|Home | About | Journals | Submit | Contact Us | Français|
About 20% of all human cancers are caused by chronic infection or chronic inflammatory states. Recently, a new hypothesis has been proposed for prostate carcinogenesis. It proposes that exposure to environmental factors such as infectious agents and dietary carcinogens, and hormonal imbalances lead to injury of the prostate and to the development of chronic inflammation and regenerative ‘risk factor’ lesions, referred to as proliferative inflammatory atrophy (PIA). By developing new experimental animal models coupled with classical epidemiological studies, genetic epidemiological studies and molecular pathological approaches, we should be able to determine whether prostate cancer is driven by inflammation, and if so, to develop new strategies to prevent the disease.
Prostate cancer is the most common non-cutaneous malignant neoplasm in men in Western countries, responsible for the deaths of approximately 30,000 men per year in the United States1. The number of afflicted men is increasing rapidly as the population of males over the age of 50 grows worldwide. Therefore, finding strategies for the prevention of prostate cancer is a crucial medical challenge. As men in South East Asian countries have a low incidence of prostate cancer that increases rapidly after immigration to the West, this disease is not an intrinsic feature of ageing. The pathogenesis of prostate cancer reflects both hereditary and environmental components. What are the environmental factors and genetic variations that have produced such an epidemic of prostate cancer? Approximately 20% of all human cancers in adults result from chronic inflammatory states and/or chronic inflammation2–4 (BOX 1), which are triggered by infectious agents or exposure to other environmental factors, or by a combination thereof. There is also emerging evidence that inflammation is crucial for the aetiology of prostate cancer. This evidence stems from epidemiological, histopathological and molecular pathological studies. The objective of this Review is to take a multidisciplinary approach to present and analyse such studies. Because several reviews related to these topics have been published5–7, here we will focus on new findings and ideas with the purpose of sparking innovative areas of investigation that might ultimately lead to the prevention of prostate cancer.
Chronic inflammation is implicated in the development of a diverse range of human cancers, with overwhelming evidence causally linking it to cancer of the liver, stomach, large intestine, biliary tree and urinary bladder, and significant evidence to link it to cancer of the oesophagus, lung and pancreas. Many of these cancers are associated with infectious agents and/or a defined environmental exposure(s). Inflammation often collaborates with environmental exposures, such as dietary derived toxins, to increase cancer risk even further132. The molecular mechanisms that underlie the pathogenesis of inflammation-associated cancer are complex, and involve both the innate and adaptive immune systems3,133,134,135. Although viral oncogenes can contribute directly to neoplastic transformation, neither infection nor pathogen-encoded oncogenes are required for inflammatory cells to induce cancer136. Indeed, highly reactive chemical compounds, including superoxide, hydrogen peroxide, singlet oxygen and nitric oxide are released from activated phagocytic inflammatory cells of the innate immune system, and can cause oxidative or nitrosative damage to DNA in the epithelial cells, or react with other cellular components such as phospholipids, initiating a free-radical chain reaction2. The result is that many host epithelial cells are damaged and killed, and in order to preserve the barrier function of epithelia, these cells must be replaced by cell division from resident progenitor and/or stem cells. Epithelial cells that undergo DNA synthesis in the setting of these DNA-damaging agents are at an increased risk of mutation. That oxidant or nitrosative stress is important for driving prostate cancer formation is bolstered by epidemiological data, which indicate that the consumption of certain types of dietary antioxidants is associated with reduced prostate cancer risk137. Inflammatory cells also secrete cytokines that promote epithelial cell proliferation and stimulate angiogenesis138. In terms of disease progression, inflammatory cells migrate readily through the extracellular matrix as a result of the release of proteolytic enzymes and their inherent motile nature. Therefore, they might facilitate epithelial cell invasion into the stromal and vasculature compartments and, ultimately, the metastasis of tumour cells133,134. In another mechanism, the disruption of cytokine production and regulation, including cytokine deficiencies, lead to increased inflammation and cancer, whether in response to infection with a commensal organism or to chemical carcinogens139. A final mechanism is that certain immune responses can directly dampen cell-mediated anti-tumour immune surveillance mechanisms, thereby averting an immune reaction against the tumour that could potentially eliminate the cancer90.
As in other cancers, prostate cancer develops through the accumulation of somatic genetic and epigenetic changes, resulting in the inactivation of tumour-suppressor genes and caretaker genes, and the activation of oncogenes8,9 (TABLE 1). There is also evidence for an underlying genetic instability that might facilitate tumour progression10,11. Although these genetic and epigenetic changes are crucial for our understanding of ‘how’ prostate cancer arises, another key remaining question is ‘why’ prostate cancer is so common. The most consistent risk factors for the development of prostate cancer are advancing age, family history and race — diet is thought to be an emerging risk factor. To answer the question of why prostate cancer is so prevalent, several puzzling facts regarding its occurrence must be explained. The first enigma is the striking organ selectivity of prostate cancer within the genitourinary system: whereas there are approximately 280,000 new cases of prostate cancer in the US each year, there have been less than 50 reported cases of primary seminal vesicle carcinoma in the English literature12. The second unexplained issue is the geographic variation in the incidence of prostate cancer: as compared with the US and Western Europe, the incidence and mortality rates for prostate cancer are much lower in Southeast and East Asia13. Chinese and Japanese men who immigrate to the west acquire higher prostate cancer risks within one generation14, supporting an effect of the environment on prostate cancer development. A third key unsolved problem is the zonal predilection of prostate cancer. Most cancer lesions occur in the peripheral zone of the gland, fewer occur in the transition zone, and almost none arise in the central zone15 (FIG. 1).
