Relationship between Ventilation/Perfusion Ratio Mismatch and Capnogram Phase III Slope in Acute Bronchoconstriction and Its Modulation by Salbutamol Assessed by Synchrotron Imaging in Rabbit Lung

Abstract

Background:

The slope of phase III (SIII) of the volumetric capnogram is proposed as a noninvasive marker of ventilation/perfusion ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$) mismatch. However, direct comparisons of SIII with measurements of lung ventilation and perfusion distribution are lacking, particularly in acute bronchoconstriction. This study was developed to assess the relationship between SIII and distribution of lung ventilation (${\dot{\mathrm{V}}}_{\mathrm{A}}$) and perfused blood volume (V_B), a surrogate of perfusion, using synchrotron radiation K-edge subtraction imaging in a rabbit model of methacholine-induced acute bronchoconstriction.

Methods:

Five anesthetized, mechanically ventilated rabbits underwent imaging of ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B using K-edge subtraction imaging computed tomography before and after methacholine-induced bronchoconstriction and subsequent salbutamol administration. Mean ${\dot{\mathrm{V}}}_{\mathrm{A}}$, V_B, ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B, and their heterogeneity were quantified. SIII was measured from exhaled volumetric capnograms. Multiple linear regression was used to evaluate the relative contributions of ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B to SIII.

Results:

Methacholine-induced ventilation defects decreased both ${\dot{\mathrm{V}}}_{\mathrm{A}}$ (mean difference [d] = 0.22, 95% CI, 0.12 to 0.51; P < 0.001) and V_B (d = 1.9 ⋅ 10⁻³; 95% CI, 1.0 × 10⁻³ to 2.8 × 10⁻³; P < 0.001) and increased their heterogeneity (d = 0.42; 95% CI, 0.17 to 0.67; P < 0.001; and d = 0.12; 95% CI, 0.05 to 0.19; P < 0.001, respectively). Salbutamol thereafter improved respiratory elastance (d = 726.2; CI, 130.9 to 1,321; P = 0.04) but decreased ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B (d = 47.3; 95% CI, 26.5 to 68.1; P = 0.028), leading to increased shunt fraction (d = 0.044; 95% CI, 0.039 to 0.049; P < 0.001). SIII was significantly related with ${\dot{\mathrm{V}}}_{\mathrm{A}}$ (r = 0.49; 95% CI, 0.12 to 0.739; P < 0.001) but not with V_B. The strongest association was observed between SIII and mean ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B (r = −0.65; P < 0.0001). Multiple regression demonstrated that ${\dot{\mathrm{V}}}_{\mathrm{A}}$ had a larger effect on SIII than V_B.

Conclusions:

SIII is determined not only by ${\dot{\mathrm{V}}}_{\mathrm{A}}$ heterogeneity but also by ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ and its heterogeneity. The lack of improvement or paradoxical worsening in gas exchange despite mechanical improvement after salbutamol in acute bronchoconstriction is due to increased perfusion of low ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ regions and shunt fraction. These findings provide new insight into monitoring mechanical ventilation and acute bronchoconstriction with capnography.

Synchrotron radiation imaging allows quantitative evaluation of the distribution of lung ventilation, blood volume, and tissue morphology at a subacinar spatial resolution. Using methacholine to induce bronchospasm (causing ventilation defects) followed by salbutamol in this rabbit model, the authors demonstrated that phase III is determined not only by ventilation heterogeneity but also by ventilation/perfusion heterogeneity. Although salbutamol improved respiratory elastance, a lack of improvement (or worsening) in gas exchange is likely due to increased perfusion of low V̇_A/Q̇ regions and increased shunt fraction.

Editor’s Perspective

What We Already Know about This Topic

• Volumetric capnography of the expiratory carbon dioxide waveform is widely used for patient monitoring in the operating room and intensive care unit

• Consisting of three separate phases, it reflects emptying of the conducting airways (phase I), a mixture of gas from the alveoli and the conducting airways (phase II), and finally the alveolar gas (phase III)

• Phase III is postulated to reflect not only ventilation inhomogeneity but also perfusion inhomogeneity

• The relative contribution of these has been difficult to study

What This Article Tells Us That Is New

• Synchrotron radiation imaging allows quantitative evaluation of the distribution of lung ventilation, blood volume, and tissue morphology at a subacinar spatial resolution

