The Diet as a Cause of Human Prostate Cancer

Abstract

Asymptomatic prostate inflammation and prostate cancer have reached epidemic proportions among men in the developed world. Animal model studies implicate dietary carcinogens, such as the heterocyclic amines from over-cooked meats and sex steroid hormones, particularly estrogens, as candidate etiologies for prostate cancer. Each acts by causing epithelial cell damage, triggering an inflammatory response that can evolve into a chronic or recurrent condition. This milieu appears to spawn proliferative inflammatory atrophy (PIA) lesions, a type of focal atrophy that represents the earliest of prostate cancer precursor lesions. Rare PIA lesions contain cells which exhibit high c-Myc expression, shortened telomere segments, and epigenetic silencing of genes such as GSTP1, encoding the π-class glutathione S-transferase, all characteristic of prostatic intraepithelial neoplasia (PIN) and prostate cancer. Subsequent genetic changes, such as the gene translocations/deletions that generate fusion transcripts between androgen-regulated genes (such as TMPRSS2) and genes encoding ETS family transcription factors (such as ERG1), arise in PIN lesions and may promote invasiveness characteristic of prostatic adenocarcinoma cells. Lethal prostate cancers contain markedly corrupted genomes and epigenomes. Epigenetic silencing, which seems to arise in response to the inflamed microenvironment generated by dietary carcinogens and/or estrogens as part of an epigenetic “catastrophe” affecting hundreds of genes, persists to drive clonal evolution through metastatic dissemination. The cause of the initial epigenetic “catastrophe” has not been determined but likely involves defective chromatin structure maintenance by over-exuberant DNA methylation or histone modification. With dietary carcinogens and estrogens driving pro-carcinogenic inflammation in the developed world, it is tempting to speculate that dietary components associated with decreased prostate cancer risk, such as intake of fruits and vegetables, especially tomatoes and crucifers, might act to attenuate the ravages of the chronic or recurrent inflammatory processes. Specifically, nutritional agents might prevent PIA lesions or reduce the propensity of PIA lesions to suffer “catastrophic” epigenome corruption.

Prostate cancer is the most common cancer diagnosed in men in the United States (US) and in Western Europe. Once rare in the rest of the world, the disease appears also to be on the rise throughout Asia and in many other developing nations. Diet and lifestyle, along with risk factors such as age, family history, and sex steroid hormones, have long been thought to contribute to prostatic carcinogenesis [57]. However, more recently, molecular pathology insights have indicted chronic or recurrent epithelial cell injury, accompanied by innate and adaptive inflammatory responses, in the early steps of prostate cancer development [18]. As a consequence, dietary components capable of inducing such injury, such as the heterocyclic amines created by over-cooking meats, loom large as candidate prostate carcinogens, while dietary components able to limit cell and genome damage and/or attenuate prostate inflammation, may protect against prostate cancer development. The mechanism(s) by which dietary components, inherited susceptibility, and sex steroid hormones cause epithelial damage and/or drive inflammatory processes that lead to cancer as men age, if better understood, could provide new opportunities for prostate cancer prevention, improved prostate cancer screening strategies, and perhaps even better prostate cancer treatment outcomes.

1 Proliferative Inflammatory Atrophy: A Lesion that Links Epithelial Injury to Prostate Cancer

In regions of the world with high prostate cancer incidence, prostate inflammation is essentially ubiquitous [18]. Though mostly asymptomatic, particularly if affecting the peripheral zone of the prostate where cancers arise, prostatitis has long been known to drive prostate cancer diagnoses independent of its propensity to cause the disease, because it tends to elevate serum prostate-specific antigen (PSA) levels. In the inflamed prostate, damage to the barrier function of the prostate epithelium stereotypically causes backflow of prostate secretions, including secreted proteins like PSA, into the prostate parenchyma and ultimately into the bloodstream. Because the detection of PSA in serum serves as the primary trigger of prostate biopsy for prostate cancer detection and diagnosis, prostate inflammation is responsible for a significant fraction of more than 30 million PSA tests, leading to more than a million prostate biopsies looking for cancer, performed in the United States each year [36]. Nonetheless, serum PSA elevations as early as age 40 years are associated with an increased risk of prostate cancer later in life [23, 24].

