INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

From studies using bronchoalveolar lavage and bronchial mucosal biopsies it is well known that inhalation of ozone causes an influx of neutrophils and an increase in the concentration of several cytokines and mediators of inflammation within the airways (1). Because of their invasive nature, these methods are not suited for studies that include multiple measurements or investigations of large groups of subjects. In contrast, the method of sputum induction is a noninvasive approach, which has been shown to be safe, reproducible, and sensitive in the detection of airway inflammation (2). Sputum samples are likely to be derived from more central airways, whereas samples of bronchial wash and bronchoalveolar lavage represent more peripheral areas of the lung (3, 4). The observation that ozone causes effects in the central airways (5, 6), in addition to the peripheral airways, justifies the use of induced sputum to investigate the effects of ozone.

To our knowledge, two studies have been published using the method of induced sputum in ozone exposure (7, 8). Both involved a relatively high concentration of ozone (400 ppb) and demonstrated an influx of neutrophils 4 and 12 h after exposure. From the range of responses, as observed in a single exposure using lavage techniques, it has been suggested that individuals differ in their sensitivity to ozone in terms of inflammatory parameters (9). In contrast to acute lung function responses (10), however, it is not known whether the inflammatory response to ozone is reproducible.

Therefore, we studied whether the individual inflammatory response to ozone as detectable in induced sputum was reproducible in healthy subjects as well as in those with mild asthma in two independent exposures to 250 ppb ozone. To investigate the dose-dependence of the inflammatory response, subjects were additionally exposed to 125 ppb ozone. Sputum inductions were performed 1 and 24 h after exposure to compare immediate and delayed responses to ozone.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

We studied 15 subjects with mild bronchial asthma (Table 1). The diagnosis was based on internationally accepted guidelines (11). The subjects showed a positive skin prick test against at least one common airborne allergen (Allergopharma, Reinbek, Germany) and were hyperresponsive, their provocative concentration of methacholine (PC₂₀FEV₁) producing a 20% fall in FEV₁ being below 8 mg/ml. None of the subjects were receiving inhaled corticosteroids during the course of the study or had used steroids within the 3 mo before the study. Short-acting ₂-adrenoceptor agonists were withheld for at least 8 h before each test. As a control group, we included 21 healthy subjects (11 female, 10 male; mean age ± SD, 28 ± 5 yr; FEV₁, 106 ± 11% of predicted; geometric mean [SD] PC₂₀FEV₁, 16.0 [2.5]) into the study. Nine of the control subjects reported symptoms of allergic rhinitis and showed a positive skin prick test to at least one common allergen, but they had no history of asthma symptoms. All subjects were studied outside their respective season. Four of the control subjects were mildly hyperresponsive to inhaled methacholine, their PC₂₀FEV₁ being above 2 mg/ml.

All subjects were nonsmokers and had not experienced a respiratory tract infection within the 4 wk preceding each test. The tests were approved by the Ethics Committee of the Chamber of Physicians of the State of Schleswig-Holstein, and all subjects gave their informed written consent.

Study Protocol

On the first visit the subject's history was taken, a physical examination, baseline lung function measurement, methacholine inhalation challenge, skin prick test for 20 allergens, and electrocardiography were performed. Lung function measurements and methacholine inhalation challenges (12, 13) were performed according to standardized protocols. On the second visit a sputum induction was performed to assess the subject's ability to produce sputum of sufficient quality. At least 1 wk later, on visits 3 to 6, subjects inhaled either filtered air (FA) or 125 ppb ozone or, on 2 separate days, 250 ppb, in randomized order, in a single-blind manner and at the same time of the day. Study days were separated by at least 1 wk and the median (interquartiles) interval between these days was 19.0 (14 to 37) d. Lung function was measured before, during, and after the exposure. One hour and 24 h after the exposure a sputum induction was performed.

Ozone Exposure

Details of the exposure system have been described in detail (14). Ozone generated from 100% oxygen was added to purified air using a mass flow controller and monitored continuously to achieve concentrations of 0, 125, and 250 ppb, with peak deviations of less than 5%. The ozone analyzer was calibrated regularly by the Environmental Protection Agency of Hamburg, Germany. A two-way valve was attached to the inspiratory limb, and each subject breathed through a mouthpiece while wearing a noseclip. Temperature and relative humidity of the inspired air did not differ between exposures, mean ± SD values being 26.6 ± 1.2° C and 56.0 ± 6.8%, respectively.

