INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A recently published study in normal human subjects by two of us (P.S, N.Z.) showed a significant positive correlation between expiratory flow rate () and the quantity of nitric oxide (NO) exhaled per unit time (), and a significant negative correlation between and exhaled NO concentration (CE) (1). We have reanalyzed these data, and it is apparent that an even better correlation exists between the decreases in CE and the increases in which result from the increases in . This latter relationship can be readily explained by considering diffusion of NO from airway walls into the flowing air. As the NO concentration in the expired air falls with increasing flow, the gradient for diffusion between the airway wall and lumen increases, and the quantity of NO diffusing increases in direct proportion to this gradient.

This simple explanation for the relationships among , CE, and is analogous to transfer of heat to air flowing through a heated pipe, a system which has been thoroughly analyzed (2). As airflow increases, the temperature of the air leaving the pipe falls, but the quantity of heat transferred increases. If the relationship between outflowing air temperature and the quantity of heat leaving the pipe per unit time is determined at several flow rates, the temperature of the wall of the pipe and the determinants of heat transfer between the wall of the pipe and the air can be quantified. An increase in the temperature of the outflowing air at the same flow could be caused by either an increase in wall temperature, or an increase in the transfer of heat at the same wall temperature. The latter could occur if the surface area for heat transfer were increased by increasing the length of the pipe.

In asthmatic subjects, elevated exhaled NO concentrations are well documented (3). Explanations for this finding have frequently focused on increased NO production, which would increase NO concentration in airway walls, and therefore expired air. However, it is also possible that exhaled NO concentrations were elevated owing to increased effectiveness of NO transfer from airway wall to lumen in asthma. In this study, we attempt to distinguish between these two possibilities.

We have developed a model that quantifies the amount of NO exhaled per unit time () in terms of the NO concentration in the walls of the airways (C_w) and the diffusing capacity of NO transfer from the airways to the expired air (DNO). With this model, we were able to estimate the relative contribution of these two factors to NO output by analyzing the relationship between and expired NO concentration (CE) measured at different expiratory flow rates. When we applied this model to asthma, we found that the increase in CE was due to increased DNO, whereas the reduction in CE produced by inhaled corticosteroids was due to decreased C_w. Moreover, the product of DNO and C_w measured in asthmatics after steroids, presumably reflecting maximal diffusion of NO produced in airway walls by constitutive NO synthase (cNOS), was positively correlated with the provocative concentration of methacholine causing a 20% reduction in FEV₁ (PC₂₀) and FEV₁/ FVC measured before or after steroids.

Model

Our model (Figure 1) assumes that exhaled NO derives from alveoli and conducting airways, including the oropharynx. Nasal NO is excluded because the restricted breath technique employed in our studies closes the velum (1). During expiration, NO concentration in alveolar gas (Calv) is constant at a very low value, probably less than 5 parts per billion (ppb) (6), because NO, produced in the alveoli or delivered to alveoli from the inhaled gas or upper airways during inspiration, binds rapidly to hemoglobin in the pulmonary capillary bed. As alveolar gas ascends the airway, NO concentration in the airway lumen (C_a) rises progressively above Calv because NO diffuses from airway walls in proportion to the concentration gradient (C_w  C_a). Although C_w may vary with anatomic site in the airway, for simplicity we assume that C_w is constant and uniform along the airways at a value equal to the mean NO wall concentration driving diffusion. Because observations in human subjects indicated that CE achieved constant plateau values during constant-flow exhalations (1), we also assume that steady-state conditions exist. Given these assumptions, C_a would rise monotonically toward C_w as distance from the alveoli increases, achieving the measured value of CE at the airway exit. Recent findings support this possibility (10).

View larger version (18K): [in this window] [in a new window]   Figure 1.   Idealized diagram of an alveolar compartment containing NO at concentration Calv, and a conducting airway. Alveolar gas ascending the airway is conditioned with NO diffusing from the airway wall under a concentration gradient = Cw (the wall NO concentration)  Ca (the lumenal NO concentration). Ca increases monotonically from Calv at time zero to CE at the mouth. In an increment of time dt, the lumenal gas with volume dV ascends the airway and its concentration increases by dCa.

In Figure 1, the airways contributing NO to the expired air are depicted as a cylinder, whose volume is less than or equal to the anatomic dead space. The relation between C_a and distance along the airways, expressed as time for exhaled gas to move from alveoli to specific loci within the airways (transit time), is also shown. At constant expiratory flow, the longer the transit time, the greater the distance from the alveoli, and the greater the volume traversed. At a specific locus, the time for expired air to move from the alveoli to that locus is the ratio of the volume traversed (V) to the flow (). During the time interval dt, shown by the hatched bar in Figure 1, NO concentration increases by dC_a. The quantity of NO added during this interval (dq) equals the product of dC_a and the volume traversed such that

(1)

where dV = dt. The quantity of NO added per unit time by diffusion is proportional to airway wall surface area (A), the difference in NO concentration between airway wall and lumen (C_w  C_a), and the transfer coefficient (), which expresses the quantity of NO diffusing from airway wall to lumen per unit time per unit surface area, such that

(2)

where dA = surface area of dV in contact with the lumenal airway wall. Combining Equations 1 and 2, we have

(3)

With rearrangement,

(4)

If the single equivalent tube were a cylinder of radius r, then dA/dV = 2/r, and we have:

(5)

Integrating Equation 5 yields

(6)

where Calv = C_a at t = 0.

