Perspectives On Animal Research

Volume 1, Supplement

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

Adenocarcinoma of the Colon (Rats)

(1)

Description of the Model:

In 1963, Lacqueur observed that cycacin, a hydrazine component of the cycad nut, induced occasional colon cancer in rats. Druckrey et al. subsequently screened similar hydrazine compounds for more potent colon carcinogens, and in 1967 they reported that 1,2-dimethyhydrazine (DMH) had many desired attributes.(2) Unlike other hydrazine compounds, DMH consistently produced tumors of the colon, and it rarely caused neoplasms elsewhere, Newberne and Rogers concluded:

The tumors are, in both gross and microscopic morphology, good models for human colon carcinoma. The model is a useful tool for investigations into the pathogenesis and the metabolic and biochemical derangements associated with the induction of colon carcinoma. Use of the DMH rat model may be particularly helpful in studies about refined carbohydrate diets, altered transit time, bacterial populations of the gut, bile salt metabolites and other environmental factors as they relate to colon carcinoma.(1)

Criterion I: Concordance between the Animal Model and the Human Disease

Clinical Presentation:

Newberne and Rogers noted, "Rectal bleeding, obstruction and intussusception, and clinical presentations of colon carcinoma in man are also observed in the DMH rat model."(1) Despite these similarities, there are important differences in clinical presentation. Compared to human patients with spontaneous colon cancer, rats with DMH-induced cancer are much more likely to have multiple primary lesions and to have involvement of the proximal intestinal tract. Perhaps most importantly, the rats tend to die of complications of colon obstruction, and they have a very low rate of metastases. On the other hand, humans who die of colon cancer usually succumb to the metastatic disease, not the primary lesion.

Pathogenesis:

Although there are some gross similarities between colon cancer in humans and DMH-induced tumors in rats, there are several fundamental differences in pathophysiology. First, DMH does not cause cancer in humans. This synthetic compound is not found naturally, and it is not produced for industrial or household use. Furthermore, there is indirect evidence that the mechanism of inducing colon cancer in humans differs from that of rats given DMH. Hill noted, "Cycasin has been consumed by humans on numerous occasions ... and there is still no evidence that it gives tumors of any site in humans (although it is undoubtedly hepatotoxic...)"(3)

In order to obtain an overview of the relationship between human and experimental colon cancer, we evaluated a review article of this model. All of the investigators who used this model acknowledged problems with it, but they concluded that it is valuable. A representative recent review of this model is a 1987 article by Ahnen, entitled "Are animal models of colon cancer relevant to human disease."(4)

Ahnen noted the similar pathologic features of DMH-induced colon tumors and human colonic neoplasms. There are similar epithelial cell proliferation characteristics, and similarities in mucin synthesis alterations due to tumor. However, there are several findings in the animal model that cannot be compared to the human disease because of insufficient data. Ahnen wrote:

Whereas a great deal is known about the metabolism of DMH and its metabolic conversion to the proximate carcinogen, the proximate carcinogen in human colon cancer is unknown. The finding that DMH-induced cancers are less frequent in germ-free animals and animals treated with antibiotics suggests an important role for the colonic flora in DMH-induced carcinogenesis; similar data is not available for human colon cancer. Similarly, selenium, ascorbic acid, disulfiram, cis-retinoic acid, and butylated hydroxyanisole have all been shown to inhibit the induction of cancer in the DMH models, but there is inadequate information to determine whether similar effects are found in human colon cancer. Certainly none of these features can serve as a basis for determining if the DMH model effects a disease similar to human colon cancer.(4)

Ahnen discussed three areas where the DMH-induced model and human colon cancer differ. These are, "...the effect of dietary fat and fiber on colon cancer risk, the presence of an adenoma as a precancerous lesion, and the metastatic potential of colon cancer."(4) Regarding dietary fat, Rogers and Nauss wrote, "Epidemiological studies have consistently implicated dietary fat as a promoter of colorectal cancer in human beings."(5) Rogers and Nauss noted that early animal studies also correlated colon cancer with a high-fat diet, but, "In some of these studies, there was a concomitant lipotrope deficiency, the excess fat was added to lab chow, or the diets were not isocalorically balanced, resulting in decreased consumption of nutrients per calorie by animals fed the high-fat diets."(5) Nauss et al. used equivalent nutrient-to-calorie ratios in low- and high-fat diets, and they found no effect of dietary fat on DMH-induction of colon cancer.(6)

