If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Assistance Publique-Hôpitaux de Paris, Département Hospitalo-Universitaire Thorax Innovation, Service de Pneumologie, Hôpital Bicêtre, Université Paris-Sud, INSERM Unités Mixte de Recherche 999, Le Kremlin-Bicêtre, France
Department of Respiratory and Sleep Medicine, John Hunter Hospital, Hunter Medical Research Institute, Kookaburra Circuit, New Lambton Heights, NSW, Australia
Efficacy pharmacogenetics of mepolizumab treatment in patients with severe asthma was investigated.
•
A candidate polymorphism analysis and an unbiased genome-wide association analyses were performed.
•
No genetic effect on mepolizumab treatment response was identified.
Abstract
Background and objectives
Treatment with mepolizumab, a humanized monoclonal antibody to interleukin-5, reduces the rate of asthma exacerbations and the requirement for systemic glucocorticoids while maintaining asthma control. Treatment decisions are guided by predictors of response, including blood eosinophil thresholds in patients with frequent exacerbations despite intensive anti-inflammatory and controller treatment. Identification of additional predictors of response could aid treatment decisions. We investigated genetic associations that may predict response to mepolizumab-treatment.
Methods
In this post hoc analysis of DREAM and MENSA, association of genetic markers was tested in patients with severe asthma treated with mepolizumab who provided consent for pharmacogenetic research. Association was tested in a tiered approach with alpha spend differing for candidate genetic markers selected for prior history of association with relevant traits or pathways and in a genome-wide analyses (p < 4.7 × 10−4 and p < 5 × 10−8, respectively). Efficacy endpoints included: clinically significant exacerbation rate (tested using a negative binomial model), time to first exacerbation (tested with a Cox proportional hazards model), change in exacerbation rate, change in eosinophil count, and change in IgE level (tested by linear regression).
Results
No genetic marker was significantly associated with the primary endpoint, clinically significant exacerbation rate. One genetic marker was associated with time to first clinically significant exacerbation, but this association was driven by the DREAM data and was not supported in additional sensitivity analyses by treatment regimen/dose.
Conclusion
No genetic effect on mepolizumab-treatment response was identified in this population on intensive asthma treatment, with history of frequent exacerbations and pre-selected for airway eosinophilia.
Treatment with mepolizumab, a humanized monoclonal antibody to interleukin-5 (IL-5), reduces the rate of asthma exacerbations and the requirement for systemic glucocorticoids while maintaining asthma control [
]. Treatment benefit is guided by predictors of response based on specific phenotypic characteristics. These include blood eosinophil thresholds in patients with frequent exacerbations despite intensive anti-inflammatory and controller treatment. However, identification of additional predictors of response could aid with treatment decisions. In the Dose Ranging Efficacy and Safety with Mepolizumab (DREAM [
]) study, key phenotypic characteristics were associated with response to mepolizumab. These characteristics guided enrollment in the Mepolizumab as Adjunctive Therapy in Patients with Severe Asthma (MENSA [
]) study. In this post hoc analysis of DREAM and MENSA, we used clinical samples and data from these two studies to investigate potential genetic associations that may predict efficacy response to mepolizumab treatment in patients with severe eosinophilic asthma.
The intent-to-treat (ITT) study population included 1192 patients enrolled in two studies: DREAM (NCT01000506), n = 616, and MENSA (NCT01691521), n = 576. The exploratory pharmacogenetic (PGx) study sample included ITT patients who provided informed consent and blood samples for PGx investigations (DREAM, n = 589; MENSA, n = 464). To reduce confounding due to genetic admixture, primary PGx analyses were limited to DNA samples and clinical datasets from non-Hispanic white patients (DREAM, n = 468; MENSA, n = 352).