The transit time from the alveoli to the outflow end of the single equivalent cylinder is t = V/, where V = volume of the cylinder and = flow. C_a at the outflow of the tube is the expired concentration (CE). Substituting into Equation 6, we have

(7)

Let the NO diffusing capacity of the cylinder wall representing the airways (DNO) = A where A = surface area = 2rL and L = length of the cylinder. Substituting DNO /A for  in Equation 7 and simplifying yields

(8)

Solving Equation 8 for CE yields

(9)

Measurement of C_E at different levels of allows DNO, C_w, and Calv to be determined from Equation 9 by nonlinear least-squares regression.

During constant expiratory flow, the quantity of NO leaving the lungs per unit time () is equal to CE. Thus,

(10)

Because is the sum of NO diffusing from the airway wall surface per unit time (D) and NO leaving the alveoli by convection per unit time (Calv),

(11)

Thus, substituting in Equation 10, we have

(12)

The model was applied to data obtained in normal subjects and 25 asthmatic subjects before and after inhaled corticosteroids.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Studies in Normal Subjects

Ten nonsmoking nonasthmatic subjects with no respiratory disease (8 male, age 15 to 64) had exhaled NO measured at nine separate expiratory flow rates using a restricted breath technique. Some of these data have been previously published (1), but we have included data from individual subjects in this study (Table 1). The breathing circuit consisted of a mouthpiece connected to a Hans-Rudolph valve, through which subjects inhaled clean compressed air from a reservoir and then exhaled via an expiratory resistance while targeting a fixed mouth pressure of 20 mm Hg displayed on a pressure gauge. This technique was shown to close the velum thus excluding nasal NO during expiration (1). The fixed mouth pressure and resistance creates a constant expiratory flow rate. Subjects exhaled via nine separate resistances in turn while maintaining the same expiratory pressure giving nine flows of 4.2, 8.5, 10.3, 17.2, 20.7, 38.2, 75.6, 850, and 1,550 ml/s measured by a downstream pneumotachygraph. The exhaled NO single-breath profile showed a washout phase followed by a steady plateau; the latter gave the exhaled NO value. NO was detected with a chemiluminescent analyzer (270B; Sievers, Boulder, CO) with a lower limit of detection of 1 ppb and NO sampling rate of 250 ml/min. Daily 2-point calibration was performed with a zero gas and a standard 2 parts per million (ppm) NO gas (Matheson, ON, Canada).

Studies in Asthmatic Subjects

In two separate studies, we recruited a total of 25 nonsmoking asthmatics not taking systemic steroids or inhaled corticosteroid therapy in the 6 wk before the study. Asthma was defined according to the criteria of the American Thoracic Society. In the first asthma study, there were 10 subjects (4 female) age 18 to 28 and in the second, there were 15 subjects (7 female) age 19-33. Baseline pulmonary function and PC₂₀ did not differ significantly between the two studies: baseline methacholine PC₂₀ (geometric mean and SEM) was 0.3007 ± 0.1083 mg/ml. Spirometric data (mean ± SD) data for the 25 asthmatics were FEV₁ = 0.8174 ± 0.1664, FVC = 1.0052 ± 0.1236, and FEV₁/FVC = 0.8112 ± 0.1167, all expressed as fraction of predicted values. Predicted FEV₁ and FEV₁/FVC were significantly different from the normal subjects (p < 0.00005).

At baseline, all subjects had exhaled NO measured at expiratory flow rates of 20.2 and 45 ml/s (using the identical technique and equipment as with the normal subjects) followed by determination of methacholine PC₂₀ and spirometry. In the first asthma study, subjects received 1,000 µg/d beclomethasone (Beclovent-Glaxo, Mississauga, ON, Canada) from a metered-dose inhaler for 2 wk with repeat PC₂₀ spirometry and exhaled NO after this. In the second asthma study, subjects received 400 µg/d for 1 wk followed by 800 µg/d beclomethasone (Beclovent-Glaxo) for 1 wk followed by repeat exhaled NO, PC₂₀, and spirometry.

Statistics

The hypothesis that the data of a specific variable were normally distributed was tested by chi-square analysis. If the hypothesis could not be rejected at the 95% confidence level, an arithmetic mean and the standard error of the mean (SEM) were used unless otherwise specified. If the hypothesis that the distribution was normal could be rejected, the data of that variable were transformed to natural logarithms. The antilogarithm of the mean logarithm (geometric mean) was then used as the statistical parameter of the specific variable, and the standard error of the geometric mean was the product of the standard deviation of the natural logarithms and the geometric mean divided by the square root of the number of observations. The hypothesis that the major parameters of the model (DNO and C_w) and methacholine PC₂₀ were normally distributed could be rejected, but not the distribution of the natural logarithms. Therefore, all mean data for the model parameters shown in Tables 234 and PC₂₀ were transformed to natural logarithms.

The parameters of the model (Calv, DNO, and C_w) in the normal subjects were determined by nonlinear regression (nonlinear regression analysis program, Philip H. Sherrod, ). In addition to the estimation of DNO and C_w from the nonlinear regression at nine different flow rates, DNO and C_w were also determined from the linear regression of six flows and two flows. The estimates of DNO and C_w obtained by these three methods were compared by analysis of variance (ANOVA) and by Bland and Altman analysis (11).

