Robin J Svensson* (1), Jakob Ribbing* (1), Naoki Kotani (2), Michael Dolton (2), Shweta Vadhavkar (2), Dorothy Cheung (2), Tracy Staton (2), David F Choy (2), Wendy Putnam (2), Jin Jin (2), Nageshwar Budha (2), Mats O Karlsson (1,3), Angelica Quartino (2), Rui Zhu (2) *RJS and JR contributed equally to this work
(1) Pharmetheus, Sweden (2) Genentech, South San Francisco, USA (3) Department of Pharmaceutical Biosciences, Uppsala University, Sweden
Objectives: In early stage clinical development for severe asthma treatments, go/no go decisions can be guided by changes in biomarkers or functional endpoints (such as forced expiratory volume in 1 sec (FEV1)), instead of asthma exacerbations (which is often the registrational endpoint) which require a longer treatment duration due to rare events. Knowing which biomarkers and functional endpoints are the most relevant predictors of asthma exacerbation hazard could benefit decision-making for early stage asthma trials. Furthermore, if the relationship between the most relevant predictors and asthma exacerbations were to be quantified, it would allow predictions of asthma exacerbations from early stage data. The objective was to describe the longitudinal asthma exacerbations in patients as a function of baseline and time-varying biomarkers and functional endpoints using a population repeated time-to-event (RTTE) approach.
Methods: A RTTE model for asthma exacerbations was developed using 502 patients with uncontrolled severe asthma from study GB39242 (clinicaltrials.gov: NCT02918019). Patients were randomized in a 1:1:1:1 ratio to receive subcutaneous placebo or MSTT1041A (monoclonal antibody selectively targets the IL-33 receptor, ST2) as 70, 210, or 490 mg every 4 weeks (Q4W) for 52 weeks. The dependent variable was asthma exacerbations, included in the data without interval censoring. An asthma exacerbation was defined as new or increased asthma symptoms that resulted in treatment with systemic corticosteroids for at least 3 days with or without hospitalization/emergency department visit. The dataset included biomarkers and functional endpoints, treated as time-varying covariates, a) based on daily diary (peak expiratory flow and Asthma Daytime Symptom Diary (ADSD)-based symptom score, short-acting rescue medication use and nighttime awakenings) and b) collected during scheduled visits (asthma quality of life questionnaire (AQLQ), FEV1, fraction exhaled nitric oxide (FeNO), blood eosinophils and soluble ST2). Patient demographics, other baseline measurements and individually predicted MSTT1041A exposure were also included in the dataset.
Exponential, Weibull and Gompertz distributions were explored to describe the baseline hazard. Full random effects modeling (FREM)[1-3] was used to explore the effect of 20 baseline covariates on the hazard, followed by stepwise covariate modeling (SCM) to explore 9 exploratory time-varying covariates. Following SCM, the effect of MSTT1041A exposure on hazard was explored. A Markovian element and presence of increased hazard in relation to clinic visits (i.e. a ‘visit effect’ where asthma exacerbations were detected) were investigated. Models were selected based on difference in OFV (dOFV), RSEs, Kaplan-Meier visual predictive checks (VPCs) and posterior predictive checks (PPCs) of asthma exacerbation rate. The analysis was conducted in NONMEM 7.3 [4] assisted by PsN [5].
Results: A Weibull distribution best described the baseline hazard. A Markovian element of lower hazard following an asthma exacerbation event was included, with a time-dependent return to the baseline hazard. The number of asthma exacerbations in the past year and symptom score were the most influential baseline covariates. Symptom score (dOFV=83.7), short-acting rescue medication use (dOFV=33.5) and FEV1 (dOFV=6.9) were selected as time-varying covariates. MSTT1041A exposure was not statistically significant at the 5% level but was included as a conservative measure with dOFV=2.1, since it could not be regarded as irrelevant given the effect size; Exposure was included as a step function and predicted a 28.3% lower hazard when the predicted trough concentration was above the lower limit of quantification.
Conclusions: The final model described the observed data well according to VPCs and PPCs. Diary-based symptom score and short-acting rescue medication use as well as FEV1 were identified as important time-varying covariates for asthma exacerbation hazard; Individual changes in these covariates were the main drivers for the predicted differences in asthma exacerbation rate between treatment arms. The analysis using RTTE further demonstrates the utility of these biomarkers and functional endpoints as key predictors of exacerbation and provides a tool to assess drug effect in early clinical trials.
References:
[1] Karlsson MO. A full model approach based on the covariance matrix of parameters and covariates. PAGE 21 (2012) Abstr 2455 [www.page-meeting.org/?abstract=2455]
[2] Yngman G, Nyberg J, Jonsson EN & Karlsson MO. Practical considerations for using the full random effects modeling (FREM) approach to covariate modeling. PAGE 26 (2017) Abstr 7365 [www.page-meeting.org/?abstract=7365].
[3] Nyberg J, Jonsson EN, Karlsson MO, Häggström J & members of the HBGDki Community. Properties of the full random effect modelling approach with missing covariates. bioRxiv 2019 https://www.biorxiv.org/content/early/2019/06/07/656470.full.pdf
[4] Beal S, Sheiner LB, Boeckmann A & Bauer RJ. NONMEM Users Guides. 1989-2013. Icon Development Solutions, Ellicott City, Maryland, USA.
[5] Keizer RJ, Karlsson MO & Hooker A. Modeling and simulation workbench for NONMEM: Tutorial on Pirana, PsN, and Xpose. CPT Pharmacometrics Syst Pharmacol 2013; 2: e50
Reference: PAGE () Abstr 9341 [www.page-meeting.org/?abstract=9341]
Poster: Oral: Drug/Disease Modelling