Indirect bioequivalence assessment using network meta-analyses.

Autor: Ring A; Department of Mathematical Statistics and Actuarial Science, University of the Free State, P.O. Box 339 (IB75), Bloemfontein, South Africa, RingA@ufs.ac.za., Morris TB, Hohl K, Schall R
Jazyk: angličtina
Zdroj: European journal of clinical pharmacology [Eur J Clin Pharmacol] 2014 Aug; Vol. 70 (8), pp. 947-55. Date of Electronic Publication: 2014 May 20.
DOI: 10.1007/s00228-014-1691-0
Abstrakt: Aims: For market approval, new drug formulations (test) must demonstrate bioequivalence (BE) to at least one approved formulation (reference). If several formulations of a drug are already on the market, one might have to show BE to more than one reference formulation. Similarly, if several test formulations have shown BE to a reference formulation, it will be of interest whether the test formulations are bioequivalent to each other.
Methods: An enhanced statistical model to assess BE indirectly through a network meta-analysis is provided. Statistical properties of a parallel and a bridging approach are derived, in particular the relative statistical efficiency of the two approaches. The analysis is illustrated using individual subject data from two 3×3 crossover trials of metformin formulations, which have one of the formulations in common.
Results: The parallel estimate of relative bioavailability is confounded with between-trial differences, while the bridging estimate is not. The standard errors of the formulation differences using the bridging approach are smaller than the standard errors using the parallel approach if the within-subject correlation in each trial of the network is larger than 0.5. This is the condition for a crossover trial to be more efficient than a parallel trial, and thus is usually fulfilled in pharmacokinetic crossover trials.
Conclusions: Indirect BE assessment offers the opportunity to efficiently determine the relative bioavailability of drug formulations that have not been studied in the same randomized BE trial. The methodology developed here allows estimating formulation differences across a larger network.
Databáze: MEDLINE
Externí odkaz: