Inhaled Corticosteroid Reduction and Elimination in Patients With Persistent Asthma Receiving Salmeterol:  A Randomized Controlled Trial FREE

In patients with persistent asthma inadequately controlled by treatment with a low dose of inhaled corticosteroids (ICSs), the addition of a long-acting β₂-agonist^1- 4 provides an incremental improvement in asthma control exceeding that achieved by increasing dosages of ICS. We anticipated that once this improvement occurred, many patients would question whether their dosage of ICS could be reduced or even eliminated. Evidence both supporting^5- 6 and refuting^7- 9 the concept that continual long-acting β₂-agonist therapy treats symptoms but not the underlying disease⁶ has generated controversy. Although some studies^5,10 have evaluated ICS reduction in patients treated with salmeterol xinafoate, the disparate and complex nature of the reduction and elimination strategies, the outcome measures used, and the small numbers of patients evaluated have not provided sufficient information for clinical decision making. To address these issues, the National Heart, Lung, and Blood Institute (NHLBI) Asthma Clinical Research Network (ACRN) conducted the clinical trial Salmeterol ± Inhaled Corticosteroids (SLIC). The trial tested the hypothesis that in patients with persistent asthma whose symptoms are suboptimally controlled with a regularly scheduled ICS (triamcinolone acetonide) but subsequently controlled following the addition of a scheduled long-acting β₂-agonist (salmeterol), the dosage of ICS can be reduced or eliminated without increasing the risk of treatment failure. The trial, which used clinically relevant ICS reduction and elimination strategies that are readily adaptable to patient care, allowed us to show that while being treated with salmeterol, triamcinolone dosages could be safely reduced but not eliminated.

We conducted a 24-week, randomized, controlled, blinded, double-dummy, parallel group trial from February 1997 to January 1999 at the 6 ACRN clinical centers (Figure 1). The study was approved by the ACRN protocol review committee and human subjects review boards at each participating institution. Patients with asthma as defined by the American Thoracic Society guidelines¹¹ who met recommended criteria for treatment with ICSs¹² were recruited at each ACRN clinical center. Written informed consent was obtained from all patients enrolled in the study. The study entry criteria and the initial triamcinolone run-in period are described in an accompanying article.¹³ Briefly, entry criteria included being aged 12 through 65 years and having persistent asthma defined for patients not receiving ICSs at study entry as having a forced expiratory volume in 1 second (FEV₁) of 80% of the predicted value or less and a 12% or greater increase after treatment with aerosolized albuterol. For patients already receiving ICSs, entry criteria included an FEV₁ of 40% or more of the predicted value, and if FEV₁ was 40% to 80% of the predicted value, a 12% or more increase in FEV₁ after treatment with aerosolized albuterol. If FEV₁ was greater than 80% of the predicted value, patients needed to demonstrate a 20% reduction in FEV₁ in response to a provocative concentration of inhaled methacholine of 8 mg/mL or less (PC₂₀ FEV₁ ≤8 mg/mL). Exclusion criteria included smoking (total lifetime smoking history of ≥10 pack-years or smoking within the last year), regular use of other medications except oral contraceptives and nasal beclomethasone, respiratory tract infection or asthma exacerbation within 6 weeks of the run-in period, and serious medical illnesses in addition to asthma.

After a 6-week run-in period with open-label triamcinolone acetonide 400 µg (4 puffs) twice per day via metered dose inhaler (MDI) with built-in spacer and chlorofluorocarbon propellant and rescue albuterol by MDI as needed, patients whose asthma was not well-controlled during the final 2 weeks of the run-in period entered the SLIC trial. Suboptimal asthma control was defined as FEV₁ 80% or less of the predicted value, or if FEV₁ was greater than 80% of the predicted value, an average variability in peak expiratory flow (PEF) greater than 20%, calculated as [(PM PEF − AM PEF)/(PM PEF + AM PEF)/2] × 100. Patients whose asthma was well controlled according to preestablished criteria after the triamcinolone run-in period entered the Salmeterol or Corticosteroids (SOCS) trial.¹³

