Oral Glucocorticoid–Sparing Effect of Benralizumab in Severe Asthma

Trial Design

Figure 1. Figure 1. Trial Design.

Patients entering the run-in phase had their oral glucocorticoid dose reduced until the minimum effective dose without loss of asthma control was reached. All the patients then underwent randomization and entered the induction phase, during which the dose established in the run-in phase was maintained. The oral glucocorticoid dose was further reduced, every 4 weeks, in the intervention period from weeks 4 to 24. The oral glucocorticoid dose that was reached at week 24, or the complete discontinuation of oral glucocorticoid therapy, was maintained until week 28. The last dose of benralizumab or placebo was administered at the week 24 visit.

This randomized, double-blind, parallel-group, placebo-controlled trial comprised an enrollment visit, a run-in phase that included stabilization of the oral glucocorticoid dose, a randomized intervention period, and a follow-up visit (Figure 1). The intervention period consisted of the following: an induction phase, during which patients continued receiving their oral glucocorticoid dose as established during the run-in phase; a dose-reduction phase, during which the oral glucocorticoid dose was reduced at regular intervals; and a dose-maintenance phase, during which the reduced oral glucocorticoid dose was maintained or, in patients in whom oral glucocorticoid therapy was discontinued, no further oral glucocorticoids were received.

All the patients had been treated continuously with oral glucocorticoids for 6 months or more before enrollment and were receiving oral prednisone or prednisolone at trial entry. Patients who were receiving any other oral glucocorticoid at enrollment were switched to an equivalent dose of oral prednisone or prednisolone. During the run-in phase, the oral glucocorticoid dose was adjusted to the minimum dose that could be received without loss of asthma control (Table S1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). No additional asthma-controller medication was initiated unless it was used for the treatment of exacerbations (medication was provided by the trial investigators).

After the run-in phase, patients underwent randomization at week 0 (baseline) to receive benralizumab or placebo and then entered the 4-week induction phase, during which the oral glucocorticoid dose established in the run-in phase was maintained. In the subsequent dose-reduction phase (weeks 4 to 24), the daily oral glucocorticoid dose was reduced every 4 weeks by 2.5 to 5.0 mg (Table S2 in the Supplementary Appendix). Only patients who had an oral glucocorticoid dose of 12.5 mg or less per day at the end of the run-in phase (baseline) were eligible for a 100% dose reduction (discontinuation of oral glucocorticoid therapy). If the criteria for dose reduction were not met, the oral glucocorticoid dose was returned to a previous level, which was maintained until the end of the trial. If a patient’s asthma worsened during the maintenance phase (weeks 24 to 28), the final dose was deemed to be one adjustment increment greater than the dose at which the worsening started.

Trial Oversight

The trial was designed by AstraZeneca in collaboration with the first author. Trial data were collected by the clinical investigators and analyzed by employees of AstraZeneca. The first author and two of the authors who are employees of the sponsor vouch for the accuracy and completeness of the data and analyses and for the fidelity of the trial to the protocol (available at NEJM.org). All the authors reviewed the data. Writing and editing assistance, including preparation of a draft manuscript under the direction and guidance of the authors, incorporation of author feedback, and manuscript submission, was provided by Endpoint Medical Communications (funded by the sponsor) and the sponsor. The independent ethics committees of the trial centers or the central institutional review boards approved the trial protocol. The statistical analysis plan is available with the protocol. The trial was conducted in accordance with the principles of the Declaration of Helsinki, and all the patients provided written informed consent.

Patients

Adults were eligible to participate in the trial if they had a blood eosinophil count of 150 cells or more per cubic millimeter and had asthma that had been treated with medium-dose to high-dose inhaled glucocorticoid and LABA therapy for at least 12 months before enrollment and treated with high-dose inhaled glucocorticoid and LABA therapy for at least 6 months before enrollment. Patients had been receiving oral glucocorticoid therapy for at least 6 continuous months directly before enrollment (equivalent to a prednisolone or prednisone dose of 7.5 to 40.0 mg per day). The full eligibility criteria are provided in Section 2 in the Supplementary Appendix.

