Abstract
The aim of this study was to evaluate the use of limited blood sampling and Bayesian analysis to estimate the pharmacokinetics (PK) and tailor the dose of factor VIII (FVIII) in an individual patient. In a Bayesian analysis, PK parameters are estimated from only a few plasma concentration measurements, using a previously established PK model. First the necessary model was created using intense blood sampling FVIII data from 10 patients. Then FVIII data from another 21 patients were used for 'clinical' evaluation. Three scenarios were created retrospectively by reduction of the original 7-sample data set; blood sampling at 4, 24 and 48 h, at 8 and 30 h and at 24 h after the infusion. PK parameters were estimated for each individual using Bayesian analysis and compared with those obtained using conventional methods from the full data. The accuracy of predictions of FVIII levels during prophylactic treatment 5-17 months later and implications for dose tailoring were also investigated. Blood sampling at 4, 24 and 48 h was found to give practically the same PK information as a full, conventional (7-10-sample) study. Even a single 24-h FVIII level provided adequate data for initial dose tailoring and gave predictions of FVIII levels 5-17 months later that were not appreciably worse than predictions based on the full PK analysis. By contrast, dose tailoring based on body weight failed completely. In conclusion, PK-based dose tailoring of FVIII can be performed using limited blood sampling during prophylactic treatment.