Topographic Basis of Bimodal VentilationeCPerfusion Distributions during Bronchoconstriction in Sheep
http://www.100md.com
美国呼吸和危急护理医学 2005年第4期
Department of Anesthesia and Critical Care and Department of Medicine (Pulmonary and Critical Care Unit), Massachusetts General Hospital
Harvard Medical School, Boston, Massachusetts
ABSTRACT
The distribution of ventilationeCperfusion (A/) ratios during bronchoconstriction measured with the multiple inert gases elimination technique is frequently bimodal. However, the topographic basis and the cause of that bimodality remain unknown. In this article, regional A/ is quantified by three-dimensional positron emission tomography (PET) imaging of methacholine-induced bronchoconstriction in sheep. Regional A/ ratios were calculated from the imaged kinetics of intravenously injected 13NN-saline bolus, assembled into global A/ distributions, and used to estimate gas exchange. During bronchoconstriction, large regions with impaired tracer washout were observed adjacent to regions of normal ventilation. PET-derived A/ distributions during bronchoconstriction were consistently bimodal, with areas of low A/ receiving a large fraction of . The standard deviation of the A/ distribution was 38% lower if small-scale (subresolution) heterogeneity (< 2.2 cm3) was ignored. Arterial blood gases predicted from PET data correlated well with measured values for PaO2 (r2 = 0.91, p < 0.01) and PaCO2 (r2 = 0.90, p < 0.01). We conclude that the bimodality of A/ distributions in bronchoconstriction reflects the involvement of large contiguous regions of hypoventilation with substantial subresolution intraregional A/ heterogeneity. Assessment of the subresolution A/ heterogeneity is therefore essential to accurately quantify global gas exchange impairment during bronchoconstriction.
Key Words: lung imaging positron emission tomography pulmonary gas exchange theoretical models
Heterogeneity in the distribution of alveolar ventilation (A) to perfusion () is the main determinant of gas exchange impairment during bronchoconstriction in humans and animals. Using the multiple inert gases elimination technique (MIGET), Wagner and coworkers (1) observed bimodal blood-flow distributions of A/ ratios in most patients with asymptomatic asthma.
Bimodal or broadened unimodal A/ distributions were also observed in patients with chronic (2) and acute severe asthma (3). In dogs, using MIGET, Rubinfeld and colleagues (4) showed broadened unimodal A/ distributions in mild bronchospasm and bimodal distributions in severe bronchospasm. Collateral ventilation of units distal to closed airways was invoked as a mechanism explaining the bimodality of the distributions. A/ distributions obtained with MIGET provide accurate description of gas exchange. However, MIGET-derived complex A/ distributions may be nonunique (5, 6), and the bimodality of the A/ distribution during bronchoconstriction has not been confirmed by other methods. Furthermore, because MIGET does not contain topographic information on the measured A/, the structural basis for the bimodal distributions has remained elusive.
Airway resistance in asthma is significantly affected by small airways constriction (3), although constriction may occur at different levels of the airway tree (7, 8). It follows that A/ heterogeneity in the lungs during bronchoconstriction may be manifested at several length scales. Several imaging methods have been used in the attempt to directly and topographically assess ventilation or A/ distributions during airway obstruction (7, 9, 10). Studies with inhaled radionuclides (7, 11, 12), magnetic resonance (13), single-photon emission computed tomography (14), and computed tomography (15) have all shown significant spatial ventilation or A/ heterogeneity in patients with asthma. However, those imaging techniques have not been able to quantify A/ ratios to the point of predicting the level of global gas exchange impairment, and, as a result, there are limited data on the functional impact of the topographic A/ changes during airway obstruction.
We recently described a method to compute A/ heterogeneity based on positron emission tomography (PET) data (16), which provided satisfactory agreement between experimental arterial blood gases and estimates from PET-derived A/ distributions. The method follows the regional kinetics of an intravenous bolus of 13NN-labeled saline and not only quantifies the imaged topographic heterogeneity in A/ but also the heterogeneity occurring at length scales below the spatial resolution of the imaging instrument (2.2 cm3). Lung structures underlying 2.2 cm3 correspond to secondary pulmonary lobules and thus include multiple terminal acini that, on theoretic grounds, have been linked to the bimodal shape of the A/ distribution during bronchoconstriction (17, 18). Subresolution (intraregional) A/ heterogeneity is taken into account by analyzing the tracer kinetics at the level of the smallest image volume component (voxel), and exploring its compatibility to either a heterogeneous (two-compartment) or a homogeneous (one-compartment) washout pattern.
We reasoned that A/ heterogeneity occurring at length scales smaller than 2.2 cm3 could have a significant contribution to the overall A/ heterogeneity and to the bimodality in A/ distribution observed during bronchoconstriction. To test this hypothesis, we assessed the contribution of A/ heterogeneity originating from length scales below the imaging resolution to global A/ heterogeneity in a sheep model of methacholine-induced bronchoconstriction. Some of the results of these studies have been previously reported in the form of abstracts (19, 20).
METHODS
Animal Preparation
The study protocol was approved by the Massachusetts General Hospital Committee on Animal Care. Six normal sheep weighing 17 kg (range 15eC21 kg) were anesthetized; intubated; mechanically ventilated; prepared with a femoral arterial catheter, a pulmonary artery catheter, and an independent central venous catheter for tracer injection; and placed in the prone position. The ventilator (Harvard Apparatus, Millis, MA) was set at an inspired oxygen fraction (FIO2) = 0.49 ± 0.02, positive end-expiratory pressure = 5 cm H2O, VT = 17 ± 2 cc/kg, and inspiratory time of 30% of the breathing period. The respiratory rate (12 ± 2 bpm) was set to normocapnic arterial blood gases at the beginning of the experiment and maintained at that value for the rest of the experiment. Alveolar ventilation was computed as A = (VT eC VD) x respiratory rate, where VD = dead space was estimated from the animal's weight (21).
PET Imaging Protocol and Processing
Emission and transmission scans were performed as previously described (16, 22). The animal was positioned in the PET camera with the most caudal slice adjacent to the diaphragm dome. The PET camera (Scanditronix PC4096; General Electric, Milwaukee, WI) collected 15 transverse cross-sectional 6.5-mm slices over a 9.7-cm-long axial field.
The tracer 13NN gas ( 10-minute half-life) was generated by a cyclotron and dissolved in degassed saline. The imaging protocol for the emission scans started with a tracer-free lung. The ventilator was turned off at end exhalation and a bolus of 13NN saline solution was injected into a central vein. Simultaneously, collection of six consecutive images of 5-second duration was started while the animal was kept in apnea for 30 seconds. Subsequently, mechanical ventilation was resumed and four images of 30 seconds and two images of 60 seconds were collected. Resulting images consisted of an interpolated matrix of 128 x 128 x 15 voxels of 6 x 6 x 6.5 mm. Images were low-pass-filtered to 13 x 13 mm in the image plane, and a two-point moving-average filter was applied in the z plane to a final volumetric resolution of approximately 2.2 cm3.
Once the animal was positioned in the PET camera and in steady state, a set of physiologic measurements, transmission, and emission scans was collected (control). After these, methacholine solution (25 mg/ml) was delivered through the inspiratory line with an ultrasonic nebulizer (Ultra-Neb 99; DeVilbiss Co., Somerset, PA). The methacholine dose was titrated to double control peak airway pressure (Ppeak), and nebulization was maintained during subsequent imaging. A new set of physiologic, transmission, and emission scans was obtained in this new condition (bronchoconstriction) approximately 35 minutes after control measurements. Physiologic measurements were performed within 10 minutes of PET measurements once steady-state conditions (i.e., stable arterial and pulmonary blood pressures and Ppeak) were obtained.
Regional Perfusion and A/
Regional perfusion (r) and shunt were computed from the apneic phase of the tracer kinetics (23, 24). r was estimated from the mean activity during the apneic plateau phase.
To assess the relevance of subresolution (intraregional) A/ heterogeneity to global A/ heterogeneity, we used two methods to compute regional A/ as follows: (1) homogeneous regional A/, where the voxel was considered a uniform compartment and regional A/ was computed from the integral of tracer activity along the washout, including an exponential correction in the cases of incomplete washout at the end of the imaging period (16); (2) heterogeneous regional A/, in which regional A/ was calculated with a recently described technique (16) where the regional tracer washout plot was used to classify (Figure 1) and compute intraregional specific ventilation (sr) and r/sr. A voxel was classified as homogeneous when the semilog washout plot approached linearity and was treated as a single compartment whether ventilating or complete air trapping. Conversely, a voxel was classified as heterogeneous when the semilog washout plot was curvilinear and was analyzed as made of two compartments, either both ventilating or one ventilating and one air trapping. Distributions of A/ in absolute units were estimated using thermodilution cardiac output, total shunt, VT, estimated anatomic dead space, and respiratory rate. Total shunt was estimated by adding the shunt derived from global tracer kinetics to the component corresponding to regional subcompartments with complete gas trapping (A/ = 0). We assumed that sr was proportional to regional ventilation r and thus r/sr was proportional to regional A/.
Gas Exchange Computations
Regional A/ compartments were grouped into 100 bins of equal log (A/) width (0.05) ranging from eC3 to 2. The square root of the second moment of the perfusion-weighted A/ distribution about its mean on a log scale (SD) was used as an indicator of A/ heterogeneity (2). Global alveolar and blood O2 and CO2 partial pressures were estimated by computing partial pressures and contents from the obtained A/ distributions (16, 25).
Statistical Analysis
Data were expressed as mean ± SD. Comparisons between values before (control) and during bronchoconstriction were made with a two-tailed Student's t test for paired samples. Linear correlation and biases (26) were used to summarize the relationship between the measured and estimated blood gases. Statistical significance was taken at the p < 0.05 level.
