Journal of Clinical Oncology, Vol 23, No 12 (April 20), 2005: pp. 2840-2855
© 2005 American Society of Clinical Oncology.
DOI: 10.1200/JCO.2005.09.051
COX-2: A Molecular Target for Colorectal Cancer Prevention
Joanne R. Brown,
Raymond N. DuBois
From the Vanderbilt-Ingram Cancer Center, Nashville, TN
Address reprint requests to Raymond N. DuBois, MD, PhD, Professor of Medicine and Cancer Biology, 694 Preston Research Building, The Vanderbilt-Ingram Cancer Center, Nashville, TN 37232; e-mail: raymond.dubois{at}vanderbilt.edu
 |
ABSTRACT
|
|---|
Cyclooxygenase (COX), a key enzyme in the prostanoid biosynthetic pathway, has received considerable attention due to its role in human cancers. Observational and randomized controlled studies in many different population cohorts and settings have demonstrated protective effects of nonsteroidal anti-inflammatory drugs (NSAIDs; the inhibitors of COX activity) for colorectal cancers (CRCs). COX-2, the inducible isoform of cyclooxygenase, is overexpressed in early and advanced CRC tissues, which portends a poor prognosis. Experimental studies have thus identified important mechanisms and pathways by which COX-2 plays an important role in carcinogenesis. Selective COX-2 inhibitors have been approved for use as adjunctive therapy for patients with familial polyposis. The role of COX-2 inhibitors is currently being evaluated for use in wider populations.
|
CANCER PREVALENCE
|
|---|
In 2000, estimates for global statistics on cancer rates indicate that there were 10.1 million new cases of cancer, 6.2 million deaths resulting from cancer, and 22 million people living with the disease.1 Often, 10 to 25 years may pass before a neoplastic cell has gone through enough cell divisions and molecular changes to reach a clinically detectable lesion. Cancer can occur at any age, but risk increases in the elderly. An aging population means that cancer incidence will increase; projected rates for 2050 estimate 57% of those cancer cases detected will be in people 65 years of age or older.1 The impact of cancer in our aging world population illustrates the critical need to identify and apply effective preventive measures.
Colorectal cancer is ranked the third most common form of cancer worldwide in terms of incidence (estimated to result in 945,000 new cases, 9.4% of the world total) and mortality (492,000 deaths, 7.9% of the total) in 2000.1 High incidence rates are found in Western Europe, North America, and Australia, with the lowest rates found in the sub-Saharan Africa. The National Cancer Institute estimates 146,940 new colorectal cases will be diagnosed in 2004, with an estimated 56,730 deaths in the United States.2 Screening procedures such as colonoscopy can reduce the risk of colorectal cancer mortality by 50%.3 However, routine screening procedures are received by only approximately 50% of Americans age 50 years or older.4 In the United Kingdom, there are 25,000 to 30,000 new cases of colorectal cancer each year.5 Of these new cases, 17,000 result in death, of which 15,000 are in those age 60 years and older. In the United Kingdom, a considerable number of patients present with metastatic disease, and have less than a 50% chance of survival despite aggressive adjuvant therapy.6 Based on data from the United States, average-risk patients account for approximately 75% of all colorectal cancers and include persons older than 50 years with no other known risk factor; moderate-risk patients account for 15% to 20% of all colorectal cancers and are identified as having a diagnosis of colorectal adenomatous polyps, a personal history of resected colorectal cancer, or a positive family history of colorectal adenomatous polyps or cancer; and high-risk patients account for 5% to 15% of all colorectal cancers and include those with familial cancer syndrome—familial adenomatous polyposis (FAP), hereditary nonpolyposis colorectal cancer or long-standing inflammatory bowel disease.7 The majority of colorectal cancers are thus nonhereditary and sporadic, which makes early detection important. Unfortunately, it is not yet economically feasible to identify all individuals in the general population who are at the highest risk for developing colorectal cancer.8 Therefore, most patients present clinically with advanced disease. Advanced metastatic disease remains incurable for the most part, with current treatment regimens having little effect on 5-year survival. Hence, another important approach to reduce the overall morbidity and mortality of colorectal cancer involves its prevention.
