Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cell Cycle. Author manuscript; available in PMC Jan 28, 2013.
Published in final edited form as:
Published online May 15, 2010.
PMCID: PMC3557519
Dietary n-3 polyunsaturated fatty acids fail to reduce prostate tumorigenesis in the PB-ErbB-2 x Pten+/− preclinical mouse model
Sarada Vissapragada,1 Anup Ghosh,1 Lymor Ringer,1 Patricia Salinas,1 Amanda Brophy,1 Daniel Peaceman,1 Bhaskar Kallakury,1,2 Partha P. Banerjee,3 Stanley T. Fricke,4 William Helfrich,5 Yi Chien Lee,1 Richard Pestell,6 Philipp Scherer,7 Herbert B. Tanowitz,8 Maria Laura Avantaggiati,1 Leena Hilakivi-Clarke,1 Michael P. Lisanti,6 Olga C. Rodriguez,1* and Chris Albanese1,2*
1Lombardi Comprehensive Cancer Center and Department of Oncology; Georgetown University Medical Center; Washington, DC USA
2Department of Pathology; Georgetown University Medical Center; Washington, DC USA
3Department of Biochemistry and Molecular & Cellular Biology; Georgetown University Medical Center; Washington, DC USA
4Children’s National Medical Center; Washington, DC USA
5University of Illinois; Champaign-Urbana, IL USA
6Kimmel Cancer Center; Thomas Jefferson University; Philadelphia, PA USA
7UT Southwestern Medical Center; Dallas, TX USA
8Department of Pathology; Albert Einstein College of Medicine; Bronx, NY USA
*Correspondence to: Olga C. Rodriguez and Chris Albanese; rodriguo/at/ and ; albanese/at/
These authors contributed equally to this work.
Diet and obesity, and their associated metabolic alterations, are some of the fastest-growing causes of disease and death in America. Findings from epidemiological studies correlating obesity, the sources of dietary fat and prostate cancer (PCa) are conflicting. We have previously shown that 15% of PB-ErbB-2 x pten+/− mice developed PCa and exhibited increased phosphorylated 4E-BP1, but not the key PI3-kinase intermediary phospho-protein, mTOR, when maintained on unrefined mouse chow.
We report herein that 100% of animals fed refined, westernized AIN-93-based diets containing corn oil developed PCa by 12 months of age. Increases in visceral fat and mTOR activation in the tumors were also observed. Furthermore, nuclear cyclin E levels were significantly induced by the AIN-93-corn oil-based diets versus chow. Replacing 50% of the corn oil with menhaden oil, with 21% of its triglycerides being n-3 PUFA’s, had no effect on tumorigenesis, fat deposition, cyclin E or mTOR. Phosphorylated BAD levels were similar in the tumors of mice in all three diets.
Our data demonstrated that in the context of our preclinical model, components of crude chow, but not dietary n-3 PUFAs, protect against PCa progression. In addition, these data establish phosphorylated mTOR, nuclear cyclin E and visceral fat deposits as possible biomarkers of increased dietary risk for PCa.
