Abstract
The phages used to treat bacterial infections in phage therapy are commonly chosen based on their abilities to form plaques on the infecting bacterium-on host range. In practice, phage therapy is not always successful, leaving room for improvement. Here, we use computational models to investigate whether some standard phage properties (burst size, lysis rate, adsorption rate constant, intrinsic decay rate, and growth rate) might serve as predictors of treatment success. As our measure of treatment success, we deviate from many other approaches by calculating the number of phages needed to suppress bacterial densities 100-fold in the short term, given that the patient's immune system is expected to regain control once bacterial numbers are reduced. Numerical analysis of single-phage trials across 2400 combinations of phage phenotypes reveals that, on average, adsorption rate constant and growth rate provide the most useful predictive values, decay rate provides some value, whereas burst size and lysis time offer essentially little or no value. Bacterial density is especially informative of the number of phages required for treatment. There is nonetheless often considerable variation around average behavior for a single phenotype. These results raise the possibility that the adsorption rate constant and growth rate may be especially important in phage therapy performance for both high and low bacterial densities. Given that therapeutic phages are often evolved in vitro for broad host ranges rather than for individual hosts, it should be considered that selection for broad host range may have a downside of compromising adsorption to and growth rate on individual bacterial hosts.