Objectives: Shingles/herpes zoster (HZ) occurs in elderly and immunocompromised individuals when latent varicella zoster virus (VZV) is reactivated. Shingles causes painful rashes with blistering and can lead to post-herpetic neuralgia in severe cases [1]. The only shingles vaccine currently approved for use in the US is a two-dose, adjuvanted recombinant zoster vaccine (RZV). RZV was found to be 70% effective at preventing HZ for at least 10 years (>50-year-old cohort) [2]. While an immunological “correlate of protection” for vaccine efficacy is not established, cell-mediated immunity appears to be most predictive of protection from HZ [3]. Model-informed drug development (MIDD) approaches are yet to be routinely applied to inform vaccine development. For example, optimal vaccine dose and dose regimens for pivotal trials are currently informed by empirical approaches using sparse early phase data and could be better-informed by quantitative MIDD [4,5]. An end-to-end MIDD platform relating vaccine dose to persistence of key immunological markers and ultimately vaccine efficacy could significantly de-risk clinical development decisions by informing candidate selection and pivotal trial design. Here, as proof of concept, we used RZV clinical data [2,6-8] to demonstrate how modeling could aid in vaccine development.
Methods: We adapted a quantitative systems pharmacology (QSP) model of the immunogenic response to therapeutic monoclonal antibodies (mAbs) [9] to reproduce clinical immunogenicity data for RZV. Leveraging mAb immunogenicity models is a natural way to inform and accelerate vaccine immunogenicity model development [4]. Key model features include antigen presentation by dendritic cells, T-cell activation and memory, B-cell activation and memory, and the generation of antibody-secreting plasma cells. Furthermore, we demonstrate a framework to predict future HZ incidence rates [2,8] from vaccine-induced antibody titers, motivated by Probability of Disease modeling [5].
Results: A virtual population (VPop) was parameterized to match the immunogenic response to RZV including the dose and dose regimen response and associated variability up to 36-months post-vaccination [6,7]. The model was validated using RZV single-dose data [6], long-term persistence data [7], and dosing interval data [10]. Finally, we linked immunogenicity measures to future HZ incidence rates using published summary-level data [2,8] and demonstrated how this could be used to estimate vaccine efficacy from model projections.
Conclusions: The QSP vaccine immunogenicity model reproduces key features of the immunogenic response to RZV with the potential to predict HZ incidence rates. Such an MIDD framework could improve confidence in decision-making to accelerate the clinical development of novel vaccines.
Citations: [1] Kawai et al., BMJ Open, 2014 [2] Strezova et al., Open Forum Infect Dis., 2022 [3] Bharucha et al., Hum Vaccin Immunother, 2017 [4] Giorgi et al., CPT PSP, 2021 [5] Dudasova et al., NPJ Vaccines, 2021 [6] Chilbek et al., Vaccine, 2014 [7] Schwarz et al., Hum Vaccin Immunother, 2018 [8] Cunningham et al., J Infect Dis., 2018 [9] Chen et al., CPT PSP, 2014 [10] Poder et al., Open Forum Infect Dis., 2016
