INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The analysis of phase III slopes obtained in single-breath washout (SBW) maneuvers provides information about ventilation distribution in the lung periphery. In particular, when He and SF₆ tracer gases are included in the inspiratory gas mixture, their respective alveolar slopes in the washout curves are indicative about the site where ventilation inhomogeneities occur (1). In general, larger ventilation inhomogeneities give rise to steeper alveolar slopes. A marked alveolar slope increase has been observed in smokers with chronic obstructive pulmonary disease-type emphysema, with a positive correlation between the magnitude of the slope and the degree of emphysema (4, 5). Potentially, such a slope increase can be produced by airspace enlargement or by other airways alterations usually associated with emphysema (such as fibrosis, smooth muscle hypertrophy, and mucus) or by both (3, 4, 6, 7).

Smokers' emphysema usually has a panacinar component and centriacinar component that differ in respect to the distribution of lung destruction and in mechanical properties (8). In centriacinar human emphysema, small airways diseases can explain some of its functional aspects, such as lower compliance, in comparison with panacinar emphysema (8). In this respect, small airways diseases could also be responsible for the increased N₂ phase III slopes found in smokers with emphysema (5, 7). A difficulty in the interpretation of ventilation distribution studies in patients is that other factors such as impaired gas exchange and/or modifications in the interregional ventilation distribution can also contribute to the observed alveolar slope changes. In this respect the postmortem rat lung has the advantage of having slopes generated primarily by diffusion-convection ventilation inhomogeneities (9, 10). Indeed, it has been shown that the physical formulation of diffusion-convection interactions in the rat lung geometry can quantitatively reproduce the observed alveolar slopes in normal rats (11). This suggests that the study of panacinar and centriacinar emphysema is probably appropriate in rats, as long as its extrapolation to human subjects is handled with care.

In animal models in general, elastase has been frequently employed to induce a lesion that functionally and morphometrically resembles human panacinar emphysema (12). Centriacinar emphysema can also be induced in animals by cadmium-chloride (CdCl₂) inhalation. This drug induces peribronchiolar and interstitial fibrosis that can evolve to airspace enlargement (12), which is predominantly localized around respiratory bronchioles (13). The combination of CdCl₂ inhalation with the oral intake of the lathyric agent -aminopropionitrile (-APN), which inhibits lysyl oxidase (i.e., the enzyme responsible for cross-linkings between elastin and collagen), leads to a more emphysematous pattern (14). We have induced panacinar-like and centriacinar-like emphysema in two separate groups of rats in order to evaluate how each morphologic deterioration relates to indices of ventilation distribution. In particular, N₂, He, and SF₆ slopes were monitored for different breathing maneuvers, including end-respiratory breath-hold, in order to investigate the diffusion-convection-dependent ventilation distribution in the modified lung structure.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Materials and Subjects

We have studied ventilation inhomogeneity in 87 male Wistar rats that were originally classified in five groups: (1) control group for elastase (n = 15) orotracheally instilled with saline, (2) elastase group (n = 11) orotracheally instilled with elastase, (3) control group for CdCl₂ (n = 18) inhaling saline aerosol, (4) CdCl₂ group (n = 18) inhaling CdCl₂, and (5) CdCl₂ + -APN group (n = 25) receiving oral administration of -APN in addition to the CdCl₂ inhalation treatment. We will further refer to the rats with induced lung disease as E for elastase treatment, CD for CdCl₂ treatment, and CD for the combination of CdCl₂ and -APN treatment.

All treatments were initiated either 8 or 6 wk prior to the study day and such that the ventilation studies were done when the rats were 14 wk old. The elastase administration was performed in a group of 6-wk-old rats by orotracheal instillation of 50 IU of porcine pancreatic elastase (Boehringer Mannheim, Germany) in 0.5 ml saline. These rats were instilled in the supine position and shaken in order to improve the distribution of elastase. After instillation, the rats were returned to their cages, where food and water were provided ad libitum, for an 8-wk-period until the study day. The CdCl₂ (Sigma, St. Louis, MO) was administered to a group of 8-wk-old rats by aerosol in saline (0.1%) at a rate of 1 h/d during a total period of 15 d. This was done by placing the conscious rats in exposure chambers that were continuously alimented with a CdCl₂ aerosol, generated by use of a compressed air nebulizer (Hudson). After each exposure, the rats were returned to their cages. The aerosol treatment period was followed by a 6-wk interval with food and water ad libitum until the study day. Finally, -APN (Fluka, Switzerland) administration, in addition to the CdCl₂ treatment, consisted in mixing the -APN with the powdered food (0.3%), starting 48 h before the first day of CdCl₂ aerosol administration until the study day.