• Using methacholine to induce bronchospasm (causing ventilation defects) followed by salbutamol in this rabbit model, the authors demonstrated that phase III is determined not only by ventilation heterogeneity but also by ventilation/perfusion heterogeneity

• Although salbutamol improved respiratory elastance, a lack of improvement (or worsening) in gas exchange is likely due to increased perfusion of low ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ regions and increased shunt fraction

Volumetric capnography describes the evolution of exhaled carbon dioxide partial pressure as a function of expired volume during a breath. The shape of the expired capnogram is determined by the sequential emptying of gas from the conducting airways (phase I), a mixture of gas from the alveoli and conducting airways (phase II), and that of alveolar gas (phase III). This latter phase typically shows a slope of the volumetric capnogram (SIII) even in normal lung, which reflects continuous carbon dioxide diffusion into a progressively decreasing alveolar volume during expiration. SIII was suggested to reflect ventilation inhomogeneity.^1,2 However, because carbon dioxide evolves by passive diffusion into the alveoli from the capillary blood, perfusion inhomogeneity has also been proposed as a mechanism of phase III slope.^3–5 SIII has been shown to change in several clinical conditions associated with mismatching of the lung ventilation/perfusion ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$), such as atelectasis,^6,7 pulmonary embolism,⁸ and asthma.^1,9,10 Owing to these characteristics, volumetric capnography appears as an attractive noninvasive method for monitoring gas exchange efficiency during mechanical ventilation^7,11,12 and/or small airway function.² However, both experimental and theoretical studies have suggested that SIII can arise solely from intra-acinar inhomogeneity due to airway asymmetry (convection and diffusion-dependent inhomogeneity) and sequential emptying of larger lung units (convection-dependent inhomogeneity).^13,14

Acute airway obstruction during an asthma exacerbation causes remarkable spatial heterogeneity in ventilation and perfusion.^2–4 Bronchoconstriction also causes reductions in regional lung perfusion that can more or less spatially coincide with ventilation defects due to the mechanisms assuring ventilation–perfusion (${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$) matching.³ In severe bronchoconstriction, ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ mismatch reduces gas exchange efficiency particularly due to alveolar hypoventilation causing low ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ areas.¹⁵ This can increase physiologic shunting leading to hypoxemia. Only a few previous studies have assessed the relation between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$,^16,17 and no such relationship has been evaluated in the presence of bronchoconstriction. Furthermore, in these earlier studies, the measurements of ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ were performed indirectly based on gas exchange, and no direct measurements of ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ based on functional lung imaging techniques in relation to SIII are available.

In this article, we used synchrotron radiation K-edge subtraction (KES) imaging to directly map the distribution of lung ventilation and perfused blood volume (V_B) as a surrogate of perfusion, and their ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B). The KES imaging technique allows investigation of the distribution of lung ventilation, blood volume, and tissue morphology quantitatively and with a subacinar spatial resolution.^18–20 We hypothesized that SIII is determined by ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B in the presence of ventilation inhomogeneity in acute bronchoconstriction. The aim of this study was therefore to assess the relationship between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}$, V_B, and ${\dot{\mathrm{V}}}_{\mathrm{A}}$/ V_B in a model of acute bronchoconstriction to gain insight into the relative effects of ventilation and perfusion inhomogeneities. Additionally, we assessed the effect of salbutamol administration.

Materials and Methods

Animal Preparation

The care of animals and the experimental procedures were in accordance with directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes, complied with the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines,²¹ and approved by the Internal Evaluation Committee for Animal Welfare in Research of the European Synchrotron Radiation Facility in Grenoble, France (#MD-580). The experiments were performed on five male New Zealand rabbits (2.9 ± 0.1 kg). Venous access was secured by a 22-gauge catheter inserted into the marginal ear vein under local anesthesia, using 5% topical lidocaine. Anesthesia was induced by the intravenous injection of 25 mg/kg thiopental sodium. The animal was tracheotomized with a no. 3, Portex tube (Smiths Medical, United Kingdom) after the injection of local anesthetics by lidocaine. The left carotid artery and jugular vein were catheterized for hemodynamic monitoring, blood gas measurements (radiometer; ABL77, Denmark), and drug delivery. Anesthesia was then maintained with 0.2 mg · kg⁻¹ · h⁻¹ IV midazolam. After ensuring adequate anesthesia from the hemodynamic parameters, continuous IV infusion of atracurium (1.0 mg · kg⁻¹ · h⁻¹) was started. The animal was immobilized in the vertical position in a cylindrical polyvinyl chloride custom-made holder.

