INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Inflammation of the airways is a critical component of the pathogenesis of asthma. The mechanisms by which inflammation is controlled are therefore of considerable interest. Moreover, most agents effective in controlling asthma, such as inhaled corticosteroids and leukotriene modifiers, also have important anti-inflammatory effects. Segmental allergen challenge (SAC) is a model of airway inflammation which can be quantitated and assessed using bronchoalveolar lavage (BAL), and which is relevant to asthma. Considerable insight has been gained into the mechanisms of airway inflammation in asthma by use of these techniques.

The inflammatory response that follows SAC in allergic individuals can be modulated by factors that influence asthma severity. Such effects have been demonstrated with viruses (1), nedocromil (2), leukotriene antagonists (3), and inhaled corticosteroids (4). Thus, this model can be used to assess the effects of selected factors on airway inflammation in asthma.

The effects of ₂-receptor agonists on this process have not been fully examined. There are data which suggest ₂-agonists might have anti-inflammatory effects by interfering with activation of inflammatory cells. For example, histamine release by basophils and mast cells is inhibited in a concentration-dependent manner by ₂-agonists (5, 6). Functional activation of neutrophils for lysosomal enzyme release is also inhibited by -agonists in a concentration-dependent manner (7). In addition, salmeterol specifically was shown to reduce intracellular calcium influx, production of oxidants, and platelet-activating factor (8). These observations would suggest that chronic administration of a short-acting ₂-agonist, or administration of a long-acting ₂-agonist such as salmeterol, might be associated with reduced inflammatory cell function, and therefore reduction or inhibition of airway inflammation. Finally, Eickelberg and colleagues have demonstrated translocation of corticosteroid receptors in cells treated with salmeterol, suggesting that long-acting -agonists might have effects on nuclear transcription that would be anti-inflammatory (9). This in vitro "Eickelberg effect," if present in vivo, would predict increased effect of inhaled steroids in patients treated with long-acting bronchodilators such as salmeterol.

However, there are also intriguing in vitro data to suggest that ₂-agonists may have proinflammatory activity. From an epidemiologic standpoint, the suggestion has been made that increasing asthma mortality might relate to increasing use of ₂-agonists (10). In addition, both short- and long-acting ₂-agonists may increase the slope of methacholine dose-response curves, which may lead to more precipitous asthma attacks (11). Although a mechanism for this relationship has not yet been convincingly demonstrated, there are data which suggest that ₂-agonists may interfere with the action of corticosteroids. These data have recently been reviewed (12). The binding of glucocorticoid receptor protein to DNA was reduced by albuterol and the long-acting ₂-agonist fenoterol (13). This mechanism, if relevant in asthma, might account for reduced effectiveness of inhaled corticosteroids in patients requiring repeated dosing of short-acting or long-acting ₂-agonists. The clinical relevance of this mechanism has not been established, although Wong and colleagues have suggested that terbutaline, when added to budesonide, eliminated the protection against allergen-induced bronchoconstriction afforded by budesonide alone (14). Collectively, the data on a proinflammatory or anti-inflammatory effect of long-acting ₂-agonists are sufficiently equivocal that additional empiric evidence is required.

Finally, there are a number of studies which suggest that salmeterol, of itself, has little important effect on airway inflammation (15, 16). In a recent study by Li and colleagues, the effects of added placebo, fluticasone propionate, and salmeterol to a basal dose of inhaled beclomethasone or budesonide were evaluated using markers of inflammation derived from BAL and bronchial biopsy. More than 20 such markers were evaluated, and only one, the expression of the general eosinophil marker EG1, was reduced by salmeterol. Of note in this study, added fluticasone, a potent and topically active corticosteroid, had no effect on this parameter of airway inflammation.