|
CHEMOPREVENTION AND NSAIDs
|
|---|
General preventative measures so far can be divided into those that depend on an alteration of lifestyle and those that depend on direct treatment with nutraceutical and/or pharmaceutical agents. Lifestyle alterations have an impact on colorectal cancer; evidence suggests the intake of high fiber and fish oils and a reduction in saturated fats is sufficient to reduce overall risk.9,10 Unfortunately, evidence from cardiovascular disease studies has indicated the difficulty in devising dietary strategies with high compliance that could be applied on a large scale.10
The International Agency for Research on Cancer (IARC; part of WHO) uses the term chemoprevention to refer to interventions with pharmaceuticals, vitamins, minerals or other chemicals (natural or synthetic) at any of the multiple stages of carcinogenesis to reduce cancer incidence.11 Since colorectal cancer develops over a 10- to 25-year period, a large-scale clinical trial using colon cancer incidence or mortality as an end point would be time consuming, expensive, and difficult to instigate.8 Adenomatous polyps are used as a surrogate end point in some clinical trials in order to determine the efficacy of NSAIDs as chemopreventive agents.7 Prompted by experimental studies, several factors were taken into account when evaluating the antitumor actions of NSAIDs in clinical trials: source of the study population, outcome of interest (adenomatous polyp, cancer incidence, mortality), and reliability of the information regarding NSAID exposure. Prevention of colon cancer by NSAIDs has created considerable interest because of the enormous amount of clinical experience with NSAIDs in the general population.12
It is estimated that at least 20 to 30 billion aspirin tablets are purchased annually in the United States alone, and that 1% to 2% of the world population consumes at least one aspirin tablet daily.13 At present there are at least 15 NSAIDs on the market, being prescribed at a rate of 70 million NSAID prescriptions per year for individuals suffering from chronic inflammation and pain; for example 10 to 15 million individuals afflicted with rheumatoid arthritis (RA) and osteoarthritis in the United States.13 However, use of aspirin and other NSAIDs accounts for $5 to $10 billion in hospitalization charges and lost work time, and 26,000 deaths per year. The adverse effects attributable to NSAIDs are thought to arise from the inhibition of the constitutive isoform of the target enzyme, cyclooxygenase (COX-1). COX-1 has been hypothesized to function as a housekeeping gene responsible for example, for the production of cytoprotective prostaglandins (PGs) in the gastrointestinal (GI) tract; whereas COX-2 is an immediate-early gene thought to be involved in inflammation, mitogenesis and/or specialized signal transduction mechanisms.14 Hence, the anti-inflammatory, analgesic, and antipyretic properties of NSAIDs are attributed to the inhibition of COX-2, and the GI ulcerations induced by NSAIDs are thought to be due to the inhibition of COX-1. This relationship between NSAIDs, COX-1, and COX-2 has provided the impetus to develop NSAIDs devoid of the adverse effects associated with traditional NSAIDs but which retain COX-2 selectivity to inhibit inflammatory parameters. However, we now know that this original hypothesis was overly simplistic and that even the use of highly selective COX-2 inhibitors is associated with some GI toxicity. Despite these limitations, targeted COX-2 inhibition may have an important role for use in the chemoprevention of colorectal cancer.
|
BIOCHEMISTRY OF COX-2
|
|---|
Arachidonic acid (5,8,11,14-eicosatetraenoic acid) is derived either directly from the diet or from modification of linoleic acid, and resides mostly in an esterified form in cell membranes (Fig 1). Metabolites of the arachidonic acid cascade include products of COX (or PGH synthase [prostaglandin endoperoxide synthase; prostaglandin G/H synthase]; EC1.14.99.1) isoenzymes which utilize membrane-bound arachidonic acid as its substrate. COX has to compete for substrate with lipoxygenase and cytochrome P450 enzymes, the metabolites of which are collectively known as eicosanoids, meaning "twenty" because they are derived from the 20 carbon arachidonic acid. Arachidonic acid is released via the action of cellular phospholipases, liberating free substrate to the active site of COX enzymes.15 COX is a bifunctional (it sequentially catalyzes cyclooxygenase and peroxidase reactions), membrane-bound hemeprotein that catylases the bisoxygenation of arachidonic acid to form PGG2 and the peroxidative reduction of PGG2 to form PGH2.16 COX inserts two molecules of oxygen into the sn-position of arachidonic acid, producing the unstable PGG2, which is rapidly reduced by the hydroperoxidase activity of the COX enzyme to PGH2, the precursor for all prostaglandins. PGH2 is unstable and is subsequently converted to PGs by a family of prostaglandin synthases that are specific for each prostaglandin. The end products of this pathway generated by various cell types reflect the particular cellular abundance of synthase enzymes that utilize PGH2 as a substrate (Fig 1). 17
View larger version (83K):
[in this window]
[in a new window]
|
Fig 1. The arachidonic acid cascade. Arachidonic acid is released via the actions of phospholipases, and is channeled to the cyclooxygenase (COX) enzymes. Cyclooxygenase sequentially catalyzes cyclooxygenase and peroxidase reactions to form PGG2 followed by PGH2, the precursor for all prostaglandins (PGs). TXA2, thromboxane A2.
|
|
NSAIDs are chemically unrelated compounds that share a common therapeutic action. The best known of these are aspirin, ibuprofen, piroxicam, indomethacin, and sulindac. Aspirin is an irreversible inhibitor of the COX active site, covalently modifying the COX protein by acetylating a single serine residue in the substrate-binding channel, blocking the approach of arachidonic acid. Indomethacin, piroxicam, ibuprofen, and sulindac are competitive inhibitors that noncovalently bind to the protein in the substrate channel. The structural differences between COX-1 and COX-2 have been exploited by pharmaceutical companies to develop selective COX-2 inhibitors.18-20 The active site of COX-2 is larger than that of COX-1 with an additional side-pocket18 that can accommodate larger structures. The replacement of isoleucine by valine in the binding site of COX-1, for example, removes constriction in the mouth of this secondary pocket, which results in access by more bulky molecules.21 Hence, compounds specifically designed to bind and fit this additional space are usually potent and selective COX-2 inhibitors.
