Celecoxib With Chemotherapy in Colorectal Cancer

April 1, 2002

Cyclooxygenase-2 (COX-2) is the enzyme that normally synthesizes prostaglandins during an inflammatory response.

ABSTRACT: Cyclooxygenase-2 (COX-2) is the enzyme that normally synthesizesprostaglandins during an inflammatory response. Many primary and metastaticcancers express COX-2, and its presence is correlated with tumor angiogenesis,more invasive tumor phenotype, resistance to apoptosis, and systemicimmunosuppression. The expression of COX-2 is associated with a worse prognosis.Inhibition of prostaglandin synthesis may be beneficial in human malignancy.Regular consumption of nonsteroidal anti-inflammatory drugs (NSAIDs) decreasesthe incidence of, and mortality rate resulting from, a number of types ofgastrointestinal cancers. Premalignant colonic lesions regress following theadministration of nonspecific COX inhibitors, such as sulindac (Clinoril).Advanced solid tumor patients treated with indomethacin (Indocin) survive twiceas long as do such patients who receive supportive care alone. The U.S. Food andDrug Administration has approved specific COX-2 inhibitors for the treatment ofarthritis, pain, and familial adenomatous polyposis. Preclinical studies showthat these drugs block angiogenesis, suppress solid tumor metastases, and slowthe growth of implanted gastrointestinal cancer cell lines. The COX-2 inhibitorshave safely and effectively been combined with chemotherapeutic agents inexperimental studies. Ongoing clinical trials are currently assessing thepotential therapeutic role of COX-2 inhibitors in both prevention and treatmentof a diverse range of human cancers. [ONCOLOGY 16(Suppl 3):17-21, 2002]

Colorectal cancer is asignificant cause of morbidity and mortality for men and women in the United States. Overall, almost 6% ofAmericans will develop this disease during their lifetime. Estimates for 2001called for 138,900 incident cases and 57,100 deaths from colorectal cancer,making it the second most common visceral malignancy and the third most commoncancer killer in both genders.[1] Twenty percent of patients present withmetastatic disease,[2] and approximately 30% of patients ostensibly cured bysurgical resection will develop unresectable locally recurrent or distantdisease. The 5-year survival rate for patients with metastatic disease is only6%,[2] suggesting a need for more effective chemotherapy for advanced tumors.

Routine chemotherapy for metastatic colorectal cancer has beenunsatisfactory, although fluorouracil (5-FU)-based chemotherapy has been usedfor 5 decades. The current standard therapy for patients with untreatedcolorectal cancer is the 5-FU plus leucovorin combination given with irinotecan(CPT-11, Camptosar), a topoisomerase I inhibitor.[3] "Standard" NorthAmerican dosing (each drug given weekly × 4 every 6 weeks) achieves responserates of 39% and a median overall survival of 14.8 months, with grade 3/4diarrhea occurring in 23% and severe to life-threatening neutropenia occurringin 54% of patients. Administering at least some portions of the 5-FU bycontinuous infusion decreases toxicity and may be associated with longer mediansurvival.[4] Overall, most patients with metastatic colorectal cancer die within2 years; there is clearly significant room for improvement in outcome.

Arachidonic Acid andEicosanoids: Roles in Carcinogenesis and PotentialChemoprevention

Chemopreventive agents play a role in impeding the development of colorectalcancer, and some of these drugs might be useful in treating established diseaseas well. The arachidonic acid cascade contains enzymes linked to colorectalcancer development, and existing chemopreventive agents impair those reactions.Arachidonic acid, which is derived from the diet, resides in cell membranes inester form with phospholipids. High saturated fat diets promote colorectalcancers initiated by chemical carcinogens,[5] and while the mechanism is not entirely understood, tumor promotionalso may be related to a change in the composition of the colorectal cancer cellmembranes.

