The effect of gas exchange on multiple-breath nitrogen washout measures of ventilation inhomogeneity in the mouse

Abstract

Inert-gas washout measurements using oxygen, in the lungs of small animals, are complicated by the continuous process of oxygen consumption (V̇o₂). The multiple-breath nitrogen washout (MBNW) technique uses the alveolar slope to determine measures of ventilation inhomogeneity in the acinar (S_acin) and conducting (S_cond) airway regions, as well as overall inhomogeneity, as determined by the lung clearance index (LCI). We hypothesized that measured ventilation inhomogeneity in the mouse lung while it is alive is in fact an artifact due to the high V̇o₂ in proportion to alveolar gas volume (Va), and not ventilation inhomogeneity per se. In seven male C57BL/6 mice, MBNW was performed alive and postmortem to derive measures with and without the effect of gas exchange, respectively. These results were compared with those obtained from an asymmetric multibranch point mathematical model of the mouse lung. There was no statistical difference in S_acin and LCI between alive and postmortem results (S_acin alive = 0.311 ± 0.043 ml⁻¹ and S_acin postmortem = 0.338 ± 0.032 ml⁻¹, LCI alive = 7.0 ± 0.1 and LCI postmortem = 7.0 ± 0.1). However, there was a significant decrease in S_cond from 0.086 ± 0.005 ml⁻¹ alive to 0.006 ± 0.002 ml⁻¹ postmortem (P < 0.01). Model simulations replicated these results. Furthermore, in the model, as V̇o₂ increased, so did the alveolar slope. These findings suggests that the MBNW measurement of S_cond in the mouse lung is confounded by the effect of gas exchange, a result of the high V̇o₂-to-Va ratio in this small animal, and not due to inhomogeneity within the airways.

the distribution of ventilation is inhomogeneous, even in the healthy lung. In lung disease, however, ventilation inhomogeneity increases. The degree of inhomogeneity, especially within the small airways, has been demonstrated as an important feature in a number of common respiratory conditions (33, 34), in terms of both pathophysiology and patient-centered outcomes, such as symptoms and disease control (10, 33).

One of the most widely used noninvasive methods for measuring ventilation inhomogeneity in humans is the multiple-breath nitrogen washout (MBNW) technique. This test consists of inspiring several consecutive breaths of 100% oxygen (O₂) until the nitrogen (N₂) concentration in the lungs is reduced to <2%. Each breath generates an alveolar slope, the evolution of which has been used to measure ventilation inhomogeneity that is a result of convective gas flow (termed S_cond) the site of origin which is thought to be proximal to the so-called diffusion front. The technique also generates a measure that results from inhomogeneity in gas mixing due to diffusion inhomogeneity (termed S_acin) occurring peripheral to the diffusion front. S_acin results from the interaction between convection and diffusion gas transport in the background of heterogenous airway architecture, that is, unequal narrowing of parallel airways or differences in volume subtended by two daughter branches (39).

The MBNW technique has been standardized (24) and used in numerous physiological and clinical studies involving human adults and children (23). Additionally, MBNW has been applied to large animals (8, 11, 12, 25), as well as very small laboratory animals (15, 18, 19, 22, 30, 38), as a means of investigating animal models of human lung disease. However, difficulties arise when interpreting measurements of gas mixing efficiency on small animals due to the relative degree of gas exchange that occurs and the potential effects it may have on the measured alveolar slope during MBNW. Although not studied extensively, previous washout studies in small animals have avoided the influence of gas exchange by performing experiments postmortem (14, 38). Clearly, however, there are situations in which MBNW experiments performed on live animals are desirable, such as for longitudinal studies.

