Optimizing polymyxin antibiotics using combination therapy and mechanism-based models
Ly, Neang S.
MetadataShow full item record
The emergence of multidrug-resistant (MDR) Gram-negative 'superbugs' coupled with a rapid decline in the development of new antibiotics has led to a serious, global health crisis. Polymyxin antibiotics, which include polymyxin E (colistin) and polymyxin B, were introduced over 50 years ago and were seldom used due to nephrotoxicity. However, recently, the polymyxins have re-emerged in the clinic as an antibiotic of last-resort. There are increasing reports of polymyxin resistance and failure during monotherapy with polymyxin. While polymyxin based combination therapies are a promising alternative, there are significant gaps regarding optimal use of polymyxin combinations. The primary goal of this dissertation is to rationally optimize polymyxin combination therapy based on pharmacokinetic/pharmacodynamic (PK/PD) principles. Mechanism-based PK/PD models (MBMs) were employed to evaluate mechanisms of antibiotic synergy. A secondary goal was to define the mechanisms of polymyxin tolerance and resistance utilizing genetically modified knockout strains and develop MBMs that quantify the impact of two-component regulatory systems and quorum sensing on antibiotic PD. Combination therapy has become a common practice for treating a number of diseases including cancer, arthritis, and certain infections (e.g., HIV and tuberculosis). However, there is a paucity of data to guide optimal combination therapies for the vast majority of bacterial infections. In Chapters 3, 4, 5, and 8, we use different antibiotic combinations to target multidrug-resistant P. aeruginosa strains with different phenotypes. In Chapter 3, the combination of colistin and doripenem was investigated against three P. aeruginosa strains to understand the type and time-course of interaction via a MBM. The final MBMs contained both subpopulation and mechanistic synergy with up to three bacterial subpopulations that had different susceptibility to colistin and doripenem. Our findings may serve as a framework for designing and optimizing new antibiotic combination therapies. Furthermore, in Chapter 4, we extended the in vitro PD model results from Chapter 3 by incorporating patient PK to project the efficacy of doripenem and colistin combinations in critically ill patients. Our simulations suggested that clinical doses of colistin monotherapy were insufficient against P. aeruginosa infections with a high bacterial density, while colistin combined with doripenem resulted in rapid bacterial killing. High colistin concentrations in combination with doripenem were essential to maximize the benefits of combination regimens. There are multiple bacterial mechanisms that may affect the PD of antibiotics. These include pathways involved in spontaneous mutations, two-component regulatory systems, and quorum sensing regulators. We evaluated the influence of different resistance mechanisms on the PD of polymyxins and other antibiotics via in vitro experiments using bacterial strains with specific genetic mutations and via novel mathematical models. First, we defined the impact of 'mutator' phenotypes of P. aeruginosa (deficient in genes critically involved in DNA repair mechanism) on polymyxin activity by evaluating the PD of polymyxin B monotherapy and combinations. We characterized bacterial killing and resistance via MBM PD models (Chapter 5) and showed the benefit of polymyxin-based combinations against this 'mutator' P. aeruginosa strain. We observed that therapy with a high initial dose of polymyxin B ('short-term burst') in combination with doripenem is beneficial to rapidly eradicate this mutator strain and minimize resistance. Second, we assessed the impact of deficiency in the PhoP or PmrA regulatory system on the PD of polymyxins (Chapter 6). The PhoP or PmrA regulatory system has been previously reported to be an important determinant of resistance to polymyxins. We found a minimal change in colistin (i.e. polymyxin E) activity between the wild-type and knockout strains, suggesting that other regulatory systems are primarily responsible for developing resistance to colistin. Next, we investigated the impact of quorum sensing on polymyxin PD at different initial bacterial inocula and studied a quorum sensing inhibitor (azithromycin) in combination with polymyxin B (Chapters 7 and 8). Polymyxin B displayed greater killing against quorum sensing deficient strains at high bacterial densities, suggesting that quorum sensing contributes to polymyxin tolerance at high bacterial densities. The combination of azithromycin and polymyxin B did not enhance killing against the wild type P. aeruginosa PA01 strain. This suggested that inhibition of quorum sensing by azithromycin was incomplete, which may be regulated by alternative pathways. Further studies are warranted to determine the impact of quorum sensing inhibitors on the activity of polymyxin B and other antibiotics to support the potential future clinical use of such combinations. An important theme throughout this dissertation and specifically in Chapters 3, 5, and 7 was the development of novel mechanism-based PK/PD models to elucidate the mechanisms of antibiotic interaction and to define the impact of bacterial regulatory systems on antibiotic PD. Two approaches for targeting multi-drug resistant pathogens were investigated: traditional combinations of two antibiotics and combinations of an antibiotic and a sensitizing agent targeting quorum sensing. Genetically modified knockout strains were used to determine the impact of specific regulatory systems on antibiotic PD. New MBMs were developed to describe and predict the time-course of bacterial killing and resistance for antibiotic mono- and combination therapy. Our MBM were combined with published population PK models to rationally optimize combination dosage regimens that maximize bacterial killing and minimize resistance in patients. While our approach focused on maximizing synergistic bacterial killing and prevention of resistance, we did not incorporate models for minimizing nephrotoxicity associated with polymyxin therapy and for the impact of the immune system. In addition, we assumed that the free plasma drug concentration determines pharmacodynamic activity; however, this assumption may not be appropriate for certain sites of infections. These considerations present potential limitations of our approach. Our studies in knockout strains suggested that the PD of polymyxins were not primarily determined by a single gene mutation but rather by a collection of regulatory systems, genes, and environmental factors. Therefore, redundancy of bacterial pathways may be critical for designing antibiotics with a new target site. Network analysis of bacterial gene regulatory systems might help to better understand regulatory pathways and to identify essential antibiotic targets whose inhibition maximizes bacterial killing. Collectively, the approaches developed in this thesis can be used as a platform for evaluating and rationally optimizing combinations of antibiotic and chemotherapeutic agents based on PK/PD principles and to quantitatively characterize the mechanisms of drug action and resistance via dynamic in vitro models and novel MBM.