Pharmacokinetics in mice and growth-inhibitory properties of the putative cancer chemopreventive agent resveratrol and the synthetic analogue trans 3,4,5,4′-tetramethoxystilbene

Resveratrol (trans-3,5,4′-trihydroxystilbene) is a naturally occurring polyphenol with cancer chemopreventive properties in preclinical models of carcinogenesis, including those of colorectal cancer. Recently, a variety of analogues of resveratrol have been synthesised and investigated in in vitro assays. One analogue, 3,4,5,4′-tetramethoxystilbene (DMU 212), showed preferential growth-inhibitory and proapoptotic properties in transformed cells, when compared with their untransformed counterparts. As part of a chemoprevention drug development programme, the pharmacokinetic properties of DMU 212 were compared with those of resveratrol in the plasma, liver, kidney, lung, heart, brain and small intestinal and colonic mucosa of mice. DMU 212 or resveratrol (240 mg kg−1) were administered intragastrically, and drug concentrations were measured by HPLC. Metabolites were characterised by cochromatography with authentic reference compounds and were identified by mass spectrometry. The ratios of area of plasma or tissue concentration vs time curves of resveratrol over DMU 212 (AUCres/AUCDMU212) for the plasma, liver, small intestinal and colonic mucosa were 3.5, 5, 0.1 and 0.15, respectively. Thus, resveratrol afforded significantly higher levels than DMU 212 in the plasma and liver, while DMU 212 exhibited superior availability compared to resveratrol in the small intestine and colon. Resveratrol was metabolised to its sulphate or glucuronate conjugates, while DMU 212 underwent metabolic hydroxylation or single and double O-demethylation. DMU 212 and resveratrol inhibited the growth of human-derived colon cancer cells HCA-7 and HT-29 in vitro with IC50 values of between 6 and 26 μM. In the light of the superior levels achieved in the gastrointestinal tract after the administration of DMU 212, when compared to resveratrol, the results provide a good rationale to evaluate DMU 212 as a colorectal cancer chemopreventive agent.

Resveratrol (trans 3,5,4 0 -trihydroxystilbene, for structure see Figure 1) is a phytoalexin generated in response to environmental stress or pathogenic attack in grapes, mulberries, cranberries, peanuts and plants of the Cassia quinquangulata family. Resveratrol inhibits diverse cellular events associated with the three major stages of carcinogenesis: initiation, promotion and progression (Jang et al, 1997). Since this discovery, resveratrol has been the subject of a large number of preclinical and mechanistic studies. The cancer chemopreventive potential of resveratrol has been demonstrated in the models of carcinogenesis in vivo and in cells in vitro. It inhibits the proliferation of a variety of cancer cell lines (for a review, see Gusman et al, 2001), formation of preneoplastic lesions in the 7,12-dimethylbenz(a)anthracene-(DMBA) induced mouse mammary organ culture model (Bhat et al, 2001) and benzo(a)pyrene-induced transformation of rat tracheal epithelial cells (Jang et al, 1997). In animal studies, resveratrol interfered with the formation of azoxymethane-(AOM) induced aberrant crypt foci in rat colon (Tessitore et al, 2000), attenuated oesophageal carcinoma formation in rats that received N-nitrosomethylbenzylamine (NMBA) , decreased the number of adenomas in the small intestine and suppressed tumour formation in the colon of Apc Min þ mice (Schneider et al, 2001) and reduced mammary tumour formation in N-methyl-Nnitrosourea-(NMU) treated rats (Bhat et al, 2001). Resveratrol has been reported to possess a variety of anti-inflammatory, antiplatelet and both pro-and antioestrogenic effects (Bertelli et al, 1996;Gehm et al, 1997;Jang et al, 1997;Uenobe et al, 1997). It exerts a wide variety of biological effects germane to cancer chemoprevention, including the inhibition of cytochrome P450 enzyme expression activity (Chun et al, 1999;Guengerich et al, 2003), induction of apoptosis (Mahyar-Roemer et al, 2001, modulation of components of the cell cycle machinery (Schneider et al, 2000;Wolter et al, 2001), decrease in cyclooxygenase 1 (COX-1) activity and COX-2 expression (Subbaramaiah et al, 1998;Mutoh et al, 2000;Li et al, 2002), antioxidation (Sgambato et al, 2001), inhibition of activities of protein kinase C and D (Haworth Revised 17 October 2003;accepted 17 November 2003 andAvkiran, 2001;Slater et al, 2003) and decrease in the activity of transcription factors NFkB and AP-1 (Surh et al, 2001;Banerjee et al, 2002). As is the case with many polyphenols with putative cancer chemopreventive properties, the systemic bioavailability of resveratrol is probably poor. This notion