Of the selective COX-2 inhibitors, three in the United States and global markets include celecoxib (Searle/Pharmacia/Pfizer, New London, CT), rofecoxib (Merck, Whitehouse, NJ), and valdecoxib (Pharmacia/Pfizer). All have been approved for use in the management of pain and inflammation associated with RA, osteoarthritis, and primary dysmenorrhea. Not only have all three drugs shown similar efficacy to the traditional NSAIDs in treating acute and chronic inflammatory conditions, but these agents have also significantly improved GI tolerability and platelet safety profiles.
|
EVIDENCE OF NSAIDs AS CHEMOPREVENTIVE AGENTS
|
|---|
During the development of colorectal cancer, PG production is significantly increased in malignant tissue22,23 compared with adjacent normal mucosa, particularly PGs of the E series.24 Sulindac treatment of patients with FAP provided the first insights into the potential chemopreventive actions of NSAIDs. Waddell & Loughry25 showed that sulindac treatment for 1 year could decrease polyp multiplicity and induce regression of polyps in patients with FAP and Gardner’s syndrome. A follow-up report 6 years later demonstrated that cessation of treatment resulted in adenoma recurrence,26 suggesting a cytostatic rather than a cytotoxic effect. Since these studies, a large body of evidence has now emerged revealing a 40% to 50% reduction in colorectal cancer in individuals taking NSAIDs regularly, either in the context of sporadic colorectal cancer or in FAP patients27 (Tables 1 and 2, respectively). These studies comprise observational as well as randomized controlled studies in many different population cohorts and settings demonstrating strong protective properties of NSAIDs; mostly aspirin, but including ibuprofen, indomethacin, and piroxicam against the risk of developing colorectal cancer and reducing mortality related to colorectal cancer (Table 1). Thun et al28 conducted the largest observational study to date on the use of aspirin and colon cancer, in which aspirin use reduced the relative risk of cancer mortality by approximately 40% when consumed more than 16 times per month for at least 1 year. Two recent randomized, double-blind, controlled studies have both confirmed these observational studies, showing aspirin can protect against new adenoma formation thus delaying adenoma development.29,30 With regard to optimal dose regimen, Baron et al30 showed that aspirin only taken at a dose of 81 mg/day and not at the higher dose of 325 mg/day conferred a reduction in advanced adenoma incidence. The reason why no significant antitumor effect in the 325 mg/day group was seen is not certain, because Sandler et al29 showed 325 mg/day was effective. Moreover, patient populations with FAP are sensitive to selective COX-2 inhibition31,32 (Table 2), such that a significant reduction in the number and burden of colorectal polyps provided supporting data for the approval of celecoxib as adjunctive therapy to endoscopic surveillance and surgery for FAP patients. The effect of selective COX-2 inhibition for the chemoprevention of common sporadic colorectal cancers remains to be investigated but is the focus of many international ongoing clinical trials, as well as the effectiveness of other NSAIDs (Table 3). An issue of the long-term safety of selective COX-2 inhibitors also needs to be considered in these future trials, since a potential prothrombotic effect of these inhibitors is of intense debate.33 Randomized controlled studies (VIGOR34 [Vioxx Gastrointestinal Outcomes Research] and CLASS35 [Celecoxib Long-term Arthritis Safety Study]) revealed good upper GI protection compared with other NSAIDs when administered at therapeutic doses. However, the potential inhibition of endothelial cell–derived COX-2 activity and subsequent PGI2 production, which may shift the homeostatic balance toward more thromboxane A2 (TXA2) effects, promoting platelet aggregation and leading to increased cardiovascular events remains controversial.36 These drugs would require clinical and mechanistic safety assessments in patients with known atherosclerotic disease. In this regard, a recent observational study using 54,475 older adults (age 65 years or older) showed a significantly elevated relative risk of acute myocardial infarction (AMI) with rofecoxib > 25 mg/day compared with celecoxib (odds ratio [OR], 1.24; 95% CI, 1.05 to 1.46) or no NSAID treatment (OR, 1.14; 95% CI, 1.00 to 1.31) only during the first 90 days of drug exposure.37 These increased risks were observed in a dose-specific manner, whereas celecoxib was not associated with an increased risk of AMI. Similarly, a large population-based study showed individuals age 66 years or older were at higher risk of admission for congestive heart failure when taking rofecoxib compared with those taking either nonselective NSAIDs or celecoxib.38
Patients with RA who regularly take NSAIDs also have a reduced incidence of colorectal cancers.39 However, as mentioned in preceding paragraphs, arthritis appears to be a disease that predisposes the patient to the development of ulcers.40 Trujillo et al41 outlined a risk-benefit scenario based on the known effects of aspirin on cancer prevention and intestinal complications. Of 100 individuals older than 50 years, as many as 40 will develop adenomatous polyps, and five will develop colon cancer in their lifetime. If these 40 polyp-bearing individuals were treated with aspirin as a chemopreventive agent, 1.2 patients per year will develop a serious intestinal complication as a result, rising to six individuals within 5 years. Although aspirin could prevent 2.5 colon cancer–related deaths, the risk of a serious ulcer will have exceeded the potential lifetime benefit of preventing five colon cancers. These estimates indicate the need to determine whether long-term usage of selective COX-2 inhibitors would retain their antitumor activities while being devoid of GI complications.
