Content uploaded by Khosrow Kashfi
Author content
All content in this area was uploaded by Khosrow Kashfi
Content may be subject to copyright.
724 Biochemical Society Transactions (2005) Volume 33, part 4
Is COX-2 a ‘collateral’ target in cancer prevention?
K. Kashfi* and B. Rigas†1
*Department of Physiology and Pharmacology, City University of New York Medical School, NY 10031, U.S.A., and †Division of Cancer Prevention,
Department of Medicine, SUNY at Stony Brook, Stony Brook, NY 11794-5200, U.S.A.
Abstract
NSAIDs (non-steroidal anti-inflammatory drugs) prevent colon and other cancers. The fact that NSAIDs
inhibit the eicosanoid pathway prompted mechanistic drug-developmental work focusing on COX (cyclo-
oxygenase) and its products. The increased prostaglandin E2levels and the overexpression of COX-2 in colon
and many other cancers provided the rationale for clinical trials with COX-2 inhibitors for cancer prevention
or treatment. However, one COX-2 inhibitor has been withdrawn from the market because of cardiovascular
side effects, and there are concerns about a class effect. Evidence suggests that COX-2 may not be the
only, or the ideal, target for cancer prevention; for example, COX-2 is not expressed in human aberrant
crypt foci, the earliest recognizable pre-malignant lesion in the colon; COX-2 is expressed in less than half
of the adenomas; in vitro data show that NSAIDs do not require the presence of COX-2 to prevent cancer;
in familial adenomatous polyposis, the COX-2 inhibitor, celecoxib, had a modest effect, which was weaker
than that of a traditional NSAID; and COX-2-specific inhibitors have several COX-2-independent activities,
which may account for part of their cancer-preventive properties. The multiple COX-2-indpendent targets,
and the limitations of COX-2 inhibitors, suggest the need to explore targets other than COX-2.
Introduction
Cancer prevention, at present a better option than cancer
treatment, is entering an era when it appears to be a realistic
possibility. The seminal epidemiological observation that
NSAIDs (non-steroidal anti-inflammatory drugs) prevent
colon, and possibly other, cancers has led to the unambiguous
demonstration that aspirin does prevent colon cancer. Two
randomized interventional studies using polyp recurrence as
a general end point demonstrated the chemopreventive effect
of aspirin [1,2]. The relative risks following administration of
aspirin ranged between 0.59 and 0.96, depending on the speci-
fic end point and aspirin dose. Although specific aspects of
this effect appear unclear at this point, these studies, neverthe-
less, constitute proof-of-principle for pharmacological cancer
prevention. However, NSAIDs are ill-suited for widespread
application as chemopreventive agents. Their two prohibi-
tive limitations concern their safety (among patients using
NSAIDs, up to 4% per year suffer serious gastrointestinal
complications) and efficacy (NSAIDs can prevent at best 50%
of colon cancer) (reviewed in [3]). To these, one should add
the need to have more stringent criteria for safety and efficacy
for chemoprevention, as opposed to chemotherapy, when one
deals with a life-threatening cancer.
Considerations of safety and efficacy have prompted the
search for a ‘better NSAID’, with coxibs, selective inhibitors
of COX-2 (cyclo-oxygenase-2), being the most notable out-
come. Coxibs have been developed based on the notion that
inhibition of COX-2, the induced isoform of COX, will
Key words: apoptosis, carbonic anhydrase, cell cycle, coxib, cytochrome c, NAG-1 (non-steroidal
anti-inflammatory drug-activated gene).
Abbreviations used: COX, cyclo-oxygenase; FAP, familial adenomatous polyposis; LOX,
lipoxygenase; NSAID, non-steroidal anti-inflammatory drug; PG, prostaglandin; PGI2,prostacyclin;
TxA2, thromboxane A2.