Within the 3-h exposure periods, subjects performed six cycles of breathing at rest and during exercise. In each cycle they inspired air from the mixing system for 15 min at rest and 15 min during bicycle exercise. Once every hour, lung function values as well as symptom scores were assessed. After the sixth cycle, lung function was measured repeatedly over 1 h, and symptom scores were assessed immediately and 1, 6, and 24 h afterwards. Work load was adjusted on an individual basis to achieve a minute ventilation of 14 L /(min · m²), known from previous studies to be well tolerated for 3 h. The same work load was chosen on all exposure days.

Assessment of Symptoms

The severity of symptoms comprising nose and throat irritation, cough, chest tightness, shortness of breath, headache, nausea, thirst, and dizziness was recorded in a written questionnaire on a five-point scale. Afterwards, symptoms of upper respiratory tract, lower respiratory tract, and general symptoms were added to give three summary scores. Changes in symptoms were expressed as differences between scores assessed before and immediately after exposures because ozone effects were most pronounced at this time.

Sputum Induction and Analysis

Our method of sputum induction and processing has been described in detail (15). After premedication with 200 µg salbutamol, subjects inhaled hypertonic saline from an ultrasonic nebulizer (NE-U12; Omron, Tokyo, Japan) (mean ± SD delivered amount was 1.72 ± 0.2 ml/ min; mass median diameter was 4.9 µm) for three consecutive 10-min periods. The sputum samples produced after the first (3% saline), during the second (4% saline), and during the third (5% saline) 10-min inhalation period were processed separately. Adequate (15) plugs of sputum were selected to reduce contamination with saliva (16). After homogenization and dilution, the samples were centrifuged, the supernatants were collected and stored at 80° C until further analysis, cell counts were assessed, the viability was determined by trypan blue exclusion, and a portion of 30,000 cells was used to prepare cytospin slides.

Differential cell counts were assessed by two independent observers from 400 to 500 nonsquamous cells on coded Giemsa-stained cytospin slides and averaged. Results were expressed as percent of nonsquamous cells. Albumin was analyzed by RIA (Kabi Pharmacia, Freiburg, Germany) in pooled samples, eosinophil cationic protein (ECP) by the CAP system (ECP FEIA; Kabi Pharmacia), and IL-8 by ELISA (CLB, Amsterdam, The Netherlands).

Statistical Analysis

We computed arithmetic mean values and standard errors of the mean (SEM) for lung function parameters and percent cell numbers; geometric mean values and geometric SEM (expressed as a factor) were calculated for cell numbers per milliliter sputum and fluid-phase concentrations. For exposure conditions and subjects' characteristics, mean and standard deviations (SD) were calculated; for symptom scores, median values and percentiles were calculated. Correlations were expressed as Pearson's linear correlation coefficient (r), and reproducibility was expressed by intraclass correlation coefficients (Ri), which were derived from one-way analysis of variance tables as the ratio of variance among subjects to total variance.

To achieve comparability with conventional sputum analysis, we computed pooled mean values of mediators from the three consecutive periods by multiplying the individual values for each period by the weight of the respective sputum plugs. Correspondingly, pooled mean values of percent cell numbers were calculated by taking into account the absolute cell number per sputum period.

For statistical comparisons, data were logarithmically transformed when necessary to meet the requirements of homoskedascicity and normal distribution in the analysis of variance (ANOVA). Distribution was tested by the analysis of residuals. Repeated-measures ANOVA was performed for the pooled data (filtered air, 125 ppb, and the average of both exposures to 250 ppb ozone), with exposure conditions as within-factor, group (control and asthma) as between-factor, and an interaction term between exposure conditions and group. In addition, the repeated-measures ANOVA was performed for both groups separately (Table 3) and combined without a between factor. Data of the three separate consecutive induction periods were analyzed in the same way.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ventilation and Heart Rate during Exposure

Minute ventilation and heart rate during resting and exercise periods did not show significant differences between groups and exposures. Mean (± SD) minute ventilation during rest was 6.8 ± 2.3 L/min, with a corresponding heart rate of 80 ± 9. Mean (± SD) minute ventilation during exercise was 26.2 ± 6.4 L/min, corresponding to a minute ventilation per body surface of 14.0 ± 3.0 L/(min · m²). The mean (± SD) heart rate during exercise was 115 ± 10 beats/min.

