Animal studies of colon carcinogenesis and altered epithelial cell cliff erentiation

Chronic inflammatory bowel disease (IBD) appears to predispose 
to subsequent colon cancer. Factors that influence the degree of this risk require 
definition since the reported incidence of malignant change varies widely. 
Differing environmental factors such as diet may be critical, and several approaches 
have been used to explore the role of specific variables in colon cancer 
pathogenesis; one has employed the use of animal models. Naturally occurring 
models of colon cancer exist including cotton-topped tamarins with colitis. Best 
studied, however, are animal models of colon cancer induced with specific 
chemical carcinogens. Cycasin and hydrazine derivatives, eg, 1,2-dimethylhydrazine, 
are most widely used. After parenteral administration of an active 
carcinogen, metabolic activation occurs, resulting in colonic adenocarcinomas. 
Sessile and polypoid neoplasms may be induced, particularly in the distal colon, 
similar to human colon cancer. Using this model, the effect of differing dietary 
and therapeutic variables has been explored. Studies with purified single dietary 
fibres, such as microcrystalline cellulose and hemicellulose, but not pectin, have 
demonstrated reduced numbers of colonic tumours; these in vivo observations 
correlate with in vitro effects of fibres on rat luminal and fecal mutagenic 
activities. Specific therapies used in IBD have also been evaluated - metronidazole, 
for example, a bacterial mutagen, enhances the development of chemically 
induced rodent colon cancer. In addition, a significant increase in colonic 
tumour development occurs after intestinal resection or bypass, two procedures 
used in the surgical management of IBD. In this setting, surgical sutures, particularly 
nonabsorbable materials including stainless steel, may play a critical 
role. Although the extent and duration of disease in patients with chronic IBD 
may be important in colon cancer pathogenesis, other variables, including diet 
and treatment, may be critical modulating factors.

C HRONIC INFLAMMATORY BOWEL disease (IBD) appears to predis, pose to subsequent colon carcinoma (1).Although risk seems to be greatest fo r some individuals with longstanding pan col iris (2), the precise magnitude of this risk in specific populations or geographic areas is not clear.Indeed, some studies from referral centres may have overemphasized colon cancer ask in patients with ulcerative colitis.For example, high rates reported from the Mayo C linic in Rochester (2) and Mount Sinai Hospital in New York (3) contrast with lower rates in published studies even from the same areas (4,5).
Risk for cancer in colitis may also re• fleet geographic variables.Studies from the United States (2,3) and the United Kingdom (6) reporting high risks contrast with lower risks from other countries.In Copenhagen , 783 patients were observed for a median period of 6. 7 years and colonic cancer was seen in only seven with a calculated cumulative risk after 18 years wiLh ulcerative colitis of only 1.4% (7).Similar data were recorded by Gilat et a l (8) from Tel Aviv, who fo llowed 504 patients with ulcerative colitis for a mean penod of 7.54 years.A very low incidence of colonic carcinoma of 0.6% was noted, comparable to a si milarly low incidence of 0.5% from Prague, Czechoslovakia (9).Thus, the incidence of malignant Etudes chez l'animal de la carcinogenese colique et de la differenciation alteree des cellules epitheliales: Application possible aux MU RESUME: Les maladies inflammaroires de l'intcstin (Mil) semblent predisposer au cancer colique.La survcnue des transformations maltgnes etant rapportec de fa~on rres diverse, ii est necessaire de definir lcs facteurs qui influent sur le dcgre de risque.Divers facteurs d'environnement tels quc l'alimentation peuvent erre d'imporrance essenrielle ct plusicurs approchcs, parmi lesquellcs l'experimcnrarion sur ['animal, ont servi a explorer le role de ccrtaincs variables ,pecifiques Jans la pathogenel>e du cancer colique.II exiscc des model es de cancer colique survenant naturellemenr chez !'animal les tamarins pinches attemt5 <le colire notamment.Ceux qui se prerent le mieux a l'etude sont routefois les modeles animaux de cancer colique provoque par des carcinogencs chuniques spfofiqucs.La cycasine