Human prostate cancer risk factors§

Authors


  • This review describes the consensus proceedings of a conference entitled “Human Prostate Cancer Risk and the Environment” sponsored by the U.S. Environmental Protection Agency (EPA), Washington, DC (U.S. EPA contract 2W-2493-NATX dated June 25, 2002), Bostwick Laboratories (Richmond, VA), and the Virginia Prostate Cancer Coalition (Crystal City, VA), September 28, 2002.

  • The original draft was prepared by experts independently identified by the Eastern Research Group, Inc. (Lexington, MA), under EPA contract 68-5-0022. Further updated drafts were prepared by Bostwick Laboratories (Richmond VA) under EPA contract 2W-2493-NATX.

  • §

    The views expressed in this article are those of the authors and do not constitute U.S. EPA policy or endorsement. Mention of any trade name or commercial product also does not constitute endorsement or recommendation for use by the authors or by their affiliated institutions.

Abstract

Prostate cancer has the highest prevalence of any nonskin cancer in the human body, with similar likelihood of neoplastic foci found within the prostates of men around the world regardless of diet, occupation, lifestyle, or other factors. Essentially all men with circulating androgens will develop microscopic prostate cancer if they live long enough. This review is a contemporary and comprehensive, literature-based analysis of the putative risk factors for human prostate cancer, and the results were presented at a multidisciplinary consensus conference held in Crystal City, Virginia, in the fall of 2002. The objectives were to evaluate known environmental factors and mechanisms of prostatic carcinogenesis and to identify existing data gaps and future research needs. The review is divided into four sections, including 1) epidemiology (endogenous factors [family history, hormones, race, aging and oxidative stress] and exogenous factors [diet, environmental agents, occupation and other factors, including lifestyle factors]); 2) animal and cell culture models for prediction of human risk (rodent models, transgenic models, mouse reconstitution models, severe combined immunodeficiency syndrome mouse models, canine models, xenograft models, and cell culture models); 3) biomarkers in prostate cancer, most of which have been tested only as predictive factors for patient outcome after treatment rather than as risk factors; and 4) genotoxic and nongenotoxic mechanisms of carcinogenesis. The authors conclude that most of the data regarding risk relies, of necessity, on epidemiologic studies, but animal and cell culture models offer promise in confirming some important findings. The current understanding of biomarkers of disease and risk factors is limited. An understanding of the risk factors for prostate cancer has practical importance for public health research and policy, genetic and nutritional education and chemoprevention, and prevention strategies. Cancer 2004. © 2004 American Cancer Society.

1.0 GOALS

Prostate cancer poses a greater risk for American men, especially African-American men, than any other nonskin cancer. In 2003, it is estimated to account for 220,900 new cancer diagnoses and 28,900 deaths, approximately 1 every 15 minutes.1 This veritable epidemic of prostate cancer has resulted in part from successful efforts at early detection with the use of the serum prostate-specific antigen (PSA) test, thereby narrowing the still enormous gap between the clinical incidence (8% lifetime risk) and autopsy-based prevalence (80% by age 80 years). Most men die with prostate cancer rather than from it, yet physicians are unable to stratify patients accurately into those who will have progressive cancer and those who will not. An equally great problem is determining which men are at greatest risk for developing clinically apparent prostate cancer. An understanding of the risk factors for cancer has practical importance for public health efforts and genetic and nutritional education.

The process of determining risk begins by collecting data accumulated from epidemiologic studies, animal models, clinical trials, and basic investigations of cancer at the biologic and the molecular biologic levels. This review is a contemporary and comprehensive, literature-based analysis of the genetic and environmental risk factors for human prostate cancer. It emphasizes all aspects of our current understanding of risk, including epidemiology, pathology, animal and cell culture models, biomarkers of exposure, and mechanisms of risk exposure. Substantial gaps exist in the data regarding each of these aspects of risk, limiting our understanding of the complex mechanisms that contribute to the greatest cancer epidemic of our time.

Epidemiology of Prostate Cancer Risk

Epidemiologic studies have provided the greatest amount of information to date regarding risk of prostate cancer. However, epidemiology is a relatively crude tool for examining what may prove to be an unusually complex etiology. Most of these studies have significant problems with exposure and disease characterization.

The incidence of prostate cancer has increased in the past 50 years, with recent dramatic increases most likely due to early detection methods, such as the measurement of serum PSA, rather than true differences in underlying risk; a slight decline in the past few years most likely resulted from depletion of the pool of detectable cases. There is considerable international variation in the incidence of clinically detected prostate cancer, but comparisons are distorted by lead-time, case identification, detection, and reporting biases. Unlike clinical incidence, the age-specific prevalence of prostate cancer found at autopsy is relatively uniform across countries and ethnic groups, with contemporary studies indicating a rate as high as 80% by age 80 years.

Prostate cancer shares a number of features with benign prostatic hyperplasia (BPH) and the putative precursor of cancer, prostatic intraepithelial neoplasia (PIN). All increase in prevalence with host age, all require androgens for growth and development, and all respond to androgen-deprivation treatment. The continuum that culminates in high-grade PIN and early invasive cancer is characterized by progressive basal cell layer disruption, abnormalities in markers of secretory differentiation, increasing nuclear and nucleolar alterations, increasing cell proliferation, variation in DNA content, and increasing genetic instability. Some biomarkers show up-regulation or gain in the progression from benign prostatic epithelium to high-grade PIN and cancer, whereas others are down-regulated or lost. Existing data indicate that more biomarkers are up-regulated, but the relative importance of each is unknown. There is a significant increase in microvessel density in PIN and carcinoma compared with normal prostatic tissue.

Risk factors can be classified as endogenous or exogenous, although some factors are not exclusively one or the other (e.g., race, aging, oxidative stress). Recognizing that, some factors may reflect both endogenous and exogenous influences, and this is noted for those instances.

Endogenous Risk Factors

Endogenous risk factors for prostate cancer, include the following.

Family history

Family history is associated significantly with prostate cancer risk in epidemiologic studies but may be influenced by detection bias. The clinical and pathologic features of familial cancer are similar to nonfamilial cancer.

Hormones

Androgens significantly alter prostate cancer growth rates, and progression of prostate cancer from preclinical to clinically significant forms may result in part from altered androgen metabolism. Elevated concentrations of testosterone and its metabolite, dihydrotestosterone, over many decades may increase prostate cancer risk, but results have been inconsistent. Hormone levels may be affected both by endogenous factors (e.g., genetics) and by exogenous factors (e.g., exposure to environmental chemicals that affect hormone activity).

Race

Differences in prostate cancer risk by race may reflect three factors: differences in exposure, such as dietary differences (exogenous factors); differences in detection (reflecting exogenous factors); and genetic differences (endogenous factors). The highest incidence rates for prostate cancer in the world are among African-American men, who have a higher risk of prostate cancer than white American men. However, racial differences may reflect differences in access to care (exogenous factors), differences in the decision-making process of whether to seek medical attention and follow-up, and differences in allelic frequencies of microsatellites at the androgen receptor (AR) locus or polymorphic variation.

Aging and oxidative stress

Prostate cancer theoretically may result from an increase in oxidative stress, but supportive evidence is limited. Clinical studies indicate that intake of antioxidants, such as selenium, α-tocopherol (vitamin E), and lycopene (a carotenoid) offers protection against prostate cancer. Our current knowledge of the relation between aging and prooxidation-antioxidation homeostasis of the human prostate remains virtually nonexistent.

Exogenous Risk Factors

Exogenous risk factors for prostate cancer include the following.

Diet

A wide variety of dietary factors have been implicated in the development of prostate cancer according to descriptive epidemiologic studies of migrants, geographic variations, and temporal studies. Fat consumption, especially polyunsaturated fat, shows a strong, positive correlation with prostate cancer incidence and mortality, perhaps resulting from fat-induced alterations in hormonal profiles, the effect of fat metabolites as protein or DNA-reactive intermediates, or fat-induced elevation of oxidative stress. Retinoids, including vitamin A, help regulate epithelial cell differentiation and proliferation, with a positive association with prostate cancer risk. Vitamin C is a scavenger of reactive oxygen species (ROS) and free radicals, but there is no consistent association of intake and prostate cancer risk. Vitamin D deficiency may be a risk factor for prostate cancer; the hormonal form, 1-25-dihydroxyvitamin D, inhibits invasiveness and has antiproliferative and antidifferentiative effects on prostate cancer. Vitamin E (α-tocopherol) is an antioxidant that inhibits prostate cancer cell growth through apoptosis, and daily intake decreased the risk of prostate cancer by 32% in a large, controlled, clinical trial from Finland. Zinc (Zn) concentration is higher in the prostate than in any other organ in the body; although it is reduced > 90% in prostates with cancer; the relation of dietary zinc and prostate cancer risk is uncertain. Selenium is an essential trace element that inhibits viral and chemical, carcinogen-induced tumors in animals; a chemopreventative role for selenium is plausible, but the evidence in humans is limited. Alcohol intake has no significant association with prostate cancer risk. Consumption of cruciferous vegetables is associated with a decreased risk of many cancers, but there is no evidence of a protective effect for prostate cancer. Lycopene, an abundant constituent of tomato-based products and the most efficient carotenoid antioxidant, has a significant protective effect.