Pressure-controlled mechanical ventilation was delivered using a custom-made apparatus, described in detail previously.²⁰ The ventilator allowed synchronizing mechanical ventilation with the image acquisition. The pressure was set to obtain a tidal volume of 7.0 ml/kg at baseline with a maximum threshold pressure of 30 cm H₂O, a baseline inspired concentration of oxygen of 21% and no positive end-expiratory pressure, with the respiratory rate set (50 to 60 breaths/min) to achieve an arterial PCO₂ close to 40 Torr. During ventilation imaging, the animal breathed a mixture of xenon (70%) and oxygen (30%). The respiratory gas flow was monitored by using a heated pneumotachometer (Hans Rudolph, USA). The endotracheal pressure was monitored continuously. All monitored signals were amplified, digitized at 400 Hz (PowerLab; ADI Instruments, United Kingdom), and recorded on a computer.

Synchrotron Radiation Computed Tomography Imaging

The experiments were performed at the Biomedical Beamline of the European Synchrotron Radiation Facility (Grenoble, France). Mono-energetic beams with an energy difference of 250 eV were produced from the continuous synchrotron radiation spectrum by a bent silicon crystal monochromator. The beams focused and crossed at the animal position, beyond which they diverged and were recorded by a liquid nitrogen–cooled high-purity germanium dual-line detector (Eurisys Measure, France).¹⁸ The horizontal pixel size of the detector was 0.35 mm, and the vertical beam height was 0.7 mm. Image reconstruction was performed using the filtered-back-projection algorithm, using the Interactive Data Language (RSI, France). Tuning of the beam energy to bracket Xe or I K-edge was performed beforehand, and monochromator parameters were recorded. This allowed switching from one pair of energies to the other within 2 min.

The KES imaging technique allows quantitative measurements of specific ventilation(s), as well as lung tissue density. A detailed description of the methodology and instrumental setup has been extensively discussed in previous studies.¹⁹ This imaging technique uses dual x-ray beams at slightly different energies above and below the K-edge of a contrast element. X-rays from a synchrotron radiation source are required since, as opposed to standard x-ray sources, they allow the selection of monochromatic beams from the full x-ray spectrum while conserving enough intensity for imaging with sufficient temporal resolution. Two computed tomography images are thus simultaneously acquired (2 s per 360° scan) during the inhalation of the xenon-oxygen gas mixture or infusion of iodine contrast (Iomeron 350; Bracco Imaging, France). Visualization and quantitative measurement of xenon within the airways or iodine in the vasculature is based on the property that the attenuation coefficient sharply increases when the energy of the incident x-ray beam rises above the K-edge of the contrast element. A material decomposition algorithm is then used to compute separate tissue- or contrast-density images as described previously.¹⁸ Dynamic KES imaging during xenon wash-in allows the measurement of the regional distribution of ventilation.²⁰ Static KES images obtained in 2 s per axial image slice after an intravenous bolus injection of iodine contrast (3 ml followed by a 10-ml/h infusion) yield quantitative maps of blood volume distribution. This regimen allowed obtaining constant blood iodine concentrations in preliminary experiments. A “tissue-density” image obtained from the same data allows quantitative measurement of the regional tissue density.

Image Processing and Analysis

The images were processed by using the MATLAB programming package (MathWorks Inc., USA). Lung tissue was selected within the tissue-density computed tomography images, by region growing segmentation. The local specific ventilation or ventilation normalized to the gas volume within the voxel(s) was calculated from the time constant of the xenon wash-in using a single-compartment model fit of xenon concentration versus time¹⁹ of 10 brief (2 s) subsequent expiratory pause images, each acquired after three breaths of 70% xenon in oxygen. A 5 × 5 voxel moving average window was used to avoid artefacts due to image misregistration between time points. Local voxel ventilation (${\dot{\mathrm{V}}}_{\mathrm{A}}$) was computed as s × voxel gas volume. Local voxel blood volume (V_B) was computed as the ratio of voxel to blood iodine concentration, measured within a region of interest placed in the pulmonary artery × voxel tissue volume. The heterogeneity of ventilation or blood volume distribution was calculated as the coefficient of variation (CV) of ${\dot{\mathrm{V}}}_{\mathrm{A}}$ within the lung area contained in the image slice.

