A spatially resolved pharmacokinetic/pharmacodynamic model for bystander effects of the hypoxia-activated prodrug PR-104A

Abstract

Background: Tumour hypoxia can be exploited by hypoxia-activated prodrugs (HAPs), but most HAPs are activated only under extreme hypoxia whereas more moderately hypoxic cells probably confer resistance to radiotherapy and chemotherapy. The ability of active metabolites to diffuse locally from severely hypoxic zones (bystander effect) may therefore be critically important for their use. Here, a spatially resolved pharmacokinetic/pharmacodynamic (SR-PK/PD) model was developed, tested and used to evaluate the contribution of bystander effects to activity of the HAP PR-104A, which is reduced to cytotoxic PR-104H and PR-104M with half-maximum activation at 0.13 μM O2 (KO2). The corresponding phosphate ester, PR-104, releases PR-104A systemically and is currently in clinical trial. Methods: SR-PK/PD models were developed for SiHa and HCT116 tumours. Intra/extracellular partitioning, metabolism and potency of PR-104A, PR-104H and PR-104M were determined in single cell suspensions and diffusion coefficients were measured using multicellular layers (MCLs). Plasma PK of all compounds was determined in mice by LCMS/ MS. Spatial gradients of O2, PR-104A, PR-104H and PR-104M concentrations in tumours were calculated using a digitised tumour microvascular network, and used to predict average clonogenic cell kill, which was compared with measured values. Results: Single cell suspension and MCL data could be described by a transport model considering intra- and extracellular compartments, using different parameter sets for SiHa and HCT116. The SR-PK/PD model was broadly predictive of measured PR-104 antitumour activity and estimated that bystander effects contribute ~30% and ~50%, respectively, to killing in SiHa and HCT116 tumours. Simulations showed that higher bystander effects decrease activity with radiation but moderately increase monotherapy activity, with washout of active metabolites into the circulation increasingly limiting activity. The model suggests that low-KO2 HAPs should have high activation rate constants and relatively short-range bystander effects for optimal complementation of radiotherapy. Moreover, it indicates that macroregional heterogeneity in hypoxia and HAP reductase expression in tumours limits killing because bystander effects operate on smaller spatial scales. Conclusions: The PR-104 SR-PK/PD model improved our understanding of the pharmacology of HAPs with implications for other targeted anticancer prodrug approaches, and identified potential strategies to optimise HAPs and their clinical application.