COX-2 Inhibition in Clinical Cancer Prevention

March 2, 2001

Colorectal cancer is an excellent model for studying cancer prevention by means of secondary (eg, polypectomy to remove a precursor adenoma) and primary

ABSTRACT: Colorectal cancer is an excellent model for studying cancerprevention by means of secondary (eg, polypectomy to remove a precursor adenoma)and primary (chemoprevention) strategies. Evidence has shown that regular usersof aspirin and other nonsteroidal anti-inflammatory drugs (NSAIDs) have areduction in risk of colorectal cancer. A possible mechanism of this benefit isdecreased prostaglandin production, which is achieved through inhibition ofcyclooxygenase (COX) activity, and possibly other pathways. Two isoforms of COX—COX-1and COX-2—have been identified. COX-2 is expressed in colorectal adenomas andcarcinomas, both in humans and rodents. Inhibition of COX-2 has been shown todecrease the incidence of carcinogen-induced neoplasia in rats and to lower theincidence of adenomas in murine models. Several COX-2 inhibitors, with thepotential for less toxicity than that associated with traditional NSAIDs, areunder development. This paper reviews potential chemoprevention of colorectalcancer using COX-2 inhibitors in patients at increased risk, eg, patients withfamilial adenomatous polyposis, hereditary nonpolyposis colorectal cancer, andsporadic adenomas. Included are the rationale for use of such agents, results ofa study showing a significant reduction in adenoma burden in familialadenomatous polyposis patients who received the selective COX-2 inhibitorcelecoxib (Celebrex), and the design of other ongoing or planned clinicaltrials. [ONCOLOGY 15(Suppl 5): 21-26, 2001]


In patients who develop adenocarcinomas, there is typically aprogression from normal epithelium, through some inflammatory, metaplastic, or otherintermediate stage, to dysplasia and invasive cancer. The progression is by nomeans invariable and indeed may be reversible. Colorectal cancer exemplifiesthis progression and serves as an excellent model for investigatingopportunities in cancer prevention by means of various secondary (eg,polypectomy to remove the precursor adenoma) and primary (chemoprevention)strategies.

We and many others have begun to explore opportunities in coloncancer chemoprevention through clinical trials involving groups at increasedrisk of colorectal cancer: patients with familial adenomatous polyposis,hereditary nonpolyposis colorectal cancer, and sporadic adenoma. This articlebriefly reviews data from our large trial of a selective COX-2 inhibitor,describes the designs for our current hereditary nonpolyposis colorectal cancerand familial adenomatous polyposis trials (as well as ongoing sporadic adenomatrials), and includes comments on trials with other agents.

Familial Adenomatous Polyposis

Patients with familial adenomatous polyposis have a germlinemutation in the adenomatous polyposis coli (APC) gene.[1] The consequence ofthis is the development of hundreds of adenomas, typically during adolescence.If untreated, as by prophylactic colectomyor proctocolectomy, the risk of colorectal cancer is nearly 100%. Followingcolectomy, subjects remain at risk of duodenal carcinoma and, if proctectomy isnot performed, of rectal cancer; these areas have hence been targeted insurveillance and chemoprevention interventions.

Historically, cancer prophylaxis has consisted of colectomyfollowed by proctosigmoidoscopic surveillance and ablation of recurrent rectalpolyps, or more recently and aggressively, prophylactic proctocolectomy withrestorative ileal pouch reservoir/anal anastomosis. Management of risk ofduodenal neoplasia has been particularly vexing because of the variable naturalhistory of duodenal adenomas; the lack of effective, safe endoscopic measures toablate the flat, spreading lesions; and the morbidity associated with aggressivesurgical intervention—typically pancreaticoduodenectomy.