Commenting on the effects of dietary fiber, Rogers and Nauss observed, "Dietary content of fiber and risk for colon cancer correlated inversely in some epidemiological studies, but others have shown no correlation ... current experimental results also are contradictory."(5) Thus, the lab data on dietary risk factors has often conflicted with the best available data on the risk factors for humans. Ahnen wrote:

Is there a serious discrepancy between the animal model and human colon cancer with regard to the influence of diet? I don't believe so because I don't believe that the human data is strong enough to prove an independent association between a high-fat diet or a low-fiber diet and colon cancer risk.(4)

Haenszel and Kurihara's classic study contradicted this opinion. They found that the rate of colorectal cancer in Japanese migrants to Hawaii and California was much higher than that of comparable populations in Japan. Other studies of migrant populations implicated dietary changes as responsible for the increased rates of cancer.(7) Moore and LaMont concluded, "It seems evident that ... adoption of Western dietary habits by migrant populations underlies their increased risk."(7)

Regarding the pathogenesis of colon cancer, the second problem area, Ahnen noted, "It is generally, although not universally, accepted that most human colon cancers arise in benign adenomatous polyps."(4) However, several investigators have not found this sequence in DMH-induced cancer. For example, Pozharisski found, "Experimental intestinal adenocarcinomas developed de novo or, at least, were not preceded by adenomatous polyps."(8) Thus, there appears to be a fundamental difference in the pathophysiology between human colon carcinoma and the experimental model. Nevertheless, Ahnen objected:

Again, I do not believe that a significant disparity between the animal model and human colon cancer exists with regard to the adenoma-carcinoma sequence. In this case, the data for the sequence for human colon cancer seems adequate, but the data suggesting that such a sequence does not occur in the DMH model is clearly inadequate.(4)

Ahnen cited studies indicating that an adenoma-carcinoma sequence may exist in the DMH rodent model. While it may be true that "...the studies in the animal model appear to be inadequate to exclude the adenoma-carcinoma sequence...",(4) it appears that the majority, if not all, DMH-induced carcinomas do not arise from adenomas.

Ahnen last commented on what is perhaps the most problematic issue with this animal model:

There does appear to be a significant difference between the animal models and human colon cancer with regard to metastatic potential of the colonic adenocarcinomas. In human colonic cancer, approximately 1/2 of patients have evidence of lymphatic metastases and 1/3 have evidence of hematogenous metastases at the time of presentation ... Most studies of the DMH models of colon cancer suggest a very low incidence of metastases. The tumors which are capable of metastasis are almost exclusively the mucinous adenocarcinomas of the proximal colon. The adenocarcinomas of the distal colon have not been shown to be capable of metastases. In addition, the metastases that do occur are generally found in regional lymphatics or on the peritoneal surface. Whereas liver metastases are commonly found in the human disease, they are rarely, if ever, found in the DMH models.(4)

This difference in natural history suggests fundamentally different disease processes. Furthermore, because of the lack of metastases, studies with the animal model on the efficacy of therapeutic regimens are profoundly compromised. Nonmetastatic colon cancer in man is successfully treated by resection of a portion of the bowel. Adjuvant therapy, such as chemotherapy or radiotherapy, may be needed to treat patients with metastatic disease, and many patients present with evidence of metastases. However, the DMH model can provide little information about the efficacy of treatments against metastases.

Despite these problems, Ahnen defended this animal model as useful for several types of studies. First, he maintained, "Studies designed to assess the effect of dietary manipulation or chemoprevention can be done more efficiently and rapidly in the animal model than in human colon cancer."(4) While the animal studies are certainly easier and faster, their reliability and applicability is highly questionable. The different effects of dietary fat and fiber on the induction of human and rat colon adenocarcinomas have been discussed above. It is speculative at best that substances that prevent DMh-induced cancer (a condition that never occurs in man) will protect people against the proximate human carcinogen(s), whatever they may be.

Second, Ahnen asserted, "Studies of carcinogen metabolism are possible in short-term cultures of human mucosa, but serial observations or more long-term studies require an appropriate animal model."(4) However, serial observations on a long-term basis are already done in people who are at high risk for cancer, including patients with familial polyposis or ulcerative colitis. Ahnen did not clarify how experimental data might add to the readily available human data. Ahnen continued, "Similarly, studies of colon cancer promoters cannot be performed in the human."(4) However, epidemiologic research can help identify colon cancer promoters.