The effect of gas exchange on MBNW measurements has only indirectly been studied in rats. Verbanck et. al. (38, 40) demonstrated that the rat postmortem does not have a measurable S_cond. The authors proposed that the reason for this was due to the relatively short path lengths within the rat lung; therefore, the diffusion front was likely to be located very proximally. In a series of other studies in humans, Cormier and Belanger (4, 5) looked at the effect of gas exchange on the alveolar slope. In those studies, it was demonstrated that gas exchange accounts for up to 10–28% of the alveolar slope under normal conditions. However, this was demonstrated for only the single-breath washout (SBW) method, which involves a single vital capacity inspiration, in contrast to MBNW, which requires consecutive 1-liter inspirations in adult humans (24).

Previous studies have suggested that, while the respiratory exchange ratio (RER) is similar between small and large animals, the rate of oxygen consumption (V̇o₂) per unit alveolar gas volume (Va) is significantly greater in the small animal. It has been demonstrated that V̇o₂ generally increases as the body surface area-to-mass ratio increases (16). Consistent with these findings, mouse experiments have shown significantly higher weight-corrected V̇o₂ values than for larger mammals (9).

The mouse is commonly used to model a number of human lung pathophysiologies, such as asthma. Accordingly, measures of ventilation inhomogeneity that are recognized in human studies are valuable is evaluating such mouse models. Since the MBNW technique requires inspiration of O₂-rich gas, we hypothesized that, when V̇o₂ in proportion to Va is high, such as in the mouse, it would effectively concentrate the resident N₂. This would result in a steeper alveolar slope, which is an artifact and not related to inhomogeneity in gas mixing due to heterogenous airway structure. The aim of the study was to perform the classic human MBNW technique in the mouse and assess whether gas exchange has an effect on indexes of ventilation inhomogeneity in this widely used laboratory animal. The alveolar slope in mice was measured while the mice were alive and postmortem to, respectively, derive indexes with and without gas exchange. The experiment results were then compared with those generated from a mathematical model of the lung in which V̇o₂ could be altered.

METHODS

MBNW was performed in seven male C57BL/6 mice weighing 27.2 ± 0.5 g. All mice were healthy and 12–14 wk of age. The study protocols complied with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and was approved by the animal ethics committees of Monash University and the Alfred Medical Research & Education Precinct.

Animal preparation.

MBNW measurements were performed within 15 min post-anesthetic induction using a combination of ketamine and xylazine (75 mg/kg) agents that were mixed and delivered intraperitoneally (12 μl/g body wt). A tracheotomy was performed by a transverse cut between two tracheal rings in the upper third of the trachea. A pediatric double-lumen central venous catheter (CS-15402-E, Arrow) was used to provide isolated inspiratory and expiratory gas channels to minimize dead-space gas mixing. The catheter was advanced ∼1 cm into the trachea and then secured by ligation (silk 6.0). The animal was placed on a heat plate, which was enclosed within a small-animal whole body plethysmograph (FinePointe Series RC System, Buxco Research Systems). The key instruments and arrangement of the experiment apparatus for MBNW in the mouse are schematically illustrated in Fig. 1.

Fig. 1.

Schematic illustration of the apparatus and ventilation circuit for performing multiple-breath nitrogen washout (MBNW) on the anesthetized and mechanically ventilated mouse. The delivery of room air and 100% O₂ was achieved with two separate ventilators, and the switching between these gases was performed with two three-way stopcocks for the inspiratory (I) and expiratory (E) partitions of the ventilation circuit. Arrows indicate the direction of gas flow.

Following the tests from when the mice were alive, experiments were subsequently repeated postmortem. The mice were euthanized by CO₂ asphyxiation. Postmortem tests were completed no longer than 10 min posteuthanasia to avoid the effects of rigor mortis (38).

Washout protocol.

Washouts were performed using two mechanical ventilators (Fig. 1) set at a tidal volume (Vt) of 0.25 ml and a respiratory rate of 120 breaths/min. A positive end-expiration pressure of 1 cmH₂O was also applied to minimize end-expiratory airway collapse (26).

