Journal of Applied Physiology
Vol. 82, No. 5, pp. 1531-1541, May 1997
GAS EXCHANGE, MECHANICS, AND AIRWAYS

Partitioning airway and lung tissue resistances in humans: effects of bronchoconstriction

David W. Kaczka, Edward P. Ingenito, Bela Suki, and Kenneth R. Lutchen

Department of Biomedical Engineering, Boston University, Boston 02215; and Pulmonary Division, Brigham and Women's Hospital, Boston, Massachusetts 02115

ABSTRACT

INTRODUCTION

METHODS

RESULTS

DISCUSSION

ACKNOWLEDGEMENTS

FOOTNOTES

REFERENCES

ABSTRACT

Kaczka, David W., Edward P. Ingenito, Bela Suki, and Kenneth R. Lutchen. Partitioning airway and lung tissue resistances in humans: effects of bronchoconstriction. J. Appl. Physiol. 82(5): 1531-1541, 1997.The contribution of airway resistance (Raw) and tissue resistance (Rti) to total lung resistance (RL) during breathing in humans is poorly understood. We have recently developed a method for separating Raw and Rti from measurements of RL and lung elastance (EL) alone. In nine healthy, awake subjects, we applied a broad-band optimal ventilator waveform (OVW) with energy between 0.156 and 8.1 Hz that simultaneously provides tidal ventilation. In four of the subjects, data were acquired before and during a methacholine (MCh)-bronchoconstricted challenge. The RL and EL data were first analyzed by using a model with a homogeneous airway compartment leading to a viscoelastic tissue compartment consisting of tissue damping and elastance parameters. Our OVW-based estimates of Raw correlated well with estimates obtained by using standard plethysmography and were responsive to MCh-induced bronchoconstriction. Our data suggest that Rti comprises ~40% of total RL at typical breathing frequencies, which corresponds to ~60% of intrathoracic RL. During mild MCh-induced bronchoconstriction, Raw accounts for most of the increase in RL. At high doses of MCh, there was a substantial increase in RL at all frequencies and in EL at higher frequencies. Our analysis showed that both Raw and Rti increase, but most of the increase is due to Raw. The data also suggest that widespread peripheral constriction causes airway wall shunting to produce additional frequency dependence in EL.

airway resistance; inhomogeneities; airway wall shunting; forced oscillations; methacholine

INTRODUCTION

SEVERAL INVESTIGATORS have attempted to quantify the relative contributions of airway resistance (Raw) and tissue resistance (Rti) to total lung resistance (RL) in humans (1, 7, 16, 23, 24, 26, 33) and to determine how these contributions are altered in lung disease (2, 9). These earlier studies showed poor agreement in their results, with Rti reported to be between 2 and 40% of RL. Such discrepancies have been attributed to the different techniques used to partition Raw and Rti: ventilation with gases of different densities and viscosities (9, 24), interruption technique (26), volume displacement plethysmography (1, 2), and pressure plethysmography (16, 23). All of these techniques did not fully account for factors that are known to affect the measurement of Rti: breathing frequency, lung volume, and volume history. Thus, to date, the relative contributions of Raw and Rti to RL in humans are poorly understood.

Due to tissue viscoelasticity, Rti decreases quasi-hyperbolically with frequency and has been shown to approach zero near 2-4 Hz in some species (10, 21). However, Raw is fairly constant with frequency, provided that peak flows are small. Owing to these distinct frequency responses, it should be possible to partition Raw and Rti by using the frequency response of the lungs alone. Lutchen et al. (21) showed that in open-chest dogs, airway and tissue properties can be accurately partitioned by fitting low-frequency lung impedance data (ZL) with a lumped element model consisting of homogeneous airway and viscoelastic "constant-phase" tissue compartments. They relied on an optimal ventilator waveform (OVW) to estimate ZL (22). Partitioning ZL data into airway and tissue components with this modeling technique provides results that are consistent with data obtained by using alveolar capsules, even after induction of bronchoconstriction. In a subsequent study, Suki et al. (34) showed that the OVW approach provides reliable 0.1-5.0 Hz ZL data and hence partitioning of RL to Rti and Raw for in situ conditions and use of an esophageal balloon to establish pleural pressure.

The purpose of this study was to examine the partitioning of airway and tissue properties under typical tidal breathing conditions by using the OVW approach in awake healthy and bronchoconstricted humans. We measured ZL in nine subjects from 0.156 to 8.1 Hz. In four of the subjects, measurements were obtained at baseline and during methacholine (MCh)-induced bronchoconstriction.