Methods used to obtain lung functional, ventilation distribution, and morphometric parameters have been reported in great detail in our previous work (15). In the following sections, we describe the general measurement procedures.

Lung Function Tests

The functional study was performed in a breathing assembly for small animals identical to the one previously used for a similar study (15). The rats were anesthetized with sodium pentobarbital (60 mg/kg), tracheotomized, put into a plethysmograph, and fitted to a cannula that allowed communication with the breathing assembly. As soon as the rat was connected to the breathing equipment, it was paralyzed with 0.2 mg of pancuronium bromide and then artificially ventilated at a rate of 58 breaths/min with a tidal volume of 4 ml (Harvard miniature ventilator; Harvard, Edenbridge, Kent, UK).

Lung volumes were obtained as follows. The inspiratory capacity (IC) was considered as the change in lung volume between an airway pressure of 0 and 30 cm H₂O. Functional residual capacity (FRC) was determined immediately after the death of the rat, using a rebreathing test where an N₂-free gas mixture was rebreathed for 30 s by a tidal airway pressure change between 0 and 20 cm H₂O. The relation between initial N₂ and final N₂ concentration was used to determine the initial lung volume at 0 cm H₂O, i.e., FRC. The total lung capacity (TLC) was obtained by adding IC and FRC.

For the measurement of diffusing capacity, the rat lung was inflated from 0 to 20 cm H₂O with a test gas mixture of 0.25% C¹⁸O and 10% Ar in air. After 2 s of breath-holding, the rat lung was emptied by the mass spectrometer (Marquette Electronics, Milwaukee, WI) at a constant expiratory flow of 1.1 ml/s, during which the gas concentrations were recorded continuously. The diffusing capacity (DL_CO) and diffusing coefficient (K_CO) were calculated as previously described (15).

For the quasistatic pressure-volume curves, the rat was inflated with air to an airway pressure of 30 cm H₂O and slowly deflated by the mass spectrometer at a constant expiratory flow of 1.1 ml/s down to residual volume. Compliance (C_L) was considered as the steepest slope of the deflation pressure-volume curve. The curvature of the pressure-volume curve was computed by fitting a mono-exponential function with constant K to the part of the pressure-volume curve above the inflection point.

Finally, we induced flow-volume curves by inflating the rat with air to 30 cm H₂O and inducing a forced expiration by applying a pressure of 40 cm H₂O using a vacuum reservoir. Forced vital capacity (FVC), forced expiratory flows at 75% of FVC (F75), and specific forced expiratory flows (F75/FVC) were determined from these curves.

The above-described series of tests were performed twice, after which the rats were killed by inflating the lungs to TLC with 100% N₂. This method of killing was chosen in order to avoid possible atelectasis and to wash off O₂ from the lungs. After that, ventilation distribution tests were performed within the hour following killing.

Ventilation Distribution Tests

SBW tests consisted of breathing a gas washout mixture containing 5% He, 5% SF₆, and 90% O₂. The maneuvers started from a given pre-inspiratory lung volume (PILV), and the inspiratory volume (IV) of washout gas was injected slowly with the syringe, at a rate of approximately 1 ml/s. The expiration was performed by the mass spectrometer, which emptied the lungs down to residual volume at a flow rate of 1.1 ml/s. A series of SBW tests were carried out with varying IV between 4 ml and IC, varying PILV between FRC and FRC + 4 ml, and also varying end-inspiratory breath-holding time (tBH) between 0 and 20 s. The SBW test with IV = 4 ml, PILV = FRC, and tBH = 0 s was taken as the reference maneuver. Whenever either one of these three parameters was varied in a given range, the other two parameters remained set to their reference values.

All SBW maneuvers were performed twice. Alveolar slopes were computed in phase III of the N₂, He, and SF₆ washout expirations by linear regression between 40 and 80% of the expired volume between the beginning of expiration and the residual volume. Although inspired gases (He and SF₆) produced negative phase III slopes, whereas the lung resident gas (N₂) showed upward phase III slopes; all slopes were considered as absolute values. Finally, the slopes were normalized by mean expired concentration (N₂) or inspired minus mean expired concentration (He, SF₆) of the expiration where the slope was computed.