This tendency for prostate inflammation to elevate serum PSA has greatly undermined attempts to test causal associations between prostatitis and prostate cancer in population studies. Restricting cancer association studies to symptomatic prostatitis has not been very helpful either. Symptomatic prostatitis typically reflects inflammation of the transition zone near the urethra where benign prostatic hyperplasia (BPH) arises, leading to irritative voiding symptoms which prompt as many as 2 million physician visits in the United States each year, each accompanied by serum PSA tests [50]. As a consequence, apparent correlations between symptomatic prostatitis and prostate cancer in epidemiologic studies, which are numerous, have been frequently attributed to the bias that such men might be more likely subjected to prostate biopsy [19]. In addition, since symptomatic prostatitis does not reflect inflammation in the prostate peripheral zone where cancers arise, the condition presumably serves as a poor surrogate for pro-carcinogenic inflammatory processes. To circumvent these limitations and more directly test whether peripheral zone inflammation might be correlated with prostate cancer, an analysis of end-of-study prostate biopsies collected from placebo-treated subjects participating in the Prostate Cancer Prevention Trial of finasteride for asymptomatic men with serum PSA values <3 ng/mL was undertaken [83]. The results of this study, expected soon, should yield definitive insight into whether the presence of inflammation is directly associated with cancer within prostate glands.

Most of the inflammation in the prostate is a consequence of damage to the prostate epithelium, which can be caused by dietary carcinogens, estrogens, and inflammatory oxidants [18]. Microscopic examination of prostate tissues, particularly tissues from glands which contain prostate cancers, yields evidence of longstanding chronic or repeated tissue insults and innate immune responses. As an example, large numbers of corpora amylacea, microscopic laminated bodies containing calprotectin, myeloperoxidase, and α-defensins, from neutrophil granules, are often spread out through such prostate glands [71]. These bodies form as a consequence of neutrophil discharge during acute inflammation, and remain, like spent ammunition casings, even after the acute inflammatory process has subsided. Of interest, prostate tissues from rats prone to prostate inflammation and to prostate cancer also contain abundant corpora amylacea [32]. Characteristically, when the prostate epithelium suffers injury, focal atrophy lesions with dilated glandular lumens and immature epithelial cells appear [16, 17]. The epithelial cells act as if the normal secretory cell differentiation pathway has been abandoned in favor of a stress response with induced expression of α- and π-class glutathione S-transferases, cyclooxygenase-2, and other mediators of genome damage defense and cell survival [60, 86, 94]. Focal atrophy lesions, surrounded by corpora amylacea, mark prostate tissues that have suffered cell and tissue damage, and define a pro-carcinogenic milieu.

Proliferative inflammatory atrophy (PIA) lesions (Fig. 1), which comprise a subset of focal atrophy-type injury responses in the prostate, exhibit very high epithelial proliferation rates and infiltration by inflammatory cells [67]. This type of epithelial damage response is reminiscent of a number of other conditions in other organ sites, such as atrophic gastritis, hepatitis, and cirrhosis, that are known to lead to cancer development. Not surprisingly, PIA lesions also appear to be cancer precursor lesions, giving rise to prostatic intraepithelial neoplasia (PIN) or to prostate cancer directly [64]. The evidence that epithelial damage, leading to PIA, may initiate prostate cancer development, has accumulated over the past decade or so to include: (1) the detection of somatic genome and epigenome defects in PIA lesions that are identical to those seen in PIN and prostate cancer, (2) the visualization of direct morphological transitions between PIA and PIN and between PIA and prostate cancer, (3) the frequent appearance of somatic genetic and epigenetic alterations characteristic of PIN and cancer in PIA lesions, often to an intermediate degree between normal and neoplasia, (4) the greater prevalence of PIA lesions in prostate peripheral zone regions in areas of the world with high prostate cancer risk, and (5) the propensity for preclinical models of prostatic carcinogenesis, including those using dietary carcinogens, to first show PIA lesions before cancers arise [18].

Fig. 1.

