Modern medicine and the pharmaceuti- cal industry rely heavily on animal mod- els for drug-target validation. Preclinical drug development is performed almost exclusively in inbred mice and rats. This affords substantial savings by increasing data reproducibility because of the small genetic heterogeneity of the animals. However, these savings are often offset by failed clinical trials, sometimes reflect- ing the inadequacy of inbred animals to model genetic variation in humans. The solution could include using mouse strains closely related to wild mice and more elaborate comparative genomics of the natural variations of new drug- target genes in humans and mice. Animal models, in particular mouse models, for human diseases have made a significant contribution to modern medicine. During the second half of the preceding century, mice have become the most widely used species in biomedi- cal research. It would be impossible to envisage modern medicine without the availability of transgenic mice, knockout mice and inbred strains of mice with de- fined traits, such as obesity, susceptibility to stroke or seizures, and so on. Inbred strains of animals are gener- ated by mating male and female siblings of >21 generations. The founders for a new strain are sometimes selected for a particular trait, such as susceptibility to a certain disease. This ensures that descen- dants of the original pair are almost homogenous genetically and, therefore, respond similarly to treatments provok- ing the onset of disease (such as diet) as well as to drugs designed to combat it. Indeed, marked strain differences exist in the susceptibly of mice to atherosclero- sis 1 , autoimmune diseases 2 , stroke 3 and asthma 4 . Inbred mice: insufficient representation for natural genome diversity Although the advantages of the genetic homogeneity of inbred mice for medi- cal and pharmacological research are obvious, the pitfalls are sometimes overlooked. Complex diseases involve numerous modifying genes that govern disease severity, age of onset, progres- sion rate and the efficacy of available drugs. The most common monogenic disease, cystic fibrosis, is highly variable even in patients carrying the same specific mutation, demonstrating the significant role of modifying genes in its etiology 5 . The contribution of different genes to phenotypic expression and the progression of complex diseases reflects the patient’s share of the natural human- genome diversity, in addition to the contribution of non-genomic factors. Using a single strain of inbred mice cannot reflect the natural variation of the human patient population. For example, using genetically homogenous mice for studying complex human disorders, such as cancer, asthma and diabetes, might obscure the input of crucial human modifying genes. Such pitfalls are also evident when using transgenic and knock- out mouse models, as these are typically produced on specific backgrounds, such as 129/Sv or C57BL6 1 . This might inter- fere with the identification of new drug targets or, worse, lead to heavy invest- ments in drugs that eventually fail in clinical trials. Indeed, <10% of new drugs tested in clinical trials receive Food and Drug Administration (FDA) approval. Many failures are related to unwanted side effects, sometimes reflecting the different drug metabolism in mice and humans; however, some failures are sim- ply a result of lack of efficacy. This might, in some cases, reflect the inadequacy of animal models to mirror human genome variation, and contribute to the escalat- ing cost of new drugs. A study of drug efficacy using a disease model in a single inbred mouse strain could be compared with a clinical drug trial performed in an isolated South Pacific island population. Clearly, such trials in a non-representa- tive selected sample cannot depict the huge intricacy and heterogeneity of human complex diseases. Indeed, Phase III trials ideally involve several medical centers in different countries to allow, among other factors, generalization for different ethnic backgrounds. Likewise, disease models using a single strain of inbred mice cannot reflect the natural variation of the mouse genome faithfully, and the full spectrum of interactions between products of the major disease- related genes and the disease-modifying genes. The need for comparative genomics in humans and mice The realization that our genome is highly variable is changing modern medicine 6 . Humans are much more homogenous than other mammal species because all update feature DDT Vol. 6, No. 15 August 2001 1359-6446/01/$ – see front matter ©2001 Elsevier Science Ltd. All rights reserved. PII: S1359-6446(01)01874-8 766 Animal models and human genome diversity: the pitfalls of inbred mice David Gurwitz, National Laboratory for the Genetics of Israeli Populations, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv 69978, Israel, tel/fax: +972 3 640 7611, e-mail: gurwitz@post.tau.ac.il, and Abraham Weizman, Laboratory of Biological Psychiatry, Felsenstein Medical Center, Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv 69978, Israel