Environmental agents

One class of environmental agents that has received a lot of attention is the endocrine disrupting chemicals (EDCs). An EDC can be defined as an environmental agent that positively or negatively alters hormone activity (these are “endocrine-active”) EDCs) and ultimately leads to effects on reproduction, development, and/or carcinogenesis, particularly of reproductive organs. EDCs have been identified that elicit effects on estrogen, androgen, and/or thyroid activities. Although it has been shown that the majority of the well studied EDCs are estrogen agonists, which bind the to estrogen receptor (ER), thereby increasing estrogen activity, it has been shown that a number of EDCs affect other hormone activities. For example, it has been shown that the active metabolite of the pesticide vinclozolin is an androgen antagonist, binding to the AR and decreasing the expression of androgen-regulated genes,2 and an androgen agonist was identified in water downstream of pulp mills.3 Studies have shown that certain pesticide residues on foods, chemicals used in plastics production, and phytoestrogens in dietary plant products (e.g., soy) behave as EDCs. Exposure to EDCs can occur through ingestion of food or water or through inhalation. High-level exposure to estrogen agonists is unusual, but men may have chronic exposure to low doses of a mixture of EDCs. Individuals or groups with relatively high endogenous estrogen or androgen concentrations (serum or prostate tissue levels) may have a greater susceptibility to EDC exposure, because exposure to an EDC could add effectively to the endogenous activity.2

Cadmium is a significant environmental contaminant that has been linked to prostatic cancer in some, but not all, epidemiological studies. It is worth noting that the carcinogenic potential of cadmium may be modified by zinc.

Occupation and other factors

Many industrial and occupational exposures have been studied in relation to prostate cancer risk, but the findings are inconclusive; of greatest concern is farming and, to a lesser extent, working in the rubber industry. Numerous other factors have shown inconsistent results, negative associations, or have very limited data with prostate cancer risk, including smoking, energy intake, sexual activity, marital status, vasectomy, social factors (lifestyle, socioeconomic factors, and education), physical activity, and anthropometry.

Animal and Cell Culture Models for Prediction of Human Risk

Numerous animal bioassays and resultant cell lines have been developed for predicting human risk of spontaneous, inducible, and transplantable prostate cancer. Each bioassay has strengths and weaknesses as a model for human prostatic carcinogenesis.

Rodent models

Rodent models have many similarities to human prostate cancer and are easy and economic to use. Cancer may occur spontaneously, but most models rely on cancer induction or implantation, both of which occur quickly and predictably, often with dysplastic precursor lesions similar to human PIN. Differences exist regarding the site of origin in spontaneous and transplantable models.

Transgenic mouse models

Transgenic mouse models allow investigators to answer fundamental questions of prostate cancer biology and cancer progression. Constructs are made containing a promoter of a gene that is expressed specifically in the prostate and a gene of interest, allowing study of the phenotypic effects of expression of that oncogene in the prostate. Constructs have been developed consisting of promoters of genes that are activated oncogenes or genes that function to disrupt tumor suppressor genes, including the simian virus 40 (SV40) large-T antigen, which binds to both the retinoblastoma gene and the p53 tumor suppressor gene. Concerns persist regarding the validation and correlation of these models with human prostate cancer.

Mouse prostate reconstitution models

The mouse prostate reconstitution (MPR) model consists of dissected mouse urogenital sinus tissue, which is modified genetically in vitro, reconstituted with mesenchymal tissue, implanted under the renal capsule of mice, and allowed to differentiate. Grafts grow in the mouse into which they are grafted and organize ductal structures similar to those of the normal mouse prostate. This technique revealed that mesenchyme has the ability to influence the development of genital organ systems in mice.

Severe combined immunodeficiency syndrome mouse models

Human prostate cancer or cell lines can be implanted into mice with severe combined immunodeficiency syndrome (SCID). The percentage of SCID mice with cancer growth varies from 33% to 100%, depending on the strain. Hence, these mice allow growth of human prostatic adenocarcinoma effectively from pieces of cancer, and androgen-independent cancer can be selected that mimics the development of androgen-independent cancer from initially androgen-sensitive tumors. This model may have significant utility for testing cancer chemotherapeutic agents.

Canine model

The dog is the only nonhuman, nonrodent species in which spontaneous prostate cancer occurs frequently. Prostate cancer in dogs is aggressive clinically, with frequent metastases to regional lymph nodes, lung, and bone. Many cases occur in castrated dogs, suggesting that some tumors are independent of testicular androgens. Identification of PIN in the prostates of dogs with carcinoma provided the first evidence of spontaneously occurring PIN in animals. Microvessel density, a quantitative measure of angiogenesis, was greater in PIN and carcinoma in dogs than in benign prostatic epithelium. The dog prostate may provide an opportunity to evaluate rapidly the efficacy of promising chemopreventive agents.

Xenograft models

Xenograft models allow comparison of patient-derived specimens and the influence of the microenvironment on gene expression, cancer growth, and metastasis. Research involving xenograft models has focused on four areas: 1) characterizing the steps of cancer progression, including the emergence of androgen-independence; 2) assessing the efficacy of anticancer therapies; 3) identifying novel prostate cancer markers or phenotypes; and 4) evaluating metastatic potential of cell lines or carcinoma from patients.

Cell culture models

A variety of androgen-dependent and androgen-independent human and rodent cancer cell lines have been established, each of which exhibits unique properties while retaining the parental phenotype, offering novel opportunities for comparative studies. Nontransformed cell lines also have been established by viral immortalization of normal or hyperplastic prostatic cells.

Biomarkers of Prostate Cancer Risk, Exposure, and Effect

A wide variety of biomarkers have been identified in benign prostatic epithelium, PIN, and prostate cancer, many of which provide important predictive information for cancer progression and patient outcome. Most of these markers have not been evaluated as risk factors, perhaps with the exception of PSA. Markers reflect changes in cell morphometry; DNA ploidy; chromosomal gains and losses; cytoplasmic differentiation; cytoskeletal proteins; cell adhesion (such as E-cadherin), proliferation, and apoptosis; growth factors and their receptors; oncogenes and tumor suppressor genes; AR gene mutations; and metastasis suppressor genes.

Mechanisms of Action

The mechanisms of action that result in prostate cancer are uncertain but may be classified as genotoxic or nongenotoxic.

Genotoxic

Genetic mutations may be inherited due to familial or racial predilection, or they may be induced by a wide variety of agents; it is probable that molecular cloning of activated oncogenes from human prostate cancer and sequencing mutations will provide information on whether specific mutational spectra occur in particular genes. One possible genotoxic trigger may be continuous cell division, driven by hormones such as testosterone, resulting in the accumulation of spontaneous mutations, thereby activating select oncogenes and inactivating tumor suppressor genes. Another possible trigger may be small amounts of dietary carcinogens, such as 2-amino-1-methyl-6-phenylimidazo (4,5-b) pyridine (PhiP) in cooked fish and meat, cause mutations in prostate tissue over a lifetime. If this finding is confirmed in humans, then it would suggest that consumption of foods that contain PhiP over a lifetime could result in the consumption of a substantial prostate carcinogen. Because PhiP is mutagenic, forms DNA adducts, and is carcinogenic in the rat prostate, this may provide evidence for a genotoxic mechanism of human prostate carcinogenesis. However, currently, there is no strong evidence that known human chemical carcinogens or endocrine-disrupting chemicals play a role in the induction or evolution of human prostate cancer.1

Figure 1.

The combination of cell proliferation, protein damage, and DNA damage over time results in multiple effects, which ultimately result in prostate cancer.

Nongenotoxic

One hypothesis proposes that testosterone plays a significant role in the evolution of human prostatic cancer by acting as a stimulus for prostate cell growth. It may function as a mitogen or a tumor promoter. Testosterone induces cell division; and, over a lifetime, the large number of cell divisions may result in the accumulation of spontaneous mutations in prostate cells. Testosterone, at the least, is a necessary factor for prostatic carcinogenesis, but it may serve as a cofactor rather than the ultimate trigger. Another trigger may be oxidative stress induced by smoldering, long-standing, chronic inflammation and the unique biochemical milieu of the prostate (e.g., high citrate and zinc levels), ultimately resulting in mitogenesis.

2.0 INTRODUCTION

A risk factor separates patients without known disease into those at increased or decreased likelihood for disease. The benefit of a risk factor is related directly to the strength of its prediction for an individual patient. There is a great need for such factors for prostate cancer, particularly for men who will develop aggressive cancer. The prudent use of risk factors, including those acquired by epidemiologic study, animal and cell culture models, or evaluation of prognostic biomarkers, would be valuable in filling this need. However, there is a complex interaction between risk factors, and each may affect the prediction to a different degree. Therefore, multivariate analysis must be used to identify those factors that independently contribute to or define risk after controlling for other variables. A combination of independent covariables then may be used to enhance the accuracy of predicting risk, and clinical prediction models can be constructed.

Any study of risk of prostate cancer must account for five unique characteristics that contribute to its often unpredictable and paradoxical clinical behavior. First, prostate cancer tends to be slow-growing, with a typical doubling time of 3–4 years, significantly longer compared with the doubling time for cancers that affect the breast or colon. This slow growth rate most likely accounts for the extended latency of prostate cancer, which, in turn, contributes to the 10-fold greater prevalence than clinical incidence. Second, prostate cancer is remarkably age-related, rarely appearing before age 40 years and typically identified in men around age 70 years. This extraordinary linkage with age suggests that prostate cancer results from accumulation of genetic damage, perhaps due to oxidative stress or other endogenous or exogenous factors. Third, prostate cancer usually is multifocal, so that most men have prostate cancers, not just one cancer. Similarly, the likely precursor, PIN, usually is multifocal and often is in intimate spatial association with cancer. Fourth, prostate cancer and PIN are heterogeneous in their morphology and genotype; virtually the entire genome participates in prostatic carcinogenesis, with no apparent unique pathways like those observed with retinoblastoma, clear cell renal cell carcinoma, and other cancers. This remarkable heterogeneity suggests that multiple pathways and perhaps multiple mechanisms result in prostatic cancer. Finally, prostate cancer has the highest prevalence of any nonskin cancer in the human body, with similar likelihood of neoplastic foci found within the prostates of men around the world regardless of diet, occupation, lifestyle, or other factors. Essentially all men with circulating androgens will develop microscopic prostate cancer if they live long enough. What factor or combination of factors creates a universal risk for prostate cancer?

