Evaluation of curve comparison metrics applied to pharmacokinetic profiles and relative bioavailability and bioequivalence

Abstract

The goal of this dissertation was to investigate three new curve comparison metrics, the Rescigno Index, fl, and the Chinchilli Metric as tools to compare pharmacokinetic profiles for the assessment of assess relative bioavailability (BA) and bioequivalence (BE). The specific objectives were to (1) compare the relative sensitivity of the new metrics to detect differences in AUC and Cmax as a function of the pharmacokinetics of the drug products, and (2) to estimate relative bioavailability and bioequivalence. Methods. Retrospective analysis of experimental data and Monte Carlo simulations of bioequivalence trials were used to evaluate the relative sensitivity of the metrics to detect profile differences. The experimental data study involved determining the degree of discordance with typical criteria when judging individual profiles to be the same or different, and then examining the relationship between the degree of discordance and the pharmacokinetics of the drug product. The simulation studies involved determining the proportion of clinical studies failing bioequivalence under different pharmacokinetic models. Product bioequivalence was estimated using data from 35 typical 2 treatment-2 period bioequivalence study experimental datasets. Three different bioequivalence limits were applied to the curve metrics. Results. The new metrics more effectively detect differences in absorption time lags but less effectively detect differences in Cmax under some conditions. The relative sensitivity to Cmax depends on the shape of the curve, where increasing the ka(ref)/ke(ref) increases the disparity across the metrics. The curve metrics show increased sensitivity to variability in disposition, elimination, and random residual error, but comparable sensitivity to differences in bioavailability. Fourteen of the 35 studies failed typical criteria (AUC and Cmax). Applying bioequivalence limits of 25%, 21 and 26 studies failed the Chinchilli and fl criteria respectively. At the 30% limit, 14 and 20 studies failed the Chinchilli and fl criteria respectively. The specific studies failing each criterion varied. The within-subject variability of the Chinchilli Metric was higher than Cmax. Both the Chinchilli Metric and fl showed a tendency toward extreme values. Conclusions. The metrics differ in pharmacokinetic sensitivities and differ in statistical properties. There are advantages and disadvantages associated with these differences.