A randomized, placebo‐controlled trial evaluating effects of lebrikizumab on airway eosinophilic inflammation and remodelling in uncontrolled asthma (CLAVIER)

Abstract Background The anti‐interleukin 13 (IL‐13) monoclonal antibody lebrikizumab improves lung function in patients with moderate‐to‐severe uncontrolled asthma, but its effects on airway inflammation and remodelling are unknown. CLAVIER was designed to assess lebrikizumab's effect on eosinophilic inflammation and remodelling. Objective To report safety and efficacy results from enrolled participants with available data from CLAVIER. Methods We performed bronchoscopy on patients with uncontrolled asthma before and after 12 weeks of randomized double‐blinded treatment with lebrikizumab (n = 31) or placebo (n = 33). The pre‐specified primary end‐point was relative change in airway subepithelial eosinophils per mm2 of basement membrane (cells/mm2). Pre‐specified secondary and exploratory outcomes included change in IL‐13‐associated biomarkers and measures of airway remodelling. Results There was a baseline imbalance in tissue eosinophils and high variability between treatment groups. There was no discernible change in adjusted mean subepithelial eosinophils/mm2 in response to lebrikizumab (95% CI, −82.5%, 97.5%). As previously observed, FEV1 increased after lebrikizumab treatment. Moreover, subepithelial collagen thickness decreased 21.5% after lebrikizumab treatment (95% CI, −32.9%, −10.2%), and fractional exhaled nitric oxide, CCL26 and SERPINB2 mRNA expression in bronchial tissues also reduced. Lebrikizumab was well tolerated, with a safety profile consistent with other lebrikizumab asthma studies. Conclusions & Clinical Relevance We did not observe reduced tissue eosinophil numbers in association with lebrikizumab treatment. However, in pre‐specified exploratory analyses, lebrikizumab treatment was associated with reduced degree of subepithelial fibrosis, a feature of airway remodelling, as well as improved lung function and reduced key pharmacodynamic biomarkers in bronchial tissues. These results reinforce the importance of IL‐13 in airway pathobiology and suggest that neutralization of IL‐13 may reduce asthmatic airway remodelling. Clinical Trial Registration: NCT02099656.


| INTRODUC TI ON
Asthma, a chronic, heterogeneous disorder affecting ≈ 300 million people worldwide, 1,2 is characterized by variable airflow obstruction, airway inflammation, mucus hypersecretion and tissue remodelling, including subepithelial fibrosis. Patients whose asthma remains uncontrolled despite treatment represent a substantial unmet clinical need and are at risk of acute disease worsening. 3,4 Guideline-based standard-of-care therapy includes inhaled corticosteroids (ICS) plus a second controller medication. 4 Airway eosinophilic inflammation is a key feature of asthma driven by type 2 (T2) inflammation. 5 Eosinophil counts in the blood and different airway compartments can be discordant. [6][7][8] ICS treatment can decrease airway mucosal eosinophils, although eosinophilic airway inflammation persists in some patients. 7,9 Interleukin (IL)-13 is a pleiotropic cytokine thought to play a key role in T2-driven inflammation, including eosinophilic inflammation in severe asthma, 10 has been implicated in promoting eosinophil survival, activation and recruitment, [11][12][13] and may also mediate features of airway remodelling relevant to asthma, such as subepithelial fibrosis. 14,15 Lebrikizumab is a humanized monoclonal antibody that binds soluble IL-13 to block downstream signalling. 16,17 Lebrikizumab treatment is associated with increased peripheral blood eosinophils in some patients with asthma, which may have been due to reduced eosinophil trafficking to tissue. 16,18,19 In phase 2 studies, lebrikizumab reduced the number of exacerbations and improved lung function in patients with moderate-to-severe uncontrolled asthma, particularly in those with higher levels of T2 biomarkers such as periostin, blood eosinophils and fractional exhaled nitric oxide (FeNO). 16,20,21 However, replicate phase 3 trials in adult patients with uncontrolled asthma only partially supported these findings. Lebrikizumab significantly reduced the rate of asthma exacerbations over 52 weeks in biomarker-high patients (defined as periostin ≥50 ng/mL or blood eosinophils ≥300 cells/ μL) in LAVOLTA I (NCT01867125), but this effect was inconsistently observed in LAVOLTA II (NCT01868061). 18 Nevertheless, both phase 3 trials showed improvements in forced expiratory volume in 1 second (FEV 1 ). 18 CLAVIER (NCT02099656) was a phase 2 bronchoscopy trial that investigated the effects of lebrikizumab on airway inflammation and remodelling in patients with uncontrolled asthma. Based on the mixed efficacy results of the LAVOLTA studies, the lebrikizumab asthma programme was terminated by the sponsor; therefore, CLAVIER drug dosing was terminated and enrolment closed before the planned sample size was achieved. All enrolled patients were invited to complete the study, and here, results from enrolled participants with available data are reported.