Nicholson et al analyzed the fatty acid content of normal colonic mucosa andtumor mucosa from Wistar rats.[6] Weanling rats were fed a low- or highsaturated fat diet, and a subsample of rats in each group received thecarcinogen azoxymethane intraperitoneally. After humane killing, colon andrectum were excised, and fatty acid methyl esters in the cell membranes wereanalyzed. There was a significantly higher proportion of arachidonic acid intumor cell membranes as compared with normal colorectal tissues, regardless ofdietary composition. The higher saturated fat diet was associated with greatertumor promotion than was the low-fat diet.

Eicosanoids are 20-carbon arachidonic acid metabolites that take the form ofprostaglandins, thromboxanes, and leukotrienes. Series-2 prostaglandins arespecific substances hypothesized to have a role in colorectal carcinogenesis,since they modulate the growth of several cell types. Indirect evidence tosupport this exists: arachidonic acid mobilization is linked to a wide varietyof biologic signal transduction pathways,[7] and this process is fairly tightlyregulated in the gastrointestinal (GI) tract. Prostanoid synthesis is enhancedby a variety of growth factors, and PGH2 synthase (another name forcyclooxygenase [COX]) is homologous to the product of a proliferation-associatedgene.[8] Human colonic mucosa is known to have the ability to synthesizemultiple eicosanoids, and tumor cells produce larger quantities of certainprostaglandins than does surrounding mucosa.

Other non-growth-regulated mechanisms for prostaglandin-induced tumorinitiation and promotion exist. For example, tumor growth is enhanced in thesetting of immunosuppression. Colony-stimulating factors released by tumors cancause mononuclear cells to secrete PGE2, which influences activity of T cellsand natural killer cells, the cells that may be involved in immunesurveillance.[9] Prostaglandins regulate platelet function, and tumor-plateletaggregates are proposed to activate cancer cells for vascular attachment,promoting metastases.[10] PGI2, a platelet inhibitor, inhibits metastases ofcolon carcinoma. Also, E-series prostaglandins are angiogenic, and tumor-inducedangiogenesis is strongly tied to growth and metastasis.[11]

Cyclooxygenase-2 (COX-2) Expression and Inhibition

At least two COX enzymes are present in humans: COX-1 and COX-2.[12] COX-1 isa constitutive enzyme involved in homeostasis of tissues such as gastric andrenal epithelium. COX-2 is an enzyme induced by a variety of mitogens,cytokines, and growth factors; it is associated with PG production at sites of inflammation. Eberhart et aldemonstrated that COX-2 gene expression (mRNA) was found in low-to-undetectablelevels in normal colorectal mucosa, but was increased in the majority ofadenocarcinomas studied.[13] Interestingly, up-regulation of COX-2 also wasdemonstrated in a subset of adenomas, the precursors of adenocarcinomas. COX-2expression has been demonstrated in tumors from a variety of GI sites, includingsquamous cell carcinoma and adenocarcinoma of the esophagus, gastric cancer,pancreatic cancer, and well-differentiated hepatocellular carcinoma. It also hasbeen seen in non-GI cancers, such as squamous cell tumors of the head and neck,transitional cell carcinoma of the bladder, non-small-cell lung cancer,cervical cancer, retinoblastoma, prostate cancer, and glioma.

More importantly, COX-2 overexpression is likely related to the acquisitionand maintenance of an invasive metastatic phenotype. Tsujii and DuBois showedthat COX-2 overexpression in rat intestinal epithelial (RIE) cells led tophenotypic changes that could enhance their tumorigenic potential.[14] Thesecells appeared resistant to apoptosis, or programmed cell death, although theresistance could be overcome by adding sulindac (Clinoril), a nonspecific COXinhibitor. Tsujii also transfected human colon cancer cells (Caco-2) with aCOX-2 expression vector and selected cells that constitutively express the COX-2gene.[15] Those cells demonstrated potent extracellular matrix degradingabilities, leading to their becoming sixfold more invasive than control cells.Again, sulindac almost completely inhibited this new property.