Efficacy endpoints used for PGX analyses were the clinical study primary endpoint, clinically significant exacerbation rate (CSE; number of asthma exacerbations resulting in emergency room visit, hospitalizations, or at least 3 days of steroid use), time to first CSE, change in CSE rate, change in eosinophil count, and change in immunoglobulin E (IgE) level.
Two analysis approaches were used. Candidate genetic variants (105) from 53 gene regions previously linked to asthma, asthma severity and/or within candidate IL-5 and eosinophil pathway genes (Table E1) were analyzed. Other variants were tested in an exploratory genome-wide association study (GWAS). We selected genetic variants described in asthma or nasal polyp literature and in IL-5/eosinophil literature and reported in IL-5 pathways. This selection used a combination of systematic text mining via the Linguamatics I2E platform (Linguamatics Ltd, Cambridge, UK; www.linguamatics.com [
]) and pathway analysis using QIAGEN's Ingenuity Pathway Analysis (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity) and MetaCore™ from GeneGo (a Thomson Reuters business). DREAM samples were genotyped using the OmniExpressExome BeadChip (Illumina, San Diego, CA, USA), by Expression Analysis (Morrisville, NC, USA) and using KASP assays (LGC Genomics, Hertfordshire, UK). MENSA samples were genotyped using the Axiom Biobank Plus array (Affymetrix, Santa Clara, CA, USA) by BioProcessing Solutions Alliance (Piscataway, NJ, USA). Missing genotypes were imputed [
Associations with each genetic variant were analyzed using regression models, assuming an additive genetic model. Generalized linear regression, assuming a negative binomial outcome with log-link function, was used for CSE, Cox regression was used for time to first CSE, and normal linear regression was used for change in CSE, eosinophil count, and IgE. Three GWAS principal components, corticosteroid use and number of exacerbations prior to commencement of treatment were included in each regression model to account for population ancestry and clinical heterogeneity. Subject level data was analyzed separately for the two studies, then meta-analyzed.
The false-positive rates were set to 5%, with statistical significance declared at p < 4.7 × 10−4 for candidate genetic variant and p < 5.0 × 10−8 for GWAS analysis, respectively [
]. No further adjustments were made for the number of phenotypes analyzed.
Power (assuming an additive genetic model) was evaluated for changes in exacerbation rate ratio, ranging from 1.1 to 2.5, with variant frequencies between 10% and 50%. The candidate variant analysis had 80% power to identify differences in exacerbation rate ratios of ≥1.6, for common genetic variants with minor allele frequency >20% (power to detect smaller genetic effects and/or lower frequency alleles is limited). The power for this PGx study was estimated assuming an effect equivalent to or higher than the clinical study effect size for each trial (Fig. E1).
Key covariates and demographic factors such as age, sex, baseline eosinophil count, baseline oral corticosteroid use, or baseline exacerbation count in the 12 months prior to study enrollment are summarized for the ITT and PGx analysis samples, and by placebo- and mepolizumab-treated subgroups, for DREAM (Table E2A) and MENSA (Table E2B). These summaries suggest that the PGx cohort is representative of the ITT population in each study.
After multiple test adjustment in either candidate gene variant or GWAS analyses (Fig. E2), no variant was significantly associated with the primary endpoint, CSE.
However, genomic regions with association evidence just below the GWAS threshold for statistical significance (5 × 10−8 <p-value ≤ 10−6) were evaluated. Four variants in chromosome 6 and two variants mapping to chromosome 9 met this criterion. The four variants in chromosome 6 (rs114633080, rs137893217, rs78517277 and rs117220641) are in complete Linkage Disequilibrium (LD) so indistinguishable in effect. These variants are all intergenic. The nearest genes to these variant location include UTRN (which encodes a cytoskeleton protein, utrophin) and EPM2A (which encodes a phosphatase, laforin), but there is no clear biologic justification for involvement of either of these genes with asthma or mepolizumab response. The two variants on chromosome 9 also meeting this criterion (rs10811516 and rs10811517) are in complete LD, and map to an intergenic region flanked by type I interferon genes including IFNA14 and a pseudogene of unknown function, IFNA22P. Type I interferons are cytokines with antiviral, antiproliferative, and immunomodulatory properties. Group 2 innate lymphoid cells (ILC2) are involved in respiratory virus-associated pulmonary type 2 immunopathology. Although a specific role for interferon alpha 14 has not been defined in infection associated immune responses, other type I interferons directly regulate the activation of ILC2. In the recent publication of Duerr et al., 2016 [
], cultures of ILC2 cells produced less IL-5 when treated with a different type 1 interferon, Interferon β. Although this result is intriguing, a role for interferon alpha 14 in ILC2 cell activation is yet to be established.