METHODS

Subjects

Table 1. Physical data for 9 healthy subjects examined Subject Age, yr Gender Ht, cm Wt, kg HG 26 F 163 49 EI 37 M 175 72 JR 38 M 188 77 MH 26 F 163 66 TL 21 F 168 66 NC 23 F 170 64 KL 40 M 182 91 WO 35 M 183 82 TH 22 F 168 50 F, female; M, male; Ht, height; Wt, weight.

OVW

Table 2. Frequency, magnitude, and phase information for 2 OVWs Frequency, Hz OVW-A OVW-B Relative magnitude Phase, rad Relative magnitude Phase, rad 0.156250 1.000 4.95 0.171 5.20 0.390625 0.300 3.82 1.000 5.42 0.859375 0.250 4.37 0.214 5.55 1.484375 0.250 3.67 0.214 2.59 2.421875 0.175 4.05 0.143 5.71 4.609375 0.175 4.13 0.143 5.44 8.046875 0.125 4.02 0.100 4.00 OVW, optimal ventilator waveform; OVW-A and OVW-B, OVWs with peak energy concentrated at 0.156 and 0.391 Hz, respectively.

Data Acquisition

Protocol

Finally, in four subjects (EI, TH, MH, and NC) we made OVW measurements at baseline and during bronchoconstriction induced by MCh inhalation. Each subject inhaled three breaths of increasing concentrations of a MCh-aerosolized solution. The aerosols were generated with a nebulizer system (S&M Instrument). Three minutes after each MCh administration, a measurement of forced expiratory volume in 1 s (FEV₁) was obtained and followed by an OVW measurement ~1 min later. The MCh administrations were continued up to 25.0 mg/ml or until FEV₁ decreased below 80% of the baseline value. Each subject then received two puffs of aerosolized albuterol, and FEV₁ and OVW measurements were repeated.

Data Processing

Data Analysis

(1)

(2)

(3)

(4)

The model properties were estimated by using a nonlinear gradient search technique that minimized the performance index

(5)

Here, N is the number of frequencies used in the estimation. The Re(k) and Im(k) represent the real and imaginary components, respectively, of ZL at the kth frequency, with subscripts to denote actual data (d) and model predicted values (m). An estimate of the "goodness-of-fit" index (20) can thus be obtained as

(6)

where P is the number of free parameters in the model. Following Lutchen and Jackson (20), we can then estimate a parameter covariance matrix, the diagonal of which contains the squares of the SE for each parameter.

RESULTS

Healthy Subjects

Table 3 shows the estimated airway and tissue properties for all nine subjects. The range of values of Raw (0.99-2.48 cmH₂O · l¹ · s) and tissue elastance parameter H (4.68-12.65 cmH₂O/l) were reasonable for healthy lungs. The tissue damping parameter G ranged from 0.56 to 2.46 cmH₂O/l. Values of G for healthy humans have not been previously reported (see DISCUSSION). The tissue hysteresivity  ranged from 0.10 to 0.39 with a mean of 0.19, in close agreement with previously reported values for human  (8). The large spread in  appears to be the result of one outlier, subject EI, whose  was over two SDs above the mean value. The values of for the other eight subjects were all with in one SD of the mean. Repeatability of these estimates is summarized in Table 4. The parameter that demonstrated the greatest day-to-day variability was Raw, possibly reflecting differences in glottal aperture between measurements. In general, we found less day-to-day variability in the tissue parameters G, H, and , with two subjects (TH and TL) showing almost none (especially considering the estimated SE of the parameters).

Table 3. Summary of model parameter values estimated from ZL data for all 9 healthy subjects Subject Raw, cmH2O · l1 · s Iaw, cmH2O · l1 · s2  G, cmH2O/l H, cmH2O/l   HG 1.30 0.014 2.46 11.18 0.22 EI 0.99 0.018 2.63 6.70 0.39 JR 2.48 0.016 1.45 7.05 0.21 MH 2.29 0.014 1.68 8.46 0.19 TL 1.97 0.027 1.76 8.91 0.19 NC 2.04 0.022 0.56 5.95 0.10 KL 1.94 0.012 0.79 4.68 0.17 WO 0.79 0.008 0.89 7.33 0.12 TH 1.19 0.019 1.53 12.65 0.12 Mean ± SD 1.67 ± 0.61  0.016 ± 0.006  1.52 ± 0.71  8.10 ± 2.52  0.19 ± 0.09 ZL, lung impedance; Raw, airway resistance; Iaw, airway inertance; G, tissue damping; H, tissue elastance; , tissue hysteresivity.

