INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Asthma is a prevalent disease in developed countries (1) and causes an economic burden to patients, their families, and society (2). In Australia the direct cost of asthma associated with medical care has been estimated at $A250 million (3). First line treatment for asthma is usually with inhaled corticosteroids (ICS) (4). There is clear evidence for the positive clinical effects of ICS in symptomatic patients taking bronchodilator therapy (5, 6). As ICS are expensive and can have systemic effects, it is important to keep the dose at the lowest level possible that maintains good asthma control and minimizes side effects. However, reduction in ICS dose may lead to unstable asthma, emergency visits and hospital admissions. Hospitalization accounts for 20-25% of the direct costs of asthma (6, 7). Because most hospitalizations for asthma are emergency admissions, inadequate disease control can be assumed to be present (6). Therefore to avoid this situation, it would be very helpful to have predictors for failure or success of a planned ICS reduction.

According to current asthma management guidelines (8), the level and adjustment of anti-inflammatory asthma treatment are guided by symptoms and lung function. However, there are patients in whom airway hyperresponsiveness (AHR) and airway inflammation persist, although they are apparently clinically well controlled (9). Airway hyperresponsiveness is a characteristic feature of asthma, is considered to be related to airway inflammation (10, 11), and its severity changes with ICS treatment (12). Airway hyperresponsiveness may be measured using challenge tests with direct agonists such as histamine (13) or indirect agonists such as mannitol (14). The severity of airway inflammation can be measured directly by counting the numbers of inflammatory cells in sputum (15). Sputum eosinophils are increased in patients with asthma exacerbations (16), and they are reduced following ICS treatment (17). Airway inflammation may also be reflected by the levels of exhaled nitric oxide (eNO). Exhaled NO is increased during exacerbations of asthma (18), and reduced in subjects taking inhaled corticosteroids (18, 19).

To identify predictors for failure of ICS dose reduction with loss of control, we measured AHR to histamine and mannitol and noninvasive markers of airway inflammation as measured by sputum eosinophils and eNO in patients with stable asthma.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The patients were recruited from the Asthma Clinic of the Royal Prince Alfred Hospital, Australia. Fifty subjects with asthma using ICS to control their asthma who had a past history of wheezing and chest tightness and who had asthma previously diagnosed by a physician were studied. The subjects' characteristics are summarized in Table 1. Asthma severity was graded on the basis of lung function (8). Information on atopic status was available for 44 subjects, 41 of whom (93%) were atopic. Eight subjects were using long-acting -agonists (LABA) and all used short acting -agonists when needed. All subjects were clinically stable. In the 4 wk before the study, subjects had asthma symptoms no more than twice a week, did not wake at night because of asthma, and had no respiratory tract infection. They had no changes in their dose of ICS in the last 4 wk and no major changes in dose (> 1,000 µg daily) in the last 3 mo. Exclusion criteria were current smoking and the use of oral steroids within the previous 6 mo.

The study was approved by the Central Sydney Area Health Service Ethics Committee (protocol no. X97-0230). The trial was carried out under the Clinical Trial Notification Scheme of the Therapeutics Goods Administration of Australia (CTN No. 1997/373). All subjects signed a consent form prior to commencement of the study.

Study Design

This was a prospective study including two study periods: A 4-wk run-in phase (baseline) and a dose-reduction phase, in which the subjects' current ICS dose was halved every 8 wk. Before the start of the study, subjects were screened for eligibility on the basis of the inclusion and exclusion criteria. On the screening day, the clinical diagnosis of asthma was confirmed by a staff physician by examination and history and informed consent was obtained. During the run-in period, subjects attended for two study visits within 2 wk at the end of the run-in period and monthly during the dose-reduction phase. They were also asked to refrain from taking short acting -agonists for 6 h, LABA for 24 h, and antihistamines for 3 d prior to each study day. No ICS were taken on the day of the study. An "indirect" challenge with mannitol powder was performed, at the first visit of the run-in period and a "direct" challenge with histamine was performed at the second visit. Exhaled NO and spirometry were measured before the challenge tests on both days and sputum was collected during or after the mannitol challenge.

