A Bayesian Approach to Parallelism Testing in Bioassay

• Published in: Statistics in biopharmaceutical research Vol. 4; no. 4; pp. 357 - 374
• Main Authors: Novick, Steven J., Yang, Harry, Peterson, John J.
• Format: Journal Article
• Language: English
• Published: Taylor & Francis Group 01.10.2012
• Subjects: Equivalence; Gibbs sampling; Prior distribution; Similarity; Stand curve
• ISSN: 1946-6315; 1946-6315
• DOI: 10.1080/19466315.2012.707085

Parallelism is a prerequisite for the determination of relative potency in bioassays. It involves testing of similarity between a pair of dose-response curves of a reference standard and a test sample. Methods for parallelism assessment that are currently in use include p-value-based significance tests and interval-based equivalence tests. These methods make statistical inference about the similarity between the model parameters of the dose-response curves based on the sampling distribution of the estimates of these parameters. Although the methods have some merits for parallelism testing, there is a major drawback to these approaches, namely that the similarity between the model parameters does not necessarily translate into the similarity between the two dose-response curves. As a result, a test may conclude that the model parameters are similar, yet there is little assurance on the similarity between the two dose-response curves. In this article, we reformulate the parallelism testing problem as testing the hypothesis that the test sample is a dilution or concentration of the reference standard. We propose a Bayesian approach to directly test the hypothesis. When the dose-response curves are linear, a closed-form solution is obtained. For nonlinear cases, we render a solution based on a simple two-dimensional optimization routine. The empirical properties of the method are evaluated and compared with existing methods through a simulation study based on real-life examples. It is shown that the method overcomes the shortcomings of the current approaches and is a viable alternative to parallelism testing.