Etiological factors for prostate cancer. Prostatic carcinogenesis follows damage to the prostate epithelium, regenerative proliferation, and chronic/recurrent inflammation, leading to proliferative inflammatory atrophy lesions. These lesions sprout cells with c-Myc activation, telomere shortening, and epigenetic gene inactivation, which give rise to prostatic intraepithelial neoplasia and prostate cancer, carrying targeted gene rearrangements, mutations, and an extensively corrupted epigenome

Prostate damage, followed by a maladaptive innate immune response and tissue injury response, may be the pathway that connects environmental exposures, including the diet, to prostate cancer development. Certainly, both in population studies and in molecular pathology analyses, the association between prostate inflammation, PIA, and prostate cancer is clear. Yet, the mechanisms by which dietary habits, or other exposures, lead to prostate damage have not been fully elucidated.

2 The Diet: A Source of Carcinogens that can Damage the Prostate Epithelium and Cause PIA

Epidemiology studies of prostate cancer have strongly implicated the diet as a major modulator of prostate cancer risk. Prostate cancer incidence and mortality varies among different geographic regions, with high prostate cancer risk in the United States and in Europe and low prostate cancer risk in Asia, yet immigrants from low-risk regions to high-risk regions typically adopt higher prostate cancer risks, particularly with cultural assimilation [27, 73]. This likely reflects dietary differences: either dietary habits in high-risk regions promote prostate cancer, dietary habits in low-risk regions prevent prostate cancer, or both. When examined in greater detail, the most consistent dietary association for prostate cancer appears to be intake of red meats and/or animal fats [26, 43]. For red meats, cooking at high temperatures or char-broiling creates both heterocyclic aromatic amine and polycyclic aromatic hydrocarbon carcinogens [39, 44]. These cooking practices are also associated with an increased prostate cancer risk and may partially explain an increased propensity for prostate cancer development among African–American men versus Caucasian men in the United States [37, 82].

The best studied of these carcinogens for prostate cancer is 2-amino-1-methyl-6-phenylimidazopyridine (PhIP), the most abundant of the >20 heterocyclic amines that can appear in over-cooked meats [8, 85, 87]. In rat models, PhIP ingestion causes PIA and prostate cancer [74]. By itself, PhIP is non-toxic and non-mutagenic, but after activation, first to N–OH-PhIP by CYP1A1 or CYP1A2 in the liver or elsewhere and then to more reactive species in prostate epithelial cells by sulfotransferases or by a kinase/phosphatase, PhIP can cause marked prostate epithelial cell damage, elicit inflammatory responses, and form pro-mutagenic adducts with DNA [51]. Epithelial injury, accompanied by an inflammatory response, is a critical feature of PhIP prostate carcinogenesis in rats. Rats fed PhIP show genome mutations in ventral, dorsolateral, and anterior prostate lobes, but exhibit epithelial cell damage, inflammation, and PIA in only the ventral lobes [51]. For epidemiology studies, precise estimates of PhIP exposure from food frequency questionnaires, even with added questions about food preparation preferences, have proven difficult. This is likely because cooking practices can generate a wide variation of heterocyclic amine levels, though frequent intake of meat by men who prefer the meats to be well-done, pan-fried, or grilled, appears to be accompanied by an increased risk for prostate cancer [15]. More recent studies have suggested that germ line variants in PhIP-metabolizing enzymes, including cytochromes p450 enzymes, sulfotransferases, and UDP-glucuronide transferases, might affect prostate damage upon over-cooked meat consumption [40].

The influence of meat consumption may not be restricted to the initiation step of prostate cancer development. The natural history of human prostate cancer encompasses many years, with small prostate cancers appearing as early as age 20–30 years in the United States in autopsy studies, diagnoses of localized cancers seen at age 60–70 years, and death from prostate cancer occurring at age 70 years and older. This provides a large window of opportunity for chronic PhIP intake to influence the pathogenesis of prostate cancer. The rat models of PhIP prostate cancer development feature carcinogen exposure for a limited period of time after puberty, followed by observation for many months, isolating the action of PhIP on the initiation and early promotion steps of carcinogenesis [74]. However, during the pathogenesis of human prostate cancer, the most common somatic gene defect, epigenetic silencing of GSTP1, results in loss of π-class glutathione S-transferase (GST) expression, a phenotype that may be remarkably sensitive to PhIP-mediated cell and genome damage [56]. This change first appears in PIA lesions and persists in prostate carcinoma cells throughout metastatic dissemination [52, 91]. Thus, not only might PhIP cause the epithelial injury that leads to the formation of PIA lesions, but PhIP might also continue to inflict cell and genome damage for decades, driving ongoing malignant prostate cancer progression.