Table 4. Model parameters for 5 subjects measured on different days Subject Date Raw, cmH2O · l1 · s Iaw, cmH2O · l1 · s2  G, cmH2O/l H, cmH2O/l   EI 22 Sep 95  1.7 ± 0.34  0.007 ± 0.009  1.7 ± 0.62  6.2 ± 0.43  0.28 ± 0.11  13 Oct 95  1.0 ± 0.13  0.017 ± 0.004  2.6 ± 0.24  6.7 ± 0.18  0.39 ± 0.03   9 Jul 96  1.1 ± 0.11  0.021 ± 0.003  1.3 ± 0.23  5.8 ± 0.18  0.22 ± 0.04  TH 25 Mar 96  1.2 ± 0.18  0.019 ± 0.006  1.5 ± 0.38  12.7 ± 0.33  0.12 ± 0.03   4 Apr 96  1.5 ± 0.12  0.019 ± 0.003  1.9 ± 0.23  13.9 ± 0.18  0.14 ± 0.02  NC 13 Jun 96  2.0 ± 0.17  0.022 ± 0.005  0.56 ± 0.31  5.9 ± 0.26  0.10 ± 0.05  20 Jun 96  2.4 ± 0.24  0.017 ± 0.007  0.76 ± 0.43  4.7 ± 0.36  0.16 ± 0.09  TL 23 May 96  2.0 ± 0.13  0.027 ± 0.004  1.8 ± 0.27  8.9 ± 0.22  0.19 ± 0.03  31 May 96  3.1 ± 0.26  0.017 ± 0.008  1.9 ± 0.49  7.7 ± 0.41  0.24 ± 0.07  KL 11 Jul 96  1.9 ± 0.12  0.012 ± 0.004  0.79 ± 0.25  4.7 ± 0.21  0.17 ± 0.06  29 Aug 96  2.0 ± 0.10  0.011 ± 0.003  0.56 ± 0.21  4.4 ± 0.18  0.13 ± 0.05 Values are means ± SE (20).

For some subjects, the model estimated values of Raw compared very well with those from the plethysmographic approach (Table 5). In three subjects (HG, EI, and KL) there was no difference, but the estimates tended to be higher in the other three (JR, MH, and NC). We point out that the plethysmographic technique determines Raw during panting at high frequency and low tidal volume, whereas our OVW measurements are made during physiological tidal excursions. The latter tends to increase Raw by narrowing glottal aperture (31).

Table 5. Comparison of Raw values from model estimate and plethysmograph Subject Raw (Model Estimate), cmH2O · l1 · s Raw (Plethysmograph), cmH2O · l1 · s HG 1.30 1.30 EI 0.99 1.09 JR 2.48 1.28 MH 2.29 1.66 NC 2.04 1.62 KL 1.98 1.94

Figure 5 illustrates Rti and Raw vs. frequency simulated for healthy lungs by using the mean parameter estimates in Table 3 and Eq. 3. Also shown are the percent contributions of Rti and Raw to RL. Our data suggest that during typical breathing frequencies (~0.25 Hz), Rti is a substantial component of RL in healthy humans (~40%), much higher than reported in most earlier studies (1, 2, 7, 9, 16, 23). Note, however, that the Raw estimates also include upper airway structures (oropharynx, glottis, and larynx), which may contribute roughly one-third of total RL at breathing frequencies (7, 31). Thus our data suggest that the contribution of Rti to intrathoracic RL is considerable during breathing (~60%).

MCh-Challenge Subjects

Figure 7 shows the dose-response results for FEV₁ and the estimated values of the airway and tissue properties for four subjects. We also show the values at baseline and after albuterol inhalation. A postalbuterol FEV₁ measurement was not obtained for subject MH. Using a paired t-test, we detected a significant increase from baseline in Raw for all subjects after even the lowest dose of MCh (P = 0.007). A significant increase in G was observed only at the highest dose of MCh (P = 0.017). For subject NC, who showed the largest drop in FEV₁, a clear trend of increasing Raw and G with MCh dose was observed. None of the subjects showed significant changes in H even at peak doses. After inhalation of albuterol, all parameters return close to baseline.

DISCUSSION

Airway vs. Tissue Properties: Healthy Lungs

Initial attempts to determine Raw and Rti in humans relied on ventilating the lungs with gases of different densities and viscosities (9, 24) and showed considerable disagreement because the kinematic viscosities of the foreign gases varied widely. McIlroy et al. (24), by using gases with kinematic viscosities equal to that of air, found that Rti contributed to 30-40% of RL during quiet breathing in healthy adult men, an estimate in agreement with ours.

