Researchers and patients have a common desire for safe and efficacious medicines. As a result, all of the tools that aid in the process of creating new drug products are highly valued. Software tools won’t replace data-gathering pharmacokinetic and toxicology experiments; however, researchers can leverage the data mining and computer modeling capabilities using already amassed information to make predictions to facilitate drug discovery and development.
Understanding and modeling the basic absorption, distribution, metabolism, and elimination (ADME) processes for organisms exposed to drugs contributes to the safety assessment of a drug product, which requires not only the evaluation of a drug substance, but also safety profiles of its metabolites.
The ability to predict the potential risk associated with metabolite vulnerabilities can be used to help redesign and select more promising candidates for further and more rigorous testing throughout the phases of discovery and development. Obviously, simple and confident estimation of candidate risk is desired.
In vitro studies, however, still involve extrapolations, and conducting all experimental tests for candidates would require prohibitive investments of animals, time, and money. Overall costs should be alleviated by early rejection of compounds with problematic metabolites and reallocation of resources toward more promising candidates. These factors increasingly emphasize the value of reliable in silico property prediction as part of the process.
Each organization has its own preferred and dynamic battery of experiments and computational assessments for new chemical entities and derived metabolites, including objectives to:
- Predict basic physicochemical properties (logP/logD, pKa) that determine the distribution of metabolites in an organism and their clearance;
- Predict the likelihood that a metabolite will be further metabolized by cytochrome P450 enzymes;
- Predict inhibition of cytochrome P450 isoforms, and thereby, possible interactions of metabolites with concomitant drugs;
- Predict hERG inhibition, genotoxicity, and other toxicity endpoints.
Despite the complexity of biological systems in organisms, software enabling the prediction of ADME properties from molecular structure is available. While a variety of approaches to individual predictions have been expounded, it is possible to find an extensive set of prediction capabilities, including those addressing the objectives mentioned in this article. The use of logP, logD, solubility, and pKa for predicting ADME properties such as permeability, oral bioavailability, and distribution are well documented and will not be discussed further.
Hepatic metabolism by cytochrome P450 enzymes is the major clearance route for xenobiotics. These biotransformations can lead to either an increase or decrease in toxicity. Using in silico tools such as the cytochrome P450 predictive modules in the ACD/ADME Suite from Advanced Chemistry Development (ACD/Labs), researchers can identify whether compounds will be substrates and/or inhibitors of the five major drug metabolizing enzymes—CYP3A4, CYP2D6, CYP2C9, CYP2C19, and CYP1A2.
Furthermore, the regioselectivity module predicts the sites most likely susceptible to metabolism in human liver microsomes and proposes possible biotransformation reactions. These predictors also offer the ability to add experimental data to extend chemical space coverage and improve prediction accuracy.