Distal bowel selectivity in the chemoprevention of experimental colon carcinogenesis by the non-steroidal anti-inflammatory drug nabumetone
Article first published online: 15 MAR 2001
Copyright © 2001 Wiley-Liss, Inc.
International Journal of Cancer
Volume 92, Issue 4, pages 609–615, 15 May 2001
How to Cite
Roy, H. K., Karolski, W. J. and Ratashak, A. (2001), Distal bowel selectivity in the chemoprevention of experimental colon carcinogenesis by the non-steroidal anti-inflammatory drug nabumetone. Int. J. Cancer, 92: 609–615. doi: 10.1002/ijc.1226
- Issue published online: 9 APR 2001
- Article first published online: 15 MAR 2001
- Manuscript Accepted: 11 DEC 2000
- Manuscript Received: 12 OCT 2000
- American Society of Clinical Oncology
- non-steroidal anti-inflammatory drugs;
- colon cancer;
Use of non-steroidal anti-inflammatory drugs (NSAIDs) for chemoprevention of colon cancer has been hindered by their potential gastro-intestinal toxicity. Nabumetone, which is approximately 10 to 36 times safer than conventional NSAIDs, was evaluated in 2 models of experimental colon carcinogenesis. In azoxymethane (AOM)-treated Fisher 344 rats, nabumetone caused dose-dependent inhibition of aberrant crypt foci (ACF), with 750 and 1,500 ppm resulting in 15% and 37% reductions, respectively (p < 0.05). Moreover, complex ACF were reduced by 48% in the latter group. MIN mice studies confirmed the chemopreventive efficacy of nabumetone, with 900 ppm suppressing approximately half of the intestinal tumors. Interestingly, inhibition of intermediate biomarkers in both models was markedly greater in the distal than the proximal bowel. To mechanistically evaluate this regional selectivity, we assessed cyclo-oxygenase-2 (COX-2) expression in the uninvolved mucosa and demonstrated a 3- to 4-fold excess in the distal relative to the proximal bowel in both MIN mice and AOM-treated rats. We then investigated another putative NSAID target, peroxisome proliferator–activated receptor-δ (PPAR-δ) and demonstrated up-regulation during AOM-induced colonic tumorigenesis. Furthermore, in pre-neoplastic mucosa, there was a 3-fold excess of PPAR-δ in the distal colon. We demonstrate that nabumetone is an effective protective agent in both experimental models of colon carcinogenesis. The striking distal predilection of nabumetone may be, at least partially, explained by distal bowel over-expression of COX-2 and PPAR-δ. © 2001 Wiley-Liss, Inc.
Despite significant medical advances in diagnosis and therapy, colorectal malignancies remain the second leading cause of cancer mortality in the United States,1 underscoring the need for effective chemopreventive strategies. A variety of agents have been purported to protect against colon cancer; of these, the non-steroidal anti-inflammatory drugs (NSAIDs) have demonstrated the greatest promise.2 The epidemiological and experimental evidence indicates that NSAIDs can reduce the risk of colorectal cancer by up to 50%;3 however, utilization of NSAIDs for colon-cancer prevention has been hindered by their propensity to cause gastro-intestinal injury.4 To minimize these untoward effects, novel NSAIDs have been developed with selectivity for cyclo-oxygenase-2 (COX-2), the COX isoform not expressed in normal gastro-intestinal epithelium.5 COX-2, but not COX-1, is markedly up-regulated early during colon carcinogenesis.3 Therefore, selective targeting of COX-2 may afford protection against colorectal cancer without the gastro-intestinal toxicity associated with depletion of COX-1-dependent mucosal cytoprotective prostaglandins.