et les derives de !'hydrazine tels que le dimcthylhy<lrazine ,ont les plus utilises.Ces carcinogencs soot administrcs par voie parencerale er leuracr1varion metaholiquc chez l'hore donnc lieu a des adenomes coliqucs.Des neoplasmes sessiles ou polypoYdes similaires au cancer colique chez l'homme peuvent ainsi erre provoques, surtout a hauteur du colon distal.L'effet de di verses \'ariables d'ordre alimentaire et therapeurique a ains1 etc etudic.Des ctudes utilisant des fibres alimenraircs uniques purifices relies que la cellulo •e microcristalline er l'hemicellulose, mats non la peccine, one demonrre un nombre rcduit de tumeurs coliques; ces observations in vivo rcvelenr qu'il ya correlation enrre lcs effets in vitro des fibres et les activires mutagcnes lurninales et fecal es etud1ees chez le rat.Les tra1tements specifiqucs utilises dans !es MJI Ont ere cgalcmcnt evalues.Le metroni<lazole, par excmple, un mutagene bactt'!nen, favorisc le dcveloppcmcnt du cancer colique produit par les agents ch1m1ques chez !es rongeurs.De plus, on note une augmentation significative du dcveloppcment tumoral au niveau du colon apres resection ou derivation mtcstinale, deux procedures utilisees dans le traitement chirurgical des MIi.En l'occurence, ii se poL .ateque les sutures jouent un role essenriel, surtour lorsqu'elles se composcnt de materiaux non absorbables comme l'acier inoxydable.Bien que l'etendue et la duree des enteropathies inflammatoires chroniques soienr importances dans la pathogenese du cancer colique, d'autres facteurs relies que l'alimentation et les variables rherapeutiqucs pourraienr intervenir de fa<;on determinante.change in ulcerative colitis may vary widely in different population groups; the reasons tor the different incidence rates arc unknown.Geographic vanables may be important, possibly reflecting the influence and interaction of differing generic and environmental factors thought to be critical m the development of de novo colon cancer unrelated to chronic IBD.A number of approaches have been used to explore the role of specific variables m the pathogenesis of colon cancer; both naturally occurring o r experimentally induced animal models of carcinogenesis may be employed.

COTION-TOPPED TAMARIN MODEL
Colon cancer was initially observed in a colony of cotton-topped ramarins (Sagumus oedipus) h oused m Oak Ridge, Tennessee (10).These new world primates were nattve to South America, largely northern Colombia, and were subsequently imported to the United States and Europe for research purposes as well as zoological displays.Since the initial identification of colon cancer m the cotton-topped ramarin over 100 add1t1onal ammals have been described with colonic adenocarc inoma from several animal colonies m different countnes ( 11 ).These findings have been considered strongly suggestive of a largely genetica lly determined neoplasm in this animal species.After the initial recognition of colon cancer, the observation of underlying colitis was also recorded (12).
Emaciation, palpable tumours and lymphadenopathy may be present and banum contrast studies may reveal suspect neoplastic lesions.The rumours Altered epithelial cell differentiation and IBD may vary in size from microscopic foci to large 5 to 6 cm tumours, and these can apparently arise de novo from mucosa! epithelium without prior adenomatous polyp formation; often an acute, chronic or quiescent form of colitis 1s present.A very high percentage of tamarin colon cancers appear co metastasize to regional lymph nodes as well as other tissues, including lung.Metastases to the liver arc apparently rare, possibly due to some anatomical differences in the lymphatic drainage of che large bowel and the apparent lack of portal vein invasion.The colon cancers have been ohserved in both imported and colony-born cotton-topped tamarins from horh sexes, and their intracolornc d1stribut1on approximates 60% from cecum to splenic flexure and 40% from splcnic flexure to the rectum ( 13 ).The histology of colon cancer occurring in t he cotton-topped tamarin has recently been reviewed (14).The carcinoma 1s most typically poorly d1fferenttated, although periodic acid Schiff stains reveal mucin-producing malignant cells.Pre-existing colttts 1s usually present, cancers are often mulcicentric and, interestingly, multiple ileal sites may also show carcinoma.

