Introduction

Severe non-allergic, adult-onset or intrinsic asthma is frequently associated with pronounced eosinophilia [1,2,3,4]. Interleukin-5 (IL-5) is a potent activator of eosinophils and enhances their viability [5]. IL-5 activity has been demonstrated to inversely correlate with pulmonary function in patients with asthma [4], and anti-IL-5 treatment has been shown to improve asthma control in patients with severe asthma and eosinophilia [6].

Reslizumab is an IgG4-kappa humanized monoclonal antibody targeting IL-5 [7]. In Phase 3 clinical trials intravenous (IV) reslizumab dosed at 3 mg/kg once every 4 weeks (q4w) was associated with a significant reduction in the risk of clinical asthma exacerbations (CAEs) and improved asthma control, lung function, and quality of life in patients with inadequately controlled asthma with blood eosinophil levels (EOS) ≥ 400 cells/µL and a history of CAEs [8]. Reslizumab has been indicated as add-on maintenance treatment for adult patients with severe eosinophilic asthma [9].

Among the entry criteria for these trials, patients had to have EOS ≥ 400 cells/µL at screening. In addition, based on traditional concepts to substantiate the presumed diagnosis of asthma, all patients had to demonstrate a reversibility of their forced expiratory volume in 1 s (FEV1) in response to the inhalation of a β2-agonist (albuterol 200 µg) of ≥ 12%. A recent longitudinal cohort study in young adults, with and without asthma, has shown that elevated EOS are associated with airflow obstruction [10].

We therefore hypothesized that any substantial increase in FEV1 following the administration of a β2–agonist might reflect asthma dominated by smooth muscle contraction rather than eosinophilic inflammation. On the other hand, a poor response to β2–agonists in patients with severe asthma and eosinophilia might predict a better response to anti–IL-5 therapy. To substantiate this we performed a post hoc analysis of data from two Phase 3 trials of IV reslizumab (NCT01287039 and NCT01285323) [8] to assess the relationship between EOS, reversibility of airway obstruction and treatment response to reslizumab therapy in a well-characterized population of patients with inadequately controlled moderate–to–severe asthma with EOS ≥ 400 cells/µL.

Methods

Study design and patients

The two duplicate trials enrolled patients aged 12–75 years with inadequately controlled asthma (Asthma Control Questionnaire-7 [ACQ-7] score ≥ 1.5) on medium-to-high doses of inhaled corticosteroids (ICS), and who had screening EOS ≥ 400 cells/μL, ≥ 1 CAE in the previous year, and FEV1 reversibility of ≥ 12% with albuterol [8]. The selection criteria for FEV1 reversibility were chosen based on National Asthma Education and Prevention Program guidelines available at the time of study design [11].

Both trials were conducted in accordance with Good Clinical Practice guidelines, the Declaration of Helsinki, and local regulatory requirements. All patients provided written informed consent, and the relevant health authorities and local ethics committees or institutional review boards approved the study protocols.

Following a 2–4-week screening period, patients were randomized (1:1) to receive IV reslizumab (3.0 mg/kg) or matching placebo q4w for 52 weeks. Patients continued their usual asthma treatment during the screening, run-in and treatment periods. Pre-bronchodilator spirometry, Asthma Symptom Utility Index (ASUI), and ACQ-7 were assessed q4w at the scheduled clinic visits, from day of randomization to the end of treatment. Possible cases of CAEs were assessed by questioning of the patient at every scheduled monthly visit. Asthma Quality of Life Questionnaire (AQLQ) score was assessed at baseline and weeks 16, 32, and 52.

Outcome measures

FEV1 reversibility at baseline (during screening) was assessed according to EOS category at baseline (day of first dose). Categories for baseline FEV1 reversibility were arbitrarily set at < 14%, 14 to < 16%, 16 to < 20% and ≥ 20%, with baseline EOS categories set arbitrarily at < 150 cells/µL, 150 to < 400 cells/µL, 400 to < 700 cells/µL and ≥ 700 cells/µL. Given that blood eosinophil counts are known to be variable over time, assessment of blood eosinophil count at baseline allowed for selection of patients with persistently elevated blood eosinophils ≥ 400 cells/µL at two timepoints (screening and baseline).

The effect of reslizumab versus placebo on asthma clinical outcomes was also assessed in the subgroup with baseline low FEV1 reversibility (12–14%) and pooled high EOS (≥ 400 cells/µL). Assessment of lung function comprised FEV1, FEV1% predicted, forced vital capacity (FVC) and forced expiratory flow at 25–75% of pulmonary volume (FEF25–75%), and other asthma clinical outcomes assessed were CAEs, ACQ-7, AQLQ and ASUI.

