Modeling the plasma concentration – time profile in a single rising dose study using the fractional polynomials approach.

Abstract

Preclinical toxicology studies identify a safe exposure range (NOAEL) to a novel chemical or biological entity. When going into man, it is important not to exceed this boundary. Therefore, an accurate prediction of the area under the curve (AUC) is crucial in a dose rising study. The estimation of the AUC can be obtained using a compartmental PK model or using an empirical approach, i.e., trapezoidal method.

Our gaol is to explore the behaviour of a third method by modelling the plasma concentration – time curve using a fractional polynomials approach. Fractional Polynomials widen the class of ordinary polynomials and offer more flexibility within the linear framework. Since the number of polynomials in infinite, the possible exponents in the model are limited to a prespecified set. In order to improve the quality of the estimators, we do not restrict to a single model but we will select a prefixed number of model, each of which will give an estimate and the final estimator will be obtained by model averaging. The weight of each term will be determined by the posterior probability of the model.

Given the dose escalation paradigm of our case study, a Bayesian approach was chosen to fit our models. The accuracy and precision of the three different methods is compared and presented.

References.

·  Aerts Marc, Faes Christel, Geys Helena and Molenberghs Geert,

Model averaging using fractionnal polynomials to estimate a safe level of

exposure. Risk Analysis, vol 27, No. 1, 2007

·  Sabanes Bove, D; Held, L,

Bayesian fractional polynomial, Statistic andComputing 2010, :Epub ahead of print

·  Aerts Marc, Boelaert Frank, Faes Christel, Greys Helena, Hens Niel,

Laevens Hans , Mintien Koenmm, Shkedy Ziv,

Estimating the herd-specific force of infection using random effects models for

clustered binary data and monotone fractional polynomials. Journal of the Royal

Statistical Society, Serie C, 55, 593-613, 2006