Though prophylactic surgical interventions are well accepted infamilial adenomatous polyposis, the possibility of medical approaches has beenexplored. The nonsteroidal anti-inflammatory drug (NSAID) sulindac has beenreported to cause complete or near-complete regression of rectal adenomas,initially in uncontrolled trials[2-4] and later in placebo-controlledinvestigations.[5] More modest regression of rectal adenomas has been reportedin two larger placebo-controlled studies.[6,7] No cases of complete regressionwere observed, and adenomas recurred within several months of cessation of thesulindac.[5,6] No long-term efficacy studies of sulindac have been carried out,and there are case reports of cancers occurring while taking sulindac.[8]

Hereditary Nonpolyposis Colorectal Cancer

Hereditary nonpolyposis colorectal cancer has until recentlybeen a clinical diagnosis involving the familial pattern of early-onset and/ormultiple primary colorectal cancer, with or without the presence of certainextracolonic tumors. In many families colon tumors cluster in the right colon.Adenomas may be completely absent and rarely number more than a few. The mode ofinheritance for hereditary nonpolyposis colorectal cancer is autosomal dominant,with penetrance for colorectal cancer estimated at about 80% by age 70. The meanage at cancer onset is approximately 45 years but ranges from 20 to 80+ years,thus overlapping with the age distribution of sporadics.

Stomach and small bowel adenocarcinomas occur in excess, but aresufficiently infrequent that surveillance is not usually recommended. Inaddition to colorectal and other gastrointestinal tumors, hereditarynonpolyposis colorectal cancer includes a number of extraintestinal tumors.Lacking any pathognomonic features, the significance of any given tumor in aparticular patient is problematic. Tumors most commonly involve the endometrium,followed by the ovary and the uroepithelium (ureter and renal pelvis). Peculiarskin tumors (sebaceous adenoma, carcinoma, and keratoacanthoma) occur in asubset of families with the so-called Muir-Torre syndrome.


As in familial adenomatous polyposis, the management ofhereditary nonpolyposis colorectal cancer involves recognition of risk, followedby appropriate surveillance and surgical intervention. All must be enhanced,compared to the average-risk patient. In a sufficiently striking family,colorectal cancer risk to offspring of affected parents approaches 50%.Molecular genetic testing will detect mutations in one of the "hereditarynonpolyposis colorectal cancer genes" in up to 85% of families.

Assuming such a mutation is identified, offspring of affectedparents can be segregated into two groups: those at population risk(noncarriers) and those whose risk approaches 100% (carriers). In carriers,colonoscopy is recommended, beginning at age 20 to 30 and repeated at intervalsof 1 to 5 years (the broad range reflects a lack of hard data and a lack ofconsensus among experts). Noncarriers will require no further enhancedevaluation, assuming accuracy of the genetic testing.

When adenocarcinomas are detected, subtotal colectomy withileorectal anastomosis is urged. Residual risk to the rectum existspostcolectomy, but its magnitude is uncertain and probably not great enough towarrant proctectomy. The approach to the patient with an adenoma is uncertain.Most would perform simple endoscopic polypectomy, but the possibility ofprophylactic colectomy may be increasingly considered in known mutationcarriers, particularly when difficult-to-remove right-sided sessile lesions areinvolved. Surveillance for extracolonic tumors has received little attention.

Molecular Advances

The chronology of molecular advances in hereditary nonpolyposiscolorectal cancer is interesting. Unlike familial adenomatous polyposis, therewas no good clue as to the possible location of a susceptibility locus, thoughseveral "candidate" loci, such as the APC, p53, and DCC (deleted incolorectal carcinoma) genes, were evaluated and excluded. Rather, establishinggenetic linkage in hereditary nonpolyposis colorectal cancer required a searchthrough the human genome, facilitated by the developing library of known, linkedgenetic polymorphisms.

In 1993 such linkage was established to a locus on chromosome 2and quickly confirmed. Within a year, the gene had been cloned and found to showhomology in nucleotide sequence to a member of a yeast DNA "mismatchrepair" gene (MutS). Because linkage studies at this locus had failed toaccount for a majority of families that appeared to have hereditary nonpolyposiscolorectal cancer, additional loci were evaluated and a second gene identified.When cloned, this gene was also found to show homology to a member of the yeastmismatch repair family of genes.

It was then concluded that perhaps other human genes from thismismatch repair family might also account for cases of hereditary nonpolyposiscolorectal cancer. The library of human DNA sequences was searched to determineif there were other areas that showed significant homology to the otherrepresentatives of the mismatch repair genes of lower species. Severaladditional human mismatch repair genes were identified and found to account fora small proportion of hereditary nonpolyposis colorectal cancer families.