is borne out by studies in mice, rats and dogs, which suggest consistently that resveratrol is well absorbed but avidly glucuronidated and sulphated both in the liver and in intestinal epithelial cells (Asensi et al, 2002;Juan et al, 2002;Marier et al, 2002). One study in humans also hints at a poor bioavailability of resveratrol (Goldberg et al, 2003). In the wake of the discovery of the interesting pharmacological properties of resveratrol, the trihydroxystilbene scaffold has become the subject of imaginative synthetic manipulations by medicinal chemists with the aim of generating novel congeners of pharmacological interest and to characterise structural features, which impart activity to the molecule. These structural alterations have been aimed at the optimisation of the cytochrome P450 enzyme-inhibitory and antimutagenic potencies of the molecule Kim et al, 2002), its antioxidant activity (Lu et al, 2001), its apoptosisinducing and growth-inhibitory activity (Lu et al, 2001;Nam et al, 2001;Kim et al, 2002) and its ability to inhibit cell transformation (She et al, 2003). These chemical synthetic attempts have predominantly been concerned with the introduction of additional hydroxy moieties into the trihydroxystilbene framework and with various degrees of methylation of the phenol groups. An especially auspicious finding concerning these analogues is the fact that 3,4,5,4 0 -tetrahydroxystilbene, resveratrol with an additional hydroxy moiety, and its O-methylated congener, 3,4,5,4 0 -tetramethoxystilbene (DMU 212, for structure see Figure 1), were capable of preferentially interfering with the proliferation and survival of transformed human lung-derived cells, with much lower growthinhibitory and apoptotic properties in their untransformed counterparts (Lu et al, 2001). In contrast, resveratrol did not possess this discriminatory potential. Furthermore, DMU 212 is currently under preclinical evaluation as a potential antitumour prodrug that undergoes metabolic activation by certain cytochrome P450 enzymes (Potter et al, 2002a). In the light of the availability of pharmacologically interesting stilbene analogues, it seems appropriate to compare the cancer chemopreventive potential of resveratrol in preclinical models with those of its congeners, which have been shown to possess increased potency in relevant in vitro assays. Before embarking on such efficacy studies, it is desirable to find out if resveratrol congeners, to be evaluated as potential chemopreventive agents, possess adequate bioavailability in the tissues in which malignancies are to be prevented. Such pharmacokinetic exploration should be an essential part of the chemopreventive drug discovery process. Mindful of these considerations, we chose DMU 212, one of the most interesting resveratrol analogues described thus far (Lu et al, 2001;Potter et al, 2002a), and compared its levels in murine tissues after oral administration with those of resveratrol. Thus, the hypothesis was tested that a replacement of the phenol functionalities in resveratrol by methoxy moieties and an addition of a further methoxy group impinge on the pharmacokinetic properties of the parent molecule. Additionally, DMU 212 was compared with resveratrol in terms of their metabolism in the mouse in vivo and in liver homogenate preparations in vitro. The mouse is the animal species frequently used in cancer chemoprevention efficacy studies. Finally, we compared the ability of DMU 212 and resveratrol to interfere with the growth of human-derived colon cancer cells. Overall, the study was designed to help rationalise the choice of resveratrol analogues for further testing for potential usefulness as cancer chemopreventive agents.

Treatments
Animal experiments were conducted as stipulated by Project Licence 40/2496 granted by the UK Home Office. Experiments were vetted and approved by the Leicester University Animal Welfare Committee and complied with the UKCCCR guidelines for the welfare of animals in experimental neoplasia. Mice aged 8 weeks (20 -22 g) received resveratrol or DMU 212 (240 mg kg À1 body weight, equivalent to 1 mmol kg À1 resveratrol or 0.8 mmol kg À1 DMU 212) via the intragastric route (three animals per time point). The vehicle was glycerol formal, and the dose volume approximately 10 ml kg À1 . In the case of DMU 212, the vehicle also contained 10% DMSO. Mice were killed by terminal gaseous anaesthesia at 10, 30, 60 or 120 min postdosing. Blood was collected by cardiac puncture, and plasma was obtained by centrifugation. The liver, kidney, lung, heart, brain and gut were excised, and scrapings were obtained from small intestine and colon. Tissues were snap-frozen (liquid nitrogen) and stored at À801C until analysis.