Keywords: ErbB-2, cyclin E, Pten, polyunsaturated fatty acid, mTOR, prostate, MRI
Epidemiological studies correlating the influence of diet, and body mass with prostate cancer (PCa) have been at times conflicting, however once prostate cancers arise, increased body mass may tend to be associated with more aggressive cancers and less favorable clinical outcomes.14 Correlations may also exist between polyunsaturated fatty acid (PUFA) ratios and PCa. Clinical data based on serum profiles of n-6 vs. n-3 PUFA’s in prostate cancer patients suggested that n-6 PUFA’s stimulated while n-3 PUFA’s inhibited prostate cancer growth.5,6 The n-6 PUFA linoleic acid induces proliferation of transformed PCa cells in vitro through arachidonic acid (AA), cyclooxygenease 2 (COX2) and proinflammatory/mitogenic cytokines, such as prostaglandin E. Conversely, n-3 PUFAs eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) have been shown to inhibit AA and COX2, leading to decreased PCa cell proliferation. In mice homozygously deleted of pten in the prostate epithelium, dietary supplementation with purified DHA and EPA was reported to suppress tumorigenesis, in part through reducing the phosphorylation-mediated inhibition of the pro-apoptotic protein, BAD (p-BAD).7 Conversely, prospective studies in humans found no appreciable protective effects of fish consumption in preventing PCa, however there was a possible link to increased survival.8
Our group9 and others have previously shown that levels of the proto-oncogene ErbB-2 are increased in human PCa. We established that a probasin-driven ErbB-2 transgene mice model with heterozygous loss of pten mouse model (PB-ErbB-2 x Pten+/−) developed prostate adenocarcinomas at approximately 15% penetrance, while still retaining Pten expression.10 Mechanistically, we demonstrated that while activation of PDK1 and p70S6K and an inactivation of the eIF4E-binding protein-1 occurred in the PCa lesions, the phosphorylation-dependent activation of the key signaling mTOR (p-mTOR) was not required.10
To investigate the possible effects of altering dietary n-3 to n-6 PUFA ratios on prostate tumorigenesis, PB-ErbB-2 x Pten+/− mice were either kept on normal rodent chow as previously described, or placed on AIN93G-based purified diets with defined ratios of n-3 and n-6 PUFAs. We present evidence herein that mice fed refined AIN93-based diets exhibited enhanced prostate adenocarcinoma formation with concomitant induction of nuclear cyclin E levels compared to the chow diet used in the previous studies. Additionally, levels of visceral fat deposition, as measured by MRI, and p-mTOR levels were induced in the PCa lesions of mice fed the refined diets, and an increased dietary n-3/n-6 PUFA ratio failed to suppress prostate tumorigenesis. Our studies indicate that a shift from chow to the more westernized, refined AIN93-corn oil diets increased visceral fat deposition, cyclin E and p-mTOR levels as well as PCa incidences and increased availability of dietary n-3 PUFA’s failed to affect any of these parameters. These data raise the possibility that cyclin E abundance, p-mTOR status and visceral fat content may represent biomarkers of increased dietary influence on PCa progression, especially in a background of increased disease risk.
A shift from laboratory chow to the refined diets increased PCa formation
We have previously established that the PCa incidence rate of PB-ErbB-2 x pten+/− mice fed chow is 15% by 12–14 months of age.10 In order to investigate the possible effect of different fat sources on prostate cancer formation, male mice were separated at weaning and randomly assigned either chow (Diet 20) or one of two AIN93G-based purified diets, AIN93-corn oil (AIN93-CO) or a diet containing equivalent amounts of corn oil and fish oil (AIN93-CO:FO) (Table 1). The mice remained on the assigned diet until the termination of the study at one year of age. The food was provided fresh each week and consumption was measured weekly. The mice were also weighed throughout the course of the study, and their weights plotted in each dietary group. MRI for fat deposition was performed at the termination of the study.
Table 1
Table 1
Diet nutritional information
While the AIN93-based diets contained slightly more energy per gram than chow (3.8 kcal/g vs. 3.4 kcal/g, respectively), neither growth rates nor final body weights differed significantly (p ≥ 0.05) between diets (Fig. 1A), despite the observation that chow-fed mice consumed significantly (p ≤ 0.01) more food per week (Fig. 1B). In addition, MRI performed on animals at the termination of the study revealed that the AIN93-fed mice showed increased levels of visceral fat vs. chow (Fig. 1C), which was significant at p ≤ 0.05 on the AIN93-CO:FO diet. The visceral fat content in the AIN93-CO mice (50.7%), while slightly higher that on AIN93-CO:FO (49.8%), failed to reach statistical significance due to a larger standard deviation (0.9% versus 0.3%, respectively).
Figure 1
Figure 1
Influences of diets on animal grown and weight. PB-ErbB-2 x pten+/− mice were randomly placed at weaning on one of three diets, CO, AIN93-corn oil, CO:FO, AIN93-CO:FO or Chow (Diet 20). The effect of diet on (A) growth rate, (B) food consumption (more ...)
We have previously established that the PCa incidence rate of PB-ErbB-2 x pten+/− mice fed chow was 15% by 12–14 months of age.10 At 12 months of age the mice fed the refined diets were sacrificed and comprehensive pathological and immunohistochemical analyses were performed. Mice fed either AIN93-CO or AIN93-CO:FO diets both exhibited PCa incidence rates of 100% and representative examples are shown (Fig. 2A and B). Lower magnification images are shown in Supplemental Figure S1.