Nabumetone, an NSAID in widespread clinical use, has a 7-fold higher affinity for COX-2 than COX-1.6 Moreover, a meta-analysis indicates that nabumetone is 10 to 36 times safer than conventional NSAIDs.7 Although there are few data on nabumetone's chemoprotective efficacy, we have previously reported that it caused a dose-dependent suppression of tumors in the MIN (multiple intestinal neoplasia) mouse.8 The strength of this experimental model is that the initiating genetic events of human colorectal carcinogenesis are reproduced through a germline mutation in the adenomatous polyposis coli (APC) tumor-suppressor gene.9APC truncation leads to over-expression of its downstream effector, β-catenin.10 β-Catenin is a regulator of COX-2 as well as peroxisome proliferator–activated receptor-δ (PPAR-δ), another important target for NSAIDs in the chemoprevention of colon cancer.11 The MIN mouse, however, is somewhat limited as a model of colonic carcinogenesis in that most of the tumors are small bowel adenomas rather than colorectal cancers. Colon cancers can be induced by treatment of rats with the carcinogen azoxymethane (AOM) or its precursor 1,2-dimethylhydrazine.12 In this model, the APC gene generally remains wild-type, but the high frequency of activating β-catenin mutations produces a functionally equivalent result.13 COX-2 is generally up-regulated in AOM-induced tumors;14 however, no reports exists on PPAR-δ expression in this model. While tumorigenesis in the AOM model generally requires approximately 35 weeks, aberrant crypt foci (ACF) are detectable within 2 weeks after carcinogen treatment.15 The ACF assay is well validated and frequently used as an intermediate biomarker in experimental colon cancer-prevention studies. Indeed, the data support the biological importance of ACF in human colon carcinogenesis.16
Given the limited data on the chemoprotective efficacy of nabumetone, our objective was to evaluate the ability of this NSAID to suppress AOM-induced ACF in rats as well as intestinal tumors in MIN mice. In light of reports that emphasize the biological differences between proximal and distal colon cancers, we were particularly interested in assessing whether nabumetone would have regional selectivity in chemoprotection. Finally, to understand the mechanisms for the regional variability in nabumetone's efficacy, we evaluated both proximal and distal expression of the putative NSAID targets COX-2 and PPAR-δ.
MATERIAL AND METHODS
All animal studies were done in accordance with the Institutional Animal Care and Use Committee of the University of Nebraska Medical Center (Omaha, NE).
Forty-five male, 150 to 200 g Fisher 344 rats (National Cancer Institute, Bethesda, MD) were randomized to either a control AIN 93a diet (Teklad, Madison, WI) or nabumetone (99.6% pure) supplementation at 750 or 1,500 ppm (kindly provided by Smith, Kline & Beecham, Collegeville, PA). After 2 weeks, animals were injected twice with AOM (99% pure; Sigma, St. Louis, MO), 15 mg/kg s.c. separated by 1 week. Fifteen weeks after the second injection, rats were killed and their colons removed, rinsed with saline and bisected. Colon segments were fixed in 10% formalin, stained with 0.2% methylene blue and examined under 40× magnification. ACF were scored for location and size by an observer blinded to treatment group.
Twenty-four 5- to 6-week-old MIN mice (Jackson Laboratory, Bar Harbor, ME) were randomized to an AIN 93a diet alone or nabumetone supplementation at 600, 900 or 1,200 ppm. After 10 weeks, animals were killed, their intestines were removed and rinsed and the small bowel was bisected. Tumors were scored under magnification by an observer blinded to treatment group. For analysis of regional tumor distribution, data were combined with an earlier MIN mouse experiment (24 mice randomized to AIN 93a alone or nabumetone supplementation at 300, 600 or 900 ppm), for which total tumor numbers were previously reported.8
COX-2 and PPAR-δ expression in non-neoplastic mucosa
Rat colonocytes were prepared from F344 rats 15 weeks after the second injection of AOM (15 mg/kg s.c.) by a modification of the Weiser technique, which utilizes divalent cation chelation followed by gentle mechanical dissociation.17 MIN mouse intestinal mucosa was lightly scraped and frozen at –80°C. Cellular lysates were obtained by sonicating in Laemmli buffer, and protein concentrations were determined by the Amido black method.18 Forty micrograms of protein were separated on a 10% SDS-polyacrylamide gel electrophoretically, transferred to polyvinylfluoride membranes and stained with India ink to confirm equal loading of proteins. Membranes were blocked with 5% non-fat milk and probed with a polyclonal antibody to COX-2 or PPAR-δ or a monoclonal antibody (MAb) to β-tubulin (all from Santa Cruz Biotechnology, Santa Cruz, CA). Xerograms were developed by enhanced chemiluminescence (Amersham-Pharmacia, Piscataway, NJ) and quantitated with a scanning laser densitometer.