Incubation with liver microsomes
Microsomes were prepared by differential centrifugation of mouse liver homogenate first at 9 Â 10 3 g (20 min, 41C), then at 10 5 g (1 h) in a Beckman L-8-60 ultracentrifuge (Beckman Coulter UK Ltd, High Wycombe, Buckinghamshire, UK). The microsomal pellet was suspended in Tris buffer (50 mM, pH 7.4), recentrifuged at 10 5 g (1 h) and resuspended in Tris buffer. Liver microsomes from two to four mice were pooled. Microsomes (0.5 mg protein ml À1 ) were incubated at 371C with NADPH (1 mM), MgCl 2 (1 mM) and resveratrol or DMU 212 (1 mM) for 20 min (final volume: 0.2 ml). The addition of one volume of ice-cold methanol terminated the reaction. The mixture was vortexed (30 s), centrifuged (3 min, 13 400 g) and the supernatant was collected and analysed by HPLC. For the biosynthesis of resveratrol glucuronide for use as a reference compound, microsomes were incubated with resveratrol (1 mM) as described above, except that NADPH was replaced by uridine-diphosphoglucuronic acid (1 mM).

Extraction of agents from plasma and tissues
Tissues were homogenised (1 : 1 volume to tissue mass ratio) in 50 mM Hepes buffer (Sigma, Poole, UK) using a hand-held glass homogeniser. An aliquot (250 ml) of homogenate, to which an internal standard had been added was vortexed, followed by the addition of acetonitrile (1 ml). After vigorous shaking, mixtures were kept on ice (5 -10 min) and centrifuged (2800 g, 10 min, 41C). The supernatants were dried under a stream of nitrogen and reconstituted in mobile phase (100 ml).
Internal standards were carbamazepine and 4 0 -methoxy-4methyl-trans-stilbene for resveratrol and DMU 212, respectively. Resveratrol and its metabolites was analysed using UV detection (325 nm), and DMU 212 and its metabolites were determined with fluorescence detection (335 nm excitation, 395 nm emission). The injection volume of samples reconstituted in the mobile phase was 50 ml. The preliminary characterisation of metabolites was achieved by cochromatography using either synthetic standards or biosynthetically generated resveratrol glucuronide. Quantification of resveratrol and DMU 212 was performed using standard curves constructed with relevant drug concentrations. The curves were characterised by regression coefficients of R 2 ¼ 0.99 or above. The extraction efficiencies (in percent) for resveratrol and DMU 212 were as follows: from plasma 102717 and 6772, respectively, from tissues 8677 and 78715, respectively (mean7s.d., n ¼ 5 -7). The recovery of the resveratrol conjugates was approximately 50% that of the DMU 212 metabolites 60%. Therefore, metabolites were not quantitated.
The area under the plasma or tissue concentration vs time curve (AUC) values were calculated using the trapezoidal rule (Win-Nonlin v.1.1, Scientific Consultants, USA).

Mass spectrometry
Mass spectrometry was performed using a Quattro Bio-Q tandem quadrupole mass spectrometer upgraded to Quattro MK II specifications (Micromass, Manchester, UK) with a pneumatically assisted electrospray interface. Samples were analysed in positive ion mode. The temperature was maintained at 1201C, and the operating voltage of the electrospray capillary was 3.88 kV and the cone voltage was 32 V. HPLC conditions used for the on-line HPLC-mass spectrometric analyses were as described for DMU 212 above.

Effect of agents on growth of HT-29 and HCA-7 cells
Cells were seeded (10 4 per well) in 24-well plates and cultured in DMEM containing Glutamax I (Life Technologies, Paisley, UK) and glucose (4.5 g l À1 ) and 10% foetal calf serum (Gibco, Paisley, UK). Resveratrol or DMU 212 (1 -100 mM) dissolved in DMSO was added to cellular incubates 24 h postplating. Cells were counted 72, 96, 120, 144 and 168 h postaddition of agents using a Z2 Coulter Particle Count and Size Analyzer (Beckman Coulter, High Wycombe, UK). Control cultures contained the vehicle only. The amount of DMSO added to the incubate did not exceed 0.1%, which on its own failed to affect the cell growth. The IC 50 values inserted in Figure 5 were calculated from the linear portion of the cell number-vs agent concentration curves at the 168 h time point.