The purpose of this investigation was to establish the effects of the long-acting ₂-agonist salmeterol on indices of airway inflammation after SAC in asthmatic patients.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject Selection and Screening

Thirteen subjects with allergic asthma were selected for study. All had asthma meeting American Thoracic Society (ATS) criteria, with bronchodilator responses of at least 12% in FEV₁, used inhaled short-acting -agonists alone for asthma control, and reported symptom frequency consistent with mild persistent asthma (17, 18). All were atopic by demonstration of at least two positive (> 3 mm) skin test responses to a panel of common aeroallergens. None used oral or inhaled corticosteroids, inhaled nedocromil or cromolyn, or oral theophylline. Other characteristics of the subjects are shown in Table 1.

At screening, all subjects provided informed consent to a protocol approved by the Biomedical Institutional Review Board at the University of Pittsburgh. History and physical examination were completed, and skin testing was performed with a panel of common aeroallergens, and histamine phosphate positive control, using the prick-puncture method. Spirometry and bronchodilator responses were performed with equipment meeting ATS standards for accuracy and reproducibility (Med Graphics, Minneapolis, MN) before and 15 min after inhalation of two puffs of albuterol from a metered-dose inhaler (Ventolin; Glaxo Wellcome, Research Triangle Park, NC).

Aerosol allergen challenge was performed as described to determine the provocative dose for allergen that produced a 20% decrease in FEV₁ (APD₂₀) (19). Briefly, doubling dilutions of allergen solution were prepared in sterile 0.9% NaCl. After measuring baseline pulmonary function, 5 breaths of saline (0.02 ml per breath) were administered by nebulizing dosimeter (Pulmonary Data Systems, Littleton, CO). Subsequently, 5 breaths of increasing concentrations of allergen were administered by nebulizing dosimeter, with pulmonary function monitoring 10 min later. The study was completed when the FEV₁ fell by at least 20%, or a maximal allergen concentration of 10 mg/ml was administered. The APD₂₀ was determined from these data using a commercial program (PD₂₀; Madison Scientific Software, Wexford, PA). If the FEV₁ did not decrease by 20%, the APD₂₀ was assigned a value of 10,000 for purposes of calculating the dose for SAC (one subject; Table 1).

Experimental Design

This was a randomized, placebo-controlled, double-blind, crossover study (Figure 1). The order of salmeterol or placebo administration was randomized. Each active treatment arm was 7 d in duration, and the two treatment arms were separated by a washout period of 1 mo. During treatment periods, salmeterol or placebo was administered by a standard pressurized metered-dose inhaler at a dosage of 2 puff twice daily. All study devices were identical in appearance, and were differentiated only by coded information. Bronchoscopy, BAL, and SAC were performed on the fifth day of treatment (first intervention: "baseline," and "immediate response"), and BAL was repeated 48 h later on the seventh day of treatment (second intervention: "late response") to assess the inflammatory response.

View larger version (50K): [in this window] [in a new window]   Figure 1.   Experimental design. Patients were screened (Point S) with history, physical examination, spirometry with bronchodilator response, and skin testing. One day later (within the Screening period) aerosol allergen challenge was performed to establish PD20 for allergen. One month later, subjects were randomized (Point R) to active treatment or placebo treatment for 7 d. On Day 5, the first intervention was conducted (bronchoscopy, SAC, and BAL), and on Day 7 (48 h afterwards), the second intervention was performed (BAL). Subjects then underwent washout for 1 mo, and repeated the study with the other treatment.

Bronchoscopy, BAL, and SAC

Bronchoscopy was performed as described (1, 19) with minimal premedication consisting of atropine 0.6 mg intramuscularly with or without midazolam 1 mg intramuscularly. Topical anesthesia was accomplished with 4% cocaine and 1% lidocaine jelly. The fiberoptic bronchoscope was passed transnasally, with minimal additional 1% lidocaine solution delivered through the bronchoscope for topical anesthesia.