Changes in Symptoms after Ozone Exposure

No significant differences between the study groups were found. Overall, upper respiratory tract symptoms and general symptoms were not significantly altered by exposures. The median (interquartiles) change of all subjects combined in lower respiratory tract symptoms was 0 (0 to 0) after FA, 1 (0 to 2) after 125 ppb ozone, and 1 (0 to 2.5) after the exposures to 250 ppb ozone; there was a significant increase after exposure to 250 ppb ozone as compared with FA (p < 0.01).

Changes in Lung Function after Ozone Exposure

Changes in FEV₁ and VC after exposures did not significantly differ between groups (Table 2). In both groups, a significant (p < 0.01 each) decrease in FEV₁ and VC was observed 1 h after exposure to 250 ppb on both exposure days, but not after 125 ppb ozone and FA. Twenty-four hours after exposure no effect of ozone was detectable. Changes in lung function were not related to changes in symptoms.

Ozone-induced Inflammatory Changes in Sputum Composition

Fifty-five (6.8%) of the 804 single sputum samples collected in the consecutive inhalation periods could not be evaluated because of insufficient quality. Squamous cell contamination did not differ significantly between groups, exposure conditions, or time points (1 h versus 24 h), overall median values (interquartiles) being 2.4% (1.1 to 5.0), 1.6% (0.7 to 3.6), and 3.0% (1.2 to 5.6) for the three consecutive inhalation periods. Furthermore, viability did not differ significantly between groups. One hour and 24 h after exposure to 250 ppb ozone viability was increased as compared with FA when both groups were analyzed combined (p < 0.01). Under all exposure conditions, the total number of nonsquamous cells per milliliter sputum decreased during the consecutive induction periods in both groups (data not shown). Mean values of cell concentrations as pooled over the three induction periods are presented in Table 3. These data demonstrate that in healthy subjects, cell concentrations were increased 24 h after exposure to 250 ppb ozone as compared with FA (p < 0.05).

Neutrophils. Compared with FA, pooled percentages of neutrophils (Table 3) were significantly increased 1 h and 24 h after exposure to 250 ppb ozone in healthy (p < 0.001, p < 0.01, respectively) and asthmatic (p < 0.05, p < 0.05) subjects. The healthy subjects did not show an effect of 125 ppb ozone, but the asthmatic subjects demonstrated increased percentages of neutrophils (Figure 1), this effect being significant (p < 0.01) when both time points (1 h and 24 h) were combined in the analysis. In both groups, percentages of neutrophils were elevated 24 h after exposure to filtered air or ozone as compared with 1 h after exposure (p < 0.01 each) (Figure 2). Neutrophil numbers per milliliter of sputum were significantly increased 1 h and 24 h after exposure to 250 ppb in healthy subjects (p = 0.01, p < 0.001) but not in asthmatic subjects.

View larger version (47K): [in this window] [in a new window]   Figure 1.   Dose dependence. Effect of filtered air (FA), 125 ppb ozone, and 250 ppb ozone on the pooled percentage of neutrophils (panel A) and the pooled concentrations of IL-8 (panel B) 1 h after exposure. For the percentage of neutrophils, mean and SEM are given; for the concentration of IL-8, geometric mean and SEM are given. The average values of both exposures to 250 ppb ozone are presented.

View larger version (15K): [in this window] [in a new window]   Figure 2.   Sequential sputum analysis and time-dependence of inflammatory changes. Mean (SEM) percentages of neutrophils in sputum samples produced during three consecutive 10-min periods of hypertonic saline inhalation (connected by lines). The upper panels show the data for healthy subjects and the lower panels for asthmatic subjects. The average values of both exposures to 250 ppb ozone are presented 1 h after exposure (closed circles) and 24 h after exposure (open circles).

The separate analysis of the three consecutive sputum samples demonstrated that percentages of neutrophils were highest in the first portion and that the effect of ozone was evident in all portions and, within the scatter of the data, resulted in a parallel upward shift of neutrophil numbers (Figure 2). Furthermore, 24 h after exposures, asthmatic subjects showed a smaller change in percent neutrophils over the three consecutive induction periods (p < 0.05).