PPAR-δ immunohistochemistry in AOM-treated rats
Archived paraffin-embedded sections from F344 rats killed 35 weeks after the second injection of AOM (15 mg/kg s.c.) were prepared for immunohistochemistry using standard methods. After blocking with serum and quenching endogenous peroxidases, sections were incubated with a polyclonal antibody to PPAR-δ for 1 hr at 37°C. Slides were developed with the Vectastain ABC kit (Vector, Burlingame, CA).
Tolerance of nabumetone
Nabumetone administration was well tolerated, with no macroscopic or histological evidence of gastro-intestinal, hepatic or renal injury. Body weight was assessed as another surrogate gauge of toxicity. F344 rats continued to gain weight throughout the experimental period, and AOM treatment did not significantly affect body weight (data not shown). Nabumetone at either 750 or 1,500 ppm did not cause any significant change in weekly body weight (Fig. 1a). The body weight of MIN mice increased for the first 6 weeks and then began to decline as the tumor burden increased. Nabumetone supplementation, even at 1,200 ppm, did not cause a significant decrease in body weight when compared to mice on the control diet (Fig. 1b). Indeed, there was a slight, but non-significant, trend toward an increase in weight in mice treated with nabumetone, especially during the early phases of the experiment.
Effect of nabumetone on intermediate biomarkers of colon cancer
Dietary nabumetone supplementation resulted in a dose-dependent reduction in AOM-induced ACF (Fig. 2a). In animals on the control diet, there were 293 ± 9 ACF including 8 ± 3 complex ACF/colon (defined as ≥10 crypts/foci). Nabumetone at 750 ppm caused a 15% decrease in total ACF, while at 1,500 ppm it caused a 37% reduction. With regard to complex ACF, there was only a significant reduction with 1,500 ppm treatment (48%).
The total number of tumors in MIN mice on the AIN 93a diet alone was 44 ± 3.5/animal, with 90% located in the small intestine (Fig. 2b). All tumors sampled for histological analysis were adenomas. Treatment with 600 ppm nabumetone resulted in a 13% reduction in tumors; however, this failed to reach statistical significance (p < = 0.16). The maximal tumor-inhibitory dose of nabumetone was 900 ppm (43 ± 9%, p = 0.003), with the 1,200 ppm group demonstrating less tumor inhibition (38 ± 4%, p = 0.001 vs. control).
Regional chemopreventive efficacy of nabumetone
In F344 rats on the control diet, 56% of ACF were located in the distal colon. While nabumetone decreased ACF throughout the colon, suppression was accentuated distally (Fig. 3a). This reduction in ACF number in the distal vs. proximal colon with nabumetone was seen in both the 750 (15 vs. 30, p < 0.03) and the 1,500 (30 vs. 60 p < 0.05) ppm groups. This resulted in an increase in the proximal to distal ratio, from 0.78 in the AIN 93a alone group to 1.05 in the 1,500 ppm group (p < 0.05).
MIN mice on the control diet also demonstrated a distal predominance of tumors; indeed, only 21% of tumors were located in the proximal intestine. The reduction in tumors was more marked in the distal small intestine (Fig. 3b). This differential decrease in tumor number was seen at 600, 900 and 1,200 ppm. When expressed as a percentage of the AIN 93a alone group, tumor inhibition in all 4 treatment groups was still greater distally (80 ± 14%, 67 ± 11%, 50 ± 6% and 60 ± 6%, respectively) when compared to proximally (138 ± 23%, 134 ± 20%, 77 ± 16% and 70 ± 16%, respectively). However, statistically significant proximal vs. distal tumor suppression was achieved only with the 2 lower doses of nabumetone.
