Introduction
Biologic therapies have revolutionised Severe Asthma (SA) management, heralding a potential era of personalized medicine (1) that better addresses SA’s heterogeneity. In the United Kingdom (UK), four biologics are currently approved for use in SA (2). The two in longest use are the anti-immunoglobulin E (IgE) biologic, Omalizumab (OMA) and the anti-interleukin-5 (IL-5) biologic, Mepolizumab (MEPO). Both agents proved highly effective in phase-III randomised controlled trials (RCTs) (3–6) and have been widely adopted into clinical practice. However, RCT populations are not reflective of real-world patients (7,8). Brown et al. highlighted that only 9.8% of their SA cohort met enrolment criteria of phase-III asthma biologic RCTs (9). Thus, as the portfolio of biologics continues to expand, real-world data on these drugs is urgently needed (10) to better understand their place in real-life SA management. Additionally, an evolving paradox of biologic choice places greater onus on clinicians to “get it right first time”, to save costs and improve patient outcomes. To guide asthma biologic selection, it is imperative to understand patient phenotypes best suited for individual drugs, identify features associated with response, and also consider how best to judge clinical impact of these therapies. Therefore, to help address these needs, we present parallel real-world clinical data on the two widely used asthma biologics, OMA and MEPO, from the thoroughly characterised, longitudinal Wessex AsThma CoHort of difficult asthma (WATCH) (11).
MethodsWATCH is a prospective observational study of patients managed in a tertiary asthma clinic at University Hospital Southampton, UK with “high dose therapies” and/or “continuous or frequent use of oral corticosteroids (OCS)” as per the British Thoracic Society Adult Asthma Management Guidelines 2016 (12) . Detailed study methodology is described elsewhere (11). The study had ethical approval (REC reference: 14/WM/1226) and written informed consent was obtained for all participants.
We conducted a retrospective study of biologic naïve patients from WATCH who were started on OMA or MEPO between June 2006 and May 2019. Biologic eligibility was based on National Institute for Health & Care Excellence (NICE) guidance (13,14) [supplementary table E1]. Our aims were three-fold. First, to define phenotypic characteristics of patients commencing OMA or MEPO against SA patients who remained biologic naive. Second, to assess clinical responses to both agents from biologic trial data and identify features associated with response. Finally, to assess the spectrum of responses to these biologics.
Biologic trial data were collected at baseline (first biologic visit) and at subsequent treatment visits: 2-4 weekly for OMA and 4-weekly for MEPO. This included Asthma Control Questionnaire-6 (ACQ6), incidence of exacerbations (exacerbations requiring an acute oral corticosteroid [OCS] course / increase in maintenance OCS [mOCS]), incidence of emergency department or hospital admissions (Acute Healthcare Encounters [AHE]), current mOCS dose and Clinic percent predicted FEV1 (FEV1%). Hospital anxiety and depression scale (HADS) and Asthma Quality of Life Questionnaire (AQLQ) were collected at baseline and final visits for MEPO. Fractional Exhaled Nitric Oxide (FENO) was collected at all MEPO visits, but only at the baseline OMA visit. Co-morbidity, anthropometric and demographic data were extracted from the WATCH database. Maximum peripheral blood eosinophil count (PBE), total IgE, baseline exacerbations and AHE were from the 12 months prior to biologic approval. Exacerbations and AHE were annualised for comparisons.
To describe the biologic-naïve characteristics of the biologic treated groups, a common comparator ‘severe asthma, non-biologic’ (SNB) group was extracted from WATCH. SNB subjects (n= 178) were participants who either had ≥4 exacerbations or ≥1 AHE or were on mOCS in the past year but did not commence biologic therapy during the study period. Comparisons were made using baseline biologic data for the biologic treated groups and WATCH enrolment data for the SNB group. Additionally, biologic treated groups were mapped onto four age-of-onset/sex clinical clusters [male/early-onset (<18 years), female/early-onset, male/adult-onset (≥18 years), female/adult-onset] that we recently described (15), to further characterise their phenotypic features.
Biologic response was determined by the clinical MDT (Multi-Disciplinary Team), based on NICE guidance (13,14). For OMA, response was assessed using the Global Evaluation of Treatment Effectiveness (GETE) (16–18). For MEPO, response was defined as a ≥50% reduction in exacerbations or in mOCS dose without loss of asthma control. In borderline OCS responders, factors such as change in AHE, symptom control or quality-of-life would additionally guide MEPO continuation. Treatment trials typically ended at 16 weeks for OMA and 12 months for MEPO. However, equivocal trials were extended up to 32 weeks for OMA and 18 months for MEPO. We defined ‘super-response’ separately for both biologics. OMA super-responders were defined as 16-week responders who either had the top quartile of percentage reduction in mOCS dose while being exacerbation and AHE free; or if not on mOCS, were exacerbation and AHE free. For MEPO, super-responders were defined as 12-month responders who either had the top quartile of percentage reduction in mOCS dose, while having a synchronous reduction in exacerbations; or if not on mOCS, had the top quartile of percentage reduction in exacerbations. Statistical analysis was performed with SPSS 26 (IBM Corp, NY, USA), GraphPad Prism 9 (GraphPad Software, California, USA) and R (R Foundation, Vienna, Austria). Continuous variables were presented as Mean (Standard deviation[SD]) or Median (Interquartile range[IQR]). Categorical data were presented as percentage(frequency). Data were analysed using paired and unpaired t-tests, Mann-Whitney U test, Wilcoxon-Signed Rank test, Chi-square test, McNemar test or Fisher’s exact test as appropriate. Multiple logistic regression (backward variable selection) was performed using variables trending towards significance (P<0.2). Statistical significance was set at P-value < 0.05.