Does Mutated K-RAS Oncogene Attenuate the Effect of Sulindac in Colon Cancer Chemoprevention?

Colorectal cancer is the third most common cancer in the United States and is a leading cause of cancer-related death worldwide (1). If detected early, colorectal cancer is a treatable cancer with a 5-year survival rate of 90% (1). However, the 5-year survival rate drops to as low as 11% if the disease is found in advanced stages. Key factors for early detection and prevention of colon cancer include defining risk factors and understanding the stages of this disease.

Colon cancer develops gradually in normal colon epithelium through precancerous aberrant crypt foci (ACF), which can progress to early and intermediate adenomas and then malignant tumors via an accumulation of chromosomal abnormalities, genetic mutations, and epigenetic changes (2–4). Inactivation of tumor suppressor genes and DNA mismatch repair genes, and activation of oncogenes are important genetic mutations in colon neoplastic progression (5). K-RAS mutations are frequently found in colorectal cancer along with mutations of the adenomatous polyposis coli (APC) and p53 genes (5, 6). K-RAS is mutated in about one quarter of advanced colon adenomas (7) and in nearly one half of colon cancers (5, 8). Constitutive activation of K-RAS in colon cancer generally occurs by substitution of amino acid residues in codons 12 and 13 and, less frequently, in codon 61 (6, 9, 10). K-RAS mutations can be detected early in carcinogenesis, between the stages of early and intermediate adenoma, and maintain a constant incidence in late adenoma and in carcinoma (11, 12). K-RAS is a member of three highly homologous oncogenes and encodes a monomeric protein of 21 kDa (p21^ras), which is able to bind and hydrolyze GTP. This membrane-associated guanine nucleotide binding protein acts as a molecular switch for signal cascades that modulate many aspects of cell behavior, including proliferation, differentiation, motility, and death (13). Activated RAS triggers uncontrolled proliferation and morphologic alteration, contributing to the malignant phenotype of transformed cells. Oncogenic RAS influences different cellular processes in a cell type- and disease-specific manner via a variety of signaling pathways, particularly those that include the serine/threonine protein kinase Raf-1, Rho small GTPases and the AKT proto-oncogene (14, 15).

NSAIDs, including sulindac, inhibit the cyclooxygenase isozymes COX-1 and COX-2, which catalyze the rate-limiting step in the metabolic conversion of arachidonic acid to prostaglandins. Sulindac, a nonselective COX-2 inhibitor, is a prodrug that is metabolized by the liver and intestinal flora to the sulfide and sulfone forms (16). Sulindac sulfide is the active anti-inflammatory metabolite, primarily responsible for blockage of prostaglandin synthesis, whereas sulindac sulfone has no COX-inhibitory activity (16, 17). Sulindac is known to induce apoptosis in colon cancer cells (18, 19) and to cause regression of colon polyps in patients with familial adenomatous polyposis (FAP) and with sporadic adenomas (20). At the same time, long-term treatment with sulindac in FAP patients does not inhibit the progression of polyps toward malignancy (21). Analysis of molecular changes in adenomas of FAP patients that failed to respond to sulindac therapy suggested that K-RAS mutations may be associated with this resistance (22). Mutant K-RAS–associated resistance to the sulindac metabolite sulindac sulfide also has been reported in vitro, in the rat intestinal cell line IEC-18 transformed with K-RAS oncogene (23). In this study, our goal was to directly assess the effect of sulindac treatment on the formation of colon adenomas that express the mutant K-RAS oncogene. We conducted this study using a novel mouse model of colon cancer with the colon-specific and inducible expression of mutant K-Ras^G12D allele. We applied minimally invasive optical coherence tomography (OCT) to monitor the development of colon adenomas in this mouse model through time serial imaging, which enabled quantification of the chemopreventive effect of sulindac during the course of wild-type (wt) and mutant K-RAS–driven colon tumorigenesis.