The following t tests (2-tail) were performed. The fraction of the spirometric indices to the predicted values were compared with the hypothesized value of 1. The difference between the 25 asthmatic and 10 normal subjects was determined from a pooled estimate of variance. The difference in means before and after treatment with steroids was compared by paired observations.

From the 2-point estimation of DNO and C_w, any error in one of the parameters led to an error in the opposite direction of the other parameter; thus, there were significant negative correlations between DNO and C_w obtained from the 2-point method. The product of the two parameters (DNOC_w) provided an unbiased estimate because the two errors canceled each other. To determine an unbiased effect of DNO or C_w alone, multiple regression using these two variables as the independent variable was performed; and the degree of association between the two independent variables and the dependent variable was given as a partial correlation coefficient. The significance of the partial correlation coefficient was the significance of the regression coefficient. The effect of DNO and C_w as independent variables was determined for PC₂₀ and FEV₁/FVC, both before and after steroids, with the analysis performed on the natural logarithms of the variables (Table 4).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Normal Subjects

Analysis of mean data. Data from individual subjects are shown in Table 1. As shown in Figure 2, the relationship between mean values of CE and was markedly alinear. The curve through the points was obtained by fitting Equation 9 to mean CE and values using nonlinear regression analysis. This analysis yielded values for C_w (150.8 ± 5.5 ppb), Calv (5.3 ± 0.7 ppb), and DNO (5.55 ± 0.35 nl/s/ppb × 10³) that minimized the sum of the squared residual values. Substituting these values of C_w, Calv, and DNO back into equation 9, along with the values of employed in normal subjects, allowed prediction of CE at each . These predicted values were very close to the actual values of CE (r² = 0.9991).

View larger version (17K): [in this window] [in a new window]   Figure 2.   Plot of expired NO concentration against expiratory flow rate. The curve through the points was obtained by fitting Equation 9 to mean CE and values (Table 1) using nonlinear regression analysis (r 2 = 0.9991).

Figure 3A shows the relationship between expired NO output (, the product of CE and ) and expiratory flow () determined from this analysis. As shown in Figure 3B, which has expanded and scales, total expired NO output () increased rapidly and alinearly as expired flow increased from 0 to 20 ml/s. This rapid increase was due mainly to increasing diffusion of NO from airway walls (D, Equation 12). Thus at = 20 ml/s, D constituted about 85% of , and was 85% of its maximal value, calculated as the product of DNO and C_w. Thus, as exceeded 20 ml/s (Figure 3A), the resulting rise in was increasingly due to convection of NO from alveoli, calculated as the product of and Calv. For example, at = 1,000 ml/s, D was only about 13% of . Thus, exhaled NO output in these normal subjects was due mostly to diffusion from airway walls at the lower flows (e.g., 4.2, 8.5, 10.3, and 17.2 ml/s), and mostly to convection from alveoli at the higher flows (e.g., 75.6, 850, and 1,550 ml/s).

View larger version (14K): [in this window] [in a new window]   Figure 3.   Plot of total NO output () and output from airway NO diffusion ( D) alone against expiratory flow rate: (A) expiratory flows between 0 and 1.0 L/s; (B) expiratory flows between 0 and 0.020 L/s.

Besides examining the relationship between and , an important relationship between and CE is shown in Figure 4. When changes, CE can vary between C_w (its value at zero flow) and Calv (its value at infinite flow). At high levels of CE, where was low and was largely due to diffusion from airway walls, increased with decreasing CE because the diffusion gradient (C_w minus mean C_a) increased and was approximated by C_w  CE. At flows low enough for NO diffusion to dominate, the negative slope of a straight line through the mean -CE points was close in value to DNO, and the x-axis intercept of this line was close to C_w. Specifically, the linear regression of mean and mean CE obtained from the six lowest flows (4.2 to 38.2 ml/s) had a slope of 5.58 nl/s/ppb × 10³ (r = 0.98), compared with a DNO of 5.55 nl/s/ppb × 10³ as derived from Equation 9, and a concentration-axis intercept of 191 ppb, compared with a C_w of 151 ppb also derived from Equation 9. Figure 4 also shows the relation between CE and D, the NO output from diffusion. The negative slope of a straight line through the DCE relation over this range was somewhat less than true DNO because the NO diffusion gradient (C_w minus mean C_a), would have been greater than the difference between C_w and CE. However, with respect to the line through the (total NO output)-CE points, this error was offset by convection of NO from alveoli, which increased as CE decreased (and increased). As a result, the slope of the -CE line was slightly steeper than the slope of the D-CE line, and had a negative value closer to true DNO. The concentration-axis intercept of the -CE line overestimated true C_w by 27% because the true relationship between and CE at high CE (or very low flow) was highly curvilinear, falling suddenly and steeply to its value at zero flow (C_w). This steep fall occurred because the actual NO diffusion gradient (and therefore ) could still be finite when C_w  CE = 0 as lumenal NO concentration (C_a) at all points in the airway would be less than CE. Stated otherwise, a plot of D against mean C_a would have been linear with a slope nearly equal to the 6-point linear regression line.