In this review, we describe the current state of knowledge regarding putative risk factors for human prostate cancer. Virtually all of the existing evidence has been obtained in the past 2 decades, with the greatest advances made in the past few years. The first section discusses the epidemiology of prostate cancer, including the putative precursor PIN, endogenous factors (family history, hormones, race, aging, and oxidative stress) and exogenous factors (diet, EDCs, occupation, and other factors). The second section surveys animal and cell culture models for prediction of human risk. These models include numerous rodent models, transgenic models, mouse reconstitution models, SCID mouse models, canine models, xenograft models, and cell culture models. The third section explores the abundance of biomarkers in prostate cancer, most of which have been tested only as predictive factors for patient outcome after treatment rather than as risk factors. The final section addresses the important issue of mechanisms of action for prostate cancer and considers the classification of genotoxic and nongenotoxic effects.

This review represents the culmination of months of investigation and interaction by a multidisciplinary team of epidemiologists, scientists, molecular biologists, pathologists, and clinicians to arrive at a balanced evaluation of the available evidence and current knowledge. Most of the data regarding risk relies, of necessity, on epidemiologic studies, but animal and cell culture models offer promise in confirming some important findings. Our understanding of biomarkers of disease risk and exposure is limited. This review is intended to serve as a benchmark against which future progress can be measured.

3.0 EPIDEMIOLOGY

3.1 Incidence

3.1.1 Geographic variation

Prostate cancer is the most common noncutaneous cancer in American men.1 The prevalence of latent cancer at autopsy is constant across countries and ethnic groups4–7; but to our knowledge there exists considerable international variation in the incidence of clinically detected prostate cancer. American men, and African-American men in particular, have the highest incidence of prostate cancer in the world.1 European and Canadian rates are lower than in the U.S.,8 but these rates are rising and are expected to double in the next few decades.9, 10 Incidence rates are considerably lower in Africa (Movare, Zimbabwe: 29 per 100,000; Kvadondo, Uganda: 28 per 100,000) compared with the rate of 137 per 100,000 for U.S. African-Americans reported by the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program.8 The lowest incidence rates are in Asia8, 11 and North Africa.8 This variation among populations may reflect the differential use of detection methods, such as PSA screening,11 differences in underlying risk, or in the biologic behavior of resultant tumors.

Based on SEER data for 1991–1995, there was an almost 2-fold difference in prostate cancer incidence between the U.S. region with lowest rate (Hawaii) and the region with the highest rate (Metropolitan Detroit).12, 13 Regional race-specific rates differed by < 20%.

3.1.2 Temporal trends

From 1947 to 1984, the incidence of clinically diagnosed prostate cancer in the U.S. increased by 74%.14 Large gains in incidence rates have occurred in the past decade,15 increasing by 61% among whites and by 65% among blacks from 1989 to 1992; the largest increases were observed in younger men. The proportion of men diagnosed with clinically localized cancer increased from 53% in 1986 to 74% in 1996.16, 17 Between 1992 and 1994, incidence rates declined 24%.18

3.1.3 Sources of error and bias

It is impossible to obtain accurate worldwide prostate cancer incidence estimates, because, where they do exist, the quality of registration systems differ from country to country. Some of the differences in reported incidence rates between African Americans and Africans likely reflect important differences in quality of cancer registration.

In addition to case registration, there is the problem of case identification. Most cases of prostate cancer are undetectable clinically, and any procedure, such as transurethral resection of the prostate (TURP) or PSA testing,15, 19 that increases the probability of identifying such cases will result in a higher incidence rate and will create lead-time bias.

There is a strong correlation in the U.S. and Canada between prostate cancer incidence rates from 1973 to 1986 and increasing use of TURP, most likely due to increased detection of prostate cancer by TURP, especially among whites.19, 20 The international variation in incidence rates may reflect differential use of detection methods, such as PSA screening,11 as well as differences in underlying risk. PSA screening is the likely basis for the large increase in incidence rates in the past decade and also for the declines in the past few years. The introduction of new screening modalities can result in lead-time bias and length-time bias; this occurs because screening identifies cases that otherwise would not be identified until later by other methods, if at all. The number of new prostate cancer cases has declined as the pool of detectable cases was exhausted. Similar effects have been noted for mammography and breast cancer.21

The use of TURP has declined in this decade with the introduction of pharmacologic and other nonsurgical means of treating BPH. It is possible that this also has contributed to the recent decline in prostate cancer incidence rates in the U.S.

3.2 Prevalence

3.2.1 Geographic variation

It is difficult to obtain accurate prevalence estimates of prostate cancer. The age-specific prevalence of prostate cancer found at autopsy is relatively uniform across countries4, 5, 22, 23 and ethnic groups.22 Several autopsy studies reported a high prevalence of latent prostate cancer ranging up to 80%, depending on age.22, 24, 25 Data from the Connecticut Tumor Registry estimated the age-standardized prevalence rate of prostate cancer at 841.6 per 100,000 in 1994, an increase of 126% over the 1982 rate.26 Prevalence rates per 100,000 in 1994 varied from 379.8 for men ages 50–59 years to 7852.8 for men age ≥ 70 years.

3.2.2 Sources of error and bias

The vast majority of prostate cancers are latent, nonlife-threatening cancers.27 Consequently, the reported prevalence estimates most likely grossly underestimate the true prevalence. Although a greater proportion of cancers are detected now due to increased screening, it is unknown how many incidental prostate cancers escape detection because of deaths from other causes.

3.3 Mortality

3.3.1 Geographic variation

Prostate cancer mortality varies considerably from country to country. High rates have been reported in the U.S., particularly among African Americans, whereas low rates have been found in China and Japan.28, 29 Mortality from prostate cancer among African-American men is more than double that of white men30 and almost 10 times greater than that for men in Hong Kong and Japan. The lifetime (age 40–90 years) risk that an American man will die from prostate cancer is approximately 3%, whereas the risk of dying with the cancer is as high as 72%.27 Race-specific mortality rates within the U.S. are relatively comparable. Overall rates are highest in states with large African-American populations, such as the District of Columbia, and lowest in Hawaii, with its large Asian-American population.13

3.3.2 Temporal trends

From 1971–1973 to 1991–1993, the U.S. mortality rate from prostate cancer rose from 21.4 per 100,000 to 26.8 per 100,000, an increase of 25%.13 Between 1991 and 1995, prostate cancer mortality declined from 26.7 per 100,000 to 24.9 per 100,000 population.13 Rates peaked in the early 1990s and have declined modestly thereafter.13 The 5-year relative survival rates for regional disease increased from 70% in 1973–1977 to 94% in 1988–1993.21 Similar improvements in survival were observed for men with localized cancer, but only modest improvements in survival were noted for men with distant disease.21

3.3.3 Sources of error and bias

There are advantages and disadvantages in using mortality data to examine the underlying risk of prostate cancer. Incidence data often are unavailable, so mortality is a commonly used surrogate. However, mortality is a function of both incidence and survival. International differences in mortality may reflect differences in the underlying risk of developing prostate cancer, but they may also reflect differences in survival.

More men die with prostate cancer rather than from prostate cancer.11, 27 Consequently, prostate cancer may not be stated as the cause of death on death certificates. Prostate cancer is a disease of the elderly, and the quality of information on death certificates for the elderly is less reliable than for younger adults, in part because the cause of death may be difficult to determine when there are multiple comorbid conditions. The autopsy rate among the elderly usually is lower than in younger adults.

Temporal trends in mortality are different from those observed for incidence. Recent prostate cancer incidence data from the U.S. have been highly variable, whereas mortality data have been quite stable. Improvements in prostate cancer therapy have confounded the use of mortality data, but it remains likely that, over the past decade, mortality data reflect trends in the underlying risk of prostate cancer better than incidence data. The basis for this decline is unclear. Possible explanations include chance, changes in patterns of reporting prostate cancer on death certificates (unlikely), improvements in treatment (unlikely), and early detection with PSA screening (likely).

3.4 Pathologic Findings in the Prostate

3.4.1 Benign Prostatic Hyperplasia (BPH)

Benign enlargement of the prostate (BPH, nodular hyperplasia, or adenofibromyomatous hyperplasia) consists of overgrowth of the epithelium and fibromuscular tissue of the transition zone and periurethral area. Lower urinary tract symptoms are caused by interference with muscular sphincteric function and obstruction of urine flow through the prostatic urethra. There is a weak, positive correlation between the amount of hyperplastic tissue and clinical symptoms.

The prevalence of histologic BPH increases rapidly beginning in the 4th decade of life, culminating in nearly 100% prevalence in the 9th decade. The age-specific prevalence is remarkably similar in populations throughout the world.31 Epidemiologic studies have shown that the risk of undergoing TURP for BPH is four-fold greater in first-degree relatives of young men with BPH compared with controls.32 The concordance rate for BPH among identical (monozygotic) twins is greater than among nonidentical (dizygotic) twins, suggesting a hereditary influence in BPH.33

The development of BPH includes three pathologic stages: nodule formation, diffuse enlargement of the transition zone and periurethral tissue, and enlargement of nodules. In men age < 70 years, diffuse enlargement predominates; in older men, epithelial proliferation and expansile growth of existing nodules predominates, most likely as the result of stimulation by androgens and other hormones.