The mouse model with inducible colon-specific expression of the mutant K-ras allele (Vil-Cre-ER^T2 K-ras ^G12D) was generated and characterized as described in Supplementary Materials. Animals were housed at the University of Arizona's Animal Care Facility in groups of one to five in microisolator cages under fluorescent lighting on a 12-hour cycle in accordance with the University of Arizona Institutional Animal Care and Use Committee guidelines. Mice of three genotypes were used in this study: Vil-Cre-ER^T2–positive K-ras–mutant mice (K-ras^G12D), Vil-Cre-ER^T2–positive K-ras wild-type mice (K-ras wt), and Vil-Cre–negative K-ras mutant mice (Vil-Cre–negative K-ras^G12D).

Experimental protocols were approved by the Institutional Animal Care and Use Committee. The study timeline is illustrated in Fig. 1A. Mice in each of three genotypes were divided into four experimental groups: control no sulindac, control sulindac, AOM-induced no sulindac, AOM-induced sulindac. The groups and number of mice per group are presented in Fig. 1B.

Mice were fed the defined pelleted synthetic diet AIN93G or AIN93G formulated with 100 ppm (0.1 g/kg) sulindac-TD.08740 (Harlan Teklad). Mice in sulindac-treated groups were provided sulindac-containing diet beginning at 11 weeks of age, which was 1 week post-AOM treatment. The approximate daily consumption of sulindac was 25 mg/kg based on average mouse weight of 20 g and food consumption of 5 g per day. The dietary application of sulindac has been successfully used in the murine cancer models in many prior chemoprevention studies, and amounts are based on the recommended human doses (24–26). Treatment with the colon-specific carcinogen azoxymethane (AOM, Sigma) was performed to induce adenomas. AOM was injected intraperitoneally at concentration of 10 mg/kg in saline once a week for 4 weeks beginning at 7 weeks of age. Control mice received 0.2 mL normal saline using the same regime of injections. Irradiated reverse osmosis water was available ad libitum for the duration of the experiment.

The endoscopic ultrahigh resolution OCT system employed in this study has been previously described in detail (27). The light source was a superluminescent diode (Superlum Broadlighter, Moscow, Russia) centered at 890 nm with a 150-nm full-width-at-half-maximum (FWHM) bandwidth. The endoscope optics consisted of a fiber, gradient index lens, and right angle prism that directed the light out the side of a 2-mm diameter endoscope, which could be inserted approximately 30 mm into the colon. Linear and rotational motion of the optics within a protective glass envelope enabled sampling of the distal colon. Cross-sectional images 2 mm depth by 30 mm length (1,024 × 5,000 pixels) were obtained at 45-degree increments, resulting in 8 images around the circumference of the colon lumen. The generated images have an axial resolution of 3.5 μm in air and a lateral resolution of 5 μm. Depth of imaging was limited to 1.4 mm; consequently, the deep boundary of larger tumors was not completely visible.

Eighteen- to 24 hours prior to imaging with the OCT system, food was withheld from the mice in an effort to clear the colon. The mice were removed from the standard individually ventilated cages (IVC) and placed into empty microisolator cages with wire bottom inserts to keep them dry and to prevent coprophagia. Mice had free access to nonflavored pediatric electrolyte replacement supplement solution. The mice were then transferred to standard IVC cage and bedding and transported for colonic OCT imaging. Mice were anesthetized prior to imaging with a mixture of 100 mg/kg ketamine, 10 mg/kg xylazine administered intraperitoneally (IP), or 2.5% Avertin IP. The distal colon was lavaged with 0.9% sodium chloride irrigation solution, USP, to clear residual fecal material. Following colon lavage, mice were placed on a warming pad and imaged with the OCT system. They were kept on warming pads until recovered from anesthesia. Mice were imaged starting at 13 weeks of age through 34 weeks of age at 4-week intervals.