Histologically, most lesions that contain either acute or chronic inflammatory infiltrates in the prostate are associated with atrophic epithelium or focal epithelial atrophy16–18. Perhaps correspondingly, focal areas of epithelial atrophy are common in the ageing prostate16,19, and often encompass a large fraction of the peripheral zone, where atrophy most often occurs19,20. Compared with normal epithelium, there is an increased fraction of epithelial cells that proliferate in focal atrophy lesions18,21,22, and we have proposed the term proliferative inflammatory atrophy (PIA) for most of these atrophic lesions18,23. Not all focal prostate atrophy lesions show increased inflammatory cells, and for these the term proliferative atrophy might be used. In morphological studies we and others have observed transitions between atrophic epithelium and adenocarcinoma16,24,25, and frequent transitions between areas of PIA and/or proliferative atrophy with high grade prostatic intraepithelial neoplasia (PIN)18,26. Although there is evidence for somatic genetic changes in PIA and proliferative atrophy, it seems from the studies published so far that most PIA and proliferative atrophy lesions do not harbour clonal genetic alterations (BOX 2). Tissue samples from patients with benign prostate hyperplasia (BPH), which occurs in the transition zone of the prostate (FIG. 1), have areas with markedly increased numbers of chronic inflammatory cells. In these areas, in almost all cases, the epithelium seems to be atrophic, indicating that these regions can be considered PIA of the transition zone.
In terms of somatic DNA alterations, although normal appearing epithelium (even from cancer patients) does not contain methylated glutathione S-transferase P1 (GSTP1) alleles, approximately 6% of focal atrophy lesions contain epithelial cells with methylated GSTP1 (REF. 25). Another group found mutated p53 (REF. 140) and androgen receptor alleles141 in post atrophic hyperplasia (a form of focal atrophy), prostatic intraepithelial neoplasia (PIN) and carcinoma, but not in normal prostate tissue — albeit these mutations in post atrophic hyperplasia were apparently non-clonal.
Others have used fluorescent in situ hybridization (FISH) to show that there are increases in chromosome 8 centromere signals142,143, loss of chromosome 8p143,144 and a gain of chromosome 8q24 in focal atrophy (REF. 143), indicating that chromosomal abnormalities similar to those found in PIN and carcinoma occur in a subset of these atrophic lesions. However, there were no atrophy cases in which clonal alterations were identified. Consistent with this, we recently found no evidence for clonal alterations indicative of chromosome 8 centromeric region gain, 8p loss or 8q24 gain in focal prostate atrophy, and infrequent 8p loss and absent 8q24 gain in PIN27. Therefore, the non-clonal mutations and non-clonal chromosomal alterations are probably indicative of genomic damage and/or the emergence of genomic instability in proliferative inflammatory atrophy (PIA) and proliferative atrophy.
In addition to the molecular evidence, other evidence to indicate that PIA and proliferative atrophy represent a field effect in the prostate is that these lesions are often quite extensive in the peripheral zone, often merge directly with high-grade PIN, at times merge directly with small carcinoma lesions and can be directly induced in the rodent prostate by exposures to the dietary carcinogen 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP).
Several key molecular pathways involved in prostate cancer have also been shown to be altered in PIA lesions (BOX 3). For example, the protein products of three prostate tumour-suppressor genes: NKX3.1 (REF. 27), CDKN1B, which encodes p27 (REFS 18,23), and phosphatase and tensin homologue (PTEN) (A.M.D. and D. Faith, unpublished observations) are all downregulated in focal atrophy lesions. These genes are highly expressed in normal prostate epithelium, and frequently decreased or absent in PIN and prostate cancer. In addition, one allele of their corresponding genetic loci is frequently deleted in carcinomas (TABLE 1), and forced overexpression of each of these genes causes decreased growth of prostate cancer cells in culture. Finally, animal models with targeted disruption of either one or two alleles of the corresponding mouse genes develop prostate hyperplasia, PIN and/or invasive carcinoma28.
In terms of molecular modes of action, p27 functions as an inhibitor of cell-cycle progression by inhibiting the activity of cyclin–cyclin dependent kinase complexes in the nucleus. Interestingly, p27 levels are generally reduced but not absent in human proliferative inflammatory atrophy (PIA), prostatic intraepithelial neoplasia (PIN) and prostate cancer23,145–147. The fact that p27 levels are not lost entirely (or biallelically inactivated by mutations) in cancer might be explained by recent findings that indicate that cytoplasmic p27 levels, which are increased by signalling through the MET receptor tyrosine kinase, are required for cell migration in response to hepatocyte growth factor signalling through MET and in response to increased cyclin D1 levels148. Therefore, although high levels of nuclear p27 can prevent cell-cycle progression, cytoplasmic p27 might be required for optimal tumour cell motility, which is a key feature of malignant transformation and tissue repair.
Phosphatase and tensin homologue (PTEN) is a dual protein and lipid phosphatase that is responsible for the dephosphorylation and inactivation of phosphatidylinositol 3,4,5-trisphosphate (PIP3), a second messenger produced after the activation of PIP3 kinase in response to the ligation of several growth factor receptors, including the insulin-like growth factor 1 receptor (IGF1R)149. PIP3 is required for the activation of the protein kinase AKT. AKT activation results in the inhibition of apoptosis and/or increased cell proliferation through several different effector mechanisms, such as the activation of mammalian target of rapamycin (mTOR) and S6 kinase.
