ADME computational models

Our experience with trying to build computational models based on experimental permeability screening in Caco-2 cell culture illustrates the problem introduced by multiple mechanisms. We found that deviation from a single mechanism could arise either in the assay per se or could arise from the compounds that were screened in the assay. One aspect of the multiple mechanism problem is the presence of active multiple biological transport mechanisms for both enhancing and reducing absorption in cell culture assays. This issue is well documented and is outside the scope of this chapter. 

---

To date, many of the reported ADME/Tox models have been rule based. For example, some research groups have used relatively simple filters like the rule of 5 [93] and others [94] to limit the types of molecules evaluated with in silico methods and to focus libraries for HTS. However, being designed as rapid computational alert tools aimed at a single property of interest, they cannot offer a comprehensive picture when it comes to understanding ADME properties. 

Ekins S, Swaan PW. Computational models for enzymes, transporters, channels and receptors relevant to ADME/TQX. Rev Comp Chem 2004 20 333-415. 

Computationally modeled or calculated properties (ADME, safety, physchem properties, potency, selectivity) ... 

In a hit triage decision making process that blends the use of experimental data with expected general property trends and principles, there are situations where it is not feasible to obtain sufficient data to identify experimentally property trends for ADME or safety endpoints (either due to a small number of hit compounds in a series, or due to limited experimental capacity). Computational models for these parameters may provide some useful information when integrated with other known information [101],... 

There has been extensive work on computational modeling of ADME and safety properties in recent years, but the field is still evolving [102, 103]. There are two key limitations on the use of models for these endpoints. One limitation is technical - the quality and accuracy of the models for the chemical space of interest [104, 105]. 

Ekins S, Swaan PW (2004) Development of computational models for enzymes, transporters, channels, and receptors relevant to ADME/tox. In Lipkowitz KB, Larter R, Cundari TR (eds) Reviews in computational chemistry, vol. 20. Wiley, Hoboken, NJ, chap 6... 

Prediction of ADME properties should be simple, since the number of descriptors underlying the properties is relatively small, compared to the number associated with effective drug-receptor binding space. In fact, prediction of ADME is difficult The current ADME experimental data reflect a multiplicity of mechanisms, making prediction uncertain. Screening systems for biological activity are typically single mechanisms, where computational models are easier to develop [1],... 

Many advances have been made in computational ADME modeling. For many ADME properties, models now exist which provide reasonably good predictive quality and can be deployed to aid medicinal chemists in drug discovery projects. 

Sean Ekins and Peter Swaan, Development of Computational Models for Enzymes, Transporters, Channels and Receptors Relevant to ADME/Tox. 

Klopman G, Stefan LR, Saiakhov RD (2002) ADME evaluation 2. A computer model for the prediction of intestinal absorption in humans. Eur. J. Pharm. Sci. 17 253-263. 

Key Words CYP P450 cytochrome P450 docking structure-based drug discovery pharmacophore QSAR homology models databases computational models ADME/T. 

In Silico ADME Modeling 3 Computational Models to Predict Human Intestinal Absorption Using Sphere Exclusion and kNN QSARMethods. QSAR Combinatorial Science, 5, 653-668. 

Gunturi, S.B., Narayanan, R. and Khandelwal, A. (2006) In silico ADME modelling 2 Computational models to predict human serum albumin binding affinity using ant colony systems. Bioorganic and Medicinal Chemistry, 14, 4118 129. 

Egan, W.J. (2007) Computational models for ADME. Annual Reports in Medicinal Chemistry, 42, 449—467. 

The development of predictive models for drug-likeness, frequent hitters, ADME processes, and toxicological endpoints has so far yielded a great deal of soft filters (see discussion above and the compilation of ADMET computational models by Yu and Adedoyin [66]), and the trend still continues to improve both accuracy and... 

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. 

In parallel to the experimental ADME field, computational models are developed for specific in vitro assays or even, in some cases, for specific pathways represented in each system (Fig. 10.2). The computational models are based on the experimental data, and since there are still not consistent data available for some of the properties shown in Fig. 10.2, they have not yet been addressed. 

---