RODENT MODELS
Despite the lengthening !1st of natural and experimentally induced animal models of IBD detailed elsewhere ( 15 ), supenmposed development of colornc cancer m rodent models is exceedingly rare.Carrageenan, a sulphated polysaccharide fibre (m contrast to other forms of dietary fib re) derived from the red seaweed, has heen associated with the development of an inflammatory process in the intestinal tract of mice, guinea pigs, rabbits and primates ( 16-1 8).Following oral administration, a clinical-pathological syndrome similar to human ulcerative colitis results, except that h1srolog1cal c hanges are reported to be localized to the cecum.Interestingly, chronic carrageenan administratton in the presence, or even absence, of a chemical carcinogen may induce colorecral cancer in these animals ( 19,20).
Other sulphated products of high mo lecular weight, including sulphated amylopectin (21) and sod ium lignosulphonate ( 22) cause lesions similar co those observed in human ulcerative colitis in the cecum and proximal colon.Recently, colon cancer was noted to develop in rats following induction of ch ronic colitis with 2% dextran sulphate (23 ).At three months, two of e igh t rats had carcinoma in the cecum and/or ascending colon .At six months, 11 of 13 rats had carcinoma in the cecum and/or ascending colon and one of 13 had carcinoma of the rectum.

EXPERIMENTALLY INDUCED COLON CANCER
Diffe rent chemical and physical agents have been used to induce de novo colonic neoplasia in experimental animals (15).No agent has produced effects exclusively in the colonic epithelium or been completely reproducible in different animal species.The colonic mucosa of rats, hamsters and mice is highly sensitive to tumour induction with certain chemicals, particularly 1,2-methylhydrazine; in contrast, other agents produce tumours in the colon only incidentally (15 ).Nonnally, small L aboratory rodents have a very low rate of spontaneous colon tumour d evelopment (24 ).T umour inductio n depends on the specific agent used, dosage, route of administration, and time of autopsy fo llowing administration of t he carcinogen.In addition, the species, strain, degree of inbreeding, and sex of the animals are important variables.Besides chemical agents, colonic neoplasia may be induced in experimental animals with ionizing radiation administered in different forms (25,26).
In 1963, colorectal adenocarcinoma was first observed in experimental animals administered cycad meals derived from a t ropical plant, Cycas circinales (29).Subsequently, cycasin, the glucoside derivative of methylazoxymethanol, was observed to be the critical carcinogen in cycad meals (30).In 1967, intestinal tumours were induced in rats with the symmetrical hydrazine derivative, 1,2-dimethylhydrazine (3 1).Dimethylhydrazine is believed co be metabolically activated within the host to an active carcinogen , ie, it is a procarcinogen (32).lt is thought to undergo at least two hepat ic ox idation reactions, although nonhepatic tissues may also be involved (28).The first oxidation step is thought to result in azomethane, an agent expired by the host animal in its respiratory gases.The second oxidation step converts azomethane to azoxymethane, another agent commonly used in experimental colon carcinogenesis.Azoxymethane is t hen n-hydroxylated to methylazoxymethanol, which may decompose spontaneously or be altered by certain tissues enzymatically.In vivo studies indicate that an alkylating intermediate, methyldiazonium, is formed during this decomposition.Methyldiazon ium or a subsequent derivative appears to be capable of methylating cellular nucleic acids, suggesting a possible explanation for the mutagenicity of these agents.Intraluminal factors may play some role, possibly after binary excretion of one or more carcinogen derivatives (28).Luminal bacteria seem co be in volved, since tumour production is decreased in germ-free rats treated with dimethylh ydrazine, but increased in those administered azoxymethane.In add it ion, carcinogen may be delivered to colonic cells by the blood stream.
