INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Determining individual susceptibility to the health effects of air pollution is of importance in developing protective strategies against such effects. Exposure to ozone at or below the current National Ambient Air Quality Standard of 0.12 ppm causes symptoms, changes in lung function, and airway inflammation (1). Marked differences exist in individual susceptibility to ozone effects on lung function, but individuals are consistent in their responses to subsequent ozone exposures (2). Changes in lung function with exposure to ozone appear to be primarily a reduction in inspiratory capacity, possibly attributable to neurally mediated involuntary inhibition of inspiration (3).

Ozone exposure also causes cellular and biochemical changes in the upper and lower respiratory tract that are characteristic of an acute inflammatory response. Studies in humans, using bronchoalveolar lavage (BAL), have shown ozone-induced increases in polymorphonuclear leukocytes (PMN), soluble markers of inflammation and repair, and markers of epithelial permeability (4). All of these markers are increased at 18 h after exposure to 0.4 ppm ozone, a time when the changes in lung function induced by such exposure have usually resolved. Nasal lavage (NL) has revealed inflammation and changes in permeability in the upper airway (10). In contrast to the extensive data base on responses of pulmonary function to ozone, little information is available about susceptibility to the airway inflammatory effects of ozone exposure.

Recent evidence suggests that the intensity of the inflammatory response at 18 h after exposure is not correlated with the degree of functional decrement immediately after exposure (11). However, sampling at 18 h may miss the peak inflammatory response in some subjects. For example, Koren and colleagues (7), comparing two separate studies of the effects of exposure to 0.40 ppm ozone for 2 h, found that the influx of PMN was 2.3-fold greater at 1 h after exposure than at 18 h after exposure. Thus, the relationship between changes in lung function and the development of airway inflammation following exposure to ozone has not been clearly established.

Airway inflammation could result from ozone-induced stimulation of neuropeptide-containing nerves within the airways; increased levels of substance P have been found in the BAL fluid (BALF) of humans immediately after 1-h exposure to 0.25 ppm ozone (12). If changes in lung function and inflammation are both mediated via intraepithelial nerve stimulation, the intensity of inflammation should correlate with lung-function responsiveness. An alternative mechanism of airway inflammation is through the release of chemotactic cytokines such as interluekin-6 (IL-6) and IL-8 from airway epithelial cells through nonneurogenic pathways, such as activation of the nuclear regulatory element nuclear factor-B (NF-B) (13). If the mechanisms mediating changes in lung function and airway inflammation are unrelated, individuals who do not experience symptoms or changes in lung function might be at greatest risk for airway inflammation from failure to avoid exposure or limit ventilation.

The primary purpose of the present study was to determine whether individuals who differ in lung-function responsiveness to ozone also differ in susceptibility to airway inflammation. We chose to study both nonsmokers and smokers, because smokers may experience a greater mortality attributable to air pollution (14) yet have decreased lung-function responsiveness to ozone in comparison with nonsmokers (15). A secondary goal was to determine whether smokers differ from responder and nonresponder nonsmokers in susceptibility to ozone-induced airway inflammation. We evaluated airway inflammatory cells and proinflammatory cytokines immediately after and 18 h after exposure for each subject, in order to determine the time course of the response.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Volunteers aged 18 to 40 yr were recruited from the local community through bulletin-board and newspaper advertisements; most were college students. Informed written consent was obtained from each subject, and the study was approved by the Research Subjects Review Board of the University of Rochester. Briefly, subjects were free of cardiorespiratory disease by history and physical examination, and had normal spirometry and sufficient physical conditioning to complete the exercise protocol of the study. None of the subjects had had symptoms of respiratory infection within the 3 wk before the beginning of the study. Nonsmokers denied any tobacco use in the 3 yr prior to the study, with lifetime tobacco use of less than 1 pack-yr. Smokers were currently smoking at least 1 pack per day, with a smoking history of at least 3 pack-yr.

Study Design

The overall experimental protocol is summarized in Figure 1. Subjects were selected for the study on the basis of lung-function responsiveness to a 4-h exposure to 0.22 ppm ozone with intermittent exercise. The methods for subject screening and the results of this initial classification exposure have been reported (15). For the purposes of the study, ozone "responders" and "nonresponders" were selected on the basis of a decrement (> 15%) or lack of a decrement (< 5%) in FEV₁, respectively. Individuals whose FEV₁ decreased by 5% to 15% were not invited to participate in the study.

