Abstract / Summary
Mycophenolic acid and tacrolimus are frequently used in kidney transplant recipients and patients with nephrotic syndrome. For both drugs, therapeutic drug monitoring is recommended. Model-informed precision dosing, not only uses population pharmacokinetic data but also increases flexibility. To optimize treatment with mycophenolic acid and tacrolimus using a limited number of samples, we evaluated model-informed precision dosing for both drugs. We performed the following steps: (1) selection of population pharmacokinetic models; (2) retrospective evaluation using real-world data; and (3) evaluation of limited sampling strategies, (4) prospective evaluation of the predictive performance by predicting an area under the concentration–time curve from time zero to 12 h at a next occasion (3 months later). Acceptable bias and imprecision were defined as a mean prediction error and a normalized root mean square error of less than 25%. A 3‑point sampling schedule ( t = 0, t = 1, t = 2 h) provided accurate area under the concentration–time curve from time zero to 12 h estimation for both drugs, with mean prediction error and normalized root mean square error of 0.1% (− 0.3 to 0.6%) and 21.0% (7.1–34.8%) for mycophenolic acid, and 0.2% (0.03–0.4%) and 7.8% (3.0–12.6%) for tacrolimus. Substantial between-occasion variability limited the prediction of future area under the concentration–time curve from time zero to 12 h. For mycophenolic acid, the mean prediction error was 15.4% (14.4–16.4%) and the normalized root mean square error was 48.6% (18.2–79.0%). For tacrolimus, corresponding values were 64.2% (59.7–68.7%) for mean prediction error and 210% (37.1–383%) for normalized root mean square error. Model-informed precision dosing-guided limited sampling strategies are feasible for estimation of current exposure using similar sampling times for mycophenolic acid and tacrolimus. Accurate long-term prediction is not possible, underscoring the need for repeated therapeutic drug monitoring. NCT06095895.