Effect of Bronchoconstriction-induced Ventilation–Perfusion Mismatch on Uptake and Elimination of Isoflurane and Desflurane

THE rising prevalence of the chronic obstructive pulmonary disease and asthma may increase the number of patients suffering from these conditions who are in need of anesthesia for surgery. This is important for clinical anesthesia, because these diseases are associated with pulmonary ventilation/perfusion (V_A/Q) mismatch that may affect uptake and elimination of volatile anesthetics.¹ 

Theoretical considerations on the basis of an electrical analog model by Eger and Severinghaus²  in the 1960s suggested that pure shunt, that is, the most extreme form of V_A/Q mismatch, would delay uptake of volatile anesthetic agents with lower blood solubility. This was later confirmed by an experimental model published by Stoelting and Longnecker.³  However, the effects on volatile anesthetic elimination were not considered. Moreover, the effects of V_A/Q scatter, which is in general the dominating cause of gas exchange disturbance in obstructive lung diseases, have not been studied in depth.

Previously presented data from a porcine lung obstructive model regarding uptake and elimination of desflurane, a volatile anesthetic with very low blood solubility (blood/gas partition coefficient, 0.498 ± 0.522 in humans and 0.502 ± 0.054 in pigs),⁴  demonstrated that both uptake and elimination are delayed by bronchoconstriction.⁵  On the basis of these considerations we hypothesized that a volatile anesthetic with a higher blood solubility (isoflurane; blood/gas partition coefficient of 1.32 ± 0.04 in humans and 1.07 ± 0.05 in pigs)⁴  would be less affected by a methacholine-induced increase in V_A/Q scatter.

Therefore, the objective of the present experimental study in piglets was to compare the uptake and elimination of inhalational anesthetics with high and low blood solubility. The null hypothesis was that methacholine inhalation-induced bronchoconstriction affects the pharmacokinetics of isoflurane and desflurane to a similar extent.

The animal ethics committee of Uppsala University (Uppsala, Sweden) approved this prospective nonrandomized animal study. The care and handling of animals were in accordance with the National Institutes of Health (Bethesda, Maryland) guidelines for ethical animal treatment.⁶ 

Juvenile, 2-month-old piglets (weight, 26.7 ± 1.5 kg) of Yorkshire–Norwegian country breeds were used in the study. The animals fasted overnight with free access to water. All of the piglets underwent the same preparatory algorithm (induction and maintenance of anesthesia and monitoring). The experiments were conducted consecutively, no randomization procedures were used to assign animals to conditions, and no attempts were made to blind experimenters to condition.

Seven of the animals were used for assessment of uptake–elimination of isoflurane (isoflurane group) and seven for measurement of uptake–elimination of desflurane (desflurane group). Parts of the data from six of the animals in the desflurane group have previously been reported.⁵ 

As described previously in detail,^5,7  anesthesia was induced by an intramuscular injection of xylazine (2.2 mg/kg; Rompun, Bayer, Germany) and tiletamine–zolazepam (6 mg/kg; Zoletil, Virbac, France). The pigs were placed in the supine position, and the trachea was intubated orally with an ID 7-mm cuffed endotracheal tube (Mallinckrodt, Ireland). After testing for hind limb reflex absence, muscle relaxation was induced with an intravenous bolus of 2 mg/kg rocuronium (Esmeron, N.V. Organon, Netherlands), followed by a continuous infusion of 2.5 mg · kg^–1 · h^–1 rocuronium. Anesthesia was maintained by continuous intravenous infusions of fentanyl (0.04 mg · kg^–1 · h^–1; Leptanal, Janssen-Cilag AB, Sweden), midazolam (0.12 mg · kg^–1 · h^–1, midazolam Actavis, Actavis Group, Iceland), and propofol (Diprivan, Astra, Sweden) via 18-G catheters (Becton Dickinson, Germany) placed in ear veins.

After intubation and during mechanical ventilation, a median tracheotomy was performed, and the orotracheal tube was replaced by an ID 9-mm cuffed endotracheal tube (Mallinckrodt). Thereafter, the lungs were mechanically ventilated with intermittent positive pressure ventilation with fractional inspired oxygen tension of 0.4 and positive end-expiratory pressures of 5 cm H₂O provided by a KION anesthesia ventilator (Maquet Critical Care). Fresh gas flow was set to exceed double minute ventilation to make delivered gas fractions close to inspired gas fractions. The tidal volume was set to 10 mg/kg, and respiratory frequencies were adjusted to achieve a normal Paco₂ of 40 to 45 mmHg.