View larger version (39K): [in this window] [in a new window]   Figure 1.   Experimental plan. Healthy smokers and nonsmokers were selected on the basis of responsiveness to 0.22 ppm ozone for 4 h with exercise. Responders with a decrease in FEV1 > 15% and nonresponders with a decrease in FEV1 < 5% were included in the study, but subjects with intermediate responsiveness were excluded. Although the initial plan was to study four groups, there were not enough smoker-responders identified, and thus smokers were studied as a single group. Each subject subsequently underwent two exposures to ozone and one to air, separated from each other by at least 3 wk. Exposures were double blinded and of randomized order. BAL and nasal lavage (NL) were performed on three occasions: immediately (early) and 18 h (late) after ozone, and either early or late after air exposure.

The original objective was to recruit four groups of 10 subjects each: nonsmoker-nonresponders, nonsmoker-responders, smoker- nonresponders, and smoker-responders. A total of 90 subjects (56 nonsmokers and 34 smokers) were classified for ozone responsiveness; 16 nonsmokers were responders and 22 were nonresponders. Of the smokers, four were responders. Because of the low rate of ozone responsiveness among the smokers (15), we were unable to recruit a sufficient number of smoker-responders to study as a separate group; therefore, smokers were considered as a single group.

A total of 60 subjects (16 nonsmoker-nonresponders, 16 nonsmoker-responders, 24 smoker-nonresponders, and four smoker-responders) were invited to participate in the BAL study. Nine subjects (two nonsmoker-nonresponders, two nonsmoker-responders, and five smoker-nonresponders) refused to participate; nine (one nonsmoker- nonresponder, one nonsmoker-responders, six smoker-nonresponders, and one smoker-responder) were unavailable; one (a nonsmoker-nonresponder) became pregnant; and one (a smoker-nonresponder) decided to quit smoking. Thus, a total of 40 subjects (12 nonsmoker-nonresponders, 13 nonsmoker-responders, 12 smoker-nonresponders, and three smoker-responders) entered the study. No subject declined participation because of ozone-related symptoms or responsiveness.

Each subject underwent a total of three exposures (two ozone and one air exposure) and three BAL and NL procedures, with each exposure-BAL sequence involving an interval of at least 3 wk between the two events. Subjects were exposed once to air and twice to ozone, with NL and BAL performed immediately after one of the ozone exposures (referred to subsequently as "ozone early") and 18 h after the other ozone exposure ("ozone late"). Since each subject had only one exposure to filtered air, half of the air exposures involved lavage directly after exposure and the other half involved lavage at 18 h after exposure. Subjects were exposed in pairs, and were randomized so that one subject was scheduled to undergo BAL immediately after exposure and the other at 18 h after exposure. A restricted randomization scheme was used in order to balance the various treatment assignments over time. Pairs of subjects were assigned to all three treatments (air, ozone early, and ozone late) through a single randomization.

All ozone exposures were to 0.22 ppm ozone for 4 h, with exercise for 20 of every 30 min during the 4-h exposure period at a level sufficient to achieve a E of approximately 25 L/min/m² body surface area (BSA). During the BAL phase of the study, exposures were double-blinded and conducted in randomized order. Smokers were not permitted to smoke during exposure, but were not advised to abstain from smoking prior to exposure. Pulmonary-function and symptom responses to ozone exposure during the study have been reported (15).

Exposure Facilities

All exposures were undertaken in a 45-m³ environmental chamber, the characteristics of which have been described, in the General Clinical Research Center at the University of Rochester (16). Exercise bicycle ergometers and pulmonary-function-testing equipment are housed within the chamber, so that subjects were not required to exit the chamber for physiologic testing. For comfort, temperature and relative humidity were maintained at 21.2 ± 0.9 (mean ± SD) °C and 37.1 ± 3.0%, respectively.

All ozone exposures were conducted at a target concentration of 0.22 ppm (430 µg/m³). Ozone generation was accomplished by passing Breathing Quality oxygen through a flow meter into a water-cooled, high-voltage discharge ozonator (Model 03V5; Ozone Research and Equipment Corp., Phoenix, AZ). The ozonator output was connected to the chamber air intake (10 m³/min) through a Venturi mixer.