RESULTS
Significant bronchoconstriction was achieved in this animal model as expressed by the marked changes in respiratory mechanics and gas exchange during inhaled methacholine (Table 1). Ppeak during methacholine inhalation was 2.3 times baseline Ppeak, as required by protocol design. In parallel with this rise in Ppeak, arterial and mixed-venous PO2 were significantly reduced, and PaCO2 was significantly increased whereas right-to-left shunt was minimally changed. Changes in hemodynamics during inhaled methacholine were a decrease in mean arterial pressure and a small increase in mean pulmonary artery pressure, with no significant changes in pulmonary artery occlusion pressure or cardiac output (Table 1). PET imaging scans during bronchoconstriction were characterized by significant amounts of residual tracer remaining in the lung fields at the end of the washout period, in contrast to the nearly complete elimination of the tracer during control conditions (Figure 2). Tracer distribution during bronchoconstriction showed large regions of tracer retention, segmental to subsegmental size, adjacent to regions of nearly complete tracer washout (Figure 2). Compared with such changes in regional ventilation, topographic changes in perfusion were much smaller in these prone sheep (Figure 2). No evidence of development of pulmonary edema was observed during the study, with average lung gas fraction estimates derived from the transmission scans amounting to 0.54 ± 0.07 before and 0.54 ± 0.09 during bronchoconstriction. The gravity dependence of lung gas fraction was variable among animals and no consistent change was observed with methacholine.
The four different patterns of voxel washout kinetics described in METHODS (Figure 1) were observed during the experiments (Table 2). Before bronchoconstriction, the predominant pattern was that of a single "fast" compartment leaving less than 10% of residual tracer at the end of the washout (Table 2). During bronchoconstriction, a significantly larger number of areas presented abnormal washout patterns with complete or partial gas trapping, two-compartment washout, and slow single-compartment washout (Table 2). The component of gas trapping was negligible in control conditions and increased to 7.0 ± 4.1% during bronchoconstriction.
In all sheep studied, PET-based A/ distributions were unimodal and narrow during control conditions (Figure 3, left panel), with small values for SD (Table 3). During bronchoconstriction, A/ distributions became significantly wider than those during control. The calculation of A/ distributions considering subresolution heterogeneity, derived with a two-compartment model of voxel tracer washout, revealed broad and bimodal A/ distributions in all animals (Figure 3, middle panel), with significant increase of SD in bronchoconstriction but not in control conditions (Table 3).
Estimates of arterial blood gases based on these computed A/ distributions were correlated with experimental measurements for both O2 and CO2. PET-based estimates of PaO2 described measured PaO2 with the regression equation: measured PaO2 = 0.965 · estimated PaO2 eC 8.8 mm Hg, r2 = 0.91 (p < 0.01; Figure 4). The mean difference between measured and estimated PO2 was eC14.3 ± 32.9 mm Hg. Similarly, PET-based estimates of PaCO2 described measured PaCO2 with the following regression equation: measured PaCO2 = 0.923 · estimated PaCO2 + 4.2 mm Hg, r2 = 0.90 (p < 0.01). Estimated PaCO2 approximated PaCO2 measurements, with a mean difference of 1.4 ± 1.5 mm Hg (Figure 5).
The analysis assuming homogeneous or heterogeneous intraregional A/ data to fit the tracer washout kinetics yielded similar estimates of SD before bronchoconstriction (Table 3). In contrast, after bronchoconstriction, the analysis assuming homogeneous intraregional A/ yielded A/ distributions with significantly lower SD values than those obtained assuming a two-compartment intraregional washout model (Table 3). On average, lack of consideration of subresolution A/ heterogeneity resulted in an underestimation of SDQ of 38% when compared with the intraregional two-compartment analysis. In terms of A/ distributions, the use of a single-compartment model to analyze voxel washout yielded unimodal A/ distributions in half of the animals and bimodal A/ distributions in the other half (Figure 3, right panel). These bimodal distributions were less marked than those obtained when intraregional A/ heterogeneity was taken into account (Figure 3, middle panel). Gas exchange estimates during bronchoconstriction obtained from A/ distributions based on one-compartment analysis of regional washout overestimated measured PaO2 by 27.7 ± 41.7 mm Hg and underestimated PaCO2 by 2.0 ± 1.9 mm Hg.
DISCUSSION
The main findings of this study are as follows: (1) severe bronchoconstriction systematically changes the PET-derived A/ distributions from narrow and unimodal to wide and bimodal and (2) the bimodal shape is caused, in large part, by the contribution of A/ intraregional heterogeneity at length scales under 2.2 cm3 (the spatial resolution of the imaging method) and by the presence of units with substantially reduced ventilation that are topographically clustered in large contiguous regions and not diffusely distributed across the lungs.
Methodologic Limitations
The assumption of a two-compartment model to quantify intraregional heterogeneity in A/ (16) is a simplification because a larger number of A/ compartments could potentially be present within 2.2 cm3 of the lung. Identification of a larger number of compartments could improve the description of intraregional heterogeneity and increase its contribution to total A/ heterogeneity. In a washout curve analysis, however, the number of identifiable compartments is usually limited to a few and depends on the separation of their respective time constants, on the number of data points collected, the length of the measured washout period, and the signal-to-noise ratio of the imaging system. Given that the signal-to-noise ratio on a PET image depends on both activity and the time of acquisition, the number of images collected during the 4 minutes of washout was set to six, limiting the number of identifiable compartments to no more than two. Longer washout acquisition times during bronchoconstriction and the use of new systems with higher signal-to-noise ratio may allow for the identification of a larger number of compartments. Yet, when compared with the assumption of a uniform region, the use of two compartments provided a significant difference in the shape of the derived A/ distribution.
The presented method is based on the intravenous injection of 13NN. Consequently, regions of very low perfusion will receive lower tracer amounts and may have their tracer kinetics importantly modified by noise and tracer rebreathing during washout. This would compromise identification of regions with high A/ and low . Because regions of high A/ are not usual during bronchoconstriction (1eC4, 27), it is unlikely that their presence affected our results. In situations when such regions would be important, the use of emission scans of inhaled tracer (22, 28) allows for their identification.
A/ Distributions during Bronchoconstriction
In these normal sheep before bronchoconstriction, ventilation- and perfusion-weighted A/ distributions derived from PET scanning followed narrow and unimodal log-normal distribution functions. These findings agree with previous results using MIGET, which showed narrow and unimodal log-normal distributions of A/ ratios in humans and animals (29, 30). After severe bronchoconstriction with methacholine, all animals in the current study showed marked A/ heterogeneity and bimodal distributions. In these distributions, one mode was centered around values of A/ ratios between 1 to 3 and the second mode at values between 0.02 and 0.6. These results are consistent with the bimodal A/ distributions derived with MIGET in severely bronchoconstricted dogs with different agents (4, 31) and in humans with asthma (1, 2, 27).
MIGET-derived A/ distributions are based on global lung data, and thus should describe A/ heterogeneity in the whole lung. However, the nonunique nature of the MIGET solution requires the use of smoothing and optimization algorithms to yield a A/ distribution that best fits measured retentions and eliminations of inert gases. In contrast, the estimates of A/, perfusion, and ventilation presented here were directly derived from regional tracer kinetics, without a priori assumption on the shape of the A/ distribution. The fact that the shape of the A/ distributions obtained from PET topographic data is consistent with those derived with MIGET serves as cross-validation of both methods to assess A/ distributions in conditions of severe bronchoconstriction. In addition to yielding global lung A/ distributions, our PET technique provides information on the anatomic origin of the A/ heterogeneity not available from MIGET measurements.
Previous studies showed minimal changes in right-to-left shunt with bronchoconstriction (1, 4, 31). We also observed minimal right-to-left shunt blood flow to nonaerated regions during methacholine, as shown by a stable tracer kinetics plot during the breath-holding period. However, we were able to detect substantial presence of aerated but nonventilating regions (i.e., gas trapping) during severe bronchoconstriction. Because these units have A 0, and thus A/ 0, they contribute to functional shunt. A similar phenomenon was observed by Rubinfeld and coworkers (4) where complete obstruction of large lung regions was seen in one dog during autopsy. The larger component of gas trapping observed in our sheep compared with that reported for dogs may be from the lower amount of collateral ventilation in sheep compared with dogs (32) or to the more severe bronchospasm induced in this study as suggested by a mean PaO2/FIO2 ratio of 173 mm Hg measured in our study as compared with 255 mm Hg reported for the severe constriction in Rubinfeld and colleagues' (4) study.
Topographic Distribution of A/ Heterogeneity during Bronchoconstriction
Results by Wagner and coworkers (29) and West (33) indicated that both ventilation and blood flow curves derived from radioactive gas measurements were considerably narrower than A/ distributions obtained with MIGET in young, normal subjects. In contrast, our PET measurements, conducted with significantly higher spatial resolution than those former scintillation counter methods, showed satisfactory agreement between measured and arterial blood gas estimates during control conditions. Moreover, the use of a single- or two-compartment fit to control tracer kinetics resulted in minimal differences in measured SD for control conditions. This finding indicates that the small A/ heterogeneity of the normal lung is predominantly caused by interregional differences of structures with volumes greater than 2.2 cm3, the volumetric resolution of our images.
During bronchoconstriction, visual inspection of the emission scans at the end of washout showed that tracer retention was topographically distributed in large and contiguous regions of the lung (Figure 2), which were adjacent to regions with nearly complete washout. The presence of these large areas of tracer retention contrasts with the concept of bronchoconstriction as a process of small airway narrowing diffusely distributed throughout the lung. The mechanism leading to this spatial distribution of low A/ units is unclear. It can be related either to obstruction of large, approximately segmental level bronchi or to clustered increase in constriction of smaller neighboring regions.
When uniform intraregional A/ (single-compartment model) was assumed to analyze the washout data during bronchoconstriction, bimodal A/ distributions were observed in three of the six studied sheep, suggesting that interregional heterogeneity at length scales above 2.2 cm3 was sufficient to account for part of the bimodal distribution. However, the presence of significant intraregional A/ heterogeneity during bronchoconstriction was evidenced by the significant increase in total heterogeneity from the voxel-by-voxel analysis of 13NN kinetics and by the large number of voxels showing multiexponential washout and/or partial or complete intraregional gas trapping. Accordingly, the increase in A/ heterogeneity obtained with a double-compartment model analysis compared with that obtained with a single-compartment model indicates that significant intraregional A/ heterogeneity was present, at dimensions smaller than the spatial resolution of the imaging method in this model of bronchoconstriction. Furthermore, the A/ distributions derived without considering intraregional heterogeneity underestimated by 38% global A/ heterogeneity and, as a consequence, systematically increased the error of estimated arterial blood gases. These results also illustrate how, despite the impossibility of visualizing with PET the A/ heterogeneity at length scales under its effective spatial resolution of 2.2 cm3, a substantial amount of functionally relevant subresolution A/ heterogeneity could be uncovered and quantified from the temporal information contained in the tracer washout.
