ADME/T predictions

Gozalbes R, Barbosa F, Froloff N, Horvath D. (2006) The BioPrint approach for the evaluation of ADME-T properties Application to the prediction of cytochrome P450 2D6 inhibition. In Physicochemical and Computational Strategies, pp. 395-415, VHCA, Zurich, Wiley-VCH, Weinheim. 

Lion Biosciences is the supplier of the iDEA Metabolism software package as well as other ADME/T services (289). The iDEA software simulates metabolism and predicts a compound s metabolic behavior in humans. The Metabolism Module consists of a data expert module to perform data fitting and analysis of collected in vitro data and the physiological metabolism model. The physiological metabolism model is constructed from proprietary database of 64 clinically tested compounds. Additionally, the metabolism module automatically calculates the Michaelis-Menten constants Km and VjIiax for the kinetic analysis of metabolism turnover (289). 

Product property prediction Limited set mainly available in daylight Limited set Very large collection, including many vendor-supplied and internally developed in silico models for ADME T end points and target SAR... 

A further increase in the interest for predictive ADME/T methods is due to the development of high-throughput screening and synthesis methods. The possibility to design, make and test millions of compounds has increased the... 

Each year, a growing number of publications report on computational methods for the development of predictive ADME/T models. However, currently available methods are not reliable enough and are limited in their application, despite the recognition of their importance in the drug discovery process. Are we able to generate such reliable models, considering the severe limitations related to the intrinsic chemical diversity, the quantity and quality of the data In this chapter, we critically review data and approaches used to develop physicochemical and biological ADME/T models, in an attempt to address this question. 

We start this chapter with an analysis of methods to predict log P and aqueous solubility. In this context, we discuss the issue of applicability domain for QSAR models and the accuracy of prediction. Data available for simple physicochemical and ADME/T properties are compared by discussing the limitations of prediction of biological ADME/T properties. We restrict ourselves to several absorption and distribution properties, without discussing ME/T models. The interested reader is referred to the relevant sections in Comprehensive Medicinal Chemistry 7/(>1100 pages). 

Analysis of in-house data in pharma companies frequently demonstrates a low prediction ability of current models for both lipophilicity and aqueous solubility. The calculated errors of these models are often around or higher than 1 log-unit, which is not sufficient for screening purposes. Thus, despite relatively large amounts of data for physicochemical properties and their simplicity compared to more complex ADME/T properties, the accuracy of prediction remains low. Let us consider factors that limit the prediction accuracy of models. 

In silico techniques have gained wide acceptance as a tool to support the drug discovery and optimization process. Binding mode predictions via docking, affinity predictions via QSAR and CoMFA, or the prediction of ADME(T) properties are routinely applied [1-3]. 

---