| Study design
CLAVIER was a phase 2, multi-centre, randomized, double-blind, placebo-controlled clinical trial that incorporated research bronchoscopy. The study consisted of a 3-week screening period, 12-week placebo-controlled treatment period, and 8-week safety follow-up period with a planned sample size of 80 patients ( Figure 1; Table S1; see Supplement for additional details). Following written informed consent, patients were screened, and bronchoscopy was performed at visit 4a to collect baseline samples. Patients were randomized 1:1 to receive lebrikizumab or placebo stratified by baseline serum periostin level (<50 or ≥50 ng/mL), baseline asthma medications (total daily dose ≥1000 μg fluticasone propionate dry powder inhaler (DPI) or equivalent plus long-acting beta agonists [LABA; yes, no], and nasosorption/sputum induction substudy participation [yes, no]).

| Patients
Eligible patients were 18-75 years old with a clinical diagnosis of asthma ≥12 months prior to visit 1, documented bronchodilator reversibility (≥12% relative improvement) within 12 months prior to or during screening, and prebronchodilator FEV 1 of 40%-80% predicted at both screening visits 2 and 3. Patients were receiving total daily dose of 500-2000 μg fluticasone propionate DPI or equivalent and were on an eligible second controller medication for ≥6 months prior to visit 1, with no changes within 4 weeks prior to visit 1. Eligible second asthma controller medications were LABAs, leukotriene receptor antagonists, long-acting muscarinic antagonists or theophylline. Doses for ICS and second controllers needed to remain stable throughout the study, except for theophylline which could be adjusted based on blood levels.
Uncontrolled asthma during screening was defined as a five-item Asthma Control Questionnaire score of ≥ 1.5 and at least one of the following: daytime symptoms >2 d/wk, night-time awakening ≥ 1 night/ wk, rescue medication use on ≥2 d/wk and/or interference with normal

Funding information
The study was sponsored by F. Hoffmann-La Roche Ltd. The sponsor was responsible for the clinical operations oversight, data management, medical monitoring, drug supply, statistical analysis, drug safety process, medical writing and journal article processing charges. All authors had full access to all the data in this study and take complete responsibility for the integrity of the data and accuracy of the data analysis. exploratory analyses, lebrikizumab treatment was associated with reduced degree of subepithelial fibrosis, a feature of airway remodelling, as well as improved lung function and reduced key pharmacodynamic biomarkers in bronchial tissues. These results reinforce the importance of IL-13 in airway pathobiology and suggest that neutralization of IL-13 may reduce asthmatic airway remodelling. An internal data monitoring committee reviewed safety data regularly throughout the trial.