Ventilation variables were measured at the proximal end of the endotracheal tube with a standard anesthesia monitor (SC 9000 XL, Siemens, Germany) and additionally assessed by a NICO₂ system (Respironics Novametrix, Inc., USA). These measurements were averaged more than 15 cycles for analysis. Volatile anesthetic concentrations were monitored continuously with an infrared analyzer (Capnomac Ultima, Datex Ohmeda, Finland), calibrated to the manufacturer’s standards.

A flow-directed pulmonary artery catheter (7.0 French, Swan-Ganz thermodilution catheter, Baxter, USA) and a central venous catheter (4.0 French, Becton-Dickinson Critical Care Systems, Singapore) were inserted via the right external jugular vein. The balloon tip of the pulmonary artery catheter was located in the wedge position for cardiac output measurements and mixed venous blood sampling. The pulmonary artery catheter was repositioned before each experimental step to ensure that the tip was always located in regions with high pulmonary blood flow. Cardiac output was repeatedly measured at every experimental time point. All of the pigs received a right carotid arterial catheter for continuous arterial pressure measurements and for blood sampling (20 G, Becton-Dickinson Critical Care Systems).

Blood gas analysis was performed immediately after bubble-free blood sampling with standard blood gas electrodes specifically set up to analyze porcine blood (ABL 500 and OSM 3, Radiometer, Denmark). Finally, a suprapubic urinary catheter (Sympakath, Ruesch AG, Switzerland) was placed to monitor urine output.

In the desflurane group, determination of V_A/Q distribution was performed in all of the animals using the multiple inert gas elimination technique (MIGET) at baseline and during nebulization of methacholine before second desflurane administration. In isoflurane pigs, MIGET was performed in three animals using the same time points to verify the methacholine effect during nebulization (fig. 1).

MIGET has been described in detail in an earlier publication.⁵  In brief, six inert gases with different solubilities in blood were infused into a peripheral vein at a constant rate. Simultaneously, arterial and mixed venous blood, as well as mixed expired gas samples, were obtained. Thereafter, blood samples were equilibrated with nitrogen in a shaking water bath. The gas samples were analyzed for their inert gas partial pressures by gas chromatography (Gas Chromatograph Model 5890, Series II, Hewlett-Packard, USA). The retention of these six gases was calculated as the ratio of arterial to mixed venous partial pressure and excretion as the ratio of mixed expired to mixed venous partial pressure. These values were processed by the algorithm of Evans and Wagner⁸  to obtain compatible V_A/Q distributions.

The calculated parameters were log SD of perfusion and log SD of ventilation describing the dispersion or broadness of the respective distribution and mean Q and mean V representing the mean values of perfusion or ventilation. The obtained data related to V_A/Q distribution were perfusion of lung regions with V_A/Q less than 0.005, perfusion of lung regions with 0.005 less than V_A/Q less than 0.1 (i.e., low V_A/Q regions), ventilation of lung regions with 10 less than V_A/Q less than 100 (i.e., high V_A/Q regions), and ventilation of lung regions with V_A/Q greater than 100 (dead space [V_D/V_T]). As a quality indicator for the fit of the model, the remaining sum of squares is reported.

Arterial and mixed venous blood samples were collected in glass syringes coated with EDTA (FORTUNA OPTIMA 5 ml, Luer-lock, Poulten & Graf GmbH, Germany) for analysis by micropore membrane inlet mass spectrometry (MIGET by MMIMS System, Oscillogy LLC, USA). The system consists of a polymer membrane confined to multiple small micropores that separate the blood sample from the mass spectrometer and high-vacuum system, and gases diffuse through this membrane into the mass spectrometer for analysis. Because tonometry is not necessary, the native blood samples flowed over the micropore membrane inlet mass spectrometry (MMIMS) probes, and the volatile gas partial pressures were analyzed directly by measuring the ion current of the mass/charge ratio (m/e) at m/e = 51 for isoflurane and m/e = 101 for desflurane.^9–12 

Following an alveolar recruitment maneuver¹³  (40 cm H₂O for 10 s) and 30 min of stabilization after instrumentation, baseline hemodynamic, ventilation, and gas exchange data were obtained.