For first-intervention procedures, four bronchopulmonary segments were identified. BAL was performed in one using two 60-ml aliquots of 0.9% NaCl warmed to 37° C as described (16). This sample was used to establish a baseline. Generally, the left upper lobe, anterior segment, was used for the baseline samples. SAC was then conducted sequentially in three separate bronchopulmonary segments with 10 ml saline, or one of two allergen doses (1% APD₂₀low dose; 10% APD₂₀high dose) diluted in 10 ml saline. Typically, the lingula was used for the saline segment, the right upper lobe, anterior segment for the low-dose allergen, and the right middle lobe for the high-dose allergen. For each challenge, the volume was instilled through the bronchoscope, and a wedged position was maintained for 5 min. BAL was then performed to assess the immediate response (saline 5 min; Low Ag5 min; High Ag5 min). Saline challenge always preceded low-dose allergen, which always preceded high-dose allergen.

For second-intervention procedures, BAL was performed in each previously challenged segment. The challenged segments were carefully identified to ensure that the same segment was entered, and thus to ensure accurate sampling of the inflammatory response.

BAL Processing and Analysis

Lavage fluid was processed as previously described (1, 20). BAL fluid (BALF) was collected, placed on ice for transport, and processed within 30 min of collection. Fluid from each segment was pooled, and the volume was measured. Each segment was processed and analyzed separately. BALF were centrifuged at 400 × g for 15 min to sediment cells. The supernatant fluid was frozen at 80° C until analyzed. Cells were washed once in Hanks' balanced salt solution (Gibco, Grand Island, NY), and resuspended at 2 × 10⁶/ml. Cytocentrifuge preparations were prepared and stained with Diff-Quik (Baxter Healthcare, McGaw Park, IL). At least 300 nucleated cells were identified as alveolar macrophages (AM), eosinophils, neutrophils, or lymphocytes. The absolute number of each type of cell was calculated by multiplying the total cell count by the differential fraction.

Cytokines in neat, unconcentrated BALF were measured by ELISA kits (Endogen, Cambridge, MA). The specific cytokines analyzed, and corresponding limits of detectability were as follows: tumor necrosis factor-alpha (TNF-; 5 pg/ml), interferon gamma (IFN-; 2 pg/ml), interleukin-1 (IL-1; 1 pg/ml), IL-4 (2 pg/ml). Histamine in neat BALF was measured using a commercial kit (Immunotech, Westbrook, ME) with a sensitivity of 28 pg/ml. Total protein was measured by microtiter modification of the Lowry assay, as previously described (20).

AM were purified from unfractionated BAL cells over a single discontinuous gradient of Percoll (Sigma, St. Louis, MO) as previously described (20). Unfractionated cells were carefully layered over a cushion of isotonic Percoll of density 1,075 mg/ml. After centrifugation at 600 × g for 30 min, the interface cells were collected, washed, and suspended for functional analysis. Viability of these populations was routinely in excess of 95%, and by morphologic criteria, were in excess of 98% AM.

Superoxide production by purified AM was measured as described (19, 20). Spontaneous and phorbol myristate acetate (PMA)-stimulated superoxide release was quantitated spectrophotometrically as the difference between experimental wells and identical wells containing superoxide dismutase at a concentration of 25 µg/ml.

Statistical Analysis

The immediate phase response to allergen challenge is characterized by mediator release. The late response is qualitatively different, with characteristics of cellular influx, cellular activation, and increased concentrations of cytokines (21). Therefore, we analyzed Day 1 and Day 3 data separately, with a two-way analysis of variance (ANOVA) for the effect of two factors (allergen and salmeterol) using a commercial package for microcomputers (SigmaStat 2.00 and 2.03; Jandel Scientific, San Rafael, CA). Logarithmic transformation was employed, if necessary, to meet the assumptions for normality testing. For samples in which a particular analyte was below the limit of detectability, we substituted one-half of the corresponding limit of detectability for purposes of statistical analysis. Post hoc testing using Student-Newman-Keuls testing was performed to determine which specific populations differed, but these analyses were carried out only if the initial two-way ANOVA demonstrated a significant effect (p < 0.05) (22).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Results of SAC 5 min after Challenge

The immediate results of SAC are summarized in Table 2. Although a reduction in BAL histamine concentration was observed in the salmeterol ( 20 µg/ml) compared with placebo ( 24 µg/ml) groups after high-dose allergen challenge, these changes did not reach statistical significance (p > 0.3). There were no statistically significant effects of salmeterol on any of the parameters measured in samples obtained 5 min after allergen challenge. Therefore, only the effects attributable to allergen are tabulated. SAC was associated with significant allergen dose-dependent release of histamine into BALF.