Macrophages. When both groups were combined, percentages of macrophages (Tables 3 and 4) were significantly decreased 1 h and 24 h after exposure to 250 ppb ozone as compared with FA (p < 0.05). The same was true for numbers of macrophages per milliliter at 1 h. Both effects were more pronounced in healthy than in asthmatic subjects.

Eosinophils. Under all exposure conditions, the mean percentage of eosinophils was significantly (p < 0.001 each) higher in asthmatic than in healthy subjects (Tables 3 and 4). In both groups, ozone did not affect the percentages of eosinophils. However, in healthy subjects a small but significant increase in the number of eosinophils per milliliter sputum was detected 1 h and 24 h after exposure to 250 ppb ozone (p < 0.01 each). In the asthmatic subjects, percentages of eosinophils decreased during consecutive periods of sputum induction (Table 4), but this effect was statistically not significant.

Lymphocytes. One hour after ozone exposure, there were no effects on the percentages of lymphocytes (Table 3). However, 24 h after exposure to 250 ppb ozone, pooled percentages of lymphocytes were increased (p < 0.05) in both groups combined, the effect being more pronounced in healthy (p < 0.005) than in asthmatic (p = 0.15) subjects. Furthermore, in healthy and asthmatic subjects, total cell numbers of lymphocytes per milliliter sputum were increased 24 h after exposure (p < 0.01 and p < 0.05, respectively).

Fluid-phase measurements. Asthmatic subjects showed higher (p < 0.01) concentrations of ECP than did healthy subjects but no significant increase after exposure to ozone (Tables 3 and 4). In healthy subjects, the increase in ECP concentration 1 h and 24 h after exposure to 250 ppb was borderline significant (p = 0.08 and p = 0.0545, respectively). The concentration of ECP decreased during sputum induction (Table 4) and was correlated with the total number of eosinophils per milliliter sputum in asthmatic subjects; 1 h and 24 h after 250 ppb ozone, pooled samples showed correlation coefficients of r = 0.46 and 0.47, respectively (p < 0.05 each). In both groups, concentrations of ECP were increased 24 h as compared with 1 h after exposure to FA (p < 0.05 each).

When both groups were combined, the concentrations of IL-8 (Table 3) were increased 1 h after exposure to 125 ppb ozone (p < 0.05), as well as 1 h and 24 h after exposure to 250 ppb ozone (p < 0.001, p < 0.05). The increase 24 h after exposure to ozone was more pronounced in asthmatic subjects. Furthermore, concentrations were higher 24 h as compared to 1 h after exposure; this was statistically significant for the exposure to 250 ppb ozone and to filtered air (p < 0.05 each). There was a significant correlation between the concentration of IL-8 in supernatants and the total number of neutrophils per milliliter; 1 h and 24 h after 250 ppb ozone, pooled samples showed correlation coefficients of r = 0.34 and 0.68, respectively (p < 0.01).

Because of the limited amount of sputum supernatants available, the analysis of albumin was performed only in pooled samples (Table 3). Twenty-four hours after exposure to 250 ppb ozone as compared with FA, values were increased in both groups (p < 0.05 each).

Correlation between Changes in Lung Function and Sputum Composition

The changes in symptoms, FEV₁, and VC observed after exposure to 250 ppb ozone were not significantly correlated with the increase in sputum neutrophils, the decrease in macrophages, or changes in the concentration of fluid-phase parameters. This analysis was performed for both groups taken together or separately, as well as for pooled sputum and for separate analysis of the consecutive samples.

Reproducibility of the Individual Response to 250 ppb Ozone

Nineteen of the 21 healthy subjects and 12 of the 15 asthmatic subjects participated in the second exposure to 250 ppb ozone. There was a significant correlation between the changes in lower respiratory tract symptoms elicited during the first and the second exposure to 250 ppb ozone, both in healthy (r = 0.74, p < 0.001) and in asthmatic (r = 0.84, p < 0.001) subjects; the same was true for the changes in FEV₁ (healthy: r = 0.77, p < 0.001; asthma: r = 0.66, p < 0.05) (Figure 3). The correlation for the changes in VC was significant in asthmatic subjects (r = 0.74, p < 0.01), but, because of one outlier, not in healthy subjects (r = 0.38, p = 0.10). Excluding this subject from analysis resulted in a significant correlation (r = 0.76, p < 0.001). Changes in lung function were not related to changes in symptom scores.