After the run-in period of 4 wk, in which disease stability was monitored by peak flow and symptom diary cards, the current ICS dose was halved every eight weeks. The ICS treatment was stopped after a dose of 200 µg of budesonide or beclomethasone, or 125 µg of fluticasone was achieved after successive reductions in steroid dose. The subjects visited the laboratory at monthly intervals, and a BPT with mannitol was performed, eNO and spirometry were measured, and sputum was collected.

Throughout the study subjects were asked to record their asthma symptoms, -agonist use, and peak expiratory flow (PEF), twice daily in a diary card before inhaling their asthma medication.

The clinician responsible for the steroid reduction and for identifying the asthma exacerbation was "blinded" to the results of the mannitol challenge test and sputum results.

Study Endpoints

The dose of ICS was halved every 8 wk, until the patient suffered an asthma exacerbation or was successfully weaned off ICS for 8 wk. An exacerbation was defined by at least one of the following criteria:

1. 1. Reduction of PEF by more than 3 standard deviations from mean PEF value obtained during the run-in period (20), and
2. 2. A sudden rapid decline in peak flow or deterioration in symptoms, suggestive in the physicians' opinion of the development of a severe exacerbation (21).

Lung Function Measurements

Spirometry was performed using a MicroLoop II Spirometer (Micro Medical Ltd, Kent, UK). Forced expiratory volume in 1 s (FEV₁) was used as an index of change in airway caliber. Forced expiratory maneuvres were repeated until two readings of FEV₁ within 100 ml were obtained, the largest of which was used in analyses. Values for FEV₁ and FVC were recorded as a percentage of the predicted values of Knudson and coworkers (22).

Bronchial Responsiveness

Histamine challenge. A bronchial challenge test with histamine was administered using the rapid method (23). Histamine diphosphate (ICN Pharmaceuticals Inc., Costa Mesa, CA) was administered using DeVilbiss No. 45 handheld nebulizers (DeVilbiss Health Care Inc., Somerset, PA), in doubling doses from 0.03 to 3.9 µmol. The test was stopped if the FEV₁ fell by 20% or more. Salbutamol aerosol was administered to aid recovery when necessary. The dose of histamine that provoked a 20% fall in FEV₁ (PD₂₀FEV₁) was estimated by interpolation. Dose-response ratio (DRR) was calculated for all subjects as the percentage fall in FEV₁ at the last dose, divided by the total dose administered, with a constant of 3 added to allow logarithmic conversion (24, 25). Airway hyperresponsiveness (AHR) was defined as PD₂₀FEV₁ < 3.9 µmol histamine, or a DRR of > 8.1%.

Mannitol capsule challenge. A bronchial challenge test with a dry powder of mannitol was administered to all subjects using the protocol previously described by Anderson and coworkers (26). In brief, a noseclip was applied and subjects then performed the challenge with doses consisting of 0 (empty capsule acting as a placebo), 5, 10, 20, 40, 80, 160, 160, and 160 mg of mannitol via a Halermatic (Rhône-Poulenc Rorer, Collegeville, PA). The 80 mg and 160 mg were given in multiple doses of 40-mg capsules. At least two FEV₁ maneuvres were performed 60 s after each dose and the highest FEV₁ was used in the calculation. The FEV₁ value measured after the 0 mg capsule was taken as the prechallenge FEV₁ and used to calculate the percentage decrease in FEV₁ in response to the mannitol challenge. If the subject had a greater than 10% fall in FEV₁ in response to a single dose, the same dose was repeated for reasons of safety. The challenge ceased when a 15% fall in FEV₁ was documented or a cumulative dose of 635 mg had been administered. AHR was defined as a PD₁₅ of 635 mg or less. Salbutamol aerosol was administered to aid recovery when necessary. DRR was calculated for all subjects as the percentage fall in FEV₁ at the last dose, divided by the total dose administered (24, 25). The provoking dose of mannitol to cause a 15% (PD₁₅) fall in FEV₁ was calculated by linear interpolation of the relationship between the precentage fall in FEV₁ and the cumulative dose of mannitol required to provoke this. A PD₁₅ mannitol was equivalent to a DRR of 0.023% fall FEV₁/mg.