Each set of 8 OCT images was analyzed for all mice in this study, at each time point, to determine tumor count and tumor burden. The criteria for adenoma identification was previously described in detail by Hariri and colleagues (28) and includes mucosal thickness twice the local average, attenuation of signal, and faint boundaries between the mucosa and submucosa tissue boundary. Adenomas were identified, counted, and the maximum width was measured. The volume of each tumor was calculated using the maximum width as the tumor diameter and assuming a spherical shape.

Mucosal thickness for each control group genotype, with and without sulindac treatment, was determined by analyzing the ventral-most OCT image from each image set at each time point. The average thickness of the proximal 10 mm of the mucosal layer of the colon was measured using custom MATLAB software. A blinded observer identified the luminal and deep boundaries of the mucosa, and the software calculated the average thickness.

After final imaging at 34 weeks of age, the mice were euthanized using CO₂ asphyxiation. A 2-mm diameter lubricated glass rod was inserted 30-mm into the colon. After the abdomen was opened and the colon exposed, tissue marking ink was used to mark dorsally, to orient with OCT positions. After excision, colons were placed on filter paper, cut longitudinally along the marked line, and flattened. The colons were examined visually and photographed. Another piece of filter paper was placed over the colon to keep it flat for histology processing. The colons were fixed in 10% buffered formalin or Histochoice (Ameresco), paraffin-embedded, and cut longitudinally into 6-μm sections every 250 μm through the entire colon. The slides were stained with hematoxylin and eosin (H&E). Histologic features of colon tumors from the entire colon were evaluated by a certified veterinary pathologist (one animal per each genotype and treatment). The grade of dysplasia for each adenoma examined histologically was determined on the basis of criteria outlined in a review on the pathology of mouse models of intestinal cancer (29).

The H&E-stained sections of the colons of all animals were analyzed by an independent observer for tumor number and burden. Tumors were counted and volume was determined by measuring the maximum diameter of each discrete tumor using the Olympus Microsuite FIVE (2007) software, and assuming a spherical tumor, as described previously (30). A comparison of final time point OCT- and histology-determined tumor number and burden was made.

An additional set of K-ras wt and K-ras^G12D mice described in Supplementary Table S3A was sacrificed at the age of 18 weeks by CO₂ inhalation, and the whole small intestine and colon segments were removed, flushed with buffered saline, opened longitudinally and laid flat, mucosal surface up. Tissues were fixed in 10% neutral-buffered formalin for 4 hours. After 4 hours, small intestines and colons were washed with cold PBS and stored in 70% ethanol at 4°C. Fixed tissues were briefly stained with 0.5% methylene blue, and the colons were evaluated for the presence of the ACF and tumors. Morphologic evaluation of ACF and tumors was performed using a brightfield Nikon microscope at ×20 magnification. Tumors in the proximal, middle, and distal portions of the small intestines were also counted where applicable. Tumors were counted starting at a diameter of 0.5 mm.

IHC was performed using the automated Ventana platform from Ventana Medical Systems (VMS) using a streptavidin DAB Detection Kit from VMSi bed as described previously (30). The tissue was stained for proliferation marker Ki-67, apoptosis marker cleaved caspase-3, and inflammation marker COX-2. Staining conditions and antibody dilution are described in Supplementary Materials. The appropriate positive control tissue was used for each antibody assay. Analysis of the Ki-67 and CC3 staining was done by manually counting the number of positively stained cells in the crypts directly adjacent to the muscularis in 50 colonic crypts per slide. The slides stained for COX-2 were read by an experienced pathologist (R.B. Nagle) with 30 years of experience, who was blinded to treatment categories. Results are presented as a long score based on the sum of intensity of staining multiplied by the percent of stained tissue area. The following scoring criteria were used: 0, no staining; 1+, week diffuse staining (may contain stronger intensity in <10% of the cells); 2+, moderate staining in 10% to 90% of the cells, and 3+, more than 90% of the cells stained with strong intensity. Staining for all endpoints was analyzed in three fields per slide.