View larger version (12K): [in this window] [in a new window]   Figure 3.   Reproducibility of individual lung function responses to 250 ppb ozone. Correlation between the effects of ozone detected after the first and the second exposure to 250 ppb ozone on the changes in FEV1. Displayed are the percent changes caused by ozone relative to the changes observed after exposure to filtered air and the line of identity: healthy subjects (open circles); asthmatic subjects (closed circles).

Reproducibility of sputum parameters was analyzed for the pooled sputum samples. As there were no significant differences between groups in the changes of percent neutrophils, both groups were analyzed together. The mean (95% CI) increase in the percentage of neutrophils 1 h after exposure to 250 ppb ozone relative to the values after filtered air was 221.0% (104.4 to 337.6) in the first test and 246.2% (89.7 to 402.7) in the second test. The individual changes in the percentage of neutrophils showed a high intraclass correlation coefficient (Ri = 0.86) and a significant correlation coefficient (r = 0.87, p < 0.001). It can be seen in Figure 4 that a high degree of reproducibility could be observed for both groups and was also maintained when the three consecutive sputum induction periods were analyzed separately. Similar results were obtained 24 h after exposure; the mean (CI) increase in pooled neutrophil percentages was 78.8% (35.2 to 122.4) after the first and 123.9% (50.0 to 197.7) after the second exposure. These results were also reproducible (Ri = 0.71, r = 0.79, p < 0.001).

View larger version (17K): [in this window] [in a new window]   Figure 4.   Reproducibility of individual inflammatory changes to 250 ppb ozone. Correlation between the effects of ozone detected 1 h after the first and the second exposure to 250 ppb ozone on the percentage of neutrophils. Displayed are the changes in the percentages of neutrophils relative to the values detected after filtered air for the three consecutive inhalation periods separately. The lines of identity are given. Scales were logarithmically transformed because of large interindividual differences: healthy subjects (open circles); asthmatic subjects (closed circles).

Reproducibility of the changes of other cell types than neutrophils were lower. Macrophages (Ri = 0.55, r = 0.54, p < 0.01) and lymphocytes (Ri = 0.47, r = 0.51, p < 0.01) showed a significant correlation 1 h after exposure to 250 ppb as well as 24 h later (macrophages: Ri = 0.46, r = 0.48, p < 0.01; lymphocytes: Ri = 0.48, r = 0.47, p = 0.01). No significant correlations were found for the percentages of eosinophils and the concentrations of ECP and albumin. However, changes in IL-8 concentration 24 h after exposures were highly reproducible (Ri = 0.72, r = 0.82, p < 0.001). The same was true 1 h after exposure to 250 ppb ozone if one subject, who responded with completely different values and in opposite directions in the two exposures, was excluded from the analysis (Ri = 0.54, r = 0.61, p < 0.01). When this subject was included, reproducibility was low (Ri = 0.27, r = 0.26, p = 0.17).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our study demonstrated an airway response to ozone, both in terms of lung function and sputum parameters, 1 h and 24 h after exposure to 250 ppb, whereas there were no or only minor effects after 125 ppb ozone. In addition, the individual response to ozone in terms of parameters of airway inflammation, in particular the number of neutrophils, was reproducible in independent exposures to 250 ppb ozone. Inflammatory parameters were not correlated with lung function responses, although these responses were also reproducible. Our observations suggest that the mechanisms of inflammatory and lung function responses to ozone are different and that the magnitude of both responses differs between individuals.

Studies using the techniques of bronchial wash and bronchoalveolar lavage have demonstrated that ozone primarily causes an influx of neutrophils and increased concentrations of prostaglandins such as PGE₂ and PGF₂, of cytokines and chemokines such as GM-CSF, IL-6, and IL-8, and of neuropeptides such as substance P (1, 5, 9, 17), as well as cellular damage and leakage as indicated by increased concentrations of total protein, LDH, albumin, and fibronectin (5). These techniques sample material from more peripheral airways as compared with the noninvasive method of induced sputum (3, 4, 15). However, since the central airways are also known to be affected by ozone (5, 6), induced sputum appears to be suitable for detection of ozone responses.