Nitric Oxide Measurement

Mixed expired nitric oxide was measured using a modification of the method of Massaro and coworkers (18). The measurement was performed with the subject standing, without wearing a noseclip. The patient took a deep breath and exhaled over 5-15 s to residual volume into an NO impermeable polyethylene bag (Scholle Industries Pty Ltd, Elizabeth West, Australia). The exhaled flow, measured by a rotameter (Dwyer Flowmeter Model VFASS-25; AMBIT Instruments Pty Ltd, Parramatta, Australia), was 10 L/min at a mouth pressure > 20 cm H₂O. The exhaled gas from a single breath was analyzed within an hour using a chemiluminescence analyzer (Thermo Environmental Instruments Model 42C in the Institute of Respiratory Medicine), which has a lower limit of detection of 1 ppb. Ambient NO in the laboratory was measured at the time of testing.

Sputum Collection

Sputum collection was carried out in conjunction with the mannitol challenge. If subjects had to cough during the mannitol challenge, we asked them to spit whatever they produced into a sterile container. At the end of the mannitol challenge, subjects were asked to cough and spit and we collected whatever was produced. All subjects rinsed their mouths with water at each collection point to remove any particles and reduce salivary contamination. All specimens were retained for later examination under the microscope, even if there were no obvious sputum plugs.

Sputum Preparation and Differential Cell Count

Sputum was processed as described by Pin and coworkers (15). Briefly, sputum plugs were picked up and four-times the volume of diluted Sputolysin (0.1%) (Sputolysin Reagent; Calbiochem, San Diego, CA) was added. The samples were placed in a shaking water bath (37° C) for 30 min and then filtered through 50-µm nylon gauze. A total cell count was performed and cytocentrifuge slides were prepared (Shandon Cytospin II, Sewickery, PA). The inflammatory cells were expressed as a percentage of the total inflammatory cell count (400 cells) on slides fixed with methanol and stained with May-Grunwald-Giemsa.

Peak Flow Home Monitoring

Subjects were asked to perform peak expiratory flow (PEF) measurements twice a day before inhaling their medication for the whole study period. The subjects blew three times into the PEF meter (Mini-Wright; Clement Clarke International Ltd., Essex, UK) while standing and the best of three values was recorded. The lowest reading for each week was calculated as a percentage of the best peak flow value achieved during the 4 wk of the run-in period (27, 28).