Intraregional heterogeneity was not only responsible for a significant fraction of the A/ mismatch but also accounted in great part for the bimodal characteristics of the A/ distributions. These findings, taken together, support the concept that changes in peripheral airways and, potentially, lung tissue are largely responsible for overall gas exchange impairment during bronchoconstriction. It is also clear that, during bronchoconstriction, consideration of subresolution heterogeneity is essential for adequate assessment of global A/ heterogeneity from imaging data.
Structural changes that can cause A/ heterogeneity during bronchoconstriction include heterogeneity in ventilation caused by airway narrowing, edema, hypersecretion, bronchial inflammation, and constriction of parenchymal smooth muscle. Two mechanisms have been invoked to explain the bimodal shape of the A/ distributions during bronchoconstriction. Initial investigations with MIGET attributed the presence of a low A/ mode to collateral ventilation (1, 4) of regions distal to occluded airways. This mechanism could explain how a condition believed to be diffusely distributed throughout the lung would consistently give rise to a bimodal pattern without significantly increasing right-to-left shunt. More recently, Anafi and Wilson (17), modeling the mechanical behavior of a constricted terminal airway during tidal breathing, discovered that it could attain two stable states: one effectively open and one nearly closed. Predictions from that model fit well in dynamic lung elastance and gas-mixing indices in subjects without asthma exposed to methacholine (18) and were consistent with the bimodal A/ distribution during bronchoconstriction.
Our results demonstrated that one source of the bimodal A/ distribution takes place within lung structures with volumes smaller than 2.2 cm3. Although it is impossible with the spatial resolution of PET to perform an analysis following strict anatomic limits, it is interesting to note that the 2.2-cm3 volume corresponds approximately to that of a secondary pulmonary lobule. This structure is an irregular polyhedral of 1 to 2.5 cm on each side, supplied by three to five respiratory bronchioles, frequently cited in the pulmonary imaging literature (34eC36). Although collateral ventilation could potentially take place at this level, complete and partial intraregional air trapping was observed in several lung areas during bronchoconstriction. This finding suggests that collateral ventilation was not effective in those regions, and it is unclear why it would play a role in other lung regions of the same animal. Also, it is well established that collateral ventilation is less important in the sheep than in the dog, particularly in the studied young sheep, which are known to have the highest resistance to collateral flow (37). Thus, it does not appear that collateral ventilation could play a main role in explaining the bimodal A/ distributions in our animal model. Bistable mechanical characteristics of the terminal airways were predicted for acinar units (17, 18), structures with volumes between 1.2 and 1.8 cm3 at total lung capacity (17, 18). This is compatible with the presence of at least two terminal units within our spatial resolution of 2.2 cm3 during tidal breathing. Thus, our data would suggest that intraregional A/ heterogeneity caused by bronchoconstriction was primarily caused by heterogeneous ventilation at or below the acinar level. The bistability of the terminal bronchioles described by Anafi and coworkers (17, 18) is in full agreement with the distributions measured in this study, where one mode was around normal A/ and another at a A/ value about 10 times lower.
The regional distribution of both inter- and intraregional heterogeneities was nonuniform during bronchoconstriction. The topographic distribution of the well-ventilated and poorly ventilated regions exhibited partial vertical dependence in these prone sheep. Nondependent regions tended to show less poorly ventilated areas than middle and dependent regions. This finding suggests a trend for development of more dependent airway narrowing and obstruction during bronchoconstriction. However, other factors in addition to the vertical gradient must be involved because, although three sheep presented a clear vertical progression of poorly ventilated regions, in another three sheep there was poor ventilation in middle regions with preservation of dependent areas. Nonuniform distribution of intraregional heterogeneity was characterized by the fact that, as compared with the results for the whole lung, regions with large amounts of residual tracer at the end of the washout (as shown in Figure 2 during bronchoconstriction) showed a higher fraction of voxels with complete trapping (25 ± 5%) and slow single-compartment (21 ± 8%) kinetics, a smaller fraction of voxels with partial trapping (4 ± 3%) and two-compartment kinetics (18 ± 5%), and a similar fraction of voxels with fast single-compartment kinetics (32%). Thus, large lung areas with significant tracer retention tended to present more regions that either emptied homogeneously slowly or not at all, whereas areas of overall smaller tracer retention tended to show more nonuniform kinetics. However, there is still significant heterogeneity in all lung regions.
Overall, our results demonstrate that A/ distributions during bronchoconstriction have two topographic components: one, caused by interregional A/ heterogeneity between large areas, of at least segmental size, where terminal units with highly reduced ventilation cluster; the other, because of intraregional A/ heterogeneity within lung structures under 2.2 cm3, probably a result of bistable terminal bronchial constriction.
Gas Exchange
There was a high correlation between PET-estimated and measured arterial blood gases. A larger dispersion around the regression line of measured versus estimated PaO2 values was observed for the higher range of PaO2. This finding may be, in part, because at high PaO2, hemoglobin is nearly 100% saturated and small errors in oxygen saturation can result in large errors in PaO2. Estimated values of PaO2 tended to overestimate measured PaO2. In MIGET studies, this type of result has been interpreted as consistent with diffusion limitation. Our measurements are based on the kinetics of the inert gas 13NN, which is significantly less affected by diffusion limitation than O2. In addition, no evidence of diffusion limitation was observed in normal lungs subjected to methacholine challenge (27, 31, 38).
One probable cause for the underestimation of PaO2 is the absence of steady-state gas exchange during the imaging of bronchoconstriction. The two animals with the lowest PaO2 had a significant drop in their arterial-mixed venous O2 content difference with unchanged cardiac output during bronchoconstriction, clearly reflecting extreme cardiopulmonary burden with noneCsteady-state conditions. In this case, studies in less extreme situations would be expected to yield better agreement between predicted and measured PaO2, as presented previously (16). Another potential cause of the overestimation in PaO2 is that A/ heterogeneity was underestimated by PET, which may have been from the following reasons: (1) Limited spatial resolution of PET; (2) Use of only two compartments to describe intraregional heterogeneity; (3) Limited identification of regions of very small, specific ventilation because of insufficient imaging time during washout; (4) Assumption that compartment ventilation was proportional to its specific ventilation (i.e., compartment effective ventilation scaling with regional volume). If this distribution were heterogeneous, an additional component of mismatch would exist that was not considered in the computations. In this case, the difference between measured and estimated PaO2 would be an indication of intraregional ventilation to volume mismatch; (5) Nonimaging of apical or basal diaphragmatic regions with lower A/ than imaged regions. We maximized the volume of imaged lung by adjusting the animal's position in the camera according to initial transmission scans to have approximately 80% of imaged lung within the axial field of 9.97 cm used in this study. In addition, the animals were in the prone position, which should reduce the heterogeneity in supradiaphragmatic regions. However, the presence of low A/ in nonimaged regions cannot be ruled out.
Heterogeneity in regional A/ affects the exchange of O2 and CO2 differently. Regions of high A/ affect CO2 exchange more, whereas regions of low A/ have a greater impact on O2 exchange. The experimental model allowed us to test the ability of our experimental and analysis methods to recover a broad range of A/ ratios. The severe bronchoconstriction induced in this sheep model produced large changes in A/ distribution with values of A/ ratios as low as 0.001. In addition, we used an FIO2 of 49%, which implies that proper quantification of A/ ratios less than 0.17 was needed to predict the reduction of PaO2 caused by bronchoconstriction. Thus, the correlation between measured and estimated PaO2 indicates satisfactory description of very low A/ ratios. Furthermore, the linear unbiased relationship between estimated and measured PaCO2 suggests that PET-based A/ distributions satisfactorily estimate regions with normal and high A/ ratios. Taken together, our results suggest that the very extreme changes in A/ distributions caused by bronchoconstriction are suitably described by PET-derived data.
Imaging of A/ in Bronchoconstriction and Asthma
There is no gold-standard technique that can be used to validate image-derived measures of local A/ during bronchoconstriction. Nuclear medicine scintigraphic ventilation and perfusion scans have demonstrated ventilation and perfusion unevenness in patients with mild asthma with different triggers (7, 11, 12) but at a spatial resolution too low to assess its functional significance. High-resolution computed tomography provides distinct anatomic information on caliber changes of airways of diameter greater than 2 mm (39, 40) and can assess the presence and location of gas trapping after an exhalation to residual volume (41), but no direct connection to regional gas exchange during breathing has been established for these techniques. Single-photon emission computed tomography with Technegas was used to assess airway closure in patients with asthma (14). The authors, however, recognized that the technique was unlikely to detect changes in lung regions of less than 51 ml of volume. More recently, magnetic resonance imaging with hyperpolarized helium was used for high-resolution visualization of ventilation defects in symptomatic and asymptomatic asthma (13). None of these imaging techniques were consistently related to independent measures of gas exchange, such as arterial blood gases. PET has lower spatial resolution than magnetic resonance imaging and computed tomography, is expensive, and uses radioactivity. Conversely, it has high sensitivity to minute tracer concentrations and enough temporal resolution to uncover subresolution heterogeneity from analysis of regional tracer kinetics. Our current and previous (16) results show that PET estimates of regional A/ are quantitatively related to gas exchange impairment. We recently implemented the technique to study humans (42) and preliminary results in patients with asthma (43eC45) confirm and expand our observations in animals. Thus, PET may be an attractive technique to study human asthma when more accurate quantification of gas exchange impairment combined with topographic information is desirable for clinical or research purposes. Given its quantitative value, PET could also serve as a validating tool in the development of novel pulmonary functional imaging methods.
