Perspectives On Animal Research

Volume 1, Supplement

An Evaluation of Ten Randomly-Chosen Animal Models of Human Disease

Discussion

Ten randomly chosen animal models of human diseases were reviewed to determine whether or not typical animal models had made valuable contributions to the understanding, diagnosis, prevention, or treatment of the human disease. Only the animal model of anti-tubular-membrane-antibody tubulointerstitial nephritis (anti-TBM TIN) appeared to provide insight into pathophysiology. Too little is known about human anti-TBM TIN, however, to be certain that the animal model is indeed relevant to the human disease process. The authors did not find evidence that any of the ten animal models provided significant contributions to the diagnosis, prevention, or treatment of the analogous human disease.

Many of the animal models appeared to have several characteristics in common with the human diseases. Nevertheless, despite extensive research with most of the models, there appeared to be little clinical impact. Although some animal research confirmed parallels between the human and animal disease processes, the key insights into human disease pathophysiology appeared to come from clinical investigation. A possible explanation is that, in all ten cases, the human condition was described before the animal model was defined in detail. The animal model was recognized or developed after the initial clinical description. Frequently, the animal model featured many of the gross characteristics of the human disease, and researchers predicted that studies with the animal model would provide valuable insights. However, as the human disease was defined in greater detail, fundamental differences between the human and animal disease became apparent. Thus, it is understandable that an animal model might be unlikely to provide important insights into pathophysiology.

For example, the Gunn rat initially appeared to be an "authentic" model of Crigler-Najjar syndrome because investigators noted that both conditions were congenital nonhemolytic hyperbilirubinemias caused by deficiencies of UDPGT. More recent studies, however, have demonstrated that there are differences in UDPGT isoenzymes between C-N patients and Gunn rats. Similarly, LGL leukemia is an aggressive neoplasm in F344 rats, but clinical investigations have revealed that it follows a relatively benign course in humans.

No animal disease is identical to any human disease. Indeed, many promising animal models, such as the osteosarcoma model or the vitamin A-deficiency hydrocephalus model, have been largely abandoned, presumably because differences between the human and animal diseases were discovered. Thus, until results are verified by clinical investigation, it is tenuous to assume that the findings from an animal model are relevant to the human disease. Given this situation, it is difficult for an animal model to provide valuable insight into pathophysiology. Our understanding of a human disease process is limited by our knowledge of the human disease. Conclusions based on the animal data without human clinical verification are speculative, and, in many cases, are inaccurate. Indeed, assuming that the animal model results are applicable to human patients without clinical verification begs the question of the validity of the animal model.

Strengths and Limitations of These Animal Models

Animal models offer several distinct potential advantages in studies of disease pathophysiology. The uniformity of presentation in animal models permits standardization of research techniques, facilitates the development of a database for further research, and allows reproduction of experimental findings. Nevertheless, an animal model that differs fundamentally from the human disease may be irrelevant to the management of the human disease.

Animal models also offer certain advantages for the study of ongoing disease processes. Animals may be killed and tissues studied before a disease reaches its terminal stages. On the other hand, autopsies of human patients may not reveal important features of the early disease process. The early pathological changes in nonterminal, chronic diseases have been, in some cases, difficult to define in human patients. New technologies, however, are helping to overcome this problem. For example, endoscopic biopsy of human colon tissues has helped elucidate the histology of malignant transformation in people, showing, in fact, that the animal model follows a different pathological sequence from that of people. Similarly, in vitro techniques have permitted researchers to study human large granular lymphocytic leukemic cells, even though LGL leukemia is a rare disease in man. Many of the biochemical studies on Crigler-Najjar syndrome and Krabbe's disease have been done using human tissues.

Several investigators commented on such attributes of animal models as "ease of handling," "rapid reproduction," and "readily available tissue." These characteristics may facilitate research; alone, they do not justify use of the animal model. Scientific validity is necessary for an animal model to be relevant for the study of human disease.

Although animal models supply some potential advantages in the study of disease pathophysiology, it is most difficult for an animal model to provide insights into the diagnosis of human diseases. Frequently, differences in disease presentation require that human data form the basis for diagnostic techniques. For all of the diseases studied here, clinical investigations provided the critical diagnostic breakthroughs.