The pathogenesis of BPH is uncertain, but multiple overlapping theories have been proposed, all of which may be operative (Table 1). Essential to all are advancing age and the presence of circulating androgens. Regression of BPH can be induced reversibly by luteinizing hormone-releasing hormone (LHRH) agonists, indicating that androgens have at least a supportive role in BPH.34

Table 1. Theories of Pathogenesis of Benign Prostatic Hyperplasia
Theory of pathogenesisProposed causative factorsComment and references
  1. NAD: nicotinamide adenine dinucleotide; NADPH: nicotinamide adenine dinucleotide phosphate diaphorase; DHT: dihydrotestosterone; BPH: benign prostatic hyperplasia; PDGF: platelet-derived growth factor.

AgingLipid peroxidation breakdown factorsIn men, as they age, there is an increase in cumulative lipid peroxidation, resulting in an increase in tissue concentration of cofactors such as NAD and NADPH. This, in turn, increases 5-α reductase concentration (sensitive to changes in NADPH) and prostatic DHT concentration, ultimately inducing epithelial and stromal growth that culminates in BPH. This theory is supported by the consistent observation of high DHT concentration in patients with BPH (see Geller, 19911466).
EstrogenSerum hormonesThe ratio of plasma estrogen to testosterone increases with age, and this may result in stromal overgrowth because of the greater amount of hormone receptors in the stroma compared with the epithelium. This theory explains why BPH is chiefly a stromal disease. Attempts to correlate the amount of BPH with serum hormone concentrations have yielded conflicting results (see Partin et al., 19911467), although testosterone and estrogen are clearly influential.
Embryonic reawakeningDHT and other androgensMcNeal suggested that the earliest lesion of BPH is a proliferation of epithelium, probably under the influence of DHT, with branching and budding due to “reawakening” of the embryonic inductive potential of the prostatic stroma during adulthood. This theory accounts for the presence of the common fibroadenomatous nodules of BPH (see McNeal, 19901468).
OxidoreductaseAndrogen-metabolizing enzymesAbnormal activity of certain enzymes may cause BPH by promoting the retention of tissue DHT, resulting in higher DHT levels. Isaacs et al. found significantly lower concentrations of two enzymes that remove DHT from tissue (17 β-hydroxysteroid and 3 α-hydroxysteroid reductase) in BPH patients than in controls (see Isaacs et al., 19831469).
Inflammation/growth factorsGrowth factorsInflammation and the release of growth factors such as PDGF may play a growth factor role in the development of BPH (see Gleason et al., 19931470). Steiner et al. found that the number of T-cells in BPH is greater than that in the normal prostate, and these T-cells are preactivated and functionally capable of producing sufficient amounts of autocrine growth factors necessary for T-cell proliferation (see Steiner et al., 1994571). Conversely, Helpap showed that there is no significant correlation of the amount of chronic inflammation and the extent of BPH (see Helpap, 19941471).

There are a number of similarities between BPH and cancer.31 Both display a parallel increase in prevalence with patient age according to autopsy studies, although cancer lags by 15–20 years. Both require androgens for growth and development, and both may respond to androgen-deprivation treatment. Most cancers arise in prostates with concomitant BPH, and cancer is found incidentally in a significant proportion (10%) of TURP specimens. BPH may be related to prostate cancer arising in the transition zone, perhaps in association with certain forms of hyperplasia, but BPH is not considered a premalignant lesion or a precursor of carcinoma.

3.4.2 Prostatic Intraepithelial Neoplasia (PIN)

PIN refers to the precancerous end of the continuum of cellular proliferations within the lining epithelium of prostatic ducts, ductules, and acini, with cytologic changes mimicking cancer, including nuclear and nucleolar enlargement.35–39 PIN coexists with cancer in most cases, but it retains an intact or fragmented basal cell layer, unlike cancer, which lacks a basal cell layer.35, 36 The term prostatic intraepithelial neoplasia has been endorsed at multiple consensus meetings, and terms such as dysplasia, malignant transformation, and intraductal carcinoma are discouraged.39, 40

There are two grades of PIN (low grade and high grade), although the term usually is used to indicate high-grade PIN. Interobserver agreement for high-grade PIN is good to excellent.41, 42

Early stromal invasion, the earliest evidence of carcinoma, commonly occurs at sites of acinar outpouching and basal cell disruption in acini with high-grade PIN.38 Such microinvasion is present in approximately 2% of high-power microscopic fields of PIN.

PIN and cancer usually are multifocal. PIN is multifocal in 72% of radical prostatectomies with cancer, including 63% of those involving the nontransition zone and 7% of those involving the transition zone, and 2% of tumors have concomitant single foci in all zones.43 The peripheral zone of the prostate, the area in which the majority (70%) of prostatic carcinomas occur, is also the most common location for PIN. Cancer and PIN frequently are multifocal in the peripheral zone, indicating a “field” effect similar to the multifocality of urothelial carcinoma of the bladder.

The only reliable method for detecting PIN is biopsy. It is a frequent finding in needle biopsies and is present in up to 25% of men.35, 36 The incidence varies according to the patient population and is higher in urology office practices than in screening populations.44 For example, Feneley et al. reported an incidence of PIN of 11% in hospital practice, 20% in men who were screened with PSA for prostate cancer, and 25% at case-finding in a urology practice.

The incidence and extent of PIN increase with patient age, predating the onset of carcinoma by > 5 years.45 The mean age of men who have isolated PIN is significantly younger compared with the mean age of men who have cancer (65 years vs. 70 years, respectively).45 High-grade PIN is identified first in the third decade and increases steadily with age.45 It is more prevalent and extensive in African-American men, apparently preceding that of white men by approximately a decade.46

The clinical importance of recognizing PIN is based on its strong association with carcinoma.47–52 PIN in quadrant-needle biopsies is a stronger predictor of cancer than serum PSA, and it indicates an approximately 15-fold greater risk compared with chance alone (risk ratio [RR], 14.93; 95% confidence interval [95% CI], 5.6–39.8).50 Among patients with a clinical suspicion of cancer and negative aspiration biopsies, follow-up aspiration revealed PIN in 31% and invasive carcinoma in 17%. The optimal repeat-biopsy strategy includes both sides of the prostate.51, 52

The continuum that culminates in high-grade PIN and early invasive cancer is characterized by progressive basal cell layer disruption, abnormalities in markers of secretory differentiation, increasing nuclear and nucleolar alterations, increasing cell proliferation, variation in DNA content, and increasing genetic instability.53 Some biomarkers show up-regulation or gain in the progression from benign prostatic epithelium to high-grade PIN and cancer, whereas others are down-regulated or lost. Existing data indicate that more biomarkers are up-regulated, but the relative importance of each is unknown. There is a prominent clustering of changes in expression for many biomarkers between benign epithelium and high-grade PIN, indicating that this is an important threshold for carcinogenesis in the prostate. Consistent with this hypothesis, PIN shows marked genetic heterogeneity and impairment of cell differentiation and regulatory control. Smaller numbers of changes are introduced in the progression from high-grade PIN to localized cancer, metastatic cancer, and hormone-refractory cancer.

PIN and prostate cancer have an abnormal nuclear DNA content.54 The mean proliferative index and proportion of aneuploid cell nuclei in high-grade PIN is similar to that in cancer and is much greater than that in hyperplastic epithelium and low grade PIN. The incidence of aneuploidy in high-grade PIN varies from 32% to 68%, somewhat lower than in carcinoma, which shows aneuploidy in 55–62% of tumors. There is a high level of concordance between the DNA content of PIN and cancer foci within the prostate. About 70% of aneuploid foci of PIN are associated with aneuploid carcinoma; conversely, only 29% of aneuploid cancers are associated with aneuploid PIN.

There is a marked decrease in the prevalence and extent of high-grade PIN after androgen-deprivation therapy compared with untreated tumors.55, 56 These findings indicate that PIN is androgen-dependent. In normal prostatic epithelium, luminal secretory cells are more sensitive to the absence of androgen control than basal cells, and these results show that the cells of high-grade PIN share this androgen sensitivity. The loss of normal, hyperplastic, and dysplastic epithelial cells after androgen-deprivation therapy most likely is due to the acceleration of programmed cell death (apoptosis) with subsequent exfoliation into glandular lumens.57 Taken together, these findings suggest that chemoprevention trials could target patients with PIN.

3.4.3 Cancer-associated pathologic findings

3.4.3.1 Luminal mucin

Acidic sulfated and nonsulfated mucins often are seen in acini of adenocarcinoma, appearing as amorphous or delicate, thread-like, basophilic secretions in routine sections. These mucins stain with Alcian blue and are demonstrated best at pH 2.5, whereas normal prostatic epithelium contains periodic acid Schiff-reactive, neutral mucin. Acidic mucin is not specific for carcinoma: It may be found in PIN but is found rarely in BPH.58 The predominant acidic mucin is sialomucin, and O-acetylation expression of these mucins is correlated inversely with cancer grade.59 Episialin, also known as MUC1, may or may not correlate with Gleason grade and microvessel density.60, 61

3.4.3.2 Crystalloids

Crystalloids are sharp, needle-like, eosinophilic structures that often are present in the lumens of well differentiated and moderately differentiated carcinoma.62 They are not specific for carcinoma and can be found in other conditions. The presence of crystalloids in metastatic adenocarcinoma with an unknown site of origin is strong, presumptive evidence of prostatic origin, although it is an uncommon finding.63

The pathogenesis of crystalloids is uncertain, but they most likely result from abnormal protein and mineral metabolism within benign and malignant acini, including loss of acidity.64 Ultrastructurally, they are composed of electron-dense material that lacks the periodicity of crystals, and X-ray microanalysis reveals abundant sulfur, calcium, and phosphorus and a small amount of sodium.62 Seminal vesicles also contain inspissated secretions in many tumors, including 24% of tumors with predominantly crystalloid morphology.64

