Abstract
The spread of multi-drug resistant (MDR) bacteria is a global public health crisis. Bacteriophage therapy (or “phage therapy”) constitutes a potential alternative approach to treat MDR infections. However, the effective use of phage therapy may be limited when phage-resistant bacterial mutants evolve and proliferate during treatment. Here, we develop a nonlinear population dynamics model of combination therapy that accounts for the system-level interactions between bacteria, phage and antibiotics for in-vivo application given an immune response against bacteria. We simulate the combination therapy model for two strains of Pseudomonas aeruginosa, one which is phage-sensitive (and antibiotic resistant) and one which is antibiotic-sensitive (and phage-resistant). We find that combination therapy outperforms either phage or antibiotic alone, and that therapeutic effectiveness is enhanced given interaction with innate immune responses. Notably, therapeutic success can be achieved even at sub-inhibitory concentrations of antibiotics, e.g., ciprofloxacin. These in-silico findings provide further support to the nascent application of combination therapy to treat MDR bacterial infections, while highlighting the role of innate immunity in shaping therapeutic outcomes.
Footnotes
In addition to minor changes to the text, we have made the following major changes: 1. Substantially expanded our discussion of the generality of the link between susceptibility to phage and antibiotic resistance mechanisms given binding to components of efflux systems, not only with P.a. but within other microbes as well. We hope that this discussion makes it self-evident that the modeling context is broadly relevant. 2. Included multiple new modeling analyses that confirm our core findings of an essential synergy between antibiotics, phage, and the immune system, while enhancing the robustness of the findings. To do so we evaluated the robustness of findings given variation in both the levels of antibiotics and state of the immune system as well as via a parameter sensitivity analysis.