View larger version (14K): [in this window] [in a new window]   Figure 4.   Plot of and D against expired NO concentration (CE). Linear regression was carried out on data from the six lowest flows (mean values from Table 1) to estimate DNO (negative slope) and Cw (x-axis intercept).

Parameters of individual subjects. Using nonlinear regression analysis to fit Equation 9 to the individual CE and data, we determined DNO, C_w, and Calv in nine of 10 subjects (Table 2, "9-Point Nonlinear"). In one subject, analysis was not possible as the function did not converge before the iteration limit was reached, and there was considerably more scatter of the data (note the much smaller linear correlation coefficient for the six lower flows shown in Table 2 under "6-Point Linear"). The regressions for the other nine subjects, however, all had r² values > 0.99. Moreover, with the exception of Calv in Subject 10 (p = 0.15), the t values determined for each of the parameters (Calv, C_w, and DNO) were highly significant (p < 0.01).

Linear regression of versus CE using data from the six lowest flows (Table 2, "6-Point Linear") yielded negative slopes and CE-axis intercepts that were not significantly different from DNO and C_w, respectively, obtained by 9-point nonlinear regression analysis (Table 2, "9-Point Nonlinear"). Estimating DNO and C_w in individual subjects from the slopes and intercepts of lines through -CE points obtained at only two flows (17.2 and 38.2 ml/s) yielded values (Table 2, "2-Point Linear") similar to estimates of DNO and C_w obtained from the 9-point nonlinear and 6-point linear regression (Table 2, "9-Point Nonlinear" and "6-Point Linear"). Indeed, ANOVA revealed no significant differences among values obtained by these three methods. Bland and Altman analyses (11), carried out by comparing each one of the three methods with each of the other two, indicated that all three methods provided unbiased estimates of the parameters, with no significant differences among the values obtained by the three methods; however, greater precision was obtained with the 6- or 9-point methods compared with the 2-point method.

Application of the Model to CE Measured at Two Low Flow Rates in Asthmatic Subjects

The above considerations indicate that the 2-point linear method can be used to obtain reasonable estimates of DNO, C_w, and DNOC_w in normal subjects. The same would be true of asthmatic patients, if C_alv in asthmatics were not significantly greater than C_alv in normals. Current unpublished measurements of C_alv, obtained at multiple expiratory flow rates in asthmatic subjects with high values of exhaled NO concentrations, indicate that this is the case.

The exhaled NO parameters obtained using the 2-point method in asthmatic subjects before and after steroids and in normal subjects are shown in Table 3. Exhaled NO concentrations at flows of 20.2 and 45 ml/s in asthmatic subjects were compared with measurements in normal subjects obtained at flows closest to these values (17.2 and 38.2 ml/s). All other things being equal, the slightly higher flows in asthmatic subjects would have been expected to produce CE values slightly lower than those in normal subjects; however, CE in asthmatic subjects before administration of inhaled steroids was significantly higher than in normal subjects as previously reported (3). DNO, estimated from the negative slope of the 2-point linear regression of NO output (the product of CE and ) versus CE, was 4-fold greater in asthmatics before steroids than in normal subjects, a highly significant difference. C_w, estimated from the CE-axis intercept of the 2-point linear regressions, was not significantly different in asthmatic and normal subjects, although the mean value was greater in the asthmatic subjects. The maximal amount of NO that could be added to the exhaled air by diffusion of NO from airway walls (DNOC_w) was more than 6-fold greater in the asthmatic subjects.

Inhaled steroids in the asthmatics caused highly significant decreases in CE, C_w, and DNOC_w, but did not significantly affect DNO. Although steroids decreased C_w to a value even below that measured in normal subjects, C_w in asthmatics was still not significantly different from C_w in normals. After steroids, DNOC_w remained more than twice the value measured in normal subjects.

Arithmetic mean expired NO output is plotted against arithmetic mean expired NO concentration measured at two levels of flow (see Table 1) in Figure 5A. There was a marked difference between normal and asthmatic subjects in the slope of the line through these points. Administration of steroids to asthmatics had no effect on slope, but produced an essentially parallel shift of the line to lower values of expired NO output ().

View larger version (21K): [in this window] [in a new window]   Figure 5.   (A) Plot of expired NO output against CE (from arithmetic mean values) at two low flows in normals and asthmatics before and after inhaled beclomethasone. The negative slopes and the x-axis intercepts of the lines joining the points gave approximate values for DNO and Cw. (B) Plot of expired NO output against CE in normals and in asthmatics before and after inhaled beclomethasone from solutions of Equation 9 with data (from arithmetic mean values) of two flows and assuming a value of Calv = 5.3 ppb.

Equation 9 can be solved for DNO and C_w with only two measurements of CE if Calv is known. Assuming Calv = 5.33 ppb (obtained by nonlinear regression of all nine arithmetic mean values of CE and in normal subjects), fitting Equation 9 by nonlinear regression analysis to the points shown in Figure 5A yielded the solutions shown in Figure 5B. Although the values of DNO and C_w differed somewhat from those estimated from linear relations of the two points (Figure 5A), they are qualitatively similar. For example, the marked difference in DNO between normal and asthmatic subjects remains. The lack of a significant difference in C_w between normal subjects and asthmatic subjects before steroids is even clearer, and the marked reduction in C_w after steroids is strikingly revealed.