Figure 2
Figure 2
Dietary influences on PCa in PB-ErbB-2 x Pten+/− mice. At the end of the study, the urogenital tracts were fixed, embedded in paraffin and sectioned for analysis. Top row, mice fed AIN93-CO. Bottom row (B, D, F and H), mice fed AIN93-CO:FO. Bottom (more ...)
The increase in PCa observed in mice fed the refined diets was accompanied by increased nuclear cyclin E
We have previously reported that approximately 30% of the epithelial cells in PCa lesions of mice on chow stain nuclear-positive for cyclin D1.10 We therefore assessed the frequency of nuclear cyclin D1 positivity in the PCa lesions of mice on the AIN93-based diets. The prevalence of cyclin D1 nuclear-positive cells was decreased in mice fed AIN-93-CO:FO, with approximately 22% (±9.5, N = 3) of the cells positive for nuclear cyclin D1 (Fig. 2D), which failed to reach statistical significance. In contrast, the prostate cancers of mice fed AIN-93-CO had approximately 18% (±7.2, N = 3) of the cells positive for nuclear cyclin D1 (Fig. 2C), a statistically significant (p ≤ 0.05) decrease versus chow. Since the decrease in cyclin D1 activity suggested that the increased tumorigenesis observed may be associated with an acceleration of the cell cycle through G1, we next assessed the levels of the late G1/S phase cell cycle regulatory protein, cyclin E, in the same sets of mice. Approximately 67% (±22.5, N = 3) of the cells stained positive for nuclear cyclin E in the PCa lesions from mice fed AIN93-CO, and 56% (±19.3, N = 3) were positive in the AIN93-CO:FO fed animals (Fig. 2D and E). In contrast, 20% (±8.2, N = 4) of the cells were positive for nuclear cyclin E in the PCa lesions of chow fed mice (Suppl. Fig. S2). The increase in cyclin E staining between either of the refined diets versus chow was statistically significant (p ≤ 0.05), but not between AIN93-CO and AIN93-CO:FO. These data establish that cyclin E may be an important target in the prostate lesions of mice fed the refined diets.
mTOR is markedly elevated in prostate tumors of mice fed refined diets
We have previously reported that the levels of phosphorylated mTOR (phosphorylated mammalian target of rapamycin or p-mTOR) were low in the PB-ErbB-2 x pten+/− adenocarcinomas of mice fed chow.10 Because of the increased disease penetrance, immunohistochemical staining for p-mTOR was performed. Frequent p-mTOR immunopositivity was seen in the PCa lesions (Fig. 2G) from mice fed the AIN93-CO diets, with 83% (±6.2, N = 3) of the tumor cells staining positive. Parallel analyses performed on prostates from mice fed the AIN93-CO:FO diet (Fig. 2H) also showed strong immunopositivity, 84% (±7.8, N = 3) while p-mTOR staining in the normal adjacent epithelium (Fig. 2I) remained low, and staining in adenocarcinomas of mice fed chow was low as previously reported,10 and diffuse (Fig. 2J).
Prevalence of the phosphorylated form of the pro-apoptotic protein, BAD
Activation of the MAPK and PI3K pathways has been shown to inhibit apoptosis in part through the phosphorylation and subsequent inactivation of the pro-apoptotic protein, BAD,16 and both pathways are induced in our model.9,10 In addition, preclinical studies reporting tumor suppression by n-3 PUFAs in a prostate-specific conditional pten knockout model7 implicated BAD phosphorylation (p-BAD) as a critical mechanism that supported apoptosis in the prostate epithelium. In order to assess whether p-BAD status was affected by diet in the PB-ErbB-2 x Pten+/− model, IHC for p-BAD was carried out on PCa lesions from mice fed either chow or the refined diets (Fig. 3). No significant differences in the number of p-BAD positive cells were observed, with 84% (±7.8, N = 3), 79% (±4.1, N = 3) or 82% (±5.6, N = 3) of the cells staining p-BAD positive in the PCa lesions of the mice fed chow, AIN93-CO or AIN93-CO:FO diets, respectively (Fig. 3A–C).
Figure 3
Figure 3
Assessment of prevalence of phosphorylated-BAD in PB-ErbB-2 x Pten+/− mice. Immunohistochemical staining for phospho-BAD in ventral prostate lesions in 12 month old mice fed (A) Chow, (B) AIN93-CO and (C) AIN93-CO:FO. Representative examples are (more ...)