Comparison between genotypes and groups (e.g., sulindac) on tumor latency, tumor number, and tumor burden was statistically analyzed using the following techniques. Cox regression was performed to determine the latency (timing of first adenoma detection). Wilcoxon rank sum /Kruskal–Wallis tests were performed to compare the tumor number derived from OCT analysis at 34 weeks between groups/genotypes. Square root transformation was applied to the tumor number, and then, linear regression was performed to determine whether the genotypes modulated the group effects on the tumor number. On the basis of the substantial number of mice without tumors, Fisher exact tests were also performed to compare the percentage of mice with at least one tumor between groups/genotypes, and logistic regression was performed to determine whether the genotypes modulated the group effects. Wilcoxon rank sum /Kruskal–Wallis tests were also performed to compare tumor burden derived from OCT analysis at 34 weeks between groups/genotypes. Linear regression was performed to determine whether the genotypes modulated the group effects on tumor burden. Similar techniques were used for comparisons of tumor number and tumor burden from histology. For tumor grade analysis, the Dunnett test and Bonferroni multiple comparison test were used. For IHC staining and analysis of mucosa thickness, linear mixed effects models with random intercept were fitted to account for within-mouse correlation and compare between groups (genotype, sulindac, and AOM).

Through the analysis of common downstream targets of mutant K-RAS oncogene, we confirmed that our model successfully recapitulates the main features of activated K-RAS signaling reported in humans (Supplementary Material; Supplementary Fig. S2).

Examples of OCT images obtained from the AOM-induced experimental groups of K-ras^G12D and K-ras wt genotypes at the final imaging time point are presented in Fig. 2 along with corresponding histology. Control animals of Vil-Cre–negative K-ras^G12D and K-ras wt genotypes did not develop any colon tumors over the course of the study, but a low number of tumors was detected in K-ras^G12D mice (Table 1A). Treatment with AOM resulted in a significant increase in the number of colon adenomas and tumor burden in K-ras^G12D animals (on average 6.91 colon tumors and 50.43 mm³ tumor burden per mouse) compared with the K-ras wt mice (average 0.83 tumor per mouse and total tumor burden of 10.89 mm³; Tables 1A and 1B, P < 0.0001).

In AOM-induced K-ras^G12D mice, colon adenomas were detected by OCT imaging beginning at 18 weeks of age (8 weeks after the last AOM injection, second imaging time point). In contrast, in AOM-induced K-ras wt, and Vil-Cre–negative K-ras^G12D mice, the first adenomas were detected at 26 and 30 weeks of age, respectively. This time difference was significant in K-ras^G12D compared with K-ras wt mice (P < 0.0001) and K-ras^G12D compared with Vil-Cre–negative K-ras^G12D mice (P < 0.0001). According to OCT image measurements (Table 1A), sulindac treatment decreased the number of adenomas for both AOM-induced K-ras wt (P = 0.02) and K-ras^G12D (P < 0.01) mice (Tables 1A and 1C, P < 0.001) and the decrease depended on the genotype of AOM-induced mice (P < 0.001; Table 1C). However, sulindac treatment did not alter the time of the first adenoma appearance in AOM-induced K-ras^G12D.

No significant differences in the rate of the tumorigenesis were noted between the AOM-induced control genotypes (K-ras wt and Vil-Cre–negative K-ras^G12D) and sulindac completely prevented adenoma formation in these animals. In contrast, 61.54% of sulindac-treated K-ras^G12D mice developed colon adenomas (Table 1C). However, sulindac treatment did significantly decrease the percentage of mice with at least one adenoma (P = 0.04) and tumor burden in AOM-induced K-ras^G12D mice by 83% (P < 0.01) (Table 1A; Fig. 3B), according to OCT image measurements. Similarly to the tumor number, the decrease for tumor burden depended on the genotype of AOM-induced mice (Table 1C, P < 0.0001).