As in the case of familial adenomatous polyposis, each of thesediscoveries carried potential implications for management. Establishment oflinkage enabled recognition of carriers of susceptibility through performance oflinkage analysis utilizing polymorphic flanking markers. Cloning of the genesenabled direct testing of individual affected members of hereditary nonpolyposiscolorectal cancer families, without having to resort to linkage analysis.


With recognition of hereditary nonpolyposis colorectal cancer asa distinct entity with predictable colorectal cancer risk, efforts inchemoprevention began. Cats and colleagues[9] administered oral calcium to asmall series of subjects at risk of hereditary nonpolyposis colorectal cancertumors. The trial end point was epithelial proliferation or labeling index, asmeasured by bromodeoxyuridine incorporation. No significant difference inposttreatment labeling index was observed between the study group receiving 1.5g of CaCO3 and the placebo group. The design of our ongoing study, utilizingcelecoxib (Cerebrex), a selective COX-2 inhibitor, is outlined below.

Another multicenter trial centered in Europe[10,11] butintercontinental in scope uses aspirin and resistant (high-amylose, fermentable)starch (Novelose) in a factorial design. Its accrual goals are very ambitious—approximately1,200 subjects from 55 institutions—and accrual is underway at this time. Itis intended to have sufficient power to identify a treatment effect, namelyreduction in adenoma incidence in hereditary nonpolyposis colorectal cancersubjects, during a follow-up period of at least 2 years for each enrolledsubject. Eligibility criteria are similar to those used in our hereditarynonpolyposis colorectal cancer celecoxib trial outlined in this article.

Sporadic Adenoma/Carcinoma

Nonfamilial, sporadic, or common adenomas and carcinomas are amuch more common problem than either familial adenomatous polyposis orhereditary nonpolyposis colorectal cancer. However, the risk of incidentneoplasia in any given subject, even one with a previous neoplasm, iscomparatively low. Consequently, evaluation of impact of intervention requires amuch larger sample size. Thus, in the National Polyp Study evaluation of theimpact of adenoma follow-up by means of colonoscopy, more than 1,400 subjectswere required to demonstrate a significant surveillance benefit. Manyepidemiologic studies have shown benefits of NSAIDs, notably aspirin, inreducing risk of sporadic colorectal neoplasia.[12-15]

Use of COX-2 Inhibitors in Colorectal Neoplasia Chemoprevention

NSAIDs have been shown in several experiments to decreasecarcinogen-induced bowel tumors in rodents.[16,17] In humans, epidemiologicstudies have shown a salutary effect of NSAIDs in reducing rates of sporadiccolorectal adenoma, cancer, and mortality attributed to colorectalcancer.[12-15] Sulindac, a common NSAID, has, in preliminary investigations[2,3]and randomized placebo-controlled trials,[5-7] induced regression of colorectaladenomas in subjects with familial adenomatous polyposis. Unfortunately, thetoxicity of traditional NSAIDs limits their long-term use for cancerprevention.[18]

NSAIDs are inhibitors of the cyclooxygenase enzymes thatcatalyze arachidonic acid metabolism to the prostaglandins, prostacyclin, andthromboxanes. One isoform, cyclooxygenase-1 (COX-1), is constitutively expressedand appears to mediate several physiologic functions. Not surprisingly, then,inhibition of COX-1 is associated with the common side effects of NSAIDs.[18,19]Another isoform, cyclooxygenase-2 (COX-2), is induced. Cytokines and growthfactors seem causally associated with its expression in inflammation andneoplasia.[20-22] Selective inhibition of COX-2, but not of COX-1, appears toreduce gastrointestinal toxicity.[19,23-25] The chemopreventive effects ofNSAIDs may be due at least in part to inhibition of COX-2,[26,27] though non-COX-2mechanisms may be a factor in the chemopreventive effects of both selective andnonselective COX inhibitors.

If NSAIDs inhibit colon carcinogenesis, it is important todetermine if their effect occurs through inhibition of the COX-1 or COX-2isoenzyme or by non-COX processes. Most data support the notion that COX-2 isthe critical mediator, though other pathways may also be involved.[26,28,29] Ithas been established that COX-2 is expressed in colorectal adenomas andcarcinomas, both in humans and rodents. In familial adenomatous polyposis, COX-2has been detected in very small adenomas in mice harboring germ-line APCmutations.[21,27,30] COX-2 inhibitors have been shown to decrease the incidenceof carcinogen-induced neoplasia in rats and to lower incidenceof adenomas in murine APC models.[27,31,32] In one elegant experiment, thepresence of the COX-2 gene was found to be associated with adenomas. In a mousemodel of familial adenomatous polyposis, knockout of one COX-2 allele decreasedadenoma formation, while homozygous deletion resulted in a greatly reducedadenoma burden.[27] This supported the hypothesis that protective action ofNSAIDs is due to inhibition of COX-2.