In the first phase of the SLIC trial (Figure 1), the salmeterol introduction phase, all patients continued to receive 400 µg of inhaled triamcinolone acetonide twice per day. Thirteen of every 15 patients were randomly assigned to receive add-on therapy with 42 µg of salmeterol xinafoate (2 puffs) twice per day via MDI (no spacer; chlorofluorocarbon propellant); and 2 of every 15 patients were assigned to receive placebo salmeterol (placebo-minus group). At the end of 2 weeks, half of the patients who had received triamcinolone and add-on salmeterol were randomly assigned to either maintain their triamcinolone dosage throughout the study (active control, salmeterol-plus group) or to undergo a blinded, 1-step, 50% reduction in triamcinolone dose for the first 8 weeks (triamcinolone reduction phase) followed by an 8 week triamcinolone elimination phase during which salmeterol was used as monotherapy (salmeterol-minus group). The patients in the placebo-minus group also underwent the same phases of triamcinolone reduction and elimination. The placebo-minus group was limited in enrollment to safely evaluate whether the rate of treatment failure following triamcinolone reduction and elimination would be similar to that reported by others and to document that patients enrolled in the SLIC trial required ICS therapy to maintain adequate asthma control.^14- 15 In all cases, triamcinolone reduction and elimination were only performed if patients did not meet criteria for treatment failure status (BOX).

Treatment failure status was defined as the occurrence of ≥1 of the following:

*Greater deteriorations in asthma control (PEF value ≤65% of baseline [mean value of AM prebronchodilator values over a 2 week period] after the first 60 minutes of rescue albuterol use, or an increase in rescue albuterol use ≥8 puffs/24 hours over baseline use or ≥16 puffs/24 hours for 48 hours) were considered asthma exacerbations and were treated by a predetermined set of rescue algorithms that included oral or parenteral corticosteroid administration. All asthma exacerbations were also considered treatment failures.

Patient randomization was performed online via an Internet connection to the computer system at the data coordinating center. Staff members entered and verified the pertinent data and received a drug packet number to give each eligible patient at the 2 randomizations. The first randomization at the end of the triamcinolone run-in period was stratified according to clinical center, and the second randomization before the triamcinolone reduction phase was stratified according to ethnic group, sex, and age.

This study was triple-blinded in that patients, clinical center personnel, and data analysts were all blinded to treatment identity and dose levels. Each patient received 2 triamcinolone canisters to be taken twice per day as 2 inhalations of each: (1) 2 canisters of active triamcinolone for all patients during the salmeterol introduction phase; (2) 1 canister of active triamcinolone and one canister of placebo drug during the reduction phase for the placebo-minus and the salmeterol-minus groups; (3) 2 canisters of placebo triamcinolone during the elimination phase for the placebo-minus and salmeterol-minus groups. Patients in the placebo-minus group received placebo canisters of salmeterol; patients in the salmeterol-plus and salmeterol-minus groups received active salmeterol canisters. All patients received albuterol for rescue therapy as needed. Medication for each patient was packaged together, labeled with a unique number, and distributed to the clinical centers. The contents of the drug packages were known only to administrative personnel at the data coordinating center.

The primary outcome measure was time-to-treatment failure as defined by preestablished criteria (BOX).¹⁶ Patients who met these criteria received 400 µg of triamcinolone acetonide twice per day by open-label inhaler and continued coded inhalers of triamcinolone, salmeterol, and/or placebo. These patients continued to participate in the study until its termination, but no further reductions in triamcinolone dosage were attempted.

