We undertook a post hoc analysis of a phase 2b study of mepolizumab in patients with severe eosinophilic asthma (DREAM [Mepolizumab for Severe Eosinophilic Asthma: A Multicentre, Double-Blind, Placebo-controlled Trial]) (1). We selected this study as it was the only mepolizumab study to assess Fe_NO and blood eosinophils at baseline.

DREAM evaluated placebo and three doses of mepolizumab (75, 250, and 750 mg i.v. once every 4 wk) for 52 weeks. Participants had a history of two or more exacerbations requiring oral corticosteroids in the previous year, and evidence of eosinophilic inflammation as reflected by one of more of the following: a PBE count ≥300 cells/μl, a sputum eosinophil count ≥3%, Fe_NO ≥ 50 ppb, and prompt deterioration of asthma control after a 25% or less reduction in regular maintenance inhaled or oral corticosteroids. As the DREAM study did not show a dose-related effect of active treatment or evidence of an interaction between dose and predictive value of biomarkers, our analysis is based on the combined effect of the different doses.

Participants were divided into subgroups depending on their baseline PBE count and Fe_NO. PBEs were defined as high (≥150 cells/μl) or low (<150 cells/μl), and Fe_NO as high (≥25 ppb) or low (<25 ppb). We chose these cut points because of preexisting evidence linking them to eosinophilic airway inflammation and response to corticosteroids (5). Baseline demographics, clinical characteristics, and annualized exacerbation rates were calculated on the basis of four biomarker subgroups: PBE high–Fe_NO high, PBE high–Fe_NO low, PBE low–Fe_NO high, and PBE low–Fe_NO low. An additional analysis was performed using a PBE cut point of 300 cells/μl.

The DREAM study was a multicenter, randomized, double-blind, placebo-controlled trial. Our primary interest was severe exacerbation rate, defined as the requirement for rescue oral corticosteroids, as the main benefit of mepolizumab treatment is to reduce the rate of exacerbations, and exacerbation rate was the primary outcome measure of the trial. We also present the change in prebronchodilator FEV₁ after 52 weeks of treatment.

A total of 606 DREAM participants had baseline blood eosinophil and Fe_NO measurements. The study population had a mean of 3.6 exacerbations per patient per year in the year before study enrollment. Lung function was reduced, with a mean FEV₁ of 60% predicted, and there was a high symptom burden with a mean asthma control questionnaire 6 score of 2.3 (with <1.5 indicating good control). The baseline demographics and clinical characteristics of the study patients across biomarker subgroups and treatment are shown in Table 1.

The risk for exacerbations was highest in placebo-treated patients with high baseline PBE count and Fe_NO. The efficacy of active treatment was most marked in this group, with mepolizumab showing 62% exacerbation rate reduction compared with 36% exacerbation rate reductions in the PBE high–Fe_NO low group. Mepolizumab did not have a significant effect on exacerbation rate in the PBE low subgroups, regardless of Fe_NO. Similar findings were seen for change in prebronchodilator FEV₁ (Figure 1) and when patients were stratified by a PBE count of 300 cells/μl (Figure 2).

Figure 1. Annualized exacerbation rates and improvement in prebronchodilator FEV₁ with 52 weeks’ treatment with mepolizumab or placebo. Equivalent exacerbation rates for dupilumab 200 mg every 2 weeks and placebo are shown for comparison.

Download Figure | Download Powerpoint

[More] [Minimize]

Figure 2. Annualized exacerbation rates and improvement in prebronchodilator FEV₁ with 52 weeks’ treatment with mepolizumab or placebo with a peripheral blood eosinophil cut point of 300 cells/μl. CI = confidence interval.

Download Figure | Download Powerpoint

[More] [Minimize]

We found that in patients with severe asthma treated with placebo who had high blood eosinophil counts and Fe_NO, the rate of severe exacerbations requiring oral corticosteroid treatment was up to twice that seen in placebo-treated patients with low or discordant biomarker results. The different biomarker groups had similar baseline lung function, symptom scores, and exacerbation risk, indicating that the increased risk for exacerbation events is independent of these other markers of asthma control and risk, indicating that biomarker profiling of patients with severe asthma adds predictive value to a traditional risk assessment.