Morphometry

After the lung function and ventilation part of the study, the chest was opened and the cardiopulmonary block was removed. The lungs were fixed by filling the lungs with 10% formalin to an airway pressure of 25 cm H₂O for 24 h. After fixation, three lung blocks from three different lobes were taken away for morphometry. From each block, 5-µm sections were stained with hematoxylin-eosin. Mean linear intercept (Lm), as a measure of interalveolar wall distance, was determined by light microscopy at a total magnification of ×100. For each rat, a total of 45 randomized selected microscopic fields were examined under a cross-hair ocular (Zeiss cross ocular 464043-9902, Zeiss, Germany). Lm was obtained by multiplying the total length of the cross-hair by the number of microscopic fields (i.e., 45) and dividing by the total number of intercepts encountered over the 45 microscopic fields.

In eight rats from each group, collagen accumulation was also studied in lung sections stained with Sirius red. Sirius red has been shown to bind selectively to collagen present in paraffin sections and to emit birefringence when analyzed under polarized light (15). Measurement of birefringence was studied in 45 fields per rat using a magnification of ×63 by point-counting in a 40-points graticule. Percentage of points falling on birefringent areas was considered as the collagen density.

Statistical Analysis

All the data are expressed as mean ± SE. Normal distribution of each variable was checked by the Kolmogorov-Smirnov test. Variance homogeneity was also checked by Cochran's C test. Analysis of variance (ANOVA) and of covariance (ANCOVA) were used for comparison between groups. Least significant differences multiple range tests were used for analysis of differences among means. Pearson correlations were also calculated (STATGRAPHICS PLUS, Manugistics, Inc., MD).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

From this section onward, the rats from the elastase control group and the CdCl₂ control group are pooled to one control group (n = 33), because they showed no significant difference with respect to any of the morphometric, lung function, or ventilation distribution parameters under study.

Morphometry

Morphometric data are shown in Table 1. In terms of Lm, both E and CD groups presented increased values that were significantly different from the control group and not significantly different from each other. The CD group did not show a significantly different Lm with respect to the control group. Qualitative observation showed that CD group presented a more irregular airspace enlargement in comparison with the E group, as can be seen in Figure 1. The collagen density quantified by Sirius red stain in the E and CD groups was only significantly increased in the CD group. The latter group presented significantly more birefringence (control group: 3.21 ± 0.32%; E group: 3.72 ± 0.24%; CD group: 7.75 ± 0.32%), as can be seen in Figure 2. Both CD and CD groups presented foci of peribronchiolar and interstitial fibrosis that led to thicker alveolar walls, which was cumulated with airspace enlargement (increased Lm) only in the CD group.

View larger version (158K): [in this window] [in a new window]   Figure 1.   Photomicrographs of hematoxylin-eosin-stained lung sections from rats of the control group (A), CdCl2-aerosolized group (B), elastase-instilled group (C ), and CdCl2 plus -APN-treated group (D). Note the airspace enlargement in lung sections in panels C and D. (Original magnification: ×40.)

View larger version (70K): [in this window] [in a new window]   Figure 2.   Photomicrographs of Sirius red-stained lung sections under polarized light from rats of the elastase-instilled group (left panel ) and of the CdCl2 plus -APN-treated group (right panel ). (Original magnification: ×63.)

Lung Function Tests

Pulmonary function parameters obtained for the control, E, CD, and CD groups are also listed in Table 1. In the E group, FRC and TLC increased with respect to the control group. In the CD group, FRC was also significantly increased but, because of a decreased IC in this group, TLC remained unchanged with respect to the control group. In the CD group, FRC and TLC were significantly increased with respect to the CD and control groups but were similar and not significantly different from each other in both emphysematous groups E and CD.

With respect to the control group, lung distensibility around the pressure-volume inflection point, as measured by CL, was significantly increased in the E group, significantly decreased in the CD group, and not significantly different in the CD group. When normalized by TLC, compliance (C_L/TLC) in the E group was no longer significantly different from the control group, whereas in both the CD and CD groups C_L/ TLC was decreased to the same extent. The exponential constant K, reflecting the pressure-volume curvature above the inflection point, was significantly higher in the E group and significantly lower in the CD and CD groups with respect to the control group. A significantly larger K decrease was seen in the CD group with respect to the CD group. In all three groups, E, CD, and CD, end-expiratory flows, both absolute and specific, were significantly decreased with respect to the control group. The most important reduction in F75 and F75/ FVC was observed in the CD group. When normalized to actual alveolar volume, diffusing capacity, i.e., K_CO, was significantly reduced only in the emphysematous groups E and CD.