Statistical Analysis

Data were analysed using the statistical package SAS (SAS Institute Inc., Cary, NC) and STATA (Stata Statistical Software, Release 6.0, 1997; Stata Corporation, College Station, TX). Analyses of PD₁₅ mannitol, PD₂₀ histamine, DRR values of both challenge tests, and eNO were carried out on log-transformed data. Summary values for both DRR values and eNO are geometric means, with their 95% confidence intervals (CI). Summary values for the other measurements are arithmetic means and 95% confidence intervals. Paired t test was used to compare the outcome measurements (DRR mannitol, eNO, FEV₁ percentage predicted, FVC percentage predicted, PEF percentage predicted, PEF lowest percentage best, sputum inflammatory cells) between run-in period and exacerbation period, and the outcome measurements between the visit before the last successful ICS reduction and visits before the failed ICS reduction. Unpaired t test was used for comparison between groups. Possible predictors for failure of ICS reduction were determined using log-rank test and Cox regression. Kaplan-Meier survival curves were used to demonstrate the probability of failure of ICS reduction between AHR positive and negative groups, which were defined at baseline or at the visit prior to exacerbation. Logistic parameters were estimated using maximum likelihood estimation and evaluated using the likelihood ratio test. Odds ratios (OR) with a confidence interval greater than 1 indicate a significantly increased risk for failure of ICS reduction. Significance was accepted at the 5% level. To examine the ability of the DRR mannitol and percentage eosinophils to predict failure of ICS reduction, receiver operator characteristic (ROC) curves were generated by plotting sensitivity against 1-specificity over a range of cut-points for both of these measures.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subject characteristics at run-in period are summarized in Table 2. Apart from a small but significant difference in PEF values, there were no significant differences between the two study days in the run-in period. Of these 50 patients with clinically well-controlled asthma taking ICS, 15 had AHR to histamine, 24 to mannitol, and 13 to both histamine and mannitol in the run-in period. Sputum was successfully collected from 31 subjects (62%) during or at the end of the mannitol challenge at baseline. Subjects older than 40 yr had significantly lower FEV₁ percentage predicted than younger subjects (76% [67.7-84.3] and 98.9% [92.5-105.4]; p = 0.0001) and longer duration of asthma (29.1 yr [22.5-36.7] and 18.9 yr [14.4-23.4]; p = 0.017).

Study Endpoints: Exacerbations and Successfully Weaned off ICS

Thirty-nine subjects suffered an exacerbation of their asthma13 subjects after the first ICS dose reduction, 19 after the second ICS reduction, 5 after the third reduction, and 2 after the fourth reduction. Seven subjects reduced their ICS dose to zero and remained well at the 2 mo follow-up appointment. Inhaled steroids were ceased after the first ICS reduction in one subject, after the second ICS reduction in two subjects, after the third reduction in one subject, and after the fourth reduction in three subjects. Four subjects dropped out of the study during the dose-reduction phasetwo after the first ICS reduction (both noncompliant) and two after the second reduction (pregnancy; noncompliant). When comparing the changes between run-in visits and visits during an exacerbation there were significant differences in ICS dose, lung function (FEV₁ percentage predicted, PEF percentage predicted, and PEF lowest percentage best), and response to mannitol (Table 3). Sputum was able to be collected in 21 out of 39 subjects at the exacerbation visit and showed a significant increase in sputum eosinophils (Table 3).

Predictors for Failure of ICS Reduction

Kaplan-Meier survival analysis using baseline data as predictors. Figure 1 shows the survival curve using AHR measurements at baseline. Being hyperresponsive to both direct (histamine) and indirect (mannitol) challenge test at baseline was a significant predictor for failure of ICS reduction, whereas being hyperresponsive to only a direct or indirect challenge test at baseline was not a predictor for failure of ICS reduction. The majority of exacerbations occurred following the first or second ICS reduction. The odds ratio for failure at or before the second ICS reduction was 2.38 (95% CI: 0.67-8.4; p > 0.05) for AHR to histamine, 2.27 (0.73-7.07; p > 0.05) for AHR to mannitol, and 4.38 (1.03-18.56; p < 0.05) for AHR to both a direct and indirect challenge test at baseline. Being older than 40 was a borderline significant predictor for failure of ICS reduction (p = 0.059). Nonpredictors for failure of ICS reduction at baseline were eNO (cut-off point 20 ppb; p = 0.61), ICS dose (cut-off point 800 µg; p = 0.621), sputum eosinophils (cut-off point 2.5%, p = 0.445, and PEF lowest percentage best (cut-off point 90%; p = 0.162; 85%; p = 0.674; and 80%; p = 0.754).

View larger version (12K): [in this window] [in a new window]   Figure 1.   Kaplan-Meier survival curve based on AHR to histamine (A), mannitol (B), or both histamine and mannitol (C) at baseline. (A, B) The dashed line represents the AHR negative subjects and the continuous line represents the AHR positive subjects. (C) The dashed line represents the AHR negative and the AHR positive (to one test only) subjects to either of the challenge test and the continuous line represents the AHR positive subjects to both challenge tests.

