Scientists using mice also should know how to find other species' databases because it is important to try to give homologous genes in different species the same or similar symbols. Some databases for species most commonly referred to in comparative studies are listed in Table 1.

Finally, I would like to return to the point that many scientists trained in the 1980s and 1990s have not really been trained in practical genetics or animal husbandry. With the current research trend moving back toward phenotype analysis, mutagenesis, and whole animal studies, there is a desperate need to provide programs that train scientists to understand, work with, and maintain genetically defined mice. We need resources to provide training in practical genetics, breeding schemes, record keeping, and mouse husbandry. I think an increasing number of investigators today recognize the effects of genetic background on phenotype and the importance of using genetically defined strains; however, many need resources to help them with the practical aspects of creating and using such strains. For example, The Jackson Laboratory has an annual course called Experimental Genetics that is geared to graduate students, post-doctoral fellows, and investigators changing their research programs to use mice. The course teaches practical Mendelian genetics, how to breed animals, how to keep records to avoid mixing up mice within the colony, and basic animal husbandry. Unfortunately, the course handles only about 30 students a year and, as far as I know, is the only course of its type in this country. We need more of this sort of course introduced into graduate schools or offered in training programs similar to that at The Jackson Laboratory.

Bailey, D. W. 1971. Recombinant-inbred strains. An aid to finding identity, linkage, and function of histocompatibility and other genes. Transplantation 11:325–327.

Fox, R. R., and B. Witham, editors. 1997. Handbook on Genetically Standardized JAX Mice. The Jackson Laboratory, Bar Harbor, Maine.

Lyon, M. F., S. Rastan, and S.D.M. Brown. 1996. Genetic Variants and Strains of the Laboratory Mouse. Oxford University Press, Oxford.

Moen, C. J., M. A. van der Valk, M. Snoek, B. F. van Zutphen, O. von Deimling, A. A. Hart, and P. Demant. 1991. The recombinant congenic strains—A novel genetic tool applied to the study of colon tumor development in the mouse. Mamm. Genome 1:217–227.

Snell, G. D., and H. P. Bunker. 1965. Histocompatibility genes of mice. V. Five new histocompatibility loci identified by congenic resistant lines on a C57BL/10 background. Transplantation 3:235–252.

Stassen, A. P., P. C. Groot, J. T. Eppig, and P. Demant. 1996. Genetic composition of the recombinant congenic strains. Mamm. Genome 7:55–58.

Taylor, B. A. 1989. Recombinant inbred strains. Pp. 773–796 in M. F. Lyon and A. G. Searle, eds. Genetic Variants and Strains of the Laboratory Mouse. 2nd edition. Oxford University Press, New York.

Joseph DeGeorge

Associate Director, Pharmacology and Toxicology

ORM, CDER/FDA

Rockville, Maryland

• Why is the issue of outbred strains of current importance from a regulatory perspective?
• Why was there virtually no interest in the issue several (for example, 10) years ago?
• What makes outbred strains important now?

I believe the answer to all three questions can be summed up in one simple sentence: There has really been a large change in the paradigm of pharmaceutical development. And that is the basis for my concern today.

The change has occurred over the last 10 to 15 years and mainly over the last 5 to 10 years. I address below some specific aspects of the changes that have led to my concern about standardization of the animal models on which we rely. One additional point to keep in mind is that as a regulatory agency, the US Food and Drug Administration (FDA) is the end user of the data from all of the pharmaceutical testing that goes on. We have to rely on it to make judgments about potential human health risks. The outcome is not a research paper but rather, marketing to millions of people around the world of a product that has gone through a particular testing process.

The first of the major changes that have occurred recently in pharmaceutical development is the globalization of the drug development process. It has been mentioned that there is an International Conference on Harmonization of Technical Requirements for International Registration for Pharmaceuticals for Human Use (ICH), where we have harmonized and in fact have agreed on certain standards. However, beyond that is the fact that pharmaceutical companies are almost no longer national. There are very few national pharmaceutical companies, and most market worldwide. Most also develop their drugs as a worldwide activity. Worldwide marketing, in fact, has reached the point where we are developing a common technical document so that the test studies—the same studies—are being submitted simultaneously around the world for marketing approval. This is an attempt to achieve drug approval in Europe, the United States, and Japan at roughly the same time.