Either isoflurane (Forene, Abbott Laboratories, USA) or desflurane (Suprane, Baxter Int., USA) was administered via the KION ventilator (Maquet Critical Care, Sweden) with the vaporizer set at 1 vol% for isoflurane or at 5 vol% for desflurane in an open system. Fresh gas flow was set to exceed double minute ventilation. Arterial and mixed venous blood samples were obtained simultaneously after 0, 1, 2, 5, 10, 20, and 30 min and after 45 and 60 min only for isoflurane (volatile anesthetic uptake, wash-in). Thereafter, the inhalation of the volatile agent was stopped and the sampling sequence was repeated (volatile anesthetic elimination, wash-out).

Methacholine (100 µg/ml in saline, acetyl-ß-methacholine chloride, Sigma-Aldrich Inc., USA) was intermittently aerosolized using an ultrasonic nebulizer (Siemens Model 63 02 595 E400E) to maintain a constant increase of respiratory resistance with doubling of peak inspiratory pressure in each pig throughout this experimental step. The uptake and elimination sampling for each volatile agent were repeated at the given time schedule.

At the end of each experiment, the animals were euthanized with an intravenous injection of potassium chloride while under general anesthesia. The workflow of the experimental protocol is presented in figure 1.

Nonlinear regression analysis of the MMIMS data was performed with Sigmaplot version 11 (Systat Software Inc., USA). The curves are displayed as means of each data point with SD. As a first step, the individual data for a single pig were calculated. For this, the end-plateau signal of the respective arterial wash-in curve of each piglet before methacholine nebulization (T₁ in fig. 1) was set as reference value (= 1.0). The arterial and mixed-venous data obtained during bronchoconstriction were scaled to this signal and fitted to a double exponential function (a + c = 1):

The measured (dry) end-tidal partial concentration in vol% for the volatile anesthetics was converted to partial pressure (millimeters of mercury) and corrected for water vapor at the respective body temperature. Correction for water vapor was as follows:

The end-tidal partial pressure was assumed to be approximately equal to arterial partial pressure at the end of the wash-in of the volatile anesthetics, before methacholine inhalation. Therefore, the maximum healthy arterial wash-in mass spectrometry signal was calibrated to the corrected end-tidal partial pressure of the volatile anesthetic. The resulting calibration factor (millimeters of mercury per microamp) was then applied to the raw mass spectrometer ion current values of the arterial and mixed venous samples to derive the partial pressures of the volatile anesthetics in the respective blood samples. The Ostwald coefficient for the respective anesthetic in pigs¹⁵  was applied and the uptake calculated analogous to Peyton¹⁶  as follows:

where p_aAn is the arterial partial pressure of the anesthetic, p_mvAn is the mixed venous partial pressure of the anesthetic, λ is the Ostwald coefficient of the anesthetic, and Q is the cardiac output. The data were calculated separately for each piglet and were averaged for the respective groups afterward.

Statistical analysis was performed with the Statistical Package for the Social Sciences (SPSS version 23, IBM Corporation, USA). The estimation of sample size was based on a previous experimental porcine study,⁵  which used an analogous experimental setup. Power calculation using two-sided t test at a significance level of 5% (α = 0.05) and a power of 80% (β = 0.20) revealed that at least four animals per group were needed to detect a difference of more than 20% in the volatile wash-in period. The change of the time period to 90% (p90) of the maximum arterial anesthetic partial pressures was defined as the primary variable.

The data were tested for normal distribution with the Shapiro–Wilk W test and are presented as means and SDs in the case of normal distribution (cardiopulmonary and ventilation variables). The analysis of normally distributed data was performed by a repeated-measures one-way ANOVA with post hoc Bonferroni correction. The sequential changes of the relative mass spectrometry signal (representing blood partial pressures of volatiles) in each group were assessed by a repeated-measures general linear model (type III sums of squares). Subsequent between-group comparisons were performed by two-way ANOVA using the independent variables group and time. Post hoc multiple comparisons were performed by the Bonferroni procedure applied to all of the pairwise comparisons.

The area under the wash-in and wash-out curves was calculated before and during methacholine-induced bronchoconstriction for anesthetic uptake and elimination, and grouped values were compared by a two-sided t test. V_A/Q distributions were directly compared for each individual measurement in each animal by the Kolmogorov–Smirnov two-sample test, treating each paired measurement as an independent observation (i.e., without accounting for intraindividual versus interindividual differences). The differences were considered to be statistically significant for all of the procedures if the P value was less than 0.05.

