Background
Asthma is a common chronic disease and affects approximately 315 million people worldwide [
1]. About 3–10% of all asthma patients suffer from severe asthma which is defined as asthma remaining uncontrolled despite treatment with high-dose inhaled glucocorticoids combined with other controllers (long-acting β2-agonist, long-acting antimuscarinic agent, leukotriene receptor antagonist or theophylline) and/or treatment with systemic corticosteroids for at least 6 months [
2,
3]. Despite the rather small percentage, patients with severe asthma are responsible for up to 50% of the direct and indirect costs associated with bronchial asthma [
4]. Due to distinct symptoms, frequent exacerbations and numerous medication side effects, severe asthma represents a substantial burden for affected patients [
5].
The first IL-5 antibody mepolizumab has been approved for over two years and has since become an established therapy for patients with uncontrolled severe asthma caused mainly by type 2 inflammation. Type 2 inflammation is characterized by the presence of IL-4, IL-5 and IL-13, produced by helper T cells leading to production, recruitment and activation of eosinophil granulocytes [
6].
In large placebo-controlled trials, treatment with mepolizumab was well tolerated, resulting in a significant reduction of exacerbations and intake of oral corticosteroids (OCS) [
7‐
9]. Meanwhile, another IL-5 antibody (reslizumab), differing from mepolizumab by the route of administration and an IL-5 receptor antibody (benralizumab) are available [
10,
11]. Besides reduction of exacerbation rates and OCS dosages, all anti-IL-5 treatments led to a small but partly significant improvement of lung function [
10,
12,
13]. As IL-5 functions as a central cytokine for activation and recruitment of eosinophils, it is not surprising that the number of eosinophil granulocytes in peripheral blood has been shown to be a predictor of clinical efficacy in targeted anti-IL-5 treatment with a greater reduction of exacerbations in patients with an eosinophil blood count of ≥150/μl [
14]. So far this is the only available biomarker for selection of patients who are most likely to benefit from anti-IL-5 treatment [
15]. The percentage of severe asthma patients presenting with high numbers of eosinophils is unknown, but studies of mild to severe asthma suggest approximately 50% [
16]. Besides the initial studies which led to approval of the drugs, experience in clinical practice and efficacy is scarce. Especially, distinct definitions of treatment response to anti-IL-5 therapy remain to be elaborated. The
National Institute for Health and Care Excellence (NICE) has published recommendations, defining the reduction of the exacerbation rate by at least 50% or a clinically reduced dose of continuous OCS as adequate response [
17]. These criteria are not applicable to all patients with severe asthma, as not all patients require continuous OCS treatment or suffer from frequent exacerbations. We propose treatment response criteria, which are easy to assess and applicable to all patients as a continuous OCS therapy as well as frequent exacerbations are not required. Based on our treatment response criteria, we report the clinical efficacy of anti-IL-5 treatment in real-life setting and analyse potential predictors for treatment response.
Discussion
This study aimed to identify response criteria for treatment with mepolizumab and to evaluate clinical efficacy of anti-IL-5 therapy in clinical practice. Furthermore, potential predictors of treatment response were analysed.
Since the approval of mepolizumab as the first available IL-5 antibody, official treatment response criteria remain to be defined [
6,
27]. Based on placebo-controlled phase III-studies [
8,
9], recommendations published by NICE define the reduction of the exacerbation rate by at least 50% or a clinically significant reduced dose of continuous OCS as adequate response criteria [
8,
9,
17]. In accordance with the literature, treatment with mepolizumab led to a significant reduction of asthma exacerbations in the responder group of our cohort. No prediction could be made concerning the definite exacerbation rate in non-responder patients as treatment in all but 2 patients was stopped before reaching the 12 months interval. As exacerbations mainly occur during winter, assessment of at least 12 months is mandatory to truly reflect the exacerbation rate [
28]. Furthermore, exacerbations are common in patients with severe asthma, but are not an ubiquitous feature [
29]. Therefore, while prevention of exacerbation represents a hallmark treatment goal, the exacerbation rate by itself hardly represents a valid criterion for treatment response in routine clinical practice. Similarly, only approximately 30% of patients with severe asthma are dependent on additional OCS treatment with many patients using OCS on demand rather than continuously [
30]. In our cohort 55% of patients were receiving continuous OCS at baseline, rendering the reduction of OCS as an ineligible treatment response criterion in this study.
As the exacerbation rate as well as OCS treatment shows a close correlation to asthma-related QoL, we chose improvement of subjective condition (“subjective treatment response”) as one treatment response criterion [
5,
31]. QoL and asthma control as traditional parameters were assessed separately with validated questionnaires, however, QoL is influenced by numerous aspects unrelated to asthma and the ACT has known limitations in severe asthma.
Korn et al. recommended reducing the ACT cut-off for uncontrolled asthma in severe asthma to 16 [
26]. Despite reduction of asthma-related symptoms following treatment, an ACT score remaining below 16 points indicates poorly controlled asthma where often there is no further option for therapy escalation. When asked about the effects of mepolizumab treatment, 32/42 patients (76%) in our cohort stated improvement of subjective condition. All 32 patients fulfilled at least two response criteria and were therefore grouped as responders, while in the non-responder group, no patient reported subjective improvement.