Survival analysis using the visits prior to each ICS reduction. Figure 2 shows the survival curve using AHR to mannitol measured at the visit prior to an ICS reduction. Being hyperresponsive to mannitol (DRR mannitol > 0.023 percentage fall FEV₁/mg) was a significant predictor for failure of ICS reduction. Failure of ICS reduction was not predicted by FEV₁ percentage predicted (p = 0.46), PEF percentage predicted (p = 0.8), or eNO (p = 0.98).

View larger version (6K): [in this window] [in a new window]   Figure 2.   Survival curve using AHR to mannitol measured at the visit prior to an ICS reduction. The dashed line represents the norm-responsive subjects and the continuous line represents the hyperresponsive subjects.

Comparison of the visit before the last successful ICS reduction with the visit before the failed ICS reduction. Twenty-six subjects had both one successful ICS dose reduction and one failed ICS reduction. There were significantly higher levels of airway responsiveness to mannitol, as described by the DRR, and sputum eosinophils before the last successful reduction compared with levels before the failed reduction. There were no significant differences in spirometric function and eNO (Table 4). Figure 3 shows the ROC curves that describe the performance of DRR mannitol and percentage eosinophils for prediction of failure of ICS reduction. DRR mannitol had 90% sensitivity at a cutpoint of 0.023% fall FEV₁/mg (equivalent to a PD₁₅ value of 635 mg) and 90% specificity at a cutpoint of 0.27% fall FEV₁/mg (equivalent to a PD₁₅ value of 53.6 mg). Sputum eosinophils had 90% sensitivity at a cut-point of 6.3%, and 90% specificity at a cut-point of 13.3%.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Receiver operator characteristic (ROC) curves showing the sensitivity and 1-specificity over a range of cut-points for DRR mannitol (open triangles) and % sputum eosinophils (crosses) for predicting failure of ICS reduction in 26 subjects The solid line indicates no discrimination.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of the present study suggest that measurements of AHR and sputum eosinophils can be used to predict the success or failure of reduction in ICS dose during backtitration. First, being hyperresponsive to both "direct" (histamine) and "indirect" (mannitol) challenges at baseline and hyperresponsive to mannitol during dose-reduction phase of the study were clear predictors for failure of ICS reduction. Second, an increase in sputum eosinophilia during backtitration, but not high levels at baseline, was a significant predictor for failure of ICS reduction. Third, subjects aged 40 and older seem to be at greater risk of exacerbation following ICS reduction. Finally, lung function and exhaled NO did not have predictive value for exacerbation following ICS reduction at any time points.

At baseline, having AHR to both histamine and mannitol was a clear predictor for failure of ICS reduction, whereas having AHR to either histamine or mannitol alone was not a significant predictor. Interestingly the combination of the tests was most strongly predictive at or before the second ICS reduction. Although the odds ratio for failure at or before the second ICS reduction was > 1, irrespective of which challenge test was used, only AHR to both challenges was a statistically significant risk factor for failure. Our findings are supported by Sont and coworkers (29), who demonstrated better asthma control if a decrease of AHR to mechacholine was used in addition to existing guidelines (optimizing symptoms and lung function). AHR probably reflects several different acute as well as chronic aspects of airway inflammation (30). There is also some evidence that direct and indirect challenge tests provide different, and probably complementary, information (12, 21, 24, 31), which can also be seen in our present study. The presence of AHR to both challenge tests in our study was a stronger predictor for failure of ICS reduction than either test alone.

We were able to collect sputum in more than 60% of the subjects, which is similar to other reported data (32). We may have been more successful had we used a wet aerosol challenge. With respect to assessment of AHR, mannitol acts as a hyperosmolar stimulus in the same way as hypertonic saline (26). There was no formal comparison made in the same subject using mannitol and hypertonic saline for inducing sputum. However, the mannitol is unlikely to have changed cell number as there is no difference in cell number reported when hypertonic and normal saline are used for sputum induction (33).

