Efficacy and Safety of Anti-Interleukin-5 Therapy in Patients with Asthma: A Systematic Review and Meta-Analysis

Background Recent trials have assessed the efficacy and safety of novel monoclonal antibodies such as reslizumab and benralizumab. However, the overall efficacy and safety anti—interleukin (IL) 5 treatment in asthma have not been thoroughly assessed. Methods Randomized controlled trials (RCTs) of anti-IL-5 treatment on patients with asthma published up to October 2016 in PubMed, Embase, and Cochrane Central Register of Controlled Trials (CENTRAL) that reported pulmonary function, quality of life scores, asthmatic exacerbation rate, blood and sputum eosinophil counts, short-acting β-agonist (SABA) rescue use, and adverse events were included. The pooled mean difference, and relative risks (RR), and 95% confidence intervals (CIs) were calculated using random-effects models. Results Twenty studies involving 7100 patients were identified. Pooled analysis revealed significant improvements in FEV1 (first second forced expiratory volume) (MD = 0.09, 95% CI: 0.06–0.12, I2 = 10%), FEV1% (MD = 3.75, 95% CI: 1.66–5.83, I2 = 19%), Asthma Quality of Life Questionnaire (AQLQ) score (MD = 0.22, 95% CI: 0.15–0.30, I2 = 0%), decreased blood, sputum eosinophils and asthmatic exacerbation (RR = 0.66, 95% CI: 0.59–0.73, I2 = 51%); peak expiratory flow (PEF) (MD = 5.45, 95% CI: -2.83–13.72, I2 = 0%), histamine PC20 (MD = -0.62, 95% CI: -1.92–0.68, I2 = 0%) or SABA rescue use (MD = -0.11, 95% CI: -0.3–0.07, I2 = 30%) were unaffected; adverse events were not increased (RR = 0.93, 95% CI: 0.89–0.98, I2 = 46%). No publication bias was observed (Egger's P = 0.78). Conclusions Anti-interleukin 5 monoclonal therapies for asthma could be safe for slightly improving FEV1 (or FEV1% of predicted value), quality of life, and reducing exacerbations risk and blood and sputum eosinophils, but have no significant effect on PEF, histamine PC20, and SABA rescue use. Further trials required to establish to clarify the optimal antibody for different patients.


Introduction
Asthma is a common chronic inflammatory disease that affects more than 300 million people worldwide, and imposes a high disease burden and economic impact globally [1][2][3]. Despite taking high-dosage inhaled corticosteroids according to the Global Initiative for Asthma (GINA) guidelines, at least 40% of patients continue to suffer from inadequately controlled symptoms, either because they are truly resistant or because they do not take them [4,5]. Patients who remain uncontrolled can use other drugs such as leukotriene-receptor antagonists, slow-release theophylline, and long-acting anticholinergics [6]. Since the anti-immunoglobulin (Ig)E humanized monoclonal antibody omalizumab became the first biological treatment approved for treating allergic asthma, many small molecules and monoclonal antibodies targeting biomolecular specificities have been investigated for treating symptomatic asthma [7]. Eosinophilic inflammatory infiltration is a central event in asthma pathogenesis. IL-5 is the chief cytokine responsible for eosinophil production, survival, maturation and recruitment and activation at allergic inflammation sites [8]. Preclinical studies have demonstrated a key role for IL-5 in murine models of allergen-induced airway eosinophilia and hyperresponsiveness [9]. Given the relationship of IL-5 to eosinophilia and asthma severity, human(ized) monoclonal antibodies targeting IL-5 have shown great promise in severe refractory asthma with persistent eosinophilia [10,11]. The anti-IL-5 agents benralizumab, reslizumab, and mepolizumab have been investigated for treating asthma [12,13]. However, their effects on lung function (especially FEV 1 ) have been less consistent. Here, we conducted a meta-analysis of randomized, controlled trials (RCTs) to assess whether anti-IL-5 monoclonal antibodies therapy is safe and effective in patients (more than 12 years) with asthma.

Literature searches and study selection
PubMed, Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL) were searched for articles published from 1946 to October 2016, using the search terms: ''anti-interleukin-5" or ''mepolizumab" or "benralizumab" or "reslizumab" or "monoclonal antibody" or "anti-IL-5", combined with ''asthma". Language restrictions were not applied. Reviews and the reference lists of relevant articles were also screened for additional articles of interest. Two independent authors (FPW and TL) screened all references according to the selection criteria.
To ensure a complete review of the available studies, the abstracts of relevant scientific meetings were also examined, but trials published solely in abstract form were excluded. Any disagreements were resolved by consensus with a third author when necessary. The details of the search strategy are displayed in S1 Table.