Plasma and tissue levels of resveratrol and DMU 212
Mice received intragastric resveratrol or DMU 212 (1 or 0.8 mmol kg À1 ), and drug levels were measured in the plasma and liver, kidney, lung, brain, small intestinal mucosa and colonic mucosa. These are all tissues in which resveratrol might prevent malignancy, or delay its onset. For comparison, levels in the heart were also studied. The results are shown in Figure 3. The plasma and tissue level data were used to calculate the respective mean AUC 0 -120 min values (Table 1). Both agents were rapidly cleared from blood and tissues within an hour of administration (Figure 3). Concentrations of resveratrol were consistently higher than those of DMU 212 in the plasma, liver and heart. Levels were similar in the kidney and lung, while resveratrol concentrations were much below those of DMU 212 in the brain, small intestinal and colonic mucosae. The most dramatic discrepancy in levels occurred in the liver, in which the AUC for resveratrol was five times higher than that for DMU 212, and in the small intestinal and colonic mucosae, in which the AUCs for DMU 212 exceeded those for resveratrol by factors of 10 and 7, respectively (Table 1). Resveratrol levels peaked after 10 min, the first of the time points chosen, and peak concentrations were 32 mM in plasma and 51, 16, 50, 1.2, 75, 960 and 30 nmols g À1 tissue in the liver, kidney, lung, brain, heart, small intestinal and colonic mucosae, respectively. DMU 212 levels reached a peak after 10 min, except in the case of the colon, in which the peak occurred after 30 min. The peak levels for DMU 212 were 5 mM in the plasma and 8, 11, 11, 5, 10, 7600 and 330 nmols g À1 tissue in the liver, kidney, lung, brain, heart, small intestinal and colonic mucosae, respectively. We performed a preliminary experiment in order to explore whether the difference in gastrointestinal levels would also be observed when the agents are administered with the diet, the route often used in preclinical cancer chemoprevention studies. Agents were admixed with the diet at 0.2% and fed for 2 weeks. Colonic levels of DMU 212 after DMU 212 administration exceeded those of resveratrol after resveratrol administration by two-fold (result not shown).
For confirmatory purposes, the metabolism of resveratrol and DMU 212 was also studied in vitro in suspensions of liver microsomes fortified with cofactors of cytochrome P450 enzymes. While HPLC analysis of extracts of the incubation mixture with resveratrol did not yield any peak in addition to that of the parent substrate, analysis of incubates with DMU 212 afforded peaks in addition to that of parent DMU 212 ( Figure 4C). In analogy to the analysis of liver tissue from mice that had received DMU 212, the incubation mixture was extracted and the extract was subjected to cochromatographic analysis. The results ( Figure 4C) suggest that 4,4 0 -di-desmethyl-DMU 212, 4 0 -desmethyl-DMU 212, 3 0hydroxy-DMU 212, 4-desmethyl-DMU 212 and 3-desmethyl-DMU 212 were products of the microsomal biotransformation of DMU 212.

Effect of resveratrol and DMU 212 on the growth of colon cancer cells
The growth-modulating ability of resveratrol and DMU 212 were investigated in a preliminary manner in HT-29 and HCA-7 colon cancer cells in vitro. The IC 50 values computed form the growth curves ( Figure 5) range from 6 to 26 mM, and DMU 212 appears to be a slightly more potent growth inhibitor than resveratrol.