3.4.3.3 Collagenous micronodules

Collagenous micronodules are a specific but infrequent and incidental finding in prostatic adenocarcinoma, consisting of microscopic nodular masses of paucicellular, eosinophilic, fibrillar stroma that impinge on acinar lumens.65 They usually are present in mucin-producing adenocarcinoma and result from extravasation of acidic mucin into the stroma. Collagenous micronodules are present in 2–13% of adenocarcinomas and are not observed in benign epithelium, nodular hyperplasia, or PIN.66 They are an infrequent finding and are present in 0.6% of needle biopsies and in 12.7% of prostatectomies.65, 66

3.4.3.4 Perineural invasion

Perineural invasion is common in adenocarcinoma and is present in 20%66 to 38%67 of biopsies; in some patients, perineural invasion may be the only evidence of malignancy found in a needle biopsy. This finding is strong, presumptive evidence of cancer but is not pathognomonic, because it occurs rarely with benign acini. Complete circumferential growth, intraneural invasion, and ganglionic invasion almost always are limited to cancer. Perineural invasion usually indicates tumor spread along the path of least resistance and does not represent lymphatic invasion. Routine light microscopy usually is sufficient to identify prostatic nerves.68

Approximately 50% of patients with perineural invasion on biopsy have extraprostatic extension, but perineural invasion has no independent predictive value for disease stage after consideration of Gleason grade, serum PSA, and amount of cancer on biopsy.67 Perineural invasion was a significant, independent, predictive factor for adverse outcome at 3 years for patients who received treatment with external beam radiation therapy; however, its value was associated only with a pretreatment serum PSA level < 20 ng/mL, suggesting that the poor prognosis associated with a elevated PSA overrides any additional information that perineural invasion may provide.69 Perineural invasion was predictive of recurrence-free survival after radical prostatectomy in univariate analyses70–72 but usually not in multivariate analyses.73–75 However, the greatest dimension of perinerual invasion may be an independent predictor of cancer volume76 and cancer recurrence,77 although it was not predictive of outcome after brachytherapy.78

3.4.3.5 Vascular/lymphatic invasion

Vascular/lymphatic invasion is present in 38% of radical prostatectomy specimens with cancer and commonly is associated with extraprostatic extension and lymph node metastases (in 62% and 67% of patients, respectively).79 Its presence also correlates with histologic grade.79 Microvascular invasion appears to be an important predictor of outcome and carries a four-fold greater risk of tumor progression and death.79 However, it is not an independent predictor of disease progression when stage and grade are included in the analysis.79

3.4.3.6 Microvessel density (angiogenesis)

There is a significant increase in microvessel density in PIN80 and carcinoma compared with normal prostatic tissue.81–84 Mean blood vessel counts are higher in tumors with metastases than in tumors without metastases,85, 86 and most82–84, 87 (but not all88) studies demonstrate a correlation with pathologic stage. Microvessel density appears to be an independent predictor of cancer progression,88–93 but this has been refuted.79 Increased microvessel density in prostatic carcinoma most likely is related to the production of proandrogenic growth factors. The cumulative data suggest that increased microvessel density contributes to extraprostatic spread of adenocarcinoma, perhaps by facilitating microvascular invasion, and also promotes the survival of distant metastases, because the tumor cells can shift androgenic balance within the distant target.

3.5 Risk Factors

3.5.1 Endogenous factors

3.5.1.1 Family history

Family history has been associated consistently with prostate cancer risk in epidemiologic studies.94–111 Prostate cancer appears to have a stronger familial aggregation than colon or breast cancer, two malignancies with well recognized familial components.112 One study noted a higher risk for family history among blacks (odds ratio [OR], 3.2) compared with whites (OR, 1.9), but the difference was not statistically significant.96 Another recent study also found that more African Americans reported a family history of prostate cancer (31.2%) than whites (22.2%). In the same study, more Hispanics had a family history of prostate cancer (25.0%) compared with whites (22.2%), but the difference was not statistically significant.108 Another recent study also found that positive family history increased the risk of prostate cancer for Hispanics and non-Hispanic whites.111

Risk is increased with greater genetic linkage of a man to an affected relative103 and with the greater number of relatives he has with prostate cancer.94, 101, 103, 110, 113 However, one study did not find an increased risk with an increasing percentage of affected family members.114 A recent case–control study in Kingston, Jamaica found that men who had an affected first-degree relative were 2 times more likely to develop prostate cancer than the general population (OR, 2.1; 95% CI, 1.1–4.4). There also may be a statistical difference in the risk of developing prostate cancer if an individual has one second-degree relative affected.99 Men with ≥ 3 first-degree or second-degree affected relatives have an 11-fold greater risk.94

A population-based case–control study in Quebec found an OR that approached 9.0 for men who had 1–4 first-degree relatives with prostate cancer.115 This high risk may have been due to the inclusion of men with more than two affected relatives, unlike many previous studies.116 A population-based case–control study in Alberta, Canada revealed that men who had an affected first-degree relative were more than three times as likely to develop prostate cancer compared with men who had no family history of prostate cancer.117 A later population-based case–control study, which also was conducted in Canada, compared symptomatic men with asymptomatic men. After adjusting for age, all men, both symptomatic and asymptomatic, who reported a history of prostate cancer in a first-degree relative were at increased risk (symptomatic men: OR, 2.41; 95% CI, 1.64–3.54; asymptomatic men: OR, 3.18; 95% CI, 2.28–4.45).105 In a study of Scandinavian twins, higher prostate cancer concordance rates were found for monozygotic twins versus dizygotic twins, suggesting a genetic influence on risk.118

A Swedish cohort study noted a higher standardized incidence ratio (SIR) for men ages 45–49 years who had a family history of prostate cancer (SIR, 3.38) compared with the SIR for men age > 80 years with a family history (SIR, 1.35).118 A more recent Swedish cohort study concluded similar results (men ages 45–49 years: SIR, 3.18; men age > 80 years: SIR, 1.45).100 Yet another study conducted in Sweden determined that men who had a family history of the disease had a 10 times, 5 times, and 3 times increased risk compared with the general population at age 60 years, age 70 years, and age 80 years, respectively.113 In a Massachusetts population-based case–control study,97 it was found that the elevated risk with a family history was confined to men age < 75 years (men age ≤ 65 years: OR, 4.0; 95% CI, 2.2–7.4; men age ≥ 75 years: OR, 0.78; 95% CI, 0.39–1.6). In a similar case–control study, 7.1% of patients age < 60 years reported at least 1 incident of prostate cancer among their first-degree relatives, whereas only 2.2% of patients age > 60 years reported similar family histories.102 A study in New York found that men who had a family history were diagnosed at an earlier mean age (64.9 years) compared with men who lacked this risk factor (66.9 years). Patients who had an affected father also were diagnosed at an earlier age compared with men who had an affected brother but not an affected father.108

One study found that family history predicted an elevated risk for aggressive prostate cancer.98 Clinical aggressiveness was defined as higher tumor grade and disease stage at diagnosis, increased likelihood of distant metastasis after surgical excision, and increased likelihood of prostate cancer mortality. Patients with localized cancer who reported a positive family history had a 23% worse outcome at 3 years and 5 years after either external beam radiotherapy or radical prostatectomy compared with patients who had sporadic cancer (at 3 years: RR, 1.4; 95% CI, 1.2–1.7; at 5 years: RR, 1.8; 95% CI, 1.3–2.4). Even in patients who had the most favorable tumor characteristics, a family history of prostate cancer still resulted in a worse outcome compared with patients who had no family history of prostate cancer.98 In contrast, Kotsis et al. found an association between family history and well differentiated prostate cancers. Those authors proposed that familial prostate cancer may be less aggressive than sporadic prostate cancer, because men with a family history may seek earlier detection and, thus, may be diagnosed with earlier stage cancer.107 Another recent study found that young men with a family history of prostate cancer were less likely to develop high-grade disease.119

Epidemiologic studies consistently note an increased risk with family history; however, one concern is that these studies may be confounded by detection bias. Individuals with a family history of prostate cancer may be screened more aggressively and may have a greater likelihood of positive findings because of the high prevalence of latent prostate cancer in the population.25 However, most studies of family history and prostate cancer were conducted before the widespread use of PSA screening, and the clinical and pathologic stage of prostate cancer is similar in men with and without a family history.94 A recent study in Australia120 has determined that the introduction of PSA testing does not appear to have altered the familial risk of prostate cancer significantly. Recent studies have suggested that the development of prostate cancer in genetically predisposed individuals is preceded by a subclinical period when PSA detection is possible.121 A study that was conducted using cases and controls from 1992 to 1994 observed a greater risk with family history among men who were identified through screening compared with men who were identified because of urologic symptoms, suggesting some detection bias; however, the risk still was elevated in men who were identified because of symptoms.97

Several studies have reported an increased prostate cancer risk among men with first-degree female relatives who had breast cancer,104, 122–124 but others failed to find this association.106, 125 One study found that the association was stronger in men age < 65 years with relatives who were diagnosed with breast cancer before age 50 years (RR, 1.65; 95% CI, 0.88–3.10) and among Jewish men (RR, 1.73; 95% CI, 1.00–2.97).124

The possible relation between parental age and subsequent risk of prostate cancer has been evaluated. Although no definite relation between maternal age and risk was found, there were some interesting findings with regard to paternal age: The fathers were split into quartiles based on their age at the birth of their son (age < 27 years, from age 27 years to < 32 years, from age 32 to < 38 years, and age > 38 years). After adjustment for age and other variables, men whose fathers were in the second, third, and fourth quartiles had 1.2 times, 1.3 times, and 1.7 times increased risk for prostate cancer, respectively, compared with men whose fathers were in the first quartile. The association with older parental age was stronger for men with early-onset prostate cancer (age < 65 years) than for men with late-onset disease (> 65 years). The authors of that study proposed that their findings may reflect an increased germ cell mutation rate in older fathers.126

3.5.1.2 Hormones

Endogenous hormones, especially androgens, are required for growth, maintenance, and function of the prostate.127 Prostate cancer growth rates often can be manipulated with hormonal therapy.5 The progression of prostate cancer from a subclinical form to a clinically important form may result in part from altered hormone metabolism.128 Patients who had prostate cancer had lower serum androgen bioactivity compared with a control group of men who had BPH in a recent study in Finland.129 The results of a prospective cohort study in France130 suggested that the absence of an association between plasma levels of androgens and prostate cancer risk may be a result of the complex and inverse associations of androgenicity to insulin-like growth factor I (IGF-I), insulin, and leptin.

