Development of an indirect response model-based meta-analysis (MBMA) framework for the description of viral antigen decline for siRNA-based anti-HBV therapies
Authors: Louis Sandra (1,2), Huybrecht T'Jollyn (1), Oliver Ackaert (1), An Vermeulen (1,2), Juan José Perez- Ruixo (1)
(1) Janssen Research & Development, Beerse, Belgium; (2) Laboratory of Medical Biochemistry and Clinical Analysis, Faculty of Pharmaceutical Sciences, University of Ghent, Ghent, Belgium
Objectives: Silencing/small interfering RNA (siRNA) represents a novel therapeutic modality in the field of hepatitis B virus (HBV) infection, aiming to degrade viral mRNA and inhibit the translation of HBV mRNA into HBV-related proteins (e.g. hepatitis B surface antigen (HBsAg)). Since persistent HBsAg seroclearance is associated with a state called ‘functional’ cure, HBsAg decline is considered an important parameter during anti-HBV therapy [1]. However, functional cure rates are low with the current state-of-the-art antiviral therapies (including interferon (IFN) and nucleoside analogue (NA) therapies) [2]. The main objective of this work was to investigate the dose-response relationship and compare efficacy of various siRNA therapeutics in a model-based meta-analysis (MBMA) framework, assessing both system- and drug-related parameters.
Methods: Data from a Boolean-based literature search were digitized using WebPlotDigitizer v4.6. Current focus was on both subcutaneously (SC) and intravenously (IV) administered siRNAs, formulated as liposomal nanoparticles, or conjugated to either cholesterol or N-acetylgalactosamine (GalNAc). Doses ranged from 0.03 to 9 mg kg-1 and all included studies were placebo controlled. The pharmacological HBV mice models expressed all clinically relevant HBV biomarkers (including HBsAg). All HBsAg profiles were baseline- and placebo-normalized. A combined kinetic-pharmacodynamic (KPD) indirect response model (IRM) MBMA framework was developed describing HBsAg profiles after siRNA administration. Model refinement was based on improvement of the objective function value (OFV), standard error of the estimates, goodness-of-fit (GOF) plots, and parsimony. Modeling and simulation analyses were conducted using nonlinear mixed-effects modeling in NONMEM version 7.4.0. The first-order (FO) estimation method was used. The exploratory and statistical analyses, diagnostic plots and postprocessing of NONMEM results were carried out in R version 3.4.1.
Results: Data from 13 siRNA therapeutics, originating from 10 studies and reporting HBsAg concentration-time data in mice, were identified. Twenty-five treatment arms (excluding placebo arms) consisting of a total of 237 observations originating from 154 mice were included in the analysis dataset. A one compartment KPD model with first-order loss rate (KDE) described the kinetics at the biophase [3]. HBsAg dynamics were characterized by a one compartment turnover model, balancing zero-order HBsAg production rate (kin) and first-order HBsAg elimination rate (kout). HBsAg baseline (HBsAg0) was assumed to correspond to the ratio between kin and kout (HBsAg0 = kin/ kout). As a consequence of baseline- and placebo-normalization, the kin was assumed to be equal to kout and HBsAg0 is equal to 1. Drug effect was characterized by an inhibitory Imax model, affecting kin [4].
Typical value of the system parameter kout is 0.8131 day-1 [RSE: 10.2%], with a between study variability (BSV) of 36.0 %. Compound-specific parameter estimates for KDE and IC50 were lumped by drug class based on the log-likelihood ratio test (LLRT), when RSE% were acceptable and individual fits were indicative of good fit. Typical KDE estimates ranged from 0.02635 day-1 to 0.3535 day-1, and were lowest for GalNAc-conjugated siRNAs compared to other siRNA classes. Typical IC50 estimates ranged from 5.841 ng to 8373 ng and were lowest for LNPs and highest for GalNAc-conjugated siRNAs. Our findings are in accordance with Brown et al., identifying a delayed but prolonged drug effect for GalNAc-conjugated siRNAs compared to LNP formulated siRNAs [5]. Additionally, LNP formulated siRNAs are characterized by a redistribution from liver back to plasma, which reduces the liver half-life, ultimately resulting in a more transient drug effect compared to GalNAc-conjugated siRNAs [6].
Conclusions: Aggregate study level HBsAg data after siRNA administration can be described using the presented KPD-IRM MBMA framework. It allows to make a quantitative comparison between different siRNA classes.
References:
[1] Song A, Lin X, Chen X. Functional cure for chronic hepatitis B: accessibility, durability, and prognosis. Virol J. 2021 Jun 3;18(1):114. doi: 10.1186/s12985-021-01589-x
[2] Loglio A, Lampertico P. How Durable Is Functional Cure (Hepatitis B Surface Antigen Loss) in Patients With Chronic Hepatitis B Treated With Current Antivirals? Hepatol Commun. 2020 Jan 2;4(1):5-7. doi: 10.1002/hep4.1476
[3] Jacqmin P, Snoeck E, van Schaick EA, Gieschke R, Pillai P, Steimer JL, Girard P. Modelling response time profiles in the absence of drug concentrations: definition and performance evaluation of the K-PD model. J Pharmacokinet Pharmacodyn. 2007 Feb;34(1):57-85. doi: 10.1007/s10928-006-9035-z
[4] Sharma A, Jusko WJ. Characteristics of indirect pharmacodynamic models and applications to clinical drug responses. Br J Clin Pharmacol. 1998 Mar;45(3):229-39. doi: 10.1046/j.1365-2125.1998.00676.x
[5] Brown CR, Gupta S, Qin J, Racie T, He G, Lentini S, Malone R, Yu M, Matsuda S, Shulga-Morskaya S, Nair AV, Theile CS, Schmidt K, Shahraz A, Goel V, Parmar RG, Zlatev I, Schlegel MK, Nair JK, Jayaraman M, Manoharan M, Brown D, Maier MA, Jadhav V. Investigating the pharmacodynamic durability of GalNAc-siRNA conjugates. Nucleic Acids Res. 2020 Dec 2;48(21):11827-11844. doi: 10.1093/nar/gkaa670
[6] Goel V, Gosselin NH, Jomphe C, Zhang X, Marier JF, Robbie GJ. Population Pharmacokinetic-Pharmacodynamic Model of Serum Transthyretin Following Patisiran Administration. Nucleic Acid Ther. 2020 Jun;30(3):143-152. doi: 10.1089/nat.2019.0841