3 Non-Dietary Exposures that Cause Prostate Epithelial Damage, Prostate Inflammation, and PIA

Sex steroid hormones, infections, and inheritance have all been thought to influence prostate cancer development. Androgenic hormones, such as testosterone and dihydrotestosterone, along with a functioning androgen receptor, are needed for normal growth and development of all sex accessory glands, including the prostate and seminal vesicles, but there is little evidence that androgens cause prostate cancer per se. Androgen levels decline steadily throughout life in adult men, reaching a peak around age 21 years and falling thereafter, as prostate cancers begin to arise [66, 68]. In the United States, African-American men suffer more prostate cancers than Caucasians, despite similar age-adjusted androgen levels [66]. Also, in the prostates of adult men, androgen signaling is required for terminal differentiation to the columnar secretory epithelial cell phenotype, promoting the transcription and translation of genes like PSA and TMPRSS2 and driving the production of secretions for the ejaculate. By acting in this way, androgens tend to suppress epithelial cell proliferation.

Of course, androgen signaling does play a significant role in the progression of established prostate cancers. The mechanism for this appears to involve the acquisition of somatic genome translocations and deletions which create fusion transcripts formed from androgen-regulated differentiation genes, such as TMPRSS2 and others, and oncogenes, such as ERG and ETV1 of the ETS family of transcription factors [84]. With these somatic genome defects, prostate cancer cells co-opt androgen signaling for the maintenance of a neoplastic phenotype. This may be the mechanistic basis for the frequent responses of advanced prostate cancers to androgen deprivation or to antiandrogens: interference with androgen signaling results in a reduction in the levels of TMPRSS2-ERG or other fusion transcripts in prostate cancer cells, attenuating cell growth and limiting cell survival. Remarkably, while androgens may act more to drive prostate cancer progression than to trigger prostate cancer initiation, new data have suggested that the initiation of androgen-target gene transcription might involve induction of DNA double-strand breaks by androgen receptor-associated TOP2B, a topoisomerase capable of resolving tangles in DNA, directed to sites in TMPRSS2 most often involved in translocations and deletions [28]. These breaks, which lead to TMPRSS2-ERG fusions, likely occur after the emergence of PIA lesions, perhaps driving PIN cells to become more invasive carcinoma cells. Topoisomerases prevent DNA tangling by catalyzing DNA breakage/rejoining reactions to permit strand passage. The enzymes are well-known to be sensitive to many compounds which can disrupt rejoining reactions and create recombinogenic DNA double-strand breaks. However, whether dietary components are responsible for any TOP2B-mediated DNA double-strand breaks during the pathogenesis of prostate cancer has not been reported.

Unlike androgens, estrogens may act to damage the prostate epithelium and promote the early steps of prostatic carcinogenesis. Breast cancers and prostate cancers are generally coincident throughout the world, leading to the hypothesis that estrogens might cause both diseases [14]. This contention is largely supported by rodent models, where estrogen exposures lead both to prostate inflammation and to prostatic cancer [59]. As an example, exposure of adult male Wistar rats to 17ß-estradiol results in prostate inflammation whether or not dihydrotestosterone is also given [55]. Male rodents given perinatal or neonatal estrogen exposures manifest prostatitis in adulthood [55, 75, 76]. Estrogens likely trigger inflammation in rodent prostates via induction of autoimmunity, as the condition can be induced in non-estrogen-treated rats via adoptive transfer of T-cells from adult male rats given 17ß-estradiol [69]. The mechanisms by which estrogens cause autoimmune prostate inflammation have not been fully elucidated, but may involve pituitary prolactin secretion, differential action of estrogenic hormones on estrogen receptor isoforms in the prostate, and/or reactive oxygen species generation by estrogen redox cycling. Also, the influence of diet habits on estrogen levels in men, largely produced in fatty tissues by aromatase action on androgens, has not been determined. Of note, however, in the United States, African-Americans, with higher prostate cancer risks, tend to have higher estrogen levels than Caucasians [66].