An ozone analyzer (Model 8810; Monitor Labs Inc., Englewood, CO) continuously sampled the ozone concentration in the chamber atmosphere through a Teflon tubing connection. By means of feedback circuitry, the analyzer regulated the ozonator output. An ozone analyzer (Model 1003-AH; Dasibi Environmental Corp., Glendale CA) that has been designated as a U.S. Environmental Protection Administration (EPA) Transfer Standard was used to calibrate the Monitor analyzer. Before each ozone exposure, the calibration procedure required that the output of the Monitor analyzer be compared with that of the Transfer Standard while both instruments were sampling the identical ozone concentration produced by a portable ozone generator (Stable Ozone Generator, Model SOG-2; Ultraviolet Products Inc., San Gabriel, CA). The output of the ultraviolet (UV) ozone generator, in turn, was validated at least bimonthly against a certified ozone standard at the Air Quality Control Station of the New York State Department of Environmental Conservation in Avon, NY.

Air exposures, and the diluent air for the ozone exposures, were done with environmental air passed through an air intake purification system (16). The quality of the purified air with regard to background ozone, nitrogen oxide, and sulfur oxide concentrations was established through use of the Dasibi Model 8810 Ozone Analyzer, an NOx analyzer (Model 8840; Monitor Labs Inc.), and a Meloy SO₂ analyzer (Model SA285E; Columbia Scientific Instruments, Jollyville, TX), respectively. Background levels of air pollutants in the intake air of the chamber were at or below detection levels with respect to particles, sulfur oxides, nitrogen oxides, and ozone, at less than 4 µg of particles/ m³, approximately 0.01 ppm NO₂, and less than 0.005 ppm for O₃ and sulfur dioxide, respectively.

Physiologic Testing

Methods for spirometry (FVC, FEV₁, and FEF_25-75%) and determination of thoracic gas volume and specific airway conductance have been reported previously (15). Minute ventilation (E) was measured initially at rest and during exercise, using inductive plethysmography (Respigraph Model PN SY01; Noninvasive Monitoring Systems, Miami Beach, FL) calibrated with a rolling seal spirometer (Model 840; Ohio Medical Products, Houston, TX).

Methacholine Challenge

Airway challenge with methacholine was performed at the time of subject screening, as reported previously (15). In brief, increasing concentrations (up to 40 mg/ml) of methacholine in normal saline were administered at 4-min intervals through a nebulizer (Model 646; Devilbiss Co., Somerset, PA) with a dosimeter (Rosenthal-French Model D-2A; Laboratory for Applied Immunology Inc., Fairfax, VA) calibrated to deliver 0.01 ml per breath. Challenge was stopped if specific airway conductance (SGaw) decreased by more than 50% from the baseline value. The concentration of methacholine that produced a 50% decrease in SGaw (PD₅₀) was determined by interpolation, using the regression line of the methacholine dose response.

BAL

Subjects were premedicated with 0.75 to 1.0 mg intravenous atropine, and topical anesthesia of the upper and lower airways was established with lidocaine spray. A fiberoptic bronchoscope (FB-19H, O.D. = 6.3 mm; Pentax, Orangeburg, NY) was passed orally, and was gently wedged in a subsegmental airway of the inferior segment of the lingula. Four 50-ml aliquots of sterile normal saline were sequentially instilled and immediately withdrawn under gentle suction, with the washings collected into a siliconized Erlenmeyer flask on ice. The lavage was then repeated in the right middle lobe. The same lingular and middle-lobe subsegments were entered during each subject's three bronchoscopies. After completion of lavage on the first four subjects, the BAL protocol was modified so that the return from the first 50-ml aliquot instilled into the lingula was identified as a "bronchial lavage" (BL) sample and counted separately, since this portion of the lavage has been shown to be more representative of changes in the proximal airways (17). The returns from the remaining three 50-ml aliquots were pooled as the alveolar lavage (AL) sample.

Nasal Lavage

In order to determine whether the nasal inflammatory response correlates with airway inflammatory responses in the categories of subjects in the groups studied, NL was performed just prior to BAL following each exposure. Five milliliters of warmed normal saline was instilled into each nostril with a syringe, held for 10 s, and discharged into a sterile container. The procedure was then repeated, and the discharged fluids were pooled, placed on ice, and transported to the laboratory for total and differential cell counts.

