Antibiotic Model-Informed Precision Dosing in Critical Illness

PROJECT SUMMARY Pathophysiologic changes during critical illness can affect drug pharmacokinetics (PK, how the body affects the drug) and pharmacodynamics (PD, how the drug affects the body), but the current paradigm of drug dosing is a standard “one-size-fits-all” approach, rather than a personalized approach. Emerging data in adults

demonstrate increased risk of morbidity and mortality with standard doses of antibiotics in critically ill patients due to lack of PD target attainment. β-lactam antibiotics are a prime example of drugs demonstrating high PK

and PD variability in critically ill patients. Thus, critically ill patients are at risk of having low antibiotic exposures leading to ineffective bactericidal activity or high antibiotic exposures resulting in toxicity. By using real-time drug concentrations, individual patient and disease factors, and population PK models, model-informed

precision dosing (MIPD) can ensure adequate antibiotic exposure while avoiding toxicity. My proposed research program will address three critical knowledge gaps that are necessary to fill prior to implementation of antibiotic MIPD. First, the appropriate patient populations who will most benefit from

precision dosing remain unknown. Implementation of MIPD for every patient admitted to the intensive care unit may be resource-intensive and costly. However, simply increasing antibiotic doses or frequency of administration without data-driven management can be dangerous as it carries risks for antibiotic-associated

toxicity, including nephrotoxicity and neurotoxicity. Therefore, it is critical to optimize the benefit-to-risk ratio of therapeutic interventions for individual patients, a fundamental concept of precision medicine. Second, there remains a knowledge gap on the association of precision dosing of antibiotics and clinical outcomes. Studies

examining antibiotic exposure and outcomes in adults have had mixed results; some show improved outcomes with PD target attainment and some show no difference in outcomes with regards to target attainment. Third, many of the antibiotic population PK models needed for MIPD have not been prospectively validated in

critically ill patients, so it is unknown which models should be used for precision dosing. To address these knowledge gaps, Project 1 will utilize innovative modeling and simulation to identify patient and disease factors associated with antibiotic under-exposure (risk of ineffective antibacterial activity) or over-exposure (risk

of toxicity) and investigate mechanisms underlying toxicities. Project 2 will investigate the effect of precision dosing on clinical outcomes by evaluating the association between PD target attainment and clinical outcomes at the individual level. Project 3 will prospectively validate our models and previously published models to

ensure accurate predictive ability in critically ill patients. With these prospectively validated models, we will lay the foundation to build MIPD decision support tools integrated with the electronic health record to generate individual patient PK and PD profiles for precision dosing in critically ill patients. These tools will be essential to

implement antibiotic MIPD to guide clinicians in dosing regimen selection for critically ill patients.