Respiratory Mechanics and Gas Exchange

The respiratory system resistance R was calculated using a multiple linear regression method based the respiratory system equation of motion:

$${\displaystyle \begin{array}{l}{\displaystyle {\mathrm{P}}_{\mathrm{a}\mathrm{w}}={\mathrm{P}}_0+{\mathrm{E}}_{\mathrm{r}\mathrm{s}}\mathrm{V}+{\mathrm{R}}_{\mathrm{r}\mathrm{s}}{{\displaystyle \dot{\mathrm{V}}}}_{\mathrm{a}\mathrm{w}}}\end{array}}$$ (1)

where P_aw is the airway pressure, P₀ is the dynamic positive end-expiratory pressure, E_rs is the respiratory system elastance, V is the lung volume, R_rs is the respiratory system resistance, and $\dot{\mathrm{V}}$_aw is the gas flow at the airway opening. The P₀, E_rs, and R_rs values were estimated for individual respiratory cycles by multiple linear regression based on selected parts of the flow signal, as described previously,^20,22 using a custom program in MATLAB. R_rs was corrected for the tracheal tube resistance.

Partial pressure of oxygen (Pao₂) and carbon dioxide (Paco₂), pH, and bicarbonate (${\mathrm{H}\mathrm{C}\mathrm{O}}_3^{-}$) were measured from arterial blood samples. Shunt fraction (${\dot{\mathrm{Q}}}_{\mathrm{s}}$) was estimated based on the arterial blood gases, assuming an arteriovenous O₂ content difference of 3 ml/dl based on the data in Lovisari et al.²³

Volumetric Capnography

Capnography was performed using a rapid-response infrared analyzer (17630 silver edition; VacuMed, USA). Sidestream capnography data were collected for approximately 20 s. At least 12 technically acceptable breaths were collected. To reduce instrumental dead space, the gas sampling catheter was connected directly to a side port of the tracheal tube. Tidal capnograms with a variation in a tidal volume greater than 10% or showing artifacts in flow or capnographic recordings were discarded. A sensor calibration was performed before each experiment with a reference gas with a carbon dioxide concentration of 5.0% (Air Liquide, France). The volumetric capnogram was automatically analyzed to partition phases II and III, and the corresponding slopes were computed and averaged over the recorded cycles, using custom software in MATLAB (MathWorks Inc.) according to the method described by Tusman et al.^10,24 and averaged over the collected cycles. Physiologic dead space was measured using the Bohr equation (VD_Bohr)²⁵ and the Enghoff modification of the Bohr equation (VD_Enghoff).²⁶

Study Protocol

The volume history was standardized by inflating the lungs up to the 30-cm H₂O pressure for 5 s twice. Arterial blood gases were then measured, followed by recording of the volumetric capnography curves. The measurements were completed by acquiring ten subsequent KES subtraction images during xenon wash-in at two different axial positions selected approximately at the sixth (middle) and eighth (caudal, dependent) thoracic vertebral levels, based on a previous thoracic projection image. The synchrotron beam energy was immediately switched to the iodine K-edge, and the caudal axial image level was scanned. Once baseline data were collected, all measurements were repeated after bronchial challenge induced by inhalation of methacholine aerosol (125 mg/ml, 1 min) using an ultrasonic nebulizer (SAM LS2000; Systam, France) and subsequent salbutamol (2 mg/ml, 5 min) aerosol administration.

Statistical Analysis

We estimated that to detect a 50% increase in R_rs, five observations per group are needed to reach 80% power at the 5% significance level. The data are summarized as means ± standard error (SE) unless stated otherwise. The Shapiro–Wilk test was used to test data for normality [${\dot{\mathrm{V}}}_{\mathrm{A}}$, V_B, ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B, CV(${\dot{\mathrm{V}}}_{\mathrm{A}}$), CV(V_B), CV(${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B), E_rs, R_rs, SIII, ${\dot{\mathrm{Q}}}_{\mathrm{s}}$, VD_Bohr, and VD_Enghoff]. Both the mechanical and image-derived parameters were normally distributed. Accordingly, one-way repeated-measures analysis of variance was used to evaluate the changes in the mechanical and functional imaging parameters. Pairwise comparisons were performed on all of the above variables by using Holm–Sidak multiple comparison procedures. Linear regression was used to assess the strength of associations between SIII and the other variables. The statistical analyses were conducted by SigmaPlot (version 11.0; Systat Software, Inc., USA). Multiple linear regression analysis was performed to assess the relative effects of ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B on SIII in R.²⁷ Statistical tests were carried out with the significance level set at P < 0.05.