Cell Quantitation and Characterization

Analysis of cells recovered by BAL was designed to detect influx of inflammatory cells or changes in the distribution of alveolar cell subpopulations in response to the exposure. For the BL and AL samples, total cell counts and viability determinations were performed separately, with a hemocytometer and trypan blue dye exclusion, respectively; the BL and AL samples were then combined into a single BAL lingular sample. Total cell counts and viability were determined for the lingular and right-middle-lobe BAL samples. Cytospin slides (Shandon Inc., Pittsburgh, PA) were prepared for each of the BAL, BL, and AL samples from aliquots of sufficient volume to contain 5 × 10⁴ cells. Slides were stained with Diff-Quick (American Scientific Products, McGraw Park, IL) for differential counts; at least 500 cells from each slide were counted. Total and differential cells counts for BAL were expressed as averages of the results from the lingula and right-middle-lobe samples. To evaluate effects on mast cells, a separate slide of cells from the right middle lobe (BAL) was stained with Mayer's hematoxylin and toluidine blue for enumeration of metachromatic cells.

T-lymphocytes recovered through BAL were quantitated with immunocytochemical staining of cytospin slides that had been air dried and stored at 70° C. All three slides from the same subject were stained simultaneously and counted without knowledge of the exposure. After thawing and fixing with acetone, 0.3% H₂O₂ in methanol was used to quench native peroxidase activity. Horse serum at a dilution of 1:20 in phosphate-buffered saline-fetal bovine serum (PBS- FBS) was then used as a blocking agent. The primary and secondary antibodies were monoclonal mouse anti-CD3 (Leu-4; Becton Dickinson, San Jose, CA) and biotinylated horse antimouse IgG (Vector Laboratories, Burlingame, CA), respectively. Freshly prepared steptavidin peroxidase was added and then developed, using an aminoethyl carbazole substrate kit (Zymed Laboratories, South San Francisco, CA). Counterstaining was performed with Meyer's hematoxylin, and CD3⁺ cells were expressed as cells/ml BALF.

Measurements of Cytokines in BALF

IL-6 and IL-8 were measured to determine their potential role in recruitment of inflammatory cells following ozone exposure.

BALFs were stored at 70° C prior to analysis of IL-6 and IL-8; determinations were done simultaneously on all samples from each subject. Immunoassays were validated for BALF through the use of serial dilutions and add-back of purified antigen to confirm accurate recovery.

IL-6 and IL-8 were determined in unconcentrated BALF, using commercially obtained immunoassay kits (R&D Systems, Minneapolis, MN). Samples were assayed in duplicate and read on a microtiter plate reader (Model EL312; Bio-Tek Instruments, Winooski, VT).

Data Handling and Statistical Methods

On the basis of comparative estimates of the total intake of ozone and magnitude of the inflammatory response in a previous study (6), we expected a roughly 6-fold increase in PMN following ozone exposure of our subjects. A sample-size computation of this effect indicted that eight subjects per group would be required to achieve 80% power at the 5% level of significance. A group size of 12 was selected to provide a sufficient margin of statistical power.

Primary data analyses were done with SAS (SAS Institute Inc., Cary, NC). Data from smokers and nonsmokers under three conditions (air exposure, ozone early, and ozone late) were analyzed through repeated-measures analysis of variance (ANOVA).

The primary analyses for the data from BAL were based on a two-way mixed-model or repeated-measures ANOVA. In the language of repeated measures ANOVA, there was one intersubject factor and two intrasubject factors. The intersubject effect was subject group. The primary intrasubject effect in the model was treatment (air, ozone early, and ozone late; three conditions). For these analyses, the air exposure was considered as a single treatment, regardless of whether subjects were lavaged early or late. This assumption was checked in a separate two-way ANOVA comparing air-early with air-late. For any variable for which air-early was different from air-late, the ANOVA was run separately for the two groups of subjects (air-early versus air-late).

The second intrasubject factor was a period effect. Because subjects were studied at different times of the year, period effects were not expected, and the effect was included as a check on this assumption. As in any repeated-measures ANOVA, a random-subject effect was also included, and was nested within groups. The analysis also included a test of interaction of the treatment effect with subject group. If this interaction was significant, we concluded that differences among the three treatments were different among the three groups of subjects. If the interaction was not significant, we examined the individual or main effects for statistical significance. No terms were included in the model for carryover effects because of the relatively long interval between repeated measurements on the same subject.