Secondary outcome measures included (1) pre–β₂-agonist FEV₁(after 8 hour albuterol hold and 48 hour salmeterol hold) and post-salmeterol FEV₁ (1 hour after administration of 42 µg of salmeterol xinafoate); (2) AM and PM PEF (Airwatch; ENACT Health Management Systems, Mountain View, Calif); (3) salmeterol-protected methacholine response (methacholine PC₂₀) measured 1 hour after a 42 µg dose of salmeterol xinafoate; (4) asthma day and night symptom scores, the scores of 5 symptoms—shortness of breath, chest tightness, wheezing, cough, and phlegm/mucus—each measured on a scale from 0 (no symptoms) to 3 (severe symptoms) and recorded on daily diary cards; (5) asthma quality-of-life scores, which were derived from a 32-item questionnaire with each item scored from 1, no limitations, to 7, totally limited. An overall asthma quality-of-life score was calculated by averaging the responses to all 32 items, and a separate average quality of life score for each of 4 individual domains was calculated¹⁷; and (6) rescue albuterol use recorded on daily diary cards. Values for each of these secondary outcome measures were compared within each group and between groups for the 3 phases of the study. All study-related tests (eg, FEV₁) were administered by ACRN-certified personnel using network standardized equipment and procedures.¹⁸ Study outcomes were reviewed by the ACRN data and safety monitoring board.

The study was designed to have 80% power to detect a difference between the expected asthma treatment failure percentages of 5% in the salmeterol-plus group and 20% in the salmeterol-minus group from the time the triamcinolone dosage was reduced (triamcinolone reduction phase) to the end of the study (Figure 1), allowing for a 10% withdrawal rate and testing at a 2.9% significance level (adjusted from 5% to account for an interim analysis at the trial midpoint based on the Pocock group sequential method¹⁹). To achieve this statistical objective, 65 patients were required in each of the salmeterol groups. The primary outcome was the percentage failing treatment according to Kaplan-Meier estimates using the log-rank test for comparison between groups. Estimates are also provided from the Kaplan-Meier curve of the percentage failing at the end of the triamcinolone reduction and elimination phases. To compare the salmeterol-plus with the salmeterol-minus treatment arms during each phase of the trial, the time-to-treatment failure was analyzed in an intent-to-treat manner by a Cox regression model with time-dependent covariates.²⁰ The treatment group in the reduction and elimination phases was modeled as a time-dependent covariate, thus allowing a separate estimate of relative risk among treatment groups within each phase. The change over each of the 3 phases was tested within and between the salmeterol groups for secondary outcomes. Secondary outcomes with a symmetric distribution were assessed with longitudinal data analyses based on fitting a mean for each treatment group at each time point. Model-based estimates of the change were used for all tests. Outcomes with a nonsymmetric or discrete distribution were analyzed by nonparametric rank tests. The change was calculated and compared within each group with a Wilcoxon sign-rank test and between groups with a Wilcoxon-Mann Whitney test. All secondary outcomes were analyzed using both intent-to-treat and last value carried-forward methods (ie, the last value prior to treatment failure was carried forward at all future time points); similar results were obtained using both methods. The 2 approaches to the secondary analyses should provide the relative extremes of the possible results. Since treatment failures were treated with increasing dosages of inhaled triamcinolone, oral corticosteroids, or both, trends in various secondary outcome measures are depicted with the use of carried-forward analysis to avoid confounding by this treatment intervention.

A total of 422 patients were eligible to enter the common 6-week run-in period for the SOCS¹³ and SLIC companion studies (Figure 1). Of these, 361 completed the triamcinolone run-in period. The 164 patients who achieved good asthma control, according to preestablished criteria, entered the SOCS trial; and the 175 who did not, entered the SLIC trial. Twenty-two patients did not qualify for either protocol or withdrew consent. Of the patients assigned to the SLIC study, 144 (82.3%) qualified by FEV₁ criterion only, 24 (13.7%) by both FEV₁ and PEF variability criteria, and 7 (4%) by PEF variability criterion alone. The characteristics of the patients in each of the 3 groups in the SLIC study prior to the second randomization are listed in Table 1; comparisons of characteristics among the treatment groups revealed no significant differences (all P values >.05).