Ventilation Distribution Tests

Ventilation distribution tests were only performed in the control group and in the two emphysema groups, E and CD. Figure 3 shows N₂ and SF₆-He slope behavior in the control, E, and CD groups with varying PILV for IV = 4 ml. The actual slope values of N₂, He, SF₆, and SF₆-He corresponding to Figure 3 can be found in Table 2. From Figure 3A, it appears that N₂ alveolar slopes are lower in the E group with respect to the control group, when expressing the abscissa PILV as the number of milliliters above FRC (as was done in Figure 3). However, when comparing N₂ slopes between the control and E groups using FRC as covariate (see Table 1 for FRC values in both groups), no significant difference could be demonstrated. In contrast, N₂ slopes of the CD group are dramatically increased with respect to the control group, and this was true irrespective of whether PILV is considered in terms of milliliters above FRC or in absolute number of milliliters.

View larger version (13K): [in this window] [in a new window]   Figure 3.   N2 phase III slope (A) and SF6-He phase III slope differences (B) as a function of pre-inspiratory lung volume (PILV) with IV = 4 ml and tBH = 0 s for the control, E, and CD groups. *Significantly different from control group, p < 0.0001.

As far as the SF₆-He slope difference is concerned (Figure 3B), its value was reversed with respect to the control group both in the E and CD groups and over the entire PILV range, but as a result of different He and SF₆ slope modifications with respect to the control group (Table 2). In the control group, the He slope is larger than the SF₆ slope (i.e., SF₆-He negative). In the E group, both He and SF₆ slopes decreased but with a larger He slope decrease to the extent that the SF₆ slope became larger than He, and the SF₆-He difference changed sign. In the CD group, He and SF₆ both increased and with SF₆ increasing far more than He, the SF₆-He slope difference also became positive.

Figure 4 shows the dependence of the N₂ slope and the SF₆-He slope difference on IV, for PILV = FRC (keeping in mind that FRC is much larger in the two emphysematous groups E and CD with respect tot he control group). Basically, the N₂ slope as well as the SF₆-He slope difference show similar dependency on IV (Figure 4) as on PILV (Figure 3) in all three groups.

View larger version (13K): [in this window] [in a new window]   Figure 4.   N2 phase III slope (A) and SF6-He phase III slope differences (B) as a function of inspiratory volume with PILV = FRC and tBH = 0 s for the control, E, and CD groups. *Significantly different from control group, p < 0.0001. Significantly different from control group, p < 0.05.

In Figure 5, N₂, He, and SF₆ slopes are normalized to their respective slope value for tBH = 0 s so that for each gas its value becomes 100% for tBH = 0 s. Within each of the three groups, the gas with the larger diffusion coefficient produces a more rapid slope decrease as a function of breath-holding time. Across groups, the comparison of the slope decrease as a function of breath-hold time for each gas shows the same pattern in the control and E groups but a very distinct pattern in the CD groups. In the latter group, He decrease is faster and SF₆ decrease is slower than in the E and control groups.

View larger version (11K): [in this window] [in a new window]   Figure 5.   N2 (A), He (B), and SF6 (C ) slopes expressed as a percentage of N2, He, and SF6 slopes for tBH = 0 s as a function of end- inspiratory breath-holding time for the control, E, and CD groups. *Significantly different from control group, p < 0.0001.

Correlations

In order to relate the airspace enlargement found in the E and CD groups to their lung functional behavior, we correlated the morphometric and lung function parameters, the results of which are shown in Table 3.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The phase III slope of the SBW is considered to reflect mainly ventilation inhomogeneities in the lung periphery and it has been used as an index of peripheral airways dysfunction in lung disease (3, 17). Major advances in the interpretation of the phase III slope have been made using mathematical models that can identify the gas-mixing mechanisms generating the alveolar slope, both in humans (18, 19) and in rats (11). In postmortem rats, we could expect to find two predominant components to ventilation inhomogeneity (18): (1) convection-dependent ventilation inhomogeneity, which acts prior to the diffusion front and generates concentration differences between relatively large lung units, for instance as a consequence of the differences in mechanical properties between these lung units, and (2) diffusion-convection-dependent inhomogeneity, which is generated by convection-diffusion interaction at branch points located in the lung generations over which the diffusion front is spread. The diffusion-convection-dependent inhomogeneities are directly related to the asymmetry of the lung structure, which can be altered in lung disease due to unequal narrowing of two airways or due to changes in volume or configuration of lung units subtended from each airway.