In cases in which examination of residuals indicated that the variance was not constant, log transformation was performed. In cases in which the data contained more than 10% zeros (e.g., BAL eosinophil counts), paired and unpaired Wilcoxon's tests or t tests were used. For a small number of endpoints, analysis of covariance (ANCOVA) was performed. A significance level of 5% was required for statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For clarity of presentation, significant values of p found in statistical comparison testing are given in the tables and figure legends.

Subjects Characteristics

A total of 40 subjects were randomized. One smoker-nonresponder withdrew and one smoker-responder did not complete the protocol because of problems with intravenous access. Table 1 shows the characteristics of the 38 subjects who completed the study. Smokers were considered as a single group, and included two ozone responders. Groups did not differ except in smoking status.

Exposure Data

During the ozone exposure sessions, the mean ozone concentration was 0.219 ± 0.005 ppm. E during exposure was similar for nonsmokers and smokers, responders and nonresponders. During exercise, E ranged from 39.4 to 45.3 L/min in nonsmokers, and from 42.6 to 45.6 L/min in smokers. Ambient outdoor ozone levels during the study were obtained from the New York State Department of Environmental Conservation, and did not exceed 0.10 ppm within the 24 h prior to any exposure session.

Pulmonary-function changes and symptoms following ozone exposure in the study subjects have been reported previously (15). Within the classification of exposure to ozone, the mean decrease in FEV₁ ± SD was 5.0 ± 13.9% for smokers, 1.4 ± 2.4% for nonresponders, and 28.5 ± 11.8% for responders. Responsiveness (or lack of responsiveness) to ozone remained consistent in smokers as well as in nonsmokers. There were no significant decrements in lung function following air exposure in any group.

Airway Inflammation

BALF and cell recovery. There were no significant effects of ozone exposure on the recovery of lavage fluid from either the BL or AL components of the lavage. Table 2 shows the recovery of fluid and cells from the AL and BL components of BAL, respectively. Fluid recovery from smokers was significantly smaller for AL but not for BL.

Alveolar-cell recovery after air exposure was 2-fold greater in smokers than in nonsmokers. In all groups, cell recovery decreased early after ozone exposure and increased at 18 h. Cell recovery at 18 h after ozone exposure remained slightly decreased relative to air exposure in smokers, but was greater than for air-exposure values in nonsmokers. When total BAL cell recovery was analyzed, including recovery from the four subjects from whom a separate BL was not collected, smokers differed significantly from the other two groups in the effects of ozone on cell recovery (p = 0.026). The pattern of response in nonsmoker-nonresponders and responders was similar.

Viability of recovered cells was generally above 90%, but was decreased by approximately 2% in BL early after ozone exposure in all groups (p = 0.026).

Differential cell recovery. Polymorphonuclear leukocytes showed the largest change in response to ozone exposure in all groups. As shown in Figure 2, recovery of PMN increased progressively following ozone exposure in both BL and AL, and was greatest at 18 h after exposure. For PMN recovered in AL, the ozone response did not differ significantly among groups. The concentration of PMN in BAL (BL and AL not separated) was slightly greater for smokers than for nonsmokers after air and early after ozone exposure. However, the concentration of PMN in BAL at 18 h after ozone exposure was similar among all groups. The PMN concentration increased similarly both early and late after ozone exposure in both the responder- and nonresponder-nonsmokers.

View larger version (35K): [in this window] [in a new window]   Figure 2.   Cell recovery from alveolar lavage (AL) and bronchial lavage (BL) fractions. Alveolar macrophages (AM) were increased in smokers as compared with nonsmokers in AL (group effect, p < 0.0001), but not in BL fractions and decreased early after ozone in both fractions (ozone effect, p < 0.0001). Polymorphonuclear leukocytes (PMN) increased progressively after ozone exposure in both lavage fractions for all groups (ozone effect, p < 0.0001). Lymphocytes were decreased in smokers as compared with nonsmokers in both AL (group effect, p < 0.0001) and BL (p = 0.0007) fractions. They also increased late after ozone in all groups in AL (ozone effect, p < 0.0001) and BL (p = 0.0096) fractions. Eosinophils (EOS) were increased in smokers as compared with nonsmokers in both AL (Wilcoxon Rank Sum, p = 0.0015) and BL (p = 0.011) fractions. They also increased late after ozone in the BL fraction for all nonsmokers (paired t test, p = 0.0098). Open bars: air exposure; crosshatched bars: ozone early; solid bars: ozone late. Data are expressed as means ± SE.