Histologic assessment of tumor number and burden in AOM-induced K-ras^G12D mice in the no sulindac group at the time of sacrifice identified an average of 8 tumors per mouse and a tumor burden of 62.88 mm³ per mouse (Supplementary Table S1A). The measurements of tumor number and burden obtained by imaging were slightly lower than those obtained from histologic evaluation. The percentage of mice with at least one adenoma identified during histologic evaluation on histology was nearly identical to that obtained by OCT imaging (Supplementary Table S1).

An increase in the mucosal thickness of the colon in animals with conditional activation of the mutant K-Ras allele was obvious from OCT imaging. K-ras^G12D mice had a significantly larger mucosal thickness than K-ras wt mice (P < 0.0001; Table 2). Sulindac treatment decreased the mucosal thickness significantly, when the values for all treated versus untreated mice were compared (P < 0.01). The effect of sulindac on thickness depended on genotype (P = 0.01): specifically, in mice of K-ras^G12D genotype sulindac significantly decreased the thickness of mucosa (P = 0.02), whereas in K-ras wt mice, the effect of sulindac was marginal (P = 0.07; Table 2B).

Measurements of mucosal thickness in AOM-induced K-ras wt and K-ras^G12D mice were not practical because of the difficulties in accurate assessment of the mucosal layer thickness when tumors were present.

Histolopathologic analysis of colon tumors in a subset of AOM-induced experimental groups was performed. As the animals for this analysis were selected randomly from three genotypes and groups, the subset is considered a random sampling of all of the animals in the study. Analysis showed that animals expressing mutant K-ras developed all low-grade adenomas (Supplementary Table S2). K-ras^G12Dmice had a significantly higher number of low-grade adenomas (6.39 ± 0.42 adenoma/mouse) compared with control genotypes (K-ras wt, 0.38 ± 0.11 adenoma/mouse P < 0.0001 and Vil-Cre–negative K-ras^G12D, 0.14 ± 0.08 adenoma/mouse, P < 0.0001; Supplementary Table S2). Sulindac treatment decreased the number of low-grade adenomas in K-ras^G12D mice (0.5 ± 0.15 adenoma/mouse) to the level comparable with the number of adenomas found in no sulindac control genotypes (Supplementary Table S2). There was no significant difference between K-ras wt and Vil-Cre–negative K-ras^G12D mice (P = 0.16).

We performed evaluation of the ACF and tumors formed in the subset of 18-week-old animals of K-ras^G12D and K-ras wt genotypes from the control and AOM-induced groups listed in Supplementary Table S3A (8 weeks after the last AOM injection, when applicable). AOM-induced K-ras wt mice developed more ACF than control mice, when fed a regular diet, and sulindac treatment lowered the number of ACF formed in this group (P ≤ 0.02; Supplementary Table S3B; Fig. 4A). The AOM-induced K-ras^G12D mice developed a similar number of ACFs as control K-ras^G12D group, when not treated with sulindac. In addition, the small intestinal and colon adenomas were detected in AOM-induced K-ras^G12D mice (Supplementary Table S3B). Sulindac-treated AOM-induced K-ras^G12D mice had a significantly higher number of ACFs formed, compared with AOM-induced no sulindac K-ras^G12D mice. However, they had less colonic adenomas formed (on average one per mouse) and no small intestinal tumors, in contrast to AOM-induced no sulindac K-ras^G12D mice (Fig. 4A, P = 0.01). No adenomas were detected in K-ras wt mice of the same category (these data were not analyzed due to the small sample size).

IHC analysis of colon tissues from different experimental groups is presented in Supplementary Table S4. Sulindac treatment did not have a significant effect on COX-2 expression (P = 0.10). AOM treatment caused a significant increase in COX-2 expression (P < 0.01). Examples of COX-2 staining in the colon of the AOM-induced Vil-Cre-ER^T2K-ras wt and Vil-Cre-ER^T2K-rasG12D mice are presented in Fig. 4B, COX-2.