Correlation of Maximal Airway NO Diffusion (DNOC_w) with PC₂₀ and FEV₁/FVC

Among the 25 asthmatic subjects, DNOC_w measured after steroids was significantly correlated with PC₂₀ and FEV₁/FVC measured before and after steroids (Figure 6, Table 4). The relations between DNOC_w after steroids and FEV₁/FVC shown in Figure 6 were essentially the same for FEV₁/predicted FEV₁, although the correlations were stronger for FEV₁/FVC. Multiple regression analysis revealed that these correlations were due to significant effects of both DNO and C_w (Table 4). In contrast, DNOC_w measured before steroids was not correlated with FEV₁/FVC or PC₂₀.

View larger version (27K): [in this window] [in a new window]   Figure 6.   The relationship between the maximal airway NO output from diffusion after steroids (DNOCw) and FEV1/FVC (left panels) and PC20 for methacholine (right panels), both before and after steroids. The lowest panels show that treatment with inhaled corticosteroids caused parallel upward shifts in the regression lines.

Steroid therapy did not alter the slopes of the lines of regression; however, it did cause small, but highly significant, upward displacements (lowest row of Figure 6). For example, paired analysis revealed that steroids increased PC₂₀ from 0.3007 ± 0.1083 to 1.0850 ± 0.4691 mg/ml (geometric mean ± SEM), an increase of 128 ± 29.6% (p = 0.0002); FEV₁ by 9.7 ± 2.4% (p = 0.003); and FEV₁/FVC by 7.1 ± 2.3% (p = 0.007) with no significant change in FVC. There were no significant correlations between changes in NO measurements induced by steroids and changes in FEV₁, FEV₁/FVC, or PC₂₀ induced by steroids.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Critique of Model

Tsoukias and coworkers (8) have developed a mathematical model of the factors controlling diffusion of NO from airway wall to lumen. Their analysis shows that the flux of NO from the tissue to the air per unit surface area (/A) is a linear function of C_a. They have shown that this linear relationship can be expressed by two constants: the intercept on the /A axis and the negative slope of the relationship between /A and C_a. Their analysis shows that these two constants are complicated functions of the molecular diffusivity of NO within the tissue of the airway wall, NO production per unit volume of tissue, the rate at which NO is consumed by chemical reactions within the tissue, and airway wall thickness. We have shown that their linear function is mathematically equivalent to our Equation 2, although their analysis does not explicitly consider the wall concentration of NO (C_w). The negative slope of their relationship between /A and C_a is our  and the intercept of this relationship is the product of  and C_w.

We treat the addition of NO to the exhaled air by diffusion as arising from a single equivalent tube with a uniform wall concentration (C_w). In this case, DNO is proportional to the product of the lumenal surface area and the transfer coefficient (). We have not attempted to partition DNO into  and surface area, but an analysis of the transient changes of CE after a breathhold or a step change in flow theoretically should allow the partition to be made.

Dweik and coworkers (6) measured the time course of NO concentration in 5- to 7-mm airways during a breathhold after a rapid forced expiration. Under these conditions, the increase in lumenal NO concentration was a first-order process with a rate constant of 0.25 s¹. Using their mean data in our Equation 6 for a bronchus 6 mm in diameter yields a  of 0.0375 cm/s. If mean DNO measured in our normal subjects (5.58 nl/s/ppb × 10³, Table 2) is divided by this estimate of , mean surface area would equal 149 cm², or approximately the surface area of the lobar bronchi, mainstem bronchi, and trachea (12). These results are compatible with the studies of Silkoff and coworkers (10), suggesting that the major site of NO diffusion in the normal subject is the large airways.

Because the airways are not a single homogeneous tube, we have analyzed how a distributive, nonhomogeneous pattern of NO diffusion in the airways would influence the parameters calculated from treatment of the system as a single equivalent tube. In a nonhomogeneous, distributive system, DNO would be the sum of all NO diffusing capacities at individual loci where NO was diffusing, and C_w would be the diffusion-weighted mean of the wall concentrations at these loci. If C_w varied along the course of the airways, maximal diffusion of NO from wall to lumen at each locus would be reached when expiratory flow was great enough for lumenal NO concentration to approximate Calv. When this occurred, total NO diffusion (D) would equal DNOC_w, the sum of the products of diffusing capacity and wall concentration at all loci where NO was diffusing. As flow approached zero, CE would approach the wall concentration of the most downstream locus. If the most downstream locus had a higher wall concentration than any other loci, the slope of the D-CE relationship would underestimate DNO and overestimate C_w, but the total maximal diffusible NO (DNOC_w) would be unaltered. Just the opposite would occur if the most downstream locus had the lowest wall concentration. These reciprocal changes in DNO and C_w due to nonhomogeneous distributive aspects of the system would not affect DNOC_w.

Comparison with Other Models of Exhaled NO

R. W. Hyde and his group at the University of Rochester, Rochester, NY (personal communication) assumed that airway walls add NO to exhaled air at a constant rate regardless of expiratory flow or lumenal NO concentration. They did not consider the addition of NO to the exhaled air as a process primarily determined by the difference in the wall and lumenal concentration of NO. Nevertheless, the mathematical relations of their assumptions can be shown to be identical to those we are using. They have made available to us data on seven normal subjects. When analyzed with our equations, the results of DNO, C_w, and DNOC_w were not significantly different from those of our normal subjects.