During the salmeterol introduction phase, 21 patients (12%) were assigned to the placebo-minus group and 154 (88%) to the combined salmeterol group. At the beginning of the triamcinolone reduction phase, 19 patients remained in the placebo-minus group. Of the 154 patients assigned to receive salmeterol, 148 completed the salmeterol introduction phase, and 74 were then randomly assigned to the salmeterol-plus and salmeterol-minus groups (Figure 1). During the triamcinolone reduction and elimination phases, 13 patients (8.8%) in the salmeterol groups withdrew for personal reasons, none citing dissatisfaction with asthma control. Frequency of withdrawal was not significantly different among groups. The fraction of weeks in which patients were adherent to protocol-defined treatment for more than 70% of the days was 3351/3673 (91.2%), with no significant differences among groups.

Of the 167 patients who completed the salmeterol introduction phase, 50 (29.9%) went on to meet one or more of the preestablished criteria for treatment failure (BOX and Table 2). Failure to achieve before and after salmeterol FEV₁ values of 80% or greater of the reference baseline uniquely accounted for 20 (40%) of treatment failures; there were only 2 treatment failures (4.0%) for which clinical safety judgment was the sole reason (Table 2). Seventeen patients (34.0%) with treatment failure (1 in the placebo-minus group; 4, salmeterol-plus group; and 12, salmeterol-minus group) developed asthma exacerbations (defined in BOX). Two patients (4%) in the salmeterol-minus group with treatment failure required brief hospitalization to optimize their asthma control; both episodes occurred during the triamcinolone elimination phase.

Nine patients in the placebo-minus group experienced treatment failure during the reduction and elimination phases of the study (47.4%; 95% confidence interval [CI], 24.5%-70.3%), 9 patients in the salmeterol-plus group (12.2%; 95% CI, 4.6%-19.8%), and 32 patients in the salmeterol-minus group (43.2%; 95% CI, 31.7%-54.7%). Analysis of the percentage of patients experiencing treatment failure during the triamcinolone reduction and elimination phases vs the time-to-treatment failure for the 2 primary comparison groups showed a significant difference between the groups (P<.001, log-rank test) (Figure 2). However, independent analysis of treatment failure for the triamcinolone reduction and elimination phases showed differences in terms of the phase of the study in which the majority of failures occurred. For the reduction phase, the proportion of treatment failures was 2.8% (95% CI, 0%-7%) in the salmeterol-plus group and 8.3% (95% CI, 2%-15%) in the salmeterol-minus group. At the end of the elimination phase, however, the difference in the proportion of treatment failures in the 2 groups substantially increased, with values of 13.7% (95% CI, 5%-22%) for the salmeterol-plus group and 46.3% (95% CI, 34%-59%) for the salmeterol-minus group. The relative risk of treatment failure for patients in the salmeterol-minus group compared with the salmeterol-plus group was 2.2 (95% CI, 0.5-9.2) during the triamcinolone reduction phase (P = .27; Cox regression model); and during the elimination phase, the relative risk of treatment failure in the salmeterol-minus group increased further to 4.3 (95% CI, 2.0-9.2), and was significantly greater than in the salmeterol-plus group (P<.001).

Salmeterol Introduction Phase. After the addition of salmeterol to triamcinolone therapy in both salmeterol groups, AM PEF significantly increased, daily asthma symptom scores and daily use of rescue albuterol decreased, pre-salmeterol FEV₁ values increased, and asthma quality-of-life scores improved overall and for each domain (Table 3 and Figure 3).

In contrast to these improvements in asthma control, the salmeterol-protected methacholine response PC₂₀ significantly decreased. Although the post-salmeterol FEV₁ values decreased in both salmeterol groups, the change was significant only for the salmeterol-plus group. When changes during the salmeterol introduction phase were compared between the 2 salmeterol groups, no significant differences were noted. No change in any secondary outcome measure was observed in the placebo-minus group.