Results

Representative images of lung ${\dot{\mathrm{V}}}_{\mathrm{A}}$, V_B, and ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B, in a middle axial level are represented in figure 1. Methacholine challenge induced ventilation defects and increased ventilation inhomogeneity, as well as ventilation redistribution to the other lung regions. Methacholine inhalation also reduced V_B. These changes resulted in a heterogenous ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B. After salbutamol administration, ${\dot{\mathrm{V}}}_{\mathrm{A}}$ remained significantly altered, while there was some recovery in V_B. There was no apparent improvement in ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B.

Fig. 1.

(A) Images of regional ventilation (${\dot{\mathrm{V}}}_{\mathrm{A}}$, in µl/min), blood volume (V_B, in µl), and ventilation/blood volume ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$) distributions at baseline, after methacholine, and after salbutamol inhalation in a representative rabbit at end expiration. Note the appearance of ventilation defects, ventilation redistribution, the reduced V_B, and the increased ${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$ after methacholine. After salbutamol, there is some recovery in V_B without any apparent improvement in ${\dot{\mathrm{V}}}_{\mathrm{A}}$, and ${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$ heterogeneity. (B) Sample volumetric capnograms (blue line) and corresponding phase III slope (dashed red line) in the three experimental conditions.

Overall, mean ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B were significantly reduced after methacholine (fig. 2, A and B). Salbutamol increased V_B, while ${\dot{\mathrm{V}}}_{\mathrm{A}}$ remained low. The inhomogeneity of both ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B significantly increased after methacholine and remained high after salbutamol (fig. 2, C and D). Consequently, although mean ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B did not significantly change after methacholine, while it dropped after salbutamol (fig. 2E). The ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B heterogeneity rose after methacholine and remained high after salbutamol (fig. 2F).

Fig. 2.

(A to D) Ventilation (${\dot{\mathrm{V}}}_{\mathrm{A}}$; A), blood volume (V_B; B), and changes in their respective inhomogeneity (C, D) measured by the coefficient of variation (CV). (E) Ventilation/blood volume ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$; E) and CV of ${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$ (F) at baseline, after methacholine (MCH), and after salbutamol inhalation. Gray lines and dots represent individual data points (duplicate measurements in n = 5 animals, one missing value), while the black line and error bars show the means ± standard error. *P < 0.05 versus baseline; #P < 0.05 versus MCH, by one-way analysis of variance with Holm–Sidak post hoc tests.

Expectedly, methacholine significantly increased respiratory resistance and elastance, that only partially recovered after salbutamol (fig. 3, A and B). In addition, SIII showed a slight increase after methacholine that remained after salbutamol (fig. 3C).

Fig. 3.

(A) Respiratory elastance (E_rs). (B) Respiratory resistance (R_rs). (C) Exhaled carbon dioxide phase III slope (SIII). Gray lines and dots represent individual data points (duplicate measurements in n = 5 animals, one missing value), while the black line and error bars show the means ± standard error. *P < 0.05 versus baseline; #P < 0.05 versus methacholine (MCH), by one-way analysis of variance with Holm–Sidak post hoc tests.

The estimated shunt fraction significantly increased with induced bronchoconstriction, and remained high after salbutamol, while neither the Bohr nor the Enghoff dead space significantly changed (fig. 4).

Fig. 4.

(A) Estimated shunt fraction (${\dot{\mathrm{Q}}}_{\mathrm{s}}$). (B) Bohr’s dead space (VD Bohr). (C) Dead space calculated with Enghoff’s modification of the Bohr equation (VD Enghoff). Gray lines and dots represent individual data points (single measurements in n = 5 animals for Q_s, two missing values at baseline; duplicate measurement of VD, four missing values at baseline), while the black lines and error bars show the means ± SE. *P < 0.05 versus baseline versus methacholine (MCH) by one-way analysis of variance with Holm–Sidak post hoc tests.