Recovery of AM in BAL and AL was greater in smokers than in nonsmokers. Recovery of AM in BL was similar among groups. In all groups, AM recovery decreased early after ozone exposure and returned to baseline at 18 h. There were no significant differences among groups in the effects of ozone on AM recovery in any lavage component.

Lymphocyte recovery was increased at 18 h after ozone exposure in all lavage components in all groups. There were no significant differences between groups in the effect of ozone on lymphocyte recovery, although fewer lymphocytes were recovered from smokers. Immunohistochemical staining of cytospin slides with monoclonal anti-CD3 antibody confirmed a more than 2-fold increase in BAL T lymphocytes at 18 h after ozone exposure in nonsmokers (Figure 3).

View larger version (35K): [in this window] [in a new window]   Figure 3.   Epithelial cells, metachromatic cells, and T lymphocytes. (A) Concentration of epithelial cells in bronchial lavage (BL). (B) Epithelial cells as a percentage of total cells recovered from BL. The percentage, but not concentration of epithelial cells, increased early after ozone exposure in all groups (ozone effect, p = 0.0002). (C ) Total number of metachromatic cells per cytospin slide prepared from equal numbers of cells obtained from BAL and stained with Mayer's hematoxylin and toluidine blue. Stained cells increased early after ozone in all groups (ozone effect, p = 0.045). (D) Concentration of cells staining with anti-CD3 is increased late after ozone exposure for nonresponders (p = 0.008) and responders (p = 0.02), as compared with air exposure. Open bars: air exposure; crosshatched bars: ozone early; solid bars: ozone late. Data are expressed as means ± SE.

Eosinphils were more numerous in lavage fluids from smokers than from nonsmokers after air exposure (Figure 2). Eosinophil concentration in BL was increased at 18 h after ozone exposure in all groups, but the increase was significant only for nonsmokers considered as a single group (paired t test, p = 0.0098).

Metachromatic cells in BAL increased early after ozone exposure in all groups (Figure 3). A further increase was seen at 18 h in nonresponders, but not in responders or smokers. ANOVA indicated that the main effect of ozone was marginally significant (p = 0.045), with no significant differences between groups.

Epithelial-cell recovery in BL was assessed as an indicator of airway epithelial injury. Figure 3 shows both percent and concentration of epithelial cells in BL. There was no significant change in BL epithelial-cell concentration following ozone exposure. The percentage of epithelial cells increased early after ozone exposure in all groups, possibly because of the decrease in recovery of AM and total cells at this time point (see Table 2).

Inflammatory cytokines. Levels of the inflammatory cytokines IL-6 and IL-8 were measured in BL and AL components of lavage fluid in order to examine their potential role in modulating the inflammatory response following ozone exposure. Both IL-6 and IL-8 (Figure 4) increased early after ozone exposure and decreased toward baseline at 18 h, and similar effects were seen in BL and AL. The largest change was in IL-6, which increased by more than 10-fold in nonsmokers. The increase was greater in nonsmokers than in smokers in both AL and BL. For IL-8, the response to ozone was not significantly different between groups.

View larger version (39K): [in this window] [in a new window]   Figure 4.   Concentrations of IL-6 and IL-8 in alveolar lavage (AL) and bronchial lavage (BL) fractions. IL-6 increased early and returned to baseline late in both AL and BL fractions (ozone effect, p < 0.0001). The effect of ozone was greater in nonsmokers than in smokers in both AL and BL fractions (interaction between group and ozone effects, p = 0.024 and p = 0.0002, respectively). IL-8 increased early in all groups in both AL and BL fractions (ozone effect, p < 0.0001). Open bars: air exposure; crosshatched bars: ozone early; solid bars: ozone late. Data are expressed as means ± SE.

To examine the possible role of IL-6 and IL-8 in the recruitment of PMN following ozone exposure, ANCOVA was performed, using changes in these inflammatory cytokines in AL early after ozone exposure as independent variables, and changes in BAL PMN counts at 18 h after ozone as the dependent variable. Early increases in both IL-6 and IL-8 correlated significantly with the late increase in PMN (p = 0.002 and p = 0.007, respectively). There was a marginally significant difference between groups for the IL-8 association, (p = 0.046), but not for IL-6. These findings support the hypothesis that both IL-6 and IL-8 may play a role in recruitment of PMN following ozone experience.