Oimeth ylhydrazine produces tumours in t he small and/or large intestine.In the small intestine, most t umours are found in the proximal few centimetres, often near the region of the bile ducts, providing additional evidence that either bile itself or some component excreted into bile or pancreatic juice is important.T umoufb in the small in testine are polypoid or sessile lesions and L ocally invasive.In the large intestine, rat epithelial tumours are very similar to h uman neoplasms in both histologic type and distribution.These are pcdunculateJ or sessile, and exhibit varying degrees of cellular atypia and mucus content.Most tumours are found in the distal colon, altho ugh they may also occur in proximal colon; freq uently, tumours are multiple.Most have tubular or tubulo• villous elements; serial sections may reveal local invasion t hrough the muscularis mucosa.Small microulcerations occur on the surface of colonic tumou~ and, occasionally, large ulcerations may be observed with malignant epithelial cells, connective tissue elements, inflammatory exudate and van• able amounts of necrotic debris.Occult Luminal blood loss may be Jetected along with frank rectal bleeding.Most invasive t umours arc well differentiated adenocarcinomas, particularly in the more distal colon.T hese neoplasms form tubular, acinar or papillary scruc• cures, and often t he t umour:, are nonuniform in structure with occasional foci of poorly d iffere ntiated cells.Some tumours produce serosal umbilication.Pedunculated masses may lead to intus• susception and obstruction; protrusion through t he anal verge may occur.More ad vanced lesions may involve local lymph nodes, pericolonic fat and mesen tery, altho ugh metastases to more distant sites, including t he liver or lungs, is less common.W hile most carcino mas occur in the d istal colon, administration of L ower dosages of carcinogen reportedly produces an apparen t shift in distribution to the more proximal colon.Even with large dosage schedules, however, tumours can be fo und in the prox imal colon (33).
In th is model, mucinous adenocarcinomas are also observed.These tend to occur more commonly in the proximal colon, parallelling the dis, t ribution of mucinous or collo id-type adenocarcinomas in humans (34).Usually these tumours are sessile masses, often ulcerated, and may be associated with mucosa!and/or submucosal lymphoid aggregates.In these tumours, clusters of epithelial cells tend to fonn glandular structures.Large ductular spaces and varying amounts of nonepithelial stromal elements and inflammatory cells may be present; the most prominent features of these tumours are considerable amounts of extracellular mucus that appear to distend glandular structures.All muscle coats and serosa •may be involved with these lesions, as they tend to be less well d ifferentiated and more invasive.Undifferentiated or signet ring cell types with intracellular mucus also occur but are rare.Scanning and transmission electron microscopy as well as freeze fracture techniques have provided more refined ultrastructural descriptions of these colonic tumours induced with hydrazine compounds (35,36).Tumour cells, although highly variab le in size and shape, tend to be smaller and more rounded than normal colonic epithelial cells, and their microvilli appear to be fewer, blunted and smaller.In addition, fewer ap ical surface membrane particles are found than particles associated with the microvill i of surface membranes of normal colonic epirhelial cells.Fewer tumour cell surface mucopolysaccharides seem to be present as assessed by ruthen ium red staining, compared to normal colonic epithelial cells.Other histochemical studies have revealed reductions in total mucus content of the tumours and altered staining reactions, similar to findings in humans (37).Chemical and biochemical observations in dimethylhydrazine-treated animals indicate that the carbohydrate content of colonic tumours is reduced compared with noncancerous colonic mucosa from both normal and carcinogen-treated rats (38).In addition, reduced activities of glycosyltransferases are seen, and t he activities of glycosidases are reduced or remain unaltered in colonic tumours (39).