Infectious causes of prostate cancer have been more difficult to pin down. Many infectious agents have been detected in prostate tissue specimens or prostate secretions, but which of these actually cause prostate damage or trigger prostate inflammatory responses has not been systematically determined [70]. The best studied have been sexually transmitted infections, where gonorrhea and chlamydia have been reported to elevate serum PSA levels, an indicator of damage to the epithelial barrier function of the prostate, in at least 32 % of cases [78]. Remarkably, despite effective antibiotic treatment, PSA values can remain elevated after such infections for many months, providing evidence for chronic epithelial damage and dysfunction. This is consistent with population studies finding an increased prostate cancer risk with mild increases in serum PSA at early ages and/or with a history of sexually transmitted infections [23, 24, 79]. This complicates the search for infectious etiologies for prostate cancer in population cohort studies. If a prostate infection can damage the prostate epithelium in a young man, leading to PIA and chronic prostate inflammation even without persistent colonization as the man ages, then definitively establishing which infectious pathogen was responsible for this “hit-and-run” phenomenon will be very difficult. Nonetheless, direct inoculation of rodent prostates with bacteria or viruses triggers marked inflammatory responses and subsequent prostate cancer precursor lesions [21, 22].

Finally, some of the inherited susceptibility to prostate cancer development may be explained by genes encoding participants driving the activation and intensity of innate inflammatory responses. Two such genes, RNASEL and MSR1, appear responsible for some familial clusters of prostate cancer [11, 90]. RNASEL encodes a ribonuclease that participates in an interferon-inducible RNA destruction pathway activated in response to viral infection or other cellular damaging stress; MSR1 encodes subunits of a macrophage scavenger receptor that binds bacterial lipopolysaccharide and lipoteichoic acid. Diminished function of either protein in mice reduces the ability to fully clear various infections [80, 96]. In population studies, fairly consistent associations have been seen between prostate cancer and polymorphic variants of genes encoding toll-like receptors (TLRs), such as TLR4 and the cluster TLR1-TLR6-TLR10 [77, 95]. TLRs can bind a broad range of pathogens and/or damaged cell components, acting via NF-κB signaling to promote vigorous innate immune responses [12].

Even though estrogens and infections seem to be able to cause prostate epithelial damage which might lead to PIA and prostate cancer in the absence of dietary influences, each of these processes could be impacted by dietary habits common in high-risk prostate cancer regions of the world. Estrogen levels tend to be higher in men with increased fatty tissues. The microbiome, a source of infections or of colonization resistance to infections, varies widely with dietary practices. By influencing the propensity for estrogens and infections to inflict prostate damage, the diet can indirectly act to promote prostate cancer. Similarly, the degree of dietary heterocyclic amine-mediated prostate damage is likely to be subject to the same host genetic factors that regulate the intensity of host responses to prostate infections. RNASEL can degrade human RNA as well as viral RNA, leading to apoptosis [89]. MSR1 helps clear circulating oxidized serum lowdensity lipoproteins [41]. TLRs are activated by damaged human cell components [12]. Thus, the diet likely exerts direct and indirect effects on human prostate cancer development.

4 Inflammation, PIA, and the Molecular Pathogenesis of Prostate Cancer

Life-threatening human prostate cancer cells contain 3,866 mutations (20 non-silent coding mutations), 108 rearrangements, 5,408 regions with DNA hypermethylation, shortened telomere sequences, and activated c-Myc protein [7, 58, 92]. Notably, the somatic mutations do not seem to have singled out any common “driver” of prostatic carcinogenesis, nor hinted at any base change signature more consistent with one type of carcinogen versus another. Instead, mutations seem to accumulate over time in individual cancers, influenced by whether acquired mismatch repair gene abnormalities have appeared and/or whether pro-mutagenic treatments have been used. The more consistent somatic genetic defect is the translocations described above, particularly those involving gene targets of androgen signaling fused to cancer genes, such as TMPRSS2-ERG, which may be attributed to errors in initiation of transcription in response to androgen action leading to TOP2B-associated DNA double-strand breaks [28]. This somatic genome defect appears to occur in PIN lesions and likely underlies the invasiveness characteristic of carcinoma. Of all of the somatic changes in prostate cancer cells, the most consistent and earliest seem to involve epigenetic gene silencing, telomere shortening, and c-Myc induction [58].