NKX3.1 is a prostate-restricted homeodomain protein encoded within a region of chromosome 8p21 that often contains single copy deletions in prostate cancer150. In addition to suppressing the growth of prostate cells, decreased NKX3.1 protein levels result in increased oxidative DNA damage151.
PIA and proliferative atrophy also show increased BCL2 protein expression18, a gene product that is a potent suppressor of apoptosis. Other gene products that are increased in PIA and proliferative atrophy include those that are induced by oxidant and electrophilic stress, or by signals associated with cell activation and proliferation, including glutathione S-transferase P1 (GSTP1), GSTA1, cyclooxygenase 2 (PTGS2) and p16 (REFS 18, 152, 153). The fact that these stressed cells are undergoing tissue repair is supported by the finding that several proteins known to be involved in tissue repair and cell motility, such as MET23 and HAI1 (REF. 154), have increased expression in PIA and proliferative atrophy.
In most cases, the cause of prostatic inflammation is unclear. Various potential sources exist for the initial inciting event, including direct infection, urine reflux inducing chemical and physical trauma, dietary factors, oestrogens, or a combination of two or more of these factors (FIG. 2). Furthermore, any of these could lead to a break in immune tolerance and the development of an autoimmune reaction to the prostate.
Many different pathogenic organisms have been observed to infect and induce an inflammatory response in the prostate. These include sexually transmitted organisms, such as Neisseria gonorrhoeae29, Chlamydia trachomatis30, Trichomonas vaginalis31 and Treponema pallidum32, and non-sexually transmitted bacteria such as Propionibacterium acnes33 and those known to cause acute and chronic bacterial prostatitis, primarily Gram-negative organisms such as Escherichia coli34. Although each of these pathogens has been identified in the prostate, the extent to which they typically infect this organ varies. For example, T. pallidum is a very rare cause of granulomatous prostatitis32, which is itself a rare pattern of prostate inflammation. In the pre-antibiotic era before 1937, a large proportion of other sexually transmitted infections (STIs, predominantly gonorrhea) resulted in severe prostatic inflammation or prostatic abscess29. However, since the introduction of antibiotics this proportion has decreased dramatically, presumably owing to treatment before progression to the prostate. Despite this decline, asymptomatic infection and inflammation of the prostate can still occur. In their study of gonorrhea, Handsfield and colleagues35 cultured N. gonorrhoeae in expressed prostate fluid after urination in 93% of men with asymptomatic gonorrhea.
Viruses can also infect the prostate, and human papillomavirus (HPV), human herpes simplex virus type 2 (HSV2), cytomegalovirus (CMV) and human herpes virus type 8 (HHV8) have been detected in the prostate36–38. How frequently these agents infect the prostate, and whether they elicit an inflammatory response, is largely unknown. In conclusion, many different pathogens can infect the prostate. Whereas some of these are associated with inflammation, others have not been detected in association with inflammation. Because many additional bacterial sequences39, and now a new viral sequence40, can be found in prostate tissue in the absence of an ability to culture any of these organisms using traditional means, it is still possible that in analogy to H. pylori gastritis, researchers have missed a previously unidentified pathogen associated with most inflammatory lesions in the prostate.
Several epidemiological studies of STIs and prostate cancer have been undertaken (BOX 4). Adding weight to the argument for a link between inflammation and prostate cancer are data indicating that users of anti-inflammatory agents have a reduced risk of prostate cancer. Prospective and case–control studies, including a relatively small prospective analysis that we conducted, suggest a reduction of ~15–20% in the risk of prostate cancer in regular users of aspirin or nonsteroidal anti-inflammatory drugs (NSAIDs) compared with non-users41,42,43; however, in a large study by Jacobs et al. the effect was seen only in long-term users44.
Epidemiological studies of sexually transmited infections (STIs) and prostate cancer initially focused on gonorrhea and syphilis. Dennis and Dawson155 combined the results of these studies and estimated summary odds ratios (ORs) of 1.4 for the development of prostate cancer in patients with a history of any STI, 2.3 for a history of syphilis and 1.4 for a history of gonorrhea. Similar estimates were also calculated in a subsequent meta-analysis156. Another recent case–control study reported that a history of both gonorrhea and more than 25 previous sexual partners were associated with an increased risk of prostate cancer46. The significance of many of these studies is limited, as most were small case–control designs that may have been susceptible to selection, recall and interviewer bias. In terms of other STIs, we recently observed that men who carried antibodies against Trichomonas vaginalis had a higher risk of prostate cancer than men who did not, and the association was stronger in men who rarely used aspirin157.
In another inquiry we conducted a longitudinal study of young (median age <31) STI clinic patients by measuring serum prostate specific antigen (PSA) as a marker of prostate infection and damage. Men with an STI were more likely to have a ≥40% increase in serum PSA than men without an STI diagnosis (32% versus 2%, P <0.01)158. Increases in PSA levels were strongly suggestive of direct prostate involvement by the infectious agent, with resultant epithelial cell damage (either due to the organism itself or the inflammatory response to the organism) resulting in the release of PSA into the blood stream. As only about 32% of patients with acute STIs showed increased levels of PSA, it is apparent that either these agents do not always infect the prostate, or they do not illicit a strong inflammatory response that damages prostate tissues, or rapid antibiotic treatments prevent full-blown prostate involvement.