In summary, we showed that kinetics analysis of intravenously injected 13NN measured with PET reveals significant A/ heterogeneity with a consistent bimodal distribution of A/ ratios during severe bronchoconstriction. This bimodality was caused by a combination of inter- and intraregional A/ heterogeneity. Interregional A/ heterogeneity was manifested as clusters of bronchoconstricted units in large contiguous areas of reduced ventilation. Intraregional A/ heterogeneity was evidenced as heterogeneity at subresolution length scales. This subresolution A/ heterogeneity represented an important component of total A/ mismatch and proved, therefore, to be essential to quantify gas exchange alterations during bronchoconstriction.
Acknowledgments
The authors thank Steven B. Weise, Sr., Research Technologist, PET Imaging Laboratory, Massachusetts General Hospital, for his expert support in the acquisition of images and Tobias Schroeder, Department of Anesthesia, Massachusetts General Hospital, for assistance with data analysis.
Presented, in part, at the International Conference of the American Thoracic Society, Atlanta, Georgia, May 22, 2002, and at the Annual Meeting of the American Society of Anesthesiologists, Orlando, Florida, October 15, 2002.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
REFERENCES
Wagner PD, Dantzker DR, Iacovoni VE, Tomlin WC, West JB. Ventilation-perfusion inequality in asymptomatic asthma. Am Rev Respir Dis 1978;118:511eC524.
Wagner PD, Hedenstierna G, Bylin G. Ventilation-perfusion inequality in chronic asthma. Am Rev Respir Dis 1987;136:605eC612.
Ferrer A, Roca J, Wagner PD, Lopez FA, Rodriguez-Roisin R. Airway obstruction and ventilation-perfusion relationships in acute severe asthma. Am Rev Respir Dis 1993;147:579eC584.
Rubinfeld AR, Wagner PD, West JB. Gas exchange during acute experimental canine asthma. Am Rev Respir Dis 1978;118:525eC536.
Kapitan KS, Wagner PD. Linear programming analysis of VA/Q distributions: limits on central moments. J Appl Physiol 1986;60:1772eC1781.
Evans JW, Wagner PD. Limits on VA/Q distributions from analysis of experimental inert gas elimination. J Appl Physiol 1977;42:889eC898.
Munkner T, Bundgaard A. Regional V/Q changes in asthmatics after antigen inhalation. Eur J Respir Dis Suppl 1986;143:44eC47.
Sekizawa K, Sasaki H, Shimizu Y, Takishima T. DoseeCresponse effects of methacholine in normal and in asthmatic subjects: relationship between the site of airway response and overall airway hyperresponsiveness. Am Rev Respir Dis 1986;133:593eC599.
Munkner T, Bundgaard A. Regional V/Q changes in asthmatics after histamine inhalation. Eur J Respir Dis Suppl 1986;143:22eC27.
Kuethe DO, Caprihan A, Gach HM, Lowe IJ, Fukushima E. Imaging obstructed ventilation with NMR using inert fluorinated gases. J Appl Physiol 2000;88:2279eC2286.
Agnew JE, Bateman JR, Pavia D, Clarke SW. Radionuclide demonstration of ventilatory abnormalities in mild asthma. Clin Sci (Lond) 1984;66:525eC531.
Vernon P, Burton GH, Seed WA. Lung scan abnormalities in asthma and their correlation with lung function. Eur J Nucl Med 1986;12:16eC20.
Altes TA, Powers PL, Knight-Scott J, Rakes G, Platts-Mills TA, de Lange EE, Alford BA, Mugler JP III, Brookeman JR. Hyperpolarized 3He MR lung ventilation imaging in asthmatics: preliminary findings. J Magn Reson Imaging 2001;13:378eC384.
King GG, Eberl S, Salome CM, Young IH, Woolcock AJ. Differences in airway closure between normal and asthmatic subjects measured with single-photon emission computed tomography and Technegas. Am J Respir Crit Care Med 1998;158:1900eC1906.
Jensen SP, Lynch DA, Brown KK, Wenzel SE, Newell JD. High-resolution CT features of severe asthma and bronchiolitis obliterans. Clin Radiol 2002;57:1078eC1085.
Vidal Melo MF, Layfield D, Harris RS, O'Neill K, Musch G, Richter T, Winkler T, Fischman AJ, Venegas JG. Quantification of regional ventilation-perfusion ratios with PET. J Nucl Med 2003;44:1982eC1991.
Anafi RC, Wilson TA. Airway stability and heterogeneity in the constricted lung. J Appl Physiol 2001;91:1185eC1192.
Anafi RC, Beck KC, Wilson TA. Impedance, gas mixing, and bimodal ventilation in constricted lungs. J Appl Physiol 2003;94:1003eC1011.
Vidal Melo MF, Harris RS, Call D, Layfield JDH, Venegas JG. V A/ Q distributions during bronchoconstriction derived from positron emission tomography (PET) data . Am J Respir Crit Care Med 2002;165:A706.
Vidal Melo MF, Harris RS, Layfield J, Venegas JG. Contribution of inter and intraregional V/Q heterogeneity to gas exchange impairment during bronchoconstriction assessed by PET . Anesthesiology 2002;96:A1311.
Stahl WR. Scaling of respiratory variables in mammals. J Appl Physiol 1967;22:453eC460.
Treppo S, Mijailovich SM, Venegas JG. Contributions of pulmonary perfusion and ventilation to heterogeneity in V(A)/Q measured by PET. J Appl Physiol 1997;82:1163eC1176.
Galletti GG, Venegas JG. Tracer kinetic model of regional pulmonary function using positron emission tomography. J Appl Physiol 2002;93:1104eC1114.
Mijailovich SM, Treppo S, Venegas JG. Effects of lung motion and tracer kinetics corrections on PET imaging of pulmonary function. J Appl Physiol 1997;82:1154eC1162.
Vidal Melo MF, Loeppky J, Caprihan A, Luft UC. Alveolar ventilation to perfusion heterogeneity and diffusion impairment in a mathematical model of gas exchange. Comput Biomed Res 1993;26:103eC120.
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;327:307eC310.
Rodriguez-Roisin R, Ferrer A, Navajas D, Agusti AG, Wagner PD, Roca J. Ventilation-perfusion mismatch after methacholine challenge in patients with mild bronchial asthma. Am Rev Respir Dis 1991;144:88eC94.
Vidal Melo MF, Harris RS, Layfield D, Musch G, Venegas JG. Changes in regional ventilation after autologous blood clot pulmonary embolism. Anesthesiology 2002;97:671eC681.
Wagner PD, Laravuso RB, Uhl RR, West JB. Continuous distributions of ventilation-perfusion ratios in normal subjects breathing air and 100 per cent O2. J Clin Invest 1974;54:54eC68.
Wagner PD, Laravuso RB, Goldzimmer E, Naumann PF, West JB. Distribution of ventilation-perfusion ratios in dogs with normal and abnormal lungs. J Appl Physiol 1975;38:1099eC1109.
Rodriguez-Roisin R, Bencowitz HZ, Ziegler MG, Wagner PD. Gas exchange responses to bronchodilators following methacholine challenge in dogs. Am Rev Respir Dis 1984;130:617eC626.
Kuriyama T, Wagner WW Jr. Collateral ventilation may protect against high-altitude pulmonary hypertension. J Appl Physiol 1981;51:1251eC1256.
West JB. Gas exchange. In: West JB, editor. Regional differences in the lung, 1st ed. New York: Academic Press; 1977. pp. 201eC243.
Itoh H, Nakatsu M, Yoxtheimer LM, Uematsu H, Ohno Y, Hatabu H. Structural basis for pulmonary functional imaging. Eur J Radiol 2001;37:143eC154.
Heitzman ER, Markarian B, Berger I, Dailey E. The secondary pulmonary lobule: a practical concept for interpretation of chest radiographs: II. Application of the anatomic concept to an understanding of roentgen pattern in disease states. Radiology 1969;93:513eC519.
Heitzman ER, Markarian B, Berger I, Dailey E. The secondary pulmonary lobule: a practical concept for interpretation of chest radiographs: I. Roentgen anatomy of the normal secondary pulmonary lobule. Radiology 1969;93:507eC512.
Terry PB, Menkes HA, Traystman RJ. Effects of maturation and aging on collateral ventilation in sheep. J Appl Physiol 1987;62:1028eC1032.
Putensen C, Rasanen J, Lopez FA. Improvement in VA/Q distributions during inhalation of nitric oxide in pigs with methacholine-induced bronchoconstriction. Am J Respir Crit Care Med 1995;151:116eC122.
Brown RH, Zerhouni E. New techniques and developments in physiologic imaging of airways. Radiol Clin North Am 1998;36:211eC230.
Herold CJ, Brown RH, Mitzner W, Links JM, Hirshman CA, Zerhouni EA. Assessment of pulmonary airway reactivity with high-resolution CT. Radiology 1991;181:369eC374.
Lynch DA, Newell JD, Tschomper BA, Cink TM, Newman LS, Bethel R. Uncomplicated asthma in adults: comparison of CT appearance of the lungs in asthmatic and healthy subjects. Radiology 1993;188:829eC833.
Musch G, Layfield JD, Harris RS, Melo MF, Winkler T, Callahan RJ, Fischman AJ, Venegas JG. Topographical distribution of pulmonary perfusion and ventilation, assessed by PET in supine and prone humans. J Appl Physiol 2002;93:1841eC1851.
Harris RS, Musch G, Layfield D, Vidal Melo MF, Venegas JG. Analysis of hypoventilating (HV) regions in asthmatic subjects using positron emission tomography (PET) . Am J Respir Crit Care Med 2002;165:A784.
Harris RS, Musch G, Winkler T, Vidal Melo MF, Layfield J, Tgavalekos NT, Lutchen KR, Venegas JG. Imaging of hypoxic pulmonary vasoconstriction (HPV) in asthmatic subjects during bronchoconstriction . Am J Respir Crit Care Med 2003;167:A698.
