Can we predict who will benefit most from biologics in severe asthma? A post-hoc analysis of two phase 3 trials

Abstract

Background: Individualized prediction of treatment response may improve the value proposition of advanced treatment options in severe asthma. This study aimed to investigate the combined capacity of patient characteristics in predicting treatment response to mepolizumab in patients with severe asthma.

Methods: Patient-level data were pooled from two multinational phase 3 trials of mepolizumab in severe eosinophilic asthma. We fitted penalized regression models to quantify reductions in the rate of severe exacerbations and the 5-item Asthma Control Questionnaire (ACQ5) score. The capacity of 15 covariates towards predicting treatment response was quantified by the Gini index (measuring disparities in treatment benefit) as well as observed treatment benefit within the quintiles of predicted treatment benefit.

Results: There was marked variability in the ability of patient characteristics to predict treatment response; covariates explained greater heterogeneity in predicting treatment response to asthma control than to exacerbation frequency (Gini index 0.35 v. 0.24). Key predictors for treatment benefit for severe exacerbations included exacerbation history, blood eosinophil count, baseline ACQ5 score and age, and those for symptom control included blood eosinophil count and presence of nasal polyps. Overall, the average reduction in exacerbations was 0.90/year (95%CI, 0.87‒0.92) and average reduction in ACQ5 score was 0.18 (95% CI, 0.02‒0.35). Among the top 20% of patients for predicted treatment benefit, exacerbations were reduced by 2.23/year (95% CI, 2.03‒2.43) and ACQ5 score were reduced by 0.59 (95% CI, 0.19‒0.98). Among the bottom 20% of patients for predicted treatment benefit, exacerbations were reduced by 0.25/year (95% CI, 0.16‒0.34) and ACQ5 by -0.20 (95% CI, -0.51 to 0.11).

Conclusion: A precision medicine approach based on multiple patient characteristics can guide biologic therapy in severe asthma, especially in identifying patients who will not benefit as much from therapy. Patient characteristics had a greater capacity to predict treatment response to asthma control than to exacerbation.

Trial registration: ClinicalTrials.gov number, NCT01691521 (registered September 24, 2012) and NCT01000506 (registered October 23, 2009).

Keywords: Biologics; Mepolizumab; Prediction; Severe asthma; Treatment response.

Fig. 1

Importance of individual standardized covariates for predicting the reduction of severe exacerbations and the 5-item Asthma Control Questionnaire (ACQ5) score with 1-year mepolizumab treatment. Left panel, reduction in rate of severe exacerbations in 365 days of follow up. Right panel, reduction in ACQ5 scores in 365 days of follow up. The x-axis shows the absolute value of the regression coefficient for each covariate. BMI, body mass index, FEV_1, forced expiratory volume at 1 s, FVC, forced vital capacity, OCS, oral corticosteroids

Fig. 2

Lorenz curve for the heterogeneity of treatment benefit comparing 75 mg mepolizumab versus placebo in severe asthma. Left panel, predicted reduction in rate of severe exacerbations in 365 days of follow up. Right panel, predicted reduction in ACQ5 scores in 365 days of follow up. The Gini Index captures the level of inequality in predicted treatment benefit

Fig. 3

Averaged treatment effect comparing mepolizumab 75 mg versus placebo in severe asthma across quintiles of predicted treatment benefit. Left panel, observed treatment benefit in terms of average reduction in the rate of severe exacerbations per year; Right panel, observed treatment benefit in terms of average reduction in ACQ5 scores