Optimal exposures of ceftazidime predict the probability of microbiological and clinical outcome in the treatment of nosocomial pneumonia

Objectives: The %fT > MIC of ceftazidime has been shown to correlate with microbiological outcome of Gram-negative bacteria (GNB) in preclinical studies. However, clinical data are still lacking. We explored the relationship of ceftazidime exposure and outcome in patients with nosocomial pneumonia using data from a recent randomized, double-blind Phase 3 clinical trial. Patients and methods: Pharmacokinetic (PK) and demographic data from three clinical trials were used to construct a population PK model using non-linear mixed-effects modelling. Individual concentration-time curves and PK/pharmacodynamic indices were determined for individual patients. The MICs used in the analyses were the highest MICs for any GNB cultured at baseline or end of therapy. Results: A two-compartment model best fit the data, with creatinine clearance as covariate on clearance and age on the central compartment. Classification and regression tree analysis showed a breakpoint value of 44.9% (P<0.0001) for GNB in 154 patients. The Emax model showed a good fit (R-2=0.93). The benefit of adequate treatment increased from an eradication rate of 0.4848 at %fT > MIC of 0% to 0.9971 at 100%. The EC50 was 46.8% and the EC90 was 95.5% for %fT > MIC. Exposure correlated significantly with both microbiological and clinical outcome at test-of-cure. Conclusions: We conclude that exposures to ceftazidime predict microbiological as well as clinical outcome, and the %fT > MIC required to result in a likely favourable outcome is >45% of the dosing interval. This value is similar to that observed in animal models and underscores the principle that adequate dosing can be predicted and is beneficial to patient care.