Univariable regression showed a statistically significant association between SIII and ventilation inhomogeneity (CV): r = 0.45 (95% CI, 0.07 to 0.72; P = 0.023). The association of SIII with blood volume inhomogeneity was not statistically significant: r = 0.27 (95% CI, −0.141 to 0.600; P = 0.194). We found a statistically significant inverse association between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B (fig. 5A); r = −0.65 (95% CI, 0.343 to 0.832; P < 0.0001). The SIII was also associated with CV($\dot{\mathrm{V}}$/V_B): r = 0.43 (95% CI: 0.042, 0.70, P = 0.034; fig. 5B). Lower mean ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B values were associated with higher SIII. This association was stronger than that of SIII with ventilation (r = −0.51; 95% CI, −0.753 to −0.144; P < 0.0001). Furthermore, shunt fraction was strongly associated with CV(${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B): r = 0.81 (95% CI, 0.610 to 0.913; P < 0.001).

Fig. 5.

(A) Relation between ventilation/blood volume ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$) and exhaled carbon dioxide volumetric phase III slope (SIII). (B) Relation between the coefficient of variation (CV) of ${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$ and SIII. Solid lines show linear fit of the data. Dashed lines represent 95% CI. Note that SIII increases as ${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$ is lowered, and CV of ${\dot{\mathrm{V}}}_{\mathrm{A}}/{\mathrm{V}}_{\mathrm{B}}$ increases. BL, baseline; MCH, methacholine.

Table 1 summarizes the arterial blood gas parameters obtained under the baseline condition and after methacholine and salbutamol. PO₂ dropped and PCO₂ elevated after methacholine and remained altered after salbutamol. No statistically significant change in ${\mathrm{H}\mathrm{C}\mathrm{O}}_3^{-}$ was detected during the experiments.

Table 1.

Gas Exchange Data, Mechanical Ventilation Parameters, Time Constant of Xenon Wash in τ, and Its Coefficient of Variation, CV(τ)

Condition	Pao2, mmHg	Paco2, mmHg	pH	${\mathrm{H}\mathrm{C}\mathrm{O}}_3^{-}$, mmol/l	VT, ml	RR, breaths/min	P0, cm H2O	Τ, s	CV(τ)
Baseline	117.8 ± 14.2	30.4 ± 5.8	7.492 ± 0.05	20.7 ± 3.9	19.6 ± 3.4	50	1.6 ± 1.2	4.3 ± 0.3	0.19 ± 0.04
MCH	52.4 ± 13.1*	42.6 ± 7.9	7.30 ± 0.07*	20.5 ± 2.6	19.1 ± 2	50	1.1 ± 0.8	6.4 ± 2*	0.59 ± 0.25*
Salbutamol	51.6 ± 15.7*	41.2 ± 11.3	7.33 ± 0.06*	20.8 ± 3.9	18.1 ± 1.4	50	2.3 ± 1.1	7.2 ± 2.1*	0.71 ± 0.25*

P₀ is the estimate given by the multiple linear regression method (mean ± SD).

^*P < 0.05 versus baseline, by one-way ANOVA with Holm–Sidak post hoc tests.

ANOVA, analysis of variance; CV, coefficient of variation; MCH, methacholine; RR, respiration rate; VT, tidal volume.

The multiple regression analysis performed to assess the relative effects of ventilation and blood volume on SIII is summarized in Table 2. This analysis supported the opposite and stronger effect of ${\dot{\mathrm{V}}}_{\mathrm{A}}$ versus V_B on SIII.

Table 2.

Multiple Linear Regression Analysis

Value	β	SE	95% CI	P Value
Intercept	9.88E−17	1.56E−01	−0.32 to 0.32	1
${\dot{\mathrm{V}}}_{\mathrm{A}}$	−6.51E−01	1.70E−01	−1.00 to −0.30	< 0.001
V B	4.82E−01	1.70E−01	0.13 to 0.83	0.009

A 1 SD decrease in ${\dot{V}}_A$ was associated with an 0.65 SD increase in SIII, while a 1 SD increase in V_B was associated with a 0.48 SD increase in SIII.

β, standardized coefficient; SE, standard error.