Of the 105 candidate genetic variants, only 8 variants had weak evidence of association (p < 0.05, Fig. 1). No suggestive associations were found with change in eosinophils or IgE in either GWAS or candidate gene analyses. However, the intronic variant, rs1021621 (A > G), in POU2F1, was significantly associated with time to first CSE (p = 2 × 10−5) when both studies were meta-analyzed, with a 20% reduction in CSE rate in patients treated with mepolizumab (p = 0.007, not statistically significant). This POU2F1 variant was further tested in patient subsets by dose. This association was driven mainly by the DREAM data (p = 1.5 × 10−4) over the MENSA data (p = 0.037), and in particular, a small subset of patients who were treated with 750 mg (intravenously), the highest experimental dose (p = 0.002, n = 119). Given the small number of samples, this suggestive association will require replication in larger sample sets.
Fig. 1Candidate Variant/Gene Association Results – with p < 0.05 on the primary endpoint (CSE) with corresponding data on secondary endpoints.
In summary, we conducted a PGx efficacy analysis for mepolizumab-treatment response in patients with severe eosinophilic asthma with a history of frequent exacerbations. The available sample size, though the largest reported to date, only had statistical power to identify large and common genetic effects; this study was therefore not powered to detect small genetic effects or effects from rare variants. No PGx effects on mepolizumab-treatment response were identified. However, given that we had 80% power to detect common genetic effects (minor allele frequency >20%) in non-Hispanic white patients, similar in size to the overall treatment effect, it is possible that less common and/or small genetic effects on mepolizumab-treatment response will be missed.
Funding
This study was funded by GSK (GSK ID: 117395, 201318).
Author contributions
LC, MC, HO, NB, SY and SG were involved in the concept and design of the study. EH was involved in the acquisition of the data. All authors were involved in data analysis and interpretation, preparation and review of the manuscript and approved the final version to be submitted.
Conflict of interest
MC, NB, SY, and SG are employees of GlaxoSmithKline (GSK) and own stock/stock options in GSK. LC and EH are former employees of GSK and own stock/stock options in GSK. HO is a former employee of GSK. ERB has received grants from AstraZeneca, MedImmune, Boehringer Ingelheim, Pfizer, Cephalon/Teva, Forest, Genentech, GSK, Johnson & Johnson (Jansen), Novartis, Sanofi, and Regeneron. He has also served as a paid consultant for AstraZeneca, MedImmune, Boehringer Ingelheim, Pfizer, GSK, Forest, Novartis, Regeneron, and Sanofi. MH has received grants from GSK and Sanofi, and has received personal fees for consultancy/advisory boards from GSK, AstraZeneca, Novartis, Roche, and Sanofi. PJT has received consultancy fees from GSK, Novartis, CSL Behring, Boehringer Ingelheim, and AstraZeneca. PG has received grants and personal speaker's fees from GSK, Novartis, and AstraZeneca.
Acknowledgments
Editorial support (in the form of grammatical editing and styling) was provided by Elizabeth Hutchinson, PhD, of Fishawack Indicia Ltd, funded by GSK.
Appendix A. Supplementary data
The following is the supplementary data related to this article:
30358-X&id=gr1.jpg){kind=link}