Handling Below Limit of Quantification Data in Optimal Trial Design
2014 (English)In: Journal of Pharmacokinetics and Pharmacodynamics, ISSN 1567-567X, E-ISSN 1573-8744Article in journal (Other academic) Accepted
Methods that perform well in handling limit of quantification (LOQ) data exist in estimation of parameters for non-linear mixed effect models but are not well developed in experimental design. The aim of this work was to evaluate existing methods and to explore new methods of handling LOQs in Optimal Design (OD). Seven different methods were implemented in PopED 2.13: D1 (Ignore LOQ), D2 (Non-informative Fisher information matrix (FIM) for median response below LOQ), new D3 (Non-informative FOCE linearized FIM for individual response below LOQ), D4 (Addition of a homoscedastic variance), new D5 (Simulation & Rescaling), new D6 (Integration & Rescaling) and new D7 (joint likelihood using the Laplace approximation). Predictive performance of D1-D7 was first assessed and sample time optimization was performed for a number of different LOQ levels. Resulting designs were evaluated for bias and imprecision, robustness and predictability from multiple stochastic simulations and estimations (SSE) in NONMEM using the M3 method. Evaluated determinants of the FIM for all methods, except D1 and D4, were in good agreement with SSE-derived covariance. In optimization, D6 provided the most accurate and precise parameter estimates and the designs with the best predictive performance under the M3 method. Methods D1 and D2 resulted in the least robust designs for estimation. Method D4 was shown to be insensitive to LOQ levels. For the scenarios investigated, method D6 showed the best compromise in terms of speed and accuracy. The use of OD methods anticipating LOQ data in planned designs allows better parameter estimation.
Place, publisher, year, edition, pages
Population Optimal Design, LOQ, BQL data, NLME, Pharmacometrics, Population modeling
Research subject Pharmaceutical Science
IdentifiersURN: urn:nbn:se:uu:diva-233441OAI: oai:DiVA.org:uu-233441DiVA: diva2:752556