Triamcinolone Reduction Phase. During the triamcinolone reduction phase, 2 statistically significant interval changes occurred. First, the daily symptom score increased in the salmeterol-minus group (P = .03; increase of 0.01 on a 3-point scale), the clinical relevance of which is questionable. Second, the salmeterol-protected methacholine response PC₂₀ significantly increased within the salmeterol-plus group (Table 3; Figure 3D). However, there was no significant difference between the interval change for methacholine response (P = .11) or for any other outcome among the salmeterol groups.

Triamcinolone Elimination Phase. During the triamcinolone elimination phase, outcomes in the salmeterol-minus group significantly deteriorated. Daily asthma symptom scores and daily rescue albuterol use increased, and before and after salmeterol FEV₁ values and quality of life scores (overall and for each individual domain) decreased. In contrast, no significant deterioration occurred in any outcome measure in the salmeterol-plus group (Figure 3 and Table 3). Significant differences in between-group interval change comparisons were noted for daily symptom scores (P<.01), daily rescue albuterol use (P<.01), and quality-of-life scores (overall and for each domain; P<.01).

Sequential Effects. To determine what the overall anticipated effect would be in clinical practice following (1) the addition of salmeterol to triamcinolone therapy and (2) either a reduction in or elimination of or no change in triamcinolone dosages, we compared the interval change for 4 outcome measures from the start of the salmeterol introduction phase to 2 later time points: (1) the end of the triamcinolone reduction phase, and (2) the end of the elimination phase (Figure 3). For the first interval comparison, asthma symptom scores improved for the salmeterol-plus (P<.01) and salmeterol-minus (P = .04) groups and average daily rescue albuterol use for both salmeterol groups decreased (P<.001). For pre-salmeterol FEV₁ values, only the salmeterol-plus group demonstrated significant improvement (P = .02). Salmeterol-protected methacholine response PC₂₀ decreased significantly in both the salmeterol-plus (P = .01) and salmeterol-minus (P<.001) groups.

For the second interval comparison, the salmeterol-plus group demonstrated significant improvement in AM PEF values, daily asthma symptom scores, daily rescue albuterol use, and overall quality-of-life scores (P<.01, all comparisons); salmeterol-protected methacholine response PC₂₀ decreased (P<.01). No improvement in any outcome measure was noted for the salmeterol-minus group. Salmeterol-protected methacholine response PC₂₀ decreased (P<.001) with a trend for a significant decrease in pre-salmeterol FEV₁ during this interval (P = .07).

The idea that treatment with salmeterol could allow a reduction in ICS dosages in patients with persistent asthma was based on the availability of long-acting β₂-agonists for the treatment of asthma,²¹ the demonstration of the safety of regularly scheduled use of inhaled albuterol,¹⁸ and the results of controlled clinical trials demonstrating that the addition of long-acting β₂-agonists to a fixed dosage of ICS improves asthma control more than increasing dosages of ICS.^1- 2,4 To provide information for clinical decision making, we designed a clinical trial to determine whether the dosages of ICS can be reduced and subsequently eliminated in patients treated with low to moderate dosages of ICS whose asthma control had improved with the addition of salmeterol treatment. Our data indicate that in patients treated with add-on salmeterol, the ICS dosage required to achieve asthma control can be safely reduced, but total elimination of ICS results in an unacceptably high rate of treatment failure.

The SLIC protocol included 3 phases to evaluate whether salmeterol use could allow ICS reduction, elimination, or both. The salmeterol introduction phase was designed to parallel as closely as possible 2 previously published clinical trials.^1- 2 Although the SLIC study population was somewhat younger, baseline pulmonary function values (mean [SD] FEV₁ % predicted 70.4 [8.4]) and ICS doses were comparable. The significant improvements in pulmonary function and asthma quality of life that we observed during the salmeterol introduction phase of the trial (Table 3) were comparable to those observed in the patients evaluated by Woolcock et al¹ and Juniper et al.²² Our replication of these findings in the first phase of the SLIC trial provided the foundation for our test of the hypothesis that the introduction of salmeterol could enable reduction and elimination of ICS.