In terms of viral STIs and prostate cancer, Strickler and Goedert36 concluded that those studied to date are unlikely to contribute to prostate carcinogenesis, although they did suggest the possibility of a causal relationship between an as-yet unresearched and unidentified infectious agent and prostate cancer. Urisman and colleagues40 recently identified a novel γ-retrovirus in prostate tissues primarily from patients with germline RNASEL mutations. This intriguing finding is a proof of concept that specific infectious agents might persist in the prostate as a result of heritable changes in genes responsible for the clearance of these agents.
Although most studies investigating clinical prostatitis in relation to prostate cancer reported a positive association45,46, many of these studies might have been susceptible to detection bias. In our work, there was no association between clinical prostatitis and prostate cancer among men with an equal opportunity for prostate cancer screening by serum prostate specific antigen (PSA) testing, except in men diagnosed with cancer at a young age47. Although it is unclear why the effect was seen only in early-onset prostate cancer, it is possible that clinical prostatitis is associated with only a subset of prostate cancers that manifest at a relatively young age.
To determine whether inflammation is related to prostate cancer independent of clinical symptoms, it will be crucial to compare the patterns and extent of inflammation in prostate biopsy samples from men with and without carcinoma. As inflammation is so common in prostate specimens, these measurements will need to be quantitative, and will require large sample sizes. There is a US National Institutes of Health (NIH) consensus grading system48 for histological prostate inflammation, and we will be using this system in a nested case–control study to determine whether asymptomatic prostatic inflammation is associated with prostate cancer using needle biopsy specimens from the Prostate Cancer Prevention Trial (PCPT)49. This trial is a large (approximately 18,000 men) study that was carried out to determine whether the 5 α-reductase inhibitor, finasteride, could reduce the period prevalence of prostate cancer. We will measure the pattern and extent of prostate inflammation and relate these to the presence or absence of prostate cancer.
Chemical irritation from urine reflux has been proposed as an aetiological agent for the development of chronic inflammation in the prostate50. Although urine contains many chemical compounds that might be toxic to prostate epithelium, uric acid itself might be particularly damaging51. In support of this, recent work has implicated crystalline uric acid as a ‘danger signal’ released from dying cells, and it has been shown to directly engage the caspase-1-activating NALP3 (cyropyrin) inflammasome present in cells of the innate immune system (primarily macrophages), resulting in the production of inflammatory cytokines that can increase the influx of other inflammatory cells52. In addition, urine reflux of injurious chemicals can function in conjunction with infectious agents to increase prostate inflammation. Another manner by which prostate inflammation might occur is the development of corpora amylacea53 (FIG 2). Corpora amylacea have been proposed to contribute to prostate inflammation54, persistent infection55 and prostate carcinogenesis56 because they are frequently observed adjacent to damaged epithelium and focal inflammatory infiltrates54. In terms of epidemiological research and corpora amylacea, few studies have been conducted, one of which observed a higher proportion of calculi in prostate tissue from patients with prostate cancer compared with patients with BPH56, whereas others observed no association57,58. In support of the concept that ‘flushing’ of the prostate might be beneficial, several studies have found that increased ejaculation frequency, which might determine the rate of intraluminal corpora amylacea formation and the contact time between chemical agents such as uric acid or other urinary carcinogens and prostatic epithelium59, is related to decreased prostate cancer incidence (REF. 60 and references therein). Spermatozoa have also been localized to prostate tissues, and the retrograde movement of sperm cells into the prostate has been found in association with inflammation54,61 and with PIA lesions61.
Epidemiological studies have revealed a link between prostate cancer incidence and mortality and the consumption of red meat and animal fats62–64. One mechanism by which meats might stimulate cancer development could be related to the formation of hetero cyclic amines (HCAs)65. The exposure of laboratory rats to dietary 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP)66 results in carcinomas of the intestine in both sexes, in the mammary gland in females and in the prostate in males65. Rodent prostates contain four different lobes that do not correspond anatomically to the zones of the human prostate, and PhIP induces cancer only in the ventral lobe of rats67. In a recent study we exposed laboratory rats to PhIP and found a similar increase in the mutation frequency in all lobes of the prostate, yet the ventral lobe selectively responded with increased cell proliferation and cell death68. Therefore, PhIP functions as both a lobe-specific classical ‘tumour initiator’ as well as a ‘tumour promoter’. We also found that only the ventral lobe showed an increase in stromal mast cells, and stromal and intraepithelial macrophages68. After 12 weeks of PhIP exposure, the ventral lobe developed widespread epithelial atrophy; later, PIN and intraductal carcinomas were observed to develop directly from the atrophic epithelium (A.M.D., Y.N. and W.G.N., unpublished observations). Others have recently reported similar findings, in that PhIP treatment was found to induce inflammation and atrophy before inducing PIN and intraductal cancers69. Although it is not yet known whether the lobe-specific increase in mast cells and macrophages has a role in the neoplastic process, mast cells have been shown to stimulate cancer formation in several animal models, probably as a result of the release of factors such as tumour necrosis factor-α (TNFα) and various proteases, which might have an important role in tumorigenesis70–72.
Another line of research into the causes of prostate inflammation and prostate cancer is the study of oestrogenic exposures in the prostate. Oestrogens are strongly linked to autoimmune processes in women, who are much more predisposed to autoimmune diseases than men. Increased levels of oestrogens, whether from environmental or developmental exposures, have long been linked to the development of prostate cancer73,74. Oestrogens affect the growth and development of the prostate, and this occurs through indirect routes on the hypothalamic–pituitary–gonadal axis through prolactin, and also by direct effects mediated by oestrogen receptor-α (ERα), which is expressed primarily in the stroma, and oestrogen receptor-β(ERβ), which is expressed primarily in the epithelium73–76. Oestrogens given to neonatal rodents result in an ‘imprinted state’ or ‘developmental oestrogenization’ in which there are developmental defects, including a reduction in prostatic growth. This treatment also results in the development of lobe-specific inflammation, hyperplasia and dysplasia or PIN77,78. Virtually all of these effects are mediated through ERα79. Therefore, it is quite plausible that chronic inflammation in the adult human prostate might reflect an autoimmune reaction caused, at least in part, by oestrogens.