|
EXPERIMENTAL EVIDENCE
|
|---|
In regard to experimental evidence, nonselective NSAIDs have been known to inhibit cancer formation in rodent models of colorectal cancer since the 1980s, and the development of selective COX-2 inhibitors has shown equal promise in rodent models and in models of colonic carcinogenesis, genetically modified tumor development, or human cells grown in nude mice.42-52 For example, selective COX-2 inhibitors can reduce tumor formation in Apc 716 (adenomatous polyposis coli716) mice46 and in Min (multiple intestinal neoplasms) mice.47 However, sometimes the doses needed to exhibit an antitumor effect were in excess of anti-inflammatory doses. For example, Reddy et al, 48 using celecoxib in azoxymethane (AOM)-induced mouse carcinogenesis, showed the doses needed to inhibit aberrant crypt foci (ACF; preneoplastic lesions) incidence and multiplicity reached plasma drug concentrations of 3.5 µg/mL (approximately 9 µmol/L; administered in the diet at 1,500 parts per million), whereas the plasma concentration of 0.3 µg/mL (0.8 µmol/L) was sufficient to inhibit adjuvant-induced arthritis. Lower doses, reaching 0.5 µg/mL (1.3 µmol/L) plasma levels did not affect ACF.49 This drug achieves IC50 (inhibitory concentration 50%) values for the inhibition of COX-1 and COX-2 in insect cells in vitro of 13 µmol/L and 0.04 µmol/L, respectively.48 This group (Reddy et al) extended these findings further by demonstrating the antitumor effect of celecoxib in the same model, with reduced tumor multiplicity, incidence and volume,44 even when administered during the promotion or progression stages.50 This finding suggests that celecoxib can retard the growth and/or development of pre-existing neoplastic lesions. Again, the lowest plasma concentration was 2.3 µg/mL (5.8 µmol/L) and the highest was 4.3 µg/mL (11.3 µmol/L). Similarly, nimesulide has shown to dose-dependently inhibit tumor multiplicity, incidence, and size in the AOM-induced colonic cancer model in mice, using doses equivalent to 21 and 39 mg/kg in the diet,51 and inhibit polyp multiplicity in Min mice at the equivalent dose of 39 mg/kg.43 Sawaoka et al52 also showed that COX-2–expressing and non–COX-expressing tumor cells were sensitive to NS-398 through the inhibition of angiogenesis at 10, 30, and 100 mg/kg orally in mice. These doses are sufficiently high to be nonselective, which suggests that mechanisms other than COX-2 inhibition may be involved.53,54 It has also been shown that celecoxib at 1,250 mg/kg of diet fed to nude mice reaches mean plasma levels of 2 to 3 µmol/L concentration, whereas 2 µmol/L in vitro did not affect cell toxicity.55 Interestingly, recent clinical studies indicted that celecoxib caused a statistically significant inhibition of FAP polyps at a dose of 800 mg/day but not at the recommended anti-inflammatory dose of 100 to 200 mg bid.31 The question of COX-independent effects occurring in association with the antitumor properties of selective COX-2 inhibitors at high doses has been reviewed elsewhere56 and includes inhibition of phosphodiesterases and IKKß (inhibitor of nuclear factor kappa ß kinase, beta subunit) kinase by sulindac sulfone; celecoxib can inhibit 3-phosphoinositide-dependent kinase (PDK1) downstream of Akt activity, whereas many NSAIDs can inhibit PPAR (peroxisome proliferator activated receptor delta) activity and matrix metalloproteinases (MMPs), inhibit and/or activate various protein kinases, and downregulate ß-catenin.
More compelling evidence for the role of COX-2 in the formation of colorectal cancers has been provided by genetic studies in mice. Genetic inhibition of COX-2 negated the development of colonic polyps in the APC716 murine model.42 Also, the maintenance of COX-2–bearing xenografts requires host COX-2 but not COX-1 expression for the production of angiogenic stimuli.57 In a model of mammary tumorigenesis, overexpression of COX-2 alone in mammary glands was sufficient to induce cellular transformation and resulted in the formation of breast carcinomas.58 Similarly, COX-2 overexpression in basal keratinocytes induces hyperplasia, skin premalignancy59 and skin malignancy.60 These studies provide strong evidence supporting the use of selective COX-2 inhibitors for the chemoprevention of some epithelial-derived cancers.
The potential role of COX-1 has not been neglected. Recent studies have shown an equal importance of this isoform in colorectal and skin tumorigenesis. A recent comprehensive study identified COX-1 expression detected in human colorectal adenomas as well as COX-2, which highlights the importance of using several analytic techniques to draw conclusions about COX expression.61 Deficiency in either the COX-1 or COX-2 gene was sufficient to inhibit the formation of intestinal polyps in Min mice.62 Selective COX-1 inhibition can reduce ACF in AOM-induced colorectal carcinogenesis.63 Similarly, the genetic deficiency of either COX-1 or COX-2 can prevent terminal differentiation of initiated keratinocytes and reduce skin tumorigenesis.64 It has been postulated that COX-1 may protect cells from the initiated DNA-damaging effects during the early tumorigenesis, whereas COX-2 contributes to tumor promotion, particularly after the loss of the APC gene in colorectal tumorigenesis.65 Cervical66 and ovarian67 cancers are other epithelial-derived malignancies where studies have recently highlighted a role for COX-1 in tumor formation through the production of angiogenic stimuli. Endothelial cell–derived COX-1 has also been shown to be important in angiogenesis,68 such that overexpression leads to malignant transformation in endothelial cells.69 The role of COX-1 in vascular biology was recently extended because selective inhibition of COX-1 prevented malignant cells from primary breast tumors metastasizing to the lung.70 These results highlight the fact that both COX-1 and COX-2 carry out identical reactions leading to the production of PGs such as PGE2.
|
REGULATION OF COX-2 EXPRESSION
|
|---|
COX has a molecular weight of 72 kDa71,72 made up of a dimeric complex of two polypeptides,73,74 each of which requires one molecule of heme for maximal catalytic activity.71 COX-1 was cloned in 1988 by three independent groups.75-77 In 1990, COX gene expression was shown to be increased, producing elevated levels of PGs from mammalian cells in vitro.78,79 In the following year, an inducible isoform of COX was identified using mitogen-stimulated chicken fibroblasts,80 phorbol-ester81 and serum-stimulated82 murine fibroblasts (termed TIS10). This COX was a 4.1 kb mRNA transcript that encoded a protein with 59% sequence homology to the ovine COX. The noninducible or constitutive isoenzyme was renamed COX-1 and the new inducible isoenzyme was designated COX-2.
