Introduction To TestSmart--Pharmaceuticals: An Efficient and Humane Approach to Predictors of Potential Toxic Effects of Drugs
Proceedings
Background
In May, the Johns Hopkins Center for Alternatives to Animal Testing (CAAT) organized a three-day workshop on pharmaceutical development. "TestSmart-Pharmaceuticals: An Efficient and Humane Approach to Predictors of Potential Toxic Effects of Drugs" examined the current state of pharmaceutical safety testing and considered how recent scientific advances and new technologies will change what we do today.
CAAT initiated the "TestSmart" program in 1999 to suggest a more humane and efficient approach to regulatory testing. Each TestSmart workshop brings leading scientists from industry, academia, and the regulatory community together with members of animal welfare organizations to consider the "Three Rs" of alternatives: replacement, reduction, and refinement. Replacement refers to the use of non-whole animal methods instead of animal tests; reduction means using neither more animals than are necessary nor fewer than are essential to get meaningful results. And refinement, in the alternatives field, refers to the modification of a procedure to eliminate or decrease any associated pain or distress.
Overview
The workshop began with a broad overview of pharmaceutical development, risk assessment, and the potential impact of new technologies by two speakers: James Mayne of Pfizer Inc. and James MacGregor of the U.S. Food and Drug Administration (FDA) National Center for Toxicological Research (NCTR).
In the past, it was very difficult to predict risk, so emphasis was placed on avoiding it. Today, however, it is sometimes possible to predict toxicity and manage it. This is where the pharmaceutical industry stands to make its greatest gains--informed risk management, Mayne said.
Developing a new drug is a complex process. Of 10,000 compounds that show promise, only one will usually end up becoming a candidate for a drug. In turn, about 30% of these candidates will be eliminated in pre-clinical testing; 50% of the remaining drugs are eliminated in the first phase of clinical testing. Nearly 70% of those compounds that survive until the second phase of clinical testing will never make it to market.
At each stage of discovery and development, researchers learn something new about the compound. The traditional approach to toxicology has been to treat animals with a chemical for various durations and observe and measure any change. Newer, less invasive technologies allow this data to be collected earlier in the process, using fewer animals, and with greater accuracy and detail about the mechanisms underlying toxicity.
Toxicity begins with some change in gene expression, which leads to either an increase or decrease in protein synthesis and/or protein modification. This, in turn, affects the ability of cellular systems to adapt, repair damage, or carry out critical activities, producing organ dysfunction and system toxicity. Recent scientific advances and new technology may be able to give us information about each step of this process.
The impact of these advances will lie in our ability to collect early mechanistic data and construct better discovery studies, MacGregor said. These will lead to better early candidate selection, making the animal testing stage both more efficient and more humane. There also will be a better link between non-clinical and clinical testing, because industry can develop more efficient "bridging" biomarkers.
Regulatory issues
The second day of the workshop began with a review of the regulatory environment that controls drug development. MacGregor and Robert Osterberg of the FDA's Center for Drug Education and Research (CDER) presented information about U.S. and international testing requirements, as well as new efforts to streamline the approval process.
Current practice for pharmaceutical development and approval is shaped by a need to establish both safety and efficacy in new drugs. This process is, in turn, defined by four authorities: laws, regulations, guidance documents, and consensus practices. In the development of potential new assays for drug safety and efficacy, we need to know: the relationship of new endpoints to health, the relationship of outcomes in new assays to those from established methods, the relationship between laboratory models and human beings, and the reproducibility, accuracy, sensitivity, and robustness of the test. The only way to obtain this information is through research.
When will the FDA accept data from new methods? When there is scientific consensus on the method within both the scientific community and the responsible FDA center. A critical need exists, MacGregor said, to provide funds for the regulatory and transitional research that will allow this consensus to be developed. The best way to achieve this funding is through industry-academia-government collaboration.
Collaboration among governments and international regulatory agencies is also key to making testing both humane and efficient, said Osterberg. The International Conference on Harmonization seeks to improve the efficiency in the development and regulation of new drugs by removing the need to duplicate animal tests and, therefore, reducing their use in safety and efficacy testing. This process allows drugs to be brought to the international market sooner, because companies are not forced to wait while each country or political region conducts its own regulatory tests. A website provides information on the ICH process.
In the United States, Osterberg said, the FDA is seeking to educate industry about the ICH recommendations and provide guidance on reduction and replacement of animal tests with other approaches (e.g., data sharing, in vitro tests) as well as identifying instances when no animal test is necessary (e.g., drugs with low exposure or high toxicity). Information is available on the FDA website here under "International Conference on Harmonization."
New approaches
New approaches to testing abound. A number of speakers discussed ongoing efforts to provide new tools or new methods for safety testing:
- The use of human cells in in vitro tests has the potential to reduce the use of animal tests while at the same time improving the rate of success in identifying drug candidates for clinical trials, said Albert Li of In Vitro Technologies, Inc. In particular, assays using liver tissue or cells (hepatocytes) offer great potential, due to the liver's key role in metabolism, toxicity, and drug interaction. Technology now allows researchers to cryopreserve hepatocytes, making their use more practical as early drug toxicity screens. Future efforts must focus on expanding the diversity of hepatocyte donors to allow for a more robust population and developing validation approaches to allow comparison of in vitro results with data collected in vivo. Li added that it is important to pick the right in vitro system for use, even with human cells, because otherwise results will not reflect what happens in vivo.
- Most carcinogenic studies involve large numbers of rats and mice, are very long, very expensive, and don't always produce results that apply to human beings. But little data is available on alternative tests. To address this issue, the Health and Environmental Sciences Institute of the International Life Sciences Institute (ILSI) decided to organize a collaborative study of transgenic alternatives to the standard rodent bioassays, said Denise Robinson, executive director of ILSI. HESI organized 53 sponsors to conduct five in vivo tests and one in vitro test on 21 well-understood chemicals. The data produced from 138 assays formed the basis of a large and growing database. In general, the tests were shorter, cheaper and required fewer animals. However, the test results demonstrated that alternative tests are not 100% predictive of human risk, because of the use of other species. Robinson said the results suggest these assays can best be used as part of a "weight of evidence" approach, with a battery of both traditional and alternative tests allowing decision-makers to evaluate drugs. Alternative tests may play a role in prescreening compounds, as substitutes for certain bioassays, or as adjuncts to mechanistic studies.
- Noninvasive imaging techniques offer considerable potential to reduce the numbers of animals required in pharmaceutical testing, and to refine necessary animal tests, said David Lester, associate director of CDER's Office of Testing and Research. They offer the added benefit of applicability to both human and animal subjects. Currently available imaging techniques allow researchers to measure metabolism, molecular function, physiological parameters, and changes in anatomy in real time, on living animals. Noninvasive imaging is faster than traditional tests, more comprehensive, and provides far more information from the same animal. It also allows the development of earlier, and therefore more humane, endpoints. This approach also allows the animal to serve as its own control.
- Biophotonic imaging offers these same advantages and has been developed by Xenogen Corp. specifically as a tool for improved animal research, said Anthony Purchio. Xenogen has used this technology in infectious disease research, oncology and toxicology studies, and drug development, and in each case it has proved superior to traditional, more invasive tests. This technology also may contribute to the development of better in vitro assays.
More information/fewer animals
The second day concluded with several presentations that focused on approaches to help scientists maximize the information they collect from testing, while minimizing animal use:
- The development of high-throughput screening technology, recent advances in combinatorial chemistry, and the success of mapping the human genome all promise an unprecedented increase in the number of potential new drug compounds available for development. Industry needs methods that are rapid and efficient to screen drugs and identify potential toxicity while minimizing animal use. CDER has attempted to create an integrated set of databases, a "knowledge base," drawing upon information submitted to the FDA for review, said Joseph Contrera, director of CDER's Office of Informatics and Computational Safety Analysis. This knowledge base contains information on metabolism, pharmcokinetics, and toxicology, as well as chemical structure. CDER has developed a process to compare an unknown chemical to a known one, using chemical structure as the key field linking databases. This knowledge base, constructed from studies of thousands of compounds, has allowed CDER to score and quantify "carcinogenic potency" of chemicals. Currently, CDER has a contract with NIH to screen all their new pharmaceutical compounds for their potential to act as cancer agents. Ongoing areas of concern include a critical need for uniform compound identification and for better search and retrieval capability within and across databases, as well as issues of quality control and sharing proprietary data.
- A tiered approach to compound development is another way researchers can maximize speed, safety, and efficiency while minimizing animal use, said Oliver Flint, manager for cellular toxicology at Bristol-Myers Squibb Co. Beginning with in silico, or computer-based, approaches, many companies then try simple in vitro models, followed by more complex in vitro models, in vivo animal tests, and finally human clinical studies. So far, no commercially available software achieves all of the goals of the first stage, but they remain useful tools if their limitations are well understood. Validation of new in vitro assays remains a critical issue; while many new assays using human cell lines offer cheaper, more predictive alternatives, they continue to present complications. Emerging research in fields like proteomics will allow new assays to be developed that more closely mimic function in vivo.
- With the advent of commercially available filter arrays and microarrays, it is now also possible to evaluate the effects of compounds with toxicological and carcinogenic properties on gene expression, said Daniel Casciano of NCTR. This tool offers a reasonable alternative to in vivo rat studies. It may play a role in identifying early biomarkers, allowing species comparisons, or developing new in vivo or in vitro models more relevant to human beings.
- The combination of low-, medium-, and high-throughput screens for proteins may provide essential information about the development and mechanisms of diseases like cancer, ultimately leading to new and better drugs, said Emmanuel Petricoin, senior investigator at the FDA Center for Biological Evaluation and Research. In particular, microarrays combined with sophisticated artificial intelligence programs that can "learn" from data sets will allow the scientist to discover new patterns. CBER has applied this approach successfully to studies of human solid tumors as a model for the study of human disease progression.