| Outcomes
The pre-specified primary stereologically assessed efficacy endpoint was the placebo-corrected adjusted mean relative (per cent) change in the number of airway subepithelial (basal lamina plus submucosa) eosinophils per mm 2 of basement membrane from baseline to week 12. The placebo-corrected adjusted mean change was defined as the difference in adjusted mean changes between the lebrikizumab and placebo groups.
Pre-specified secondary stereologically assessed efficacy end-points, pharmacodynamics and exploratory remodelling-related stereologically assessed end-points were also evaluated (see Supplement for full list).

| Statistics
Due to our interest in exploring multiple parameters and subgroup analyses in this study, no formal hypothesis tests were performed.
Results were reported as point estimates and associated 95% CIs, with no adjustment for multiple comparisons. Details of sample size calculations, analysis populations, stratification and adjustment analyses are provided in the Supplement.

| Patients
Enrolment began on 6 November 2014, and the study was concluded on 13 October 2016. Patients already enrolled at the time All patients completed assessments up to and including week 12. One patient receiving four doses of lebrikizumab discontinued the study at week 16 (safety follow-up), withdrawing consent due to study termination by the sponsor.
The treatment arms in the ITT population were generally balanced with respect to baseline demographic and disease characteristics, though a racial imbalance was present ( Table 1)

| Sample quality
The tissues sampled by endobronchial biopsy met quality expectations, with 88% and 60% of biopsies passing criteria for lamina propria and epithelium, respectively (Table S2). Stereological precision was considered acceptable for subepithelial measurements, with coefficient of errors of <0.5 for 95% of subepithelial eosinophil counts and 93% of the associated basement membrane surface area measurements. Coefficient of errors for the epithelial measurements was <0.5 for 32% of epithelial eosinophil counts and 95% of the associated basement membrane surface area measurements.

| Primary efficacy end-point
The baseline mean subepithelial eosinophil count per mm 2 of base-   (Figure 3; Figure S2).

| Secondary efficacy end-points
No substantial overall or biomarker subgroup differences were ob-  (Table S3).
A clinically meaningful increase in placebo-corrected mean FEV 1 was observed in the FeNO-and blood eosinophil-high subgroups, with less pronounced increase in the overall population ( Figures S5A,B). A meaningful placebo-corrected decrease in mean FeNO levels was observed overall and in the periostin-high subgroup ( Figures S5C,D).

| Exploratory histological end-points
To account for the extracellular eosinophil peroxidase (EPO) signal present in some biopsies, a presumed consequence of eosinophil activation and secretory granule secretion, we pursued a non-stereological method to quantify mean signal intensity per image pixel of subepithelium or epithelium as a post hoc exploratory analysis. As

| Safety
Safety analyses were based on the safety-evaluable population and included the 12-week treatment period and 8-week safety follow-up. The proportion of patients with at least one AE was similar between the lebrikizumab and placebo arms (69.7% and 67.7%, respectively), and events were mostly of mild or moderate intensity (  No eosinophil-associated AEs or herpes infections were reported. There were no treatment-emergent elevations in peripheral blood eosinophils of grade ≥2 (>1500 cells/μL).
The baseline prevalence of ATAs was 1.6% (one of 62 patients).
Post-baseline prevalence of ATAs was 6.9% (two of 29 patients) in patients randomized to lebrikizumab treatment. The two ATApositive patients developed low ATA responses, and there was no apparent impact on safety and drug exposure.