Indeed, two studies have been published that demonstrated the effect of ozone on sputum composition. Hiltermann and colleagues (8) exposed six healthy and six asthmatic subjects to 400 ppb ozone for 2 h of intermittent exercise [minute ventilation per body surface area, 20 L/(min · m²)], performed sputum induction 12 h later, after an additional methacholine challenge, and observed a significant increase in neutrophil numbers. Fahy and colleagues (7) studied 10 healthy subjects who were exposed to similar conditions [400 ppb, 2 h, 25 L/ (min · m²)], with sputum induction 4 h later. In addition to the increase in neutrophil numbers, the concentration of MPO, as a marker of neutrophil activation, was elevated by ozone. In a preliminary report using a lower concentration of ozone (260 ppb, 2 h) and including 14 subjects with asthma, again increased percentages of neutrophils in sputum occurred 6 h after exposure (20). Therefore, sputum induction is capable of monitoring inflammatory airway responses to ozone.

In accordance with these studies, we observed an influx of neutrophils into the airways, both 1 h and 24 h after exposure to 250 ppb ozone during 3 h of intermittent exercise. Our findings indicate a fast initial response and a persistent inflammation at the time when lung function was already back to normal, in accordance with the results of lavage studies (21). In our study, the increase in the percentage of neutrophils after exposure to 125 ppb ozone in asthmatic subjects was statistically significant when exposure and both time points were considered as within-factors in a repeated-measures ANOVA. Although this effect was small and questionable with respect to its clinical significance, the higher sensitivity of asthmatic subjects is in accordance with previous observations (22, 23).

The exposure level in our study was lower than that in previous studies using induced sputum (7, 8) because of the lower ventilation rate and lower ozone concentrations, and it might be noteworthy that a cellular response to inhalation of 125 ppb ozone, after mild to moderate exercise for a total of only 1.5 h, has not been detected before using lavage techniques. It is unlikely that the mouthpiece exposure resulted in much higher delivered doses of ozone than those given in chamber exposures at the same concentration because the major amount of ozone was inhaled during exercise, with increased ventilation rates, where it is natural to breathe through the mouth.

Ozone exposure did not alter the fall in neutrophil numbers that occurred over the three consecutive sputum induction periods. As previously described (15), this fall was more pronounced in healthy than in asthmatic subjects. The fact that the percentage of neutrophils was lower in asthmatic than in healthy subjects has also been observed previously (7, 15, 16, 24). In addition, these results are in accordance with data from proximal airway lavages (17, 23).

The percentage of neutrophils was increased 24 h as compared with 1 h after exposure to filtered air. Using the identical method of sputum induction, we have shown that this effect also occurred when two inductions were performed 24 h apart, without exercise in between (24). The magnitude of the difference in neutrophil numbers was similar both with and without exercise. These data suggest that the first sputum induction, but not exercise, was responsible for the increase in neutrophil numbers and ECP levels. Similar findings have been described for shorter intervals of time between inductions (25). Because of this, we always compared the data obtained after ozone exposure with the data after filtered air exposure at the corresponding time points 1 h or 24 h. The significant correlation between the percentages of neutrophils measured 1 h and 24 h after exposure to 250 ppb ozone (r = 0.45, p < 0.05) suggests that the effect of the sputum induction was homogeneous throughout subjects.

Our data may be compared with data using proximal airway lavages in healthy and asthmatic subjects (17, 23). These studies demonstrated similar percentages of neutrophils in lavage samples as known for induced sputum, suggesting the same airway compartments are sampled by the two techniques. The percentages of macrophages were inversely related to those of neutrophils, and their increased number during consecutive induction periods is in accordance with the interpretation that consecutive sputum samples are derived from increasing depths of the airways (15, 26), although it has been suggested that the decrease in the percentage of macrophages after exposure to ozone could be due to changes in cell adherence (21).

The subjects with mild asthma showed higher percentages of eosinophils in induced sputum than did the healthy subjects. In asthmatic subjects, the proportion of eosinophils and their cell numbers per milliliter were not affected by ozone; there was a small but significant increase in the number of eosinophils per milliliter sputum in healthy subjects. Although the clinical significance of this finding is questionable, it might explain the increased concentrations of ECP after ozone. The fact that the effects of ozone on eosinophil numbers were small or absent is in line with other studies using induced sputum or lavage (8, 21).