The relevance of synthetic and natural hydrazine compounds to human cancer is not known, although these are present in the environment as industrial and food contaminants (40,41).1,1-Dimethylhydrazine is present in tobacco plants, although a correlation between smoking and colon cancer is not evident (40,42).Hydrazines are found in wild and cultivated mushrooms including Gyromir:ra esculenta and possibly some Helvella species ( 40,43 ).n-Methyl-nformylhydrazine may be converted to methylhydrazine under conditions analogous to those present in the human stomach (44).Hydrazine compounds may be found in rocket propellants while hydroxyethylhydrazinc has been used as a ripener for plants (45,46).Hydrazine and its many derivatives are also commercially used in pesticides, herbicides, blowing agents for plastics and water treatment.Several of these hydrazine analogues can produce tumours in experimental animals (40).

STUDIES USING THE DIMETHYLHYDRAZINE MODEL
The dimethylhydrazine rodent model has been used to examine factors that influence the pathogenesis of colon cancer, including dietary and other therapeutic variables, that may be relevant in patients with IBD.Studies with pure dietary fibres, for example, were previously reviewed (47).Reduced numbers of colon tumours were found in animals fed d ifferent amounts of cellulose or hemicellulose, but not pectin (48-51).These in vivo effects also correlated with the observed in vitro effects of these fibres on rat fecal mutagens (52).Pharmacologic Attered epithelial cell differentiation and IBD agents used in the treatment of IBD have also been evaluated in this model.Metronidazole, for example, an agent first reported for use in 197 5 for IBO (53 ), particularly in perianal Crohn's disease, is known to be a bacterial mutagen (54).In one study, long term use of metronidazole was reported to enhance the development of chemically induced rodent colonic cancers (55).Besides drug therapy, patients with ulcerative colitis or Crohn's disease may require some form of surgical intervention du ring the course of their disease.Two procedures commonly used in the past in this setting have included resection or bypass of diseased intestinal segments.As very significant morphological and functional change may be observed in the residual intestine or bypassed loops of small bowel following these procedures, detailed studies of the effects of both procedures in carcinogen-treated rats were performed.
The incidence, distribution, size and histopathology of rat small and large bowel tumours induced by sequential administration of 1,2-dimethylhyd razine followed by either small bowel transsection, 50% jejunoileal resection or 50% jejunoileal bypass were examined (56).Table 1 shows some of the results from these studies.Even after limited small bowel resection or bypass, intestinal neoplasia was enhanced in both the ~mall and large intestine of the rat.Additional studies were done to define possible mechanisms for changes in tumour incidence observed.No differences in transit times were detectable, but increased luminal ( ie, bacterial) beta-glucuronidase activities in both the cecum and distal colon of the rescctcd, but not bypassed, rats were observed.In addition, an apparent subsite redistribution of small bowel tumours to ileum and large bowel tumours ro more proximal colon in bypassed rats further suggested that the mechanisms involved for this tumour enhancement differed substantial ly from those in resected rats.Interestingly, m this study, at least a portion of the increased numbers of tumours in the small bowel seemed to be related directly to the presence of anastomotic suture lines.

FREEMAN
As a result of this observation, the effects of sequential administration of 1,2-dimethylhydrazine followed by cecal placement of one of six different types of suture materi als were systematically examined (Table 2) (57).S lowly absorbed and/or n onabsorbable suture materials in the absence of a surgical anastomosis promoted local tumour induction in the rat cecum, a site within the rat colon where carcinogen-induced tumour induction is distinctly rare.In addition, cecal suture material composed of multifilament stainless steel wire enhanced tumour development at a 'downstream' site in the distal colon .This was associated with increased fecal ( ie, bacterial) beta-glucuroni<lase activities indicating a possible luminal-mediated mechanism fo r distal colon tumour development in this animal model.

CONCLUSIONS
Although patients with extensive and longstanding ulcerative colitis appear to be at increased risk fo r sub-