Regional expression of COX-2 in MIN mice and AOM-treated rats
COX-2 was expressed in both colonocytes of AOM-treated rats and the uninvolved intestinal epithelial scrapings of MIN mice. In pre-malignant rat colon, COX-2 was detected faintly in the proximal mucosa but more robustly in the distal mucosa [Fig. 4a(i)]. A proximal to distal COX-2 gradient was evident regardless of dietary group (308 ± 108% for AIN 93a alone vs. 218 ± 55% for nabumetone 1,500 ppm) (Fig. 4b), consistent with our previous observation that COX-2 levels were not altered by nabumetone treatment.8 The MIN mouse small bowel also had markedly more COX-2 expression distally (368 ± 107%, p < 0.03) (Fig. 4b).
Regional expression of PPAR-δ in AOM-treated rats
We assessed PPAR-δ expression in AOM-treated rats as another putative NSAID target in the chemoprevention of colon carcinogenesis. Since no previous reports have assessed PPAR-δ in AOM-induced tumors, we assessed 10 tumors and detected strong immunoreactivity in the adenomas and adenocarcinomas [Fig. 5a(i,ii)]. Moreover, PPAR-δ was clearly demonstrated in the uninvolved mucosa, and the staining was more pronounced near the luminal surface in comparison to the base of the crypt [Fig. 5a(iii,iv)]. We then performed Western blotting for PPAR-δ in proximal and distal colonocytes from rats (Fig. 5b). Immunoblot analysis demonstrated marked induction of PPAR-δ in the distal relative to the proximal colon (151 ± 1.7% of proximal colon, p < 0.02) (Fig. 5c).
Our study confirms that nabumetone has an excellent safety profile, with no evidence of toxicity or body weight alterations even at the highest doses utilized (Fig. 1). Our data are consistent with nabumetone's excellent clinical safety record.7 Its improved safety profile is believed to be multifactorial but at least partly due to its relative COX-2 selectivity. Newer NSAIDs with even greater COX-2 selectivity (celecoxib, rofecoxib) also appear to be efficacious in protecting against experimental colon cancer,19 but their lack of anti-platelet activity (mediated through COX-1) implies that they would not have a cardiovascular protective effect. This issue is crucial from a public health perspective since over 80% of the survival benefit from NSAIDs is through protection from myocardial infarctions.20 Prostacyclin, which inhibits platelet aggregation, is derived from endothelial COX-2, whereas the pro-aggregatory thromboxane is derived from platelet COX-1.21 Moreover, COX-2 is important in the cardiac protective effects of ischemic pre-conditioning.22 Therefore, theoretically, highly selective COX-2 inhibitors may actually increase the incidence of acute coronary events. In support of this hypothesis, there were significantly more myocardial infarctions in patients treated long-term with the highly selective COX-2 inhibitor rofecoxib compared to the conventional NSAID naprosyn.23 Other potential benefits of COX-1 inhibition include enhancement of protection against colon carcinogenesis.24 Thus, nabumetone, with its safety and anti-platelet activity,25 appears to be an excellent NSAID for use in cancer prevention.
Nabumetone inhibited intermediate biomarkers in both of these well-validated models of colon cancer. The data support the importance of ACF in colon carcinogenesis and its potential utility as a clinical screening tool. Endoscopic detection of ACF correlated well with the presence of colonic adenomas and carcinomas. Moreover, in humans, ACF were very sensitive to inhibition by NSAIDs such as sulindac.16 Nabumetone's dose-dependent suppression of ACF was comparable to that observed with other NSAIDs. For example, sulindac 320 ppm (the highest tolerated dose) caused a 36% reduction in ACF, while the maximum dose of SC-58635, a highly selective COX-2 inhibitor, resulted in a 41% reduction.25a Our data, however, are in contrast to a report that was unable to demonstrate inhibition in ACF with nabumetone 80 mg/kg injected s.c.26 This failure of protection may be secondary to both the intermittent dosing of nabumetone and the parenteral route of administration, which would be expected to result in a significantly lower concentration of nabumetone in the colonic lumen. Furthermore, the methodological shortcomings were underscored by the inability to demonstrate chemoprotection with indomethacin, an NSAID for which there are many reports affirming its ability to inhibit ACF formation.27 Our AOM-rat data are further supported by our MIN mouse studies, in which 900 and 1,200 ppm markedly suppressed intestinal adenomas. The approximately 50% reduction with nabumetone is equivalent to reports on other NSAIDs using this mouse strain. For instance, treatment with high-dose sulindac (84 mg/l of drinking water)28 and aspirin 500 ppm (the highest dose tested)29 caused MIN mouse tumors to be reduced by 39% and 48%, respectively.