While a role for COX-2 is now clarified, less is known about theintracellular pathways that mediatesuch effects. COX-2 appears to mediate signaling of mitogenic growth factors. Italso seems to downregulate apoptosis.[33-35] In familial adenomatous polyposis,in which apoptosis is decreased, upregulation throughCOX-2 inhibition ought to be beneficial.[36]

COX-2 Inhibition in FAP

The first concrete evidence of a favorable COX-2 impact on humancolorectal neoplasia came in our familial adenomatous polyposis adenomaregression trial (Figure 1).[37] In order to establish whether a COX-2 inhibitorcould induce regression in the size and/or number of adenomas in familialadenomatous polyposis, investigators at The University of Texas M. D. AndersonCancer Center in Houston and St. Marks Hospital in London carried out arandomized, double-blind, placebo-controlled study of celecoxib.[25] Of 113subjects evaluated endoscopically, 28 were ineligible due to insufficientadenoma burden, while one subject each had a rectal cancer and a large sessileadenoma requiring surgery.

Seventy-five patients were initially randomized to oralcelecoxib at 100 mg twice daily, to oral celecoxib at 400 mg twicedaily, or to a look-alike placebo twice daily for 6 months; the study durationand adenoma regression end point were based on previous trials of sulindac thathad shown efficacy.[2-7] Genetic testing for APC gene mutations was performedwith a positive result in 90% of subjects. Comprehensive compliance and patientsafety monitoring were performed throughout the trial with adverse events gradedaccording to National Cancer Institute (NCI) Common Toxicity Criteria.[38]

Colonoscopy or sigmoidoscopy (if previous colectomy had beendone) and duodenoscopy were carried out at baseline and off-study at month 6,immediately upon completion of the 6-month course of drug or placebo. The examswere documented by videotape and a series of photographs. A very involvedprocess was employed to reliably and quantitatively score polyp density. Theinvestigator performing the scoring was not involved in the endoscopy, and wasblinded as to treatment arm and as to whether the exam was pre- orpost-treatment. This scoring relied on still photographs, though videotapes wereemployed to resolve ambiguities in the photos.

A second, qualitative scoring system was also employed. A globalassessment of the colorectal, rectal (in postcolectomy cases), and duodenaladenoma burden was conducted by each of five endoscopists or surgeonsexperienced in familial adenomatous polyposis. This was done, in the interest ofexpediency, during joint videotape-review sessions, though discussion was notallowed during the viewing and sealed scorings were submitted. A score of"better," "worse," or "same" was required incomparing the videos that were presented in random pairs, ie, baseline vs 6months, and blinded as to temporal sequence and treatment.

Following 6 months of therapy at the highest celecoxib dose, ie,400 mg twice daily (twice the usual antiarthritic dose), there was astatistically significant (P = .003) reduction in adenoma burden compared withbaseline, as measured by still photographs in designated regions of interest.Specifically, the high-dose group experienced, on average, a 28% reduction inadenoma count, compared with a 4.5% reduction in the placebo group, and anintermediate reduction of 12% in the 100-mg twice daily group. This effectpersisted after adjusting for age, sex, previous surgery (colectomy vs intactcolon), baseline polyp burden, and investigating institution.

Applying the more qualitative videotape method of polyp scoring,significant improvement occurred in the 400-mg group in all colorectal regions.In the low-dose (100 mg twice daily) group, there was a nonsignificant trendtoward a treatment response. The celecoxib was well tolerated, with 68%, 56%,and 57% of subjects in the placebo, low-dose, and high-dose groups,respectively, reporting NCI grade 2 or worse adverse event,[38] includingdiarrhea and abdominal pain. Adverse events requiring subject withdrawal fromthe trial included suicide (celecoxib 100-mg arm, in a patient with a previoussuicide attempt), acute allergic reaction (celecoxib 400-mg arm), and dyspepsia(celecoxib 400-mg arm, though no ulcer was observed on upper gastrointestinalendoscopy). No significant changes occurred in hematologic or chemical profiles.