- Incorporating a search for alternative methods into the traditional literature search scientists conduct can minimize the use of animals while identifying new approaches, said Lisa Libowitz, Director of Communications at CAAT. Altweb, a free online resource managed by CAAT, provides scientists with a useful tool for conducting these searches. The web site (http://altweb.jhsph.edu) offers scientists help through tutorials, a search engine, and links to specialized databases, including a new database on pain management with information on analgesia and anesthesia.
Recommendations and Conclusions
The third day of the meeting was devoted to five breakout sessions. Each group met for several hours in the morning, then presented its conclusions and recommendations to the whole audience:
Group 1: Building a Hypothetical Program Using Novel Technologies and Approaches to Drug Discovery and Development.
- Consider discovery and development as a continuous process, rather than separate phases. Gather as much information about toxicity as possible in the earliest stages, by in silico approaches where possible. Structure information gathering so that when a candidate moves from discovery to development, it brings:
- a working therapeutic index (efficacy/toxicity), developed in silico
- a description of its toxic liabilities and efficacy, based on all models
- information about metabolism, kinetics, etc.
- In the discovery phase, gather information using:
- biomarkers
- bioassays
- transgenics/humanized models
- different species
- In the discovery phase, look at pharmacology and target pathways and identify any problems that may be triggered by the compound; in other words, anticipate toxicity.
Group 2: Ways to Implement 3Rs in the Pharmaceutical Industry
- There is a need to document scientist/non-scientist perspectives on alternatives and the 3Rs and communicate them better, beginning with the definition of terms. For example, scientists use the term "refinement" to mean better, more efficient tests. The animal welfare community and those in the alternatives field use refinement to refer to new methods that decrease or eliminate pain and/or distress in a procedure.
- Reduction issues needing to be resolved include:
- classical toxicology testing versus safety pharmacology
- transgenics: will they increase or decrease animal use?
- harmonization: U.S. versus international testing requirements
- Transgenics is an issue of both reduction and refinement. When groups have good, well-established strains, the numbers of animals used decreases. Until then, however, numbers often increase and refinement issues become much more of a concern. Transgenic animals raise significant questions because it is difficult to predict their likelihood of developing painful conditions or of suffering more than non-transgenic animals.
- Data sharing has led to a decrease in the use of animals, and it is important to support international harmonization efforts.
- OECD's decision to remove the classical LD50 as a test for acute toxicity has led to a decrease in the use of animals.
- Most of the new innovative, non-invasive technologies may be refinements, and most of these are less expensive as well.
- Transgenics may offer opportunities for refinement, but it is too soon to assume this to be true.
- Refinement is a continuing process.
- In vitro models, particularly human in vitro systems, offer great opportunities. But enormous work remains to be done, and all sides of the scientific community (academia, industry, and government) need to encourage this work and support it financially.
- More research needs to focus on human endpoints for in vitro tests. Not much is available currently to replace regulatory tests, although many new tests will reduce the use of animals.
- High-throughput screens have allowed the elimination of many compounds that, formerly, would have been tested on animals.
Group 3: Genomics, Proteomics, and Bioinformatics
- With genomics,
- We need a database on gene expression
- We need a guidance document to development of assays based on the humane genome, to ensure consistency
- The effectiveness of chip arrays for toxicity will depend on how well this approach is used and how well results are analyzed. It has the potential to decrease animal use.
- With proteomics,
- We need consistency in platforms (currently lots of data is coming from multiple platforms)
- Researchers are looking for patterns, rather than known endpoints.
- Work needs to be done to determine which proteomes, which targets to focus on.
- Training is needed on computational approaches.
- Technology is advancing faster than our understanding of its implications
- We need gene versus protein studies
- We need comparison studies of rats versus dogs versus primates versus human beings
- We need protein-protein interaction studies
Group 4: Transgenic Animals: Current and Future Applications in Pharmaceutical Safety Assessment
- Very few transgenic models (4-5) have been assessed for use by the pharmaceutical industry. Lack of experience with these models keeps some companies from using them, and there are concerns about regulatory interpretations of results with new models. However, their use is increasing.
- Transgenic models are not stand-alone models. However, when used as part of a weight-of-evidence approach, they can:
- add certainty or reduce uncertainty
- potentially replace the mouse two-year study
- provide an opportunity to assess the cancer potential of a drug at an early stage or prescreen drugs
- Transgenic models can be both reduction and refinement models, because they can decrease the number of animals in a traditional two-year test and they diminish the suffering connected to aging in two-year mouse studies.
- Additionally, these models may be less costly and provide quicker results.
- More information is needed, including
- peer-reviewed publication of ILSI's study
- public availability of the ILSI database
- a broader database on non-genotoxic models of action
- databases for mechanistic understanding of tumor responses
- Protocol refinement issues still to be resolved include
- the number of animals per dose group
- the number of dose groups
- study duration (will an increase to nine months improve clarity of the study outcome? How will a change of duration effect the value/usefulness of data based on a six-month study?)
- use of positive controls
- questions about strain effects
Group 5: Application of Human Cells in the Evaluation of Drug Toxicity
Human cells and tissues can:
- predict human metabolism and drug-drug interactions
- predict human toxicity
- allow for quantitative analysis of pharmacokinetics of toxicity