Clémence Boivin-champeaux 1, Stephan Schmidt 1, Francine Johansson Azeredo 1, Scott Balsitis 2, Justin Feigelman 3
1 University Of Florida (Orlando, United States), 2 Stealth mode biotech (San Mateo, United States), 3 Gilead Sciences (Foster City, United States)
Objectives: Chronic hepatitis B (CHB) remains a major global health burden, and functional cure (sustained HBsAg loss off therapy) is rarely achieved with current therapies. Building on disease-progression modeling principles outlined in our HBV tutorial [1], we established a mechanistic multi-compartment QSP model for acute hepatitis B (AHB) [2] with a structure designed to extend to chronic HBV. We aimed to (i) qualify the AHB model using a virtual population (VPop) to reproduce inter-individual variability in viral and biomarker kinetics and (ii) apply the same platform to immune-tolerant CHB by constructing an immune-tolerant VPop baseline and evaluating nucleos(t)ide analogue (NA) treatment response against published trial mean profiles in normal-ALT/high-viremia cohorts [3,4].
Methods: An ODE model was implemented in MATLAB/SimBiology with liver, plasma, and lymph nodes, integrating hepatocyte infection and biomarker generation (HBV DNA, HBsAg/HBeAg), inter-compartment transport, and immune modules for dendritic-cell sensing/priming, CD8 effector differentiation/trafficking, CD4 help and regulation (Tregs), and B-cell/plasmablast dynamics for antibody production. For AHB, literature-informed parameters were assigned where available, and remaining uncertain parameters were manually calibrated to qualitatively reproduce clinical kinetics across HBV DNA, ALT, and serologic markers. Performance was then evaluated quantitatively using predefined kinetic metrics (e.g., peak magnitude, time-to-peak, and decline slopes) computed from clinical time-course data and matched to simulations. Mechanistic credibility was assessed with perturbation analyses (pathway knockouts and increased Treg differentiation), followed by local sensitivity analysis (LSA) to identify dominant drivers and prioritize parameters for variability. An AHB VPop was constructed by sampling LSA-prioritized parameters within physiological/biological ranges and selecting virtual patients meeting metric-based acceptance criteria. For immune-tolerant CHB, a separate VPop was generated by sampling chronic-relevant parameters and accepting virtual patients that satisfied steady-state plateau constraints and immune-tolerant biomarker bounds at a defined baseline (~day 4000), then applying NA effects via exposure-driven Emax inhibition of viral replication and comparing simulated mean trajectories to trial mean profiles [3,4]. A peginterferon-α mechanism module was implemented; however, its evaluation in immune-tolerant CHB was constrained by the limited availability of peginterferon-α monotherapy data in this cohort.
Results: The AHB model reproduced hallmark features of acute infection: exponential viremia rise to peak HBV DNA, immune-mediated clearance with a transient ALT flare, and resolution with ALT normalization and hepatocyte recovery. Simulated hepatocyte loss at peak infection remained limited (up to ~10.8%), consistent with preservation of essential liver function, and subviral particle production exceeded infectious virions by 100- to 10,000-fold. Delayed CD4+/CD8+ activation emerged from lymph priming dynamics; effector CD8+ timing aligned with HBV DNA decline and ALT elevation, while antibody responses emerged later and stabilized post-clearance. Cytokine dynamics were biologically plausible, including low IFN-α with transient spikes and a transient IL-10 elevation near peak viremia. Perturbation analyses produced clinically interpretable “failure-to-clear” phenotypes: loss of mDC priming or CD8 differentiation eliminated clearance; impaired B-cell/TFH processes yielded partial decline followed by stabilization; increased Treg differentiation produced low-level persistence with normal ALT. LSA highlighted adaptive immune drivers (CD8 differentiation; B-cell/plasmablast maturation and antibody production) alongside key viral replication parameters, and the resulting AHB VPop captured both central tendencies and inter-individual variability in HBV DNA, HBsAg, and ALT across clinical datasets. The immune-tolerant CHB VPop met steady-state and biomarker-bound criteria at baseline and maintained stable pre-treatment antigenemia. NA simulations reproduced the immune-tolerant pattern of strong HBV DNA suppression with comparatively limited HBsAg change, consistent with published trial profiles [3,4].
Conclusions: A VPop-qualified multi-compartment QSP model links HBV life-cycle biology and immune regulation to clinically measured biomarkers and captures both resolving AHB trajectories and chronic persistence phenotypes. Using a shared structural platform enables coherent extension from acute qualification to immune-tolerant natural history and NA response evaluation, providing a foundation for future phase-specific reparameterization and assessment of emerging therapies and combinations toward functional cure.
References:
[1] Boivin-Champeaux C, Schmidt S, Balsitis S, Johansson Azeredo F, Feigelman JS. A Quantitative Systems Pharmacology (QSP) Model of Acute Hepatitis B Virus Infection: Mechanistic Insights and Foundations for Future Extensions. CPT Pharmacometrics Syst Pharmacol. 2026;15(2):e70172. doi:10.1002/psp4.70172.
[2] Boivin-Champeaux C, Velez de Mendizabal N, Jones A, Balsitis S, Schmidt S, Feigelman JS, Johansson Azeredo F. Disease Progression Mathematical Modeling With a Case Study on Hepatitis B Virus Infection. CPT Pharmacometrics Syst Pharmacol. 2025;14(3):420–434. doi:10.1002/psp4.13298.
[3] Rosenthal P, Ling SC, Belle SH, et al. Combination of Entecavir/Peginterferon Alfa-2a in Children With Hepatitis B e Antigen-Positive Immune Tolerant Chronic Hepatitis B Virus Infection. Hepatology. 2019;69(6):2326–2337. doi:10.1002/hep.30312.
[4] Chan HLY, Chan CK, Hui AJ, et al. Effects of tenofovir disoproxil fumarate in hepatitis B e antigen-positive patients with normal levels of alanine aminotransferase and high levels of hepatitis B virus DNA. Gastroenterology. 2014;146(5):1240–1248. doi:10.1053/j.gastro.2014.01.044.
Reference: PAGE 34 (2026) Abstr 11932 [www.page-meeting.org/?abstract=11932]
Poster: Drug/Disease Modelling - Infection