Harris RS, Isasi AG, Musch G, Winkler T, Vidal Melo MF, Tgavalekos NT, Bellani G, Venegas JG. Imaging aerosol deposition in normal and asthmatic subjects with positron emission tomography . Am J Respir Crit Care Med 2004;169:A545.(Marcos F. Vidal Melo, R. )
Harvard Medical School, Boston, Massachusetts
ABSTRACT
The distribution of ventilationeCperfusion (A/) ratios during bronchoconstriction measured with the multiple inert gases elimination technique is frequently bimodal. However, the topographic basis and the cause of that bimodality remain unknown. In this article, regional A/ is quantified by three-dimensional positron emission tomography (PET) imaging of methacholine-induced bronchoconstriction in sheep. Regional A/ ratios were calculated from the imaged kinetics of intravenously injected 13NN-saline bolus, assembled into global A/ distributions, and used to estimate gas exchange. During bronchoconstriction, large regions with impaired tracer washout were observed adjacent to regions of normal ventilation. PET-derived A/ distributions during bronchoconstriction were consistently bimodal, with areas of low A/ receiving a large fraction of . The standard deviation of the A/ distribution was 38% lower if small-scale (subresolution) heterogeneity (< 2.2 cm3) was ignored. Arterial blood gases predicted from PET data correlated well with measured values for PaO2 (r2 = 0.91, p < 0.01) and PaCO2 (r2 = 0.90, p < 0.01). We conclude that the bimodality of A/ distributions in bronchoconstriction reflects the involvement of large contiguous regions of hypoventilation with substantial subresolution intraregional A/ heterogeneity. Assessment of the subresolution A/ heterogeneity is therefore essential to accurately quantify global gas exchange impairment during bronchoconstriction.
Key Words: lung imaging positron emission tomography pulmonary gas exchange theoretical models
Heterogeneity in the distribution of alveolar ventilation (A) to perfusion () is the main determinant of gas exchange impairment during bronchoconstriction in humans and animals. Using the multiple inert gases elimination technique (MIGET), Wagner and coworkers (1) observed bimodal blood-flow distributions of A/ ratios in most patients with asymptomatic asthma.
Bimodal or broadened unimodal A/ distributions were also observed in patients with chronic (2) and acute severe asthma (3). In dogs, using MIGET, Rubinfeld and colleagues (4) showed broadened unimodal A/ distributions in mild bronchospasm and bimodal distributions in severe bronchospasm. Collateral ventilation of units distal to closed airways was invoked as a mechanism explaining the bimodality of the distributions. A/ distributions obtained with MIGET provide accurate description of gas exchange. However, MIGET-derived complex A/ distributions may be nonunique (5, 6), and the bimodality of the A/ distribution during bronchoconstriction has not been confirmed by other methods. Furthermore, because MIGET does not contain topographic information on the measured A/, the structural basis for the bimodal distributions has remained elusive.
Airway resistance in asthma is significantly affected by small airways constriction (3), although constriction may occur at different levels of the airway tree (7, 8). It follows that A/ heterogeneity in the lungs during bronchoconstriction may be manifested at several length scales. Several imaging methods have been used in the attempt to directly and topographically assess ventilation or A/ distributions during airway obstruction (7, 9, 10). Studies with inhaled radionuclides (7, 11, 12), magnetic resonance (13), single-photon emission computed tomography (14), and computed tomography (15) have all shown significant spatial ventilation or A/ heterogeneity in patients with asthma. However, those imaging techniques have not been able to quantify A/ ratios to the point of predicting the level of global gas exchange impairment, and, as a result, there are limited data on the functional impact of the topographic A/ changes during airway obstruction.
We recently described a method to compute A/ heterogeneity based on positron emission tomography (PET) data (16), which provided satisfactory agreement between experimental arterial blood gases and estimates from PET-derived A/ distributions. The method follows the regional kinetics of an intravenous bolus of 13NN-labeled saline and not only quantifies the imaged topographic heterogeneity in A/ but also the heterogeneity occurring at length scales below the spatial resolution of the imaging instrument (2.2 cm3). Lung structures underlying 2.2 cm3 correspond to secondary pulmonary lobules and thus include multiple terminal acini that, on theoretic grounds, have been linked to the bimodal shape of the A/ distribution during bronchoconstriction (17, 18). Subresolution (intraregional) A/ heterogeneity is taken into account by analyzing the tracer kinetics at the level of the smallest image volume component (voxel), and exploring its compatibility to either a heterogeneous (two-compartment) or a homogeneous (one-compartment) washout pattern.
We reasoned that A/ heterogeneity occurring at length scales smaller than 2.2 cm3 could have a significant contribution to the overall A/ heterogeneity and to the bimodality in A/ distribution observed during bronchoconstriction. To test this hypothesis, we assessed the contribution of A/ heterogeneity originating from length scales below the imaging resolution to global A/ heterogeneity in a sheep model of methacholine-induced bronchoconstriction. Some of the results of these studies have been previously reported in the form of abstracts (19, 20).
METHODS
Animal Preparation
The study protocol was approved by the Massachusetts General Hospital Committee on Animal Care. Six normal sheep weighing 17 kg (range 15eC21 kg) were anesthetized; intubated; mechanically ventilated; prepared with a femoral arterial catheter, a pulmonary artery catheter, and an independent central venous catheter for tracer injection; and placed in the prone position. The ventilator (Harvard Apparatus, Millis, MA) was set at an inspired oxygen fraction (FIO2) = 0.49 ± 0.02, positive end-expiratory pressure = 5 cm H2O, VT = 17 ± 2 cc/kg, and inspiratory time of 30% of the breathing period. The respiratory rate (12 ± 2 bpm) was set to normocapnic arterial blood gases at the beginning of the experiment and maintained at that value for the rest of the experiment. Alveolar ventilation was computed as A = (VT eC VD) x respiratory rate, where VD = dead space was estimated from the animal's weight (21).
PET Imaging Protocol and Processing
Emission and transmission scans were performed as previously described (16, 22). The animal was positioned in the PET camera with the most caudal slice adjacent to the diaphragm dome. The PET camera (Scanditronix PC4096; General Electric, Milwaukee, WI) collected 15 transverse cross-sectional 6.5-mm slices over a 9.7-cm-long axial field.
The tracer 13NN gas ( 10-minute half-life) was generated by a cyclotron and dissolved in degassed saline. The imaging protocol for the emission scans started with a tracer-free lung. The ventilator was turned off at end exhalation and a bolus of 13NN saline solution was injected into a central vein. Simultaneously, collection of six consecutive images of 5-second duration was started while the animal was kept in apnea for 30 seconds. Subsequently, mechanical ventilation was resumed and four images of 30 seconds and two images of 60 seconds were collected. Resulting images consisted of an interpolated matrix of 128 x 128 x 15 voxels of 6 x 6 x 6.5 mm. Images were low-pass-filtered to 13 x 13 mm in the image plane, and a two-point moving-average filter was applied in the z plane to a final volumetric resolution of approximately 2.2 cm3.
Once the animal was positioned in the PET camera and in steady state, a set of physiologic measurements, transmission, and emission scans was collected (control). After these, methacholine solution (25 mg/ml) was delivered through the inspiratory line with an ultrasonic nebulizer (Ultra-Neb 99; DeVilbiss Co., Somerset, PA). The methacholine dose was titrated to double control peak airway pressure (Ppeak), and nebulization was maintained during subsequent imaging. A new set of physiologic, transmission, and emission scans was obtained in this new condition (bronchoconstriction) approximately 35 minutes after control measurements. Physiologic measurements were performed within 10 minutes of PET measurements once steady-state conditions (i.e., stable arterial and pulmonary blood pressures and Ppeak) were obtained.
Regional Perfusion and A/
Regional perfusion (r) and shunt were computed from the apneic phase of the tracer kinetics (23, 24). r was estimated from the mean activity during the apneic plateau phase.
To assess the relevance of subresolution (intraregional) A/ heterogeneity to global A/ heterogeneity, we used two methods to compute regional A/ as follows: (1) homogeneous regional A/, where the voxel was considered a uniform compartment and regional A/ was computed from the integral of tracer activity along the washout, including an exponential correction in the cases of incomplete washout at the end of the imaging period (16); (2) heterogeneous regional A/, in which regional A/ was calculated with a recently described technique (16) where the regional tracer washout plot was used to classify (Figure 1) and compute intraregional specific ventilation (sr) and r/sr. A voxel was classified as homogeneous when the semilog washout plot approached linearity and was treated as a single compartment whether ventilating or complete air trapping. Conversely, a voxel was classified as heterogeneous when the semilog washout plot was curvilinear and was analyzed as made of two compartments, either both ventilating or one ventilating and one air trapping. Distributions of A/ in absolute units were estimated using thermodilution cardiac output, total shunt, VT, estimated anatomic dead space, and respiratory rate. Total shunt was estimated by adding the shunt derived from global tracer kinetics to the component corresponding to regional subcompartments with complete gas trapping (A/ = 0). We assumed that sr was proportional to regional ventilation r and thus r/sr was proportional to regional A/.
Gas Exchange Computations
Regional A/ compartments were grouped into 100 bins of equal log (A/) width (0.05) ranging from eC3 to 2. The square root of the second moment of the perfusion-weighted A/ distribution about its mean on a log scale (SD) was used as an indicator of A/ heterogeneity (2). Global alveolar and blood O2 and CO2 partial pressures were estimated by computing partial pressures and contents from the obtained A/ distributions (16, 25).
Statistical Analysis
Data were expressed as mean ± SD. Comparisons between values before (control) and during bronchoconstriction were made with a two-tailed Student's t test for paired samples. Linear correlation and biases (26) were used to summarize the relationship between the measured and estimated blood gases. Statistical significance was taken at the p < 0.05 level.