Interventions

Patients received subcutaneous injections of benralizumab at a dose of 30 mg every 4 weeks, benralizumab at a dose of 30 mg administered every 4 weeks for the first three doses and then every 8 weeks (with placebo administered at the 4-week interim visits; hereafter referred to as the group that received benralizumab every 8 weeks), or placebo administered every 4 weeks. All the trial agents were provided by AstraZeneca. Patients underwent randomization in a 1:1:1 ratio, with the use of an interactive Web- or voice-response system, and were stratified according to eosinophil count (≥150 to <300 cells per cubic millimeter vs. ≥300 cells per cubic millimeter) and country. Investigators and patients were unaware of the trial-group assignments. Patients continued their prescribed high-dose inhaled glucocorticoid and LABA therapies, as well as any other asthma-controller medications aside from oral glucocorticoid therapy (including leukotriene modifiers, long-acting muscarinic antagonists, and theophylline), in an unchanged fashion throughout the trial. Short-acting β₂-agonists were permitted as rescue medications.

Assessments and Procedures

Worsening of asthma was defined as new or increased asthma symptoms or clinical signs that were troubling to the patient or were related to an electronic Asthma Daily Diary alert (see the Supplementary Appendix). An asthma exacerbation was defined as worsening of asthma that led to a temporary increase in the systemic glucocorticoid dose for at least 3 days to treat the symptoms, an emergency department visit resulting from asthma that led to treatment with a systemic glucocorticoid in addition to the patient’s regular maintenance medications, or an inpatient hospitalization because of asthma.

Patients recorded their lung-function measurements (using an electronic handheld spirometer) and asthma symptoms in the Asthma Daily Diary. Patients used the diary to complete the Asthma Control Questionnaire 6 (ACQ-6)¹⁷ and the Asthma Quality of Life Questionnaire (standardized) for persons 12 years of age or older (AQLQ[S]+12).¹⁸ The total asthma symptom score (a composite of morning assessments of asthma symptoms, nighttime awakenings, and rescue medication use and an evening assessment of activity impairment) is assessed on a scale from 0 to 6, with higher scores indicating a greater symptom burden. The ACQ-6 is scored on a scale from 0 to 6, with lower numbers indicating better control of asthma), and the AQLQ(S)+12 is scored on a scale from 1 to 7, with higher scores indicating better asthma-related quality of life. Score changes of 0.5 or more points were considered to be clinically meaningful for the ACQ-6 and AQLQ(S)+12. Blood and induced sputum (from a subgroup of patients) were obtained for the analysis of eosinophils.¹⁹ Adverse events were monitored throughout the trial. These procedures are described fully in Section 3 in the Supplementary Appendix.

End Points

The primary end point was the percentage reduction in the oral glucocorticoid dose from baseline (randomization at week 0) to the final dose at the end of the maintenance phase (week 28) while asthma control was maintained. Secondary end points included the percentages of patients who had a reduction in the average daily oral glucocorticoid dose of 25% or more, of 50% or more, or of 100% (discontinuation of oral glucocorticoid therapy) from baseline to end of the maintenance phase and the percentage of patients with an average final oral glucocorticoid dose of 5.0 mg or less per day while asthma control was maintained.

Additional end points included the annual asthma exacerbation rate, the time to the first asthma exacerbation, the percentage of patients with at least one asthma exacerbation (including exacerbations associated with emergency department visits or hospitalization), the forced expiratory volume in 1 second (FEV₁) before bronchodilation, the total asthma symptom score, the ACQ-6 score, and the AQLQ(S)+12 score. Exploratory end points were used to investigate the effect of blood and sputum eosinophilia on the efficacy of the trial drug. Safety end points included frequencies of adverse events.

Statistical Analysis

We estimated that 70 patients per group would be required for the trial to detect a difference in the primary end point between each benralizumab group and the placebo group with 86% power by means of a Wilcoxon rank-sum test with a two-sided level of 5%. Our estimation was based on simulations that used data from the Steroid Reduction with Mepolizumab Study (SIRIUS), which yielded a median percentage reduction from baseline of 50% in the glucocorticoid dose in the active-treatment group, as compared with no reduction in the placebo group.¹⁴ We aimed for approximately 60 patients with a blood eosinophil count of at least 150 cells to less than 300 cells per cubic millimeter and 150 patients with a blood eosinophil count of 300 cells or more per cubic millimeter to undergo randomization. Efficacy analyses were conducted with an intention-to-treat approach in the full analysis set.

