Introduction and Overview

ALAN G. E. WILSON

Lexicon Pharmaceuticals Inc., Drug Metabolism, Pharmacokinetics, Toxicology and Pathology, 8800 Technology Forest Place, The Woodlands, Texas 77381-1160, USA

1.1 Introduction

This book provides an expert and comprehensive discussion on the current status and future directions of predictive toxicology and its applications. It covers areas of in silico, in vitro and in vivo approaches, and how they are being used and integrated to improve toxicity prediction and risk assessment of chemicals. The various chapters in this book also reflect the growing need for improvements in our technologies and ability to predict the toxicities of pharmaceutical and industrial chemicals to ensure product safety and the protection of public health. It brings together leading experts in the fields of toxicology and risk assessment to discuss their views on the current status of toxicity prediction as well as their visions of the future needs and challenges.

It is becoming increasingly critical to identify potential toxicities early in the development of new chemical entities, be they for the pharmaceutical or chemical industries. It is quite apparent, however, that our current approaches have limitations and the utility of animal data may not always ensure the safety of a chemical or pharmaceutical in humans. Ultimately, it will be how we use and evaluate all of the data and approaches that will assist in making the breakthroughs that are needed in assessing risks and improving human safety of chemicals, and thereby reducing attrition of drug candidates due to safety issues and potential human health hazards associated with chemical exposures.

1.2 The Growing Need for Improvement in Toxicity Prediction

1.2.1 Global Regulatory Initiatives

In recent years we have seen the introduction of several initiatives by regulatory agencies to improve the safety assessment of pharmaceutical and industrial chemicals. For example, the European Commission (EC) introduced legislation in 2003 for a new chemical management system, REACH (Registration, Evaluation, Authorisation and Restriction of Chemicals). In 2004, the US Food and Drug Administration (FDA) launched the Critical Path Initiative, which is focused on improving drug and medical device development, the quality of data generated during development, and the outcomes of clinical use. The major focus of these initiatives is to ensure reasonable product safety, while also facilitating the translation into commercial products. Thus, these separate regulatory initiatives highlight a common theme of ensuring that we are maximizing our ability to predict the safety of pharmaceutical and industrial chemicals in humans.

There have also been a number of initiatives by regulatory agencies to improve our abilities in toxicity prediction, and to facilitate the development and validation of computational in silico approaches. In the European Union (EU), efforts have been ongoing for a number of years to develop quantitative structure–activity relationships (QSARs) and other models to aid in toxicity prediction, and the implementation of REACH is expected to accelerate the process to develop important data sources to further improve these models. Similarly, the US FDA has been engaged in a number of programs focused on developing in silico models for various toxicities, and has a long-standing interest in, and much experience of, working with industry on the evaluation of in silico approaches of toxicity prediction. In addition, in 2007 the US Environmental Protection Agency (EPA) initiated ToxCast™, a project in computational toxicology. ToxCast™ is designed to apply high-throughput screening (HTS) data to building computational models to forecast the potential human toxicity of environmental chemicals. The hazard predictions obtained with ToxCast™ are expected to provide the US EPA regulatory programs with science-based information helpful in prioritizing chemicals for higher tier toxicological evaluations, leading to a more efficient use of resources, and expediting the process to evaluate existing environmental chemicals.

1.2.2 Industrial Initiatives

For the pharmaceutical industry, safety issues remain a major concern and a cause of late-stage attrition, low success rate and withdrawal of products from the market. Only about 30% of possible drugs that have acceptable animal testing make it into Phase III of clinical testing, with approximately 30% of those failing in Phase III. Discovering safety issues late in development is potentially devastating financially for a company. It is therefore imperative that potential safety issues are identified early. Thus, the current focus and urgency in the pharmaceutical industry is to shorten the timelines for all aspects of drug discovery and to improve the ability to filter out potential safety issues earlier. We are therefore seeing an increasing interest and focus on predictive approaches. These predictive approaches are generally employed early in drug discovery often before efficacy screening. It is expected that these approaches will gain increasing importance in the coming years in allowing early identification of potential toxicity issues. In addition, the continuing understanding and advancement of, "-omics" (i.e., genomics, transcriptomics, proteomics, metab(ol)onomics) technologies are expected to identify biomarkers that will assist in the prediction of different toxicities.

In the case of industrial chemicals, the extent of environmental exposure and potential safety issues of many chemicals to which humans might be exposed, has not been well studied. Thus, in recent years we have seen an increasing focus on the need to improve our ability to more accurately screen for potential safety issues and improve the translation of safety data to humans. In order to expedite the process for assessing risks and ensuring the human safety of the chemicals already in market, it is imperative to improve our technologies and ability to predict their toxicities. The pressure to reduce the use of experimental animals further enhances the needs to innovate our strategies for predicting toxicity of chemicals in humans.

