Graham Lockwood, Gary Maier
Sanofi Research Division, Collegeville, PA; U.S.A.
A highly variable drug has been defined as one with an intra-subject CV of 25-30% or greater (Eur J of Drug Metab and Pharmacokinetics 18:225 1993). Two formulations of tiludronate were given to 24 subjects in oder to demonstrate bioequivalence; a standard randomized two way cross-over design was utilized. Although there was no statistically significant difference in AUC when the two formulations were compared, the 90% confidence interval exceed the regulatory accepted range of 80-125%.
Since both formulations were given to a combined total of approximately 150 subjects; NONMEM was used to determine the relative bioavailability of the test dosage form by evaluating all clinical studies performed in healthy volunteers. The selected model was "verified" by comparing estimates of relative bioavailability and 90% confidence intervals, obtained in a test data set using NONMEM against those calculated utilizing standard methodologies.
Initial evaluation of the base pharmacokinetic model suggested that a two compartment model with a term for relative bioavailability best described the data. The next step was to create a data base of all healthy subjects studied to date and determine relative bioavailability along with the 90% confidence interval of the estimate. The NONMEM analysis confirmed that there was no statistically significant difference between the two formulations but the 90% confidence interval exceeded commonly accepted bioequivalence guidelines as shown in the smaller 24 subject cross- over study. We then used this analysis as part of an overall petition to the FDA to reconsider bioequivalence guidelines for highly variable drugs.
Advan 2 and 4 were used along with Trans 3