We have previously reported a multiple-breath washout analysis, which was aimed at separating convection-dependent inhomogeneities and diffusion-convection-dependent inhomogeneities both in healthy rats (9) and in rats with emphysema (20). It was shown that in normal rats as well as in rats with panacinar-like or centriacinar-like emphysema, convection-dependent inhomogeneities were negligible.

In this work, we focus on the diffusion-convection-dependent inhomogeneities by means of different SBW maneuvers that are thought to influence the alveolar slope in a particular way. In order to relate the obtained results to the rat lung structure, we first describe the two different types of lesions obtained in rats in terms of morphometry.

Morphometry

As in our previous study (15), elastase treatment (E group) led to airspace enlargement regularly distributed through all the parenchyma without apparent involvement of airways or fibrosis (Figure 1). The absence of fibrosis was verified by in situ quantification of collagen (Figure 2), which was not significantly increased in the E group with respect to the control group. When airspace enlargement was quantified in terms of mean linear intercept Lm values were significantly increased by 22% in the E group (Table 1).

The rat lungs exposed to CdCl₂ (CD group) presented foci of fibrosis preferentially around bronchioles that occasionally extended through alveolar walls (Figure 1), in accordance with the morphologic descriptions published by other investigators (13, 21). In terms of Lm, the CD group was not significantly different from the control group (Table 1), in line with observations of Lai and Diamond in rats (24), but in contrast to the Lm increases of the order of 32% found by Snider and associates (21) and 25% by Niewoehner and Hoidal in hamsters (14).

The oral intake of -APN in addition to the CdCl₂ treatment (CD group) made the lesion evolve to an emphysematous pattern, with irregular airspace enlargements primarily detected adjacent to fibrotic areas (Figure 1). The emphysematous lesion in the CD group was reflected in a 29% larger Lm than in the control group (Table 1), a significant Lm increase that was also obtained by Niewoehner and Hoidal, using the same method of administration (14). Lai and Diamond (24) did not find any effect of -APN over the lesion induced by CdCl₂, in terms of Lm, but maybe their method of administration, by peritoneal injection, is less effective than oral intake through diet.

Although the Lm increases for the two emphysematous groups, E and CD, were not significantly different from each other with respect to the control group (Table 1); microscopic examination of these lungs showed that only the CD group had a more irregular distribution of airspace enlargement mainly located around bronchioles, as has been observed by other investigators (13, 21), and presented fibrosis in terms of in situ collagen quantification. These features are similar to those encountered in human panacinar and centriacinar emphysema (8, 16). Thus, based on our morphometric data, we can indeed consider our E and CD groups to represent panacinar- and centriacinar-like emphysema, respectively. This also is in accordance with the classification based on the anatomical localization of the lesion induced with CdCl₂ (13) and with elastase (12).

Lung Function

The lung function parameters obtained in the two types of induced emphysema show characteristics that are in line with observations in human subjects with panacinar or centriacinar emphysema (8). The rats with centriacinar-like emphysema (CD group) presented lower end-expiratory flows, lower quasi-static compliance, lower specific compliance, and a smaller exponential constant K than the rats with panacinar-like emphysema (E group). The K_CO was significantly decreased in the E and CD groups and DL_CO, although decreased in both, was only significantly different from the control group in the CD group. To our knowledge, the only report of diffusion capacity measured in panacinar and centriacinar human emphysema separately is by Kim and colleagues (8). These investigators did not find a decreased K_CO in either group, although DL_CO tended to be lower in the centriacinar group; K_CO was not reported in that study.

With regard to correlations between Lm and lung compliance parameters, a positive correlation between Lm and C_L in the E group and the absence of a significant correlation in the CD group (Table 3) is also in line with observations in patients with panacinar and centriacinar emphysema (8). In conclusion, we may state that from a lung function point of view the rats from the E and CD groups show similarities with observations in humans with panacinar and centriacinar emphysema, respectively.

Ventilation Distribution

Having obtained two different models of emphysema, we now discuss the effect of each type of lung disease on ventilation distribution patterns through analysis of normalized phase III slopes in the SBW maneuvers. These tests have been performed with varying PILV, IV, and tBHparameters that are known to influence diffusion-convection-dependent inhomogeneities in a particular fashion. Indeed, model simulations of diffusion-convection-dependent effects in humans (25) and rats (11) predict that for PILV = FRC, N₂, He, or SF₆ slopes decrease with IV, PILV, and tBH in an exponential pattern. In particular, the changes in the SF₆-He slope difference as a function of these parameters are indicative of diffusion-convection-dependent inhomogeneities between very peripheric lung units (2).

