W.E.A. de Witte (1), K. Danneels (1), B. Buday (2), M.L. Sargentini-Maier (1), A. Brochot (1)
(1) Ablynx, a Sanofi Company, Ghent, Belgium, (2) Sanofi R&D, Frankfurt, Germany
Introduction: Systemic Lupus Erythematosus (SLE) is an autoimmune disease that can affect multiple organs, including skin and kidney. Vobarilizumab, a half-life extended anti-IL6R NANOBODY® drug was tested clinically as a treatment for SLE. This study was a 48-week, randomized, double-blind, placebo-controlled, dose-range finding phase II study of Vobarilizumab administered subcutaneously on top of standard of care in subjects with moderate to severe active, seropositive SLE. The aim of the study was to assess the efficacy, safety, pharmacokinetics, pharmacodynamics, immunogenicity, flare rate, steroid reduction and health-related quality of life, with different dose regimens of Vobarilizumab. SLEDAI-2K (Systemic Lupus Erythematosus Disease Activity Index 2000) and BILAG (British Isles Lupus Assessment Group) disease activity scores were determined every 4 weeks. mSRI (modified SLE Responder Index, based on SLEDAI-2K and BILAG scores) was analyzed as response criterium for this analysis.
Objectives:
- To characterize the exposure-response (ER) relationship for the effect of Vobarilizumab on the time course of the modified SLE responder index (mSRI) in subjects with moderate to severe active seropositive SLE;
- To identify the potential covariates that may affect the ER relationship and enable the characterization of subpopulations with distinct drug effect profiles.
Methods: An exposure-response dataset was constructed with exposure metrics per individual and per timepoint as predicted by a previously developed PK model, mSRI response scores, and measured baseline and time-varying covariates that could relate to disease progression or exposure-response relationship. These data were fitted to a newly constructed continuous time Markov model for SLE with dropout, treatment failure and missing records models in NONMEM version 7.3. The model structure was based on published modelling of American College of Rheumatology’s (ACR) response scores in Rheumatoid Arthritis patients.1 Model performance was evaluated with visual predictive checks for mSRI scores over time, missing data over time, mSRI score transition frequency over time and tailor-made model accuracy and precision diagnostic plots over time. Graphical exploration of possible model covariates was performed using R software, version 3.3.1.
Results: A continuous time Markov model was developed, which described the mSRI time course data across dose groups well. Additional missing data models were successfully identified for dropout, treatment failure and missing mSRI records. No exposure-response relationship could be established on the mSRI score transition rates. The final model included time, region and the baseline modified SLEDAI-2K score as covariates for the transition rate constant from mSRI 4-5 to mSRI 5, time and last observed mSRI score as covariate on the probability of missing mSRI records and last observed mSRI score as covariate on the probability of dropout records. Parameter estimates had relative standard errors below 40%, except for 2 covariates (region and baseline modified SLEDAI-2K had 54% and 82% RSE, repsectively), and eta shrinkage was below 40%.
Conclusions: The developed model provided an accurate description of the mSRI data and their time course as well as the probability of dropout, treatment failure and missing mSRI records over time. No exposure-response or dose-response relationship could be identified and time, region and the baseline modified SLEDAI-2K score were included as covariates. The developed diagnostic plots for accuracy and precision are helpful to evaluate the model fitting of inter individual variability.
References:
[1] Lacroix DB, Karlsson M, Friberg LE. Simultaneous Exposure–Response Modeling of ACR20, ACR50, and ACR70 Improvement Scores in Rheumatoid Arthritis Patients Treated With Certolizumab Pegol. CPT: Pharmacometrics & Systems Pharmacology 3, e143 (2014).
Reference: PAGE 29 (2021) Abstr 9611 [www.page-meeting.org/?abstract=9611]
Poster: Drug/Disease Modelling - Other Topics