Ahnen commented that previous studies have been hampered by the requirement that animals must be killed in order to obtain pathological specimens. He wrote:

The need for sacrifice precludes the possibility of directly demonstrating that the preneoplastic mucosa would have subsequently developed into a colon cancer. The mere finding of a particular characteristic in the DMH mucosa prior to the development of identifiable cancers is insufficient to conclude that the alteration is a marker of mucosa at risk.(4)

He continued:

We have determined that serial colonoscopy and biopsy (every two weeks) of the distal rat colon can be performed in the rat ... We believe that the ability to make serial observations will greatly enhance the utility of the rat model of colon cancer."(4)

Ten years previously, Narasawa et al. described colonoscopy in rats with experimental colon cancer.(9) Meanwhile, colonoscopy has been a mainstay in the prevention and diagnosis of colon cancer in human patients. Serial examinations have been critical to the understanding of the pathogenesis of colon cancer in man. This technique in experimental animals should be no more accurate nor more valuable than colonoscopy in humans.

In summary, Ahnen advocated use of the DMH rodent model, but the problems he cited appear to be much more damaging to the usefulness of the model than acknowledged. In particular, the absence of metastases suggests that the host response to the rat tumor differs fundamentally from that of humans. This raises doubts about the reliability of the animal model data for studies of tumor pathogenesis. Furthermore, since metastatic disease is the primary cause of death in human colon cancer, the animal model is of questionable relevance for studies of tumor treatment.

Criterion II: Citations

We located 201 English articles in the Science Citation Index from 1967 to 1988, which cited the 1967 paper by Druckrey.(2) (See Appendix A.) Of these, only four were studies that used human tissues.

1. Trotta and Balis (10) compared tumor-associated alterations in the molecular characteristics of the enzyme adenosine deaminase in both human and rat colon cancers. They summarized:

In normal human colon most of the adenosme deaminase exists as a high molecular weight form (mol wt > 100,000). Colon adenocarcinomas in distinction are characterized by a high proportion of species of apparent average molecular weight 35,000 and 72,000. The proportion of these forms is found to vary from tumor to tumor. Mean specific adenosine deaminase activity is also shown to be elevated in the tumor compared with the surrounding normal tissue ... Chemically induced colon tumors in the rat demonstrate a new main electrophoretic variant of ADase which has the same apparent molecular weight as the normal enzyme (ca. 35,000) but a somewhat higher isoelectric point. These tumors display a specific enzymatic activity about one-half normal.(10)

Thus, while adenosine deaminase activity increases in human colon cancers, it decreases in rodent tumors. Nonetheless, Trotta and Balis believed that ADase abnormalities will be relevant for research with chemotherapeutics. They suggested:

The fact that forms of ADase seem to be tumor-specific makes plausible the development of corresponding tumor-specific inhibitors. A combination of such an inhibitor with an anti-tumor nucleoside which is normally inactivated by deamination would be expected to enhance greatly the specificity of the drug.(10)

Clinical benefit from this observation is, at this time, speculative. But, because the specific deaminases of human beings and rats differ, it is unclear how such studies with this animal model might yield effective chemotherapeutics for humans.

2. Fiala et al. cited previous research in which gamma-glutamyl transferase (GGT) levels were elevated in DMH-induced rat colon carcinomas, and they reported accumulated GGT in a human colon cancer specimen. They wrote:

The similarity between human and rat adenocarcinomas of the colon makes it probable that a detectable increase in GOT activity could be found in preneoplastic human colon just as it was seen in the preneoplastic stage of DMH-induced colon carcinogenesis.(11)

Although elevated GGT levels were found in both rat and human colon carcinomas, we question the value of the animal research. While it appears that this phenomenon was first observed in rodents, it was the choice of the researchers, and not necessity, that prompted them to assess GGT levels in rodent cancers rather than in human tumors, because human colon cancer tissue was readily available.

3. Hirai et al. also cited Druckrey's article in the following passage:

The concept that a majority of human colon cancers develops from preexisting adenomatous polyps is known as the adenoma-carcinoma sequence. This concept has been supported by many clinical and experimental studies, especially by those on familial polyposis coil and villous adenoma."(12)

Two of the clinical reports cited, dated 1925 and 1930, preceded all the experimental studies cited. Furthermore, as was discussed above, it is unclear if the adenoma-carcinoma sequence occurs at all in the DMH rat model.