There were no differences regarding biometric variables between both groups, and there were also no intraoperative difficulties that might have affected the data.

The wash-in of isoflurane up to 1 vol% or desflurane (5 vol%) in the inspired gas in normal piglets before methacholine-induced bronchoconstriction had no significant effect on hemodynamics, respiratory mechanics, ventilation, or global gas exchange variables as compared with the initial baseline data with intravenous anesthesia (tables 1 and 2).

Hemodynamics, alveolar ventilation, and gas exchange variables were significantly changed after the administration of methacholine by inhalation. Mean pulmonary artery pressure increased, whereas mean arterial pressure, cardiac output, and systemic vascular resistance remained constant. Pulmonary vascular resistance nearly doubled by methacholine nebulization in all of the piglets.

The respiratory system resistance increased more than fourfold with methacholine; the peak airway pressure was at least doubled, and alveolar ventilation significantly decreased. Gas exchange was severely impaired, as indicated by lower Pao₂ and mixed venous oxygen tension; increased end-tidal pressure of carbon dioxide, Paco₂, and mixed venous carbon dioxide tension; and increased venous admixture. However, these critical alterations caused by methacholine inhalation were not altered by either isoflurane or desflurane uptake and elimination (tables 3 and 4). Venous admixture was lower and Pao₂ higher with isoflurane than desflurane during methacholine nebulization, and these differences remained throughout the bronchoconstriction phase.

The V_A/Q scatter was broadened during isoflurane and desflurane anesthesia (table 5). The ventilation distribution was widened during methacholine inhalation and shifted toward regions with a higher V_A/Q ratio. The main mode of perfusion distribution was also broader, and shunt and perfusion in areas of low V_A/Q ratios increased up to 17%; the sum of them was similar with either anesthetic (fig. 2; table 5).

The inspired volatile anesthetics partial pressures reached a stable plateau after 5 min of uptake. At the end of volatile anesthetic wash-in, the end-expired partial pressures were 5.6 ± 0.4 mmHg for isoflurane and 35.4 ± 0.3 mmHg for desflurane in healthy pigs, which resulted in ratios of expiratory to inspiratory partial pressures of 0.93 ± 0.03 for isoflurane and 0.95 ± 0.02 for desflurane. The ratios of mixed venous to arterial partial pressures were 0.86 ± 0.05 (isoflurane) and 0.85 ± 0.03 (desflurane), respectively.

The arterial and mixed venous partial pressure course for both volatile anesthetics could be expressed by double exponential functions depicting the elimination phase before and during methacholine inhalation. Without methacholine, arterial isoflurane partial pressure reached 90% of the plateau (p90) within 16.4 min during wash-in of the anesthetic, whereas desflurane needed only 4.4 min (P < 0.001). The elimination measurements of the volatiles revealed that 90% had been eliminated from the circulation after 27.8 min for isoflurane (fig. 3) and 5.7 min for desflurane (P < 0.01; fig. 4).

During methacholine-induced bronchoconstriction, the uptake and elimination of both volatile anesthetics were delayed but more significantly for desflurane. Thus, for 1 vol% isoflurane, 90% uptake and elimination (p90) were reached after 35 min and 44 min, respectively. The p90 for uptake of 5 vol% desflurane was reached within 14.8 min, and 90% of elimination was detected after 14.9 min. The time to 50% of the maximum arterial partial pressure increased only in desflurane piglets by methacholine but not in animals that received isoflurane (figs. 3 and 4; table 6).

The uptake of isoflurane (fig. 5A) peaked with 0.36 ± 0.09 ml (vapor) · kg^–1 · min^–1 in healthy piglets at the 2-min time point versus 0.28 ± 0.06 ml (vapor) · kg^–1 · min^–1 during methacholine inhalation (P = 0.11). In contrast, desflurane (fig. 5B) peaked at the 1-min time point and was 1.87 ± 0.24 ml (vapor) · kg^–1 · min^–1 before and 0.90 ± 0.25 ml (vapor) · kg^–1 · min^–1 during methacholine-induced bronchoconstriction, thus only 48% of the uptake during the nonobstructed state (P < 0.01).