In 1956, Marshall and Dubois (23) used a plethysmograph to partition Raw and Rti and found that Rti contributed 18% of RL in healthy humans. However, their subjects were required to breathe at ~100 breaths/min (1.6 Hz) and tidal volumes <300 ml. Subsequent studies by Bachofen (1, 2) used the plethysmographic approach to measure Raw, Rti, and Edyn during spontaneous breathing at several frequencies. We computed the effective constant-phase model parameters by using Bachofen's data (1) from five healthy humans at breathing rates of ~20 breaths/min and tidal volumes around 850 ml (Table 6). Values of Raw and H for his subjects were very similar to our estimates, whereas G and  were ~30% lower. The Bachofen technique, however, is prone to errors because of the thermal artifacts of gas warming and wetting in the airways (29).

Table 6. Comparison of constant-phase model parameters from 3 different studies Study No. of Subjects Raw, cmH2O · l1 · s Iaw, cmH2O · l1 · s2  G, cmH2O/l H, cmH2O/l   Present 9 1.67 ± 0.61  0.016 ± 0.006  1.52 ± 0.71  8.10 ± 2.52  0.19 ± 0.09  Bachofen (1) 5 1.53 ± 0.64  0.91 ± 0.47  7.08 ± 3.58  0.13 ± 0.03  Hantos et al. (11) 5 (all male) 1.08 0.013 0.94 4.35 0.22 Values are means ± SD for all subjects, except those in the Hantos et al. study (11), which are parameter estimates from reported mean ZL spectra from all 5 subjects. For Bachofen's data (1), we computed effective constant-phase model parameters for his measured breathing frequencies of ~20/min (or   2 × (20/60) rad/s). By definition,  = Rti/Edyn (8). Then, according to Eqs. 3 and 4, G = Rti and H = Edyn 1, with  calculated from  according to Eq. 2.

In 1986, Hantos and co-workers (11) used small-amplitude pseudo/random noise from 0.25 to 5 Hz to estimate ZL in five healthy male subjects during voluntary apnea. We fit the constant-phase model to the reported mean of these ZL spectra, and these parameter estimates are also shown in Table 6. Estimates of Raw, G, and H tended to be lower than ours, but  was similar. We point out that all of their subjects were men. The mean H from our male subjects was 6.14 cmH₂O/l, closer to the value of 4.4 cmH₂O/l computed from their study.

Finally, one could estimate G and H from the study of Suki et al. (33) which measured ZL from 0.01 to 0.1 Hz in healthy adults by superimposing small-amplitude forced oscillations on spontaneous breathing in a whole body chamber. However, their RL data showed no frequency dependence beyond 0.03 Hz, which is inconsistent with ours and previous studies (3, 11, 35). The reasons for this are not clear.

Airway vs. Tissue Properties: Constricted Lungs

What are the mechanisms contributing to the increase in Rti and the increased frequency dependence of EL at maximum MCh dose? Several theories have been proposed in the past, such as airway-tissue interdependence (27), small airway closure (14), and interstitial contractile elements (15). More recently, several studies have questioned whether this is a real physiological response or a modeling artifact due to airway inhomogeneties or airway wall shunting (4, 12, 18, 19, 21). We realize that our model is limited in that it assumes that the airways are a homogeneous system and that all of the frequency dependence in RL is due only to the viscoelasticity of the parenchymal tissues. However, parallel time constant inhomogeneities and airway wall shunting might also be expected to contribute to frequency dependence in RL and EL during constriction (4, 12, 18, 19). If so, fitting the homogeneous airway constant-phase model to such data will result in an overestimation of G and, consequently,  (18, 19, 21). To explore this issue further, we considered two alternative models. To compensate for parallel inhomogeneities, one model contained two separate Raw pathways, both leading to identical inertances and constant-phase tissues (Fig. 8B). With the other model, we divided the homogeneous Raw-Iaw airway system into two equal halves with a shunt airway compliance (Caw) to account for nonrigid airway walls (Fig. 8C). Each of these models contains only one additional parameter compared with the original constant-phase model. Both were applied to the data obtained from the four bronchoconstricted subjects at maximum MCh dose.