Statistical analysis

An analysis of covariance was used to model change from baseline at Week 52 in lung function and patient-reported outcomes (ACQ-7, AQLQ and ASUI) with fixed factors for treatment arm, sex, oral corticosteroid use at baseline (Yes or No), region (USA or Other), and a continuous covariate for height.

CAEs counted are those which occurred between the completion of the first dose of study drug and 2 weeks after the end of treatment/early withdrawal visit. CAE rates, CAE rate ratio, and confidence intervals (CIs) and p values are based on a negative binomial regression model adjusted for baseline usage of oral corticosteroid (Yes or No) and region (USA or other).

All analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results

A total of 953 patients were randomized in the two duplicate Phase 3 studies (reslizumab: n = 477; placebo: n = 476). Patient demographics and clinical characteristics at baseline were similar between reslizumab and placebo groups (Table 1).

Table 1 Patient characteristics during the baseline period
Full size table

Baseline eosinophil categories and FEV1 reversibility

During the screening period, all patients were required to have EOS ≥ 400 cells/µL. However, on the day of the first reslizumab dose, 65 patients had EOS < 150 cells/µL, 179 patients had EOS 150 to < 400 cells/µL, 365 patients had EOS 400 to < 700 cells/µL, and 344 patients had EOS ≥ 700 cells/µL. At baseline, 149 patients had an FEV1 reversibility of < 14% (between 12 and 14%), 104 had reversibility between 14% and < 16%, 172 had reversibility of 16–20%, and 528 had reversibility of ≥ 20%. Across EOS subgroups, baseline mean FEV1 was numerically lowest in the EOS ≥ 700 cells/µL subgroup for both reslizumab and placebo (Table 1). There was no clear relationship between baseline mean FEV1 and FEV1 reversibility subgroup (Table 1). Baseline mean FEV1 was generally comparable between reslizumab and placebo treatment arms within patient subgroups (Table 1).

Baseline FEV1 reversibility according to eosinophil group

Those patients who had low baseline EOS (< 150 cells/µL or 150 to < 400 cells/µL) had a higher mean FEV1 reversibility and a higher proportion of patients who were highly reversible to inhaled β2-agonists (≥ 20% reversibility, 60% and 62.6% of the subgroup populations) compared with patients with higher EOS (Fig. 1). The proportion of patients who responded relatively poorly to β2-agonists (< 14% improvement) was largest in the EOS ≥ 700 cells/µL group (17.4%) compared with other EOS groups, and this high EOS group had the numerically lowest mean reversibility (Fig. 2).

Fig. 1
figure 1

Mean baseline FEV1 reversibility according to baseline blood eosinophil category. EOS blood eosinophil level; FEV1 forced expiratory volume in 1 s; LS least-squares; SE standard error

Full size image
Fig. 2
figure 2

Proportions of patients in each category of baseline FEV1 reversibility according to baseline EOS. EOS blood eosinophil level; FEV1 forced expiratory volume in 1 s

Full size image

Reslizumab treatment effect on lung function measures

Figure 3 shows the observed treatment effects for reslizumab on FEV1 versus placebo at 52 weeks in the group comprising patients with high EOS and the lowest FEV1 reversibility (EOS ≥ 400 cells/µL, < 14% reversibility) compared with the remaining overall population excluding those with EOS ≥ 400 cells/µL and < 14% reversibility. Both groups experienced a clinically significant improvement in FEV1 at 52 weeks with reslizumab versus placebo (mean: +174 mL [95% CI 1–348] and +139 mL [95% CI 76–202], respectively). Interestingly, despite the relatively poor response to β2-agonists, in the EOS high/β2-agonist reversibility low group there was a marked improvement compared with placebo, with a numerically greater treatment effect compared with the remaining population. The absolute increase in FEV1 in mL from baseline after 52 weeks in the high EOS/low β2-agonist reversibility group was numerically higher than the change from baseline in the remaining patient population with both reslizumab treatment (mean: + 439 mL [standard error [SE] 105] and + 270 mL [SE 36]) and placebo (mean: + 265 mL [SE 98] and + 130 mL [SE 37]). However, numerical differences in treatment effect for FEV1 between the EOS high/β2 agonist reversibility low group and the remaining population did not reach statistical significance. Baseline values and treatment effects in these two groups on FEV1, FVC and FEF25–75% are shown in Table 2.