Expired N₂ concentration (%) was measured using an emission spectrophotometry N₂ analyzer (MedGraphics) with <50-ms response time (10–90% full scale). Directly attached to the ionization chamber of the analyzer was the needle valve, adjusted to a sample flow rate of 6 ml/min. The ventilation circuit was initially set to a ventilator providing only room air (Model 683 Small Animal Ventilator, Harvard Apparatus). Once a stable Vt was achieved, the circuit was then promptly switched at end expiration to a separate ventilator (FinePointe Series RC System, Buxco Research Systems), which delivered 100% O₂ to wash out the N₂ from the lungs. The ensuing washout was recorded until ∼2% N₂ was measured, at which point the circuit was switched back to the room-air ventilator. All measurements were performed three to four times.

MBNW analysis.

The calculation of S_acin and S_cond was based on a previously described method (6, 21). Briefly, this analysis method consisted of initially plotting expired N₂ concentration against expired volume for each breath. The alveolar slope of each breath was then calculated by fitting a linear regression to the phase III portion of the N₂ expiration. The alveolar slope was divided by the average expired N₂ concentration of that breath to compute the normalized alveolar slope (Sn). S_cond was quantified as the slope of the best fit linear regression of Sn as a function of turnover (TO), over the range of 1.5–6.0 TOs. S_acin was quantified by multiplying S_cond by the first lung TO and then subtracting the value from the normalized slope of the first breath (39). As a measure of overall ventilation inhomogeneity, the lung clearance index (LCI) was also calculated as the cumulative net expired volume divided by functional residual capacity (FRC) derived from MBNW (3).

Mathematical model.

To facilitate the interpretation of the experiment results, a mathematical model of the mouse lung was employed to simulate the MBNW experiments described above. The transport and consumption of O₂ in the mouse lung was computationally modeled following the methods of Swan et. al. (31) for the human lung, with appropriate parameterization to the mouse. Details of the model have been published elsewhere; only the methods for its adaptation to the mouse are given here.

Geometric model.

Conducting and respiratory airway models with dimensions consistent with Oldham and Robinson (20) were used. A 12-generation single-path symmetric model was adopted for the conducting airways (2,048 acini). The number of acini was midway between the estimate of Oldham and Robinson (20) for Balb/c mice (∼1,500) and the estimate of Vasilescu et al. (37) for C57BL/6 mice (∼5,300 and ∼3,800 acini in the lungs of relatively young and old mice, respectively). The lengths and diameters of the first three generations were set, and other values chosen to give intrathoracic (i.e., excluding the upper airway) dead space volume of 60 μl, and terminal bronchiole length (0.16 mm) and diameter (0.11 mm) consistent with Oldham and Robinson (20). The equipment dead space in the experimental study was assumed to increase the total dead space by 60 μl. The typical path (symmetric) respiratory airway model from Oldham and Robinson (20) was appended to a two-trumpet model with the minimum asymmetry that gave first breath normalized slope consistent with the first breath numbers seen in these mice (see Fig. 3). Dimensions in the model were scaled to give total respiratory volume at FRC of 300 μl.

Fig. 3.

Alveolar slope normalized (Sn) as a function of lung turnover (TO) from MBNW experiments (average ± SD) in anesthetized mice when alive (□) and postmortem (○). The observed difference in the Sn rise as a function of TO is due to the effect of gas exchange.

Functional model: parameterization and assumptions.

The parameters for the O₂ transport and exchange model of Swan et al. (31) [gas exchange surface area, capillary blood volume, baseline metabolic V̇o₂, and baseline metabolic CO₂ production (V̇co₂)] were determined for a 27.2-g mouse, using equations from Geelhar and Weibel (13) and Desai et al. (9). The mixed-venous Po₂ (Pv̄_O2) was initialized to 40 Torr. Minute ventilation and Vt were set as per the physiological experiments described above. Cardiac output was assumed to be equal to alveolar ventilation.