1To whom correspondence should be addressed (email basil.rigas@sunysb.edu).
diminish the pro-inflammatory activities of COX, whereas
sparing COX-1, the constitutive isoform of COX, will
diminish the gastrointestinal, and perhaps other, side effects
of NSAIDs [4]. Recent concerns on the safety of coxibs, es-
pecially after their long-term use, justify a re-examination of
the fundamental tenet underlying their use in cancer, namely
that COX-2 is central to the pathogenesis of several cancers,
and that its inhibition would prevent them and regress those
already established.
The rationale for COX-2 as a molecular
target for cancer prevention
The initial response of many investigators, including our-
selves, to the epidemiological data showing that NSAIDs
are associated with a decreased incidence of cancer, was that
NSAIDs act by inhibiting COX, an important enzyme in
the eicosanoid cascade that ultimately leads to PGs (prosta-
glandins) and related compounds [4]. Thus we demonstrated
that, in human colon cancers, PGE2levels were strikingly
increased compared with uninvolved mucosa [5]. Subsequen-
tly, Tunni and DuBois [6] demonstrated overexpression of
COX-2 in 45% of colon adenomas and 85% of colon
carcinomas. COX-2 is overexpressed to varying degrees in
several more human cancers, including gastric, breast, lung,
oesophagealandhepatocellularcarcinomas.Additionalmech-
anistic studies showed that PGE2increases colon cancer
cell proliferation [7] and suppresses apoptosis [8]. The role
of eicosanoids in carcinogenesis has been expanded further
by studies demonstrating that, in certain cases, LOX (lipoxy-
genase) products may also play a role in carcinogenesis [9]
(Figure 1). The conclusion that inhibition of COX-2 would
arrest carcinogenesis has been supported by a constellation
of cell culture, animal and human studies, culminating in
C
2005 Biochemical Society
The Molecular Biology of Colorectal Cancer 725
Figure 1 Overview of the arachidonic acid metabolic pathway
Arachidonic acid, derived from diet, is released from membrane phospholipids through the action of PLA2(phospholipase
A2) or is synthesized from linolenic acid. The COX pathway produces various PGs and TxA2, and the LOX pathways produce
LTs (leukotrienes) and HETEs (hydroxyeicosatetraenoic acids). Some 5-LOX and 12-LOX metabolites, such as LTB4and
12-HETE, appear to enhance tumorigenesis, whereas 13-HODE [13(S)-hydroxyoctadecadienoic acid], a product of 15-LOX-1,
is anticarcinogenic.
celecoxib receiving FDA (U.S. Food and Drug Adminstra-
tion) approval for cancer prevention in patients with FAP
(familial adenomatous polyposis).
Studies using genetically modified animals have indicated
that COX-2 may be required for tumorigenesis. Deletion of
COX-2, and, importantly, of COX-1 as well, decreased signi-
ficantly the number of intestinal tumours in Apc716 mice [10].
Overexpression of the human COX-2 gene in the mammary
glands of female mice led to focal-mammary-gland hyper-
plasia, dysplasia, and transformation into metastatic tumours
[11]. Overexpression of COX-2 in basal epidermal cells of
transgenic mice was either insufficient for tumour induc-
tion (although it sensitized the tissue to carcinogens) [12] or,
rather surprisingly, protected them from developing tumours
that were induced by an initiation/promotion protocol [13].
Alternatively, numerous animal studies have shown that
coxibs prevent tumours arising from a variety of tissues [6].
The limitations of current coxibs
The APPROVe (Adenomatous Polyp Prevention on Vioxx)
study was designed to evaluate the efficacy of rofecoxib in
preventing colon cancer. During the trial, which involved
2600 subjects with a history of colorectal polyps, 3.5% of
rofecoxib recipients and 1.9% of placebo recipients suffered
myocardial infarctions or strokes. This led to the termination
of this and all related trials and the permanent withdrawal of
rofecoxib. It is still controversial whether other coxibs share
this side effect, but concerns for a ‘class (side) effect’ have
been voiced [14].