Another potential mechanism of self-perpetuating chronic inflammation in the prostate that could relate to all of the above-mentioned modes of prostate injury is that damaged prostate epithelial cells might release antigens that result in a break of the apparent immune ‘tolerance’ to the prostate. For example, many prostate antigens are not expressed until after puberty, when the gland undergoes androgen-stimulated growth and development. This is likely to result in a lack of physiological immune tolerance to these antigens. Therefore, when released during prostate injury, these antigens could prime an immune response resulting in a specific reaction to prostate-restricted antigens. Indeed, a T-cell immune response to PSA in patients with chronic prostatitis has been reported80.
In summary, many non-infectious mechanisms might lead to prostate epithelial cell and stromal damage. Injured cells are known to signal a ‘danger response’ that results in acute inflammation. Crystalline uric acid is particularly intriguing in this regard, as it directly interacts with a receptor that is part of a molecular pathway within innate immune cells that can potently stimulate inflammation. The fact that PhIP induces prostate inflammation and atrophy is also of great interest, as this might link diet to these processes in the prostate carcinogenesis pathway. Continuous exposure to the injurious agent can also set up the prostate for chronic inflammation that can lead to a sustained inflammatory response and cancer. Finally, all of these mechanisms of chronic epithelial injury might also result in a decreased barrier function, that could facilitate the growth of infectious agents that might further increase the inflammatory response, and allow toxic urinary metabolites into the prostatic interstitium, where they could further stimulate an inflammatory reaction. This is certainly an exciting area for continued research into the mechanisms of prostate carcinogenesis.
The normal prostate, like all other organs, contains endogenous inflammatory cells consisting of scattered stromal and intraepithelial T and B lymphocytes81,82, macro phages and mast cells. However, most adult prostate tissues contain increased inflammatory infiltrates, albeit the extent and type of inflammation are variable (for a review, see REF. 83). In terms of the biology of the inflammatory cells and the nature of the immune response in the prostate, most of the work has focused on BPH tissues in comparison with samples from the normal transition zone, and sometimes with carcinoma samples that have occurred in this region. Steiner et al. have examined the immunophenotypic and biological properties of chronic inflammatory cells in BPH and normal prostate tissues84–86. They have shown that of the increased CD45+ cells (all leukocytes express CD45 and non-leukocytes do not), 70–80% of these are CD3+ T lymphocytes, whereas 10–15% are CD19+ or CD20+ B lymphocytes. Macrophage numbers were also increased in these inflammatory lesions. In terms of the phenotype of the T cells, there is a reversed CD8:CD4 ratio, such that most T cells present in the normal areas expressed CD8, but most T cells in the inflamed areas expressed CD4. In terms of T-cell receptors (TCRs), 90% of the cells represent ‘standard’ αβ T cells (which express TCRαβ), with less than 1% representing γδ T cells. Class II major histocompatability antigen (HLADR), which indicates whether T cells are ‘activated’ by antigen signalling, is present on approximately 40% of the CD3+ T cells, and many of these T cells expressed CD45RO, indicating that these are ‘antigen experienced’ T cells85. None of the T cells in the normal prostate epithelium showed evidence of either activation or of being antigen experienced T cells.
CD4+ T cell responses can be divided into several different types that are classified according to their cytokine profile. TH1 cells produce interferon-γ and TNFα, whereas TH2 cells produce interleukin 4 (IL4), IL5 and IL13. Regulatory T (TReg) cells, which can suppress adaptive T-cell responses and autoimmunity, are characterized by the expression of CD25 and the transcription factor FOXP3, and they secrete transforming growth factor-β (TGFβ). In BPH, Marberger’s group determined that the T-cell response is complex, in that although TH0 (T cells that do not express any of the indicated cytokines) and TH1 cells were predominant in the inflammatory lesions of BPH and in carcinoma, some features of a TH2 response were also present. Unfortunately, at this point similar experiments have not been performed in the other zones of the prostate, or in areas of focal atrophy or PIN of the peripheral zone. The need for further understanding in this area is crucial, as is illustrated by the findings that microbially-driven inflammation can lead to colon cancer in mice, and that the prior transfer of TReg cells that express CD4 and CD25 prevents the inflammatory response that leads to colon cancer in these animals87. Recently Miller et al.88, have shown that CD4+ and CD25+ T cells, with properties of TReg cells including the expression of the FOXP3 protein, are present in increased numbers in clinically localized prostate cancer tissues, compared with normal prostate tissues. Exciting new data from several groups suggest the importance of a new subset of CD4-effector T cells known as TH17 cells, which develop through distinct cytokine signals (especially IL23) with respect to those involved in TH1 and TH2 responses, and are characterized by the production of IL17 (REF. 89). These cells are required for inflammation in arthritis and encephalitis models89, and IL23 is required for skin cancer formation in response to carcinogen exposure in mice90. A potential role for TH17 cells in prostatic inflammation had already been demonstrated by Steiner et al. before the TH17 cell lineage had been recognized as being as distinct. They showed that activated T cells in BPH tissue and in prostate cancer express high levels of IL17 (REF. 85). Further work to more fully elucidate the phenotypic and biological properties of all T-cell subsets in the prostate is required before we can understand the significance of acquired cell-mediated immunity in prostate carcinogenesis. Methods such as the quantitative image analysis of immunohistochemically stained inflammatory cell subsets, as well as flow cytometry for these subsets using tissues isolated from histologically defined areas, will be crucial to obtain such data.