To simulate the experimental protocol, the model was first ventilated on room air (inspired O₂ fraction = 0.21) until a pseudo-steady state was reached (defined as a <1% difference in the end-expired Po₂ between subsequent breaths). The ventilated air was not humidified during the experiment; however, air was assumed fully humidified (by evaporation from the airway walls) on entry to the respiratory airways. During the initialization breaths, the Pv̄_O2 was adjusted at the end of each breath using the Fick equation to maintain the specified V̇o₂. Once the pseudo-steady state was reached (typically <10 breaths), a washout was simulated by setting inspired O₂ fraction = 1.00 and recording the expired gas concentrations over eight breaths. Pv̄_O2 did not change during the washout.

During ventilation, a volume change due to O₂ flux was calculated as the difference between instantaneous V̇o₂ and V̇co₂ within the model.

Statistical analysis.

The Wilcoxon matched-pairs test was used to detect differences in S_acin, S_cond, and LCI between alive and postmortem experiments. The statistical significance level was set at P = 0.05.

RESULTS

A typical trace of N₂ vs. time and flow vs. time from a mouse is shown in Fig. 2. MBNW derived FRC = 357 ± 40 μl from antemortem experiments, and FRC = 359 ± 50 μl from postmortem experiments (nonsignificant).

Fig. 2.

Representative time course of the N₂ concentration (%; top) and flow signal (bottom) throughout a MBNW experiment from an anesthetized mouse. While the mouse is ventilated with room air, the inspiratory gas is rapidly switched to 100% O₂. The mean expired N₂ concentration diminishes in a typical exponential-like manner.

Sn (mean ± SD) of each breath is represented against lung TO in Fig. 3 for both alive and postmortem washouts. It is possible to recognize purely from visual inspection the distinct differences in the evolution of Sn as the washout proceeds for the two conditions (alive and postmortem). There was a progressive increase in Sn when tests were performed alive. In contrast, there was an initial increase in Sn beyond which Sn remains unchanged during postmortem testing.

Figure 4, A–C, provides individual data for calculated S_acin, S_cond, and LCI for each mouse in both conditions. There was no statistical difference in mean S_acin between the mice alive (gas exchange present) and postmortem (S_acin alive = 0.311 ± 0.043 ml⁻¹ and S_acin postmortem = 0.338 ± 0.032 ml⁻¹) or LCI (LCI alive = 7.0 ± 0.1 and LCI postmortem = 7.0 ± 0.1). However, there was a significant decrease in the calculated S_cond from 0.086 ± 0.005 ml⁻¹ alive to 0.006 ± 0.002 ml⁻¹ (P < 0.01) postmortem (Fig. 4B).

Fig. 4.

Experimental change in index of acinar airway ventilation inhomogeneity (S_acin; A), index of conductive airway ventilation inhomogeneity (S_cond; B), and lung clearance index (LCI; C) for each mouse (mean ± SD) with O₂ consumption (V̇o₂) equal to 100% and 0% of predicted normal. V̇o₂ had a significant effect on S_cond, but not on S_acin or LCI. ns, Nonsignificant.

Figure 5 is a plot of Sn against TO from the mathematical model at different V̇o₂ values. Also overlaying the multiple Sn TO curves are the mouse experiment data. The Sn vs. TO curve for the mouse data and model data was closest when the model simulations were executed with a V̇o₂ 100% of that previously published for a mouse (9). Importantly, as V̇o₂ increases in the model, so does the slope of the Sn TO curve; however, there was no change in the offset (S_acin) of the Sn TO curve. Therefore, as V̇o₂ was increased in the mathematical model, the calculated S_cond increased, but there was no change in the calculated S_acin.

Fig. 5.

Sn as a function of lung TO from MBNW experiments (□ = alive, ○ = postmortem) and mathematical model simulations with varying V̇o₂ values (lines). As V̇o₂ is increased in the simulation, the evolution of Sn becomes greater. A typical value for V̇o₂ for mice in this study is ∼1.36 ml/min. Values are means ± SD.