Until now, an ozone-induced increase in the proportion of lymphocytes has been reported only for studies employing bronchial and bronchoalveolar lavage (21, 27). The effect was most pronounced in the bronchial sample 18 h after ozone exposure, corresponding to our observation at 24 h. However, the effect was small and only statistically significant in healthy subjects and when all subjects were analyzed together. Previous studies using induced sputum did not observe a significant effect of ozone on lymphocyte numbers, possibly because of lower numbers of subjects studied or because of a different method of sputum induction or analysis.

We found higher concentrations of ECP in the sputum of asthmatic subjects in parallel with the higher number of eosinophils. The relatively high concentrations of ECP in healthy subjects are difficult to explain, but the increase after exposure to ozone could be related to the increase in the number of eosinophils, despite the fact that the numbers of these cells were very low. Possibly a different degree of eosinophil activation or different sources of ECP besides sputum eosinophils were responsible for the high ECP levels in healthy subjects. In either case, the observation that the number of eosinophils per milliliter was significantly correlated with the concentration of ECP lends confidence to the measured values.

In accordance with lavage studies (5, 21), ozone caused a significant increase in the concentration of IL-8, 1 h as well as 24 h after exposure. Because of the method of selecting sputum plugs from saliva, we found much higher concentrations of IL-8 than a previous study (7), and this might have resulted in an improved power. The concentration of IL-8 correlated with the number of neutrophils per milliliter sputum. As IL-8 is a major chemokine to attract neutrophils, this is consistent with results obtained before (21). Furthermore, it has been shown that the IL-8 concentration measured 5 min after an allergen challenge correlated with the number of neutrophils detected 4 h after challenge (28). We could also observe a statistically significant (r = 0.50, p < 0.01) correlation between the concentration of IL-8 1 h after the exposure and the number of neutrophils 24 h after exposure to 250 ppb ozone. This might suggest that IL-8 increased rapidly after exposure and was responsible for the influx of neutrophils after ozone, the main source probably being epithelial cells and macrophages (29). The increased concentrations of IL-8 observed 24 h as compared with 1 h after exposure to filtered air suggest an effect of the sputum induction 1 h after exposure similar to its effect on ECP concentrations (24).

The ozone-induced increase in the concentration of albumin has been described previously (5) and indicates cellular damage and epithelial or vascular leakage and increased permeability. Although it has been suggested (30) that increased permeability leads to stimulation of sensory nerves and release of neuropeptides such as substance P, which are linked to lung function responses, it has been proposed by others (19) that increased substance P concentrations, which are caused by ozone-induced reduction of neutral endopeptidase activity, are responsible for the observed permeability.

Because we performed a separate analysis of the sputum samples obtained in the three consecutive periods of saline inhalation, the amounts of sputum available were smaller than in conventional sputum analysis (2); this may have contributed to the fact that the material was insufficient in 6.8% of samples. However, all subjects produced at least one of three sufficient samples during a single sputum induction.

The individual changes in symptom scores, the range of interindividual differences, and the fact that there were no correlations with lung function responses are compatible with findings of a previous study using a similar ozone exposure (14). In addition, the result that ozone-induced decreases in FEV₁ were reproducible is in line with previous results (10, 14), suggesting that the magnitude of the acute lung function response to ozone is an intrinsic characteristic that shows a wide interindividual variation, resulting in responders and non-responders to ozone in terms of lung function responses (10, 17).

In addition to lung function, we demonstrated that the inflammatory response to ozone, particularly the influx of neutrophils and the increase in the concentration of IL-8, was reproducible within independent tests, suggesting that persons are characterized by their different intrinsic susceptibility to the inflammatory effects of ozone. The highest reproducibility was observed 1 h after exposure to 250 ppb ozone, but it was still significant 24 h later. Furthermore, we could show the correlation for both healthy subjects and asthmatics separately and for the sequential sputum samples. One possible mechanism could be genetic predisposition, as would be suggested, e.g., by the genetically controlled different susceptibility to ozone-induced inflammation in inbred mice (31). In accordance with most findings using lavage techniques, inflammatory changes and changes in lung function were not correlated with each other (5, 17, 32). Therefore, the intrinsic susceptibilities to lung function responses and to inflammatory responses appear to represent two independent individual characteristics. The mechanisms underlying the interindividual differences in the response to ozone are unknown.

In summary, our data demonstrate dose-dependent airway responses to ozone in healthy and asthmatic subjects using the method of induced sputum. The inflammatory and the lung function response were both reproducible but not related to each other. Therefore, they appear to represent two independent factors that differ between individuals.