For the primary end point, benralizumab was compared with placebo by means of a Wilcoxon rank-sum test. To control the overall type I error rate, we accounted for multiple comparisons by means of the Hochberg procedure. A sensitivity analysis for the assessment of the primary end point was conducted with a proportional-odds model, with controls for trial group, geographic region (Asia, Central Europe and Eastern Europe, Western Europe and Turkey, North America, and the rest of world), and baseline oral glucocorticoid dose.

We used a Cochran–Mantel–Haenszel test, with adjustment for geographic region, to analyze secondary end points regarding reductions in the oral glucocorticoid dose. A negative binomial model, with adjustment for trial group, geographic region, and number of exacerbations in the previous year, with an offset term of the logarithm of the follow-up time was used to calculate annual exacerbation rates in the trial groups. Treatment effects were described with the use of rate ratios. Details of the models that were used to analyze other secondary end points are described in Section 5 in the Supplementary Appendix. The trial was not powered to assess secondary end points. The analyses of the secondary end points were not controlled for multiple comparisons and are presented with nominal P values. Results for exploratory variables were analyzed with the use of descriptive statistics according to trial group, unless otherwise indicated. Data were analyzed with the use of SAS software, version 9.2 (SAS Institute).

Patients

Table 1. Table 1. Demographic and Clinical Characteristics of the Patients at Baseline.

From April 2014 through November 2015, a total of 369 patients were recruited and entered screening (Fig. S1 in the Supplementary Appendix). Of these, 98 patients did not enter the run-in phase for the adjustment of the oral glucocorticoid dose, most commonly because they did not fulfill the eligibility criteria (84 patients). An additional 51 patients did not complete the run-in phase, most commonly because they did not meet the eligibility criteria (35 patients). A total of 220 patients underwent randomization and started the intervention phase. A total of 11 patients did not complete the trial, including 5 who withdrew of their own decision. All the patients who underwent randomization were included in the full analysis set. Table 1 provides details about the characteristics of the patients at baseline (additional information is provided in Table S3 in the Supplementary Appendix).

Primary Outcome

Table 2. Table 2. Primary and Secondary Outcomes.

Figure 2. Figure 2. Change from Baseline in the Oral Glucocorticoid Dose and Asthma Exacerbations.

Panel A shows the median percentage change from baseline to week 28 in the oral glucocorticoid dose in patients who received benralizumab or placebo. Error bars represent 95% confidence intervals. Values are slightly offset from each other at each time point for clarity. Panel B is a Kaplan–Meier cumulative incidence curve for the time to the first exacerbation in patients who received benralizumab as compared with those who received placebo. Benralizumab administered every 4 weeks was associated with a longer time to the first exacerbation than was placebo (hazard ratio, 0.39; 95% confidence interval [CI], 0.22 to 0.66; P<0.001), and benralizumab administered every 8 weeks was also associated with a longer time to the first exacerbation than was placebo (hazard ratio, 0.32; 95% CI, 0.17 to 0.57; P<0.001).

With respect to the prespecified primary outcome, the median reduction from baseline in the final oral glucocorticoid dose was 75% in patients who received either of the benralizumab regimens, as compared with a reduction of 25% in the patients who received placebo (P<0.001 for both comparisons) (Table 2 and Figure 2A). A total of 24 patients (33%) who received benralizumab every 4 weeks and 27 patients (37%) who received benralizumab every 8 weeks had a reduction of 90% or more from baseline in their final oral glucocorticoid dose, as compared with 9 patients (12%) who received placebo (Table 2). More patients in the group that received benralizumab every 4 weeks and in the group that received benralizumab every 8 weeks than in the group that received placebo had a reduction of 75% or more from baseline (38 patients [53%] and 37 patients [51%], respectively, vs. 15 patients [20%]); similar results were observed with respect to a reduction of 50% or more (48 patients [67%] and 48 patients [66%], respectively, vs. 28 patients [37%]). The odds of a reduction in the oral glucocorticoid dose were 4.09 times (95% confidence interval [CI], 2.22 to 7.57) as high with benralizumab administered every 4 weeks as with placebo (P<0.001) and 4.12 times (95% CI, 2.22 to 7.63) as high with benralizumab administered every 8 weeks as with placebo (P<0.001) (Table 2).