Jörres and coworkers have analyzed NO diffusion with equations that are essentially identical to our own (personal communication). Some of their findings have been presented as an abstract (13). They estimated transfer factor (equivalent to our DNO) and effective mucosal NO concentration (equivalent to our C_w), using nonlinear regression analysis to fit their equation to -CE data from normal subjects. Their transfer factors were considerably higher than diffusing capacities found by us and Hyde and coworkers (personal communication), and their effective mucosal concentrations were considerably lower than our C_w; however, when Jörres analyzed our data with his equation, he obtained the same values as we, suggesting that the discrepancies were caused by differences in data, not models. Differences in data may have been due to the different ranges of flow employed; i.e., Jörres and coworkers used no flow rate below 33 ml/s, whereas the lowest flows used by us and Hyde and coworkers (personal communication) were 4.2 and 6 ml/s, respectively. It is apparent from Figures 3 and 4 that the higher the , the less is determined by diffusion, and, as discussed subsequently, the greater the difficulty in determining parameters of NO diffusion from - and CE- relationships. Indeed, if we limited our nonlinear regression analysis to  38.2 ml/s, mean DNO increased 3-fold, and the variance of DNO increased more than 25-fold (6.31 ± 1.09 to 18.1 ± 5.55 nl/s/ppb × 10³). In addition, there was a more than 100-fold increase in the variance of C_w.

Figure 3 illustrates that increases in which accompany increases in are determined exclusively by convection of NO in alveolar gas when  > DNOC_w; therefore, the slope and intercept of the relation between and at high should give reliable estimates of Calv and DNOC_w, respectively. Tsoukias and coworkers (7) used this approach at  > 200 ml/s to estimate the maximal rate of NO diffusion from airway walls (our DNOC_w) and Calv. The mean value they obtained for maximal NO diffusion from airway walls (0.714 ± 0.117 nl/s) was not significantly different from values obtained by us and Hyde's group (personal communication). Although their estimate of Calv (5.6 ± 1.17 ppb) was not significantly different from our arithmetic mean (5.1 ± 0.3 ppb), it was significantly greater than Hyde's value (2.2 ± 0.37 ppb, p < 0.05).

When we determined the slope and intercept of the - relation from data at only two flows (38.2 and 850 ml/s), the arithmetic mean values for Calv and DNOC_w were 5.06 ± 0.30 ppb and 0.836 ± 0.199 nl/s, respectively. These values are not different from those obtained by 9-point nonlinear regression analysis (Table 2, paired analysis), and are very close to values obtained from - slopes and intercepts in seven normal subjects studied by Tsoukias and coworkers (7).

These results suggest that we can make reasonable estimates of Calv and the maximal rate of NO output from airway walls using data from only two flows as long as these flows are in the range where changes in are determined predominantly by alveolar convection (Figures 3 and 4); however, in this case we would not be able to estimate DNO accurately. Similarly, if we choose flows low enough so that is determined predominantly by diffusion from airway walls, two points on the -CE relation would be sufficient to provide reliable estimates of DNO and C_w, but not Calv (Table 2). It follows that estimation of all three parameters (Calv, DNO, and C_w) requires data over a wide range of , from levels where is nearly exclusively due to diffusion of NO from airway walls, to levels where is nearly exclusively due to convection of NO in alveolar gas.

These considerations support the previous recommendations of Silkoff and coworkers (1) that exhaled NO concentrations be measured at constant expiratory flow rates. Furthermore, they emphasize the potential utility of making these measurements at multiple levels of flow.

Comparison of Normal and Asthmatic Subjects

In asthmatic patients, we used data at two low flows to estimate DNO, C_w, and DNOC_w. In comparison with normal subjects (Table 3, Figure 5), asthmatics before steroid treatment had much higher DNO and DNOC_w. Although C_w was 1.7 times greater in asthmatics than in normal subjects, this difference did not achieve statistical significance. If C_w was not different, the higher CE measured in asthmatic subjects must be predominantly explained by the higher DNO. Thus, the findings suggest that the increased exhaled NO concentrations in asthma are largely the result of the increased effectiveness of NO transfer from the airway wall to lumen.

These findings are consistent with the preliminary report of Högman and coworkers (9), who determined the intercept (DNOC_w) and slope (Calv) of the - relation at flows ranging between 50 and 350 ml/s in normal and asthmatic subjects. The intercept was nearly 5-fold greater in the asthmatic subjects, in keeping with our findings of a nearly 6-fold increase in DNOC_w in steroid-naive asthmatics (Table 3). The slopes were not different in normal and asthmatic subjects, and the value in normal subjects was comparable to our estimates of Calv in normals.

The increased DNO in asthma could be due to either an increase in  or an increase in the lumenal surface area over which diffusion is occurring. From the analysis of Tsoukias and George (8), there are only two factors affecting : airway wall thickness, which varies inversely with ; and the rate of NO catabolism, which varies directly with . Because airway wall thickness is significantly increased in asthma, it seems likely that the increased DNO observed in asthmatics was caused by an increase in surface area, rather than an increase in .

