Purpose: Tigecycline is a novel glycylcycline antibiotic demonstrating broad-spectrum in vitro activity. A previously developed population pharmacokinetic (PK) model was utilized to generate individual predicted AUC0-12 estimates in patients with complicated skin and skin-structure infections (cSSSI) or complicated intra-abdominal infections (cIAI) in Phase 3 studies for use in future exposure-response analyses of safety and efficacy.
Methods: In 4 Phase 3 studies, tigecycline was infused over 0.5 hour (cIAI patients, n=155) or 1 hour (cSSSI patients, n=24) as a 100-mg loading dose followed by 50 mg every 12 hours for up to 14 days. A population PK model (2-compartment, zero-order input and first-order elimination; clearance as a function of creatinine clearance, weight, and sex) was previously developed using steady-state data from Phase 2 studies of patients with cSSSI and cIAI. The final Phase 2 parameter estimates were used as population priors. The model was applied separately to the Phase 3 cSSSI and cIAI data in NONMEM® to estimate empirical Bayesian PK parameters. Model performance was evaluated for both datasets by assessing goodness-of-fit, bias (PE%), and precision (⎪PE⎪%) for individual predicted tigecycline concentrations and steady-state AUC0-12 values.
Results: The patient populations in Phase 2 and 3 cSSSI and cIAI studies had similar demographic characteristics and steady-state tigecycline concentrations. Goodness-of-fit plots revealed that the Phase 2 model provided a relatively unbiased fit to both Phase 3 datasets: the median PE% was ± 3% and 4%, while ⎪PE⎪% was within 10% and 4% for the individual predicted tigecycline concentrations and AUC0-12 values, respectively.
Conclusions: The feasibility of using a previously developed population PK model to predict individual AUC0-12 in a new population with similar demographic and pharmacokinetic characteristics was explored. The Phase 2 population PK model for tigecycline provided an acceptably precise and relatively unbiased fit to the Phase 3 data, and resulted in unbiased estimates of AUC0-12, without the need for further model refinement.
American Association of Pharmaceutical Scientists (AAPS), Nashville, Tennessee, November 2005
By Scott Van Wart, Joel S. Owen, Alison K. Meagher, Brenda B. Cirincione