Through a variety of approaches, including family and twin studies and segregation analyses, an important role for an inherited component of prostate cancer risk has been documented (recently reviewed by Schaid91). These studies have set the stage for efforts to identify prostate cancer susceptibility genes using linkage analysis and, more recently, association-based approaches. Despite strong evidence for a genetic component to prostate cancer risk, few reliable genetic risk factors for prostate cancer have been identified. In this section we will focus on a relatively new area of investigation in this field: the possibility that allelic variants of genes involved in innate and acquired immunity play an important part in determining inherited prostate cancer risk. If chronic inflammation is indeed an important aetiological factor for prostate cancer, then allelic variants of the genes involved in inflammatory pathways are logical candidates for genetic determinants of prostate cancer risk. As a result of space limitations, we can only review what we consider the most well-studied examples to date.
Following up genomic regions of interest identified by linkage studies of prostate cancer families, two genes involved in innate immunity unexpectedly emerged as candidate prostate cancer susceptibility genes. Inactivating mutations (E265X and M1I) in ribonuclease L (RNASEL) segregate with prostate cancer in two prostate cancer families: E265X with one of European descent and M1I with a family of African descent92,93. RNASEL, which is located at 1q25, is a component of the innate immune system that is required for the antiviral and antiproliferative roles of interferons94,95. Lymphoblasts from carriers of either one of the mutations mentioned above were found to be deficient in enzymatic RNase activity, although, other than prostate cancer, additional phenotypic manifestations were not obvious. Subsequent studies examining the role of RNASEL as a prostate cancer susceptibility gene have provided mixed evidence, some confirmatory91 and others not91,96–99. Although the association of this infection with prostate cancer development has yet to be shown40, carriers of a common, hypomorphic allele of RNASEL (R463Q) were found to be at risk for prostatic infection by a new γ-retrovirus. Interestingly, when RNASEL is activated in cells by its cognate interferon-inducible ligands, 2′,5′-linked oligoadenylates, mRNA species are consistently induced, one of which is encoded by the (macrophage-inhibitory cytokine 1) MIC1 gene, which is another prostate cancer susceptibility locus described below100.
The analysis of candidate genes in a different region of linkage (8p22) in prostate cancer families revealed several recurring, inactivating mutations in macrophage scavenger receptor 1 (MSR1)101. The MSR1 gene encodes a homotrimeric class A ‘scavenger receptor’, with expression largely restricted to macrophages. This receptor is capable of binding many ligands, including modified lipoproteins and both Gram-negative and Gram-positive bacteria. Mice with experimentally inactivated Msr1 are more susceptible to various types of bacterial infection102,103, although recent evidence suggests an anti-inflammatory role for this receptor, at least after exposure to certain pathogens104,105. MSR1 mutant alleles, R293X and H441R, which are found in several different prostate cancer families, code for proteins that can no longer bind bacteria or modified LDL (C.M. Ewing and W.B.I., unpublished observations). Despite the initial findings of an increased frequency of MSR1 mutations in men with prostate cancer, as with RNASEL, follow-up studies published on the possible role of MSR1 variants and prostate cancer risk have yielded inconsistent results98,101,106–111. A recent meta analysis suggests that MSR1 mutations might have a more reproducible effect on prostate cancer risk in African Americans112. Although these results do not indicate that these genes are major prostate cancer loci, they are consistent with these genes being able to modify prostate cancer risk, possibly in combination with particular environmental exposures — in this case, certain pathogens.
The Cancer Prostate Sweden Study (CAPS) is a case–control study of prostate cancer in northern Sweden. The relative genetic homogeneity of the Swedish population and the large size of the CAPS study make it an ideal platform to identify genetic variants associated with prostate cancer risk. Studying cases and controls in CAPS over the past 3 years has led to the identification of several genes in inflammation-related pathways, including MIC1, interleukin 1 receptor antagonist (IL1RN) and members of the toll-like receptor (TLR) family, with allelic variants associated with prostate cancer risk.
As key players in innate immunity to pathogens, TLRs recognize pathogen-associated molecular patterns (PAMPs)113. The engagement of TLRs results in the production of various pro-inflammatory cytokines, chemokines and effector molecules, such as reactive oxygen and nitrogen intermediates, as well as upregulation of the expression of co-stimulatory CD86 and CD80 and major histocompatability complex II (MHC II) molecules, which facilitate adaptive immune responses. Ten members of the human TLR family have been identified, and for most of these, specific classes of ligands, typically microbial components or surrogates thereof, have been identified and characterized. Recently, sequence variants in several TLR genes have been linked to prostate cancer risk, including TLR4 and the TLR1–6–10 gene cluster114,115.
Ligands that are recognized by TLR4 include Gram-negative bacterial products, including lipopolysaccharide116, and human heat shock protein 60 (HSP60)117. In the CAPS study114, a single nucleotide polymorphism (SNP) in the 3′ UTR region of TLR4 (11381G/C) was found to be associated with prostate cancer risk. Carriers of the GC or CC genotypes of this SNP had a 26% increased risk of prostate cancer, and a 39% increased risk of early-onset prostate cancer (before the age of 65 years), compared with men with the wild-type GG genotype.