Both the MIN mouse adenomas and the AOM-induced ACF were found predominantly in the distal bowel, consistent with the distribution of adenomas and carcinomas found in Western countries. Proximal colonic neoplasms appear to have distinct characteristics in histology, molecular genetics and demographics. For instance, patients with proximal colon cancers tend to be non-white, younger30 and female.31 These proximal tumors are biologically different from distal ones in that they are generally less differentiated,32 have a higher propensity for microsatellite instability (tumors whose genetic initiation is via loss of DNA mismatch repair capability with concomitant diffuse genetic instability)31 and have more frequent inactivation of the p16 tumor-suppressor gene.33 The genetic and pathological changes may correlate with clinical characteristics and natural history. Right-sided tumors have a better prognosis and are more sensitive to adjuvant chemotherapy than those on the left.31 Potential colon-cancer risk factors such as cholecystectomies,34 obesity and sedentary lifestyle35 may predispose more to proximal malignancies. This regional variability is also seen with chemopreventive agents; e.g., several epidemiological studies suggest that folate36 and possibly estrogens37 target the proximal colon. Chemoprotection in experimental models also appears to be site-specific with both oltipraz and DL-d-difluoromethylornithine protecting preferentially against distal AOM-induced tumors in rats.38
We demonstrated a marked accentuation of nabumetone's ability to suppress ACF in the distal colon. This observation was duplicated in MIN mice, where the efficacy of nabumetone was considerably greater in the distal small bowel, especially at low doses. The mechanism for the distal predilection of nabumetone in both models remains to be elucidated but may relate to regional differences in COX-2 expression, a presumptive target for NSAIDs in colon carcinogenesis. Genetic and pharmacological inhibition of COX-2 underscores its importance in colon-cancer prevention.39 In the pre-malignant mucosa of the AOM-rat, there was considerably greater COX-2 detected in distal as opposed to proximal colon. This regional distribution of COX-2 was confirmed in the uninvolved small bowel mucosa of the MIN mouse, with 4-fold greater COX-2 expression distally. Our data are supported by results demonstrating a robust increase in COX-2 expression distally in the uninvolved MIN mouse mucosa.40 The importance of gradients of COX-2 expression to the regional efficacy of nabumetone is suggested by the demonstration that genetic deletion of COX-2 decreased MIN mouse tumors more markedly in the distal than the proximal small bowel.24 Furthermore, the distal predominance of COX-2 expression appears to be important in human sporadic colon carcinogenesis, with rectal cancers over-expressing COX-2 in 90% of cases whereas more proximal tumors up-regulate COX-2 only 22% of the time.41 In our study, the cellular constituents responsible for COX-2 expression were not determined, but previous data indicate that expression of both epithelial and stromal COX-2 is important during carcinogenesis.42 While the correlation between COX-2 expression and nabumetone's chemopreventive efficacy is striking, there are several lines of evidence to suggest that this may be simplistic. For example, NSAID derivatives without anti-prostaglandin activity (i.e., sulindac sulfone, R-fluoribuprofen) were still able to protect against experimental colon cancer and induce cellular apoptosis.43, 44 PPAR-δ has been demonstrated to be a putative target, being over-expressed as a consequence of APC mutations in colorectal neoplasia.45 NSAIDs not only transcriptionally regulate this member of the nuclear receptor superfamily but also inhibit the ability of PPAR-δ to bind to DNA. We demonstrated that PPAR-δ is also up-regulated in colon adenomas and carcinomas in the AOM-rat model, where the APC gene generally remains wild-type. While we did not evaluate PPAR-δ in MIN mice, given its germline mutation in APC, we would anticipate significant over-expression during tumorigenesis. In AOM-treated rats, PPAR-δ was detectable in the uninvolved mucosa by immunohistochemistry and Western blot analysis. Moreover, expression of PPAR-δ, like COX-2, was markedly increased in the distal colon when compared to the proximal, consistent with a preliminary report that PPAR-δ and COX-2 co-localize in colorectal cancer.46