To evaluate the relationship between biomarker profile and treatment efficacy, we compared our findings with published results of the phase 3 trial of monoclonal antibody dupilumab (QUEST [A Randomized, Double Blind, Placebo-controlled, Parallel Group Study to Evaluate the Efficacy and Safety of Dupilumab in Patients With Persistent Asthma]), a monoclonal antibody that blocks IL-13 and IL-4 by binding to the IL-4 receptor-α (2). This study evaluated two doses of dupilumab (200 or 300 mg subcutaneously once every 2 wk) for 52 weeks in 1,902 patients with moderate-severe persistent, uncontrolled asthma, as per the Global Initiative for Asthma guidelines (6). Both doses had equivalent efficacy, and as combined data are not available, we present data from the 934 patients randomly assigned to dupilumab 200 mg every 2 weeks or matched placebo. Information on change in prebronchodilator FEV₁ by biomarker profile was not available for QUEST (Figure 1). The efficacy of dupilumab on exacerbations was most marked in the PBE high–Fe_NO high group with a 68% reduction. There was also an exacerbation rate reduction of 33% in the PBE high–Fe_NO low group, which was similar to the effect seen in this subgroup with mepolizumab treatment. In the PBE low–Fe_NO low group, neither biologic had a significant effect on exacerbations. In the PBE low–Fe_NO high group, a 39% exacerbation rate reduction was seen with dupilumab. Although not statistically significant, this finding contrasts the absence of effect seen with mepolizumab in this subgroup.

The study populations differed significantly, with the DREAM population having a higher exacerbation risk and a higher proportion of patients using high-dose inhaled or regular oral corticosteroids. However, the higher risk for exacerbation in patients with higher blood eosinophils and Fe_NO was seen in both populations, suggesting a true effect seen across a range of asthma severity. This is in keeping with earlier studies showing that a composite profile of biomarkers of type 2 airway inflammation provides prognostic information over and above an assessment of risk for exacerbation based on traditional asthma measures, although this earlier study used a composite score of blood eosinophils, serum periostin, and Fe_NO biomarkers (7). We extend these earlier findings by showing that patients with both high blood eosinophil counts and Fe_NO also had the greatest response to biological treatment with mepolizumab and dupilumab, indicating that biomarker profiles have predictive as well as prognostic value.

These greater prognostic values of the combined biomarker profile make biological sense, given that the biomarkers relate to different aspects of type 2 immune responses in the airway. The PBE count reflects airway and systemic IL-5 production and is reduced markedly by anti–IL-5 (1) but not dupilumab (2). In contrast, Fe_NO reflects airway IL-13 activity, as it is reduced markedly by anti–IL-13 and dupilumab (2) but not anti–IL-5 (1). Thus, the two measures provide a more complete assessment of type 2 immune responses in the airway. It is also likely that the combination of a systemic and local airway measure adds precision to an assessment in only one of these compartments. We did not find a point estimate improvement for mepolizumab in patients with high Fe_NO but low PBE count, as might be expected, as Fe_NO is a marker of IL-13 activity in the airway. In contrast, there was a trend for benefit of dupilumab in this group.

Caution is required in interpreting this post hoc subgroup analysis, as the number of patients in some subgroups is small and the populations studied in DREAM and QUEST were different. We acknowledge that the relative efficacy of mepolizumab and dupilumab treatment in subgroups may reflect the play of chance or differences in patient populations, as well as the different cytokine associations of the biomarkers. However, it is striking that the relative exacerbation rate reductions in three of the four subgroups were very similar. Our findings suggest that biomarker profiles might have value in identifying patients suitable for different biological agents. Formal head-to-head studies in similar patient populations are needed to assess this possibility prospectively. Future studies should also model the relationship between biomarker values and exacerbations more completely, allowing more accurate inferences to be drawn on individual patient responses.