In the aggregate, these data indicate that COX-2 is an importantfactor in colorectal carcinogenesis, and that its selective inhibition mayretard the formation or progression of adenomas, at least in familialadenomatous polyposis.[26] It remains to be seen whether intervention with COX-2inhibitors will prevent, or at least delay, the initial occurrence of adenomasin young familial adenomatous polyposis carriers diagnosed with familialadenomatous polyposis. If so, it may become possible to postpone colectomy orproctocolectomy for a period of years, enabling such young subjects to be moreactive participants in their disease management.

Meanwhile, patients who have already undergone colectomy and whodevelop recurrent rectal polyps may be treated with celecoxib. Most suchpatients and their surgeons are eager to avoid a second operation. If celecoxibis to be considered an adjunct to endoscopic polypectomy in such patients,careful attention must be paid to appropriate follow-up. Anecdotal cases ofprogression to overt malignancy have been documented in subjects treated withsulindac, even as regression of adenomas was documented. Clearly, it is vitalthat patients must be managed on an individual basis.

At the time celecoxib was being approved for use in familialadenomatous polyposis, the US Food and Drug Adminstration (FDA) requested thatseveral additional, postmarketing studies be performed. One such investigationwas to be a clinical trial of celecoxib to prevent the onset of first adenomasin young, genotype-positive, phenotype-negative adolescents carrying APC genemutations. Such a multicenter trial has been proposed and is currently underreview.

A second FDA-mandated study to describe the clinical benefit ofcelecoxib would comprise a registry of clinical outcomes in patients withfamilial adenomatous polyposis. Conceptually, this would entail treatingpolyp-affected adolescents (12 years or older) with the approved dose of 400 mgtwice daily oral celecoxib with an end point of "time to familialadenomatous polyposis-related events" (specifically, familial adenomatouspolyposis-related surgery, gastrointestinal cancer, desmoids, or death). Theseoutcomes would be compared with historical, untreated controls. Adverse eventswould be closely monitored. Preparations are underway for such a multicenter,familial adenomatous polyposis registry-based study.

Another familial adenomatous polyposis trial is opening toaccrual at M. D. Anderson Cancer Center and St. Marks Hospital. This will be atwo-arm, randomized, prospective, double-blind trial in familial adenomatouspolyposis subjects with residual colorectal (no previous colectomy), rectal(postcolectomy), and/or duodenal adenomas (a small group with previousproctocolectomy). Oral celecoxib at 400 mg twice daily will be compared withoral celecoxib at 400 mg twice daily plus difluoromethylornithine at 0.5 mg/m2(rounded to the nearest 250 mg). Target accrual is 152 subjects; endoscopicevaluation will be at 0 and 6 months. The primary end point will be adenomarecurrence, as in the original celecoxib trial. In addition, a region ofrepresentative, dense adenoma involvement will be cleared of adenomas atbaseline, with measurements taken of adenoma recurrence in that area.

Difluoromethylornithine is an irreversible enzyme-activatedinhibitor of ornithine decarboxylase, which in turn is rate-limiting in thepolyamine pathway.[39] Although its pathway differs from that of COX-2inhibitors, it has been shown to decrease carcinogen-induced tumors inrodents.[40] The fact that it uses a different pathway will be advantageous inachieving hoped-for synergy in reducing and preventing adenomas in familialadenomatous polyposis.

COX-2 Inhibition in HNPCC

Studies have shown that COX-2 expression occurs in colorectaladenomas and cancers. However, it may not be as great in colorectal adenomas andcancers in hereditary nonpolyposis colorectal cancer as it is in familialadenomatous polyposis sporadic colorectal cancer.[41] In our series,[41] 16 of24 hereditary nonpolyposis colorectal cancer tumors (67%) and 24 of 26 sporadics(92%) showed evidence of COX-2 immunoreactivity. If confirmed in additionalinvestigations, this would constitute another manner in which the"hereditary nonpolyposis colorectal cancer pathway," characterized byinstability in microsatellite markers and a relative paucity of tumor suppressorgene mutations and allelic losses, differs from that of both familialadenomatous polyposis and sporadic colorectal cancer. Further, to the extentthat COX-2 is relatively underexpressed, inhibitors of COX-2 ought to be lesseffective in hereditary nonpolyposis colorectal cancer.