To explain this side effect, it was suggested that inhibition
by coxibs of COX-2, the principal enzyme involved in the
production of PGI2(prostacyclin), tips the balance towards
platelet aggregation and vasoconstriction [14]. As discussed
below, this may constitute a limiting side effect of coxibs for
their required long-term application in cancer prevention.
Is COX-2 overexpression central to
carcinogenesis?
Several observations suggest that it may be worth reassessing
the notion that COX-2 is central to the pathogenesis of sev-
eral cancers, and therefore its inhibition should be the prime
target of cancer chemoprevention. Below, we outline data that
are at variance with this notion.
The pattern of COX-2 expression
Taking colon carcinogenesis as an example, it is apparent that
the pattern of COX-2 expression is not entirely consistent
with the idea that COX-2 is central to carcinogenesis. COX-2
expression is absent in aberrant crypt foci, the earliest recog-
nizable pre-malignant lesion in the colon [15], and com-
mences only at the adenoma stage (45% of them), increasing
in frequency (85%) in carcinomas [6]. An unconventional
look at the data may suggest that COX-2 expression is the re-
sult of, and not a dominant contributor to, carcinogenesis.
In support of this idea is the finding that targeted overex-
pression of human microsomal PGE synthase-1 (mPGES) in
the alveolar type II cells of transgenic mice, accompanied by
highly elevated PGE2production (12.2-fold over control),
failed to induce lung tumours [16].
NSAIDs and COX-2 in cancer prevention
NSAIDs do not require the presence of COX-2 to prevent
cancer [17]. This was based on our finding that in vitro
C
2005 Biochemical Society
726 Biochemical Society Transactions (2005) Volume 33, part 4
Figure 2 Selected COX-2-independent effects of coxibs relevant
to cancer
Coxibs modulate a large array of molecular targets, some of which are
shown here. NAG-1, NSAID-activated gene.
NSAIDs display effects compatible with cancer prevention,
such as inhibition of cell proliferation, induction of apoptosis,
inhibition of angiogenesis and many others, in the absence
of COX-1 or -2. Our initial observation is now firmly
established by the work of many investigators.
Coxibs have only limited clinical efficacy in
cancer prevention
In FAP patients, celecoxib reduced the mean number of colo-
rectal polyps by 28% and the polyp burden by 30.7% (the
respective placebo reductions were 4.5% and 4.9%) [18].
Rofecoxib had a statistically significant, but marginal, effect
on the number of polyps in FAP patients (6.8% reduction
from baseline values) [19]. In contrast, the NSAID sulindac
had a more pronounced effect on colorectal polyps in FAP
patients, being around twice as effective as celecoxib and far
more effective than rofecoxib [20].
Coxibs have several COX-2-independent
activities
This may account, at least in part, for coxibs’ cancer-pre-
ventive properties (Figure 2). For example, celecoxib inhibits
the growth of various cancer cell lines [21], including haem-
atopoietic cell lines that are COX-2-deficient. Interestingly,
celecoxib also inhibited the growth of COX-2-deficient colon
cancer xenografts in nude mice [22]. Moreover, a selective
COX-2 inhibitor reduced tumour growth and angiogenesis
in COX-2-positive pancreatic cancer, but in COX-2-negative
pancreatic cancer, it increased angiogenesis and tumour
growth [23]. Thus the chemopreventive effect of COX-2-
specific inhibitors may be due to their effect on these targets
and not on COX-2.