RESULTS
Significant bronchoconstriction was achieved in this animal model as expressed by the marked changes in respiratory mechanics and gas exchange during inhaled methacholine (Table 1). Ppeak during methacholine inhalation was 2.3 times baseline Ppeak, as required by protocol design. In parallel with this rise in Ppeak, arterial and mixed-venous PO2 were significantly reduced, and PaCO2 was significantly increased whereas right-to-left shunt was minimally changed. Changes in hemodynamics during inhaled methacholine were a decrease in mean arterial pressure and a small increase in mean pulmonary artery pressure, with no significant changes in pulmonary artery occlusion pressure or cardiac output (Table 1). PET imaging scans during bronchoconstriction were characterized by significant amounts of residual tracer remaining in the lung fields at the end of the washout period, in contrast to the nearly complete elimination of the tracer during control conditions (Figure 2). Tracer distribution during bronchoconstriction showed large regions of tracer retention, segmental to subsegmental size, adjacent to regions of nearly complete tracer washout (Figure 2). Compared with such changes in regional ventilation, topographic changes in perfusion were much smaller in these prone sheep (Figure 2). No evidence of development of pulmonary edema was observed during the study, with average lung gas fraction estimates derived from the transmission scans amounting to 0.54 ± 0.07 before and 0.54 ± 0.09 during bronchoconstriction. The gravity dependence of lung gas fraction was variable among animals and no consistent change was observed with methacholine.
The four different patterns of voxel washout kinetics described in METHODS (Figure 1) were observed during the experiments (Table 2). Before bronchoconstriction, the predominant pattern was that of a single "fast" compartment leaving less than 10% of residual tracer at the end of the washout (Table 2). During bronchoconstriction, a significantly larger number of areas presented abnormal washout patterns with complete or partial gas trapping, two-compartment washout, and slow single-compartment washout (Table 2). The component of gas trapping was negligible in control conditions and increased to 7.0 ± 4.1% during bronchoconstriction.
In all sheep studied, PET-based A/ distributions were unimodal and narrow during control conditions (Figure 3, left panel), with small values for SD (Table 3). During bronchoconstriction, A/ distributions became significantly wider than those during control. The calculation of A/ distributions considering subresolution heterogeneity, derived with a two-compartment model of voxel tracer washout, revealed broad and bimodal A/ distributions in all animals (Figure 3, middle panel), with significant increase of SD in bronchoconstriction but not in control conditions (Table 3).
Estimates of arterial blood gases based on these computed A/ distributions were correlated with experimental measurements for both O2 and CO2. PET-based estimates of PaO2 described measured PaO2 with the regression equation: measured PaO2 = 0.965 · estimated PaO2 eC 8.8 mm Hg, r2 = 0.91 (p < 0.01; Figure 4). The mean difference between measured and estimated PO2 was eC14.3 ± 32.9 mm Hg. Similarly, PET-based estimates of PaCO2 described measured PaCO2 with the following regression equation: measured PaCO2 = 0.923 · estimated PaCO2 + 4.2 mm Hg, r2 = 0.90 (p < 0.01). Estimated PaCO2 approximated PaCO2 measurements, with a mean difference of 1.4 ± 1.5 mm Hg (Figure 5).
The analysis assuming homogeneous or heterogeneous intraregional A/ data to fit the tracer washout kinetics yielded similar estimates of SD before bronchoconstriction (Table 3). In contrast, after bronchoconstriction, the analysis assuming homogeneous intraregional A/ yielded A/ distributions with significantly lower SD values than those obtained assuming a two-compartment intraregional washout model (Table 3). On average, lack of consideration of subresolution A/ heterogeneity resulted in an underestimation of SDQ of 38% when compared with the intraregional two-compartment analysis. In terms of A/ distributions, the use of a single-compartment model to analyze voxel washout yielded unimodal A/ distributions in half of the animals and bimodal A/ distributions in the other half (Figure 3, right panel). These bimodal distributions were less marked than those obtained when intraregional A/ heterogeneity was taken into account (Figure 3, middle panel). Gas exchange estimates during bronchoconstriction obtained from A/ distributions based on one-compartment analysis of regional washout overestimated measured PaO2 by 27.7 ± 41.7 mm Hg and underestimated PaCO2 by 2.0 ± 1.9 mm Hg.
DISCUSSION
The main findings of this study are as follows: (1) severe bronchoconstriction systematically changes the PET-derived A/ distributions from narrow and unimodal to wide and bimodal and (2) the bimodal shape is caused, in large part, by the contribution of A/ intraregional heterogeneity at length scales under 2.2 cm3 (the spatial resolution of the imaging method) and by the presence of units with substantially reduced ventilation that are topographically clustered in large contiguous regions and not diffusely distributed across the lungs.
Methodologic Limitations
The assumption of a two-compartment model to quantify intraregional heterogeneity in A/ (16) is a simplification because a larger number of A/ compartments could potentially be present within 2.2 cm3 of the lung. Identification of a larger number of compartments could improve the description of intraregional heterogeneity and increase its contribution to total A/ heterogeneity. In a washout curve analysis, however, the number of identifiable compartments is usually limited to a few and depends on the separation of their respective time constants, on the number of data points collected, the length of the measured washout period, and the signal-to-noise ratio of the imaging system. Given that the signal-to-noise ratio on a PET image depends on both activity and the time of acquisition, the number of images collected during the 4 minutes of washout was set to six, limiting the number of identifiable compartments to no more than two. Longer washout acquisition times during bronchoconstriction and the use of new systems with higher signal-to-noise ratio may allow for the identification of a larger number of compartments. Yet, when compared with the assumption of a uniform region, the use of two compartments provided a significant difference in the shape of the derived A/ distribution.
The presented method is based on the intravenous injection of 13NN. Consequently, regions of very low perfusion will receive lower tracer amounts and may have their tracer kinetics importantly modified by noise and tracer rebreathing during washout. This would compromise identification of regions with high A/ and low . Because regions of high A/ are not usual during bronchoconstriction (1eC4, 27), it is unlikely that their presence affected our results. In situations when such regions would be important, the use of emission scans of inhaled tracer (22, 28) allows for their identification.
A/ Distributions during Bronchoconstriction
In these normal sheep before bronchoconstriction, ventilation- and perfusion-weighted A/ distributions derived from PET scanning followed narrow and unimodal log-normal distribution functions. These findings agree with previous results using MIGET, which showed narrow and unimodal log-normal distributions of A/ ratios in humans and animals (29, 30). After severe bronchoconstriction with methacholine, all animals in the current study showed marked A/ heterogeneity and bimodal distributions. In these distributions, one mode was centered around values of A/ ratios between 1 to 3 and the second mode at values between 0.02 and 0.6. These results are consistent with the bimodal A/ distributions derived with MIGET in severely bronchoconstricted dogs with different agents (4, 31) and in humans with asthma (1, 2, 27).
MIGET-derived A/ distributions are based on global lung data, and thus should describe A/ heterogeneity in the whole lung. However, the nonunique nature of the MIGET solution requires the use of smoothing and optimization algorithms to yield a A/ distribution that best fits measured retentions and eliminations of inert gases. In contrast, the estimates of A/, perfusion, and ventilation presented here were directly derived from regional tracer kinetics, without a priori assumption on the shape of the A/ distribution. The fact that the shape of the A/ distributions obtained from PET topographic data is consistent with those derived with MIGET serves as cross-validation of both methods to assess A/ distributions in conditions of severe bronchoconstriction. In addition to yielding global lung A/ distributions, our PET technique provides information on the anatomic origin of the A/ heterogeneity not available from MIGET measurements.
Previous studies showed minimal changes in right-to-left shunt with bronchoconstriction (1, 4, 31). We also observed minimal right-to-left shunt blood flow to nonaerated regions during methacholine, as shown by a stable tracer kinetics plot during the breath-holding period. However, we were able to detect substantial presence of aerated but nonventilating regions (i.e., gas trapping) during severe bronchoconstriction. Because these units have A 0, and thus A/ 0, they contribute to functional shunt. A similar phenomenon was observed by Rubinfeld and coworkers (4) where complete obstruction of large lung regions was seen in one dog during autopsy. The larger component of gas trapping observed in our sheep compared with that reported for dogs may be from the lower amount of collateral ventilation in sheep compared with dogs (32) or to the more severe bronchospasm induced in this study as suggested by a mean PaO2/FIO2 ratio of 173 mm Hg measured in our study as compared with 255 mm Hg reported for the severe constriction in Rubinfeld and colleagues' (4) study.
Topographic Distribution of A/ Heterogeneity during Bronchoconstriction
Results by Wagner and coworkers (29) and West (33) indicated that both ventilation and blood flow curves derived from radioactive gas measurements were considerably narrower than A/ distributions obtained with MIGET in young, normal subjects. In contrast, our PET measurements, conducted with significantly higher spatial resolution than those former scintillation counter methods, showed satisfactory agreement between measured and arterial blood gas estimates during control conditions. Moreover, the use of a single- or two-compartment fit to control tracer kinetics resulted in minimal differences in measured SD for control conditions. This finding indicates that the small A/ heterogeneity of the normal lung is predominantly caused by interregional differences of structures with volumes greater than 2.2 cm3, the volumetric resolution of our images.
During bronchoconstriction, visual inspection of the emission scans at the end of washout showed that tracer retention was topographically distributed in large and contiguous regions of the lung (Figure 2), which were adjacent to regions with nearly complete washout. The presence of these large areas of tracer retention contrasts with the concept of bronchoconstriction as a process of small airway narrowing diffusely distributed throughout the lung. The mechanism leading to this spatial distribution of low A/ units is unclear. It can be related either to obstruction of large, approximately segmental level bronchi or to clustered increase in constriction of smaller neighboring regions.
When uniform intraregional A/ (single-compartment model) was assumed to analyze the washout data during bronchoconstriction, bimodal A/ distributions were observed in three of the six studied sheep, suggesting that interregional heterogeneity at length scales above 2.2 cm3 was sufficient to account for part of the bimodal distribution. However, the presence of significant intraregional A/ heterogeneity during bronchoconstriction was evidenced by the significant increase in total heterogeneity from the voxel-by-voxel analysis of 13NN kinetics and by the large number of voxels showing multiexponential washout and/or partial or complete intraregional gas trapping. Accordingly, the increase in A/ heterogeneity obtained with a double-compartment model analysis compared with that obtained with a single-compartment model indicates that significant intraregional A/ heterogeneity was present, at dimensions smaller than the spatial resolution of the imaging method in this model of bronchoconstriction. Furthermore, the A/ distributions derived without considering intraregional heterogeneity underestimated by 38% global A/ heterogeneity and, as a consequence, systematically increased the error of estimated arterial blood gases. These results also illustrate how, despite the impossibility of visualizing with PET the A/ heterogeneity at length scales under its effective spatial resolution of 2.2 cm3, a substantial amount of functionally relevant subresolution A/ heterogeneity could be uncovered and quantified from the temporal information contained in the tracer washout.