Analysis of IHC scores of the apoptosis-related protein CC3 showed that in control groups, sulindac-treated mice of all three genotypes had a significantly higher CC3 expression than untreated mice (P = 0.02). Mice in all AOM-induced groups had a significantly higher number of cells stained for CC3 than mice without AOM (P = 0.02). It is important to note that K-ras wt animals treated with sulindac had a significantly higher number of apoptotic cells, compared with sulindac-treated K-ras^G12D mice (more than 3-fold induction in CC3 staining based on cleaved caspase IHC score in these groups, although these results were not analyzed due to the small sample size). No significant difference was found between sulindac-treated control genotypes (Vil-Cre–negative K-ras^G12D and K-ras wt). Examples of CC3 staining in the colon of K-ras wt and K-ras^G12D mice are shown in Fig. 4B, CC3.

For proliferation marker Ki-67, a significantly higher number of cells positive for Ki-67 protein was observed in K-ras^G12D mice compared with K-ras wt mice (P < 0.001) and Vil-Cre–negative K-ras^G1D (P < 0.01), respectively (Supplementary Table S4, Ki-67). There was no significant difference between K-ras wt and Vil-Cre–negative K-ras^G12D mice. Mice of K-ras^G1D genotype, both control and AOM-induced, had a significantly higher Ki-67 than K-ras wt mice of the same groups (P < 0.001) and Vil-Cre–negative K-ras^G12D mice (P < 0.01). All AOM-induced groups had the higher Ki-67 IHC scores than control groups (P < 0.001). Examples of Ki-67 staining in the colon of the AOM-induced K-ras wt and K-ras^G12D mice are shown in Fig. 4B, Ki-67.

Mortality from hereditary and sporadic colorectal cancers has been significantly reduced through colorectal surveillance with colonoscopy. Colon cancer chemoprevention also has been validated as an important and efficient approach to suppress colon tumorigenesis. Multiple studies have demonstrated that the NSAID sulindac significantly inhibits colorectal polyps in patients with FAP (31, 32), although it was less effective in prevention of duodenal adenomas in these patients (33). The selective COX-2 inhibitor celecoxib showed some efficacy against duodenal and colorectal polyps but is no longer approved for treatment of these conditions due to possible side effects (34, 35). Combination treatments, such as of sulindac and the EGFR inhibitor erlotinib, showed the improved, albeit incomplete, efficacy in prevention of duodenal polyps in the FAP patients after 6-month treatment (36). Oncogenic mutations in the K-RAS gene are found in 32% to 40% of colorectal cancers and are located predominantly in codons 12 and 13 (85%–90%) and in codon 61 (6%; refs. 9, 10). Earlier studies (21, 22) strongly suggest that the presence, even at low frequency, of K-RAS mutations in patients with a high risk for developing of colorectal cancer may diminish the effect of chemoprevention by NSAIDs. A randomized double-blind, placebo-controlled phase III trial of the combination of sulindac with α-difluoromethylornithine, an inhibitor of the downstream effector of Wnt signaling pathway, showed highly significant efficacy in preventing recurrent colorectal polyps (25). However, a direct assessment of K-RAS mutations in colorectal polyps was not performed in this trial.

To directly address the efficacy of Sulindac for prevention of colon adenomas expressing the mutant K-RAS oncogene we utilized a novel mouse model of colon carcinogenesis with the conditional and inducible expression of the mutant K-ras allele in the colonic epithelium (Vil-Cre-ER^T2K-ras^G12D). Our initial analysis of expression of downstream molecular targets of mutant K-ras in K-ras^G12D model confirmed the upregulation of ERK, c-MYC, RalB GTPase and CAV-1 proteins, which are important signaling molecules that promote the proliferative signals of oncogenic K-ras in the colon. Similarly to the reported earlier mouse models with constitutive activation of the mutant K-Ras in the colon (37, 38) we did not observe any tumorigenic effects of the K-ras mutation after activation of mutant K-Ras allele by treatment with 4-OHT. We then introduced the DNA-damaging agent azoxymethane (AOM), which is commonly used to induce neoplastic progression in colon tissues. The chemically-induced mouse colon tumor model recapitulates many histopathological features associated with the multistage progression of human sporadic colorectal cancers. AOM has been successfully used to generate noninvasive in situ adenocarcinomas at the progression stage of carcinogenesis (20 weeks old, 10 weeks after the last AOM injection), specifically within the proximal colon in the mouse with constitutive expression of oncogenic K-ras (39). Our model utilizes the inducible Vil-Cre-ER^T2-driven Cre recombinase, which is expressed in the distal and proximal positions of the colon (Supplementary Fig. S1A), therefore, the adenomas were found throughout the colon.

