hide
Free keywords:
drug therapy; antibiotic resistance; linear discriminant analysis; superinfection; antimicrobial resistance; antibiotics; optimization; mathematical models
Abstract:
Author summary For life-threatening infections, antibiotics need to be administered as soon as possible. Because it takes time to acquire data about the disease causing bacteria, the immediate treatment is often empiric. In particular, there are three treatment strategies discussed in the field of empiric treatment: cycling, mixing, and combination therapy. Despite a number of clinical and theoretical studies, it still remains unclear which treatment strategy best prevents the emergence of resistance and why. To address this controversy, we present a mathematical model capturing both mono- and multi-drug therapies. We sample and analyze a large parameter space to assess the effect of parameters on treatment success, and determine which treatment strategy is the best under which circumstances. Using methods such as linear discriminant analysis and particle swarm optimisation, we find that combination therapy outperforms the other strategies by a large margin for most of the biologically relevant parameter space. We also show that the rate of de novo emergence of double resistance and the costs of resistance mutations are the most important parameters determining whether combination therapy succeeds over the others.