Intraregional heterogeneity was not only responsible for a significant fraction of the A/ mismatch but also accounted in great part for the bimodal characteristics of the A/ distributions. These findings, taken together, support the concept that changes in peripheral airways and, potentially, lung tissue are largely responsible for overall gas exchange impairment during bronchoconstriction. It is also clear that, during bronchoconstriction, consideration of subresolution heterogeneity is essential for adequate assessment of global A/ heterogeneity from imaging data.
Structural changes that can cause A/ heterogeneity during bronchoconstriction include heterogeneity in ventilation caused by airway narrowing, edema, hypersecretion, bronchial inflammation, and constriction of parenchymal smooth muscle. Two mechanisms have been invoked to explain the bimodal shape of the A/ distributions during bronchoconstriction. Initial investigations with MIGET attributed the presence of a low A/ mode to collateral ventilation (1, 4) of regions distal to occluded airways. This mechanism could explain how a condition believed to be diffusely distributed throughout the lung would consistently give rise to a bimodal pattern without significantly increasing right-to-left shunt. More recently, Anafi and Wilson (17), modeling the mechanical behavior of a constricted terminal airway during tidal breathing, discovered that it could attain two stable states: one effectively open and one nearly closed. Predictions from that model fit well in dynamic lung elastance and gas-mixing indices in subjects without asthma exposed to methacholine (18) and were consistent with the bimodal A/ distribution during bronchoconstriction.
Our results demonstrated that one source of the bimodal A/ distribution takes place within lung structures with volumes smaller than 2.2 cm3. Although it is impossible with the spatial resolution of PET to perform an analysis following strict anatomic limits, it is interesting to note that the 2.2-cm3 volume corresponds approximately to that of a secondary pulmonary lobule. This structure is an irregular polyhedral of 1 to 2.5 cm on each side, supplied by three to five respiratory bronchioles, frequently cited in the pulmonary imaging literature (34eC36). Although collateral ventilation could potentially take place at this level, complete and partial intraregional air trapping was observed in several lung areas during bronchoconstriction. This finding suggests that collateral ventilation was not effective in those regions, and it is unclear why it would play a role in other lung regions of the same animal. Also, it is well established that collateral ventilation is less important in the sheep than in the dog, particularly in the studied young sheep, which are known to have the highest resistance to collateral flow (37). Thus, it does not appear that collateral ventilation could play a main role in explaining the bimodal A/ distributions in our animal model. Bistable mechanical characteristics of the terminal airways were predicted for acinar units (17, 18), structures with volumes between 1.2 and 1.8 cm3 at total lung capacity (17, 18). This is compatible with the presence of at least two terminal units within our spatial resolution of 2.2 cm3 during tidal breathing. Thus, our data would suggest that intraregional A/ heterogeneity caused by bronchoconstriction was primarily caused by heterogeneous ventilation at or below the acinar level. The bistability of the terminal bronchioles described by Anafi and coworkers (17, 18) is in full agreement with the distributions measured in this study, where one mode was around normal A/ and another at a A/ value about 10 times lower.
The regional distribution of both inter- and intraregional heterogeneities was nonuniform during bronchoconstriction. The topographic distribution of the well-ventilated and poorly ventilated regions exhibited partial vertical dependence in these prone sheep. Nondependent regions tended to show less poorly ventilated areas than middle and dependent regions. This finding suggests a trend for development of more dependent airway narrowing and obstruction during bronchoconstriction. However, other factors in addition to the vertical gradient must be involved because, although three sheep presented a clear vertical progression of poorly ventilated regions, in another three sheep there was poor ventilation in middle regions with preservation of dependent areas. Nonuniform distribution of intraregional heterogeneity was characterized by the fact that, as compared with the results for the whole lung, regions with large amounts of residual tracer at the end of the washout (as shown in Figure 2 during bronchoconstriction) showed a higher fraction of voxels with complete trapping (25 ± 5%) and slow single-compartment (21 ± 8%) kinetics, a smaller fraction of voxels with partial trapping (4 ± 3%) and two-compartment kinetics (18 ± 5%), and a similar fraction of voxels with fast single-compartment kinetics (32%). Thus, large lung areas with significant tracer retention tended to present more regions that either emptied homogeneously slowly or not at all, whereas areas of overall smaller tracer retention tended to show more nonuniform kinetics. However, there is still significant heterogeneity in all lung regions.
Overall, our results demonstrate that A/ distributions during bronchoconstriction have two topographic components: one, caused by interregional A/ heterogeneity between large areas, of at least segmental size, where terminal units with highly reduced ventilation cluster; the other, because of intraregional A/ heterogeneity within lung structures under 2.2 cm3, probably a result of bistable terminal bronchial constriction.
Gas Exchange
There was a high correlation between PET-estimated and measured arterial blood gases. A larger dispersion around the regression line of measured versus estimated PaO2 values was observed for the higher range of PaO2. This finding may be, in part, because at high PaO2, hemoglobin is nearly 100% saturated and small errors in oxygen saturation can result in large errors in PaO2. Estimated values of PaO2 tended to overestimate measured PaO2. In MIGET studies, this type of result has been interpreted as consistent with diffusion limitation. Our measurements are based on the kinetics of the inert gas 13NN, which is significantly less affected by diffusion limitation than O2. In addition, no evidence of diffusion limitation was observed in normal lungs subjected to methacholine challenge (27, 31, 38).
One probable cause for the underestimation of PaO2 is the absence of steady-state gas exchange during the imaging of bronchoconstriction. The two animals with the lowest PaO2 had a significant drop in their arterial-mixed venous O2 content difference with unchanged cardiac output during bronchoconstriction, clearly reflecting extreme cardiopulmonary burden with noneCsteady-state conditions. In this case, studies in less extreme situations would be expected to yield better agreement between predicted and measured PaO2, as presented previously (16). Another potential cause of the overestimation in PaO2 is that A/ heterogeneity was underestimated by PET, which may have been from the following reasons: (1) Limited spatial resolution of PET; (2) Use of only two compartments to describe intraregional heterogeneity; (3) Limited identification of regions of very small, specific ventilation because of insufficient imaging time during washout; (4) Assumption that compartment ventilation was proportional to its specific ventilation (i.e., compartment effective ventilation scaling with regional volume). If this distribution were heterogeneous, an additional component of mismatch would exist that was not considered in the computations. In this case, the difference between measured and estimated PaO2 would be an indication of intraregional ventilation to volume mismatch; (5) Nonimaging of apical or basal diaphragmatic regions with lower A/ than imaged regions. We maximized the volume of imaged lung by adjusting the animal's position in the camera according to initial transmission scans to have approximately 80% of imaged lung within the axial field of 9.97 cm used in this study. In addition, the animals were in the prone position, which should reduce the heterogeneity in supradiaphragmatic regions. However, the presence of low A/ in nonimaged regions cannot be ruled out.
Heterogeneity in regional A/ affects the exchange of O2 and CO2 differently. Regions of high A/ affect CO2 exchange more, whereas regions of low A/ have a greater impact on O2 exchange. The experimental model allowed us to test the ability of our experimental and analysis methods to recover a broad range of A/ ratios. The severe bronchoconstriction induced in this sheep model produced large changes in A/ distribution with values of A/ ratios as low as 0.001. In addition, we used an FIO2 of 49%, which implies that proper quantification of A/ ratios less than 0.17 was needed to predict the reduction of PaO2 caused by bronchoconstriction. Thus, the correlation between measured and estimated PaO2 indicates satisfactory description of very low A/ ratios. Furthermore, the linear unbiased relationship between estimated and measured PaCO2 suggests that PET-based A/ distributions satisfactorily estimate regions with normal and high A/ ratios. Taken together, our results suggest that the very extreme changes in A/ distributions caused by bronchoconstriction are suitably described by PET-derived data.
Imaging of A/ in Bronchoconstriction and Asthma
There is no gold-standard technique that can be used to validate image-derived measures of local A/ during bronchoconstriction. Nuclear medicine scintigraphic ventilation and perfusion scans have demonstrated ventilation and perfusion unevenness in patients with mild asthma with different triggers (7, 11, 12) but at a spatial resolution too low to assess its functional significance. High-resolution computed tomography provides distinct anatomic information on caliber changes of airways of diameter greater than 2 mm (39, 40) and can assess the presence and location of gas trapping after an exhalation to residual volume (41), but no direct connection to regional gas exchange during breathing has been established for these techniques. Single-photon emission computed tomography with Technegas was used to assess airway closure in patients with asthma (14). The authors, however, recognized that the technique was unlikely to detect changes in lung regions of less than 51 ml of volume. More recently, magnetic resonance imaging with hyperpolarized helium was used for high-resolution visualization of ventilation defects in symptomatic and asymptomatic asthma (13). None of these imaging techniques were consistently related to independent measures of gas exchange, such as arterial blood gases. PET has lower spatial resolution than magnetic resonance imaging and computed tomography, is expensive, and uses radioactivity. Conversely, it has high sensitivity to minute tracer concentrations and enough temporal resolution to uncover subresolution heterogeneity from analysis of regional tracer kinetics. Our current and previous (16) results show that PET estimates of regional A/ are quantitatively related to gas exchange impairment. We recently implemented the technique to study humans (42) and preliminary results in patients with asthma (43eC45) confirm and expand our observations in animals. Thus, PET may be an attractive technique to study human asthma when more accurate quantification of gas exchange impairment combined with topographic information is desirable for clinical or research purposes. Given its quantitative value, PET could also serve as a validating tool in the development of novel pulmonary functional imaging methods.
