Prevention of Chemically Induced Urinary Bladder Cancers by Naproxen: Protocols to Reduce Gastric Toxicity in Humans Do Not Alter Preventive Efficacy

The NSAIDs have been a major focus in the field of chemoprevention for more than 25 years. Initially evaluated were the nonselective NSAIDs, particularly piroxicam (1), which was highly effective in preclinical models, but is now infrequently employed clinically. Approximately 20 years ago, it was shown that two different enzymes perform cyclooxygenation of arachidonic acid and were designated COX-1 and COX-2 (2). COX-1 is a constitutively expressed enzyme found in a wide variety of tissues. Although COX-2 is preferentially expressed in lymphoid cells, it can be induced by a wide variety of stimuli in many cell types. Interestingly, COX-2 has proven to be a direct target for cancer prevention (3–5). Specific inhibitors of COX-2 were synthesized and in preclinical studies were highly effective in the prevention of various types of cancer (6–8). Clinical trials of the COX-2 inhibitors were effective in clinical prevention trials of colon adenomas and squamous cell skin cancer (9, 10). However, placebo controlled adenoma trials of rofecoxib at the standard dose and celecoxib at doses higher than the standard human dose increased the incidence of adverse cardiovascular events (11). Therefore, there were significant efforts to synthesize agents with lower cardiovascular risks than the COX-2 inhibitors and certain standard NSAIDs (e.g., diclofenac). On the basis of epidemiologic data, it appeared that certain NSAIDs (e.g., naproxen) had lower cardiovascular risk than others (12), leading us to explore the chemopreventive efficacy of naproxen in the present study.

The second major toxicity associated with NSAIDs is gastric toxicity, particularly the induction of ulcers and potential life-threatening bleeds, which occurs in a very limited number of patients (13). We examined two different regimens to reduce the potential gastric toxicity of naproxen. First, naproxen was combined with the proton pump inhibitor omeprazole, as this approach has been shown clinically to substantially reduce upper gastrointestinal toxicity (14, 15). The second approach was to employ intermittent dosing with the NSAID on the expectation that it would facilitate recovery from any gastric damage associated with naproxen treatment.

The OH-BBN–induced urinary bladder cancers are highly invasive and appear histologically similar to human transitional cell carcinoma (TCC; ref. 16). The tumors appear by array analysis to have strong overlap both at the pathway level and at the specific gene level to invasive human bladder cancer (17, 18). The model has been shown to be highly sensitive to the preventive activity of a variety of NSAIDs, as well as various EGFR inhibitors (16). Interestingly, the efficacy of these two classes of agents is mediated by blocking tumor progression from microcarcinomas to the development of palpable invasive cancers, and not by blocking the formation of premalignant lesions (16).

The effects of these altered regimens to reduce gastric toxicity in humans on the efficacy of naproxen in the OH-BBN urinary bladder cancer model were examined. This study showed: (i) Naproxen administered by gavage either daily or intermittently 1 week on/1 week off, or 3 weeks on/3 weeks off, all similarly inhibited urinary bladder cancer formation 73%–82% (P < 0.005); (ii) the half-life of naproxen in the rat (roughly 3.0 hours) is much shorter than the half-life in the human; (iii) omeprazole, which by itself did not decrease bladder cancer formation, did not interfere with the efficacy of NPX; (iv) although NPX failed to alter proliferation related biomarkers, it did increase levels of the apoptotic proteins caspase-3 and caspase-7.

Naproxen and omeprazole were obtained from Sigma Chemical Company. The carcinogen OH-BBN was purchased from TCI America. Female Fischer-344 rats were obtained from Harlan Sprague-Dawley, Inc at 4 weeks of age. Diets were purchased from Teklad (Harlan Teklad) and were provided ad libitum.

The OH-BBN–induced urinary bladder cancer model was performed as previously described (16, 19). Beginning at 56 days of age, rats (30 per group) were treated twice a week with 150 mg OH-BBN/gavage for 8 weeks. The rats were weighed 1 time/week, and palpated for urinary bladder tumors 2 times/week. Intragastric administration of naproxen and/or omeprazole was initiated 2 weeks after the final OH-BBN treatment. The vehicle for naproxen was saline and for omeprazole the vehicle was polyethylene glycol 400: ethanol (90:10, v/v). The gavage volume was 0.5 mL/rat, and animals were treated 7 times/week. Beginning 2 months after the last dose of OH-BBN, rats were checked weekly for the development of palpable bladder tumors. The studies were terminated 7 months after the last OH-BBN treatment. At necropsy, urinary bladders with associated lesions were excised and weighed. Percent incidence and the percent of rats with large tumors >200 mg bladder weight were recorded as endpoints for cancer prevention. Statistical analysis of bladder weights were performed by the Wilcoxon rank test and the final tumor incidence was analyzed by the Fisher exact test (16).