In a follow up study of a North American population, homozygosity for variant alleles of eight SNPs in TLR4 (REF. 118) was associated with a statistically significantly lower risk of prostate cancer; however, the TLR4_15844 polymorphism, which corresponds to 11381G/C implicated in the CAPS population, was not found to be associated with prostate cancer. Therefore, although both published studies of this gene indicate that genetic variants of TLR4 have a role in the development of prostate cancer, the specific variants responsible for this effect might vary across different populations.
The TLR1–6–10 cluster maps to 4p14, and encodes proteins that have a high degree of homology in their overall amino-acid sequences119. TLR6 and TLR1 recognize diacylated lipoprotein and triacylated lipoprotein as ligands, respectively120,121. However, no specific ligand has been identified for TLR10. The TLR1 and TLR6 proteins each form heterodimers with TLR2 to establish a combinational repertoire that distinguishes a large number of PAMPs120,122,123.
A study of the TLR1–6–10 cluster in prostate cancer patients in CAPS identified an association of sequence variants in TLR1–6–10 with prostate cancer risk114,115. The allele frequencies of 11 of the 17 SNPs examined in this gene cluster were significantly different between case and control subjects (P = 0.04–0.001), with odds ratios for variant allele carriers (homozygous or heterozygous) compared with wild-type allele carriers ranging from 1.20 (95% CI = 1.00–1.43) to 1.38 (95% CI = 1.12–1.70). Although further studies are necessary to understand the biological consequences of the risk variants in both TLR4 and the TLR1–6–10 cluster, the observation of prostate cancer risk associated with polymorphisms in this family of genes, which is so intimately related to innate immunity, indicates that inflammation-related processes are important in prostate cancer development.
MIC1 is a member of the transforming growth factor-β (TGFβ) superfamily, and is thought to have an important role in inflammation by regulating macrophage activity. In a study of 1,383 patients with prostate cancer and 780 control subjects in CAPS, a significant difference (P = 0.006) in genotype frequency was observed for the non-synonymous change H6D between patients and controls124. Carriers of the GC genotype, which results in the H6D change, had a lower risk of sporadic prostate cancer (OR = 0.80, 95% CI = 0.66–0.97) and of familial prostate cancer (OR = 0.61, 95% CI = 0.42–0.89) than the CC genotype carriers. In the study population, the proportion of prostate cancer cases attributable to the CC genotype was 7.2% for sporadic cancer and 19.2% for familial cancer.
The protein product of the IL1RN gene belongs to the interleukin 1 cytokine family of proteins. Its primary function is as an inhibitor of the proinflammatory IL1α and IL1β. Lindmark et al. examined four haplotype-tagging SNPs (htSNPs) across the IL1RN gene in samples from patients with prostate cancer125. The most common haplotype (ATGC) was observed at a significantly higher frequency in the cases (38.7%) compared with the controls (33.5%) (P = 0.009). Carriers of the homozygous ATCG haplotype had significantly increased risk (OR = 1.6, 95% CI = 1.2–2.2). Furthermore, the association of this haplotype was even stronger among patients with advanced disease compared with controls125.
Many other genes in inflammatory pathways have been examined recently for a link to prostate cancer, generally with mixed results. For example, although McCarron et al. previously reported an association between certain alleles of IL10 and IL8 and prostate cancer126, Michaud et al. recently reported a lack of association of polymorphisms in the IL1β, IL6, IL8 and IL10 and prostate cancer in a case–control study of the Prostate, Lung, Colorectal, and Ovarian Cancer screening trial127. Further work is necessary to either confirm or refute the hypothesis that variants in genes associated with inflammation affect prostate cancer risk, and if confirmed, to understand the mechanisms that link allelic variation in inflammation genes and prostate cancer.
In a more global genome-wide approach, Zheng et al.128 proposed that sequence variants in many other genes in the inflammatory pathway might be associated with prostate cancer. They evaluated 9,275 SNPs in 1,086 genes of the inflammation pathway among 200 familial cases and 200 unaffected controls selected from the CAPS study population. They found that more than the expected numbers of SNPs were significant at a nominal P value of 0.01, 0.05 and 0.1, providing overall support for the hypothesis. A small subset of significant SNPs (N = 26) were selected and genotyped in an independent sample of ~1,900 members of the CAPS population. Among the 26 SNPs, six were significantly associated with prostate cancer risk (P ≤ 0.05). These results are consistent with the idea that variation in many genes in inflammatory pathways might affect the likelihood of developing prostate cancer.
Ideally, one would prefer to correlate the presence of specific genetic polymorphisms with the pattern and extent of intraprostatic inflammation, yet in all of the studies reported above the status of the prostate in men in terms of presence, pattern and extent of inflammation is unknown. Future studies that address these issues will be crucial in evaluating the biological effects of various polymorphisms in inflammatory pathway genes.