Because of the relatively low incidence of adenomas inhereditary nonpolyposis colorectal cancer, it would be difficult to conductclinical trials with sufficient statistical power to demonstrate a reduction inadenoma incidence. Nevertheless, we are evaluating the effect of celecoxib onvarious intermediate markers in hereditary nonpolyposis colorectal cancer.Accrual was recently completed, and included 77 subjects with either a mismatchrepair gene mutation or previous microsatellite-instability-positivecolorectal cancer in the appropriate family history setting. The study follows adesign akin to that used in the first familial adenomatous polyposis trial. Thethree arms are placebo, celecoxib at 200 mg twice daily, and celecoxib at 400 mgtwice daily for 1 year. Subjects undergo colonoscopy at baseline and theoff-study exam is carried out at month 12, immediately upon completion of the12-month course of drug or placebo.

The central feature of the study is evaluation of the presenceor absence of COX-2 modulation of markers for apoptosis and proliferation in thenormal-appearing mucosa. Indigo carmine spray for mucosal contrast is performedto identify whether any aberrant crypt foci are present. Gross surfacecharacteristics of aberrant crypt foci are evaluated with a magnifying or zoomcolonoscope and this is correlated with histology. These aberrant crypt fociwill be assessed for COX-2 expression and other markers, including such measuresof apoptosis as TUNEL (terminal deoxynucleotidyl transferase-mediateddUTP-biotin endlabeling) staining and caspase activity.[42,43]

COX-2 Inhibition in Sporadic Colorectal Adenoma

As noted above, laboratory studies in tumor cell lines and inanimal models have demonstrated that COX-2 is important in colon neoplasmformation. Our data from a study in subjects with familial adenomatous polyposisshows that COX-2 inhibition is significantly effective in reducing adenomaburden. However, whether the effect in familial adenomatous polyposis can beextended to the more prevalent problem of sporadic adenoma and cancer of thecolorectum remains to be seen. Nevertheless, these data support the conduct ofcolorectal neoplasm prevention trials in subjects with sporadic, nonfamilialadenomas.

The first large-scale COX-2 inhibitor trial involving subjectswith a history of sporadic colorectal adenoma has recently begun (personalcommunication, M. Bertagnolli, March 2000). This is a phase III, prospective,randomized, double-blind, three-arm, multicenter trial in which celecoxib atmultiple doses (200 and 400 mg twice daily) is compared with placebo. Subjectswill have undergone endoscopic polypectomy of adenoma (³ 1 cm, or two or moreadenomas of any size) within 3 months of study entry. The primary end point willbe recurrence of adenomas at 12 and 36 months following study entry. As in ourfamilial adenomatous polyposis and hereditary nonpolyposis colorectal cancertrials, this sporadic adenoma trial will include measures of surrogate endpoints in a nested subgroup of subjects. Anticipated enrollment will be 650subjects per arm (1,950 in total).


A considerable volume of preclinical data support the safety andefficacy of COX-2 inhibitors, and several clinical chemoprevention trials willsoon provide concrete information about the potential for these agents toprevent both familial and sporadic neoplasia in the colon. Already beingexplored is the possibility of a chemopreventive role for celecoxib in otherorgans. One trial to prevent recurrent bladder dysplasia is being led by a teamfrom M. D. Anderson Cancer Center (personal communication, A. Sabichi, June2000). Other trials are evaluating the effect of celecoxib on Barrett’sesophagus and actinic keratosis. More detailed information about these trials isavailable through the NCI website at


1. Kinzler KW, Nilbert MC, Su LK, et al: Identification of FAPlocus genes from chromosome 5q21. Science 253:661-665, 1991.

2. Waddell WR, Loughry RW: Sulindac for polyposis of the colon.J Surg Oncol 24:83-87, 1983.

3. Rigau J, Pique JM, Rubio E, et al: Effects of long-termsulindac therapy on colonic polyposis. Ann Intern Med 115:952-954, 1991.