The two existing isoforms of COX differ in their protein sequence and regulation of and sensitivity to NSAIDs. The major difference between the two isoenzymes is that COX-1 contains a 17-amino-acid sequence at its amino terminus that is absent in COX-2, whereas COX-2 has an additional 18-amino-acid sequence at its carboxy terminus. However, the amino sequences responsible for catalytic activity are well conserved between the two enzymes. COX-1 and COX-2 isoenzyme regulation differs at the level of both transcription and translation. The COX-1 gene is made up of larger introns compared with the COX-2 gene, accounting for lengths of 22 kb and 8 kb, respectively. The genes are transcribed into mRNA products of 2.8 kb and 4.1 kb for COX-1 and COX-2 respectively. Typical of an immediate-early gene, COX-2 contains several copies of Shaw-Kamen’s sequences in the 3'-untranslated region which confers enhanced mRNA degradation. Thus COX-2 transcripts degrade quickly and their stabilization by pro-inflammatory cytokines contributes to increased COX-2 protein levels during the inflammatory response. In contrast, corticosteroids specifically destabilize COX-2 mRNA and/or interfere with translation without affecting basal COX-1 expression (reviewed in Isakson et al83). This regulation of COX-2 at the transcriptional and translational level implies that fine control of expression is important.
Oncogenes, growth factors, cytokines, chemotherapeutics, and tumor promoters are among some of the stimuli that induce COX-2 expression. COX-2 induction has been associated with various premalignant and malignant lesions of epithelial origin in organs such as colon, lung, breast, prostate, bladder, stomach, and esophagus (reviewed in Dannenberg et al84). Although the underlying mechanisms of this elevated COX-2 expression in cancer is not known, key cis-acting elements within the start 5' region upstream of the COX-2 gene have been shown to play an important role in the regulation of COX-2. The promoter region of COX-2 consists of many transcription factor binding sites, such as nuclear factor  B (NFB), nuclear factor of interleukin-6 (NF-IL-6), cyclic adenosine monophosphate (cAMP) response element (CRE) and hypoxia-inducible factor-1 (HIF-1); most of which are known to be involved in the upregulation of COX-2 by inflammatory stimuli or tumor promoters (reviewed in Dempke et al85). It has been hypothesized that the upregulation of COX-2 prolongs the survival of abnormal cells and thereby favors the accumulation of sequential genetic changes, which increases the risk of tumorigenesis (Fig 2). 86
View larger version (58K):
[in this window]
[in a new window]
|
Fig 2. A genetic model for colorectal tumorigenesis illustrating the accumulation of mutations in oncogenes and tumor suppressor genes, which is thought to contribute to tumor progression and the propensity to metastasize. APC, adenomatous polyposis coli; COX, cyclooxygenase; MMR, mismatch repair; ras, rat sarcoma; yr, year.
|
|
However, the over-expression of COX-2 protein in colorectal cancers is likely to occur via several different mechanisms involving complex signaling pathways, since transformed epithelial cells, as well as stromal cells, have been shown to express increased levels of COX-2. For example, the COX-2 gene may be regulated by hypoxia via activation of NFB in human vascular endothelial cells.87 While COX-2 overexpression in cancerous epithelial cells may be induced through the target of normal APC (a member of the Wnt [a term derived from the genes wingless and Int-1] signaling pathway)—the ß-catenin oncoprotein (reviewed in Bright-Thomas et al88; Fig 3). The principle role of wild-type APC involves the binding and degradation of ß-catenin. ß-catenin may exist either bound to membrane or free in the cytosol. Membrane-bound ß-catenin functions through the transmembrane glycoprotein E-cadherin adhesion molecule, which is responsible for epithelial cell-cell and cell-matrix adhesion and migration. Under normal conditions, cytosolic ß-catenin is degraded by the action of wild-type APC protein complexed with two accessory proteins (axin and glycogen synthase kinase [GSK]-3ß). However, mutations in the APC gene, which encodes the vital ß-catenin-binding regions, renders the APC protein truncated and unable to control cytosolic ß-catenin. This leads to free ß-catenin, where it translocates to the nucleus and acts as a transcription factor in concert with the T-cell factor-4 (TCF-4; also known as lymphoid enhancer factor [LEF-1]) complex, a feature associated with progression along the adenoma-carcinoma sequence. The formation of transcriptionally active ß-catenin and TCF complexes binds to TCF-4 consensus sites in specific target genes, including c-myc, cyclin D1, PPAR, and COX-2. TCF binding sites (TBS) have recently been identified in the COX-2 promoter region,89 such that on modulation, COX-2 can be downregulated by wild-type APC induction and upregulated by nuclear accumulation of ß-catenin in the presence of mutant APC. Interestingly, Oshima et al42 demonstrated during tumorigenesis in Apc716 mice that induced COX-2 expression occurred either coincidentally or slightly after the loss in wild-type APC allele. This would suggest a direct role of APC loss in COX-2 overexpression. Also, PPAR expression has been recently shown to be upregulated during tumorigenesis in Min mice, and its activation can accelerate tumor growth.90
View larger version (74K):
[in this window]
[in a new window]
|
Fig 3. Cytosolic ß-catenin is degraded by wild-type adenomatous polyposis coli (APC) protein, complexed with two accessory proteins (axin and glycogen synthase kinase [GSK]-3ß). APC mutations result in truncated protein, allowing free ß-catenin to translocate to the nucleus. COX-2, cyclooxygenase-2; wnt, a term derived from the genes wingless and Int-1.