At 60 days of age, rats were administered naproxen (40 mg/kg body weight) in saline by gavage either once or for 14 consecutive days. At time points 0, 1, 2, 4, 8, and 24 hours, blood (N = 4) was obtained. The serum was frozen in liquid nitrogen and stored at −80°C. The serum was sent to Midwest Research Institute for quantitative analysis of naproxen levels. A liquid chromatography/tandem mass spectrometry (LC/MS-MS) method for the quantitation of naproxen in rat serum was employed. Rat serum samples (50 μL) were fortified with naproxen across the concentration range. Ketoprofen was added as an internal standard (17). The samples were extracted via protein precipitation before analysis. Metabolites were separated by high-performance liquid chromatography (HPLC) on a Phenomenex luna C18 column 5 μm Mobile Phase A: ammonium acetate buffer, pH 4.0: mobile phase B: 100% acetonitrile gradient type: isocratic, 25% A/75% B flow rate: 1.0 mL/minute: run time: 8.0 minutes. The resulting samples derived from HPLC separation were analyzed on an Applied Biosystems API-4000 Q—Trap Mass Spectrometer. The LC/mass spectrometry procedure employed is based on the method of Elsinghorst and colleagues, across a range of 40 to 1,000 ng/mL (20).

Female Fischer-344 rats with urinary bladder cancers induced with OH-BBN were treated intragastrically with 40 mg/kg naproxen for 5 days or with the vehicle. At the time of sacrifice, bladder cancers were quickly removed and processed for immunohistochemistry. Slides were prepared for terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) assay and immunofluorescence microscopy. Apoptosis was determined either by the TUNEL assay, using the DeadEnd Colorimetric TUNEL System (Promega) according to the manufacturer's instructions or alternatively stained for caspase 3 or caspase 7. Briefly, after deparaffinization and rehydration, the tissue sections were pretreated with 20 μg/mL proteinase K solution for 10 minutes at room temperature. Thereafter, slides were rinsed in PBS and incubated with TUNEL reaction mixture for 1 hour at 37°C in a humidified chamber. Slides were then washed with PBS followed by a stop solution for 10 minutes at room temperature. The slides were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) under glass coverslips. The stained tissue was examined at 200× magnification 9 using a Nikon Eclipse TE2000-E Confocal microscope. The specific antibodies employed immunofluorescence microscopy. Slides were baked at 60°C for 2 hours, deparaffinized with xylene, and rehydrated through a graded alcohol bath. Sodium citrate buffer (10 mmol/L, pH 6.0) was used for antigen retrieval by heating slides and buffer for 10 minutes in the microwave followed by 20 minutes of ambient cooling. Slides were washed with deionized water (DI) water followed by 1× PBS. For immunofluorescence, specimens were blocked in 5% donkey serum-1×PBS-0.3%Triton (PBST) buffer for approximately 1 hour. The primary antibody was prepared according to the manufacturer's instructions and left on overnight at 4°C. Slides were washed in 1× PBS followed by 2 hours of appropriate secondary antibody incubation prepared according to the manufacturer's instructions. Slides were washed with 1× PBS followed high salt PBS (23.38 g NaCl in 1 L 1× PBS). Fluoro-Gel II with DAPI was used to mount glass coverslips and then sealed using clear nail polish. All stained tissues were examined at 200× magnification using a Nikon Eclipse TE2000-E Confocal microscope. Antibodies employed: primaries: p-Akt (Ser473) Rb (Cell Signaling Technology): 4060S, Lot #14 p53 Total Mo (Cell Signaling Technology): 2524, Lot #4 p-AMPK (Tyr172) Rb (Santa Cruz Biotechnology): sc-33524, Lot #C1011 p-p53 (Ser392) Gt (Santa Cruz Biotechnology): sc-7997, Lot #L1809 p-mTOR (Ser2448) Rb (Cell Signaling Technology): 5536S, Lot #1 p-mTOR (Ser2448) Rb (Cell Signaling Technology): 5536S, Lot #1; caspase-3 (Cell Signaling Technology): 9664, Lot #18; caspase-7 (Imgenex): IMG 5702. The TUNEL assay was determined, using the DeadEnd Colorimetric TUNEL System (Promega) according to the manufacturer's instructions.