Our current working model (FIG. 3) suggests that repeated bouts of injury (and cell death) to the prostate epithelium occur, either as a result of oxidant and/or nitrosative damage from inflammatory cells in response to pathogens or autoimmune disease, from direct injury from circulating carcinogens and/or toxins derived from the diet or from urine that has refluxed into the prostate. The morphological manifestation of this injury is focal atrophy or PIA, which we postulate to be a signature of the ‘field effect’ of prostate carcinogenesis. The biological manifestations are an increase in proliferation and a massive increase in epithelial cells that possess a phenotype intermediate between basal cells and mature luminal cells5,6,23. In a small subset of cells, perhaps cells with an intermediate phenotype that contain at least some ‘stem cell’ properties, somatic genome alterations occur, such as cytosine methylation within the CpG island of the GSTP1 gene and telomere shortening. Both of these molecular changes can decrease the ‘caretaker’ phenotype and increase genetic instability that might then initiate high-grade PIN and early prostate cancer formation. In the setting of ongoing inflammatory and dietary insults in cells with compromised caretaker functions, additional changes such as gene rearrangements resulting in the activation of the ETS family of oncogenic transcription factors, the activation of MYC expression and the loss of tumour-suppressor genes such as PTEN, NKX3.1 and CDKN1B occur that drive tumour progression.
We reviewed evidence that in men with an underlying genetic predisposition, prostate cancer might be caused by inflammation possibly coupled with dietary factors. However, additional work needs to be done to determine whether the mechanisms proposed are correct. First, we need an improved ability to diagnose and define clinical ‘prostatitis’. Second, we need studies that quantify asymptomatic inflammation in the prostate to determine the relationship between the development of prostatic inflammation and the following: age, genotype, response to specific infectious organisms, and diet. We need an improved understanding of the types of inflammatory cells and their biological properties in the normal prostate and in the various lesions such as PIA, BPH, PIN and carcinoma. Another potential avenue for future studies is to couple improvements in imaging of the prostate, including new strategies to image inflammation and atrophy, to studies aimed at quantifying various types of inflammation in prostate biopsy specimens and quantitative analyses of cytokine profiles and inflammatory cell types in prostate fluid. These studies should also be performed in conjunction with experiments designed to identify specific infectious organisms. It will be crucial in these studies to have both the genetic information and the dietary and medical history data to correlate with the immunobiological data. Improvements in our understanding of the key molecular genetic and epigenetic events that drive prostate carcinogenesis, and the identification of the precise cell types involved (that is, whether prostate epithelial stem cells or their progeny are directly transformed) need to be applied to presumed precursor lesions to define precisely the order of events in the development of early prostate cancer. As animal models of prostate cancer continue to be developed that mimic the human disease, such as those that activate MYC or inactivate PTEN, CDKN1B or NKX3.1 (REFS 28,129), strategies for determining whether infectious agents and/or specific activated inflammatory cells are required for prostate carcinogenesis also need to be developed. Examples of such studies include crossing mice that are genetically engineered to develop prostate cancer with mice that lack specific subsets of cells of the innate and adaptive immune systems, to determine the contribution of such cells to the transformation process. In other studies, one can target inflammation to the prostate by using transgenic technologies to overexpress chemokines and/or cytokines that attract inflammatory cells to the prostate or that will activate inflammatory cells that are already resident in the prostate. As rodents are quite resistant to prostate cancer development, these studies would be potentially more informative if they were carried out on genetically altered animals already prone to developing early neoplastic lesions in the prostate.
Inflammation is a very complex process, which involves hundreds of genes. Therefore, there are many genes in the inflammatory pathways that might contribute to the development of prostate cancer. Whereas many genes in the inflammatory pathway have been shown to harbour sequence variants that may or may not be associated with increased risk of prostate cancer, larger studies in different study populations are needed to confirm and more thoroughly characterize the associations discussed above. Traditional association tests that examine one gene at a time remain valuable approaches, but they are fast being supplanted by new approaches that provide an efficient and economically feasible way to study virtually all of the genes in the whole pathway. Technologies using bead-based or chip-based arrays allow for the rapid examination of thousands of SNPs among hundreds of genes, and even genome-wide searches assessing all genes. Such approaches will provide a comprehensive evaluation of genes in inflammatory pathways, and will provide an appropriate perspective of the importance of genes in these pathways in the context of all known cellular pathways. Although the data obtained so far using genome-wide scans are promising, the challenges of such studies are significant, as many associations are expected to occur by chance. Only when specific associations are validated in multiple large cohorts will the scientific community have confidence in the purported findings from such studies. Although it is currently unknown whether or not 8q24 is related to inflammatory pathways, one very recent example of a success story stems from a set of independent studies that implicated this chromosome region in prostate cancer occurring in both families and in sporadic cases130,131.
The authors would like to thank Amelia K. Thomas for sketching the early concept designs for figure 2. Support was received from the Department of Defense Congressional Dir. Med. Research Program; The Public Health Services National Institutes of Health (NIH) and the National Cancer Institute, NIH and National Cancer Institute Specialized Programs of Research Excellence in Prostate Cancer, and philanthropic support from the Donald and Susan Sturm Foundation, B. L. Schwartz and R. A. Barry. A.M.D. is a Helen and Peter Bing Scholar through The Patrick C. Walsh Prostate Cancer Research Fund.
Competing interests statement
The authors declare no competing financial interests.
The following terms in this article are linked online to:
Entrez Gene: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene CDKN1B | CD3 | CD4 | CD8 | CD19 | CD20 | CD45 | CD80 | CD86 | ERα | ERβ | FOXP3 | GSTP1 | HSP60 | IL1RN | interferon-γ | IL4 | IL5 | IL13 | IL17 | IL23 | MIC1 | MSR1 | MYC | NKX3.1 | PTEN | RNASEL | TLR1 | TLR4 | TLR6 | TLR10 | TNFα
Angelo M. De Marzo’s homepage: http://demarzolab.pathology.jhmi.edu
Karolinska Institutet: http://ki.se/meb
Understanding Prostate Cancer: http://studentweb.usq.edu.au/home/q9210374/site/index.html
Access to this links box is available online.