|
|
COX-2 expression is also regulated by transcriptional and post-transcriptional mechanisms, although transcriptional regulation may play a more important role in the COX-2 expression of human colon cancer cells.91,92 Ras is a member of the small GTPase family and GDP/GTP-regulated signaling molecules. It is found in a mutated form in approximately 50% of large colorectal adenomas and results in the constitutive activation of Ras followed by multiple downstream signaling pathways (Fig 4). The Ras/Rac-1/MEKK/ c-Jun NH2-terminal kinase (JNK) pathway activates COX-2 transcription via the CRE element. Ras/Raf-1/ mitogen-activated protein kinase (MAPK) –extracellular regulated kinase (ERK) kinase (MEK)/ERK operates downstream of growth factor receptor activation (eg, PDGF, serum) to induce the COX-2 promoter and stabilize COX-2 mRNA (reviewed in Dixon93). The p38 MAPK pathway, usually associated with cellular stress and proinflammatory stimuli, has also been found to transform intestinal epithelial cells, in part through COX-2 mRNA stabilization. Finally, sequential activation of phosphatidylinositol 3'-kinase (PI3-K)/ 3-phosphoinositide-dependent kinase (PDK)/Akt/ protein kinase B (PKB) cascade by various growth factors promotes cell survival through the inhibition of apoptosis. This pathway has also been found to be constitutively activated colon cancer cell lines and mediates post-transcriptional stabilization of COX-2 mRNA in intestinal epithelial cells. Recently, evidence has emerged that the Wnt- and Ras-signaling pathways may actually cooperate in the regulation of COX-2 expression in colonic tumor cells.88
View larger version (104K):
[in this window]
[in a new window]
|
Fig 4. COX-2 activation pathways. NFB, nuclear factor B; APC, adenomatous polyposis coli; GSK, glycogen synthase kinase; TCF, T-cell factor 4; ras, rat sarcoma; MAPK, mitogen-activated protein kinase; ERK, extracellular regulated kinase; MKK, MAPK kinase; MEK, MAPK/ERK kinase; JNK, c-Jun NH2-terminal kinase; PI3-K, phosphatidylinositol 3'-kinase; PDK, 3-phosphoinositide-dependent kinase; Akt, also known as protein kinase B (PKB).
|
|
|
EXPRESSION OF COX-2
|
|---|
Increased levels of COX-2 mRNA and protein are found in both premalignant and malignant tissues from epithelial and nonepithelial tumors. Gastric, hepatic, esophageal, pancreatic, head and neck, lung, breast, bladder, cervical, endometrial, skin, and colorectal cancers have all shown elevated COX-2 expression when compared with nonmalignant tissue (reviewed in Koki and Masferrer94). COX-2 overexpression in the primary tumor has been associated with a poor clinical outcome. The fact that increased COX-2 was seen in premalignant tissue again illustrates that activation of COX-2 may be an early event during tumorigenesis. Recently, increased COX-2 has also been identified in pituitary tumors95 and in harmartomatous polyps in Peutz-Jeghers syndrome.96 An important issue was recently highlighted when assessing these histologic findings; the treatment status of the patients from which tumor samples are included in these studies needs to be considered because COX expression can be induced by chemotherapeutic agents and hypoxia also. Therefore, it is not known whether expression is related to the treatment regimen, surgery-induced hypoxia, or the primary disease process.97
Concerning colorectal cancer, Eberhart et al98 were the first to identify significant elevations of COX-2 expression in 85% and 50% of human colorectal carcinomas and adenomas respectively. In normal intestinal tissue, immunolocalization studies have shown the expression of both COX-1 and COX-2 in mucosal epithelial cells, mononuclear cells, vascular endothelial cells and smooth muscle.99 However, in both human and animal models of colorectal cancers, COX-2 expression is dramatically increased in malignancies when compared with adjacent normal mucosa.93 COX-1 expression appears to remain unaltered, or even reduced.100 This has been confirmed in more recent studies of human colonic cancers.93 In vivo models of colorectal cancer have displayed a similar COX expression pattern to that seen in human samples. In the AOM-induced colonic cancer model,101,102 Min mice103 and Apc716 mice,42 increased mRNA and protein levels of COX-2 (but not COX-1) were measured in colorectal tumors compared with normal adjacent tissue. In human colorectal cancer, a positive correlation was made between NFB (cytoplasmic and nuclear) and COX-2 coexpression in neoplastic epithelial cells but not in epithelial cells from normal mucosa, eluding to the possible mechanism of how the COX-2 gene is being activated.104
However, the source and type of cells responsible for this COX-2 expression has raised many questions. The tumor stroma is now known to contribute to elevated COX-2 expression in colorectal cancers.105 Chapple et al106 found increased levels of COX-2 localized in a subpopulation of macrophages, while neoplastic cells expressed COX-2 in more advanced tumors. Positively stained macrophages for COX-2 in human colorectal adenomas in addition to tumor cells, and in Min mice has also been demonstrated.107,108 Oshima et al42 in their initial study showed localized COX-2 protein to interstitial cells and not malignant cells within small polyps of heterozygote Apc716 mice. In a subsequent study, these authors extended their findings by showing fibroblasts and endothelial cells also express high levels of COX-2 in polyps from FAP patients and Apc716 mice.109 Thus both host and tumor cells may contribute to the production of PGs within the tumor microenvironment and the subsequent development of tumor growth. In keeping with this issue, activated macrophages can transform rat intestinal epithelial cells into a more neoplastic phenotype in a paracrine fashion via a COX-2-dependent mechanism.110 This implies that the increase in COX-2 expression in intestinal interstitial cells is an early event which influences surrounding epithelial cells to transform toward malignancy. Conversely, colonic cancer cells have been shown to induce COX-2 expression in monocytes/macrophages via mucin production.111
