired oxygen.

The relationship between S_PO₂ and bubble score differed significantly by F_IO₂ level (p < 0.001) (Table 2), but not by treatment group (p = 0.732). There was a significant negative relationship when F_IO₂ = 0.12 (p < 0.001; Fig. 2). There was no difference between the S_PO₂ for bubble scores 0–2 and 3–5 in room air; however, at an F_IO₂ of 0.12, S_PO₂ for bubble scores 0–2 was significantly higher than for scores 3–5 (p < 0.001; Fig. 3).

OPEN IN VIEWER

FIG. 2.

Linear mixed effect regression relationship between bubble score and oxygen saturation. S_PO₂, oxygen saturation.

OPEN IN VIEWER

FIG. 3.

Mean oxygen saturation in high and low bubble scores at each level of hypoxia (0.21, 0.14, and 0.12). Error bars indicate 95% CI. CI, confidence interval.

All participants experienced significantly impaired (p < 0.0001) peak power output during exercise in hypoxia (353.8 ± 41.8) compared to normoxia (413.5 ± 55.1 W). The decrement in power output at VO_2max (% of sea level power output in watts) ranged from a 9% to 19% below sea level values (Fig. 4). The decrement in power output in VO_2max was unrelated to Q_IPAVA in resting hypoxia with placebo (ρ = −0.33, p = 0.32) or sildenafil (ρ = −0.38 p = 0.24).

OPEN IN VIEWER

FIG. 4.

Histogram showing individual VO_2max decrement in hypoxia in all athletes.

Discussion

As a result of our intervention, we observed small, but consistent systemic effects of sildenafil. There was no effect of sildenafil on systemic BP, but HR was significantly increased with sildenafil at every level of hypoxia. This has been attributed to a sympathetic response to sildenafil's reflex vasodilatory effect and has previously been reported in rats (Fazan et al., 2008). The trend for increased VO₂ with sildenafil in hypoxia and normoxia suggests a small increase in metabolic rate that would typically be met with an increase in CO, supported by the observed increased HR. This increase in CO (compared to placebo in hypoxia where CO is limited) could raise S_aO₂ and exercise tolerance, but the difference may be very small and statistically insignificant. While elite athletes often seek marginal gains in the fraction of a percentage, the lack of an effect on endurance performance indicators is supported by several studies that found no ergogenic effect of sildenafil (Faoro et al., 2007; Jacobs et al., 2011; Kressler et al., 2011; Carter et al., 2019b).

We observed a greater bubble score under resting, HYP conditions when compared to normoxia. We interpret this to mean that exposure to increasing levels of hypoxia elicits an increase in Q_IPAVA. We discussed two possible mechanisms for sildenafil to increase Q_IPAVA, either through (i) direct vasodilation of IPAVA vessels or (ii) indirectly by increasing S_PO₂, one of the primary determinants of Q_IPAVA. Consistent with work by Elliott et al. (2014), our data showed that sildenafil did not affect bubble score in the NOR condition. This result confirms that Q_IPAVA is not directly affected by the vasoactive effects of sildenafil. Because S_PO₂ remains at normal values during the NOR exposure, the indirect effect of sildenafil on S_PO₂ is inconsequential in normoxia and no change is to be expected. The first novel finding of this study is that sildenafil did not affect Q_IPAVA during exposure to moderate hypoxia. We found no statistical difference in S_PO₂ between a 50 mg dose of sildenafil and a placebo. These results suggest the effects of sildenafil on pressure and blood flow were insufficient to increase Q_IPAVA. While we did not measure pulmonary pressure or flow, if sildenafil did decrease PAP and increase CO as previously demonstrated, HYP sensing may play a greater role in regulation of Q_IPAVA than initially believed (contrary to the pressure/flow hypothesis), consistent with work by Laurie et al. (2010) and Tremblay et al. (2015).

Our second main outcome was the significant relationship between Q_IPAVA and S_PO₂. This relationship, while indirect, may help shed some light on the IPAVA sensor mechanism. While breathing sea level air (F_IO₂ = 0.21), S_PO₂ was 98%–100% in all participants, so we would not expect to see a correlation with bubble score in normoxia based on the narrow distribution in saturation. It was not until F_IO₂ was reduced to 0.12 that we saw variability in S_PO₂ that could be compared with the corresponding variability in bubble score. When the dataset was analyzed separately for each F_IO₂, the negative correlation between S_PO₂ and bubble score became significant. This supports evidence by Laurie et al. (2010), who found a correlation between S_PO₂ and bubble score at all levels of hypoxia combined. Had we used a comparably low F_IO₂ (0.10), the correlation between S_PO₂ and bubble score at rest may have been stronger.

When participants were sorted into low (0–2) and high (3–5) bubble score categories, high bubble score participants were found to have significantly lower S_PO₂. Despite a link between S_PO₂ and bubble score in hypoxia, several participants showed evidence of significant Q_IPAVA in resting normoxia. This has been previously reported by several studies (Kennedy et al., 2012; Elliott et al., 2013; Duke et al., 2015). Kennedy et al. (2012) attribute differences in Q_IPAVA to the importance of body position and the effect of tethering in the lung. All participants in our research eventually showed some degree of Q_IPAVA below a threshold S_PO₂ of 87%. While reductions in ambient PO₂ directly result in increased Q_IPAVA, hypoxemia caused by exercise (which does not occur in some participants) is an alternative mechanism for increased Q_IPAVA.

The relationship between S_PO₂ and bubble score leads us to speculate on two possible conclusions. P_aO₂ may be the primary mediator of Q_IPAVA, especially if regional differences in P_AO₂ could potentially account for increased Q_IPAVA in normoxia during exercise when S_PO₂ is normal, as has been shown in several studies of exercise in normoxia (La Gerche et al., 2010; Lalande et al., 2012; Norris et al., 2014). A more likely (but still speculative) conclusion is that P_aO₂ is one of several modulating factors alongside P_vO₂, PAP, and CO.

Our final, and perhaps the most novel, observation was the relationship between Q_IPAVA in acute hypoxia and the effect of hypoxia on exercise performance. Exercise performance in hypoxia has been repeatedly shown to be determined by gas exchange impairments that result in variation in S_PO₂ (Chapman, 2013). Similarly, previous research and our own results presented above attribute increased Q_IPAVA to reductions in S_PO₂. If reduced S_PO₂ is implicated in both the decrement in exercise performance in hypoxia and Q_IPAVA, it is reasonable to expect that the two may be related. Despite both exercise performance in hypoxia and Q_IPAVA being modulated by S_PO₂, we found that the two are unrelated, which we believe indicates that, while blood flow through IPAVA does appear to bypass the pulmonary microcirculation, increases in Q_IPAVA are hemodynamically insignificant.

Conclusion

The results of this study show that 50 mg of sildenafil does not significantly affect Q_IPAVA in normoxia or varying levels of hypoxia (F_IO₂ = 0.14, 0.12), as measured by bubble score. We identified a threshold S_PO₂ of 87%, below which all participants displayed perfusion of IPAVA. Several participants demonstrated Q_IPAVA above zero in room air, indicating that arterial hypoxemia may not be the sole stimulus for perfusion, but perhaps, some combination of P_aO₂ and P_VO₂. Perfusion of IPAVA at rest in hypoxia does not appear to be related to the decrement in power output at VO_2max in hypoxia, despite the relationship with S_PO₂. Based on our results, we conclude that changes in Q_IPAVA do not cause changes in power output at VO_2max in hypoxia, and athletes would see no benefit from the use of sildenafil to affect the Q_IPAVA.