| D ISCUSS I ON
CLAVIER was designed primarily to evaluate the effect of lebrikizumab on airway eosinophilic inflammation in patients with moderate-tosevere uncontrolled asthma. Research bronchoscopy was employed, providing a valuable opportunity to study the effect of lebrikizumab on airway remodelling. We found no difference in change in tissue eosinophil counts in the lebrikizumab-and placebo-treated groups, although this finding is limited by significant variability, insufficient F I G U R E 3 Mean relative (%) changes from baseline in number of airway subepithelial eosinophils per mm 2 of basement membrane at week 12 in the primary analysis population. Estimates were based on a linear model that used relative change from baseline in airway subepithelial eosinophils as the response variable and included terms for treatment, number of asthma exacerbations within 12 mo of study entry and baseline asthma medications. Relative change was defined as the absolute change from baseline to week 12 divided by the value at baseline. Placebo-corrected adjusted mean change is the difference in adjusted mean changes between the lebrikizumab and placebo groups. Boxplots of the corresponding unadjusted data are provided in Figure S2 sample size and a baseline imbalance between groups. However, as part of pre-specified secondary analyses, lebrikizumab reduced subepithelial fibrosis, a novel finding suggesting a role of IL-13 in a fundamental aspect of airway remodelling in human asthma. In addition, lebrikizumab improved lung function and reduced T2 biomarkers, suggesting that this dosing regimen had the intended pharmacological and physiological effects.
A wide range of median and variance estimates in tissue eosinophil numbers have been observed in asthma studies using different methods, including omalizumab (EG2 antibody using non-stereological image analysis) and mepolizumab (gradient purification). 24,25 Given the observed variability and treatment group size in this study, approximately 90% reduction in tissue eosinophils would have been needed to reliably be detected. The variability in baseline tissue eosinophils was tenfold higher than assumed for study design power calculations based on BOBCAT study results that used an eosinophil cationic protein antibody, EG2, which stains neutrophils as well. 7 Notably, neutrophils express eosinophil cationic protein but not EPO. 26 Anti-EPO monoclonal antibody MM25-82.2 was used to detect eosinophils, which is highly specific to eosinophils. 27 Antibody specificity differences may have contributed to the unexpected higher variability in tissue eosinophils observed in CLAVIER vs BOBCAT.
To ensure high-quality analyses, quality control on bronchoscopy methods and biopsies was imposed, which resulted in high subepithelial tissue quality. Epithelial tissue quality was lower due to epithelial denudation during the procedure and processing but was consistent with what was expected. Design-based stereology was used to rigorously quantify eosinophils and demonstrate high measurement precision, suggesting that patient differences contributed to result variability. Reportedly, this is the first application of design-based stereology in support of an asthma therapeutic clinical trial.
Recent mouse studies raise the possibility of different eosinophil subsets in asthmatic airways distinguished by activation state. 28 While this pre-specified stereological enumeration F I G U R E 4 Mean adjusted relative (%) changes from baseline in thickness of subepithelial collagen at week 12 in the primary analysis population. Estimates were based on a linear model that used relative change from baseline in thickness of subepithelial collagen as the response variable and included terms for treatment, number of asthma exacerbations within 12 months of study entry and baseline asthma medications. Relative change was defined as the absolute change from baseline to week 12 divided by the value at baseline. Placebocorrected adjusted mean change is the difference in adjusted mean changes between the lebrikizumab and placebo groups. Boxplots of the corresponding unadjusted data are provided in Figure S7 TA   (Table S4). This racial imbalance in subepithelial eosinophils is aligned with the overall treatment arm imbalance observed for tissue eosinophils and could potentially be confounding the primary and secondary end-point results. Randomization for treatment assignment was stratified by serum periostin level and baseline medication but not by blood eosinophil counts nor race. Stratification by blood eosinophil count may have avoided the observed imbalance in tissue eosinophils. The study did not incorporate medication dose counters and adherence to ICS could have impacted the results, though the overall FeNO levels over time in the placebo arm ( Figure S9) do not suggest a significant impact of adherence on this relatively short duration study.
In this randomized trial, the effect of lebrikizumab on tissue eosinophils was inconclusive due to baseline imbalances in tissue eosinophils between treatment groups, lack of full recruitment and reduced statistical power, and higher than anticipated variability in tissue eosinophil measurements. Lebrikizumab inhibited the IL-13 pathway, as demonstrated by changes in key pharmacodynamic biomarkers and was associated with improved lung function and reduced degree of subepithelial fibrosis, a measure of airway remodelling.