Zufar Mulyukov, Amy Racine, Etienne Pigeolet
Novartis, Basel, Switzerland
Objectives: The progression of the wet form of age-related macular degeneration (wet AMD) and the effect of the anti-VEGF ranibizumab on disease progression in terms of best corrected visual acuity (BCVA) has been well characterized by drug-disease models [1-2]. The aim of this analysis is to develop a model describing the effect of ranibizumab on central retinal thickness (CRT) in anti-VEGF treatment naïve patients with wet AMD. CRT is one of few anatomical features used by ophthalmologists along with BCVA to decide on treatment (dis)continuation. The anti-VEGF effect on CRT is much faster than its effect on BCVA; to describe the quick onset accurately, retinal thickness assessments within first few days of treatment are needed.
Methods: We developed a nonlinear mixed effect (NLME) indirect response kinetic-pharmacodynamic (K-PD) [3] model to describe CRT change with time during treatment with ranibizumab. The effect of vitreous ranibizumab injection was modeled as inhibition of CRT production rate kin using Emax function of the drug concentration. The vitreous PK was described as linear elimination with half-life T1/2 = 9 days without inter-subject variability. The data from four clinical studies (PIER, MARINA, EXCITE and SUSTAIN) including approximately 800 patients in total are used to fit the model. The treatment arms included monthly, quarterly and pro re nata (PRN) intravitreal injections of 0.3 mg and 0.5 mg doses, as well as sham injections.
Results: The mean retinal thickness in sham treated arms did not change throughout the studies period. The effect of ranibizumab treatment on CRT was large, with a decrease of about 100 um from a baseline of approximately 320 um, and relatively fast with a decrease in thickness of about 60 um occurring within the first week after the first injection. Quarterly treatment arms demonstrated fluctuations in mean CRT between injections, leading to average loss of about half of the effect within three months after an injection.
Conclusions: The developed K-PD turnover model captured major features of CRT behavior under ranibizumab treatment in patients with wet AMD, and, together with BCVA models, may enable more accurate simulations of adaptive treatment regimens such as PRN dosing.
References:
[1] Lu, Tong, et al. “Population Analysis of wet-AMD Disease Progression and The Therapeutic Effect of Ranibizumab.” PAGE 2012. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
[2] Weber at al. “Bayesian Drug Disease Model with Stan – Using published longitudinal data summaries in population models.” PAGE 2014. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
[3] Jacqmin, Ph, et al. “Modelling response time profiles in the absence of drug concentrations: definition and performance evaluation of the K–PD model.” Journal of pharmacokinetics and pharmacodynamics 34.1 (2007): 57-85.
Reference: PAGE 25 (2016) Abstr 5918 [www.page-meeting.org/?abstract=5918]
Poster: Drug/Disease modeling - Other topics