The degree of predictivity in pilot studies on six subjects in bioequivalence trials

• Published in: Pharmacological research Vol. 49; no. 3; pp. 283 - 286
• Main Authors: Marzo, Antonio, Fibbioli, Monia, Marone, Claudio, Cerutti, Bernard
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier Ltd 01.03.2004
• Subjects: Analysis of Variance; Area Under Curve; Bioequivalence; Computer Simulation - statistics & numerical data; Confidence Intervals; Female; Humans; Male; Pilot Projects; Pilot trials; Pivotal trials; Predictive Value of Tests; Predictivity index; Therapeutic Equivalency; Predictivity index; Bioequivalence; Pivotal trials; Pilot trials
• ISSN: 1043-6618; 1096-1186
• DOI: 10.1016/j.phrs.2003.09.007

This study, based on computer simulations, analysed the degree of predictivity of pilot trials on six subjects, with the idea of a further pivotal trial on 18 or more volunteers aimed at assessing bioequivalence. Volunteers enrolled in 10 pivotal bioequivalence trials were considered. For every trial, a thousand bootstrap samples were generated to simulate trials with six subjects, while keeping a balanced design for sequence, period and formulation. Then a standard ANOVA for crossover trials, with 90% confidence intervals for the ratios, was done on C max and AUC for each simulated trial. The number of subjects needed to achieve bioequivalence was based on the residual error of the ANOVA. When this number was smaller or equal in the case of bioequivalence, or larger in the case of insufficient evidence to conclude on bioequivalence, to the number of the subjects enrolled in the original trial, the subgroup was considered predictive. Otherwise it was considered non-predictive. The average predictivity index, calculated by dividing the number of predictive findings by the total number of subgroups, in our case 1000, and multiplying the result by 100, was 71.1% for C max and 82.9% for AUC. Results of the simulations suggest that pilot trials on six volunteers can be useful for predicting the pool size of volunteers in bioequivalence trials, and for in vivo–in vitro correlation studies in pharmaceutical development strategy.