In this study, we found that a semi-purified, refined AIN93G-based diet with corn oil as sole source of fat augmented the incidences of PCa in our PB-ErbB-2 x Pten+/− sporadic model of prostate disease from 15% in the lab chow fed mice to 100%. Furthermore reformulation of the AIN93-corn oil diet to incorporate equal amounts of menhaden oil (which has 21% of its triglycerides being n-3 PUFA’s) to corn oil (which has 59% of its triglycerides being n-6 PUFA’s and very little n-3 PUFA’s17) did not reduce PCa incidences. In addition, cyclin E and p-mTOR were significantly induced in the AIN93-based diets and again the inclusion of fish oil in the diet had no appreciable effect. Therefore, in the context of this preclinical model, we conclude that unrefined chow prevents or delays the onset of PCa while refined AIN93-based diets support prostate epithelial cell cycle progression and tumorigenesis. The availability of dietary n-3 PUFAs did not influence prostate cancer progression in this model.
It is possible that the different outcomes between the current study and those of Burquin et al. may stem in part from the different mouse strains used (FVB vs. C57/Bl6). Also, our animal model retains one pten allele and supports increased growth factor signaling through ErbB-2, as opposed to PCa driven by deletion of both pten alleles. Moreover, the previous study relied on purified DHA and EPA as dietary supplements, rather than menhaden oil as the n-3 PUFA source. The previous study’s diet also delivered increased energy from fat (30% vs. 17%, which may independently promote mTOR activation. In studies performed on non-transgenic FVB mice, AIN76-based obesity-inducing diets (containing 60%of calories from fat) induced p-mTOR levels in the prostate as compared to the control diet containing 12%of its calories from corn oil.18 We observed a significant increase in p-mTOR activity in the AIN93-influenced PCa lesions versus chow, however since only 17%of calories in the AIN93G-based diets were derived from fat vs. 13%in chow, it is unlikely that the small increment would affect energy balance sufficiently to independently induce p-mTOR activity. In addition, p-mTOR induction was not observed in the normal epithelium adjacent to the PCa (Fig. 2), nor was it induced in non-transgenic mice fed the refined diets (not shown). We conclude that dietary components present in the unrefined chow protect against PCa perhaps in part through inhibiting unchecked mTOR activation. The loss of these dietary components in the highly refined, westernized AIN93-based diets thereby supports the transformation of the prostate epithelium, but is independent of the PUFA source.
The decrease in the number of cyclin D1-positive cells and the concomitant increase in nuclear cyclin E staining in mice fed the refined diets supports the possibility that the loss of protection against tumorigenesis may be a function of increased cell cycle progression. Changes in p-BAD status were not seen in prostate tumors obtained from mice fed AIN93 diets versus lab chow, suggesting that apoptosis levels may not differ between diets. Additional experiments are underway to assess the levels of apoptosis. Our data raise the possibility that abdominal obesity, levels of activated mTOR and increased nuclear cyclin E positivity may represent biomarkers of dietary influences on PCa progression, especially in at-risk individuals. Interestingly, the refined AIN-93 diets are rich in processed sugars, such as sucrose, while lacking in important plant-derived lignans and isoflavones, and mice fed these diets had larger abdominal fat deposits versus chow, despite maintaining similar overall body weighs. While it is not yet established that the increased visceral fat observed influenced the rate of PCa, these observations suggest that alterations in the metabolic state of the animals may be occurring. Detailed diet reformulation will allow for the investigation into the dietary influences of sugar sources and plant derived compounds not only on PCa but also on energy utilization, metabolism as well as circulating levels of growth factors, insulin and adipokines such as adiponectin and leptin.
The mechanisms of interaction between diet, metabolism and cell cycle progression and tumorigenesis, if they occur, remain to be elucidated. Preclinical studies such as these will be extremely useful in understanding the complex interactions between genes, diet and disease in vivo, and in developing nutritional programs for effective prostate cancer prevention.