Shao and associates described the expression of COX-2 in azoxymethane-inducedrat primary colonic tumors and human metastatic colon carcinomas.[16]Seventy-five percent of metastatic tumors displayed COX-2 immunoreactivity,while COX-1 activity was demonstrated in interstitial, but not tumor, cells.Recently, Tomozawa and associates showed that expression of COX-2 in patientswho had been potentially curatively resected of large bowel malignancycorrelated with risk of tumor recurrence.[17] "High" expression ofCOX-2 (21% of cases) was the only independent factor significantly related todisease-free survival.

Nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit COX enzymes andsubsequently reduce eicosanoid production. Pollard et al demonstrated thatSprague-Dawley rats, which develop intestinal cancers in response tointraperitoneal injection of a dimethylnitrosamine derivative, had significantlysmaller tumors if they were given drinking water containing the nonspecific COXinhibitor, nonsteroidal anti-inflammatory drug indomethacin (Indocin).[18] Inrats receiving other nonspecific NSAIDs, such as sulindac, both the number oftumors per mouse and the number of mice with tumors decreased after treatmentwith the carcinogen 1,2-dimethylhydrazine.[19] Sulindac also slowed growth oftumors present in rats.[20]

Specific COX-2 inhibitors have similarly been tested preclinically. Oshimaand associates used a selective inhibitor on Apcdelta716knockout mice, a modelfor familial adenomatous polyposis.[21] A second knockout, that of the COX-2gene, markedly decreased the size and number of intestinal polyps in offspring.Use of a selective COX-2 inhibitor also significantly decreased polyp development.Sheng and associates demonstrated that treatment of transformed rat intestinalepithelial cells with a selective COX-2 antagonist both inhibited growth andinduced apoptosis.[22]

ClinicalEvidence of NSAID Efficacy in Colorectal Cancer

Evidence shows that NSAIDs can abolish established polyps and possiblyprevent progression to cancer in humans. Frequently, colorectal adenomas areprecursors to bowel neoplasms, and patients with familial adenomatous polyposis(FAP) inevitably develop cancer unless the colon is removed. Gardner’ssyndrome consists of an autosomal dominant variation of FAP, with afflictedpatients also manifesting extra-GI soft tissue growths. Waddell and colleaguesreported the effect of sulindac on colonic polyposis and neoplasms in 11patients with Gardner’s syndrome or familial polyposis.[23] Seven patients hadsubtotal colectomy with ileorectal anastomosis, and four had intact colons. Allwere given sulindac at typical doses, and all but one were observed for 1 to 16years using frequent endoscopy or barium enema. Nearly all remaining polyps inthese groups disappeared over 6 to 12 months, and no colorectal carcinomasdeveloped in any patient. Upon drug discontinuation, two of three patients hadregrowth of polyps.

Rigau and colleagues assessed sulindac treatment in seven patients with FAP,Gardner’s syndrome, nonfamilial polyposis, and multiple hyperplasticpolyposis.[24] All had diffuse colonic polyps. A "marked reduction" inthe number and size of polyps was seen during the first 6 months of therapy, anddiscontinuation of sulindac was associated with recurrence in three of fourpatients. After resuming sulindac treatment, these three again had regression inthe number and size of polyps. In this study, PG generation (PGE2 and6-keto-PGF1alpha) in the colonic mucosa was significantly reduced after 6 monthsof therapy in all patients assessed. This trial was significant, because thepatients did not undergo surgery, which itself has been reported to affect polypgrowth.

