Evaluation of limited sampling strategies for tacrolimus

OBJECTIVE: In literature, a great diversity of limited sampling strategies (LSS) have been recommended for tacrolimus monitoring, however proper validation of these strategies to accurately predict the area under the time concentration curve (AUC(0–12)) is limited. The aim of this study was to determine whether these LSS might be useful for AUC prediction of other patient populations. METHODS: The LSS from literature studied were based on regression equations or on Bayesian fitting using MWPHARM 3.50 (Mediware, Groningen, the Netherlands). The performance was evaluated on 24 of these LSS in our population of 37 renal transplant patients with known AUCs. The results were also compared with the predictability of the regression equation based on the trough concentrations C(0) and C(12) of these 37 patients. Criterion was an absolute prediction error (APE) that differed less than 15% from the complete AUC(0–12) calculated by the trapezoidal rule. RESULTS: Thirteen of the 18 (72%) LSS based on regression analysis were capable of predicting at least 90% of the 37 individual AUC(0–12) within an APE of 15%. Additionally, all but three LSS examined gave a better prediction of the complete AUC(0–12) in comparison with the trough concentrations C(0) or C(12) (mean 62%). All six LSS based on Bayesian fitting predicted <90% of the 37 complete AUC(0–12) correctly (mean 67%). CONCLUSIONS: The present study indicated that implementation of LSS based on regression analysis could produce satisfactory predictions although careful evaluation is necessary.