Infections that do not respond to antibiotic therapy present challenges to public health. Phages are being widely studied as a potential non-antibiotic treatment for such infections. However, because phages replicate in the presence of their target, typical PK/PD analyses have limited ability to inform our use of them, particularly dosing.

The Hollow Fiber Model (HFM) is a 2-compartment in vitro model that we use to represent a localized infection and a circulatory system so that we can investigate which system parameters have the largest effect on experimental outcomes such as bacterial killing and phage resistance. The bacteria are confined to a 3.9 mL space filled with porous fibers that allow a continuous exchange of nutrients and phages between compartments. S. aureus populations are allowed to reach a steady-state concentration of approximately 1×10¹⁰ CFU/mL, then we add phages to mimic 1×10⁹ PFU being diluted into a 5 L adult blood volume. The circulating compartment and bacterial compartment are monitored for 3 weeks to quantify phage and bacterial populations.

Experiments were conducted at 37°C using a methicillin-sensitive strain (ATCC 19685) and a USA300 MRSA strain (NRS384). Our initial experiments tested the bactericidal effects of three different phage-dosing regimens when using a single phage: either phage K or a gp102 mutant that exhibited improved antibacterial activity on USA300 MRSA strains in previous studies. Each of these phages caused a transient 2-3 log₁₀ reduction in CFU/mL of ATCC 19685, which rebounded after 3 days; neither phage affected NRS384 populations. Phage concentrations were consistently higher in the bacterial versus the circulating compartment. We then used host-range and cross resistance data to design three phage cocktails, which we tested using the most intensive phage dosing regimen from the single-phage experiments. The cocktail experiments resulted in the same bactericidal outcomes as the single phage experiments. However, with the USA300 MRSA strain in particular, we saw a lot of variability in which individual phages became most abundant or persisted in the bacterial compartment, suggesting phage competition. Preliminary characterization of bacterial populations at different timepoints showed a mixture of phage-sensitive and phage-resistant cells. We are continuing to explore the effects of variables such as bacterial strain choice, phage treatment, dosing regimen, and media composition and flow rate and are working to relate these findings to observations from animal data and data from other in vitro models.