Xiao Hu (1), Yaming Hang (1), Yue Cui (1), Shifang Liu (1), Jie Zhang (1), Ivan Nestorov (1)
(1) Biogen Idec Inc
Objectives: To establish a population pharmacokinetic (PK) model of peginterferon beta-1a (PEG-IFN) in relapsing remitting multiple sclerosis (RRMS) patients and develop the relationship between PEG-IFN exposure and annualized relapse rate (ARR).
Methods: PK and ARR data were obtained from a double-blind placebo-controlled Phase 3 study in RRMS patients (n=1512), in which 125 mcg subcutaneous PEG-IFN every 2 (Q2W) or 4 (Q4W) weeks reduced ARR (primary endpoint) significantly [1]. Using post-hoc PK parameters derived from a population PK model developed using NONMEM [2], PEG-IFN exposure was represented by monthly cumulative AUC for each subject. Four models were tested to describe the distribution of relapse counts, including Poisson, zero-inflated Poisson, log-normal Poisson, and Poisson gamma (equivalent to negative binomial) models. Covariates screened for the exposure-response model included baseline relapse rate in the past 3 years, age, Expanded Disability Status Scale, sex, McDonald Criteria, T2 lesions at Week 24, gadolinium-enhanced (Gd+) lesions at Week 24. Parameters were estimated using Bayesian analysis with Gibbs sampling in WinBUGS v1.4.3 [3]. Non-informative priors were used for parameter estimates.
Results: The relapse count was best described by a Poisson gamma model. The relationship between monthly cumulative AUC and ARR was well described using a log-linear model. In general, the ARR decreased as cumulative AUC increased. The slope for ARR reduction was steep in the Q4W AUC range, especially at below median AUC. In contrast, the slope started to level off in the Q2W AUC range. The model demonstrated that the better efficacy of the Q2W dosing regimen as compared with the Q4W dosing regimen was driven by its greater PEG-IFN exposure. Among all covariates tested, both T2 lesion volume and Gd+ lesion volume at 24 week improved model prediction, indicating the efficacy observed in MRI lesion at Week 24 was related to the ARR at Year 2. Inclusion of Gd+ lesions improved the model prediction better than the T2 lesion.
Conclusions: The model suggested that greater PEG-IFN exposure in the Q2W group explain the enhanced efficacy, as accounted by ARR, observed for the Q2W group, as compared to the Q4W group.
References:
[1] Calabresi PA, Kieseier BC, Arnold DL, Balcer LJ, Boyko A, Pelletier J, Liu S, Zhu Y, Seddighzadeh A, Hung S, Deykin A; ADVANCE Study Investigators, Pegylated interferon β-1a for relapsing-remitting multiple sclerosis (ADVANCE): a randomised, phase 3, double-blind study. Lancet Neurology (2014), 13:657-65.
[2] Beal SL, Sheiner LB, Boeckmann AJ & Bauer RJ (Eds.) NONMEM Users Guides. 1989-2011. Icon Development Solutions, Ellicott City, Maryland, USA. [3] Lunn DJ, Thomas A, Best N, and Spiegelhalter D, WinBUGS – a Bayesian modelling framework: concepts, structure, and extensibility. Statistics and Computing (2000), 10:325-337.
[3] Lunn DJ, Thomas A, Best N, and Spiegelhalter D, WinBUGS – a Bayesian modelling framework: concepts, structure, and extensibility. Statistics and Computing (2000), 10:325-337.
Reference: PAGE 24 (2015) Abstr 3649 [www.page-meeting.org/?abstract=3649]
Poster: Drug/Disease modeling - CNS