The expression of COX-2 is not restricted to
tumour cells
While COX-2 is undetectable in most tissues in the absence of
stimulation, it is induced in cells such as monocytes, macro-
phages, neutrophils and endothelial cells [4]. A study in
healthy humans suggests that COX-2 is a major source of sys-
temic PGI2biosynthesis [24]. Importantly, atheromatous le-
sions contain both COX-1 and COX-2, co-localizing mainly
with macrophages of the shoulder region and lipid core
periphery, whereas smooth-muscle cells show lower levels
[25]. Inhibition of vascular COX-2 may shift the delicate
balance between TxA2(thromboxane A2)andPGI
2,which
have opposite effects on platelets and vascular tone. Shifting
this balance in the wrong direction (reduction of PGI2) could
have catastrophic effects. Indeed, this may account for the
cardiovascular side effects of coxibs. If such a mechanism is
proven, it may be of great importance to chemoprevention,
in which a chemopreventive agent against cancer will be ad-
ministered on a long-term basis to older subjects, i.e. those
likely to have atheromatous lesions.
Inhibition of COX may shift its substrate fatty
acid to a non-COX pathway and generate a
pro-carcinogenic end product
For example, inhibition of COX-2 could shift arachidonic
acid to the LOX pathway, some of whose products have pro-
tumorigenic activities. Although this possibility has not been
systematically explored, a recent study suggests that it may
not be that unlikely; in humans, under physiological condi-
tions, oral celecoxib increased leukotriene B4production in
the lung micro-environment [26].
Time to search for targets beyond COX-2
It is apparent that the central concept of a dominant role of
COX-2 in cancer prevention may have significant limitations
that make necessary its re-examination. If the cardiovascular
toxicity of coxibs is in fact due to their COX-2 effects,
then it may be difficult to envision their practical long-term
administration to individuals who, in the context of arthero-
sclerosis, may have endothelial COX-2 overexpression.
Thus alternative approaches should focus on targets beyond
COX-2.
The two main reasons justifying the search for targets other
than COX-2 are the following: first, NSAIDs prevent colon
and other cancers and do this sub-optimally, probably by
modulating several molecular targets in addition to COX-2.
NSAIDs are not reasonable candidates for chemoprevention,
owing to their safety and efficacy limitations. Secondly, coxibs
have limited clinical efficacy; it is likely that they achieve their
clinical effect by modulating targets other than COX-2, and
they may have limiting side effects.
At this time, strategies incorporating these considerations
may lead to the next, and, one hopes, final, stage in our ef-
forts to prevent cancer. Such an approach appears to be both
rational and promising.
This work was supported by the NIH (National Institutes of Health)
Grants CA92423 and CA34527, and PSC-CUNY (Professional Staff
Congress-City University of New York) Grant 65201-00 34.
References
1 Baron, J.A., Cole, B.F., Sandler, R.S., Haile, R.W., Ahnen, D., Bresalier, R.,
McKeown-Eyssen, G., Summers, R.W., Rothstein, R., Burke, C.A. et al.
(2003) N. Engl. J. Med. 348, 891–899
C
2005 Biochemical Society
The Molecular Biology of Colorectal Cancer 727
2 Sandler, R.S., Halabi, S., Baron, J.A., Budinger, S., Paskett, E.,
Keresztes, R., Petrelli, N., Pipas, J.M., Karp, D.D., Loprinzi, C.L. et al.
(2003) N. Engl. J. Med. 348, 883–890
3 Rayyan, Y., Williams, J. and Rigas, B. (2002) Cancer Invest. 20,
1002–1011
4 Simmons, D.L., Botting, R.M. and Hla, T. (2004) Pharmacol. Rev. 56,
387–437
5 Rigas, B., Goldman, I.S. and Levine, L. (1993) J. Lab. Clin. Med. 122,
518–523
6 Turini, M.E. and DuBois, R.N. (2002) Annu. Rev. Med. 53, 35–57
7 Qiao, L., Kozoni, V., Tsioulias, G.J., Koutsos, M.I., Hanif, R., Shiff, S.J. and
Rigas, B. (1995) Biochim. Biophys. Acta 1258, 215–223
8 Sheng, H.M., Shao, J.Y., Morrow, J.D., Beauchamp, R.D. and Dubois, R.N.
(1998) Cancer Res. 58, 362–366
9 Shureiqi, I. and Lippman, S.M. (2001) Cancer Res. 61, 6307–6312
10 Chulada, P.C., Thompson, M.B., Mahler, J.F., Doyle, C.M., Gaul, B.W.,
Lee, C., Tiano, H.F., Morham, S.G., Smithies, O. and Langenbach, R.