Recently, Steinbach and associates demonstrated that FAP patients given theCOX-2 inhibitor celecoxib (Celebrex) at a dosage of 400 mg orally twice dailyfor 6 months had a significantly reduced number of colorectal polyps and overallpolyp burden vs those given placebo or a lower celecoxib dose.[25]

Data from several mature epidemiologic studies support a role for NSAIDs inthe prevention of human colorectal cancer. In a large hospital-basedcase-control study, Rosenberg and associates found that regular and sustainedNSAID use reduced colorectal cancer incidence by 50%.[26] Thun and colleaguestested the chemoprevention hypothesis in a large prospective mortality study andshowed a 40% lower death rate from colon cancer in patients using aspirinregularly.[27]

Celecoxib: A Selective COX-2 Inhibitor

Highly selective COX-2 inhibitors exist and are uniquely suited tochemotherapeutic and chemoprevention trials for colorectal cancer. The COX-2inhibitor celecoxib is 300 times more active against COX-2 than it is againstCOX-1. Unlike aspirin, which covalently binds to COX, celecoxib inhibits withoutpermanently modifying the enzyme. Its anti-inflammatory activity comparesfavorably with that of indomethacin in the carrageenan footpad edema model (ED50= 50 mg/kg) in terms of decreased PG production.[28] Furthermore, selectiveCOX-2 inhibitors appear to be safer than are nonspecific NSAIDs. Celecoxib hasno effect on platelet aggregation or bleeding time, demonstrating a lack ofinhibition of COX-1. Mild to moderate side effects include dyspepsia, diarrhea,and abdominal pain, none of which occurs in more than 10% of patients. Patientswho are free of cancer have been treated with celecoxib doses ranging from 100to 400 mg given twice daily. Celecoxib has been associated with a significantlylower incidence of endoscopically documented and/or symptomatic GI ulcers ascompared with that associated with naproxen or ibuprofen.[29,30]

As with nonspecific COX inhibitors, there is a strong preclinical rationalefor the use of celecoxib in treatment of colorectal cancer. COX-2-derivedprostaglandins induce tumor neoangiogenesis, which promotes tumor growth.Celecoxib potently suppresses this prostaglandin formation and, thus, isantiangiogenic.[31] When given to nude mice implanted with human HT-29 coloncancer cells, celecoxib inhibited both primary tumor growth and lung metastases(personal communication, J. Masferrer, MD, 2000).

The U.S. Food and Drug Administration has approved celecoxib for single-agenttreatment of patients with arthritis and with FAP, and it is reasonable to testthis agent in treatment of human malignancy. Nonselective COX inhibitors havebeen safely combined with chemotherapy. A phase I trial assessed the feasibilityof combining 5-FU/levamisole (Ergamisol) with sulindac.[32] Fifteen patientswith advanced colorectal cancer were given 5-FU/levamisole as employed in theIntergroup study (450 mg/m² of 5-FU given via IV push on days 1 to 5, then onceweekly beginning on day 29; and 50 mg of levamisole orally three times daily ondays 1 to 3 of the first cycle, then repeated for 3 days on alternate weeksbeginning on day 15) plus 150 mg of sulindac twice daily beginning on the firstday of treatment. The principal toxicity was myelosuppression—four of 15patients experienced grade 3/4 neutropenia, two had grade 3 anemia, and noneexperienced severe or life-threatening thrombocytopenia. No chemotherapy-relateddeaths were reported. The opinion was that the addition of sulindac to standardadjuvant treatment did not increase short-term toxicity. An ongoing phase Itrial testing escalating doses of celecoxib in combination with the standard5-FU and leucovorin regimen of the North Central Cancer Treatment Group (NCCTG)has not demonstrated an increase in toxicity over that expected from use ofchemotherapy alone (unpublished data, Oregon Health Sciences University, 2001).

Ongoing trials are testing the role of COX-2 inhibitors not aschemopreventive agents, but as a treatment of established cancer. The HoosierOncology Group is combining celecoxib (400 mg given orally every 12 hours) withirinotecan, 5-FU, and leucovorin in the treatment of metastatic large bowelcancer. Glutamine, a conditionally essential amino acid often depleted inadvanced colorectal cancer patients, is added for its potential role inpreventing or diminishing chemotherapy-associated GI toxicities. This is aparticularly attractive regimen given the possibility that the COX-2 inhibitormight block prostaglandin and thromboxane production induced by thechemotherapy, thus also leading to less diarrhea. Planned accrual is 22patients; early findings show sufficient evidence of activity continuing pastthe first early stopping rule (personal communication, C. Sweeney, MBBS, 2001).