Inclusion and exclusion criteria
Eligible clinical trials were defined as: (1) adults/adolescents (!12 years) with diagnosis of asthma; (2) investigations of patients who received anti-interleukin-5 monoclonal antibody therapy at any dose, placebo-controlled or standard therapy; (3) randomized (parallel group) placebo-controlled trials, and (4) RCTs reporting the following outcomes: blood and sputum eosinophil count, asthma exacerbation, lung function, asthma control and quality of life scores, rescue use of SABA and adverse events. We excluded non-randomized, observational, cohort, case-control and non-blinded clinical trials. FPW and TL independently screened all references according to the selection criteria. Differences in opinion about inclusion were resolved by mutual agreement and arbitration of a third author (HM).
Data extraction and quality assessment FPW and TL independently extracted related data in blinded fashion from eligible studies based on the predefined criteria, which included the characteristics of the trials, interventions, and outcomes. The predefined primary outcomes were lung function [first second forced expiratory volume (FEV 1 ), FEV 1 % of predicted value, peak expiratory flow (PEF), histamine PC 20 ], the Asthma Quality of Life Questionnaire (AQLQ) scores, and asthma exacerbation. Asthma exacerbation was defined as a worsening of asthma requiring increased corticosteroids or albuterol dose to control symptoms and/or the need for asthma-related emergency treatment/hospitalization. Secondary outcomes were adverse events and efficacy outcomes [blood eosinophil count, sputum eosinophils (%), short-acting β-agonist (SABA) rescue use]. The risk of bias was assessed using Cochrane-recommended tools, which included: (1) adequate sequence generation; (2) allocation concealment; (3) blinding; (4) incomplete outcome data addressed; (5) free of selective reporting; and (6) free of other bias [14].

Statistical analyses
All analyses were performed with Review Manager (Version 5.3, The Cochrane Collaboration) and Stata (Version 12.0, Stata Corporation, USA), P <0.05 was considered statistically significant. If a study presented more than two interventions, we combined two or three intervention groups into a single intervention group in accordance with the Cochrane handbook. 9 Random-effects model was applied in all data analyses regardless of statistical heterogeneity. Risk ratio (RR) and 95% CIs were used to analyze dichotomous data, and mean difference and 95% CI were used for continuous data. Heterogeneity assumptions were assessed using the I 2 statistic (I 2 >50% indicates significant heterogeneity), and tested with the χ 2 statistic (P<0.05). However, the number of studies affects both the power of the heterogeneity test and the heterogeneity measures I 2 , but not H 2 M [15]. In order to the increase the power of detecting heterogeneity, the 95% CI of I 2 and H 2 M were calculated [15][16][17]. If substantial heterogeneity was identified, subgroup and sensitivity analyses were performed. Moreover, we separately performed subanalyses in different drugs for each outcome. Publication bias was determined using the Begg's funnel plot and assessed by Egger's test if the number of the studies was larger than ten.
Asthma exacerbations. Thirteen studies (6,072 participants) reported on asthma exacerbations [21- 29,33,34,36,37]. Table 1 summarizes their definitions for asthma exacerbation. Although these definitions varied, all 13 studies defined exacerbation based on increased corticosteroids or albuterol dose to control symptoms and/or the need for asthma-related emergency treatment/hospitalization. Fig 7 showed that anti-interleukin 5 monoclonal therapies were associated with a significant reduction in asthmatic exacerbation compared with placebo (RR = 0.66, 95% CI, 0.59-0.73, P<0.001), but the reporting was significantly heterogeneous (I 2 = 51%, P<0.001, 95% CI 12% to 73%, H 2 M ¼ 1:05). Secondary outcomes Blood and sputum eosinophils. 18 trials included blood eosinophil analysis and six trials compared sputum eosinophil levels between anti-interleukin 5 treatment and placebo [18,[22][23][24]28,32]. As the data were reported inconsistently (data were shown as median [range], mean [SD] or geometric mean [log10 SE]), we did not obtain a synthesized analysis of the outcomes. However, from all the results reported, a similar outcome was identified that anti-interleukin 5 significantly decreased blood and sputum eosinophils compared with placebo (S2 Table).

Risk of bias
Fig 10 summarizes the methodological domain assessments for each included study. Most trials had low risk of bias across the six domains. The allocation sequence was adequately generated and concealed in fourteen trials, [22][23][24][25][26][27][28][29][32][33][34][35][36][37]. The randomization techniques included computer generated randomization codes and minimization. The remaining trials did not report the method used, and we were unable to obtain this information. All but one study was described as double-blinded [20]. Almost all RCTs reported complete outcome data, only one trial reported on attrition insufficiently [27].  Anti-Interleukin-5 Therapy for Asthma

Subgroup analyses and sensitivity analysis
To clarify the heterogeneity or identify the optimal patients for this treatment, subgroup analyses were carried out for asthma exacerbations and FEV 1 ( Table 5). The studies were stratified according to effects model, asthma severity, asthma types, sample size, drug administration dosage, follow-up duration and published year. Subgroup analyses showed the efficacy of antiinterleukin 5 on asthma exacerbations were only influenced by asthma severity. Most subgroups showed significantly reduced exacerbations risk. Single dose anti-interleukin 5 in two studies showed no significant differences in exacerbation rates. However, the subgroup results should be interpreted with caution because of the limited sample size and potential bias inherent to subgroup analysis. The meta-analysis findings remained stable with multicenter trials. In addition, excluding the results of any single study did not alter the overall findings. Anti-Interleukin-5 Therapy for Asthma Furthermore, based on the subgroups analysis, we could know that anti-interleukin 5 can significantly improve the FEV 1 of severe asthma (MD = 0.11, P<0.001) and eosinophilic asthma (MD = 0.11, P = 0.001). This further confirmed that the severe eosinophilic asthma patients are the optimal patients for anti-interleukin 5 treatment.