Predictors of inflammation. In order to determine predictors of susceptibility to airway inflammation in response to ozone exposure, ANCOVA was performed with the change in BAL PMN count at 18 h after ozone exposure as the dependent variable and the following as independent variables: age, gender, PD₅₀, history of allergies, and subject group. Although subject group was marginally predictive of increase in the PMN with this model (p = 0.045), this statistical finding is probably a reflection of the higher PMN concentration in smokers after air exposure, and not due to differences in PMN after ozone exposure. Thus, this statistical finding does not necessarily represent a true group difference in the PMN response to ozone. With control for other factors, including subject group, age was inversely correlated with the inflammatory response (p = 0.013). Gender, PD₅₀, and allergy history were not predictive of ozone-induced increase in PMN.

Respiratory symptoms during exposure were not predictive of the intensity of airway inflammation. There was no significant correlation between the increase in BAL PMN count at 18 h after ozone exposure and the increase in cough (r = 0.14, p = 0.42) or shortness of breath (r = 0.01, p = 0.94).

Nasal Lavage

Cell recovery through NL proved variable under all exposure conditions. Although some subjects showed increases in PMN recovery in association with ozone exposure, other subjects had high numbers of PMN recovered after air exposure. Mean PMN recovery varied according to the timing of NL, and was independent of ozone exposure: mean PMN recovery was greater when NL was performed 18 h after air exposure than early after air exposure (Figure 5). NL revealed no statistically significant differences between groups or significant effects of ozone exposure on PMN recovery (in concentration or percent of cells). There was also no significant correlation between PMN concentration or percentage in NL and BAL samples.

View larger version (26K): [in this window] [in a new window]   Figure 5.   Recovery of polymorphonuclear leukocytes (PMN) in nasal lavage (NL). In some subjects, large numbers of PMN were recovered in association with air exposure, and PMN counts were greater at 18 h after than immediately after air exposure, suggesting that exposure and exercise caused temporary clearing of the nasal passages. There were no statistically significant group differences or ozone effects. Open bars: early after air exposure; stippled bars: late after air exposure; crosshatched bars: ozone early; solid bars: ozone late. Numbers of subjects are indicated above each bar. Data are expressed as means ± SE.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study confirms previous observations (4) that healthy, nonsmoking, exercising subjects develop an airway inflammatory response to ozone exposure at an ozone concentration relevant to that on summer days in a major city such as Los Angeles. The magnitude of PMN influx in our study at 18 h after ozone exposure was similar to that observed in previous studies; for example, Koren and colleagues (6) observed an approximately 8-fold increase in PMN at 18 h after exposure to 0.4 ppm ozone for 2 h with exercise. The pattern of cellular and cytokine responses in BL and AL components of lavage were generally similar, suggesting that the inflammatory response to ozone exposure involves both distal conducting airways and the alveolar space.

Our observations differ those in previous reports in two respects. First, we observed a greater influx of PMN at 18 h after ozone exposure than immediately after exposure, in contrast to previous studies (7) reporting a greater PMN influx at 1 h after exposure to 0.4 ppm ozone for 2 h. The previous reports involved separate studies on different subjects, and the findings may have been influenced by group differences. Thus, the inflammatory response to short-term ozone exposure appears to be increasing or persisting at time when symptoms and decrements in lung function have largely resolved.

Second, in contrast with previous BAL studies, our data suggest that the airway cellular response to ozone is not limited to PMN, but involves virtually all cells recovered through BAL. We observed increases in lymphocytes, eosinophils (in nonsmokers), and mast cells in association with ozone exposure. The increase in lymphocytes was confirmed through immunohistochemical staining with a monoclonal anti-CD3 antibody. These cells are important in the pathogenesis of asthma, and recent studies suggest that the airway inflammatory response to ozone may be altered or enhanced in asthmatic as compared with healthy subjects (18, 19).

We observed small decreases in recovery of AM at the early time point following ozone exposure, which could represent direct toxicity of ozone or induced programmed cell death. Either possibility is supported by the small decrease in viability of BAL cells early after exposure. Alternatively, AM recovery could be reduced as a consequence of altered membrane- adhesive properties making cells less accessible by BAL, accelerated clearance of airway cells during ozone exposure, or other airway effects altering total cell recovery.