Animal model
The PB-ErbB-2 x pten+/− mouse model mice has been previously described.911 Briefly the minimal rat probasin promoter was used to drive prostate-specific expression of an activated ErbB-2 growth factor receptor. The pten+/− mice, which harbor hemizygous inactivation of one pten allele12 were originally provided by Dr. Pier Paolo Pandolfi. The compound-engineered PB-ErbB-2 x pten+/− line exists in an FVBN background. The genotypes were established as previously described.912 Male wild type FVBN mice were also used in these studies. All animal experiments described herein were performed in accordance with the Georgetown University Animal Use Committee guidelines.
The isolated male mice were placed either on mouse chow (Labdiet 20, Table 1), or on the following AIN-93G-based purified diets beginning at weaning; AIN93-CO, containing 70 g/kg corn oil or AIN93-CO:FO, containing 35 g/kg corn oil:35 g/kg menhaden oil (Harlan). Mice were maintained on the diets until they reached 12 months of age. Fresh food was administered weekly from a supply stored at −70°C and food consumption was measured.
All MRI procedures were carried out on the 20 cm bore, 7T Bruker horizontal magnet running Paravision 4.0 (Bruker), in the Lombardi Cancer Center’s Preclinical Imaging Research Laboratory. Mice were anesthetized using 1.5%isoflurane and 30%nitrous oxide and were positioned inside a cylindrical tunable resonant radiofrequency antenna (birdcage configuration) volume coil, tuned to a center frequency of approximately 300 Mhz (the resonant frequency of water molecules when subject to a field strength of 7 Tesla) using an animal management system13 specifically modified for mouse prostate imaging, as previously described.11
Fat volumetry by MRI
We have previously established the imaging parameters required for measuring mouse fat deposition.14 Briefly, a 3-D (3-dimensional) rapid acquisition with relaxation enhancement (RARE) imaging sequence was employed with the following characteristics: echo time 27.8 ms, TR 229.5 ms, RARE factor 8, flip angle 180 and a matrix of 256 × 128 × 128. The 3-D image data set was reconstructed using ImageJ (NIH) in a coronal view and the image transformed into an 8-bit pixel mode. Lower and upper threshold values were set to segment the fat of the image. The region of interest for visceral fat was determined by the following borders: cranial-upper border of kidneys, caudal-convergence of the common iliac veins in the pelvis, left and right. Total and visceral fat voxels were quantified by ImageJ to determine percent visceral fat content.
Immunohistochemical staining
Immunohistochemical staining was performed on prostate tissue as previously described,9,10,15 using the following antibodies: cyclin D1 (Neomarkers, AB3), cyclin E (Santa Cruz, sc481), phospho-BAD (Cell Signaling, 5284S), and phospho-mTOR (Cell Signaling, 2976). The slides were blocked for 20 minutes, and incubated overnight at 4°C with the primary antibody. Detection was performed as previously described.9,10 Statistical analyses were performed using the Students t test with significant differences established as p ≤ 0.05, as previously described.
Magnetic Resonance Imaging was performed in the Lombardi Comprehensive Cancer Center’s Preclinical Imaging Research Laboratory. FACS analyses were performed in the Lombardi Comprehensive Cancer Center’s Flow Cytometry and Cell Sorting Shared Resource; microscopy was performed in the Lombardi Comprehensive Cancer Center’s Microscopy and Imaging Shared Resource while mouse tissue embedding, tissue sectioning and prostate pathology were performed in the Cancer Center’s Histopathology and Tissue Shared Resource. Grant support, R01CA129003 and AICR 05B131 (C.A.), U54CA100970-02 (L.H.-C., C.A.) and R01CA102746 (M.A.).