All quantitative results are expressed as mean values ± SD. Statistically significant differences were obtained using the Student t test or by one-way ANOVA. A P < 0.05 value was considered to be statistically significant.

We had previously found that daily dosing with naproxen (40 mg/kg), which is roughly the HED (human equivalent dose), prevented the development of urinary bladder cancers in OH-BBN–treated rats. OH-BBN was administered for 8 weeks starting at 8 weeks of age. Two weeks after the last OH-BBN dosing, rats were administered: naproxen daily (40 mg/kg body weight/day, by gavage; group A); 1 week daily naproxen/1 week vehicle (group B); 3 week daily naproxen/3 week vehicle (group C) or only vehicle (group D). At the end of the study, the number of rats with large urinary bladder cancers (> 200 mg) was: (A), 8/30; (B), 6/29; (C), 5/28; (D), 24/25; (Fig. 1; P < 0.01 for all three treatment groups vs. vehicle control). Tumor development monitored by palpation (Fig. 1) shows the increased latency and lower final incidence of the various treatment groups versus vehicle treatment. Measurement of the weights of the urinary bladder plus cancers showed that the final weights were higher in vehicle controls than in any of the three naproxen treatment groups (P < 0.025; Supplementary Fig. S1A). In contrast, none of the three treatment groups differed significantly from one another.

Starting at 2 weeks after the last dose of OH-BBN, rats were administered on a daily basis: (A) omeprazole (4 mg/kg body weight), (B) naproxen (40 mg/kg body weight), (C) omeprazole + naproxen, or (D) vehicle. Groups A, B, C, and D had large bladder cancers in 28 of 29, 8 of 30, 10 of 29, and 24 of 25 of the rats (Fig. 2) at the end of the study. Thus, naproxen alone or naproxen with omeprazole was similarly effective in reducing the incidence and increasing the latency of large bladder cancer formation. The efficacy of naproxen alone or naproxen plus omeprazole is also reflected in the final weights of the tumors (Supplementary Fig. S1B). This showed that final bladder weights (bladder + cancers) of animals treated with naproxen alone or naproxen plus omeprazole were significantly different from control (P < 0.025), but not different from one another. Interestingly, while omeprazole by itself did not alter the latency or final incidence of palpable cancers, the cancers that did develop in the rats were larger in size than tumors in rats given the vehicle alone, although it failed to reach statistical significance (P > 0.05). For reasons unrelated to the current study, we treated 10 rats with the known bladder tumor promoter rosiglitazone. As can be seen in Fig. 2, rosiglitazone greatly decreased tumor latency in this study.

Serum levels of naproxen were determined in rats treated with naproxen (40 mg/kg body weight), by gavage, for either 1 day or 14 days (Table 1). The half-life of naproxen was approximately 3 hours at either time point. The C_max was roughly 300 μmol/L, although levels greater than 100 μmol/L were probably achieved for less than 6 hours following gavage treatment. A graph of the pharmacokinetic data is presented in Supplementary Fig. S2.

The effects of limited treatment with naproxen (7 days) on a variety of potential biomarkers were examined. Naproxen increased levels of three apoptosis-related biomarkers (caspase-3 and -7, TUNEL). In contrast, it failed to alter Ki67 levels (Fig. 3). Additional biomarkers related to the AKT/PI3K/MTor pathway were also examined (Supplementary Fig. S3). Although, it did slightly, but significantly, decrease levels of phosphorylated AKT (S473) and phosphorylation of PI3K, the relative levels of inhibition were quite limited.

The NSAIDs and the Coxibs are highly effective in inhibiting the development of OH-BBN–induced invasive urinary bladder cancers (16). The OH-BBN–induced bladder cancers have significant overlap with human invasive bladder cancer at the gene expression level (17, 18). In prior studies, we presented our rationale that a bladder plus cancers that was greater than 200 mg was likely to be identified by palpation (16). Using this endpoint, we had previously shown that naproxen and other NSAIDs/Coxibs were effective when administered beginning 3 weeks or even 3 months after the final OH-BBN administration. In the late intervention, microscopic lesions already existed in the majority of rats at the time of initial treatment (16). This result implies that the major effect of the agents is on the progression of cancers from the microscopic to the palpable size.