Secondary and Exploratory Outcomes

In patients who received benralizumab, all the secondary end points related to reduction in the oral glucocorticoid dose were met. These end points included cessation of oral glucocorticoid therapy: 56% of the eligible patients who received benralizumab every 4 weeks and 52% of those who received benralizumab every 8 weeks had a 100% reduction from baseline in the final oral glucocorticoid dose, as compared with 19% of those who received placebo. The odds of cessation of oral glucocorticoid therapy were 5.23 times (95% CI, 1.92 to 14.21) as high with benralizumab administered every 4 weeks as with placebo (P<0.001) and 4.19 times (95% CI, 1.58 to 11.12) as high with benralizumab administered every 8 weeks as with placebo (P=0.002) (Table 2). Reductions in the oral glucocorticoid dose were observed regardless of the baseline oral glucocorticoid dose (Table S4 of the Supplementary Appendix).

For secondary end points related to asthma exacerbations, benralizumab administered every 4 weeks resulted in an annual asthma exacerbation rate that was 55% lower than the rate with placebo (marginal rate, 0.83 vs. 1.83; rate ratio, 0.45; 95% CI, 0.27 to 0.76; P=0.003), and benralizumab administered every 8 weeks resulted in an annual asthma exacerbation rate that was 70% lower than the rate with placebo (marginal rate, 0.54 vs. 1.83; rate ratio, 0.30; 95% CI, 0.17 to 0.53; P<0.001). The two benralizumab regimens were associated with lower odds of having at least one exacerbation than was placebo (odds ratio for benralizumab administered every 4 weeks vs. placebo, 0.32; 95% CI, 0.16 to 0.65; P=0.001; odds ratio for benralizumab administered every 8 weeks vs. placebo, 0.28; 95% CI, 0.14 to 0.56; P<0.001) and resulted in a longer time to the first exacerbation than placebo (Figure 2B, and Fig. S2 in the Supplementary Appendix). Benralizumab that was administered every 8 weeks resulted in a marginal rate estimate of annual exacerbations associated with an emergency department visit or a hospitalization of 0.02 (95 CI, 0.00 to 0.18), as compared with a marginal rate estimate of annual exacerbations of 0.32 (95% CI, 0.16 to 0.65) in the placebo group. The rate with benralizumab was 93% lower than the rate with placebo (rate ratio, 0.07; 95% CI, 0.01 to 0.63; P=0.02). The differences between the group that received benralizumab every 4 weeks and the placebo group did not reach nominal significance for this end point (Table S5 in the Supplementary Appendix).

For secondary end points related to lung function, the FEV₁ before bronchodilation at week 20 was higher than that in the placebo group by 256 ml (95% CI, 109 to 403) in the group that received benralizumab every 4 weeks and by 222 ml (95% CI, 75 to 370) in the group that received benralizumab every 8 weeks; at this time the increase in each group was significantly larger than that with placebo. By 28 weeks, there was no longer a significant difference between either benralizumab group and the placebo group in the FEV₁ before bronchodilation (Fig. S3 in the Supplementary Appendix).

For secondary end points related to asthma control, symptoms, and asthma-related quality of life, the use of benralizumab every 8 weeks was associated with a decrease in the ACQ-6 score (indicating better asthma control) from baseline to the end of the maintenance phase that was 0.55 points (95% CI, 0.23 to 0.86) greater than the decrease with placebo (P=0.001) and with an increase in the AQLQ(S)+12 score (indicating better asthma-related quality of life) from baseline to the end of the maintenance phase that was 0.45 points (95% CI, 0.14 to 0.76) higher than the increase with placebo (P=0.004) (Figs. S4, S5, and S6 in the Supplementary Appendix). By contrast, the differences at week 28 between the group of patients who received benralizumab every 4 weeks and the placebo group with regard to the end points of the ACQ-6 score (least-squares mean difference, –0.24 points; 95% CI, –0.55 to 0.08; P=0.14) and the AQLQ(S)+12 score (least-squares mean difference, 0.23; 95% CI, –0.08 to 0.53; P=0.15) did not reach nominal significance. In addition, no significant differences were found between either benralizumab group and the placebo group with regard to changes from baseline to the end of the maintenance phase in total asthma symptoms. The results of exploratory analyses that were conducted to assess blood and sputum eosinophil counts and their potential associations with efficacy outcomes are described in Tables S6 and S7 and in Figures S7 through S10 in the Supplementary Appendix.