Results of SAC 48 h after Challenge

We observed allergen concentration-dependent influx of total cells, eosinophils, protein leak, and priming of reactive oxygen species metabolism by purified AM 48 h after SAC (Table 3), consistent with previous reports of these effects (19). No significant allergen dose-dependent effect on the concentration of cytokines was observed (Table 3), although allergen dose-ordering was observed for these measures.

Effect of Salmeterol on Airway Physiology

There was a significant effect of 1 wk of salmeterol on airway physiology in subjects with mild asthma. The FEV₁ percent predicted immediately prior to the first bronchoscopy was significantly higher in the salmeterol arm (80.8% ± 2.8) than the placebo arm (76.0% ± 3.0, p < 0.007), despite the relatively mild degree of airway obstruction exhibited by our cadre of subjects.

Effect of Salmeterol on BAL Volume Recovery, Total, and Differential Cell counts

The effects of salmeterol on these parameters of inflammation are summarized in Table 4. There was no effect of salmeterol on the volume of fluid recovered during BAL, total or differential cell counts after SAC, either at 5 min or at 48 h.

Effect of Salmeterol on Soluble Factors in BALF

Five minutes after low- or high-dose SAC, BAL histamine concentrations were smaller in the salmeterol arm (low: 2.3 ± 5.5 µg/ml and high: 20.4 ± 5.5 µg/ml) than in the placebo arm (low: 4.8 ± 5.5 µg/ml and high: 23.9 ± 5.5 µg/ml), but the differences between salmeterol and placebo arms did not reach statistical significance (p > 0.3). There was no effect of salmeterol on BAL histamine concentrations 48 h after SAC.

There were no significant effects of salmeterol on the concentrations of total protein, or on TNF-, IL-1, and IFN- in BALF either 5 min or 48 h after SAC. Overall, however, salmeterol therapy for 7 d was associated with a significant reduction in IL-4 (9.57 ± 0.87 pg/ml placebo versus 6.17 ± 0.87 salmeterol, p < 0.01). This difference in IL-4 was most pronounced in the baseline samples (9.53 ± 1.34 pg/ml placebo versus 5.53 ± 1.34 salmeterol, p < 0.05), whereas there was a trend only in samples obtained 48 h after SAC (9.62 ± 1.68 pg/ml placebo versus 6.80 ± 1.16 salmeterol, p = 0.18).

Effect of Salmetrol on Cell Function

Spontaneous release of superoxide anion by unfractionated BAL cells was significantly less in the salmeterol arm compared with the placebo arm (Table 4). However, stimulated superoxide release, and both spontaneous and stimulated superoxide release by purified AM were similar in both arms.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We observed several features of the 48-h inflammatory response to SAC in asthmatic individuals which are consistent with earlier reports, including allergen concentration-dependent influx of eosinophils, and priming of purified AM for superoxide release. Thus, this model of airway inflammation induced by SAC and quantified by BAL was similar in many regards to the airway inflammation of asthma. However, we addressed specifically the question of allergen-driven airway inflammation, using the reproducible and powerful technique of SAC. Further, we studied sufficient subjects to minimize the likelihood of missing a clinically important effect. Thus, this is the first study to address the question of the influence of salmeterol on allergen-driven airway inflammation.

