Evaluation of in vivo drug release by numerical deconvolution using oral solution data as weighting function.

Determination of in vivo drug release using compartmental model analysis is hampered by problems such as flip-flop phenomena and vanishing exponential terms. The usefulness of numerical deconvolution to estimate in vivo drug release was evaluated in this study by means of simulated data comparing solid dosage forms with a solution as a reference standard. Concentration-time data were generated using the standard linear two-compartment body model with various first-order release and absorption rate constants. Random errors of 5 and 10% were added to data sets for further analysis. The results of the study using error-free data afforded excellent agreement with the theoretical values except in one case where the release rate constant was overestimated by 6%. When random error was added to the data, the resulting in vivo release profile showed considerable fluctuation and no single rate constant could be assigned. However, further analysis showed that the method does not create additional error during the calculating process, as previously suggested, but merely reflects the inherent error added to the raw data. If the raw data are poor, no useful information can be obtained without using an arbitrary technique such as smoothing or fitting. In this regard, the time course of drug release obtained after numerical deconvolution merits investigation.