Given the type-2 immunologic pathway underlying eosinophilic asthma and the IL-5 antagonizing effects of mepolizumab, reduction of eosinophils as proof of interference with one main inflammatory mechanism was chosen as a second criterion for treatment response [
32]. As IL-5 antibodies are approved for patients with blood eosinophils of ≥150 cells/μL at screening or ≥ 300 cells/μL within 12 months prior to treatment by the FDA (
Food and Drug Administration) and EMA (
European Medicines Agency), the blood eosinophil count had to be < 150 cells/μl or less than 80% from baseline to count as treatment responder in this category. A reduction to < 150 eosinophils or less than 80% from baseline could be seen in all responders except one. Interestingly, all non-responder patients except 2 dropped with eosinophil blood counts below 150/μl but did not show subjective improvement or improvement in lung function. This highlights that disease activity in refractory cases of eosinophilic asthma cannot be solely accounted for by eosinophilic inflammation. Some patients with severe asthma present with combined neutrophilic and eosinophilic inflammation of the airways detectable in sputum [
33]. These patients with a phenotype of a mixed inflammatory response appear to have the greatest disease burden and airway limitation, necessitating further research for treatment strategies for neutrophilic or mixed inflammation in asthma [
34].
Price et al. could show that patients with higher blood eosinophils are more likely to benefit from anti-IL-5 therapy but whether the statement “the higher the better” is true is not universally agreed [
35]. Post-hoc-analysis of the phase III
calima and
sirocco studies showed that improvement of lung function in patients with benralizumab treatment was proportional to the extent of blood eosinophilia [
36]. In the real-life scenario presented herein, this could not be reproduced by allocation to response groups as reduction of eosinophil count was present in both groups without significantly differing magnitude. As the blood eosinophil count represents the only available biomarker for initiation of anti-IL-5 therapy, little conclusion can be drawn from eosinophils regarding clinical response. Therefore, identification and validation of possible new biomarkers are highly desirable.
All patients with severe asthma at least intermittently show an impaired lung function with obstructive patterns. As improvement in lung function was frequently observed alongside subjective improvement and decline in eosinophil counts, we decided to include the improvement of lung function as a third response criterion using a slightly modified version of the approved bronchodilator criteria [
37]. In our cohort 26 patients (79%) showed improvement of FEV
1 of 12% or ≥ 200 ml. None of the non-responder patients showed improvement of lung function.
With these suggested response criteria 76% of our patients treated with mepolizumab could be classified as treatment responders. Treatment led to significant improvement in lung function, oxygenation, QoL and asthma control in responder vs. non-responder patients. Overall, anti-IL-5 therapy shows a favourable efficacy in patients with eosinophilic asthma in routine clinical practice.
Nevertheless, a quarter of patients with indication for anti-IL-5 therapy did not respond to treatment. In case of treatment failure, comorbidities as well as aggravating factors for severe eosinophilic asthma (such as allergic bronchopulmonary aspergillosis, bronchiectasis or eosinophilic granulomatosis with polyangiitis) should be ruled out. Demonstrated in family studies, there is increasing evidence that genetic abnormalities play a role in the pathogenesis of severe asthma [
38]. Up to now, genetic testing is however not endorsed by international guidelines in these patients. Given the fixed mepolizumab dosage, under-dosing in obese patients should be considered, considering non-responders switching to intravenous anti-IL-5 agent allowing for individual dosing. Post-hoc analysis of the
Dose Ranging Efficacy And safety with Mepolizumab (Dream)-study revealed similar reduction rates for exacerbations in obese and non-obese patients, but a recently published study demonstrated that in mepolizumab non-responders switch to intravenous, weight-adapted reslizumab can be beneficial [
23,
39]. To account for overlap phenotypes with chronic obstructive pulmonary disease (COPD) and allergic bronchial asthma, we analysed smoking history and occurrence of allergies but no impact on treatment response was evident. Prior to anti-IL-5 therapy, 5 non-responder patients were eligible and received omalizumab therapy, but none of the patients responded to treatment. There was also no significant impact by body weight in our cohort.
Studies defining the appropriate follow-up schedule or the overall duration of anti-IL-5 therapy are missing. Mainly based on the approval studies, a treatment duration for initially 12 months is recommended due to the possibility of delayed treatment response [
39]. In our cohort however, we did not observe any delayed treatment response in patients who failed to respond early after therapy initiation. Especially in regard to high treatment costs, regular assessment seems mandatory to detect treatment non-responders early [
40]. When to delay or discontinue treatment with anti-IL-5 antibodies, is also still under debate. In patients with hypereosinophilic syndrome withdrawal of anti-IL-5 therapy led to a rebound of eosinophilia after 60–90 days [
41]. In asthma patients a relapse of eosinophils to baseline could be seen after 6 months with a significant increase in exacerbations after 12 months [
42]. Based on current available data, anti-IL-5 treatment should be regarded as long-term treatment.
Limitations
There are important limitations to this analysis, mainly inherent by its retrospective design. This especially limits conclusion about exacerbation rates as exacerbations were only reported 12 months prior to treatment initiation. Value of the data and conclusions regarding the predictive power of the assessed factors are partly limited due to the small number of patients especially in the non-responder group. With the criterion of subjective treatment response, we used a non-validated assessment tool.