Gene silencing, resulting from increased DNA methylation at gene regulatory sequences, is a candidate-initiating event for human prostate cancer. As an example, GSTP1, encoding the π-class GST, an oxidant and carcinogen detoxifying enzyme, has been found to be epigenetically silenced in some 5–10 % of PIA lesions, >70 % of PIN lesions, and almost all prostate cancers [9, 52, 58]. GSTP1 silencing has been attributed to de novo DNA hypermethylation at the 5′ regulatory region of the gene [45]. Of note, since such DNA methylation changes are potentially reversible, that is, the DNA sequence remains intact; persistent maintenance of GSTP1 inactivation throughout prostate cancer progression has served as a priori evidence for a selective growth or survival advantage during prostatic carcinogenesis. The mechanism for such a selective advantage has not been elucidated. Certainly, GSTP1 serves a caretaker gene function during carcinogenesis generally, as human prostate cancer cells devoid of GSTP1 better activate PhIP to cell and genome damaging species that cells with the enzyme, and mice carrying disrupted Gstp1/2 genes develop more skin tumors in response to topical carcinogens, and more intestinal tumors in the setting of relentless inflammation, than wild-type mice [30, 65]. A clue to a selected phenotype may be in mice carrying Pb-c-Myc transgenes that develop prostate cancer [33]. Preliminary data hint that loss of π-class GST function in these mice triggers accelerated prostatic carcinogenesis.

GSTP1 is likely but one of many genes epigenetically silenced early during prostatic carcinogenesis, by some process tied to prostate inflammation and PIA [58]. Unfortunately, as of yet, the mechanism by which epigenetic silencing of any such genes occurs in the inflamed microenvironment of the prostate, or in other organs prone to cancer development, has remained elusive. Nonetheless, inflammation and epigenetic gene silencing appear to be major contributors to the earliest steps of epithelial carcinogenesis generally, with clear examples in inflammatory bowel disease, hepatitis, and gastritis in addition to PIA lesions in the prostate [81]. Presumably, the inflamed microenvironment promotes epigenetic gene silencing and aberrant DNA methylation by acting to influence the regulation of chromatin architecture, by interfering with fidelity of DNA methylation pattern preservation in such way as to promote over-methylation at certain gene sites, or by corrupting both chromatin structure and DNA methylation maintenance. In one mouse model study of intestinal tumorigenesis, new DNA methylation changes emerged in inflamed tissues at the loci of 250 genes, with 70 % of the genes known to be targeted by polycomb complexes for repression [29]. This observation nominates polycomb complex repression as a participant in the establishment of de novo DNA methylation changes in the setting of pro-neoplastic inflammation. In this type of mechanism, DNA methylation reinforces polycomb complex repression in some way to maintain epigenetic gene silencing. In support of a complementary type of mechanism, the inflammatory cytokine interleukin 1β was found to trigger gene silencing in certain cells by promoting nitric oxide generation, leading to an over-activation of DNA methyltransferases [31]. Which type of mechanism drives the epigenetic catastrophe in PIA lesions in the prostate is not clear. GSTP1, along with other stress-response genes, tends to be induced to high level expression in PIA cells, with rare PIA cells suffering loss of GSTP1 expression and de novo methylation [52]. Thus, epigenetic silencing, and DNA methylation, must occur despite transcriptional trans-activation.