In our previous studies, we have reported negative SF₆-He slope differences in healthy rats (9), consistent with model simulations in a rat lung model (11). In elastase-treated rats, the SF₆-He slope difference changed sign depending on the level of lung inflation with the He slope being the main factor for the changes of SF₆-He slope difference (15). In addition, the changes in N₂ slopes and SF₆-He slope differences between the control and elastase groups for the maneuver with PILV = FRC were shown to be a consequence of PILV differences between both groups. The results were suggestive of elastase-induced elastic alterations between acini with intraacinar structural alterations that did not contribute to changes in phase III slope. Our present SBW data obtained in the control and E groups are in agreement with these previous results. In particular, we verified that the differences in slopes between the control and E groups could be accounted for by differences in FRC by using PILV as a covariate in the analysis of variance which tests for differences in slopes between groups.

In the CD group, the extent of change in phase III slope is such that it overrides any possible effect of increased FRC (Figures 3 and 4). The magnitude of the slopes points to severe alteration of ventilation distribution, while the slope dependence on PILV and IV confirms that we are still dealing with diffusion-convection-dependent mechanisms. The tBH dependence of phase III slope (Figure 5) suggests that in contrast to the E group, which only has intraacinar structural changes, there have been interacinar structural changes in the CD group, resulting in different pathways over which diffusive equilibration can take place during an end-inspiratory breath-hold. The fact that the He slope equilibrates faster than in the control and E groups indicates that the units between which most of the slope is generated in this group are peripherally situated where distance for diffusive equilibration is small. In contrast, SF₆ decreases less as a function of end- inspiratory breath-hold because these peripheral units are situated in the region of the more distally located SF₆ diffusion front where the diffusion-convection-dependent mechanism is most effective for SF₆. This reasoning is also consistent with the fact that it is the SF₆ slope that is responsible for the large increase in the SF₆-He slope difference in the CD group with respect to control for tBH = 0 s (Figures 3 and 4).

From Figures 3-5, it can be seen that the most discriminative SBW maneuver to infer peripheral alteration is the one with PILV = FRC and IV = 4 ml (considering tBH = 0 s). For larger IV or PILV, i.e., where large end-inspiratory airway cross-sections tend to attenuate diffusion-convection-dependent effects (25), phase III slopes decrease and slope differences between groups become small. Despite the differences between normal rats and normal humans (i.e., an SF₆-He slope difference reversal between both species), the underlying diffusion-convection mechanism is the same and therefore the recommendations for the preferential maneuver to detect structural alterations of the lung periphery may be extrapolated to humans. This explains observations by Van Muylem and coworkers (3) in humans, where the classic vital capacity SBW would mask ventilation inhomogeneities between small units, which became apparent with SBW tests using small inspired volumes starting from FRC. In addition, when slope differences are small (e.g., between the control and E groups), most information can be gained from the SF₆-He slope differences, where one can speculate about more or less peripheral location of ventilation inhomogeneity depending on when the SF₆-He changes are brought about by He or SF₆ (26).

The present study also underlines the potential of studying He and SF₆ slopes with varying end-inspiratory breath-holding times not only to discriminate different types of emphysematous lesions in patients but also to help in the interpretation of the localization of the units affected by emphysematous lesions. A very different behavior of He and SF₆ slopes as a function of end-inspiratory breath-holding time with respect to normal subjects was indeed observed by Magnussen and associates in patients with lung emphysema (27), although these investigators did not report the kind of emphysema of their two subjects under study.

In conclusion, the panacinar- and centriacinar-like rat lung models presented here lead to different mechanical properties and also to different patterns of ventilation distribution. In both models, the SF₆-He slope difference becomes positive. In panacinar emphysema, this effect is due to a larger He than SF₆ slope decrease, probable as a consequence of elastic alterations between acini but with negligible effect of intraacinar structural alterations. In centriacinar emphysema, the SF₆-He slope difference becomes positive because the SF₆ slope increases more than the He slope, suggesting severe irregular structural alterations that affect the configuration of interacinar pathways. The extrapolation of the findings from this extended ventilation study in rats to humans suggests that the recommended washout maneuver for subjects with suspected emphysema would be a SBW test with a small inspired volume and pre-inspiratory lung volume near FRC.