We have examined the efficacy of naproxen in multiple animal models because it had the best clinical cardiovascular profile of any of the NSAIDs besides aspirin. Our prior data found that naproxen was highly effective in colon, bladder, and skin tumor models (16, 19–21). However, naproxen (like most nonselective NSAIDs including aspirin) causes gastric toxicity in humans (12–14). While the likelihood of a fatal hemorrhage is relatively low, the incidence of ulceration is significant; particularly with chronic and long-term dosing. We, therefore, investigated protocols that might reduce the gastric toxicity of naproxen. The first approach is quite obvious and entailed combining naproxen with omeprazole. This combination of naproxen plus a proton pump inhibitor (PPI) inhibitor is used clinically for persons with arthritis who require an NSAID for pain, but who are at cardiac risk (14, 15). As shown in Fig. 2, omeprazole by itself minimally affected the development of palpable cancers, whereas naproxen reduced urinary bladder cancer incidence by approximately 75% (P < 0.01). Most importantly, the combination of naproxen plus omeprazole was as effective as naproxen alone. When the weights of the cancers (Supplementary Fig. S1B) were evaluated, naproxen alone and naproxen + omeprazole decreased the weight of the cancers relative to the controls. Interestingly, although omeprazole did not decrease the latency of large urinary bladder cancers relative to OH-BBN alone, it did increase the final weights of the resulting cancers, although not significantly (P > 0.05). For unrelated reasons, we included 10 rats that were treated with OH-BBN and a PPARγ agonist rosiglitazone in this study. We had previously shown that this agent promoted bladder cancers in this animal model (22). The main point is that while rosiglitazone clearly decreased tumor latency in this model, omeprazole did not. These results imply that this combination of naproxen + omeprazole is highly effective and, based on clinical data, likely to decrease gastric toxicity. The gastric toxicity cannot be modeled in the animal as gavage dosing of NPX at 40 mg/kg body weight even for many months failed to induce significant gastric lesions. In fact, doses of naproxen up to three times higher similarly failed to induce significant gastric toxicity in rats. We, therefore, feel that performing a toxicity study at doses far in excess of the dose we employed in the study, which is roughly the HED, made little sense. Parenthetically, investigators have shown that the levels of omeprazole we employed (which are at the human equivalent dose) decreased the acidity of the stomach in rats (23).

Our second approach to reducing toxicity entailed intermittent dosing with naproxen. This employed either daily dosing with naproxen or intermittent dosing 1 week on/1 week off or 3 weeks on/3 weeks off. We were surprised to find that the intermittent dosing was as effective as daily dosing with the agent. Simultaneous pharmacokinetic studies showed that the half-life of naproxen in the rat was relatively short (t_1/2), about 3 hours, in contrast to the human t_1/2 of 15 hours. This argues that the efficacy observed with intermittent dosing is unrelated to extended serum levels. Interestingly, the pharmacokinetic data show that naproxen and/or certain of its conjugates are overwhelmingly excreted via the urine. This implies that naproxen may be particularly applicable for use in urinary bladder cancer. Piroxicam has shown therapeutic efficacy in urinary bladder cancer in dogs (24–26). The use of naproxen, which preferentially accumulates in urine in humans (27), might allow the use of a lower dose with more limited gastric toxicity than would be required with piroxicam. It might also suggest its use in prevention of urinary bladder cancer in humans, in which there is at least some data supporting the use of NSAIDs (28). We have recently tested intermittent dosing with naproxen in a rat model of colon cancer and have found it to be highly effective (Rao C, Steele VE, Lubet RA; unpublished data).

Cyclooxygenase inhibitors (NSAIDs/Coxibs) can prevent cancer in preclinical models, in epidemiologic studies, and in clinical trials both as primary or secondary endpoints. The clearest evidence in humans is in colon cancer (29), esophageal cancer (30), and in squamous cell skin cancer (9). However, there is also significant evidence in various other cancers (based on epidemiologic data as well as preclinical data). Regarding urinary bladder cancer (28), there is some, albeit sometimes conflicting, epidemiologic evidence. Furthermore, there is clear efficacy of standard NSAIDs and Coxibs in the treatment of bladder cancer in dogs and rodents, and a hint of efficacy in certain clinical trials (31). We have done extensive work with this class of agents in the OH-BBN–induced model of bladder cancer in rats (7, 16, 32).