Discussion

This study was undertaken to directly assess the relation between the exhaled carbon dioxide SIII and regional lung ventilation and perfusion, through functional lung imaging. We employed synchrotron radiation K-edge subtraction imaging to quantitatively map regional lung ventilation and blood volume, as a surrogate of perfusion. Our main finding was that SIII was more strongly associated with the average ventilation to blood volume ratio (${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B) than either ${\dot{\mathrm{V}}}_{\mathrm{A}}$ or V_B alone, in a model of acute bronchoconstriction induced by inhaled methacholine, followed by rescue salbutamol therapy. Although there was an acute and sharp increase in respiratory resistance and ventilation heterogeneity, SIII increase was much less pronounced. Salbutamol significantly improved respiratory mechanics but not gas exchange and increased V_B and estimated shunt fraction.

Measurements of respiratory mechanics, volumetric capnography, and synchrotron functional imaging revealed that methacholine caused a marked reduction in both alveolar ventilation and perfusion, along with an increased heterogeneity in their distribution. Although salbutamol administered after methacholine improved alveolar perfusion, lung ventilation remained reduced and inhomogeneous. The intrapulmonary shunt fraction increased persistently, while the physiologic dead space exhibited no change after these interventions. Blood oxygen levels fell, and carbon dioxide levels rose after methacholine and remained altered after salbutamol.

Methacholine inhalation induced a large increase in respiratory resistance and elastance, consistent with our previous findings in this model.¹⁹ The ventilation decreased and was significantly more heterogeneous with the appearance of regions of hypoventilation. Methacholine-induced bronchoconstriction also caused a decrease in blood volume and increased inhomogeneity of blood volume distribution. Accordingly, average ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B decreased with an increased inhomogeneity after methacholine. The direct result of these changes in ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B was a reduced gas exchange efficiency with hypoxemia and increased shunt fraction that did not improve but rather worsened after salbutamol administration. The seemingly paradoxical effects of salbutamol may be explained by its preferential delivery to well ventilated regions of the lung, which limits its effectiveness in poorly ventilated areas. This selective distribution can exacerbate ventilation–perfusion mismatch.

The effects of methacholine-induced bronchoconstriction observed here are in agreement with previous findings in the literature on the distribution of ventilation and perfusion in acute bronchoconstriction. The multiple inert gas elimination technique (MIGET) has been used to quantify and characterize ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ mismatch in asthma.¹⁵ This sophisticated technique employs the infusion and subsequent measurement of inert gases with varying solubilities to describe the ventilation–perfusion distribution within the lung. Studies employing MIGET in asthmatic subjects consistently demonstrate increased ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ heterogeneity, predominantly due to the development of poorly ventilated lung units, contributing to hypoxemia and ventilation inefficiency. Wagner et al.^15,28 have shown, through multiple MIGET-based studies, that acute bronchoconstriction induces substantial regions of low ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ ratios, reflecting airflow limitation and uneven ventilation distribution. These findings were confirmed and extended with positron emission tomography in both subjects with asthma and animal models.²⁹ Positron emission tomography studies in asthmatics have shown that during methacholine-induced bronchoconstriction, regional perfusion to ventilation defects are systematically reduced by a relative increase in pulmonary vascular resistance, possibly due to hypoxic pulmonary vasoconstriction.³⁰

In this study, while salbutamol administration improved global respiratory mechanics, it did not improve gas exchange. Indeed, low ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B regions increased, and ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B inhomogeneity worsened. These changes were associated with a concomitant increase in V_B and a tendency for an elevated shunt fraction after salbutamol. This finding is consistent with the observation of worsening gas exchange despite improved airway obstruction after β-agonist administration in some cases of severe asthma exacerbation.^28,31–33 One explanation is that airway obstruction measured in patients, as well as in the current study, is dominated by large airway resistance. In addition, inhaled salbutamol acts mainly on large airways but may not effectively reverse small airway obstruction and closure, as shown by ventilation imaging in this study. A similar observation was made by Sovijärvi et al.³⁴ in asthmatics after isoprenaline inhalation. This effect could in part be due to the heterogenous distribution of aerosol particles, and the pattern of response to airway smooth muscle contractile agonists and antagonists may be very different through the systemic route.^19,35 On the other hand, salbutamol may cause changes in cardiac output as well as pulmonary vasodilation,^36,37 which may have contributed to the extension of low ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ and a tendency for a further elevation in ${\dot{\mathrm{Q}}}_{\mathrm{s}}$.

We found a significant association between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B, in which the low ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B regions were associated with the highest SIII values. Multiple linear regression showed opposite effects of ventilation and blood volume on SIII. While a reduction in ${\dot{\mathrm{V}}}_{\mathrm{A}}$ had a strong positive effect on SIII, an increase in V_B had a moderate positive effect. This suggests that ventilation distribution has a larger effect on exhaled carbon dioxide phase III slope than perfusion distribution.