PCaprostate cancer
PUFApolyunsaturated fatty acid
AAarachidonic acid
COX2cyclooxygenease 2
mTORmammalian target of rapamycin
p-BADphosophorlyated BAD
EPAeicosapentaenoic acid
DHAdocosahexaenoic acid

Supplementary materials can be found at:
1. von Hafe P, Pina F, Perez A, Tavares M, Barros H. Visceral fat accumulation as a risk factor for prostate cancer. Obes Res. 2004;12:1930–1935. [PubMed]
2. Freedland SJ, Terris MK, Presti JC, Jr, Amling CL, Kane CJ, Trock B, et al. Obesity and biochemical outcome following radical prostatectomy for organ confined disease with negative surgical margins. J Urol. 2004;172:520–524. [PubMed]
3. Kane CJ, Bassett WW, Sadetsky N, Silva S, Wallace K, Pasta DJ, et al. Obesity and prostate cancer clinical risk factors at presentation: data from CaPSURE. J Urol. 2005;173:732–736. [PubMed]
4. Freedland SJ, Terris MK, Platz EA, Presti JC., Jr Body mass index as a predictor of prostate cancer: development versus detection on biopsy. Urology. 2005;66:108–113. [PubMed]
5. Leitzmann MF, Stampfer MJ, Michaud DS, Augustsson K, Colditz GC, Willett WC, et al. Dietary intake of n-3 and n-6 fatty acids and the risk of prostate cancer. Am J Clin Nutr. 2004;80:204–216. [PubMed]
6. Ritch CR, Wan RL, Stephens LB, Taxy JB, Huo D, Gong EM, et al. Dietary fatty acids correlate with prostate cancer biopsy grade and volume in Jamaican men. J Urol. 2007;177:97–101. [PubMed]
7. Berquin IM, Min Y, Wu R, Wu J, Perry D, Cline JM, et al. Modulation of prostate cancer genetic risk by omega-3 and omega-6 fatty acids. J Clin Invest. 2007;117:1866–1875. [PMC free article] [PubMed]
8. Chavarro JE, Stampfer MJ, Hall MN, Sesso HD, Ma J. A 22-y prospective study of fish intake in relation to prostate cancer incidence and mortality. Am J Clin Nutr. 2008;88:1297–1303. [PMC free article] [PubMed]
9. Casimiro M, Rodriguez O, Pootrakul L, Aventian M, Lushina N, Cromelin C, et al. ErbB-2 induces the cyclin D1 gene in prostate epithelial cells in vitro and in vivo. Cancer Res. 2007;67:4364–4372. [PubMed]
10. Rodriguez OC, Lai EW, Vissapragada S, Cromelin C, Avetian M, Salinas P, et al. A reduction in Pten tumor suppressor activity promotes ErbB-2-induced mouse prostate adenocarcinoma formation through the activation of signaling cascades downstream of PDK1. Am J Pathol. 2009;174:2051–2060. [PubMed]
11. Fricke S, Rodriguez O, Vanmeter J, Dettin L, Casimiro M, Chien C, et al. In Vivo Magnetic Resonance Volumetric and Spectroscopic Analysis of Mouse Prostate Cancer Models. Prostate. 2006;66:708–717. [PubMed]
12. Di Cristofano A, Pesce B, Cordon-Cardo C, Pandolfi PP. PTEN is essential for embryonic development and tumour suppression. Nat Genet. 1998;19:348–355. [PubMed]
13. Fricke ST, Vink R, Chiodo C, Cernak I, Ileva L, Faden AI. Consistent and reproducible slice selection in rodent brain using a novel stereotaxic device for MRI. J Neurosci Methods. 2004;136:99–102. [PubMed]
14. Kuo LE, Kitlinska JB, Tilan JU, Li L, Baker SB, Johnson MD, et al. Neuropeptide Y acts directly in the periphery on fat tissue and mediates stress-induced obesity and metabolic syndrome. Nat Med. 2007;13:803–811. [PubMed]
15. Lai EW, Rodriguez OC, Aventian M, Cromelin C, Fricke ST, Martiniova L, et al. ErbB-2 induces bilateral adrenal pheochromocytoma formation in mice. Cell Cycle. 2007;6:1946–1950. [PubMed]
16. Ahn S, Kim J, Hara MR, Ren XR, Lefkowitz RJ. {beta}-Arrestin-2 Mediates Anti-apoptotic Signaling through Regulation of BAD Phosphorylation. J Biol Chem. 2009;284:8855–8865. [PMC free article] [PubMed]
17. Guilbaud NF, Gas N, Dupont MA, Valette A. Effects of differentiation-inducing agents on maturation of human MCF-7 breast cancer cells. J Cell Physiol. 1990;145:162–172. [PubMed]
18. Moore T, Beltran L, Carbajal S, Strom S, Traag J, Hursting SD, et al. Dietary energy balance modulates signaling through the Akt/mammalian target of rapamycin pathways in multiple epithelial tissues. Cancer Prev Res (Phila Pa) 2008;1:65–76. [PubMed]