The principal androgenic hormone in men is testosterone.127 Elevated levels of testosterone and its metabolite dihydrotestosterone, over many decades, may increase prostate cancer risk.131 However, epidemiologic studies of hormones and prostate cancer have been inconsistent. Serum testosterone levels in men with prostate cancer may be elevated,132 depressed,133 or similar134, 135 to the levels in men without cancer. Serum levels of dihydrotestosterone were not associated with prostate cancer risk in a nested case–control study from Norway.135 Paradoxically, serum testosterone levels are declining in men at the age of peak cancer incidence.

The enzyme 5-α-reductase converts testosterone to the principal intracellular androgenic hormone, dihydrotestosterone.136 The activity of 5-α-reductase, as assessed by the measurement of 3-α-androstanediol glucuronide and androsterone glucuronide concentration, was associated with prostate cancer risk.137 Conversely, a nested case–control study136 found that neither metabolite was associated strongly with risk (3-α-androstanediol glucuronide: OR, 1.37; 95% CI, 0.73–2.55; androsterone glucronide: OR, 0.85; 95% CI, 0.44–1.65). In another nested case–control study, there was a slightly increased risk associated with levels of androstanediol glucuronide (OR, 1.16; 95% CI, 0.86–1.56) and androsterone glucuronide (OR, 1.13; 95% CI, 0.84–1.53).138 The Type II isoenzyme of 5-α-reductase is most likely a more important determinant of risk, and serum levels of 3-α-androstanediol glucuronide and androsterone glucuronide are influenced by both Type I and Type II isoenzymes.138 The resulting misclassification may have biased the OR toward the null. Clinical trials are ongoing to assess the effect of inhibition of intraprostatic 5-α-reductase activity on prostate cancer risk.139

The Massachusetts Male Aging Study recently attempted to discover the role of serum hormones in prostate cancer risk. Of the 17 hormones tested, only androstanediol glucuronide was associated significantly with the risk of prostate cancer; however, even this risk was marginal (OR, 0.2; 95% CI, 0.04–0.6). The results of that study called into question whether serum hormones collected during midlife accurately can assess the risk of prostate cancer.140

A decreased prostate cancer risk was noted among diabetics in the Health Professionals Follow-Up Study.141 Lower total and free testosterone levels142, 143 are among the numerous metabolic and hormonal aberrations associated with diabetes.

Prolactin is a pituitary peptide hormone that stimulates prostate growth in experimental models. One case–control study in Sweden144 studied the association between prolactin and prostate cancer risk; no significant association was found, and elevated levels of prolactin were not related to increased prostate cancer risk.

Leptin is an adiposity-related hormone. Men with high-volume cancer (tumors > 0.5 cc in volume or tumors with histologic evidence of extraprostatic extension but without metastases) had higher leptin concentrations at the time of prostate cancer diagnosis (OR, 2.35; 95% CI, 1.01–5.44).145 Another study found that intermediate (but not high) plasma leptin concentrations were associated with prostate cancer risk.146 A study performed in Turkey147 found that leptin influenced cellular differentiation and the progression of prostate cancer through testosterone and through factors related to obesity.

The role of female hormones in the etiology of prostate cancer is unclear. Serum estrone and free estradiol-17β were elevated in young black men relative to white men, whereas Japanese men had lower free estradiol-17β levels compared with age-matched white men.148, 149 An endocrine milieu of rising estrogenic stimulation and decreasing androgenic influence may contribute to risk by stimulating cell proliferation and altering oxidative stress.150 However, the incidence of prostate cancer appears to be lower in men with cirrhosis, a condition associated with increased circulating levels of estrogen and decreased levels of testosterone.132, 151 In one case–control study, the authors concluded that estrogen and aromatase may play a role in prostate cancer.152

3.5.1.3 Race

Race-related differences in prostate cancer risk may reflect multiple factors, including exposure differences, particularly dietary differences; differences in detection; and genetic differences. The highest incidence rates for prostate cancer in the world are among African-American men. For the period 1988–1992, race-specific U.S. incidence rates ranged from 24.2 per 100,000 for Koreans, 89.0 per 100,000 for Hispanics, 134.7 per 100,000 for whites, and 180.6 per 100,000 for African Americans. Black men in the U.S. are more likely to present with advanced stage cancers than white men, and their stage-specific mortality is worse, especially among younger men.23, 153 African-American and Hispanic men commonly are diagnosed at a significantly younger average age (mean, 63.7 years and 65.2 years, respectively) compared with white men (mean, 68.1 years).108

In recent years, prostate cancer incidence has risen considerably in many countries, including countries in which prostate cancer risk is considered low. Mortality rates also are on the rise, with the most significant increase noted in Asian countries rather than in higher risk Western countries. African Americans still present the most cases of prostate cancer, at rates that are 50–60 times that of Shanghai, China.154 This apparent increase is most likely a result of earlier detection efforts.

African-American men have a higher intake of dietary fat, and this may contribute to their higher risk.155 Japanese men consume a relatively low-fat diet; it is worth noting that, as the fat content of the Japanese diet has increased toward Western levels, the incidence of prostate cancer has increased.156 Early detection efforts and detection bias may account for some of these alterations.

Several migrant studies have found that prostate cancer rates shift toward those of the host country. For example, the incidence rate of Japanese men in the U.S. is intermediate between the low rate of Japanese men in Japan and the high rate of white men in the U.S.157, 158 When Japanese men move to the U.S., their incidence and mortality rates increase toward those of American men.159 The risk among Japanese emigrants was related inversely to age at migration and was related directly to the length of time in the new environment.159 White men in the U.S. have a higher prostate cancer rate compared with Chinese men in China, and Chinese-American men have an intermediate rate.160, 161 These studies suggest that at least some of the difference in risk relates to environmental factors. Conversely, there are considerable differences in the reporting of cancer cases around the world, and some of the variance likely is attributable to reporting and detection bias.

Some of the differences in risk between black and white Americans may reflect access to care. However, there also are race-related differences in the decision-making process regarding whether to seek medical attention and follow-up, and these differences are independent of equal access.162 Access to care may influence disease stage at diagnosis and, hence, poorer survival and higher mortality among African Americans,163 but it does not explain the higher incidence rate. PSA screening is more common among white American men than among African-American men, which should favor a higher incidence among white men, yet the opposite has been found.

Recent studies have begun to focus on PSA screening and the answers it may hold for race-related differences in prostate cancer risk. In one study, it was indicated that there were higher PSA levels in African-American men both with and without prostate cancer. The factors that cause these higher PSA levels in African-American men are not understood completely, but it is believed that they are a combination of biologic, environmental, and socioeconomic causes.164 However, another study conducted at the Albert Einstein Medical Center found absolutely no difference in PSA level between black Americans and white Americans.165 One study attempted to explain the difference in PSA level by examining the prevalence and extent of prostatic inflammation in black versus white Americans. Although the percentage of white men with prostatitis was slightly higher than the percentage of black men, the difference was not significant (P = 0.299), and the extent of inflammation was not correlated significantly with differences in the PSA level. Therefore, prostatic inflammation could not account for the racial difference in PSA levels in that study.166 A Mississippi study found racial differences in relations between the percent of free PSA and cancer detection in men with suspected prostate cancer and total PSA in the range from 2.5 ng/mL to 9.9 ng/mL.167 High-grade PIN is more common among black American men compared with white American men, but it does not contribute to racial differences in PSA concentrations among men who have no clinical or histologic evidence of carcinoma.167

Differences with regard to serum concentrations of androgens and their metabolites may explain the racial differences in prostate cancer risk. Young African Americans have higher circulating levels of testosterone than their white counterparts148;although a more recent study found no racial difference in serum androgen levels.165 Sex hormone-binding globulin (SHBG) levels are higher in young adult African Americans, and this may contribute to a subsequent increased risk of prostate cancer.168 The activity of 5-α-reductase is lower among Japanese men than among African-American and white men.137, 169 Differences in the frequency of polymorphic variation in the SRD5A2 gene, which codes for type II 5-α-reductase, exist between racial groups.170 Racial differences in the concentrations of IGF-I and IGF-binding protein-3 have been reported. White Americans had the highest level of IGF-I, whereas African Americans had the lowest level. Although differences in circulating IGF-I do not seem to account for the increased prostate cancer risk among African-American men, their lower levels of IGF-binding protein-3 may contribute to their level of risk.171 This may represent a risk factor for prostate cancer among African-American men and, on a larger scale, may be useful in predicting early-onset prostate cancer in all men.172

There are racial differences in allelic frequencies of the microsatellites at the AR locus. African Americans have a high prevalence of short CAG microsatellite alleles and a low frequency of 16 GGC repeats compared with whites and Asians. Shorter CAG microsatellite alleles of the AR gene may promote prostate cell growth, resulting in a higher prostate cancer risk.136, 173 Prostate cancer occurs 5 years later in Japanese men compared with the age of onset in Swedish men. This is attributed to a larger content of vitamin A and D receptors in the Japanese. Japanese men also have longer CAG repeats.174 Panz et al. found that, although there were small variations in the number of CAG repeats between black and white patients in their study, this factor did not appear to be a strong indicator of risk. However, those authors did conclude that the size polymorphism may be involved in the pathogenesis or aggressiveness of prostate cancer.175

Racial differences also may result from polymorphic variation in the VDBP gene, which encodes serum vitamin D-binding protein.176 One study found a significant association between the CYP17 polymorphism and prostate cancer risk among the Japanese.177

3.5.1.4 Age-related changes, including oxidative stress

The aging prostate may be the most disease-prone organ of the human body. The frequencies of BPH, PIN, and prostate cancer increase dramatically with age, beginning with low frequencies in middle-aged men and progressing to > 90% by age 90 years.5, 25, 178, 179 Although different risk factors have been identified for BPH and prostate cancer, a predominant risk factor common to both diseases is aging.