4. Finally, Quaroni et al. studied antigens in human and rat colon tumors, and they found:

In rats, these antigens were detected in all poorly and moderately differentiated adenocarcinomas of the small and large intestines ... A similar pattern of expression of these antigens was observed in all human colonic adenocarcinomas examined ... The results obtained demonstrated a marked parallelism in antigen specificities between antibodies to rat and human antigens and their expression in both rat and human tumors in situ.(13)

The authors suggested that further studies may help determine the ontogeny of colon cancer cells. They cautioned, however:

Due to the relatively few specimens of malignant, benign, and normal human colons examined, no firm conclusions could be drawn regarding the potential use of the CaCo antigens as histological markers for intestinal colon cancer; in this respect, this study must be considered preliminary.(13)

Furthermore:

...most positive samples examined showed a marked heterogeneity in antigen expression, even in histologically homogeneous regions of the tissues. Thus, although expression of the CaCo antigens could be clearly demonstrated in rat colon tumors, often in apparently large amounts, these antigens cannot be considered necessary markers of the malignant cells.(13)

The value of the CaCo antigens for the study and diagnosis of human colon cancer remains unclear. It is interesting that the DMH-induced rat colon cancers also expressed these antigens. The need of an animal model for the study of these antigens remains questionable, however, because human colon cancer tissue is readily available from colonoscopy biopsies.

Criteria III: Historical Impact

In 1976, Enker and Jacobitz observed, "Most, if not all, of the major advances in the treatment of carcinomas of the colon and rectum have been made through painstaking clinical advances without benefit of animal experimentation."(14) They advocated use of the DMH model, and indeed this animal model of colon cancer has been extensively studied. In order to assess clinical benefit from this model, we considered several recent review articles. No review article addressed all aspects of the pathophysiology, diagnosis, and treatment of this disease.

1. In 1975, Hill questioned the value of animal models in elucidating the etiology of human colon cancer. He observed:

Although there is an abundance of data on colon carcinogenesis in laboratory animals, virtually all of it is irrelevant to the subject of this review ... it must be remembered that the (animal) model systems used are highly artificial and may not relate in any way to human colon carcinogenesis.(3)

2. Moore and LaMont reviewed colon cancer diagnosis in their 1984 paper, "Colorectal cancer: Risk factors and screening strategies," published in the Archives of Internal Medicine.(7) Of the 43 references, only two were reports of animal research. Both of these studies evaluated the dietary effects on the activity of two different enzymes in normal rats. Neither study involved the DMH rat model.

Interestingly, LaMont had written a review of research with experimental models of colon cancer six years previously.(15) At that time, he wrote:

Although one must exercise great caution in extrapolating the results of animal studies to human disease, these animal models will hopefully expand our knowledge of the fundamental biology of colonic neoplasia and ultimately enhance our ability to diagnose and treat colon cancer in man.(15)

Thus; LaMont was familiar with the ongoing laboratory research. Yet, when he co-authored a review of the literature on the evaluation of risk factors for human colon cancer and on strategies for early detection, he did not cite any of the research on animal models of human colon cancer.

3. In 1987, Lise et al. reviewed adjuvant therapy for colon cancer.(16) They cited 74 papers, none of which were animal studies. In human colon cancer, nonmetastatic disease that is treated by colon resection is not a therapeutic problem. The important isue in colon cancer treatment is the management of metastatic disease. Since DMH-induced colon cancers are rarely metastatic, it is not surprising that there was no reference to the animal research in this review.

4. Gilbert suggested that the animal model is relevant for the study of chemotherapeutic regimens because animal tumors have responded to certain chemotherapy agents similar to the response of human colon cancers.(17) Specifically, he discussed the work of Sych et al., which showed that DMH-induced cancers, like human colon cancers, had a small effect in response to combination therapy with adriamycin, methotrexate, 5FU, and cyclophosphamide.18 However, there are several reasons to question the relevance of this study for human patients. First, rodent colon cancer induced by two agents other than DMH did not respond to this regimen. Consequently, it is possible that this finding with DMH was due to coincidence, rather than actual analogous response to therapy. Second, it is possible that the rat tumors responded primarily to one of the four chemotherapeutic agents, while human cancers responded to another. Third, many human patients have microscopic metastases that are not found at the time of diagnosis and that ultimately prove fatal. This rarely happens to rats because they seldom have metastases. Fourth, there was no resection of primary tumors in rats, unlike the standard treatment in patients. Comparing results from such markedly different treatment protocols is tenuous. Finally, Sych et a!. staged the cancers as "A," "B," or "C' based on gross examination at laparotomy. They then compared results to analogous Dukes' classifications, which are determined histologically, not grossly. It is inaccurate to compare lesions defined by gross evaluation with lesions defined histologically.