DISCUSSION
The results presented above suggest that the introduction of four methoxy groups into the stilbene framework, three of which replaced the hydroxy moieties in resveratrol, fails to increase the systemic availability of the molecule in comparison to resveratrol. This conclusion is borne out by the comparison between resveratrol and DMU 212 in terms of the concentrations of these agents in the plasma, liver and heart after oral administration, in which the availability of DMU 212 was inferior to that of resveratrol. In contrast, in comparison to resveratrol, DMU 212 was found to be more available in intestinal and colonic mucosae and in the brain. A priori, it is difficult to predict on theoretical grounds in what manner such a structural modification might alter the pharmacokinetic profile of the stilbene molecule. One of the physicochemical corollaries of this alteration, which undoubtedly impacts on the pharmacokinetics of the molecule, is the increase in Table 2 Features of daughter ion spectra obtained by tandem mass spectrometry of molecular ions (M+H) + of DMU 212 and four metabolites isolated from peaks in high-performance liquid chromatograms of extracts of livers of micethat had received DMU 212 (240 mg kg À1 ) and of peaks of authentic reference compounds  (100) 198 (50)  181 (62) 181 (81) 3-Desmethyl-DMU 212 287 (M+H) + (10) 287 (M+H) + (75) 227 (72) 227 (20)  195 (70) 195 (18) 4 0 -Desmethyl-DMU 212 287 (M+H) + (10) 287 (M+H) + (60) 256 (100) 256 (100)  272 (80) 272 (47) 3 0 -Hydroxy-DMU 212 317 (M+H) + (31) 317 (M+H) + (21) 225 (55) 225 (100)  286 (22) 286 (10) Only three prominent product ions are shown. Relative abundance in percentage is shown in parentheses. lipophilicity conferred on the stilbene species in DMU 212 by four methoxy functionalities vis-á-vis resveratrol, which has three hydroxy groups instead. The difference in lipophilicity between DMU 212 and resveratrol is reflected by the fact that reversed phase HPLC analysis of a mixture of both agents using the gradient system described under Materials and Methods for DMU 212 afforded a retention time of 17 min for DMU 212, while resveratrol eluted with the solvent front. Another factor, which undoubtedly determines the differential pharmacokinetic properties of resveratrol and DMU 212, is the divergence in their metabolic profile. Resveratrol is known to undergo metabolic phase II reactions involving conjugation with sulphate and glucuronic acid (Yu et al, 2002). Consistent with this report, we found a resveratrol sulphate and a resveratrol glucuronide in the liver and other tissues of mice that had received resveratrol. In contrast to resveratrol, DMU 212 was subjected to hepatic metabolic oxidation, especially single or double O-demethylation reactions in the 3, 4 or 4 0 positions of the molecule. In addition, we could identify a hydroxylated metabolic species (3 0 -hydroxy-DMU 212, DMU 214). The spectrum of Odemethylated and hydroxylated metabolites found in vivo was accurately reflected by the metabolic profile obtained on incubation of DMU 212 with NADPH-fortified liver microsomes. These findings are consistent with the results of recent in vitro experiments using cytochrome P450 isoenzyme preparations, in which DMU 212 was found to undergo both aromatic hydroxylation and O-demethylation reactions primarily catalysed by isoenzymes of the CYP1 family (Wilsher et al, unpublished). In analogy, resveratrol was recently found to undergo metabolic oxidation in vitro to piceatannol (3,5,2 0 ,4 0 -tetra-hydroxystilbene), when incubated with a source of CYP1B1 (Potter et al, 2002b). In the study described here, piceatannol was not identified as a metabolite of resveratrol in mice in vivo or in mouse liver microsomes in vitro. It is conceivable that piceatannol was present at very low levels, which might have confounded detection by the procedures used here. The ability to convert resveratrol to piceatannol has been surmised to be a relatively specific property of CYP1B1 (Potter et al, 2002b). The lack of the presence of piceatannol at detectable concentrations in our study in the mouse is consistent with the notion that CYP1B1 is not expressed at appreciable levels in mouse liver . Overall, the result described here suggests that hydroxylation to piceatannol is probably not a major metabolic route for resveratrol in the mouse.
It is conceivable that 4,4 0 -di-desmethyl-DMU 212, 4 0 -desmethyl-DMU 212, 3 0 -hydroxy-DMU 212 and 3-desmethyl-DMU 212, the DMU 212 metabolites identified here, undergo phase II metabolism to yield sulphate and/or glucuronide metabolites. However, conjugates of metabolic species derived from DMU 212 were not found in the mouse. We cannot exclude that such conjugates were formed in vivo but not detected. Nevertheless, there were no peaks indicative of such conjugates, characterised by short retention times, which could have been subjected to mass spectral investigation. It has to be stressed that the metabolism studies described here did not allow quantification of the metabolic species identified, so that it is not possible to adjudge the difference between resveratrol and DMU 212 in terms of the contribution of hepatic metabolism to the overall systemic clearance of the molecules.