Levels of responsiveness to mannitol and of sputum eosinophilia are both predictors of the likelihood of success or failure of ICS dose reduction at any given time. Our subjects were clinically well controlled and symptom free before the failed ICS reduction, suggesting that mannitol responsiveness and sputum eosinophils provide information additional to that provided by symptoms. Our findings are supported by the study by in't Veen and coworkers (21), in which the baseline severity of AHR to hypertonic saline in clinically well controlled subjects was increased in the group of subjects with frequent exacerbations compared to those without exacerbations. Further support comes from the studies of Jatakanon and coworkers (34, 35). In one study (34), they reported dose-dependent changes in sputum eosinophils and PC₂₀ methacholine during treatment with budesonide, with the maximum reduction at their highest dose of 1,600 µg budesonide. In the other study, Jatakanon and coworkers (35) induced an asthma exacerbation by reducing the dose of ICS to 200 µg budesonide and found that the change in sputum eosinophils could be a useful predictor of loss of asthma control. The predictive value for failure of ICS reduction by DRR mannitol and percentage eosinophils in sputum, however, is moderate in our study. A DRR mannitol of 0.035 percentage fall FEV₁/mg had a sensitivity of 72% and a specificity of 56%, and sputum eosinophils of 9.3% had a sensitivity of 80% and a specificity of 66.7%. Unfortunately, we cannot calculate these figures for histamine, because we did not perform a histamine challenge during the ICS dose-reduction phase. A single challenge test and sputum eosinophils do not seem to have the same predictive power as the combination of a direct and indirect challenge test.

Age older than 40 yr was a predictor of borderline significance for failure of ICS reduction. In our study population, the subjects older than 40 yr of age had a significantly lower FEV₁ percentage predicted and longer duration of asthma than those younger than 40 yr. There is some evidence that longer duration of an inflammatory state of the airways may result in anatomical and functional changes (36), especially if we take into account that the duration of the ICS use was not different between the two age groups. These findings are supported by Quardelli and coworkers (37), who found that elderly patients with asthma (> 65 yr) had significantly fewer symptom-free periods and required more frequent systemic corticosteroids; their FEV₁ percentage predicted was significantly lower than those younger than 40 yr of age. In that study, increased asthma severity in the elderly group correlated with the duration of asthma.

Values for spirometric function and exhaled NO did not have any predictive value at any time points. Treatment with ICS causes an improvement of lung function and discontinuation of treatment is often accompanied by exacerbation of the disease and decline in lung function (38). FEV₁ and PEF in our study population were also significantly lower during exacerbation than at baseline. However, spirometric function was not a predictor of failure of ICS reduction. The study of Jatakanon and coworkers (34) found dose-dependent changes in eNO levels, with the maximum reduction evident at moderate doses of budesonide (400 µg). These data suggest that eNO may be more useful when the doses of ICS used are lower. Their subjects' eNO levels were decreased to the same level as the baseline eNO value in our clinically well-controlled study population. Although, eNO might be expected to be a good predictor of response to steroid reduction, it did not change significantly during the dose-reduction phase and it did not increase significantly during exacerbation. It could be hypothesized that long-term ICS treatment leads to sustained inhibition of inducible NO synthase (19), so that eNO does not necessarily increase during exacerbations if ICS have been taken recently.

We conclude that AHR to both a direct and indirect challenge test is a good predictor for failure of ICS reduction. Age older than 40 yr, positive response to an indirect challenge test, and sputum eosinophil numbers also have some predictive value for failure of ICS reduction. However, spirometric function and exhaled NO provide no information that predict failure of ICS reduction. Therefore, for an ICS reduction of more than one halving step in patients with asthma, we would recommend having information on AHR using both a direct and an indirect challenge test and sputum eosinophils.