The primary purpose of this study was to determine whether individuals who differ in lung-function responsiveness to ozone also differ in susceptibility to airway inflammation. A secondary goal was to determine whether smokers differ from nonsmokers in their airway inflammatory response to ozone. Our study confirms and extends the finding by Balmes and coworkers (11) of a dissociation between the irritant and inflammatory changes induced by ozone exposure. In addition, our study shows that the inflammatory response to ozone is multicellular, and that it is surprisingly similar in smokers and nonsmokers, with a time course characterized by an early increase in proinflammatory cytokines followed by an influx of inflammatory cells into the proximal and distal airways.

These findings indicate that both smokers and nonsmokers may experience airway inflammation following exposure to ozone in the absence of symptoms or changes in lung function. To the extent that airway inflammation is considered an adverse health effect, this observation has implications for determining who is most susceptible to the health effects of ozone exposure.

IL-6 and IL-8 are involved in the recruitment, activation, and persistence of PMN and other cells at sites of inflammation (20). IL-8 plays a primary role in the recruitment and persistence of PMN. IL-6 is a pleiotropic cytokine that promotes the proliferation, activation, and differentiation of T lymphocytes and initiates the synthesis of acute-phase proteins by hepatocytes, among other effects. We observed increases in BAL levels of both IL-6 and IL-8 early after ozone exposure, with a return toward baseline at 18 h, when PMN and lymphocyte recovery was greatest. The time course of the cytokine response is therefore consistent with its having a contributing role in the recruitment of inflammatory cells. ANCOVA showed a significant correlation between increases in both IL-6 and IL-8 and increases in PMN when all subjects were considered together. IL-6 may have been responsible for the increase in lymphocytes, but may also play an indirect role in the recruitment of PMN. IL-6 increases expression of the nuclear regulatory protein nuclear factor interleukin-6 (NF-IL-6), and NF-IL-6-binding motifs have been identified in the functional regulatory region of the IL-8 gene (21). Consequently, it is possible that IL-6 contributes to the increased expression of IL-8 or other proinflammatory cytokines following ozone exposure.

Although the source of these cytokines in vivo remains unknown, bronchial epithelial cells release IL-8 and IL-6 in response to ozone exposure in vitro (22), suggesting that epithelial cells may be the source for the cell changes in BAL following ozone exposure. Supporting this hypothesis is that the levels of IL-6 and IL-8 were generally higher in the bronchial fraction of lavage than in the alveolar fraction. In our study, all subject groups showed similar patterns of cytokine responses, although smokers had smaller increases in IL-6 than did nonsmokers.

NL proved disappointing as a predictor of airway inflammation following ozone exposure. For some subjects, recovery of PMN appeared to be more dependent on the timing of NL than on ozone exposure (Figure 5), suggesting that exercise may have enhanced the clearance of nasal inflammatory cells prior to lavage. Our study did not specifically exclude subjects with allergic or vasomotor rhinitis, which may account in part for the variability in NL findings. In addition, mouth-breathing during exercise may have minimized exposure of the nasal mucosa. Our data indicate that findings based on NL do not reliably predict the lower-airway inflammatory response to ozone.

The findings in our study may not be representative of the response to ozone in all segments of the population. The subjects studied were young, healthy, motivated, and able to sustain exercise. Although no subject declined participation because of ozone-related symptoms, we cannot exclude the possibility of selection bias.

In our study, the prospective selection of subjects based on lung-function responsiveness to ozone, with sampling at two time points after exposure, proved to be a useful experimental design for comparing responders and nonresponders with regard to airway inflammation and injury. The striking similarity between responders and nonresponders in virtually all response indicators is worthy of note. It is clear from this and other studies that symptoms and spirometric effects do not provide sufficient bases for the assessment of risks associated with ozone exposure. Smokers and nonsmokers experience airway inflammation in response to acute ozone exposure, even in the absence of symptoms or spirometric changes. The cellular inflammatory response to ozone exposure includes cell types that have been implicated in the pathogenesis of chronic respiratory disease; recruitment of these cells to the air spaces may be mediated in part by the release of IL-6 and IL-8 from airway epithelial cells. Unfortunately, individual subject characteristics and noninvasive techniques such as NL are not helpful in determining who is most susceptible to the lower-airway inflammatory effects of ozone exposure.