A non parametric population approach for selecting biomarkers of a drug action
N. Terranova (1), F. Del Bene (2), F. Fiorentini (2), P. Magni (1), M. Germani (2)
(1) Dipartimento di Informatica e Sistemistica, Università degli Studi di Pavia, Pavia, Italy; (2) ACCELERA srl , Nerviano, Italy
Objectives: Intravenous treatment with a candidate drug administered at its optimal pharmacological dose was evaluated in healthy mice and in mice exposed to an inflammatory insult, with respect to a broad panel of inflammatory biomarkers. Three groups of animals were considered: two different groups were exposed to inflammatory insult 3h before the intravenous treatment with the investigated compound or its vehicle, one additional group was exposed to saline (that does not cause inflammation) 3h before the treatment with the same compound. Inflammatory biomarker concentrations were measured at different time points using a serial sacrifice design (6 mice for each time point for each group). The objective of the study was to determine which inflammatory biomarker was modulated by the compound under investigation.
Methods: The concentration curves, both typical and individual ones, were modeled as stochastic processes, in which the only requirement was a certain degree of regularity of the time profiles. A Markov Chain Monte Carlo algorithm was applied to perform the Bayesian estimation of typical and individual Areas Under the concentration-time Curves (AUCs) [1]. A comparison with the Bailer method was also performed.
Results: The method was applied to real data-sets. For each biomarker, the typical AUC and corresponding confidence interval in the three different groups were computed. Then, they were compared to find those biomarkers that were differently expressed in the three groups. They represent the best candidates to assess the drug effect.
Conclusions: The best experimental design and then the number of samples is usually limited by many practical factors, i.e. costs and ethical issues. This work presents an application of a nonparametric Bayesian approach for estimate the distribution of the AUC of a typical subject in a population study with a destructive sampling protocol. The method allows a sound comparison of different experimental conditions to select suitable biomarkers of the compound action. This method has a wide applicability in several (similar) situations involving sparse sampling and population studies.
References:
[1] P. Magni, R. Bellazzi, G. De Nicolao, I. Poggesi, M. Rocchetti (2002). Nonparametric AUC estimation in population studies with incomplete sampling: a Bayesian approach. J. Pharmacokinetics and Pharmacodynamics, vol. 29; p. 445-471, ISSN: 1567-567X