No Evidence of Increased Inflammatory Response

In this study, we demonstrated no evidence of potentiation of airway inflammatory responses to SAC in asthmatic individuals. There was no change in the recruitment of neutrophils, eosinophils, or AM 48 h after SAC attributable to salmeterol treatment for 1 wk. In addition, there was no significant effect of salmeterol on baseline indices of inflammation, nor were effects of salmeterol on the immediate (5-min) responses to segmental saline or allergen challenge apparent, with the exception of a 40% fall in baseline IL-4 in the salmeterol compared with placebo samples. Because of the evidence that ₂-agonists may interfere with corticosteroid signaling, perhaps even endogenous steroid signaling, and therefore might amplify inflammation, the evidence that airway inflammation in the salmeterol arm was clearly not increased suggests that any such signaling interference is likely of little consequence in vivo. Other possible explanations for the lack of an amplifying effect of salmeterol on airway inflammation include differences between endogenous and exogenous steroid effects, or differences between in vitro measures and the in vivo model of SAC.

No Evidence of Decreased Inflammatory Response

Only one measure of airway inflammation, BAL concentration of IL-4, was significantly reduced by salmeterol treatment for 1 wk. The implication of this finding is not altogether clear, but is consistent with unpublished observations of the effects of salmeterol by other investigators (J. J. Murray, personal communication). The basal numbers, and allergen-induced influx of alveolar macrophages, eosinophils, neutrophils, and lymphocytes were similar in both salmeterol and placebo arms, and activation of AM as measured by superoxide production was also comparable. A single measure of reactive oxygen species metabolism (spontaneous superoxide release by unfractionated BAL cells) was reduced, but PMA-stimulated superoxide production by those cells, and superoxide release by purified AM were similar. Thus, the effect of salmeterol on reactive oxygen species metabolism is variable, and small in magnitude. There were trends toward reductions in neutrophil infiltration and total protein concentration after high-dose allergen challenge, but the differences did not reach statistical significance.

Power Analysis

In that this is an essentially negative study, a critical question is the power of our design to detect differences in inflammation attributable to salmeterol and allergen challenge. Clearly, the study had sufficient power to detect increased inflammation caused by allergen challenge. We developed our sample size using commercial statistical software (SigmaStat) in this study to achieve a power of 90% to detect a 50% difference in histamine concentrations 5 min after SAC, and 50% reduction in superoxide release and eosinophil recruitment 48 h after SAC. Thus, this design had sufficient power to detect meaningful differences in inflammation due to salmeterol therapy.

Comparisons with Previous Studies

One of the first reports to address the question of the effect of salmeterol on airway inflammation, as assessed by BAL, was that of Gardiner and colleagues (23). In this study, nine subjects with asthma requiring inhaled beclomethasone were evaluated by BAL before and after 8 wk of salmeterol or placebo therapy. No effects of salmeterol on the proportions of resident or infiltrating inflammatory cells were seen, either in comparison to baseline values, or to placebo measures. However, this study was relatively small (n = 9), and asthmatic patients were receiving inhaled steroids, which conceivably could have minimized any existing inflammation, and left no inflammatory "signal" to modulate with salmeterol. In fact, the data suggest that this suggestion was true, as the median proportion of eosinophils in baseline BALF was less than 2%.

Pederson and colleagues studied 12 asthmatic subjects with salmeterol or placebo pretreatment, and aerosol allergen challenge (24). Salmeterol treatment did not change blood eosinophil counts, but did reduce serum concentrations of eosinophil cationic protein and eosinophil protein X. These data would suggest that in the setting of aerosol challenge, salmeterol may reduce the level of eosinophil activation. However, another study of aerosol challenge failed to identify an effect of salmeterol on circulating eosinophils or their circulating activation markers (25), suggesting that the effect of salmeterol on markers of inflammation in peripheral blood may be variable. Several differences exist between the current study and those of Pederson and Weersink. First, we used SAC, rather than aerosol challenge, which is a demonstrably more potent stimulus for inflammation (19). Therefore, the inflammatory signal may have been too intense for downregulation by salmeterol. Second, we did not assess soluble markers of eosinophil inflammation either in serum or in BALF. The Pederson study found differences in these granule proteins, but not in blood eosinophil counts. Thus, given the significant differences in design of these investigations, it is not surprising that the results of our two studies are somewhat at variance.