1.2.3 Paradigm Shift in Toxicity Testing

Current toxicity testing relies primarily on determining the potential for adverse responses in animals (typically rodents, dogs and/or non-human primates) exposed to high doses of a test agent. In vivo animal toxicity testing has been considered the benchmark to identify potential adverse effects of chemicals. However, the standard in vivo approach typically provides little information on the mode and mechanism of toxicity. Such information is frequently critical for understanding interspecies differences in toxicological responses, and in the translation of animal data to humans. In vivo studies are also expensive, time-consuming and require large numbers of animals. Hence, for a number of years we have seen global interest and initiatives to reduce, replace and refine animal testing (the three Rs) from both the standpoint of animal welfare and out of economical necessity.

The National Research Council (NRC) of the National Academy of Sciences (NAS) issued its landmark report Toxicity Testing in the 21st Century: a Vision and a Strategy in 2007. The report's vision for assessing the potential adverse effects of chemicals on human health shifts the focus from overt effects such as pathological changes in animals, to upstream effects on biological pathways. The new framework is centered on human biology and a computational systems approach, and proposes a shift from in vivo animal testing to human-cell based in vitro testing. This new paradigm suggests the possibility of accelerating the human hazard assessment of environmental chemicals by replacing the current framework based on extrapolating experimental animal data to humans, with the evaluation of the in vitro effects in human cellular systems on the toxicity pathways leading to the human diseases. The NRC report presented an ambitious vision for the future of toxicity testing with the hope that advances in new technologies would make toxicity testing quicker, less expensive, and more directly relevant to human exposure.

The excitement over the NRC vision has led to a general emphasis on "21st Century Toxicology" — a phrase drawn from the report's title and intended to encompass multiple approaches to toxicity testing that harness modern advances in biology and technology. In order to realize the new vision and strategy of toxicity testing proposed by the NRC, the regulatory agencies in the US are providing leadership on the improvement of our toxicity prediction approaches. In 2007, the National Institute of Health (NIH) Chemical Genomics Center (NCGC) of the National Human Genome Research Institute (NHGRI), the National Institute of Environmental Health Sciences (NIEHS) National Toxicology Program (NTP), and the US EPA, collaborated on a new high-throughput toxicity screening initiative entitled Tox21.

Tox21 is a collaborative effort between federal agency partners striving to develop a new paradigm of predictive toxicology by pursuing seven enabling objectives:

1) Research, develop, validate, and translate innovative chemical testing methods that characterize toxicity pathways;

2) Investigate ways to use new tools to identify chemically induced biological activity mechanisms;

3) Prioritize which chemicals need more extensive toxicological evaluation;

4) Develop models that can be used to more effectively predict how chemicals will affect biological responses;

5) Identify chemicals, assays, computational platforms, and targeted testing needed for the innovative testing methods;

6) Complete acquisition in 2010 for a library of more than 10 000 chemicals for quantitative high-throughput screening (qHTS) at the NCGC; and

7) Implement Phase II of EPA's ToxCast™ program, which will include the screening of a 700-compound subset of the 10 000-compound library in various mid- and high-throughput assays.

The goal of Tox21 is to develop automated, high-throughput molecular and cell-based approaches for determining the potential human toxicity of the thousands of chemicals currently in use in the US. These early high-throughput screening approaches can also be used in the early phases of drug discovery, and are expected to gain increasing importance in the coming years by facilitating earlier prediction and identification and prediction of toxicity issues. In 2010, the FDA also joined the Tox21 collaboration and is expected to provide additional expertise and chemical information including human clinical data to improve our ability in toxicity prediction. The data generated from the innovative chemical testing methods used by the Tox21 partnership will be publicly available for making decisions about protecting human health and the environment.

1.3 Technologies for Toxicity Prediction

The goal of toxicity testing is to develop data that can provide adequate protection of public health from the potential adverse effects of exposure to chemicals, drugs, biologics, etc. This goal provides a clear standard for judging the performance of prospective screening assays. An obvious approach for the initial screening of toxicity amongst a large number of environmental chemicals or drug candidates is to combine predictions from different high-throughput assays, and the integration of bioassay results with in silico models such that more accurate weight-of-evidence assessment can be generated. However, for any predictive technology to add value it will be critical that the relevance to humans is fully understood. Whilst no preclinical approach is expected to be totally predictive of potential human health hazards, the strategic application of in silico in conjunction with in vitro, and in vivo animal studies heralds the possibility and opportunity for improving toxicity prediction and human safety assessment.

1.3.1 Computational Modeling Approaches

We have seen over the past decade a continuing interest and focus on the development and application of in silico approaches for the prediction of toxicity. For even with the application of high-throughput screening for in vitro toxicity testing, such testing may still be precluded by cost, timing or other practical considerations. Consequently, the results from in vitro assays are increasingly being integrated with computer modeling to predict compound toxicity for untested or poorly studied chemicals.

Despite their initial promise over the past decade or so, the application of in silico technology for toxicity prediction has now settled down to a more reasonable and rational strategy for developing and applying in silico approaches. Encouragingly, in recent years we have seen increasing focus and improvement in the utility of in silico and computational modeling and simulation approaches. These range from QSAR approaches to the increasing focus on more integrated and more sophisticated systems biology approaches.

Publicly and commercially available toxicity databases represent invaluable sources of toxicology information and cover a diverse array of chemicals of widely differing chemical structures and properties, and toxicity endpoints. Such information is an important resource for the development and validation of computational toxicity modeling.