Another ongoing study is an Oregon Health Sciences University-based,multi-institutional trial of chemotherapy plus celecoxib (400 mg orally given twice daily), again in patientswith untreated metastatic colorectal cancer. Early results have demonstratedsome evidence of efficacy in terms of objective response rates; the results alsosuggested that toxicity, particularly neutropenia and diarrhea, might be lessthan that expected with the chemotherapy regimen alone.

A trial that has been proposed to replace the current Intergroup colorectalcancer metastatic disease trial would assess standard American dosing ofirinotecan/5-FU/leucovorin vs an infusional regimen (modified FOLFIRI, anotherregimen containing irinotecan, 5-FU, and leucovorin), plus or minus 400 mg ofcelecoxib bid. That trial has associated translational endpoints: investigatorswill examine polymorphism of genes involved in metabolic pathways offluoropyrimidines (TS, DPD, TP, and others) and irinotecan (UGT1A1, XRCC-1, andothers) to find out if they can be used to predict toxicity and clinicaloutcome. In addition, expression of other genes involving downstream of COX-2inhibition (E-cadherin, integrins, vascular endothelial growth factor, andinterleukin-8) will be correlated with the chemotherapeutic efficacy.


In conclusion, COX-2 inhibitors have a proven role in prevention ofneoplastic disease, and a strong laboratory-based rationale supports their usein combination with chemotherapy in the treatment of established tumors,particularly colorectal cancer. Further research will help to define their valuein colorectal cancer.


1. Jemal A, Thomas A, Murray T, et al: Cancer statistics 2002. CA Cancer JClin 52:23-47, 2002,

2. Thomas RM, Sobin LH: Gastrointestinal cancer. Cancer 75:154-170, 1995.

3. Saltz LB, Cox JV, Blanke C, et al: Irinotecan plus fluorouracil andleucovorin for metastatic colorectal cancer. N Engl J Med 343:905-914, 2000.

4. Douillard JY, Cunningham D, Roth AD, et al: Irinotecan combined withfluorouracil compared with fluorouracil alone as first-line treatment formetastatic colorectal cancer: A multicentre randomised trial. Lancet355(9209):1041-1047, 2000.

5. Nigro ND, Singh DV, Campbell RL, et al: Effect of dietary beef fat onintestinal tumour formation by azoxymethane in rats. J Natl Cancer Inst54:439-442, 1975.

6. Nicholson ML, Neoptolemos JP, Clayton HA, et al: Increased cell membranearachidonic acid in experimental colorectal tumours. Gut 32:413-418, 1991.

7. Parker J, Daniel LW, Waite M: Evidence of protein kinase C involvement inphorbol diester-stimulated arachidonic acid release and prostaglandin synthesis.J Biol Chem 262:5385-5393, 1987.

8. Xie W, Chipman JG, Robertson DL, et al: Expression of a mitogen-responsivegene encoding prostaglandin synthase is regulated by mRNA splicing. Proc NatlAcad Sci USA 88:2692-2696, 1991.

9. Pelus LM: Modulation of myelopoiesis by prostaglandin E2: Demonstration ofa novel mechanism of action in vivo. Immunol Res 8:176-184, 1989.

10. Honn KV, Busse WD, Sloane BF: Prostacyclin and thromboxanes: Implicationsfor their role in tumor cell metastasis. Biochem Pharmacol 32:1-11, 1983.

11. Ziche M, Jones J, Gullino PM: Role of prostaglandin E1 and copper inangiogenesis. J Natl Cancer Inst 69:475-482, 1982.

12. Eberhart CE, DuBois RN: Eicosanoids and the gastrointestinal tract.Gastroenterology 109:285-301, 1995.

13. Eberhart CE, Coffey RJ, Radhika A, et al: Upregulation of cyclooxygenase-2 gene expression in human colorectaladenomas and adenocarcinomas. Gastroenterology 107:1183-1188, 1994.