Our findings confirm and extend previous studies of the relation between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ inhomogeneity measured by MIGET or estimated based on gas exchange parameters. Tusman et al.³⁸ found a significant correlation between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ dispersion (inhomogeneity) measured by MIGET in mechanically ventilated pigs with lavage-induced lung injury measured at different positive end-expiratory pressures. In their study, there was a stronger and statistically significant association between ventilation dispersion and SIII than perfusion dispersion.³⁸ Furthermore, Dassios et al.¹⁷ assessed the relation between SIII and ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ estimated based on the relative position of the blood oxygen saturation (Spo₂) versus inhaled oxygen fraction (Fio₂) curve against the oxyhemoglobin dissociation curve of premature neonates. They found an inverse relationship between the ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ and SIII, in which lower ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ regions had higher SIII values, similar to that of ${\dot{\mathrm{V}}}_{\mathrm{A}}$/V_B and SIII in the current study.

Our study had strengths and limitations. A strength of our study is that we compare SIII and direct imaging measurements of lung ventilation and blood volume distribution, a surrogate of lung perfusion. We made the assumption that that blood volume distribution closely estimates that of lung perfusion. This a reasonable assumption based on previous imaging studies comparing both parameters.³⁹ Iodine and xenon images were not obtained simultaneously but within approximately 5 min of each other due to the time required to tune the energy of the incident radiation. However, ventilation and blood volume changes are very small over such short periods of time in this model. Therefore, we do not expect a significant bias due to the time delay between the ${\dot{\mathrm{V}}}_{\mathrm{A}}$ and V_B images. Our imaging data are limited to two axial image slices within the lung. Despite this limitation, we were able to demonstrate significant associations between imaging data, SIII, and gas exchange parameters. The upright position might affect regional ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ differences with an increased dead space due to reduced perfusion of the apices.⁴⁰ Xenon is a denser gas than air; therefore, ventilation imaging with this gas may yield regional ventilation values that are different than that with air on the absolute scale. However, these points should not have affected the comparison between the experimental conditions since the same inhaled gas composition and body position during imaging was maintained throughout the study. Finally, the relation between ventilation inhomogeneity and ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ matching is nonlinear. In other words, moderate ventilation inhomogeneity may be mitigated by active blood flow redistribution mechanisms so as to maintain adequate ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ matching. Perturbations in ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ may therefore inherently occur at more acute and severe levels of airway obstruction and ventilation inhomogeneity.⁴¹

In conclusion, we utilized synchrotron radiation K-edge subtraction imaging to directly assess the relationship between the volumetric phase III slope of exhaled carbon dioxide (SIII) and the quantitative lung ventilation and blood volume distributions and ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$ matching. Our data demonstrate a significant relationship of SIII with both ventilation and blood volume but a stronger inverse relationship with ${\dot{\mathrm{V}}}_{\mathrm{A}}/\dot{\mathrm{Q}}$. We found that ventilation distribution was a stronger determinant of SIII than blood volume, a surrogate of lung perfusion distribution. Physicians using exhaled carbon dioxide monitoring should be aware that active mechanisms adapting lung ventilation to perfusion may blunt the detection of ventilation inhomogeneity by SIII. Moreover, we confirm the hypothesis that the paradoxical worsening or lack of improvement in gas exchange despite mechanical improvement after salbutamol in acute bronchoconstriction is due to increased perfusion of low ${\dot{\mathrm{V}}}_{\mathrm{A}}$ regions and shunt fraction. These findings shed new light on the physiologic meaning of SIII, particularly in the context of monitoring of lung diseases with acute airway obstruction and lung heterogeneity.

Acknowledgments

The authors thank Herwig Requardt, Ph.D., (European Synchrotron Radiation Facility; ESRF, ID-17), Alberto Bravin, Ph.D., (ESRF, ID-17), and the ESRF ID-17 technical staff for their assistance.

Research Support

Supported by France Life Imaging (France) grant No. ANR-11-INBS-0006, by the European Synchrotron Radiation Facility (Grenoble, France), by INSERM UA07 STROBE (Grenoble, France), and by Labex PRIMES grant No. ANR-11-LABX-0063 (Lyon, France).

Competing Interests

The authors declare no competing interests.