Pharmacometrics to improve evaluation and individualisation of β-lactam/β-lactamase inhibitor combinations

β-lactam/β-lactamase inhibitor (BL/BLI) combinations are essential for treating infections caused by multidrug-resistant Gram-negative bacteria, particularly in critically ill patients. However, pharmacokinetic (PK) and pharmacodynamic (PD) variability complicates dosing, potentially leading to suboptimal exposure, reduced bacterial eradication, and treatment failure. Despite widespread use, BL/BLI therapy relies largely on standard dosing approaches, with limited individualisation. This thesis explores pharmacometric strategies to better understand the complexities of BL/BLI therapy in critically ill patients by evaluating drug exposure, efficacy target attainment, and dosing strategies through real-world patient data, PKPD modelling, and simulation-based approaches. Both simulated and patient-derived data were analysed. Population PK analysis was applied to characterise ceftazidime-avibactam (CAZ-AVI) disposition in critically ill patients with pneumonia and those undergoing continuous venovenous hemodiafiltration (CVVHDF). PK/PD indices and targets for avibactam were investigated in preclinical and clinical settings. In simulations, target attainment was evaluated for multiple BL/BLI regimens across different infection sites and renal function groups. Additionally, host response biomarkers were assessed for their potential role in treatment monitoring and individualisation. Significant interindividual variability in CAZ-AVI PK was observed, even after accounting for renal function, suggesting additional unexplained sources of variability. Standard dosing in CVVHDF patients resulted in lower early (0-2 h) and higher later (4-8 h) concentrations compared to non-CVVHDF patients, indicating a larger volume of distribution and the need for tailored regimens. Both fT>CT and fAUC/MIC were identified as the best PK/PD indices for avibactam, depending on bacterial strain and mode of infusion, challenging the assumption of universal PK/PD indices. Simulations revealed that insufficient BLI exposure frequently limited target attainment, underscoring the need to consider both BL and BLI concentrations in dose optimisation. Analysis of immune response biomarkers revealed dynamic changes over the course of treatment, with one identified relationship between drug exposure and host response, though further clinical validation is needed. This work demonstrates how model-based approaches can enhance BL/BLI therapy evaluation and individualisation by characterising PK variability, refining efficacy targets, and assessing dosing strategies in critically ill patients. Future research should focus on linking target attainment to clinical outcomes and integrating therapeutic drug monitoring and biomarker-guided approaches where relevant to optimise therapy.