Minimally invasive OCT quantified the growth of colonic adenomas in this mouse model over time, and traditional histologic and molecular methods were used to analyze the sulindac chemopreventive effect in the presence of K-ras mutation at the final time point (34 weeks of mice age). We have previously shown that OCT imaging is capable of identifying diseased tissue and monitoring the development of lesions individually and accurately in both chemical-induced (AOM-induced) and genetically modified (Apc^Min/+) mouse models (28, 30, 40). We also have demonstrated an advantage of this highly accurate and efficient nondestructive imaging technology to evaluate the effect of chemopreventive agents on colon tumorigenesis in different colon cancer models (30, 41). This study confirmed once more the validity of the OCT method for monitoring of colon tumorigenesis.

Analysis of the number of colon tumors using OCT showed that AOM-induced animals expressing mutant K-RAS protein developed on average 7 times higher number of tumors and had a 5-fold increase in tumor burden than their K-ras wt littermates. Analysis of the OCT images generated at multiple time points showed that sulindac was completely effective in preventing the emergence of colon adenomas after AOM induction in K-ras wt and Vil-Cre–negative K-ras^G12D mice (reduction from 42% incidence in untreated animals to 0% incidence in treated), but was partially effective in treating animals expressing mutant K-ras (reduction from 100% incidence in untreated animals to 62% in treated). At the same time, statistical analyses of tumor number, tumor burden, and the percentage of mice with at least one adenoma, demonstrated a significant efficacy of sulindac treatment against the mutant K-RAS–expressing colon tumors (Table 1C). Sulindac treatment was particularly effective in suppressing tumor burden driven by K-ras mutation (83% reduction, P < 0.01). This finding indicates that sulindac has a potential advantage for colon cancer chemoprevention in patients with colon tissue positive for K-RAS mutation. Because of the signs of ongoing colorectal decease, there may still be additional positive effects.

Image-based analysis of the tumor number and tumor burden (as a function of time (tumor growth rate) was especially valuable in this study as it provided information on colon carcinogenesis during the promotion/progression stages. In particular, it demonstrated that mutant K-ras shortens the promotion stage of AOM-induced colon carcinogenesis (time between the last AOM injection and the detection of the colon adenomas) from 26 weeks in K-ras wt mice to 18 weeks in K-ras^G12D mice. Treatment of K-ras^G12D mice with sulindac did not increase tumor-free duration, but did reduce the number of adenomas detected at the 18-week time point.

OCT imaging also facilitated the analysis of the effects of mutant K-ras on morphologic parameters of the colon tissue, for example, mucosal thickness. In control animals, expression of mutant K-ras caused a significant increase in the thickness of colonic mucosa, and sulindac prevented this increase effectively. Mucosal thickness measurement obtained in vivo with OCT may be more accurate than those measured from histology, because the shrinkage effects of fixation are eliminated, and because the amount of stretch to the colon was standardized by the insertion of the snug-fitting 2-mm diameter endoscope.

Unfortunately, there is currently no method to differentiate low-grade from advanced adenomas with OCT. Histologic evaluation showed that all adenomas in all genotypes were low grade. This finding may be explained by the fact that our mouse model is based on the monoallelic expression of the mutant K-ras, whereas high-grade colon adenomas from clinical studies have a complex mutational profile of multiple mutations (i.e., APC, SMAD, etc.) as has been assessed by the whole-exome sequencing analysis (42).