As estimated above, the surface area of the large proximal airways thought to be responsible for NO diffusion in normal human subjects was approximately 150 cm². From Weibel's anatomic model of the human lung (12), this area could be provided by airways having internal diameters of 4.5 mm and greater. Weibel's model can also give anatomic perspective to the 4-fold increase in DNO observed in asthmatic subjects. For example, if asthma caused a distal extension of NO diffusion from the 16 parallel airways with diameters of 4.5 mm to the 128 parallel airways with diameters of 2.3 mm, surface area would double. If airway wall thickness decreased by half in proportion to the decrease in airway diameter,  would also double, because  is inversely proportional to wall thickness. The combination of one-half the wall thickness and twice the surface area would have resulted in a 4-fold increase in DNO. The distal extension of NO diffusion along the course of the airways necessary to achieve this result would be only 27 mm, or an increase in distance from the hilum of approximately 1 inch.

Effects of Inhaled Steroids in Asthmatic Subjects

Treatment of asthmatic subjects with inhaled steroids decreased C_w and DNOC_w, but had no significant effect on DNO. Because steroids decrease inducible NO synthase (iNOS) activity (14) and C_w depends on the rate of NO production per unit tissue volume (8), it is likely that the decrease in C_w was caused by a steroid-induced decrease in NO production by iNOS. Furthermore, the lack of a steroid effect on DNO implies that airway iNOS activity was located within the area of NO production by cNOS, which should be unaffected by steroids. We believe that DNO was increased in asthma because the area of NO production was larger. Our studies suggest that there was an increase in the area of cNOS activity, because DNO remained elevated after steroids; however, we do not know whether the increased area of NO production was due entirely to cNOS, because the lack of a steroid effect on DNO is compatible with either an increase or no change in the area of iNOS activity, so long as no area was occupied by iNOS alone.

Because DNOC_w and C_w after steroids (but not DNOC_w and C_w before steroids) were positively correlated with FEV₁/FVC and PC₂₀ before and after steroids, we hypothesize that a large component of exhaled NO in asthmatics before steroids was derived from iNOS, but that this NO had no detectable effect on baseline pulmonary function or airways responsiveness. Indeed, this iNOS-derived NO may have obscured an underlying modulation by NO derived from cNOS which acted to increase baseline pulmonary function and decrease airways responsiveness. The cNOS responsible for this modulation may have been the constitutive neuronal NOS (Type I), which plays a major role in mediation of nonadrenergic, noncholinergic (NANC) bronchodilation in humans (15).

Several previous observations support these possibilities. First, the sites of NO diffusion from airways and the distribution of cNOS nerve fibers in airways are closely correlated. In normal human subjects, it has been reported that half of the NO expired at a flow of 45 ml/s was produced in airways proximal to the midportion of the mainstem bronchi, and that fully one-third arose from the trachea alone (10). In comparison, cNOS-containing airway nerve fibers, which are found in airway smooth muscle but not epithelium, predominate in large airways, decrease progressively as one moves towards bronchi 1 to 3 mm in diameter, and are absent in bronchioli (15). Second, iNOS was detected in bronchial epithelium in 22 of 23 airway biopsies obtained from asthmatics not treated with steroids, but only two of 20 biopsies obtained from normal subjects (18). These results suggest that steroid-induced reductions in exhaled NO in asthmatic subjects may be due to decreased activity of epithelial iNOS, and that exhaled NO in normal subjects and asthmatic patients after steroids derived from nonepithelial cNOS. Third, N^G-nitro-L-arginine methyl ester (L-NAME), a nonselective NOS inhibitor, decreased exhaled NO concentrations in both normal and asthmatic subjects, whereas aminoguanidine, a selective inhibitor of iNOS, reduced exhaled NO only in asthmatic subjects (19). This suggests that elevated exhaled NO in asthma is attributable, in part, to iNOS. Fourth, inhalation of L-NAME at low concentrations decreased CE, but did not alter airways responsiveness to histamine in patients with mild asthma; however, a 3-fold increase in L-NAME concentration, which reduced CE by a similar amount, increased histamine responsiveness (20). These data suggest that low concentrations of L-NAME inhibited bronchial epithelial iNOS, which was easily accessed by inhaled agents, whereas high concentrations of L-NAME inhibited neural cNOS, which was less accessible.

Role of cNOS in Asthma

DNOC_w measured after steroids in asthmatics was more than twice DNOC_w measured in normal subjects (Table 3). Furthermore, among asthmatics low levels of DNOC_w after steroids were associated with low levels of PC₂₀ and FEV₁/FVC (Figure 6). If higher DNOC_w after steroids reflects higher NO production by cNOS, and lower in PC₂₀ and FEV₁/FVC reflects increased severity of asthma, then asthmatics with the highest cNOS activity had the best airway function.

The relation between cNOS-derived NO and severity of asthma may be analogous to that between insulin production and the severity of non-insulin-dependent diabetes mellitus, which is thought to result from decreased sensitivity to insulin. Here, insulin production is elevated, with the highest levels found in patients with fasting blood sugar and glucose tolerance closest to normal (21). In early disease, glucose levels are maintained nearly normal by overproduction of insulin. As the disease progresses, however, ability to maintain high insulin output diminishes, resulting in progressive hyperglycemia. By analogy, the high level of cNOS-derived NO in asthma may result from a decreased sensitivity of airway smooth muscle to the relaxing effects of NO. Progression of asthma may therefore relate, in part, to a decreasing ability to maintain high cNOS activity.