Prior to the SLIC trial, no definitive guidelines for the reduction or elimination of ICS following the addition of salmeterol therapy existed. For the triamcinolone reduction phase, we reasoned that a 1-step 50% reduction in ICS dosage would be clinically relevant. We chose an interval of 8 weeks during which time we could analyze the effects of ICS dosage reduction and elimination based on reported findings that asthma exacerbation rates plateau within this period when such treatment interventions are initiated.^15- 16 Our data demonstrate that, after the introduction of salmeterol to patients receiving ICS therapy, reductions of ICS dosages by 50% were possible in the majority (>90%) of patients. During the triamcinolone reduction phase, the treatment failure rates in both salmeterol groups were low and not significantly different, even though they differed 2.2-fold. Our study did not have the power to detect a risk ratio of this magnitude when the absolute risks (8.3%, salmeterol-minus; 2.8%, salmeterol-plus groups) are this small. We doubt that these modest differences are clinically significant. On the basis of these results and the lack of clinically relevant adverse differences for any of the secondary outcome measures during this phase of the study, we propose that most patients can tolerate a 50% reduction in their ICS dosage while continuing salmeterol therapy.

In contrast, total elimination of triamcinolone therapy resulted in significant deterioration in asthma control. Specifically, during the triamcinolone elimination phase, the treatment failure rate in the group using salmeterol monotherapy was 4.3-fold greater than in the group using combination therapy. Indeed, the treatment failure rate was nearly 50%, a rate similar to that in the placebo-minus group, which is clearly unacceptably high. Moreover, we noted significant deterioration in a number of secondary outcome measures only during the elimination phase of the trial and only in those patients who were receiving monotherapy with salmeterol. To our knowledge, we are the first to observe a difference between patients receiving and not receiving ICS in terms of decreases in baseline FEV₁ values following chronic salmeterol administration.^23- 24 Our data confirm previous reports of the loss of bronchoprotection to methacholine-induced bronchoconstriction following chronic salmeterol administration.²⁵ Although this loss tended to be greater in those patients in whom ICS therapy was eliminated, the differences were not significant.

The methods in our study differ from those in published articles on the effectiveness of salmeterol as an ICS-sparing agent.^5,10 We believe the simplicity of the study design makes it a relevant model for patient care for a number of reasons. First, the SLIC trial followed what would commonly occur in clinical practice: salmeterol treatment was added, asthma control was improved, and then the triamcinolone dosage was reduced and eliminated, sequentially. Second, triamcinolone reduction and elimination was uniformly structured among the treatment groups, and evaluation of the effects of these step-downs in therapy occurred over 8 weeks, a clinically relevant and practical time interval. Third, we chose a set of criteria that would be considered by most clinicians to represent a clinically significant loss of asthma control. The safety and validity of this set of criteria were established previously in an ACRN-conducted clinical trial.¹⁶

We conclude that the addition of salmeterol therapy to ICS therapy has not only the potential to improve overall asthma control in patients with persistent asthma but it may permit ICS dosage reductions of at least 50% as well. Since it is likely that the potential for ICSs to produce adverse effects increases as dosages are increased,^26- 29 the effectiveness of salmeterol in permitting this degree of ICS dosage reduction is a clinically important feature of its pharmacologic profile. We recognize that our data, in the setting of a clinical trial, may not be directly transferable to clinical practice, but the simple message that the addition of salmeterol will aid in ICS dosage reduction but not elimination should be readily applicable to practice settings. Several studies have suggested complementary interactions between β₂-agonists and corticosteroids in that corticosteroids increase β₂-receptor synthesis and decrease β₂-receptor desensitization,³⁰ while β₂-agonists prime the glucocorticoid receptor for corticosteroid-dependent activation.³¹ However, although these synergistic interactions may facilitate clinically relevant reductions in ICS dosing in many patients, our results indicate that total elimination of ICS therapy in patients receiving salmeterol is not safe and therefore cannot be recommended.