The extent of globalization is very large, and "global" companies are headquartered throughout the world. There is a perceived need for global standards that companies can follow to make certain that if they do a study for one country, it is acceptable in another country. This standardization is actually the basis for Office of Economic Cooperation and Development (OECD) guidances, which is familiar to most of you as providing information on various test paradigms and specifically for pharmaceuticals. ICH guidance refers to specific aspects of carcinogenicity testing, genotoxicity testing, and reproductive toxicity testing for pharmaceuticals. If companies follow these guides, no matter where in the world they do the study, that study is accepted internationally—at least within the United States, Europe, Japan, Australia, Canada, Taiwan, and wherever else drugs are manufactured. Companies are looking to these standards to try to establish test systems and hopefully market a pharmaceutical.

One of the accomplishments of this harmonization of standards has been the elimination of duplicate testing. Thus, when an institution or company proposes a drug development plan for Japan, it is not necessary to complete another set of tests for the United States. In the past, that necessity might have been the case; or they would have made certain that the studies were done in both the United States and Japan. Furthermore, this harmonization has actually eliminated many specific national test requirements. The US test requirements do not differ from the Japanese or the European under this ICH process, at least for pharmaceuticals. Such standardization is one aspect of this global drug development plan that has an impact on the use of animals.

Another aspect of globalized pharmaceutical development is the segmentation of toxicology testing. International companies have, in fact, often changed

their development processes. It is often no longer an all in-house operation done in a particular facility where all the data are generated on the same colony of animals, which then undergo every test. There is an increasing use of contract facilities, a blend of contract facilities with sponsor facilities, and an increase of multisourcing outsourcing, going to the lowest bidder. In other words, if it is cheaper to do a certain study in the southwestern United States and another study more cheaply in Japan, they will do those studies in those two places, presumably using the same animal models and the same strain.

One real example is a company that conducts its chronic toxicity studies in their European facility, dose-ranging studies for carcinogenicity studies in their US facility, and then contracts out the carcinogenicity studies. Looking at the results from all those studies from different sources, we have to make a decision about how those results from the chronic study and the dose-ranging study apply in the interpretation of the carcinogenicity study. If the animals appear to be responding differently, we have a big problem. That, then, is what I mean by segmentation of toxicology testing and how it is a major issue in terms of international drug development.

A major concern in terms of carcinogenicity testing—and one reason that we are focusing now on rats rather than mice—is because (as Dr. Usui mentioned, according to the ICH guidance) our long-term carcinogenicity studies are now generally performed using rats rather than in mice. We are using rats mainly because transgenic models are available primarily in mice; so to have two-species testing, the standard 2-year bioassay tends to be done in rats. Another reason is that most pharmaceutical development strategy uses rats as the rodents and dogs as the nonrodents. Because these companies develop a large database on the effects of the pharmaceutical on the rat during their testing, they want to be able to use that information in test approach selection, dose selection, and interpretation of results.

The preceding observations lead to a couple of concerns in terms of carcinogenicity assessment, the first of which are differences in stock survival. Dr. Usui mentioned that these differences might be related to diet. In some cases, they may be related to housing conditions; some companies let their rats get very fat because they put the food in wire baskets and allow them to feed ad libitum. Other companies and facilities put their food in jars and as the rats get fat, they can no longer reach the bottom of the jar. The rats thus undergo a kind of spontaneous dietary restriction, which maintains their weight at a lower level; and those differences, although not necessarily outbred related, can affect the outcomes and interpretations of studies. Diet and body weight are two factors that can clearly affect survival characteristics of the strain and the species. If they are overlaid with differences in survival within a strain of the outbred animals, then you end up with a very difficult problem for interpretation of carcinogenicity data.

Diet, weight, and survival characteristics can also result in differences in