In addition to DNA methylation, telomere shortening and over-expression of c-Myc protein consistently accompany human prostatic carcinogenesis [33, 49]. The telomeres of chromosomes, specialized structures containing ~2,000 repeats of the sequence 5′-TTAGGG-3′ maintained by the enzyme telomerase, act to prevent loss of DNA sequences which might otherwise occur with lagging-strand DNA synthesis during replication and to reduce illegitimate recombination [48]. Critically, short telomeres, reflecting replication in the absence of the enzyme telomerase or some sort of telomere sequence damage, are characteristic of PIN lesions and prostate cancer cells [49]. Of note, short telomere sequences have also been seen associated with hepatitis and inflammatory bowel disease [2, 38]. Perhaps, oxidants elaborated at sites of inflammation can damage and shorten telomere sequences in the prostate as well. c-Myc expression, also ubiquitous in human prostate cancer cells, can drive prostate tumorigenesis in mice: forced c-Myc expression in the mouse prostate leads to the appearance of neoplastic cells with increased nuclear and nucleolar size, with blunted differentiation, and with diminished expression of Nkx3.1, a prostate-specific homeodomain transcription factor and tumor suppressor [33]. Like telomere shortening, c-Myc over-expression may be influenced by an inflammatory milieu. Mice carrying defective Apc genes prone to intestinal tumorigenesis show c-Myc phosphorylation and stabilization in response to exposure to TLR ligands from the intestinal microflora [42].

PIA lesions, which are generated in response to cell and tissue damage accompanied by an induced inflammatory response, link exposures, like dietary carcinogens and estrogens, to prostate cancer. The earliest stereotypical molecular events, epigenetic gene silencing, telomere shortening, and c-Myc activation, arise in PIA lesions. However, the precise mechanisms by which these molecular accidents occur have not been elaborated. Each may have its origins, or at least be influenced, by either the damaging exposure, for example, a dietary carcinogen, or by the inflammatory response. In this way, the prostatic carcinogenesis may resemble exposure-driven cancer development in many organ sites.

5 Rational Interventions to Prevent Prostate Cancer

If the epidemic of prostate cancer in the developed world can be explained by exposures to dietary carcinogens and/or estrogens that lead to chronic prostate inflammation, rational prostate cancer prevention approaches should involve: (1) an avoidance of exposures, (2) an attenuation of prostate cell and tissue damage inflicted by carcinogens, and/or (3) a reduction in intensity or duration of inflammation in the prostate. Not surprisingly, epidemiologic and clinical trial evidence has emerged in support of each approach.

A reduction in dietary heterocyclic amine exposure could be accomplished not only by educating individuals to avoid eating over-cooked meats, but also by attempting to modify cooking practices at a population scale. Steaming, microwaving, and marinating are all known to produce less heterocyclic amines than pan-frying, char-broiling, and retaining pan-dripping [85]. Measuring heterocyclic amine content in marketed cooked foods, along with appropriate incentives, might promote safer cooking practices by restaurants and by grocers. Until this occurs, carcinogen-inflicted prostate damage may be able to be lessened even despite continued consumption of over-cooked meats. Because heterocyclic amine carcinogens, like PhIP, need to be activated by metabolism to trigger cell and genome damage, interference with bioactivation might present an attractive strategy to limit carcinogenesis. Diets rich in inducers of phase 2 metabolic enzyme expression, which activate the Keap1-Nrf2 pathway, both reduce carcinogen damage generally in animal models and lower prostate cancer risk in human epidemiology studies [1, 13, 20]. The foods with the highest levels of phase 2 enzyme inducers, such as the isothiocyanate sulforaphane, are the cruciferous vegetables, like broccoli, Brussels sprouts, cauliflower, and others. In a study of normal human volunteers, intake of cruciferous vegetables reduced PhIP adduction to DNA in response to a cooked meat meal [88]. A reduction in estrogen exposures may be more difficult to achieve. Estrogens are produced via aromatase action and androgens in fatty tissues, a worrisome accompaniment to the obesity epidemic arising in the United States and other developed countries. Also, African-American men have higher estrogen levels than Caucasian men, even though androgen levels are similar and have higher prostate cancer risk [66]. Perhaps, if obesity can be controlled, prostate cancer risk might fall.