14. Tsujii M, DuBois RN: Alterations in cellular adhesion and apoptosis inepithelial cells overexpressing prostaglandin endoperoxide synthase 2. Cell83:493-501, 1995.

15. Tsujii M, Kawano S, DuBois RN: Cyclooxygenase-2 expression in human coloncancer cells increases metastatic potential. Proc Natl Acad Sci USA94:3336-3340, 1997.

16. Shao J, Sheng H, Aramandla R, et al: Coordinate regulation ofcyclooxygenase-2 and TGF-B1 in replication error-positive colon cancer andazoxymethane-induced rat colonic tumors. Carcinogenesis 20:185-191, 1999.

17. Tomozawa S, Tsuno NH, Sunami E, et al: Cyclooxygenase-2 overexpressioncorrelates with tumour recurrence, especially haematogenous metastasis, ofcolorectal cancer. Br J Cancer 83: 324-328, 2000.

18. Pollard M, Luckert PH: Effect of indomethacin on intestinal tumorsinduced in rats by the acetate derivative of methylnitrosamine. Science214:558-559, 1981.

19. Moorghen M, Ince P, Finney KJ, et al: A protective effect of sulindac against chemically-induced primary colonictumours in mice. J Path 156:341-347, 1988.

20. Skinner SA, Penny AG, O’Brien PE: Sulindac inhibits the rate of growthand appearance of colonic tumours in rats. Arch Surg 126:1094-1096, 1991.

21. Oshima M, Dinchuk JE, Kargman SL, et al: Suppression of intestinal polyposis in Apcdelta716 knockout mice byinhibition of cyclooxygenase 2. Cell 87:803-809, 1996.

22. Sheng GG, Shao J, Sheng H, et al:A selective cyclooxygenase 2 inhibitor suppresses the growth ofH-ras-transformed rat intestinal epithelial cells. Gastroenterology113:1883-1891, 1997.

23. Waddell WR, Ganser GF, Cerise EJ, et al: Sulindac for polyposis of thecolon. Am J Surg 157:175-179, 1989.

24. Rigau J, Pique JM, Rubio E, et al: Effects of long-term sulindac therapyon colonic polyposis. Ann Int Med 115:952-954, 1991.

25. Steinbach G, Lynch PM, Phillips RK, et al: The effect of celecoxib, a cyclooxygenase-2 inhibitor, in familialadenomatous polyposis. N Engl J Med 342:1946-1952, 2000.

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

27. Thun MJ, Namboodiri MM, Heath CW, et al: Aspirin use and reduced risk offatal colon cancer. N Engl J Med 325:1593-1596, 1991.

28. Gregory SA, Anderson GD, et al: Celecoxib: Pharmacologic inhibition ofinflammation, fever, and prostaglandin production in rodents. GD Searle &Co., Document BRD-94D-1726, November 1994.

29. Silverstein FE, Faich G, Goldstein JL, et al: Gastrointestinal toxicity with celecoxib vs nonsteroidalanti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: The CLASSstudy: A randomized controlled trial. Celecoxib Long-term Arthritis SafetyStudy. JAMA 284:1247-1255, 2000.

30. Simon LS, Weaver AL, Graham DY, et al: Anti-inflammatory and upper gastrointestinal effects of celecoxib inrheumatoid arthritis: A randomized controlled trial. JAMA 282:1921-1928, 1999.

31. Masferrer JL, Leahy KM, Koki AT, et al: Antiangiogenic and antitumoractivities of cyclooxygenase-2 inhibitors. Cancer Res 60:1306-1311, 2000.

32. Sinicrope FA, Pazdur R, Levin B: Phase I trial of sulindac plus5-fluorouracil and levamisole: Potential adjuvant therapy for colon carcinoma.Clin Cancer Res 2:37-41, 1996.