The elimination was greatest at the 1-min time point for both volatile anesthetics, 0.24 ± 0.04 (healthy) versus 0.22 ± 0.07 ml (vapor) · kg^–1 · min^–1 (methacholine, P = 0.45) before and during methacholine nebulization for isoflurane (fig. 5C) and 1.13 ± 0.19 (healthy) versus 0.72 ± 0.31 ml (vapor) · kg^–1 · min^–1 (methacholine, P < 0.01) for desflurane (fig. 5D), respectively. Thus, elimination was reduced to 92% by methacholine in the isoflurane group and much more in the desflurane group, to 64% of the nonobstructed state.

The difference in total uptake and elimination for isoflurane (calculated as area under the curve) was not significantly altered before and during methacholine-induced bronchoconstriction. For desflurane, it was decreased during uptake (25 ± 4 vs. 15 ± 3 ml [vapor]/kg; P < 0.001) and during elimination (9 ± 1 vs. 8 ± 3 ml [vapor]/kg).

The administration of methacholine by nebulization causes, as expected, stable and reproducible bronchoconstriction, characterized by increased peak inspiratory pressure and respiratory system resistance. More importantly, methacholine inhalation results in a shift of mean ventilation distribution toward regions with higher V_A/Q ratios and in a broadening of the perfusion dispersion with increased perfusion in low V_A/Q regions and intrapulmonary shunt. Based mainly on theoretical considerations and limited previous experimental data, this V_A/Q scatter could impair the uptake and elimination of volatile anesthetics.

Methacholine inhalation increased the time to 90% of the maximum arterial partial pressures for both volatile anesthetics and reduced the peak uptake and elimination of desflurane, whereas isoflurane pharmacokinetics were less affected. The data are broadly consistent with the principle that uptake of less soluble agents relies more on gas exchange in lower V_A/Q lung ratios than more soluble agents. Although our data confirmed that desflurane kinetics were always faster than isoflurane kinetics, desflurane uptake and elimination were more affected by V_A/Q heterogeneity and therefore are likely to exert a larger variability in patients with bronchoconstriction.

Previous studies modeling the kinetics of volatile anesthetics assume that the uptake is perfusion limited¹⁷  and did not consider the distribution of ventilation heterogeneity.^2,18  In the present study, we managed to avoid significant changes in total cardiac output and alveolar minute ventilation throughout the experiments, which allowed us to investigate correctly the effect of V_A/Q scatter.

The arterial kinetics of inhalational anesthetics depend on total alveolar ventilation, pulmonary perfusion, solubility of the agent, distribution of ventilation and perfusion, and mixed venous kinetics.¹⁸  Although cardiac output was constant throughout the study in all animals, the difference in uptake of the two volatile agents can be explained by their different blood solubilities, by different V_A/Q scatter, and by the extent of intrapulmonary shunt. Previous models¹⁹  and experimental data³  indicate that shunt may impair uptake and that the effect is even higher for gases with low solubility. We demonstrate here that other modifications of V_A/Q in addition to shunt (e.g., broadening and shift of the distribution modes) can have similar effects. This is of particular interest when considering a patient with bronchoconstriction, who has little or no shunt but substantial ventilation of low V_A/Q regions when awake, and shunt may be small, also during anesthesia.²⁰ 

The bronchodilating effects of desflurane and isoflurane appear different in different species. In humans, desflurane and isoflurane exert similar effects on proximal airway tissues, whereas the effect on distal airways is lower for desflurane than isoflurane.²¹  In contrast, experiments on rats revealed similar effects of the two volatile agents in protection from methacholine-induced bronchoconstriction.²²  Although we cannot exclude a differential alleviating effect of the volatile anesthetics on the methacholine-induced bronchoconstriction, it should be noted that respiratory mechanics were not influenced by the added volatile agents before and after methacholine inhalation.

Arterial oxygenation was markedly decreased during methacholine nebulization in both groups of piglets, but the effect was more pronounced in the desflurane group. This can be related to the increased shunt in desflurane piglets. Nunes et al.²³  described a shunt of approximately 23% for desflurane and 15% for isoflurane in spontaneously breathing healthy dogs, close to what we have found during methacholine nebulization. Although an influence of the volatile anesthetics themselves on shunt development cannot be excluded and is debated,²⁴  application of very high oxygen fractions is problematic, because they themselves can lead to significant amounts of shunt.^25,26  Although Nunes et al.²³  applied the volatile anesthetics in pure oxygen, in our study a fractional inspired oxygen tension of 0.4 was used, resulting in a Pao₂ of less than 225 mmHg and causing our Riley-shunt calculation to be susceptible to the influence of low V_A/Q scatter.²⁷ 

Methacholine stimulates muscarinic receptors and causes smooth muscle contraction. The M₁ receptor, located in the alveolar wall, is likely to be involved in a parenchymal response, whereas the M₃ receptor, located in the airway smooth muscles, is responsible for airway effects.²⁸  Aerosolized methacholine leads to constriction of both the airways and the parenchyma by affecting the different receptors and possibly by excessive parenchymal distortion caused by a heterogeneous response of the peripheral airways,²⁹  prompting heterogeneity of ventilation distribution and altering V_A/Q distribution.