Publication bias
Publication bias was assessed using Begg's funnel plot and Egger's test. Begg's funnel plot of the 14 studies evaluated the effect of anti-interleukin 5 on FEV1 and the Egger's test suggested no publication bias (P = 0.78, Fig 11). And also no publication bias was detected by Egger's test for other outcomes analysis (all P>0.05). However, we could not fully exclude publication bias in four outcomes (FEV 1 %, PEF, histamine PC20, SABA rescue use); we could not evaluate the potential risk of publication bias, since these tests have very low power in meta-analysis. Anti-Interleukin-5 Therapy for Asthma

Discussion
We identified 20 RCTs investigating the effect of anti-interleukin 5 monoclonal antibodies in patients with asthma. The results suggest that anti-interleukin 5 therapy was well tolerated and could significantly improve AQLQ score, FEV 1 , FEV 1 % of predicted value, and decrease asthmatic exacerbation, blood and sputum eosinophil levels, but yielded no effects in PEF, PC 20 , SABA rescue use. Additionally, reslizumab seems to be safer and more effective than the other two drugs based on all outcomes. However, since varied baseline of patients among studies, it is not possible to draw a firm conclusion. Different from previous systematic reviews that only included studies of on mepolizumab [38,39], we included trials about mepolizumab and other two anti-interleukin-5 antibodies-reslizumab and benralizumab. Additionally, the results should be interpreted with caution due to with the relatively small sample sizes and small number of included trials. Therefore, our results may be more believable. In contrast to previous systematic reviews, we found that anti-interleukin-5 treatment slightly increased FEV 1 and FEV 1 % of predicted value. But the clinical relevance of this finding to patients may not be clinically important because of the modest improvement. Only three or four studies reported detailed data, therefore we could not draw exact conclusions for these two parameters due to the insufficient data. Previous two systematic reviews failed to show a significant effect in FEV 1 , likely due to small number of trials analyzed [38,39]. Liu et al [39] converted and pooled continuous variable data such as blood and sputum eosinophils. To reduce the possible bias resulting from data conversion, we only obtained qualitative descriptions with estimations of the two outcomes. Besides, when studies with multiple intervention groups, Liu et al [39] only selected one pair of interventions and exclude the others which are not generally recommended by Cochrane handbook. Our meta-analysis found that there was a significant improvement in AQLQ score, which is consistent with previous two meta-analyses. However, as the mean change in AQLQ score is less than the clinical minimally important difference of 0.5 units, the clinical relevance of this finding may not be clinically important to patients [40]. Asthma exacerbations are associated with substantial morbidity and mortality [41]. Decreasing the asthma exacerbations rate is a key goal in asthma management. Our meta-analysis showed a significant reduction in exacerbation rates. The clinical relevance of this finding to patients may be clinically important. The inconsistency of the results between asthma exacerbations and rescue use of SABA might due to the next two reasons: 1) the slight improvement in lung functions; 2) most exacerbations in trials were predominantly those that would generally be judged severe on the basis of a need for systemic corticosteroid or requiring admission or visit to emergency. This systematic review also has limitations. First, we aimed to identify the overall effect of anti-interleukin-5 therapy on asthma, the asthma severity and baseline asthma therapy varied among studies (Table 2), so the population examined in this review was too heterogeneous to draw any conclusions about the general asthma population. Further research is needed to clarify which subgroups of patients with asthma can benefit from this treatment. Second, in accordance with the Cochrane handbook, we combined two or three intervention groups into a single intervention group regardless of different intervention dosage and administration routine. This made identifying the optimal dose and regimen for treating asthma difficult. Thirdly, although these studies shared many common issues, there were also substantial Anti-Interleukin-5 Therapy for Asthma Anti-Interleukin-5 Therapy for Asthma subgroup and study heterogeneities. Moreover, there also was significant heterogeneity among studies evaluating asthma exacerbation and adverse events; although we used a random-effects model to account for this, the correction is only partial. As for PEF, histamine PC20 and SABA rescue use, given the small number of studies being meta-analysed, it is difficult to detect heterogeneity and accurately estimate it [42]. Lastly, Ortega et al [43] re-examined baseline blood eosinophil counts from previous two studies [24,26] on mepolizumab, they found that the use of the baseline at a threshold of at least 150 cells/μL can be a reliable and simple biomarker for patient selection associated with positive clinical responses to mepolizumab treatment. However, due to the lack of individual patient data among all studies, we failed to further analysis the relationship between blood eosinophil counts!150 cells/μL at baseline and outcomes of mepolizumab, reslizumab and benralizumab treatment.