4. Winde G, Schmid KW, Schlegel W, et al: Complete reversion andprevention of rectal adenomas in colectomized patients with familial adenomatouspolyposis by rectal low-dose sulindac maintenance treatment. Advantages of alow-dose nonsteroidal anti-inflammatory drug regimen in reversing adenomasexceeding 33 months. Dis Colon Rectum 38:813-830, 1995.

5. Labayle D, Fischer D, Vielh P, et al: Sulindac causesregression of rectal polyps in familial adenomatous polyposis. Gastroenterology101:635-639, 1991.

6. Giardiello FM, Stanley RH, Krush AJ, et al: Treatment ofcolonic and rectal adenomas with sulindac in familial adenomatous polyposis. NEngl J Med 328:1313-1316, 1993.

7. Nugent KP, Farmer KC, Spigelman AD, et al: Randomizedcontrolled trial of the effect of sulindac on duodenal and rectal polyposis andcell proliferation in patients with familial adenomatous polyposis. Br J Surg80:1618-1619, 1993.

8. Niv Y, Fraser GM: Adenocarcinoma in the rectal segment infamilial polyposis coli is not prevented by sulindac therapy. Gastroenterology107:854-857, 1994.

9. Cats A, Kleibeuker JH, van der Meer R, et al: Randomized,double-blinded, placebo-controlled intervention with supplemental calcium infamilies with hereditary nonpolyposis colorectal cancer. J Natl Cancer Inst87:598-603, 1995.

10. Burn J, Chapman PD, Mathers J, et al: The protocol for aEuropean double-blind trial of aspirin and resistant starch in familialadenomatous polyposis: The CAPP study. Concerted Action Polyposis Prevention.Eur J Cancer 31A:1385-1386, 1995.

11. Burn J, Chapman PD, Bishop DT, Mathers J: Diet and cancerprevention: The Concerted Action Polyp Prevention (CAPP) studies. Proc Nutr Soc57:183-186, 1998.

12. Kune GA, Kune S, Watson LF: Colorectal cancer risk, chronicillnesses, operations, and medications: Case control results from the MelbourneColorectal Cancer Study. Cancer Res 48:4399-4404, 1988.

13. Thun MJ, Namboodiri MM, Heath Jr CW: Aspirin use and reducedrisk of fatal colon cancer. N Engl J Med 325:1593-1596, 1991.

14. Rosenberg L, Palmer JR, Zauber AG, et al: A hypothesis:Nonsteroidal anti-inflammatory drugs reduce the incidence of large-bowel cancer.J Natl Cancer Inst 83:355-358, 1991.

15. Giovannucci E, Rimm EB, Stampfer MJ, et al: Aspirin use andthe risk for colorectal cancer and adenoma in male health professionals. AnnIntern Med 121:241-246, 1994.

16. Pollard M, Luckert PH: Indomethacin treatment of rats withdimethylhydrazine-induced intestinal tumors. Cancer Treat Rep 64:1323-1327,1980.

17. Reddy BS, Rao CV, Rivenson A, et al: Inhibitory effect ofaspirin on azoxymethane-induced colon carcinogenesis in F344 rats.Carcinogenesis 14:1493-1497, 1993.

18. Wolfe MM, Lichtenstein DR, Singh G: Gastrointestinaltoxicity of nonsteroidal antiinflammatory drugs. N Engl J Med 340:1888-1897,1999.

19. Vane JR, Botting RM: Mechanism of action ofanti-inflammatory drugs. Scand J Rheumatol Suppl 102:9-21, 1996.

20. Gierse JK, Hauser SD, Creely DP, et al: Expression andselective inhibition of the constitutive and inducible forms of humancyclo-oxygenase. Biochem J 305:479-484, 1995.

21. Eberhart CE, Coffey RJ, Radhika A, et al: Up-regulation ofcyclooxygenase 2 gene expression in human colorectal adenomas andadenocarcinomas. Gastroenterology 107:1183-1188, 1994.

22. Kargman SL, O’Neill GP, Vickers PJ, et al: Expression ofprostaglandin G/H synthase-1 and -2 protein in human colon cancer. Cancer Res55:2556-2559, 1995.

23. Lipsky PE, Isakson PC: Outcome of specific COX-2 inhibitionin rheumatoid arthritis. J Rheumatol 24(suppl 49):9-14, 1997.