(2000) Cancer Res. 60, 4705–4708
11 Liu, C.H., Chang, S.H., Narko, K., Trifan, O.C., Wu, M.T., Smith, E.,
Haudenschild, C., Lane, T.F. and Hla, T. (2001) J. Biol. Chem. 276,
18563–18569
12 Muller-Decker, K., Neufang, G., Berger, I., Neumann, M., Marks, F. and
Furstenberger, G. (2002) Proc. Natl. Acad. Sci. U.S.A 99, 12483–12488
13 Bol, D.K., Rowley, R.B., Ho, C.P., Pilz, B., Dell, J., Swerdel, M., Kiguchi, K.,
Muga, S., Klein, R. and Fischer, S.M. (2002) Cancer Res. 62, 2516–2521
14 Fitzgerald, G.A. (2004) N. Engl. J. Med. 351, 1709–1711
15 Nobuoka, A., Takayama, T., Miyanishi, K., Sato, T., Takanashi, K.,
Hayashi, T., Kukitsu, T., Sato, Y., Takahashi, M., Okamoto, T. et al. (2004)
Gastroenterology 127, 428–443
16 Blaine, S.A., Meyer, A.M., Hurteau, G., Wick, M., Hankin, J.A., Murphy,
R.C., Subbaramaiah, K., Dannenberg, A.J., Geraci, M.W. and Nemenoff,
R.A. (2005) Carcinogenesis 26, 209–217
17 Hanif, R., Pittas, A., Feng, Y., Koutsos, M.I., Qiao, L., Staiano-Coico, L.,
Shiff, S.I. and Rigas, B. (1996) Biochem. Pharmacol. 52, 237–245
18 Steinbach, G., Lynch, P.M., Phillips, R.K., Wallace, M.H., Hawk, E., Gordon,
G.B., Wakabayashi, N., Saunders, B., Shen, Y., Fujimura, T. et al. (2000)
N. Engl. J. Med. 342, 1946–1952
19 Higuchi, T., Iwama, T., Yoshinaga, K., Toyooka, M., Taketo, M.M. and
Sugihara, K. (2003) Clin. Cancer Res. 9, 4756–4760
20 Giardiello, F.M., Hamilton, S.R., Krush, A.J., Piantadosi, S., Hylind, L.M.,
Celano, P., Booker, S.V., Robinson, C.R. and Offerhaus, G.J. (1993) N. Engl.
J. Med. 328, 1313–1316
21 Maier, T.J., Schilling, K., Schmidt, R., Geisslinger, G. and Grosch, S. (2004)
Biochem. Pharmacol. 67, 1469–1478
22 Grosch, S., Tegeder, I., Niederberger, E., Brautigam, L. and Geisslinger, G.
(2001) FASEB J. 15, 2742–2744
23 Eibl, G., Takata, Y., Boros, L.G., Liu, J., Okada, Y., Reber, H.A. and Hines,
O.J. (2005) Cancer Res. 65, 982–990
24 McAdam, B.F., Catella-Lawson, F., Mardini, I.A., Kapoor, S., Lawson,
J.A. and FitzGerald, G.A. (1999) Proc. Natl. Acad. Sci. U.S.A. 96,
272–277
25 Schonbeck, U., Sukhova, G.K., Graber, P., Coulter, S. and Libby, P. (1999)
Am. J. Pathol. 155, 1281–1291
26 Mao, J.T., Tsu, I.H., Dubinett, S.M., Adams, B., Sarafian, T., Baratelli, F.,
Roth, M.D. and Serio, K.J. (2004) Clin. Cancer Res. 10, 6872–6878
Received 22 March 2005
C
2005 Biochemical Society