None of the ten animal models appears to have provided important contributions to the therapy of the human disease. Even though different therapeutic regimens were tried with many of the models, the prevailing techniques were not derived from research with these models. In fact, many treatments that appeared to benefit the animals were not useful for people. For example, affinity chromatography and oral agar or cholestyramine improved the hyperbilirubinemia of Gunn rats, but these therapies have not been useful for human patients. Perhaps, the lack of therapeutic contributions from the animal models resulted from differences between the pathophysiology and clinical presentation of the human diseases and those of the animal diseases, and from differences in side effects between human patients and animals.

Validity of this Study

It was not practical to evaluate those animal models that have been discarded as inappropriate. In order to assess the value of animal models in general, it was necessary to choose a representative sample of the animal models deemed by scientists to be useful. Although the animal models chosen at random from the Handbook: Animal Models of Human Disease have been used to varying degrees, all have been established in the scientific literature and all had been used by several investigators. Three of the models, endocarditis in opossums, osteosarcoma in rats, and congenital hydrocephalus in rabbits, have been largely abandoned. Presumably, either increased knowledge about the human disease has revealed that the animal model is not relevant, or better animal models have been developed. Three of the animal models had over 100 citations, and most had more than 40. Thus, although some of the randomly chosen animal models have not been widely used by scientists, many have been mentioned quite extensively.

Because the models were chosen at random, there was no guarantee that all major disease types would be represented in this study. Fortunately, there was a wide range of disease processes. There were models of two rare, hereditary, metabolic diseases -- Crigler-Najjar syndrome and Krabbe's disease. There was an animal model of communicating hydrocephalus, a congenital condition. Four different neoplastic processes were represented. In man, LGL leukemia is very rare and carries low mortality, osteosarcoma is uncommon and carries a poor prognosis, colon cancer is common and follows a variable course, and benign monoclonal gammopathy, a frequent finding among the elderly, is asymptomatic. BMG may also be considered a degenerative condition. Infective endocarditis in the opossum and giardiasis in mice represented animal models of infectious diseases. Finally, anti-TBM TIN provided an example of an immunological disease process.

It was necessary to limit the scope of this study, and thus some relevant information might have been overlooked. For example, there might have been clinically relevant papers that cited the animal model but did not cite the paper(s) chosen for citation review. By selecting the major article(s) that described the animal model, a lengthy list of citations was usually obtained. Although this list might not have included all the work related to the animal model, it was likely to contain most of the important papers. At the least, it was likely to be a representative selection of the literature, Furthermore, using clinical review articles to evaluate historical impact made it unlikely that a major clinical development derived from research with the animal model was overlooked.

Also, we did not address potential usefulness of the animal model for conditions other than the human disease in question. It was not practical to review fully all of the possible applications that researchers have proposed. However, if the model were indeed valuable, it was likely to make some contribution to the understanding or treatment of the human disease process that the animal model was supposed to parallel.

We did not address possible basic science contributions from work with these animal models. For example, several investigators claimed that the Gunn rat has helped elucidate details of bilirubin metabolism and that antiTBM TIN has yielded important basic science understanding of autoimmune mechanisms. Basic science issues, which may be important, were outside the scope of this study. We focused on the clinical impact of the animal research.

For a given human disease, there were often several different animal models. In order to limit the scope of the study, the authors did not evaluate the other animal models in detail. Rather, we addressed only the specific model highlighted in the Handbook Animal Models of Human Disease.

Many previous attempts to determine the historical value of animal models have been very subjective and anecdotal. Although it was impossible to eliminate subjectivity altogether, this study attempted to use objective criteria as much as possible. Criterion I on concordance between the animal model and the human disease quoted both proponents and critics of the animal models. Criterion II, using the Science Citation Index, provided data on the clinical applications of the animal model. When clinical investigators cited the animal model, their reference was reviewed closely to determine whether or not they credited the animal model with any advances. Criterion III, which examined clinical review articles, while perhaps more subjective, provided a clinical, retrospective view of the impact of the animal model. The three criteria used, although not without shortcomings, seemed to be valid and appropriate tools for assessing the clinical impact of an animal model. Thus, it was unlikely that important contributions from the animal models were overlooked. However, it was impossible to determine with certainty that any given animal model had never been, nor ever would be, of clinical value. This, the null hypothesis, cannot be proven.

Conclusions:

Given the current political emphasis to cut health care costs, it is important to provide an accurate cost-benefit assessment of animal research. The results of this study raise doubts about the usefulness of animal models, particularly for the treatment of human disease, Despite the claims of scientists who used the animal models, most appeared to have little, if any, clinical value. Animal models offer certain theoretical advantages, but there are serious limitations that tend to restrict their applicability.