Safety and Adverse Events

Table 3. Table 3. Summary of Adverse Events.

A total of 166 patients (75%) had at least one adverse event during the intervention phase (Table 3). The most frequently reported adverse events were nasopharyngitis (in 17% of patients), worsening asthma (in 13%), and bronchitis (in 10%) (Table S8 in the Supplementary Appendix).

A total of 28 patients (13%) had at least one serious adverse event (Table 3). Worsening asthma was the most common serious adverse event (in 7 patients). A total of 3 patients discontinued their assigned regimen because of serious adverse events. Two patients in the group that received benralizumab every 8 weeks discontinued benralizumab: 1 because of pneumonia and 1 because of cardiac failure. One patient in the placebo group discontinued placebo because of pericarditis.

Two patients in the group that received benralizumab every 8 weeks died during the trial (Table 3). The causes of death were acute cardiac failure (the patient had hypertension and coronary artery disease at trial entry) and pneumonia (atrial fibrillation while the patient was in the hospital; this patient had concurrent hypercholesterolemia, hypertension, angina pectoris, congestive heart failure, dyslipidemia, and a history of atrial fibrillation at trial entry). Details are provided in Section 6 in the Supplementary Appendix.

Antidrug antibodies were positive in 12 of 145 patients (8%) who received benralizumab, 10 of whom had neutralizing antibody–positive status (Table S9 in the Supplementary Appendix). Among patients with a positive antidrug-antibody response, 1 of 5 patients in the group that received benralizumab every 4 weeks and 3 of 7 patients in the group that received benralizumab every 8 weeks had an increase from baseline in the blood eosinophil count (Table S10 in the Supplementary Appendix).

In this trial involving patients who relied on oral glucocorticoid therapy to control their asthma, benralizumab significantly reduced the oral glucocorticoid dose, while asthma control was maintained, in patients who had severe asthma, were receiving high-dose inhaled glucocorticoid and LABA, and had an elevated blood eosinophil count. The likelihood of a reduction in the oral glucocorticoid dose was more than 4 times as high with benralizumab as with placebo, and one half the eligible patients (those receiving a baseline prednisone dose of ≤12.5 mg per day) who were receiving benralizumab stopped the oral glucocorticoid therapy completely. In addition to clinically relevant dose reductions or the discontinuation of oral glucocorticoid therapy, patients receiving benralizumab had substantially lower rates of asthma exacerbations than did patients receiving placebo, as well as lower rates of exacerbation-related hospital visits.

The FEV₁ did not significantly increase over the entire 28-week trial period with benralizumab versus placebo. However, it did not decline, either, despite a reduction of 75% in the prednisone dose in each benralizumab group.

Targeting of the alpha subunit of the interleukin-5 receptor with benralizumab has potential advantages over existing anti–interleukin-5 therapies. By targeting the interleukin receptor rather than the cytokine, luminal depletion of eosinophils can occur,¹⁶ which may be related to greater clinical efficacy. In addition, this approach avoids the potential for autoimmune-mediated worsening of asthma, which has been reported with low-dose anti–interleukin-5 therapy.²⁰

This trial has limitations. Although we computed annualized exacerbation rates, benralizumab or placebo was administered only for 28 weeks, and longer-term studies involving patients with oral glucocorticoid–dependent asthma would be necessary before definitive conclusions could be drawn about the long-term efficacy and safety of benralizumab in this group of patients, who had undergone a reduction in the oral glucocorticoid dose.

It was unclear why approximately 20% of patients did not have any reduction in the oral glucocorticoid dose with benralizumab. In a preliminary analysis (data not shown), the baseline blood eosinophil counts in patients who did not have a response were similar to those in patients who had the greatest reductions in their final oral glucocorticoid doses. Perhaps the presence of blood eosinophilia may not identify the eosinophil as a key effector cell in some patients. Future studies may aim to investigate the characteristics of patients that are indicative of a response to benralizumab, to assess the long-term safety and efficacy of eosinophil depletion (as opposed to normalization of eosinophil counts), and to determine the effects of benralizumab on other cells that may express the interleukin-5 receptor.

In conclusion, in patients with severe eosinophilic asthma who received benralizumab subcutaneously every 8 weeks, the use of maintenance oral glucocorticoid therapy could be reduced while asthma control was maintained. No significant effect on FEV₁ was observed at the end of the trial.