Table 7 shows parameter estimates for each of the three models at the highest concentration of MCh. For all subjects, applying the inhomogeneous airway model to the data resulted in no improvement in the model fit when compared with the original homogeneous airway model. However in three subjects (EI, MH, and NC), the airway shunt model yielded a rather large and significant improvement in model fit. The values of the estimated Caw were rather large for subjects EI and MH, considering previously reported estimates of Caw (25). This may be due to some shortcomings of the model (e.g., representing airway distensibility with only a single shunt compliance, or assuming Raw and Iaw are partitioned equally on either side of Caw). For subject TH, the airway shunt model yielded negative values for both Iaw and Caw, and there was little improvement in the model fit compared with the homogeneous airways model (² = 0.37 vs. 0.38). Figure 9 illustrates all three model fits to RL and EL data obtained for subject MH at peak MCh dose.

Table 7. Comparison of 3 different models for 4 severely bronchoconstricted subjects Subject Model Raw(1) Raw2 Iaw Caw G H    2 EI HA 3.47  0.012 3.88 6.14 0.63 0.54 IA 4.75 13.34  0.012 3.19 5.87 0.54 0.60 AS 7.35  0.021 0.040 1.67 7.30 0.23 0.43 MH HA 4.08  0.015 3.36 11.01 0.31 0.29 IA 10.88 6.51  0.015 3.13 10.93 0.29 0.33 AS 5.99 0.007 0.013 2.54 12.53 0.20 0.12 TH HA 4.11 0.013 2.45 14.23 0.17 0.38 IA 8.15 8.27 0.013 2.45 14.23 0.17 0.43 AS 3.57  0.040  0.010 1.86 12.59 0.15 0.37 NC HA 5.67  0.017 2.29 8.37 0.27 0.12 IA 10.29 12.65  0.017 2.25 8.35 0.27 0.13 AS 6.57 0.023 0.006 1.69 8.76 0.19 0.06 Raw(1), Raw parameter of homogeneous airway (HA) or airway shunt (AS) model or Raw1 parameter of inhomogenous airway (IA) model; Raw2, Raw parameter of IA model; 2, model fit; Caw, airway compliance.

An important result is that the G and  estimates from the airway shunt model are significantly lower than those found with the homogeneous airway model, whereas the Raw estimates are significantly higher. However, for subjects MH and NC, the estimates of G obtained from the airway shunt model are still elevated compared with control. This suggests that their Rti is still increased at high doses of MCh but less than previously thought. Figure 10 illustrates the relative airway and tissue contributions to RL predicted by the homogeneous airway model at baseline and the airway shunt model at peak MCh dose in subject MH. At baseline, we see that Raw is roughly 60% of RL for this subject at breathing frequencies, but after MCh, Raw increases to ~70%. Thus, after MCh inhalation in humans, both Raw and Rti may increase, but Raw more so.

The superiority of the airway shunt model in three of the four subjects suggests that at high doses, MCh causes a substantial constriction of the peripheral airways, resulting in significant central airway wall shunting. This would seem to be in disagreement with animal studies that demonstrated how exogenous bronchoconstriction can lead to increased parallel airway inhomogeneities and an artifactual increase in Rti (4, 12, 19, 21). We point out, however, that the constricting agonist was delivered intravenously in these other studies, not aerosolized as in the present study. If there was a difference in the airway response to MCh in our study, it may be due to the different mode of delivery (30), although one would expect a more heterogeneous airway response with an aerosolized delivery (28). Perhaps a more important difference is that the maximum dose of the agonist delivered relative to body weight was much higher in these animal studies compared with our human study, and we terminated the MCh protocol as soon as the subject's FEV₁ dropped below 80% of the baseline value. Our data are remarkably consistent with the modeling study of Lutchen et al. (18), which demonstrated that airway inhomogeneities causes an increase in EL at low frequencies (below 2 Hz), whereas airway wall shunting increases EL at higher frequencies (out to 5-10 Hz) but only when the entire lung periphery experiences significant constriction. Moreover, the phenomenon of airway wall shunting will cause an artifactual increase in Rti, for the reasons consistent with previous arguments (12, 18). Namely, an important new parallel pathway (in this case, the airway walls) now prevents the flow at airway opening to be equivalent to the flow delivered to the tissues. This will cause an additional frequency dependence in RL and EL that is not due to tissue viscoelasticity (25).

Summary

ACKNOWLEDGEMENTS

This study was supported by National Heart, Lung, and Blood Institute Grant HL-50515 and National Science Foundation Grant BCS-9309426.

FOOTNOTES

Address for reprint requests: D. W. Kaczka, Boston Univ., Dept. of Biomedical Engineering, 44 Cummington St., Boston, MA 02215 (E-mail: dk{at}bu.edu).

Received 20 September 1996; accepted in final form 2 January 1996.

REFERENCES