Fig. 3
figure 3

Change from baseline FEV1 at 52 weeks among patients with high EOS and low reversibility, compared with the overall population excluding these patients. CI confidence interval; EOS blood eosinophil level; FEV1 forced expiratory volume in 1 s; IQR interquartile range; LS least–squares

Full size image
Table 2 Change from baseline in lung function parameters after 52 weeks
Full size table

Reslizumab treatment effect on other asthma clinical measures

At 52 weeks, mean annualized exacerbation rate was lower with reslizumab versus placebo in the high EOS/low β2-agonist reversibility group (0.63 vs 1.06, respectively; rate ratio 0.60 [95% CI 0.33, 1.09]; p = 0.0937) and in the remaining overall population excluding those with EOS ≥ 400 cells/µL and < 14% reversibility (0.59 vs 1.19, respectively; rate ratio 0.45 [95% CI 0.35, 0.57]; p < 0.0001) (Table 3). The CAE rate ratio for reslizumab versus placebo was numerically lower in the remaining overall population excluding those with EOS ≥ 400 cells/µL and < 14% reversibility compared to the high EOS/low β2-agonist reversibility group, although numerical differences did not reach statistical significance between groups.

Table 3 Clinical asthma endpoints
Full size table

Baseline values and treatment effects on ACQ, AQLQ and ASUI in the high EOS/low β2-agonist reversibility group and the remaining overall population excluding those with EOS ≥ 400 cells/µL and < 14% reversibility are shown in Table 3. Similar to the lung function findings, we observed a numerically greater treatment effect for reslizumab versus placebo across the asthma clinical outcomes in the high EOS/low β2-agonist reversibility group than in the remaining overall population excluding those with EOS ≥ 400 cells/µL and < 14% reversibility, with no statistically significant differences between groups.

Discussion

Eosinophilia is an important but often variable feature of bronchial asthma, which is highly responsive to corticosteroids and anti-IL-5 therapy. Based on our observation we can postulate that patients with the highest EOS were more likely to have a relatively poor reversibility to β2-agonists, but were as likely to have a marked response to treatment with the anti-IL-5 antagonist reslizumab as the remaining population.

Unfortunately, due to standard asthma inclusion criteria for reversibility used in this study, no patients with EOS ≥ 400 cells/µL and a β2-agonist reversibility of ≤ 12% were studied. Based on our observed results it can be speculated, however, that these patients might have benefited from treatment with reslizumab as the lack of FEV1 increase following the administration of a β2-agonist might reflect asthma dominated by eosinophilic inflammation. Therefore, this population should be specifically investigated in a future study. At present, our data suggest that high β2-agonist reversibility might not be the best predictor of response to anti-IL-5 treatment while low β2-reversibility (in the presence of a marked eosinophilia) might be indicative of eosinophil-dependent asthma. Future studies should therefore test if the combination of the two ‘biomarkers’, namely a low β2-agonist reversibility together with a high peripheral blood eosinophilia, might be a better predictor of response to anti-IL-5 therapy than either parameter alone or other markers such as fractioned exhaled nitric oxide currently used to predict treatment responses in asthma. Finally, our observation suggests that in asthma low β2-agonist reversibility in the presence of eosinophilia should not lead to the assumption of an irreversibility of airflow obstruction, but in contrast might be indicative of persistent eosinophilic inflammation rather than ‘remodeling’, which might predict a marked responsiveness to anti-inflammatory treatment with corticosteroids and/or anti-IL-5 treatments.

In this study, we noted a large increase in FEV1 from baseline in the placebo groups, particularly in the EOS high/β2-agonist reversibility low group. The marked improvement in FEV1 after 52 weeks in the EOS high/β2-agonist reversibility low group receiving placebo may be due to a ‘trial effect’ including increased adherence to ICS, which might preferentially result in improvement in eosinophilic patients. However, the improvement on reslizumab was numerically better in this group than in the remaining patients, suggesting that a possibly good response in FEV1 to ICS does not seem to interfere with the actions of reslizumab on airflow obstruction in eosinophilic patients.

Limitations of the study

Our study is limited by the post hoc nature of our analysis and the relatively small number of patients in the subgroup of interest (n = 52 in the group with the lowest reversibility to β2-agonists treated with reslizumab). Furthermore, our hypothesis that the biomarker combination of high peripheral blood eosinophilia/low β2-agonist reversibility might be an ideal predictor of response to reslizumab is limited by the inclusion criteria of FEV1 reversibility of ≥ 12% precluding assessment of patients with lower baseline reversibility values.

Conclusion

Higher baseline EOS are associated with numerically lower FEV1 reversibility in patients with inadequately controlled asthma with eosinophilia. Reslizumab treatment resulted in clinically significant improvements in asthma clinical outcomes in patients with high EOS and low FEV1 reversibility at baseline. Therefore, reslizumab may preferentially reverse airway obstruction which appears ‘fixed’ to β2-agonists (albuterol) due to action on IL-5 and airway eosinophils. The exclusion of patients with airflow obstruction poorly responsive to β2-agonists from clinical trials of biologics for severe asthma may result in an under-representation of patients with enhanced inflammation with eosinophils who might in fact particularly benefit from this treatment.