Generating Robust and Informative Nonclinical In Vitro and In Vivo Bacterial Infection Model Efficacy Data To Support Translation to Humans

Abstract

In June 2017, the National Institute of Allergy and Infectious Diseases, part of the National Institutes of Health, organized a workshop entitled "Pharmacokinetics-Pharmacodynamics (PK/PD) for Development of Therapeutics against Bacterial Pathogens." The aims were to discuss details of various PK/PD models and identify sound practices for deriving and utilizing PK/PD relationships to design optimal dosage regimens for patients. Workshop participants encompassed individuals from academia, industry, and government, including the United States Food and Drug Administration. This and the accompanying review on clinical PK/PD summarize the workshop discussions and recommendations. Nonclinical PK/PD models play a critical role in designing human dosage regimens and are essential tools for drug development. These include in vitro and in vivo efficacy models that provide valuable and complementary information for dose selection and translation from the laboratory to human. It is crucial that studies be designed, conducted, and interpreted appropriately. For antibacterial PK/PD, extensive published data and expertise are available. These have been leveraged to develop recommendations, identify common pitfalls, and describe the applications, strengths, and limitations of various nonclinical infection models and translational approaches. Despite these robust tools and published guidance, characterizing nonclinical PK/PD relationships may not be straightforward, especially for a new drug or new class. Antimicrobial PK/PD is an evolving discipline that needs to adapt to future research and development needs. Open communication between academia, pharmaceutical industry, government, and regulatory bodies is essential to share perspectives and collectively solve future challenges.

Keywords: best practices; drug development; hollow fiber system; in vitro infection models; mouse infection models; optimal design; pharmacokinetics/pharmacodynamics; progression and decision criteria; validation; workshop summary.

FIG 1

Dynamic one-compartment in vitro infection model (“chemostat”). Fresh medium is added continuously while culture contents are removed at the same rate to maintain a constant volume. (A) Chemostat model for simulating a monoexponential decline of drug concentrations after intravenous dosing; the antibiotic(s) is/are dosed into the central reservoir as bolus doses or zero-order infusions. (B) Chemostat for oral dosing, which can simulate drug concentration-time profiles with first-order absorption and elimination; typically, the antibiotic(s) is/are dosed into the antibiotic reservoir as bolus doses.

FIG 2

Dynamic two-compartment hollow fiber in vitro infection model. (A) Cross section of a hollow fiber cartridge. Many hollow fibers provide a large surface area (typically 0.2 to 0.3 m², depending on the cartridge). According to the molecular weight cutoff of the hollow fiber membrane, medium, drugs, oxygen, nutrients, bacterial metabolites (“waste products”), and other small molecules can exchange between the central circulation (which includes the interior of the hollow fibers) and the extracapillary space of the cartridge. In contrast, bacteria, other cells (if present), and large molecules are entrapped in the extracapillary space of the hollow fiber cartridge. (B) Flow of broth medium from the fresh broth to the central reservoir. From the latter, broth is circulated to the peripheral compartment (i.e., the extracapillary space of the hollow fiber cartridge) or is eliminated. Elimination occurs from the central reservoir into the waste broth reservoir. A high-precision dosing pump is used to dose drugs into the central circulation.

FIG 3

Overview of important variables which contribute to the outcome of animal infection models. These factors may need to be considered for study design and execution as well as for the data analysis and ultimate translation of rationally optimized regimens to patients. Tox, toxicity.

FIG 4

Different sources of variability that may affect the results of animal infection models. The between-system variability can be handled by appropriate choices for and the selection of experiments to be performed. The within-system variability can be split into a controllable portion and a random (i.e., usually noncontrollable) part. Experimental design choices and careful execution of animal infection model studies can minimize the controllable variability. The random, unexplained variability will necessarily include components such as between-subject variability (BSV) in pharmacokinetics, pharmacodynamics, the infection site, and the immune system.

FIG 5

Considerations and perspectives to enhance the robustness of animal infection models and ultimately better translate efficacious and reliable dosage regimens to patients.