The pharmacokinetic properties of DMU 212 have thus far been unknown. In contrast, resveratrol has been the subject of several pharmacokinetic studies, including those in the rat (Soleas et al, 2001) and the mouse (Vitrac et al, 2003), in which resveratrol has been administered via the oral route. In terms of peak levels and systemic disappearance, the results outlined above in mice are consistent with the published studies. The increased drug levels in the liver, kidney, lung and heart obtained after ingestion of resveratrol in comparison to those after DMU 212 reflect the difference in availability observed in the plasma. In contrast, the levels of DMU 212 in the brain, small intestinal and colonic mucosae after DMU 212 administration exceeded levels of resveratrol measured after resveratrol intake. The higher availability of DMU 212 in the brain suggests that it is capable of crossing the blood -brain barrier more easily than resveratrol, which is probably a consequence of the higher lipophilicity of DMU 212. The discrepancy between resveratrol and DMU 212 in concentration achieved in the small intestine and colon may be a corollary of their differential metabolic susceptibilities. The lower availability of resveratrol compared to DMU 212 may be the consequence of the high propensity of resveratrol to undergo conjugation reactions catalysed by enzymes (De Santi et al, 2000;Kuhnle et al, 2000;Marier et al, 2002), which are abundantly present in the gut (Eisenhofer et al, 1999). In contrast, on the basis of the fact that activities of oxidising enzymes in the gut are much lower than in the liver (Doherty and Charman, 2002), the ability of the gut mucosa to O-demethylate or hydroxylate DMU 212 is arguably much lower than its ability to biotransform resveratrol by conjugation. On the speculative assumption that DMU 212 and resveratrol share biochemical mechanisms germane to cancer chemoprevention and that DMU 212 is not inferior to resveratrol in intrinsic potency, this observation hints at a potential advantage of DMU 212 over resveratrol when applied as an experimental colorectal cancer chemopreventive agent. The superior growthinhibitory and apoptotic potency of DMU 212 in comparison to resveratrol in transformed human lung-derived cells (Lu et al, 2001) is consistent with this hypothesis. Peak levels of DMU 212 that were achieved in the colonic and small intestinal mucosae in mice exceeded those required to cause significant arrest of transformed lung cell growth in vitro (10 mM) by factors of 32 and 760, respectively. Furthermore, the results of our preliminary assessment of the colon cancer cell growth-inhibitory properties of resveratrol and DMU 212 suggest that the concentrations of either agent required to inhibit growth, with IC 50 values between 6 and 26 mM, are comfortably within the range achieved in the colorectal tract of mice that received these agents at the oral doses used in the study described here, 240 mg kg À1 .
In the light of the suggestion that resveratrol at 0.01% in the drinking water (constituting a dose of B15 mg kg À1 per day) decreased adenoma multiplicity in the Apc Min/ þ mouse model (Schneider et al, 2001), the results described here render a comparison of resveratrol and DMU 212 in this model appropriate, and such experiments are currently planned in this laboratory. However, it is important to add a note of caution, because the chemopreventive efficacy of resveratrol in this model is highly contentious, as borne out by two contradictory abstracts published subsequent to the paper by Schneider et al (2001). These abstracts suggest that in the same murine model, dietary doses comparable to, or much higher than, those used by Schneider et al were completely ineffective (Ziegler et al, 2001) or, in the case of a dietary daily dose of 500 mg kg À1 for 14 days, reduced adenoma load, but did so only in female mice and not at all in male mice (Gignac and Bourquin, 2001).
In conclusion, the work described here provides an initial pharmacokinetic groundwork, which can contribute to rational decision making as to the choice of resveratrol analogues that should be selected for comparative testing for cancer chemopreventive potency in preclinical models. DMU 212 showed more favourable pharmacokinetic properties than resveratrol, in that it yielded higher levels of drug in the small intestinal, colonic mucosae and brain. Buttressed by our recent finding that DMU 212 is devoid of any toxicity in rats when administered at single doses of up to 40 mg kg À1 via the i.v. route or up to 400 mg kg À1 when administered p.o. (Verschoyle et al, unpublished), the results presented here render the exploration of DMU 212 side by side with resveratrol for chemopreventive efficacy in rodent models of colorectal carcinogenesis propitious.