Methods for Non-Compartmental Pharmacokinetic Analysis With Observations Below the Limit of Quantification

Pharmacokinetic (PK) studies are conducted to learn about the absorption, distribution, metabolism, and excretion processes of an externally administered compound by measuring its concentration in bodily tissue at a number of time points after administration. Two methods are available for this analysis: modeling and non-compartmental. When concentrations of the compound are low, they may be reported as below the limit of quantification (BLOQ). This article compares eight methods for dealing with BLOQ responses in the non-compartmental analysis framework for estimating the area under the concentrations versus time curve. These include simple methods that are currently used, maximum likelihood methods, and an algorithm that uses kernel density estimation to impute values for BLOQ responses. Performance is evaluated using simulations for a range of scenarios. We find that the kernel based method performs best for most situations. Supplementary materials for this article are available online.