Our prior studies have shown that various NSAIDs/Coxibs (naproxen, NO-naproxen, sulindac, celecoxib at their HEDs and aspirin at doses in substantial excess of its HED) are effective in blocking the development of large palpable cancers even when initiated after animals have preinvasive microscopic cancers (16). One of the great hurdles to employing NSAIDs/Coxibs in a preventive setting (despite clear efficacy) is potential toxicities. Gastric toxicity is quite clear with a variety of NSAIDs and aspirin, while cardiovascular toxicity is clearer with Coxibs and certain NSAIDs (e.g., diclofenac). In fact, lower doses of aspirin are cardioprotective, whereas naproxen has the best cardiovascular profile of the standard NSAIDs (33). Presumably, the positive cardiovascular effects of low-dose aspirin, and perhaps naproxen, are due to their ability to inhibit COX-1 activity and decrease thromboxane levels. Aspirin achieves this by being a suicide substrate for COX-1, whereas naproxen is a long lived nonspecific inhibitor that inhibits both COX-1 and COX-2. Naproxen is recommended for persons with arthritis who need an NSAID, but have higher cardiovascular risk. It is administered together with a PPI inhibitor, often omeprazole. In fact, the PPI inhibitors have been shown to substantially reduce the gastric toxicity associated with NSAIDs by 50% to 75% in most studies (14, 15). The finding that the PPIs do not inhibit the antiarthritic effects of NSAIDs made us optimistic that they were not likely to interfere with the preventive efficacy of this NSAID. The rationale being that the antiarthritic effects of naproxen are likely driven by COX-2 inhibition. As this is also likely to be a primary mechanism of cancer prevention as well (3–5), we felt that omeprazole was not likely to inhibit the efficacy of naproxen. As shown in Fig. 2, omeprazole at its HED did not alter bladder cancer formation. In contrast, naproxen at its HED (40 mg/kg body weight/day) or the combination of naproxen and omeprazole at the HEDs were highly effective in the prevention of bladder tumors. Importantly, the addition of omeprazole or esomeprazole has been shown to substantially reduce any gastric toxicity associated with naproxen treatment (14, 15, 34). Parenthetically, the combination of esomeprazole and naproxen is currently available as a fixed dose product in the marketplace, and is approved for use in arthritis.

Alternatively, intermittent naproxen appears totally effective. The concept from the gastrointestinal perspective is that intermittent dosing may allow for recovery of any gastric lesions, due to regeneration of COX activity as well as natural repair mechanisms. However, cotherapy with a PPI is thought to be a safer gastrointestinal risk reducing strategy as there is demonstrated clinical efficacy (14, 15). As described above, we cannot test the ability of intermittent dosing to reduce gastric toxicity in rats as gavage dosing of naproxen (40 mg/kg body weight/day) for multiple weeks does not cause significant gastric toxicity. Given the relatively short half-life of naproxen in the rat, we are not sure why the intermittent dosing appears to be so effective. The efficacy of intermittent dosing with either naproxen or high-dose aspirin in a colon model of carcinogenesis was recently observed (C.V. Rao, V.E. Steele, and R.A. Lubet; unpublished data).

In an attempt to determine the mechanism of action of naproxen, we examined its effects on a variety of potential biomarkers. The basic approach is similar to the presurgical model of biomarkers employed in humans (35). Persons with early-stage cancer are exposed to potential agents for a limited time period before surgery. We have routinely used this approach in a breast cancer model employing either Ki67 or gene arrays to perform the analysis (36). Our laboratory used such an approach in the OH-BBN bladder model to investigate the efficacy of the EGFR inhibitor Iressa, and found alterations of a wide variety of biomarkers and genes (37). We hoped as NSAIDs in general (naproxen in particular) were effective during the later stages of tumor progression, we might observe alterations in a variety of biomarkers. As shown in Fig. 3, we observed no effects on Ki67 staining. We did, however, observe significant increases in the apoptotic biomarkers TUNEL and Caspases 3 and 7. Furthermore, we observed differences in AKT phosphorylation and PI3K phosphorylation (Supplementary Fig. S3). However, the magnitude of the changes in AKT and PI3K phosphorylation are limited, and would be difficult to use as a biomarker clinically. The changes observed in caspases and in AKT and PI3K phosphorylation are in agreement with our recent cell culture studies employing naproxen (38). As rats were treated for 5 days with naproxen it may be that all of the observed changes are secondary to alterations in prostaglandins following COX inhibition.

