In silico models for genotoxicity and drug regulation

ABSTRACT

Introduction

Whereas in the past, (Q)SAR methods have been largely used to support the design of new drugs, in the last few decades, there has been a new interest in its applications for the assessment of drug safety. In particular, the ICH M7 guideline has introduced the concept that (Q)SAR predictions for the Ames mutagenicity of drug impurities can be used for regulatory purposes.

Areas covered

This review introduces the ICH M7 conceptual framework and illustrates the most updated evaluations of the in silico approaches for the prediction of genotoxicity. The strengths and weaknesses of the state-of-the-art are presented and future perspectives are discussed.

Expert opinion

Given the growing recognition of (Q)SAR approaches, more investment will be devoted to its improvement. The major areas of research should be the expansion and curation of the experimental training sets, with particular attention to the portions of chemical space which are poorly represented. New modeling methodologies (e.g. machine-learning methods) may support this effort, particularly for treating proprietary data without disclosure. Research on new integrative approaches for regulatory decisions will also be important.

KEYWORDS:

• Alternative methods
• genotoxicity
• predictive methods
• QSAR
• structure–activity
• toxicology

Article highlights

• (Q)SARs (also referred to as in silico methods) is a promising area of computational toxicology that attempts to predict the potential adverse effects of a chemical based on its chemical structure.

• The ICH M7 guideline on drug impurities recognizes the ability of (Q)SAR to relate chemical structure and bacterial mutagenicity (Ames test).

• Development of in silico models to predict mutagenicity benefits from the underlying biological mechanisms of mutagenicity, which to a large extent are typically governed by the electrophilicity of the chemicals.

• (Q)SAR prediction rates for bacterial mutagenicity achieve an accuracy of around 80%, which is close to the reported 80 – 84% variability of the test itself.

• Most of the predictive exercises reported in the literature focus on the Ames test, and evaluations of (Q)SARs for other assays/endpoints are limited.

• Reliability of the (Q)SAR models for assays/endpoints different from in vitro bacterial mutagenicity (Ames) appears to still be far from optimal, and this could be due to the different types and quality of available biological data.

• Expert review of (Q)SAR prediction is an important process, as recognized by the ICH M7 guideline itself. The review is used to increase or decrease the reliability of the prediction and can make use of additional data generated with a third model or introduced by experts.

• An effective way to improve the predictive systems is the generation of new data as well as the curation of existing data.

This box summarizes key points contained in the article.

Declaration of interest

M Pavan and A Bassan are principal consultants at Innovatune srl (Italy), a company providing services based on computational toxicology including genotoxicity evaluations. They may occasionally use software cited in this paper. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.