COX-2–expressing cancer cells induce an angiogenic response in endothelial cells, while endothelial cell-derived COX-1 may also play a significant role in the angiogenic response. Tsujii et al68 showed COX-2 over-expressing cells were sensitive to NS-398 and indomethacin, while COX-null cells were only sensitive to indomethacin. This suggests that the COX-2 from the tumor cells induces the production of proangiogenic factors, whereas COX-1 from the endothelial cells can induce their own tube formation. This was later confirmed in vivo.52,112 Also, fibroblasts are instrumental as a source of angiogenic stimuli for the maintenance to tumor growth57,113 such that an intracrine effect in which COX-2 in stromal cells produce PGs, which increases VEGF within the same cell to have an effect on endothelial cells, may explain why host COX-2 is needed for sustained tumor growth.114 Finally, high microvessel density, IL-8 expression, and numbers of infiltrating macrophages were correlated with poor prognosis and survival time in patients with non–small-cell lung cancer.115 In this study, when macrophages were cocultured with lung tumor cells, tumor cell–derived IL-8 levels were inhibited by aspirin, indomethacin, celecoxib, dexamethasone, and other anti-inflammatory agents. Whether COX-2 (or COX-1)-derived PGs promote tumor growth through the transformed epithelial cells, endothelial cells, fibroblasts or macrophages, or a combination of all these cell types is not yet known. It seems likely that the localization of COX-1 or COX-2 expression changes during tumor progression.
|
ANGIOGENESIS AND METASTATIC POTENTIAL
|
|---|
It is now well established that tumors are dependent on a constant blood supply via neoangiogenesis,116 such that without the induction of angiogenesis, tumors remain only 1 to 2 mm3 in size. During this avascular period, tumors do not increase in size due to the higher numbers of cells undergoing apoptosis compared with those that are proliferating.117 For tumors to grow beyond this size, the balance between naturally occurring angiogenic stimulators and inhibitors is altered in favor of the induction of angiogenesis.116 Tumors acquire this angiogenic switch116 through genetic mutations of various oncogenes and tumor suppressor genes.118-121 The angiogenic component in tumors has considerable potential as a therapeutic target.116
PGs have been known to contribute to tumor development through their role in angiogenesis.122,123 Peterson124,125 was the first to demonstrate that diclofenac could inhibit the growth of transplantable tumors via the inhibition of angiogenesis. Since the identification of COX-2, its role in angiogenesis has been clearly demonstrated in the progression of colorectal cancer. COX-2–transfected human colon cells possess increased metastatic potential.126 COX-2 has been shown to be involved in models of angiogenesis,127-131 and selective inhibition of COX-2 can block tumor growth via an antiangiogenic mechanism.127,132,133 Selective COX-2 inhibition can also inhibit human colorectal cancer xenografts from metastasizing to the liver.134 Recently, COX-2 has been shown to be involved in the angiogenic switch in Apc716 mice by regulating angiogenesis through the induction of vascular endothelial growth factor (VEGF) in an EP2-receptor–dependent manner.135 However, host-derived PGE2 signaling through the EP3 receptor appears to regulate tumor angiogenesis through the induction of VEGF. 136 In human sporadic colorectal adenomas, interstitial cell–derived COX-2 expression (predominantly from macrophages) is associated with increased angiogenesis.137 A mechanism whereby PGs influence angiogenesis may be through the action of  vß3 integrin. The integrin vß3 is necessary for endothelial cell survival during angiogenesis since its antagonism can induce endothelial cells to undergo apoptosis, leaving pre-existing vascular beds unaffected.138 Similarly, selective COX-2 (and not COX-1) inhibitors can reverse the vß3 integrin–mediated angiogenic response.139 In a related study, COX-1 as well as COX-2 induction can modulate angiogenic response through vß3 integrin.140 Also, vß3 integrin antagonism can reduce liver metastasis and angiogenesis, and improve survival in mice injected with colon cancer cells.141
|
IMMUNE MODULATION
|
|---|
PGs can modulate immune function through a variety of mechanisms,142 however PGE2 in particular appears to have a clearly defined role in the regulation of humoral and cellular immunity.143 Tumor cells are vulnerable to attack by immune effector cells such as lymphokine-activated killer cell, cytotoxic T-lymphocytes (CTLs) and macrophages (reviewed in Young144). However, impairment in function of tumor-infiltrating lymphocytes, circulating T-cells, and macrophages has been demonstrated in cancer patients.145,146 One of the mechanisms attributed to the immune impairment in cancer patients has been increased PGE2 production. Immunosuppression occurs in tissues where PGE2 is high,147,148 and PGE2 can negatively regulate T-lymphocyte proliferation, cytokine production, and cytotoxicity.149 Specifically, PGE2 can mediate the suppression of macrophage-derived, TNF-induced colon cancer cell killing via the suppression of IL-10150 and increased IL-12.151 The tumor microenvironment is predominantly polarized toward TH2-like or immunosuppressive immune responses, and is a common feature of premalignant and malignant diseases (reviewed in O’Byrne et al152). Hence, indomethacin has been shown to increase the number of CTLs in cancer patients153 and stimulate mononuclear cells to increase their tumoricidal capacity.154 More recently, it has been demonstrated that this impaired mononuclear cell function can be restored with treatment with a selective COX-2 inhibitor.155 In fact, COX-2 is known to mediate the imbalance between IL-10 and IL-12 in favor of IL-10 production.156-158 Selective COX-2 inhibition serves to restore the tumor-induced imbalance between IL-10 and IL-12 and promotes antitumor responses in lung cancer159 and metastasis in colorectal cancer.160
