Model-informed development of bacteriophage therapy: bridging in vitro and in vivo efficacy against multidrug-resistant Pseudomonas aeruginosa

Abstract

Bacteriophages are emerging as promising alternatives to antibiotics for multidrug-resistant (MDR) infections. However, their unique pharmacokinetic and pharmacodynamic (PKPD) properties arising from host-dependent amplification present challenges for dose selection and clinical translation. Here, we present a mechanistic PKPD model informed by in vitro kinetic assays and in vivo mouse studies of phage therapy targeting MDR Pseudomonas aeruginosa. The model extends the classical predator-prey model by addressing dormancy-related bacterial persistence and partitioning bacterial subpopulations based on phage susceptibility profiles. Simulations revealed a non-monotonous dose-exposure curve driven by dose-dependent reduction of phage replication and the importance of cross-resistance in selecting optimal phage cocktails. In vivo, host immunity was identified as a crucial component in inhibiting bacterial regrowth, with bistable outcomes dependent on initial bacterial load and immune competence. Dose-ranging simulations under varying immune statuses suggest that long-term bacterial load is solely determined by host immune function. However, higher doses transiently reduce bacterial load to a greater extent and thereby suppress immune activation. In immunocompetent hosts, phage cocktails can enhance maximal bacterial load reduction when administered at doses higher than a critical threshold. In conclusion, our PKPD framework enables optimal selection of phage cocktails and dosing regimens, supports rational design of first-in-human trials of phage therapy, and potentially advances model-informed drug development for replication-competent biologics.IMPORTANCEIn this study, we construct an integrative model of phage-bacteria dynamics and investigate whether its calibration to in vitro kinetic assay data can inform the rational design of phage therapy regimens and cocktails. Our findings demonstrate a dose range within which lower phage doses yield higher long-term exposure, presenting a fundamentally different framework for dose optimization. Analysis of phage cocktails reveals that combining phages with low cross-resistance delays the regrowth of phage-resistant bacteria in vitro. The extended in vivo model elucidates key differences between in vitro and in vivo outcomes and highlights the importance of the host's immune response in suppressing the growth of phage-resistant bacteria. Phage cocktails to combat phage resistance are therefore of less importance in immune-competent individuals but can enhance bacterial killing when administered at sufficiently high doses. We propose that this modeling framework holds potential for model-informed drug development by quantitatively characterizing bacteria-phage dynamics using preclinical data. Furthermore, it may facilitate the interpretation of in vivo therapeutic outcomes through a mechanistic understanding derived from in vitro observations.