There is good evidence that modulation of airways by NO is less effective in animals with allergic inflammation or genetic predisposition to airways hyperresponsiveness. Mehta and coworkers (22) reported that L-NAME enhanced bronchial reactivity to histamine in control but not antigen-challenged guinea pigs, suggesting a defect in NO bronchodilator activity in the challenged animals. This defect probably did not arise from decreased NO production, because exhaled NO concentration was the same in control and challenged animals. D'Agostino and coworkers (23) and de Boer and coworkers (24) reported similar findings in isolated tracheal preparations from antigen- and sham-challenged guinea pigs. Miura and coworkers (25) showed that relaxations elicited by stimulation of NANC nerves or administration of an NO donor were attenuated in tracheal preparations from sensitized, antigen-challenged guinea pigs, but not in saline-challenged animals. Jia and colleagues (26) found that carbachol-contracted tracheal rings from a strain of hyperresponsive rats relaxed less and produced less cyclic guanosine monophosphate (cGMP) on exposure to the NO donor, nitroprusside, than rats with normal responsiveness, suggesting that guanylyl cyclase activity was decreased in the hyperresponsive strain. Subsequently, these investigators showed that the NOS inhibitor, L-NNMA, had little effect on carbachol-induced tracheal contractions and cGMP levels in the hyperresponsive strain, but markedly increased contractions and reduced the rise in cGMP in the normoresponsive strain (27). These results lend support to the possibility that a significant feature of asthma may be decreased effectiveness of, or sensitivity to, NO as a modulator of airways function.

Role of iNOS in Asthma

The role of increased iNOS expression in asthma is not clear. Considering the change in CE, C_W, or DNOC_w produced by steroids as an index of reduced iNOS activity, our studies do not indicate any significant influence of iNOS-derived NO on either FEV₁/FVC or PC₂₀, for there was no significant correlation between the magnitude of the NO changes and the changes in either FEV₁/FVC or PC₂₀. Further, there was no correlation between parameters of airway function and NO output by diffusion before steroids. Thus, while steroids clearly reduced exhaled NO and raised FEV₁/FVC and PC₂₀, there was no effect of this reduction in NO (presumably from iNOS-derived NO) on either baseline pulmonary function or airways responsiveness. Our results suggest, therefore, that the beneficial effect of inhaled steroids on FEV₁/FVC and PC₂₀ was not mediated by a reduction of iNOS expression, but rather by inhibition of some other steroid-sensitive component of allergic inflammation.

Although our studies failed to demonstrate any effect of steroid-sensitive NOS activity on airways function, two recent studies (28, 29) in patients with mild asthma untreated with steroids found that higher exhaled NO concentrations were associated with an increased sensitivity to histamine and methacholine, and that these relations did not exist after treament with inhaled steroids (28). These studies suggest, therefore, that steroid-sensitive NOS activity was detrimental to airway function

It is possible that iNOS plays a dual role. The inhalation of aminoguanidine, an inhibitor of iNOS, potentiated airways hyperresponsiveness measured in sensitized guinea pigs after the late asthmatic reaction to antigen challenge (30). In contrast, inhalation of aminoguanidine before development of the late asthmatic reaction reduced airways hyperresponsiveness and decreased the number of inflammatory cells in bronchoalveolar lavage fluid measured after the late asthmatic reaction (30). Thus, NO derived from iNOS appeared to exert both beneficial and detrimental effects on airways responsiveness, by modulating airway smooth muscle tone on the one hand and promoting airway inflammation on the other.

Summary

Using a model of NO diffusion in airways and measurements of exhaled NO concentrations at multiple flow rates in normal human subjects, we were able to estimate airways NO diffusing capacity (DNO) and the mean concentrations of NO in airway walls (C_w) and alveolar gas (Calv). Application of the model to asthmatic patients revealed a remarkable 4-fold increase in DNO which was not significantly changed by treatment with inhaled steroids. DNO may therefore quantify activity of airway cNOS, which in turn may represent the activity and extent of bronchodilating NANC nerves in the airways, because the highest values of DNO were associated with the best pulmonary function and least bronchial reactivity. Increased DNO may be an early manifestation of asthma, occurring before airways hyperresponsiveness or symptoms. Because lower baseline pulmonary function and higher airways responsiveness were accompanied by lower levels of DNO, a low DNO in well-established asthma could indicate significant pathologic alterations in airway structure and function possibly caused by some component of allergic inflammation. The assessment of DNO therefore provides a new method to evaluate airway function in health and disease and new insights into the pathophysiology of airways disease.

The decreased effectiveness of endogenous NO as a modulator of airway function may be caused by some component of airway inflammation; however, as suggested by Barnes and Liew (31), "It is also possible that the production of endogenous NO results in a long-term deleterious effect, and may be involved in the orchestration of eosinophilic inflammation that characterises asthma." Thus, the relation between eosinophilic inflammation and decreased effectiveness of endogenous NO is perhaps intimately linked and difficult to dissociate. If there were a primary decrease in the sensitivity to endogenous NO, as in the hyperresponsive rats studied by Jia and colleagues (26, 27), upregulation of NO production might occur as a compensatory response. Such upregulation from iNOS could itself be a major cause of the eosinophilic inflammation (31). Whatever the mechanism, we suggest that decreased effectiveness of endogenous NO may be a major feature of asthma.