In summary, we showed that kinetics analysis of intravenously injected 13NN measured with PET reveals significant A/ heterogeneity with a consistent bimodal distribution of A/ ratios during severe bronchoconstriction. This bimodality was caused by a combination of inter- and intraregional A/ heterogeneity. Interregional A/ heterogeneity was manifested as clusters of bronchoconstricted units in large contiguous areas of reduced ventilation. Intraregional A/ heterogeneity was evidenced as heterogeneity at subresolution length scales. This subresolution A/ heterogeneity represented an important component of total A/ mismatch and proved, therefore, to be essential to quantify gas exchange alterations during bronchoconstriction.
Acknowledgments
The authors thank Steven B. Weise, Sr., Research Technologist, PET Imaging Laboratory, Massachusetts General Hospital, for his expert support in the acquisition of images and Tobias Schroeder, Department of Anesthesia, Massachusetts General Hospital, for assistance with data analysis.
Presented, in part, at the International Conference of the American Thoracic Society, Atlanta, Georgia, May 22, 2002, and at the Annual Meeting of the American Society of Anesthesiologists, Orlando, Florida, October 15, 2002.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
REFERENCES
Wagner PD, Dantzker DR, Iacovoni VE, Tomlin WC, West JB. Ventilation-perfusion inequality in asymptomatic asthma. Am Rev Respir Dis 1978;118:511eC524.
Wagner PD, Hedenstierna G, Bylin G. Ventilation-perfusion inequality in chronic asthma. Am Rev Respir Dis 1987;136:605eC612.
Ferrer A, Roca J, Wagner PD, Lopez FA, Rodriguez-Roisin R. Airway obstruction and ventilation-perfusion relationships in acute severe asthma. Am Rev Respir Dis 1993;147:579eC584.
Rubinfeld AR, Wagner PD, West JB. Gas exchange during acute experimental canine asthma. Am Rev Respir Dis 1978;118:525eC536.
Kapitan KS, Wagner PD. Linear programming analysis of VA/Q distributions: limits on central moments. J Appl Physiol 1986;60:1772eC1781.
Evans JW, Wagner PD. Limits on VA/Q distributions from analysis of experimental inert gas elimination. J Appl Physiol 1977;42:889eC898.
Munkner T, Bundgaard A. Regional V/Q changes in asthmatics after antigen inhalation. Eur J Respir Dis Suppl 1986;143:44eC47.
Sekizawa K, Sasaki H, Shimizu Y, Takishima T. DoseeCresponse effects of methacholine in normal and in asthmatic subjects: relationship between the site of airway response and overall airway hyperresponsiveness. Am Rev Respir Dis 1986;133:593eC599.
Munkner T, Bundgaard A. Regional V/Q changes in asthmatics after histamine inhalation. Eur J Respir Dis Suppl 1986;143:22eC27.
Kuethe DO, Caprihan A, Gach HM, Lowe IJ, Fukushima E. Imaging obstructed ventilation with NMR using inert fluorinated gases. J Appl Physiol 2000;88:2279eC2286.
Agnew JE, Bateman JR, Pavia D, Clarke SW. Radionuclide demonstration of ventilatory abnormalities in mild asthma. Clin Sci (Lond) 1984;66:525eC531.
Vernon P, Burton GH, Seed WA. Lung scan abnormalities in asthma and their correlation with lung function. Eur J Nucl Med 1986;12:16eC20.
Altes TA, Powers PL, Knight-Scott J, Rakes G, Platts-Mills TA, de Lange EE, Alford BA, Mugler JP III, Brookeman JR. Hyperpolarized 3He MR lung ventilation imaging in asthmatics: preliminary findings. J Magn Reson Imaging 2001;13:378eC384.
King GG, Eberl S, Salome CM, Young IH, Woolcock AJ. Differences in airway closure between normal and asthmatic subjects measured with single-photon emission computed tomography and Technegas. Am J Respir Crit Care Med 1998;158:1900eC1906.
Jensen SP, Lynch DA, Brown KK, Wenzel SE, Newell JD. High-resolution CT features of severe asthma and bronchiolitis obliterans. Clin Radiol 2002;57:1078eC1085.
Vidal Melo MF, Layfield D, Harris RS, O'Neill K, Musch G, Richter T, Winkler T, Fischman AJ, Venegas JG. Quantification of regional ventilation-perfusion ratios with PET. J Nucl Med 2003;44:1982eC1991.
Anafi RC, Wilson TA. Airway stability and heterogeneity in the constricted lung. J Appl Physiol 2001;91:1185eC1192.
Anafi RC, Beck KC, Wilson TA. Impedance, gas mixing, and bimodal ventilation in constricted lungs. J Appl Physiol 2003;94:1003eC1011.
Vidal Melo MF, Harris RS, Call D, Layfield JDH, Venegas JG. V A/ Q distributions during bronchoconstriction derived from positron emission tomography (PET) data . Am J Respir Crit Care Med 2002;165:A706.
Vidal Melo MF, Harris RS, Layfield J, Venegas JG. Contribution of inter and intraregional V/Q heterogeneity to gas exchange impairment during bronchoconstriction assessed by PET . Anesthesiology 2002;96:A1311.
Stahl WR. Scaling of respiratory variables in mammals. J Appl Physiol 1967;22:453eC460.
Treppo S, Mijailovich SM, Venegas JG. Contributions of pulmonary perfusion and ventilation to heterogeneity in V(A)/Q measured by PET. J Appl Physiol 1997;82:1163eC1176.
Galletti GG, Venegas JG. Tracer kinetic model of regional pulmonary function using positron emission tomography. J Appl Physiol 2002;93:1104eC1114.
Mijailovich SM, Treppo S, Venegas JG. Effects of lung motion and tracer kinetics corrections on PET imaging of pulmonary function. J Appl Physiol 1997;82:1154eC1162.
Vidal Melo MF, Loeppky J, Caprihan A, Luft UC. Alveolar ventilation to perfusion heterogeneity and diffusion impairment in a mathematical model of gas exchange. Comput Biomed Res 1993;26:103eC120.
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;327:307eC310.
Rodriguez-Roisin R, Ferrer A, Navajas D, Agusti AG, Wagner PD, Roca J. Ventilation-perfusion mismatch after methacholine challenge in patients with mild bronchial asthma. Am Rev Respir Dis 1991;144:88eC94.
Vidal Melo MF, Harris RS, Layfield D, Musch G, Venegas JG. Changes in regional ventilation after autologous blood clot pulmonary embolism. Anesthesiology 2002;97:671eC681.
Wagner PD, Laravuso RB, Uhl RR, West JB. Continuous distributions of ventilation-perfusion ratios in normal subjects breathing air and 100 per cent O2. J Clin Invest 1974;54:54eC68.
Wagner PD, Laravuso RB, Goldzimmer E, Naumann PF, West JB. Distribution of ventilation-perfusion ratios in dogs with normal and abnormal lungs. J Appl Physiol 1975;38:1099eC1109.
Rodriguez-Roisin R, Bencowitz HZ, Ziegler MG, Wagner PD. Gas exchange responses to bronchodilators following methacholine challenge in dogs. Am Rev Respir Dis 1984;130:617eC626.
Kuriyama T, Wagner WW Jr. Collateral ventilation may protect against high-altitude pulmonary hypertension. J Appl Physiol 1981;51:1251eC1256.
West JB. Gas exchange. In: West JB, editor. Regional differences in the lung, 1st ed. New York: Academic Press; 1977. pp. 201eC243.
Itoh H, Nakatsu M, Yoxtheimer LM, Uematsu H, Ohno Y, Hatabu H. Structural basis for pulmonary functional imaging. Eur J Radiol 2001;37:143eC154.
Heitzman ER, Markarian B, Berger I, Dailey E. The secondary pulmonary lobule: a practical concept for interpretation of chest radiographs: II. Application of the anatomic concept to an understanding of roentgen pattern in disease states. Radiology 1969;93:513eC519.
Heitzman ER, Markarian B, Berger I, Dailey E. The secondary pulmonary lobule: a practical concept for interpretation of chest radiographs: I. Roentgen anatomy of the normal secondary pulmonary lobule. Radiology 1969;93:507eC512.
Terry PB, Menkes HA, Traystman RJ. Effects of maturation and aging on collateral ventilation in sheep. J Appl Physiol 1987;62:1028eC1032.
Putensen C, Rasanen J, Lopez FA. Improvement in VA/Q distributions during inhalation of nitric oxide in pigs with methacholine-induced bronchoconstriction. Am J Respir Crit Care Med 1995;151:116eC122.
Brown RH, Zerhouni E. New techniques and developments in physiologic imaging of airways. Radiol Clin North Am 1998;36:211eC230.
Herold CJ, Brown RH, Mitzner W, Links JM, Hirshman CA, Zerhouni EA. Assessment of pulmonary airway reactivity with high-resolution CT. Radiology 1991;181:369eC374.
Lynch DA, Newell JD, Tschomper BA, Cink TM, Newman LS, Bethel R. Uncomplicated asthma in adults: comparison of CT appearance of the lungs in asthmatic and healthy subjects. Radiology 1993;188:829eC833.
Musch G, Layfield JD, Harris RS, Melo MF, Winkler T, Callahan RJ, Fischman AJ, Venegas JG. Topographical distribution of pulmonary perfusion and ventilation, assessed by PET in supine and prone humans. J Appl Physiol 2002;93:1841eC1851.
Harris RS, Musch G, Layfield D, Vidal Melo MF, Venegas JG. Analysis of hypoventilating (HV) regions in asthmatic subjects using positron emission tomography (PET) . Am J Respir Crit Care Med 2002;165:A784.
Harris RS, Musch G, Winkler T, Vidal Melo MF, Layfield J, Tgavalekos NT, Lutchen KR, Venegas JG. Imaging of hypoxic pulmonary vasoconstriction (HPV) in asthmatic subjects during bronchoconstriction . Am J Respir Crit Care Med 2003;167:A698.
Harris RS, Isasi AG, Musch G, Winkler T, Vidal Melo MF, Tgavalekos NT, Bellani G, Venegas JG. Imaging aerosol deposition in normal and asthmatic subjects with positron emission tomography . Am J Respir Crit Care Med 2004;169:A545.(Marcos F. Vidal Melo, R. )