During aging,180, 181 there is a progressive accumulation of DNA adducts182 and an increase in DNA strand-break frequency183 in most tissues. It is believed that these age-related changes are caused by oxidative stress, which arises as a result of an imbalance in cellular prooxidant-antioxidant status.181, 184 Cellular oxidants, such as free radicals and reactive oxygen species (ROS), are produced during natural metabolic processes. ROS are highly reactive and potentially damaging to cells, because they directly damage macromolecules and organelle functions.181, 185, 186 Damage to DNA by ROS results in single-strand and double-strand breaks, apurinic and apyrimidinic sites, ring-saturation thymine derivatives, and adduct formation.182, 187 In addition, ROS can catalyze the oxidative modification of proteins, including enzymes involved in DNA repair.185 Together, these direct and indirect influences of ROS on DNA create an ideal environment for mutagenesis and tumor initiation. Indeed, Ames and Gold estimated that ROS-related DNA base modifications are in the range of ≈ 100,000 per cell per day in the rat.188 Thus, it is tempting to speculate that a fraction of these damages escape repair and accumulate in the cell.

Most cells in young, healthy individuals are equipped with adequate antioxidant defense mechanisms to protect against free radicals and ROS.184 The first line of defense is afforded by a class of ROS detoxification enzymes. Superoxide dismutases (SODs) convert superoxide to hydrogen peroxide (H2O2), whereas catalase and selenium-containing glutathione peroxidases are the major enzymes responsible for the removal of hydrogen peroxide. Overexpression of these enzymes in cells of the fruit fly prevents cellular aging and significantly extends the life span of the insect.181 The selenium-dependent glutathione peroxidases and glutathione reductases function in combination to remove H2O2 and regenerate the reduced form of glutathione. A decline in ROS detoxification enzyme activities occurs in most tissues with aging.181

Antioxidant vitamins and some nonprovitamin A carotenoids constitute a second class of intracellular antioxidants.189 α-Tocopherol is present in membranes and lipoproteins. It terminates the chain reaction of lipid peroxidation by scavenging intermediate peroxyl radicals. The resulting tocopherol radical can be converted back to α-tocopherol by vitamin C. The carotenoids lutein and lycopene possess exceptionally high antioxidant activity, and lycopene is a very effective quencher of the ROS singlet oxygen. Finally, metallothioneins (MTs), high-affinity Cd-Zn-Cu-binding proteins,190 have recently been recognized as effective scavengers of ROS, especially hydroxyl radicals, by virtue of their sulfhydryl-rich nature.191, 192 MT protects against heavy metal toxicity190 and DNA oxidative damage.193 This protein is inducible readily by Cd or Zn challenge194 and by oxidative stress inducers.191, 192 In summary, eukaryotic cells are equipped with multiple antioxidant defense mechanisms; however, these mechanisms are not completely efficient and, at times (especially with advanced age), are inadequate. Thus, oxidative damage still occurs.

Apart from the induction of genotoxicity, a chronic prooxidative state has other profound deleterious effects, including induction of apoptosis, an imbalance in cell proliferation, modification of intracellular and extracellular structural proteins, damage and leakage of cell membranes, and cellular aging. All of these are direct or indirect promoters of cancer and degenerative diseases.195–197

Age-related changes in oxidative stress status have been implicated widely in atherosclerosis, cataracts, diabetes, Alzheimer disease, muscular dystrophy, arthritis, and several cancers,181, 184, 185, 195, 196, 198 but similar changes rarely are considered as etiologic factors of prostate cancer. Hence, studies of age-dependent oxidative stress and prostatic carcinogenesis largely are lacking, although circumstantial evidence does exist.

There are higher levels of DNA base lesions and lower activities of antioxidant enzymes in BPH than in normal prostatic tissue.199 Hydroxyl radical-related DNA alterations in prostate cancer differ from those observed in normal prostate samples.200 Dietary fat is a strong risk factor for prostate cancer.155 Dietary fat may act on tumor promotion and progression by peroxidation or autoxidation of protein or DNA to create reactive intermediates,201, 202 and fat also may modulate oxidative stress in cancer cells.203, 204 In the Alpha-Tocopherol Beta Carotene Cancer Prevention Study from Finland, it was shown that daily intake of the antioxidant vitamin, α-tocopherol, offered protection against prostate cancer by reducing the incidence in smokers by 34%.205, 206 Additional evidence in support of an association between increased intake of antioxidant nutrients and decreased prostate cancer risk was provided by two recent investigations. Frequent consumption of tomato-derived food products, rich dietary sources of lycopene,207 which is a potent antioxidant nonprovitamin A carotenoid, was associated with a 35% reduction in prostate cancer risk in a large cohort of healthy men.179 Daily intake of 200 μg of high-selenium yeast significantly reduced prostate cancer incidence by 63% and mortality by > 30%.208 Taken together, these studies support the hypothesis that oxidative stress, which may increase naturally within the aging prostate, is an important risk factor for prostate cancer. However, our current knowledge of the relation between aging and prooxidation-antioxidation homeostasis of the human prostate remains virtually nonexistent.

In animal studies, Ghatak and Ho demonstrated an age-dependent increase in lipid peroxidation and a concomitant decrease in ROS detoxification enzymes in the prostate of Noble rats.209 When prostatic adenocarcinoma was induced with sex hormone administration, increased lipid peroxidation, DNA damage, and specific DNA adducts were detected in the cancerous dorsolateral prostate but not in the cancer-free ventral lobe.210, 211 Prostate cancer cell lines expressed higher levels of SOD, catalase, glutathione reductase, glutathione-S-transferase (GST), and glutathione peroxidase than primary benign prostatic epithelial cells.212 It is noteworthy that androgen treatment of androgen-dependent LNCaP cells, but not androgen-refractory DU145 cells, elevated lipid peroxidation and down-regulated ROS detoxification enzyme activities.213 Hence, results from animal models and cell studies are consistent with the concept that aging and hormones independently influence oxidative stress, which, in turn, may contribute to the pathogenesis or progression of prostate cancer.

3.5.2 Exogenous factors

3.5.2.1 Diet and nutrition

Descriptive epidemiologic studies of migrants, geographic variations, and temporal studies suggest that dietary factors may contribute to prostate cancer development.214 However, despite the large number of studies supporting this hypothesis, not all studies have shown an association between dietary factors and prostate cancer risk.215 A strong positive correlation exists between prostate cancer incidence and the corresponding rates of several other diet-related cancers, including breast cancer and colon cancer.216

3.5.2.1.1 Fat intake (dietary fat)

There is a strong, positive correlation (0.74) between prostate cancer incidence or mortality and fat consumption in multiple countries217, 218 and within the U.S.219 Many case–control studies have examined the association between dietary fat and prostate cancer,155, 220–235 although only a few studies155, 220, 225, 227–229 adjusted for energy intake. Those studies differed in terms of the selection of controls (hospital or population) and the method of dietary assessment (direct or indirect). Some studies inferred fat intake from the frequency of consumption of high-fat food, such as meat or dairy products,221, 226, 231 whereas other studies used food-composition data to approximate actual fat intake.155, 220, 222, 223, 225, 227, 228, 234, 235 Despite these differences, all but five studies223, 224, 227–229 found a positive association with total fat intake. One study229 found a positive association with dietary fat that disappeared after adjustment for energy. Two studies227, 228 were part of a Pan-Canadian investigation that adjusted for energy intake but may have suffered from selection bias, because response rates were low for cases and controls. Two studies were conducted in Japan,223, 224 where it is known that fat consumption is low compared with the United States.

A cohort study measured fat intake and adjusted for energy intake, and a significant positive association was noted between increased fat intake and the risk of advanced prostate cancer.236 Only one other cohort study measured fat based on nutrient intake,237 and none of the others adjusted for energy intake.

A positive association between the consumption of foods high in fat and the risk of prostate cancer was reported in three cohort studies.238, 239 Four other studies did not detect an association.237, 240–242 Two studies had limited food frequency data.240, 241 Severson et al.237 detected a weak association with eggs, margarine, butter, and cheese as a group, but not with total fat; however, they estimated total fat by using a 24-hour food recall survey, a method with limited usefulness in estimating individual dietary intake. Tzonou et al.231 also found an association between butter and the risk of prostate cancer. A case–control study in Greece found a positive association for dairy products, butter, and seed oils.243 Although the study by Veierod et al.242 failed to note a significant association between energy-adjusted fat consumption and prostate cancer risk, there was a significant association with consumption of hamburgers and meatballs.