Anti-inflammatory approaches to prostate cancer prevention have yielded mixed results. Chronic or recurrent inflammation can generate reactive oxygen and nitrogen species that can cause cell and tissue damage [3, 34]. For this reason, dietary antioxidants or anti-inflammatory agents should act to protect prostate cells from the ravages of ongoing injury even if PIA lesions have already emerged. Certainly, some of the dietary components consumed in more commonly or abundantly in Asia, such as soy, may have anti-inflammatory properties relevant to the prostate; in rat models, soy-rich diets have been shown to reduce prostate inflammation [72]. In addition, prostate cancer epidemiology focused on cohorts in the developed world strongly suggests that inadequate intake of any number of antioxidant micronutrients, including vitamin E, selenium, and lycopenes from tomatoes, results in increased prostate cancer risk [10, 25, 93]. However, a recent large clinical trial, the selenium and vitamin E cancer prevention trial (SELECT), revealed that supplementation with selenium alone (hazard ratio (HR) of 1.04 with 99 % confidence interval of 0.87–1.24), with vitamin E alone (HR of 1.13 with 99 % confidence interval of 0.95–1.35), or with the combination (HR of 1.05 with 99 % confidence interval of 0.88–1.25) did not reduce prostate cancer incidence [46]. Whether this trial truly targeted men with inadequate antioxidant intake, rather than over-supplementing men with adequate intake, has not yet been reported.

Dietary and anti-inflammatory drugs might provide another strategy for lowering prostate cancer risk [6]. Cyclooxygenase (COX) inhibitors, including aspirin and non-steroidal anti-inflammatory drugs (NSAIDs), have been associated with reduced incidence and mortality of many human cancers [5]. Unfortunately, these drugs also cause gastrointestinal bleeding and cardiovascular events, which even though rare, have limited widespread use for preventing cancer. The use of such drugs for prostate cancer prevention has proven even more problematic. The selective COX-2 inhibitors, including celecoxib and rofecoxib, seemed promising when used in rodent prostate cancer models, but have thus far not been found to be particularly effective in human trials [4, 53, 54]. One difference between the rodents and humans may be that PTGS2, which encodes human COX-2, is epigenetically silenced in almost all human prostate cancers, but not in any rodent prostate cancers [91, 94]. Despite this poor track record of success of COX inhibitors for prostate cancer risk reduction, epidemiology studies have detected a consistent inverse correlation between regular aspirin use and prostate cancer [35, 47, 61]. Whether this potential benefit can be attributed to non-selective COX inhibition or to some other anti-inflammatory property of the aspirin has not been resolved. The finding that statin drugs, which may also exhibit anti-inflammatory activity, are associated with lowered prostate cancer risks has also not been fully explained [62, 63].

6 Summary and Conclusions

In response to dietary carcinogens, excess estrogens, or both, chronic or recurrent prostate inflammation, induced by damage to the prostate epithelium, drives prostate carcinogenesis to epidemic rates in the developed world. This pathogenesis mechanism is revealed in the emergence of PIA, precursor lesions for PIN and prostate cancer. Cells in PIA lesions, attempting to regenerate the prostate epithelium despite ongoing inflammatory stresses, appear prone to suffer epigenome corruption, heralded by GSTP1 silencing and de novo DNA methylation, telomere shortening, and c-Myc activation. Activation of androgen signaling in such cells risks targeted translocations, caused by androgen recruitment of TOP2B to binding sites at the loci of prostate cell differentiation genes, which lead to fusion gene transcripts, such as TMPRSS2-ERG, that drive malignant prostate cancer progression. Careful attention, at a population scale, to dietary habits, with changing of cooking practices to limit carcinogen production, with increased intake of cruciferous vegetables to reduce carcinogen bioactivation, and with increased intake of anti-inflammatory and antioxidant micronutrients to attenuate prostate inflammation, might contribute to lowering the societal burden of prostate cancer.

Abbreviations

PSA

Prostate-specific antigen

BPH

Benign prostatic hyperplasia

PIA

Proliferative inflammatory atrophy

PIN

Prostatic intraepithelial neoplasia

PhIP

Phenylimidazopyridine

GST

Glutathione S-transferase

TLRs

Toll-like receptors

COX

Cyclooxygenase

NSAIDs

Non-steroidal anti-inflammatory drugs