DISCUSSION

We have demonstrated that gas exchange has a profound effect on the evolution of the normalized alveolar slope during MBNW in the mouse lung. In both an animal model and also in an anatomically realistic mathematical model of the lung, as V̇o₂ increases, so does the rate of increase in Sn. This has a direct effect on the calculation of an increasingly accepted measure of ventilation inhomogeneity, specifically the measure of inhomogeneity in the small conducting airways, S_cond. Importantly, in this circumstance, the increase in S_cond associated with increasing V̇o₂ is not due to any structural or conformational changes within the lung, but is purely a measurement artifact produced by the continuous gas exchange process. Hence the measurement of S_cond in the mouse is not a valid measure of ventilation inhomogeneity in the small conducting airways. The results from our physiological animal experiment were supported by simulations obtained from a theoretical mathematical model.

The calculated S_cond being almost absent when gas exchange is abolished is consistent with previous observations in rats (38, 40), that is, the diffusion front being located very proximally within the airway tree. Therefore, purely convection-dependent ventilation inhomogeneity (as measured by S_cond) is not an intrinsic mechanism in the mouse lung. The mechanism for the effect of V̇o₂ on the way in which the alveolar slope increases during a washout maneuver resides in the relative degree of V̇o₂ in small animals. The V̇o₂ relative to the given size of the animal, in particularly the size of Va, is significantly greater in the mouse. Previous studies have determined that, for these mice, on average, V̇o₂-to-V_A ratio (V̇o₂/Va) ∼ 1.11 min⁻¹ (27). For the average adult human (i.e., age: 57 yr, height: 175 cm, weight: 75 kg, race: Caucasian, sex: male), normative studies have determined that V̇o₂/Va ∼ 0.04 min⁻¹ (7, 35). Therefore, the V̇o₂/Va in the mouse is more than ∼27 times higher than that of the adult human. The substantially larger V̇o₂ per unit of Va in the mouse leads to the concentrating effect of N₂, such that, when an O₂-rich gas is used for N₂ washout, there is an increase in the N₂ alveolar slope on expiration.

The RER in the mouse is ∼0.80 (27), which is similar to that in humans. Therefore, the CO₂ diffusing into the alveolar replaces only 80% of the gas volume that O₂ consumed when it diffused out of the alveolus. This difference in rate of exchange of O₂ and CO₂ while alveolar gas pressure remains constant leads to a concentrating effect of N₂ (1). However, the effect is magnified in the presence of a small Va and when the Va is decreasing during exhalation. This effect is increased by the breathing of 100% O₂, leading to a non-steady-state period of gas exchange during the washout period.

Figure 6 demonstrates the degree of volume shrinkage in alveolar gas as a result of RER < 1, in the mathematical model as V̇o₂ increases. In this figure, as V̇o₂ increases, the relative Va shrinkage throughout each lung TO increases. If the percentage of Va shrinkage increases throughout the breathing maneuver, this will concentrate N₂, leading to an increase in the alveolar slope. This increase in slope is not due to inhomogeneity in gas mixing per se and is purely an artifact of the relatively efficient gas exchange in the mouse. This result is also further shown in Fig. 7, in which the alveolar slope against lung TO is plotted for 100% normal V̇o₂ (9) with an RER = 0.80 and also when instantaneous V̇o₂ and V̇co₂ are matched (i.e., RER = 1.00). From this figure, when instantaneous V̇o₂ and V̇co₂ are matched, the alveolar slope becomes close to zero. There is still some small residual increase in alveolar slope due to the uptake of CO₂ in tissues.

Fig. 6.

Percentage of alveolar gas volume shrinkage as a function of lung TO for varying V̇o₂ values from mathematical model simulations.

Fig. 7.

Sn as a function of lung TO from the mathematical model. Top line is the model executed with a V̇o₂ = 1.36 ml/min and a respiratory exchange ratio (RER) of 0.80. Bottom line is when instantaneous V̇o₂ and CO₂ production are matched (RER = 1.00).