|
APOPTOSIS
|
|---|
Tumor growth is significantly influenced by the relative balance between proliferation and apoptosis.117 Decreased cell proliferation alone is not sufficient to inhibit colonic tumor growth.161 Sinicrope et al162 studied the proliferative and apoptotic indices during the adenoma-carcinoma sequence of human colonic tumor samples and found a decrease in apoptosis during tumor progression. Compared with adenomas, lower rates of apoptosis with no change in proliferation were found in colonic carcinomas. This creates an imbalance between apoptosis and proliferation that favors net tumor growth. Colon carcinomas associated with a low apoptotic index have a poor prognosis.163,164 NSAID-induced apoptosis is a phenomenon that has been attributed to the antitumor action of NSAIDs in colon cancer in vivo.45,165-167 COX-2 is known to induce transformation in normal intestinal epithelial cells, resulting in increased bcl-2 expression, increased avidity to extracellular matrix components, reduced transforming growth factor ß (TGFß) receptor expression168 and a prolongation in the cell cycle G1 phase with increased cyclin D expression.169 COX-2 may prevent apoptosis not only by generating the antiapoptotic products PGE2170 and PGI2,171 but also by removing a proapoptotic substrate, arachidonic acid.172 Since PGE2 can activate MAPK activity,130 it is conceivable that Ras signal transduction pathways are involved.85 The overall decrease in apoptosis of these cells may predispose to exhibit an accumulation in sequential genetic changes that can increase the risk of tumorigenesis.
|
THE FUTURE IS COMBINATION THERAPY
|
|---|
COX-2 is now considered as a viable target for chemotherapy,84,85,173,174 especially since selective COX-2 inhibitors have improved safety profiles compared with nonselective NSAIDs and are much less toxic than most chemotherapeutic agents. It is estimated that approximately 26,000 persons die and another 260,000 are hospitalized per year as a consequence of adverse effects of NSAIDs in the United States alone.175 In 2000, approximately 60% of the $4.8 billion total cost for NSAID prescriptions were for selective COX-2 inhibitors (reviewed in Laine176). If long-term use of selective COX-2 inhibitors can prove to be less toxic than traditional NSAIDs, these drugs can have a 50% reduction in adverse effects. Unfortunately, the long-term safety issue with COX-2 inhibitors was recently thrown into question. Clinical trials evaluating the use of selective COX-2 inhibitors for the prevention of polyp recurrence have stopped due to increased cardiovascular and thrombotic adverse effects observed with their long-term use.176a These results dramatically alter the utility of using selective COX-2 inhibitors for chemoprevention in a low-risk population. However, additional short-term studies are underway to evaluate the utility of these inhibitors in adjuvant treatment regimens and in patients with a very high risk of developing colorectal cancer (Table 3).
For use in colorectal cancer, the role of COX-2 inhibitors in prevention still awaits assessment because their recent release means epidemiologic data will be available only in a few years (Table 3). On the basis of the cumulative experimental and clinical evidence, selective COX-2 inhibitors may be considered as cotherapeutic agents (reviewed in Blanke173). In most preclinical studies, selective COX-2 inhibitors reduce the growth rate of established tumors rather than causing tumor regression.84 Therefore, selective COX-2 inhibition has recently been investigated in various models of colorectal cancer in combination with inducible nitric oxide inhibitors,49 MMP inhibitors,177 ornithine decarboxylase inhibitors (difluoromethylornithine),178 epidermal growth factor receptor kinase inhibitors,179-181 radiotherapy,182 and chemotherapy.160 A small phase II study183 of 10 patients with metastatic colorectal cancer using rofecoxib and chemotherapy did not demonstrate increased efficacy; however, ongoing trials may further clarify the role of these agents for the treatment of colorectal cancer (Table 3). The majority of colorectal cancers over-express COX-2, and this increased expression is thought to inhibit apoptosis, induce angiogenesis, subvert the immune system, and promote tumor invasion. An understanding of the mechanism(s) whereby COX-2 mediates these phenomena awaits further studies.
|
Authors' Disclosures of Potential Conflicts of Interest
|
|---|
The following authors or their immediate family members have indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. Consultant/Advisory Role: Raymond N. DuBois, Pharmacia, Novartis. Honoraria: Raymond N. DuBois, Pfizer. For a detailed description of these categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration form and the Disclosures of Potential Conflicts of Interest section of Information for Contributors found in the front of every issue.
|
NOTES
|
|---|
Authors' disclosures of potential conflicts of interest are found at the end of this article.
|
REFERENCES
|
|---|
1. Parkin DM: Global cancer statistics in the year 2000. Lancet Oncol 2:533-543, 2001[CrossRef][Medline]
2. National Institute for Cancer (NCI), Surveillance and End Results (SEER) Cancer Statistics Review: 1975-2001. http://seer.cancer.gov/csr/1975_2001 |
TOP
ABSTRACT
CANCER PREVALENCE