Prior studies suggest that aberrant crypts in the colon of experimental animals are precursor lesions of the adenoma–carcinoma sequence (43, 44). The ACF analysis performed in 18-week-old mice showed that sulindac treatment significantly increased the number of ACF in AOM-induced K-ras^G12D mice. At the same time point, sulindac-treated AOM-induced K-ras^G12D mice did not develop any adenomas in the small intestine and had a 3-fold decrease in the colon tumor number (Fig. 4B). This suggests that sulindac inhibits the progression of ACFs to adenomas in colon mucosa, which expresses mutant K-RAS protein.

In addition, the IHC analyses of colonic tissues from different treatment groups were done to evaluate the expression of COX-2 (inflammation marker involved in the mechanism of sulindac antitumor activity), CC3 (apoptosis marker), and Ki-67 (proliferation marker).

On the basis of IHC scoring, COX-2 expression (inflammation marker involved in the mechanism of sulindac antitumor activity) was elevated in all AOM-treated animals, compared with control animals of all groups (P < 0.01). IHC scoring was not significantly altered by mouse genotype or sulindac treatment. The mechanism responsible for the antitumor activity of sulindac metabolites has been studied extensively and a number of COX-dependent, and independent targets have been identified (45). We reported previously that sulindac sulfide metabolite suppressed prostaglandin synthase E2 (PGE2) production in colon cancer cells expressing mutant K-RAS while it increased COX-2 protein level in this model, but sulindac sulfone metabolite inhibited both PGE2 levels and COX-2 expression in treated cancer cells (46). Previous studies have concluded that inhibition of COX-2 enzyme activity is required for antitumor activity of NSAIDs, although other mechanisms have been reported (45).

High rates of proliferation and apoptosis were noted in AOM-treated groups compared with control groups (P < 0.001 for Ki-67, P = 0.02 for CC3). The Ki-67 scores were not significantly affected by sulindac treatment in these groups. We found an increase in the apoptosis marker CC3 staining in sulindac-treated mice of all genotypes in both control and AOM-treated groups (P = 0.02). The lack of significant changes in Ki-67 and CC3 expression in sulindac-treated mice in the current study may be due to the relatively low dose of sulindac used (100 ppm) or could suggest a resistance to apoptosis in the tested mouse model.

Our current study is significant as it demonstrates for the first time using in vivo OCT imaging that mutant K-RAS expression results in faster development of the colonic adenomas, and sulindac chemoprevention does not inhibit the onset of adenomas carrying a K-RAS mutation.

Our findings underscore the importance of targeted and tailored therapy aimed to suppress mutant K-RAS activity in individuals with high risk of developing cancer.

No potential conflicts of interest were disclosed.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Conception and design: J.K. Barton, N.A. Ignatenko

Development of methodology: R.B. Nagle, J.K. Barton

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P.F.S. Rice, K.G. Ehrichs, M.S. Jones, R.B. Nagle, D.G. Besselsen, J.K. Barton, H. Chen, E.R. Abril

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P.F.S. Rice, K.G. Ehrichs, N.A. Ignatenko, C.-H. Hsu, J.K. Barton

Writing, review, and/or revision of the manuscript: P.F.S. Rice, C.-H. Hsu, D.G. Besselsen, J.K. Barton, N.A. Ignatenko

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.F.S. Rice

Study supervision: P.F.S. Rice, J.K. Barton

We would like to thank Dr. Sylvia Robin, Institute Curie-CNRS, Paris, France, for providing us with the inducible Vil-Cre ER^T2 strain.

This work was supported by NIH/NCIR01CA157595 (to N.A. Ignatenko), NIH/NCIR01CA109835 (to J.K. Barton), Research reported in this publication was supported by the NCI of the NIH under award numbers R01CA109835 and R01CA157595. Research in this article was directly supported by the Experimental Mouse and Tissue Acquisition and Molecular Analysis Shared Resources funded by the NCI Award P30CA023074 (to A.E. Kraft, Cancer Center Support grant).

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.