The most important question associated with these studies deal with whether the results presented are likely to reduce adverse events in human. Regarding upper gastrointestinal toxicity, omeprazole has clearly demonstrated a striking ability to decrease gastrointestinal bleeding by 50% to 75% in multiple clinical studies (13, 15). There is no reason to think that it would not be effective in a prevention study. The prior findings in arthritic patients that omeprazole reduces gastric toxicity without reducing efficacy of naproxen is particularly important as we feel that the antiarthritic effects of naproxen are related to inhibition of COX enzymes, which is the same determinant of its preventive activity (3–5). The clinical question regarding potential trials should focus on baseline gastrointestinal risk. If one precludes persons with prior bleeds or who are greater than 70 years of age (or with major gastrointestinal comorbidity), than one might expect to achieve a low incidence of clinically significant upper gastrointestinal events that would be further reduced by a PPI; for example, omeprazole. An alternative is to use intermittent dosing with an NSAID. Certainly, one could expect that dosing for 3 to 4 weeks on and 3 to 4 weeks off in humans would decrease gastric toxicity. This presumably would allow sufficient time for significant repair of any erosion that might arise. However, as gastrointestinal toxicity can occur, even short-term combinations with a NSAID and PPI inhibitor that have demonstrable clinical efficacy may be preferable. Alternative methods to reduce gastric toxicity of NSAIDs is the use of NO or H₂S analogs (32). However, these are not nearly as advanced clinically and the agents are likely to be more expensive than omeprazole or naproxen, which are off patent.

One question relevant to clinical trials with NSAIDs is the use of aspirin. There has been substantial enthusiasm for employing low-dose aspirin as a cancer preventive (39). There are a few purely practical problems associated with trying to perform aspirin trials (e.g., the exact dose to employ; ≤100 mg vs. ≥ 325); how does one randomize the high numbers of persons taking low-dose aspirin; and the fact that persons on even low dose aspirin may have significant upper gastrointestinal adverse effects. In addition, there are a few results that make a comparison of aspirin with a standard NSAID of great interest: (i) animal data from various organs (colon, bladder, skin) show aspirin is routinely ineffective at doses that correspond to human doses of 100 or 325 mg, (ii) while a wide variety of NSAIDs and Coxibs are highly effective at their HEDs. These animal models predicted the high clinical efficacy of Coxibs and the combination of NSAIDs plus difluoromethylornithine (DFMO) in colon and skin (9, 10, 40); (iii) there is clinical trial data implying potentially greater efficacy of NSAIDs versus low-dose aspirin. Thus, celecoxib resulted in roughly a 45% decrease in total colon adenomas and a 55% to 65% decrease in advanced adenomas (10), and were similarly effective in individuals taking low-dose aspirin. The combination of sulindac plus DFMO resulted in a 65% decrease in total adenomas (40), and an even higher inhibition of advanced adenomas. The inhibition observed in these trials is far greater than that of low to moderate doses of aspirin resulting in a roughly 20% to 25% decrease in total adenomas and a 35% to 40% decrease in advanced adenomas. Finally, the efficacy of NSAIDs or coxibs in treatment of urinary bladder cancer in dogs and cats encourages examination of more standard NSAIDs. These studies indirectly argue for the potential of a human trial comparing the efficacy of aspirin with more standard NSAIDs (e.g., naproxen vs. aspirin).

J.M. Scheiman is a consultant/advisory board member for AstraZeneca, Pfizer, and Pozen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R.A. Lubet, J.M. Scheiman, L. Minasian, V.E. Steele, C.J. Grubbs

Development of methodology: V.E. Steele

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.M. Bode, J. White, C.J. Grubbs

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.M. Scheiman, A.M. Bode, M.M. Juliana

Writing, review, and/or revision of the manuscript: R.A. Lubet, J.M. Scheiman, A.M. Bode, L. Minasian, D. Boring, V.E. Steele, C.J. Grubbs

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.J. Grubbs

Study supervision: R.A. Lubet, V.E. Steele, C.J. Grubbs

Other (was involved in obtaining efforts of Dr. Bode in performing the immunohistochemical analysis and getting the collaboration of Dr. Scheiman with regard to the manuscript): R.A. Lubet

The funding for the studies was provided in part by NCI Contract Number HHSN261200433001C (NO1-CN-43301) awarded to C.J. Grubbs at the University of Alabama at Birmingham (Birmingham, AL).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.