An important issue in the management of colorectal cancer is the early diagnosis of lesions in patients at high risk, particularly patients with familial polyposis or inflammatory bowel disease. Rogers and Nauss commented, 'The increased risk of cancer in patients with inflammatory bowel disease is sufficiently great that one would expect the factor(s) responsible to be identifiable in experimental models."5 Nevertheless, 'There is no acceptable rodent model for investigation of the association of intestinal cancer and inflammatory bowel disease."5

Conclusions:

It appears that the only conceivable value of this animal model is to note the differences in the biological behavior between DMH-induced rat colon cancer and human colon cancer. In our search for clinical impact from the model, we found only four clinical studies citing Druckrey's paper, and none credited the animal model with a definite clinical contribution. Furthermore, none of the recent review articles cited research with this animal model as relevant to our understanding, diagnosis, or treatment of human colon cancer.

References:

1. Newberne PM, Rogers AE: Adenocarcinoma of the rat, model no. 42, in Jones, TC, Hackel DB, Migaki G (eds). Handbook Animal Models of Human Disease Fasc. 6. Washington DC, Armed Forces Institute of Pathology, 1977.

2. Druckrey H, Preussmann R, Matzkies F, Ivankovic S: Selektive erzeugung von darmkrebs bei ratten durch 1,2-dimethyl-hydrazin. Naturwissenshaften 1967;54:285-286.

3 Hill MJ: The etiology of colon cancer. CRC Crit Rev Toxicol 1975;4:31-70.

4. Ahnen DJ: Are animal models of colon cancer relevant to human disease. Digest Dis Sci 1985;3(h1CJ3S-106S.

5. Rogers AE, Nauss KM: Rodent models for carcinoma of the colon. Digest Dis Sci 1985;30:87S-102S.

6. Nauss KM, Locniskar M, Newberne PM: Effect of alterations in the quality and quantity of dietaiy fat on 1,2-dimethyuiydrazine-induced colon tumorigenesis in rats. Cancer Res 1983;43:4083-4090.

7. Moore JRL LaMont JT: Colorectal cancer: Risk factors and screening strategies. Arch Intern Med 1984;144:1819-1823.

8. Pozharisski KM: Morphology and morphogenesis of experimental epithelial tumors of the intestines. US Nat Canc Inst J 1975;54:1115-1136.

9. Narasawa T, Wong CQ, Weisburger JH: Evaluation of endoscopic evaluation of colon tumors in rats. Am J Dig Dis 1975;20:928-934.

10. Trotta PP, Balls ME: Characterization of adenosine deaminase from normal colon and colon tumors: evidence for tumor-specific variants. Biochemistry 1978;17:270-278.

11. Fiala S, Trout E, Pragani B, Fiala ES: Increased ganuna-glutamyl transferase activity in human colon cancer. Lancet 1979;i:1145.

12. Hirai K, Ueda K, Hayaishi O: Aberration of poly (adenosine diphosphate-ribose) metabolism in human colon adenomatous polyps and cancers. Cancer Res 1983;43:3441-3446.

13. Quaroni A, Weiser MW, Lee S, Amodeo D: Expression of developmentally regulated crypt cell antigens in human and rat intestinal tumors. US Natl Canc Inst J l986;77:405-411.

14. Enker WE, Jacobitz JL: Experimental carcinoma of the colon induced by 1,2- dimethyihydrazine di-HCl: value as a model of human disease. J Surg Res 1976;21:291-299.

15. LaMont JT, O'Gorman TA: Experimental colon cancer. Gastroenterol 1978;75:1157-1169.

16. Lise M, Gerard A, Nitti D, et al.: Adjuvant therapy for colonic cancer: The EORTC experience and a review of the literature. Dis Colon Rectum 1987;30:847-854.

17. Gilbert JM: Adjuvant chemotherapy in large bowel cancer. Canc Treat Rep 1982;9195-228.

18. Sych F, Habs M, Schniahl D: Chemotherapy studies in autochthonous rat tumors. Z Krebsfors 1978;92:105-117.