The present study is in agreement with results demonstrating that, in MIN mice, the NSAID piroxicam was markedly more effective at tumor suppression in the distal and mid-small bowel when compared to the proximal small bowel.47 However, in a report utilizing the AOM-rat model, piroxicam suppressed 82% of proximal tumors and only 57% of those in the distal colon.38 Therefore, localization of these putative NSAID targets is probably not the complete answer for the site specificity of nabumetone. Another potential explanation is the differential sensitivity to apoptosis in the proximal and distal colon. Inhibition of apoptosis is a crucial early event in colorectal carcinogenesis,48 which may be reversed by NSAID treatment.49 Previous reports with the AOM-rat model have demonstrated marked regional variability in apoptosis,50 and in humans, apoptosis is markedly increased in the distal colon.51 Unfortunately, there are no data on the differential effects of NSAIDs on the proximal vs. distal bowel with regard to apoptosis per se or induction of important mediators in the NSAID-apoptosis response, such as the prostate apoptosis response-4 gene52 or 15-lipoxygenase 1.53 However, endoscopic studies in humans suggest that the pro-apoptotic protein Bak is over-expressed in the distal colon, correlating with greater basal apoptosis in this location.51 While we did not evaluate regional differences in apoptosis or apoptosis mediators, we have previously reported that nabumetone down-regulated anti-apoptotic Bcl-2 in both cell culture and MIN mice.8 Since Bak expression is greater in the distal colon, one could speculate that the balance between pro- and anti-apoptotic forces might be more greatly altered toward cell death in the distal colon.
There are a variety of other potential explanations for the regional chemopreventive efficacy. Biologically, the proximal and distal colon vary markedly in fecal bile acid exposure, inherent carcinogen-detoxifying proteins and mucin production,32 all conceivable targets for nabumetone. Also, pharmacological properties of these NSAIDs may play a role in their site specificity. Nine percent of an oral dose of nabumetone is recovered from the feces (manufacturer's data). One could postulate that the different chemical properties among NSAIDs may alter the contact time with the proximal and distal colonic epithelium. In summary, there are several biologically plausible mechanisms for nabumetone's regional selectivity in the inhibition of colon carcinogenesis, including over-expression of COX-2 and PPAR-δ. Further studies will be conducted to elucidate the mechanism(s) involved in this clinically and biologically important observation.
In conclusion, nabumetone was effective in the AOM model of colorectal carcinogenesis. These data are complemented by the demonstration that nabumetone inhibited intestinal tumors in MIN mice. In both experimental models, the efficacy of nabumetone was markedly greater in the distal than the proximal colon. This may be related, at least in part, to over-expression of COX-2 and PPAR-δ in this region. These findings have implications for the study of the biology involved in colon carcinogenesis and underscore the necessity of focusing on regional differences of chemopreventive agents in designing effective colon cancer-prevention strategies.
- 8Nabumetone inhibition of intestinal tumorigenesis in MIN mice: modulation of BCL-2. Gastroenterology 1997;112: A647., .
- 13Activation of the beta-catenin gene by interstitial deletions involving exon 3 in primary colorectal carcinomas without adenomatous polyposis coli mutations. Cancer Res 1998;1: 58: 1021–6., , , , , , et al.
- 20Aspirin in colorectal cancer chemoprevention: a decision analysis. Am J Gastroenterol 1997;92: A413., , , , .
- 34Cholecystectomy and the risk of colon cancer. Am J Gastroenterol 1999;94: 41–6., , , , .Direct Link:
- 46PPAR-δ is a putative downstream receptor of COX-2 derived prostacyclins during colon carcinogenesis. Gastroenterology 2000;118: A3749., , , , , .