24. Simon LS, Lanza FL, Lipsky PE, et al: Preliminary study ofthe safety and efficacy of SC-58635, a novel cyclooxygenase 2 inhibitor—Efficacyand safety in two placebo-controlled trials in osteoarthritis and rheumatoidarthritis, and studies of gastrointestinal and platelet effects. Arthritis Rheum41:1591-1602, 1998.

25. Geis GS: Update on clinical developments with celecoxib, anew specific COX-2 inhibitor: What can we expect? J Rheumatol 26:31-36, 1999.

26. Taketo MM: Cyclooxygenase-2 inhibitors in tumorigenesis(Part II). J Natl Cancer Inst 90:1609-1620, 1998.

27. Oshima M, Dinchuk JE, Kargman SL, et al: Suppression ofintestinal polyposis in Apc delta716 knockout mice by inhibition ofcyclooxygenase 2 (COX-2). Cell 87:803-809, 1996.

28. Kopp E, Ghosh S: Inhibition of NF-kappa B by sodiumsalicylate and aspirin. Science 265:956-959, 1994.

29. Piazza GA, Alberts DS, Hixson LJ, et al: Sulindac sulfoneinhibits azoxymethane-induced colon carcinogenesis in rats without reducingprostaglandin levels. Cancer Res 57:2909-2915, 1997.

30. Boolbol SK, Dannenberg AJ, Chadburn A, et al:Cyclooxygenase-2 overexpression and tumor formation are blocked by sulindac in amurine model of familial adenomatous polyposis. Cancer Res 56:2556-2560, 1996.

31. Reddy BS, Rao CV, Seibert K: Evaluation of cyclooxygenase-2inhibitor for potential chemopreventive properties in colon carcinogenesis.Cancer Res 56:4566-4569, 1996.

32. Kawamori T, Rao CV, Seibert K, et al: Chemopreventiveactivity of celecoxib, a specific cyclooxygenase-2 inhibitor, against coloncarcinogenesis. Cancer Res 58:409-412, 1998.

33. Coffey RJ, Hawkey CJ, Damstrup L, et al: Epidermal growthfactor receptor activation induces nuclear targeting of cyclooxygenase-2,basolateral release of prostaglandins, and mitogenesis in polarizing coloncancer cells. Proc Natl Acad Sci USA 94:657-662, 1997.

34. Vadlamudi R, Mandal M, Adam L, et al: Regulation ofcyclooxygenase-2 pathway by HER2 receptor. Oncogene 18:305-314, 1999.

35. Tsujii M, DuBois RN: Alterations in cellular adhesion andapoptosis in epithelial cells overexpressing prostaglandin endoperoxide synthase2. Cell 83:493-501, 1995.

36. Pasricha PJ, Bedi A, O’Connor K, et al: The effects ofsulindac on colorectal proliferation and apoptosis in familial adenomatouspolyposis. Gastroenterology 109:994-998, 1995.

37. Steinbach G, Lynch PM, Phillips RKS, et al: The effect ofcelecoxib, a cyclooxygenase-2 inhibitor, in familial adenomatous polyposis. NEngl J Med 342:1946-1952, 2000.

38. Common Toxicity Criteria: Cancer Therapy Evaluation Program,Division of Cancer Treatment and Diagnosis, National Cancer Institute, NationalInstitutes of Health, Department of Health and Human Services, March 1998.

39. Pegg AE: Polyamine metabolism and its importance inneoplastic growth and as a target for chemotherapy. Cancer Res 48:759-774, 1988.

40. Kingnorth AN, King WWK, Diekema KA, et al: Inhibition ofornithine decarboxylase with 2-difluoro-methylornithine: Reduced incidence ofdimethylhydrazine-induced colon tumors in mice. Cancer Res 43:2545-2549, 1983.

41. Sinicrope FA, Lemoine M, Xi L, et al: Reduced expression ofcyclooxygenase 2 proteins in hereditary nonpolyposis colorectal cancer relativeto sporadic cancers. Gastro 117:350-358, 1999.

42. Jarry A, Vallette G, Cassagnau E, et al: Interleukin 1 andinterleukin 1 (beta) converting enzyme (caspase 1) expression in the humancolonic epithelial barrier. Caspase 1 downregulation in colon cancer. Gut45:246-251, 1999.

43. Reed JC: Mechanisms of apoptosis avoidance in cancer. CurrOpin Oncol 11:68, 1999.