Modern inhalational anesthetics like desflurane have a very low blood solubility resulting in fast induction and fast emergence from anesthesia. The present data suggest that kinetics of low soluble gases are more impaired by V_A/Q mismatch and therefore could result in greater variability of kinetics between patients than higher soluble anesthetics like isoflurane, enflurane, or halothane. Even for the more soluble agents, the end-tidal volatile agent partial pressure in the presence of V_A/Q mismatch is not necessarily a reliable measure for the arterial blood level, as shown by Frei et al.³⁰  for isoflurane. The molecular weight of the volatile anesthetic could also influence the end-tidal-to-arterial gradient, as suggested by Landon et al.³¹ 

A limitation of our study is that the anatomy of piglet and human lungs differs in ways that could affect translation of results. In contrast to piglets, humans have collateral ventilation,³²  therefore clinical methacholine effects may appear less severe than observed in pigs. However, patients suffering from obstructive lung diseases have been demonstrated to have V_A/Q mismatch in distributions similar to the V_A/Q abnormalities that we have measured in our model.³³ 

Other errors could impact our calculations of uptake, including differences in solubility, variations in cardiac output, and errors in the calibration of the infrared analyzer. Although the pigs were placed on heating mats, variations of plus or minus 2°C around the normal body temperature did occur.

Two assumptions were made that allow us to derive uptake from the partial pressures and cardiac output measurements. First, the calculation of the uptake of inhaled agent across the alveolocapillary membrane was based on a calibration factor that assumed that the end-tidal anesthetic partial pressure approximated the arterial anesthetic partial pressure. We believe this to be the case because the pigs’ lungs were healthy with minimal V_A/Q abnormality, as evidenced by the absence of a significant alveolar–arterial partial pressure difference for carbon dioxide before methacholine administration. Second, we did not measure blood–gas partition coefficients but retrieved them from the literature⁴  and assumed they would match this in the animals used in this study. Because the arterial partial pressure might have been lower than indicated by the end-expired partial pressure, the data presented in figure 5 may slightly overestimate uptake and elimination. The error arising from unmeasured alveolar–arterial differences for the two volatile agents is likely to be small and will effectively only scale the uptake calculations but will not substantially alter the results.

The difference in solubility of volatile anesthetics has a significant influence on their uptake and elimination in a piglet model of bronchoconstriction and V_A/Q mismatch. The higher soluble isoflurane is affected to a lower degree than the fairly insoluble desflurane. Respiratory diseases account for a large part of morbidity and mortality and are projected to increase in the coming years.^34,35  Significantly more patients suffering from lung conditions associated with V_A/Q mismatch will thus be presented for elective surgery in the future. Therefore, an improved understanding of anesthesia induction and emergence and of anesthesia depth in patients with V_A/Q mismatch undergoing volatile anesthesia seems to be warranted.

The authors express their special gratitude to Agneta Ronéus, Karin Fagerbrink, Annelie Ryden, and Maria Svaelas (Hedenstierna Laboratory, Uppsala University, Uppsala, Sweden) for their help and support during the experiment. Their work and their assistance in instrumentation and monitoring of the animals, as well as blood sampling and data recording, are greatly appreciated. The authors especially acknowledge the work of Eva-Maria Hedin (Hedenstierna Laboratory, Uppsala University, Uppsala, Sweden), who performed the multiple inert gas elimination technique measurements.

Supported by the Swedish Research Council (X2015-99x-22731-01-04; Stockholm, Sweden), the Swedish Heart and Lung Fund (Stockholm, Sweden), and institutional sources of Uppsala University (Uppsala, Sweden) and Otto von Guericke University Magdeburg (Magdeburg, Germany).

Dr. Baumgardner is president of Oscillogy LLC (Pittsburgh, Pennsylvania), the manufacturer of the multiple inert gas elimination technique by micropore membrane inlet mass spectrometry system. The other authors declare no competing interests.