O'Connor and associates examined the effect of albuterol in a study similar in design to the current investigation (26). Although the hypothesis was somewhat different, and the study was retrospective, the findings in 48 asthmatic subjects undergoing SAC and BAL were similar in those receiving albuterol and those who did not. Further, the treatment was a single administration of albuterol, rather than 1 wk of treatment with salmeterol. Nonetheless, this study suggests that the effects of albuterol on inflammation induced by SAC are negligible.

Using the technique of sputum analysis, Pizzichini and colleagues investigated the effects of salmeterol and beclomethasone on airway inflammation induced by aerosol allergen challenge in eight subjects (27). No influence of salmeterol on airway inflammation induced by allergen challenge was seen. However, and importantly, the investigators also could not demonstrate an effect of inhaled beclomethasone on airway inflammation using sputum analysis. Thus, the study may have been underpowered, the analysis of sputum may have been insufficiently sensitive to detect airway inflammation, the single dose administration may have been insufficient to influence the biology of the airway, or, consistent with the present study, salmeterol may have negligible effect on airway inflammation.

Kraft and colleagues employed the natural "exacerbation" of nocturnal asthma to assess the effects of salmeterol on airway inflammation as determined by BAL markers (28). Ten subjects were treated with placebo and salmeterol for 1 wk in a crossover study. Bronchoscopy and BAL were performed at 4:00 A.M. and 4:00 P.M. No differences in eosinophils, neutrophils, macrophages, lymphocytes, or eosinophil cationic protein were seen. In this model of endogenous asthma exacerbation with time-varying airway inflammation, salmeterol therapy did not alter BAL markers of the circadian development of inflammation.

Roberts and colleagues employed bronchial biopsy as an index of airway inflammation (16). Treatment with salmeterol for 6 wk was associated with no significant change in any marker of airway inflammation as assessed by mucosal biopsies.

Finally, Li and colleagues (from the same laboratory as the previously cited Gardiner study) (15) evaluated the effects of added fluticasone (100 µg twice a day) or salmeterol (50 µg twice a day) to a regimen which included low- to medium-dose inhaled budesonide or beclomethasone in a cohort of 45 asthmatic patients. Using BAL and biopsy before and after a 12-wk treatment phase, they found no effect of added salmeterol on most measures of inflammation, including all indices derived from BAL. The number of total eosinophils, as detected by the EG1 monoclonal antibody in bronchial biopsies, was significantly reduced by salmeterol therapy. However, there was quite surprisingly no effect of added fluticasone on this marker. Faul and colleagues have suggested that EG1 is not sufficiently reproducible to be of value in clinical trials studies (29). Furthermore, the experimental design in the Li study was one of additivity, and was not specifically designed to assess the effects of salmeterol alone on airway inflammation. Thus, the significance of observation of decreased EG1+ eosinophils in the Li study is uncertain, and certainly has little relevance to the findings of our more direct approach.

Summary and Conclusions

In this study, we could identify no significant effects of salmeterol on airway inflammation, either on baseline BALF before SAC (after 5 d of therapy), immediate response to SAC 5 min afterwards, or the late response 48 h after SAC, with the exception of reduced spontaneous superoxide release by unfractionated BAL cells, and a 40% reduction in IL-4 in baseline samples. That stimulated superoxide release by those cells, and that superoxide production by purified AM (both spontaneous and stimulated) was unchanged by salmeterol, suggests that the influence of salmeterol on reactive oxygen species metabolism is quantitatively small and may be biologically unimportant. Our data are consistent with a variety of other empiric studies that suggest that salmeterol neither abates nor amplifies airway inflammation. However, this is the first study to examine the question of an effect of salmeterol on inflammation after SAC. We conclude that salmeterol probably does not affect the development of allergen-driven airway inflammation, either positively or negatively. Theoretical concerns about amplified inflammation resulting from salmeterol appear not to have demonstrable empiric consequences in this in vivo model of human airway inflammation.