Patrick Nolain (1), Elisa Calvier (1), David Fabre (1), Jean-Marie Martinez (1), Aurélie Brunet (1), Sonia Khier (2)
(1) Sanofi, Montpellier, France (2) University of Montpellier, France
Objectives:
Alirocumab is a cholesterol-lowering monoclonal antibody targeting proprotein convertase subtilisin kexin type 9 (PCSK9) indicated in the prevention of cardiovascular risk. We aimed to develop and qualify a population pharmacokinetic/pharmacodynamic (PKPD) model to simultaneously describe the interaction of alirocumab with PCSK9 and its impact on the fluctuation of LDL-cholesterol (LDL-C) levels, taking into account the mechanistic Target-Mediated Drug Disposition (TMDD) process previously described [1]. The model was developed from adult patients and healthy volunteers’ data and was later used to fit pediatric data using a prior approach.
Methods:
This model was developed using NONMEM version 7.4.1, on nine clinical studies (Phase I/II/III studies, N=527) of the clinical development program of alirocumab. Potential covariate-parameter relationships were investigated using a Full Fixed Effect Modeling approach (FFEM [2]) to look for clinically relevant associations. Predictive ability was validated using four further clinical studies (Phase II/III, N=2271) using a Maximum A Posteriori Bayesian approach. Subsequently, a model for pediatric patient (N=42, from 8 to 17 years of age) was built using the model without covariates as priors with the $PRIOR subroutine and testing weight as covariate on each of the model parameters.
Results:
The PK submodel consists of a TMDD model considering the quasi-steady-state approximation to characterize the interaction dynamics of alirocumab and PCSK9. The PD submodel consists of an indirect pharmacodynamic model describing the inhibition of LDL-C using PCSK9 concentrations, taking into account endogenous nature of PCSK9 and the initial hypothesis of constant LDL-C and PCSK9 levels before treatment. FFEM decision criteria for selection of covariate-parameter relationships were (i) a variation of +/- 20% of interdose area under the curve of concentrations of alirocumab at steady-state and of (ii) LDL-C decrease after twelve weeks of treatment. Similarly to previous findings [1,3], co-administration of a statin was found to increase the central volume of distribution (Vc, 1.75-fold higher with statin) of alirocumab and to increase the maximal inhibitory effect (Imax, 14% higher with statin) of the degradation of LDL-C by PCSK9. Also, baseline level of PCSK9 was linked to an increase in the IC50 parameter of the PD effect. Numerical and graphical quality criteria of the estimation demonstrated equivalent predictive performance of this simultaneous model than the sequential PKPD method employed initially [3]. The pediatric model included allometry on the volumes of distribution and alirocumab linear clearance (exponents of 1 and 0.95 respectively).
Conclusions:
This semi-mechanistic TMDD PKPD model integrates the dynamics of interaction between alirocumab and PCSK9 and accurately predicts alirocumab, PCSK9 and LDL-C concentrations in healthy subjects and patients. It allowed for a reduction in the number of included covariates without deteriorating fitting quality. Its allometerized extension was further successfully applied to a pediatric population.
References:
[1] Djebli et al., Clinical Pharmacokinetics 56, 1155-1171 (2017)
[2] Gastonguay, The AAPS Journal 2004 (6), S1, W4354.
[3] Nicolas et al., Clinical Pharmacokinetics 58, 115-130 (2019)
Reference: PAGE () Abstr 9499 [www.page-meeting.org/?abstract=9499]
Poster: Drug/Disease Modelling - Other Topics