PI‐69

BACKGROUND The aim of this work was to develop a method for estimating exposure (AUC) to a highly variable drug, based on two fixed‐time plasma samples per subject for a prospective clinical study. Other challenges in estimating AUC were limited dose proportionality and changes to dosing regimen, for which no PK data existed. METHODS Steady‐state plasma concentration‐time data from a rising, multiple dose study was used to develop a population PK model, which was evaluated for goodness‐of‐fit. PK profiles were simulated based on model‐derived individual PK parameters and compared to those of the observed data. Subsequently, PK profiles based on a new dosing regimen were simulated, from which concentrations were selected based on two fixed times. The average of these two concentrations (Cavg) for each subject was correlated with individual AUC. RESULTS The population PK model adequately described the data, but could not be used for predictions due to the large residual error (CV=36%). Thus, PK profiles were simulated based only on model‐derived individual PK parameters; these simulated profiles agreed with the observed data. There were positive linear correlations between Cavg and AUC for both observed data (r 2 =0.97) and simulated data with the new dosing regimen (r 2 =0.99). CONCLUSIONS In the case where a population PK model can not be used, an alternative method was developed and evaluated for estimating exposure based on two fixed‐time samples. This approach will be used in prospective clinical studies. Clinical Pharmacology & Therapeutics (2005) 79 , P25–P25; doi: 10.1016/j.clpt.2005.12.090