In the only other alveolar slope analysis performed in rodents, Verbanck et al. (38, 40) demonstrated that, in rats, the calculated S_cond was zero. Our results agree with this, as we also demonstrated that S_cond was zero in the mouse lung postmortem. Importantly, the apparent S_cond in a small animal while it is alive is due to the gas exchange process; hence measuring ventilation inhomogeneity as a result of structural changes in the small conducting airways is impractical using this technique.

In humans, other authors have also looked at the effect of gas exchange on the single-breath N₂ test. Cormier and Belanger (2, 4, 5) and also Van Liew and Arieli (36) have shown that gas exchange accounts for between 10 and 28% of the alveolar slope under normal conditions in humans. This was shown by comparing the alveolar slopes between a normal SBW and a reversed SBW technique, where O₂ was used as the residual gas and room air as the inspired gas. The difference in the slopes were found to be attributable to the continuous V̇o₂ during expiration. Ventilation inhomogeneity generated by a SBW test is similar to the measurement S_acin; however, the alveolar slope is generated from a vital capacity maneuver instead of a 1-liter inspiration from FRC. It should be highlighted that the alveolar slope generated from the 1-liter inspiratory breath is steeper than that from the vital capacity method. Therefore, if gas exchange has a relatively small role in producing the alveolar slope during a vital capacity SBW, then that effect should be even smaller for a 1-liter tidal breath, making for only a negligible effect of continuing gas exchange on the first breath of the MBNW, that is, S_acin.

Since the V̇o₂/Va is much smaller in the human than in the mouse, we would expect that the effect of gas exchange in the adult human lung is likely to be relatively small. This is yet to be determined experimentally for human MBNW; however, recent mathematical modeling of the human airway tree (32) demonstrated that Sn only increased progressively during the washout simulation when the gas exchange terms were included in the model, and that, when gas exchange was excluded, Sn remained unchanged from the first breath. This was found to be the case whether a symmetrical or asymmetrical acinar tree model was used.

Our findings clearly have implications when the MBNW is used to assess models of lung disease in mice and other small laboratory animals. There are also potentially important implications in human studies, specifically those involving children and infants (28). Although the V̇o₂/Va in children is similar to that in adults with values ∼0.10 min⁻¹ (17, 29), which is still considerably smaller than mice (∼1.1 min⁻¹), infants and neonates are likely to have values of ∼0.20 min⁻¹ (17). With a V̇o₂/Va of ∼0.20 min⁻¹, it is plausible that the effect on the alveolar slope that we have demonstrated in the mouse is also likely to occur to some degree in the neonate and also infant lung.

While every effort was made to accurately apply the conventional human MBNW technique to the mouse, modifications were ultimately required. It was necessary to perform the technique in the mouse with the aid of anesthesia and with artificial ventilation. These may have served to alter the Sn values obtained to some degree. Despite these technical challenges and differences to the human version of the measurement, the results compare very well to the mathematical model of the mouse lung.

In conclusion, we have demonstrated that the MBNW technique in the mouse is practical; however, it is profoundly affected by gas exchange. The reason for gas exchange having such a large effect on the MBNW measurement was due to the concentrating effect of N₂ gas as a result of high V̇o₂/Va in the mouse. Our findings suggest that, while the measure of S_cond in living mice is likely invalid, due to the effects of continuing gas exchange, S_acin and LCI derived from MBNW are valid measures for exploring ventilation inhomogeneity in the mouse.

GRANTS

Financial support was provided to B. R. Thompson from the National Health and Medical Research Council of Australia (486101 and 491103). G.K. Prisk was supported by the National Institutes of Health (Grant R01 HL104118).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: M.D. and B.R.T. conception and design of research; M.D. and S.G.R. performed experiments; M.D., G.K.P., M.H.T., and B.R.T. analyzed data; M.D., G.K.P., M.H.T., and B.R.T. interpreted results of experiments; M.D. and B.R.T. prepared figures; M.D. and B.R.T. drafted manuscript; M.D., G.K.P., S.G.R., M.H.T., and B.R.T. edited and revised manuscript; M.D., G.K.P., S.G.R., M.H.T., and B.R.T. approved final version of manuscript.