Four case–control studies found a positive association between dietary intake of polyunsaturated fats and prostate cancer risk.225, 228, 230, 231 One cohort study also noted an inverse association between polyunsaturated fat and risk, although the relation was not significant (RR, 0.78; 95% CI, 0.56–1.10).215 Conversely, one study noted that higher levels of polyunsaturated fatty acids were associated with less aggressive and extensive prostate cancer.244 There was a positive, significant association between dietary α-linolenic acid, an essential polyunsaturated fatty acid, and prostate cancer risk (RR, 3.43; 95% CI, 1.67–7.04)236 and between plasma α-linolenic acid and risk.233, 245, 246 However, no clear, linear relation was present across quartiles of exposure in the studies by Gann et al.245 and Giovannucci et al.236

A weak, inverse relation was observed by Giovannucci et al.236 and by Gann et al.245 between linolenic acid, another polyunsaturated fatty acid, and risk. Because linolenic acid and α-linolenic acid compete for some of the same enzymes, low levels of linolenic acid may exaggerate further the risk associated with α-linolenic acid.236 Harvei et al.247 noted a negative association between prostate cancer risk and the ratio of serum levels of linolenic acid to α-linolenic acid. However, a significant, negative association between linolenic acid consumption and risk was found in a case–control study of African Americans age ≥ 50 years248 and, more recently, in The Netherlands Cohort Study.215 Newcomer et al. found a 2.6-fold increased risk of prostate cancer with greater intake of α-linolenic acid.249

Dietary saturated fat was associated with an increased risk of prostate cancer in several case–control studies,155, 220, 222, 225, 230 and other case–control studies have found increased risks associated with eating animal fat,224, 232–234 high-fat animal products,239 or meat.221, 250 An inverse association with saturated fat was found by Rohan et al.227 However, response rates were low in that study, raising the possibility of selection bias. A recent case–control study in Athens, Greece, found no statistically significant association between saturated and monosaturated fat intake and risk, even though the results supported a positive association for polyunsaturated fat.231

Findings of an association between saturated fat and risk of prostate cancer from cohort studies are mixed. No increased risk was associated with dietary saturated fat in the Health Professionals cohort, although an association was noted with fat from red meat236 and with meat consumption in the Lutheran Brotherhood cohort study,241 among men of Japanese ancestry in Hawaii,237 and in The Netherlands Cohort Study.215 Meat consumption was associated with an increased risk among Adventists,238 and red meat consumption was associated positively with risk in the Physicians Health Study cohort.245 Red meat also was associated with risk in a case–control study that was conducted in Uruguay.251

Essential fatty acids in fish inhibit the growth of prostate cancer cells in vitro and in vivo252; however, there have been only three studies to date that have examined the effect of these acids (eicosapentaenoic acid [EPA] and docosahexaenoic acid [DHA]) in the epidemiology of prostate cancer risk. A case–control study in New Zealand253 and a population-based cohort study in Sweden252 found reduced risks of prostate cancer in men with the highest consumption of fatty fish. In both studies, the authors believed that this finding was a result of inhibition of arachidonic acid-derived eicosanoid biosynthesis. The third study to include EPA and DHA in its analyses found no association between the acids and prostate cancer risk.254

It is unclear how dietary fat may increase the risk of prostate cancer, but a number of mechanisms have been proposed. These include dietary fat-induced alterations in hormonal profiles,255 the effect of fat metabolites as protein or DNA-reactive intermediates,202 and dietary fat-induced elevation of oxidative stress.150 It is likely that the relation between dietary fat and prostate cancer risk is complex, involving interplay of fat with other dietary factors, such as antioxidant vitamins and minerals, as well as genetic factors that influence susceptibility.256

3.5.2.1.2 Vitamins.

Vitamin A

Retinoids include retinol and its metabolic derivatives, retinoic acid and retinyl aldehyde, and a group of structurally related carotenoids.150 Retinoids regulate epithelial cell differentiation and proliferation.257 Vitamin A refers to all substances that possess the biologic properties of retinol.258 Vitamin A may be ingested either as a provitamin or as a preformed vitamin. Provitamin A originates from certain carotenoids found in plant sources, including carrots, yellow squash, green leafy vegetables, corn, tomatoes, and oranges.259 On average, one-sixth of β-carotene and one-twelfth of the intake of other dietary provitamin A carotenoids are converted to vitamin A.

Only 50 of the approximately 700 carotenoids found in nature are converted into vitamin A. However, most carotenoids, including those with provitamin A activity, can act as antioxidants under certain conditions.259 Thus, any association between prostate cancer and carotenoids does not necessarily require conversion to vitamin A.214 Therefore, it is difficult to interpret studies that have used an index of vitamin A that combines the dietary intake of both preformed and provitamin A.179, 220, 223, 225, 227, 228, 234, 235, 260–262 It may be more useful to examine the association between preformed vitamin A and risk of prostate cancer separately from the association of carotenoids and prostate cancer.263

Preformed vitamin A is found naturally only in food from animal sources, such as liver, dairy products, and fish.259 A positive association between preformed vitamin A intake and prostate cancer was reported in five studies179, 226, 261, 262, 264; although in two studies,261, 262 the association was restricted to patients of certain ages. A slightly decreased risk with increased consumption was found in two Canadian case–control studies,227, 228 but those studies suffered from low response rates.

The association of serum vitamin A or serum retinol and prostate cancer is uncertain.262, 265–267 Increased risk was associated with lower serum retinol levels in a hospital-based case–control study from The Netherlands.265 The design of the study suggested that a treatment effect or an effect from the disease process itself could not be dismissed. Low serum retinol levels may be a metabolic consequence of cancer rather than a precursor.268 Three nested case–control studies yielded different results. One study suggested an inverse relation,262 a study of Japanese Americans in Hawaii reported no association,269 and a third group reported a positive association.266

Data from the National Health and Nutrition Examination Survey showed an increased risk of prostate cancer in men who had serum vitamin A levels in the lowest quartile compared with men who had levels in the highest quartile (RR, 2.2; 95% CI, 1.1–4.3).267 In contrast, in a Canadian study, men with serum vitamin A levels in the highest quartile were at the greatest risk (RR, 2.0; 95% CI, 1.1–3.5). Reasons for the discrepant results are not readily apparent. The two cohort studies were similar in many respects, including the time period of the study, the length of follow-up, and the overrepresentation of elderly and low-income individuals.263

The relation between dietary intake of carotenoids (primarily β-carotene) and the risk of prostate cancer has been investigated extensively in case–control studies221, 223–228, 235, 261, 270–272 and in cohort studies.179, 237, 238, 262, 264, 273, 274 Most of those studies were nutrient-based, the preferred design that takes into account the potentially confounding effect of other nutrients contained in the same food item.220 Those studies reported positive associations,226 negative associations,223, 224, 235 and null associations.179, 227, 264, 273 In two reports, the direction of the association differed by age group.225, 262 Significantly, a clinical trial of α-tocopherol and β-carotene supplements among cigarette smokers noted an overall increased risk of prostate cancer with β-carotene use.206, 275 However, men in the lowest quartile of initial plasma β-carotene had a 32% decreased risk of prostate carcinoma after β-carotene supplementation. These data suggest that men with dietary practices or metabolic profiles resulting in low plasma concentrations of β-carotene represent a cohort that may benefit from daily dietary supplements of β-carotene.276 Serum β-carotene was associated positively with prostate cancer risk266 in one study, but not in three other studies.262, 265, 271

Vitamin C

Vitamin C is a scavenger of ROS and free radicals.197 Vitamin C inhibits cell proliferation in prostate cancer cell lines.277 However, most epidemiologic studies of vitamin C and prostate cancer risk have found no significant association.117, 220, 222, 223, 229, 248, 273, 278, 279 One report found a positive association (OR, 2.32; trend P < 0.01) that was stronger in men age > 70 years (OR, 3.41; trend P < 0.05).234 The results of another study supported an inverse association (OR, 0.6; 95% CI, 0.3–0.9).233 Two other studies222, 225 reported nonsignificant, increased risk estimates of ≈ 50% among men in the highest quartile of vitamin C intake compared with men in the lowest quartile. A recent case–control study in Uruguay251 found that high consumption of vitamin C reduced risk compared with men in the lowest quartile of vitamin C intake (OR, 0.4; 95% CI, 0.2–0.8). Another study found that the level of vitamin C in men who had prostate cancer was lower, but the difference was not statistically significant compared with men who did not have cancer.280

Vitamin D

Vitamin D deficiency may be a risk factor for prostate cancer.281–285 One study failed to find an association between serum vitamin D concentration and prostate cancer; however, the lack of sufficient numbers of study participants who had low vitamin D levels may have influenced the results.286 A Japanese case–control study287 found no association between vitamin D receptor gene polymorphisms and familial prostate cancer. The hormonal form of vitamin D, 1α-25-dihydroxyvitamin D (1,25-D), inhibits invasiveness of prostate cancer cells in vitro288 and has antiproliferative and prodifferentiation effects in the Dunning rat model of prostate cancer.289

Low serum levels of 1,25-D, a vitamin D metabolite, were associated significantly with an increased risk of clinically detected prostate cancer among older men, particularly in men with low levels of 25-dihydroxyvitamin D (25-D) (OR, 0.41).282 The protective effect was attributed to seasonally lower summer levels of 1,25-D in the case group.176 However, a nested case–control study failed to confirm those findings.290 Higher levels of 1,25-D or 25-D were not associated with a reduction in risk, although a nonsignificant, inverse association (OR, 0.67) was observed among men simultaneously in the highest quartiles of both metabolites relative to those simultaneously in the lowest. Similarly, a study of prediagnostic serum vitamin D metabolite levels291 failed to support an association between 1,25-D and the risk of prostate cancer. Serum levels of 1,25-D and 25-D were comparable for cases and controls, and no difference was observed for older men or for men with advanced cancer. However, the study had limited power. Serum was collected in the autumn, and serum levels of 25-D vary by season, raising the concern that case–control differences also may vary by season.

The potential protective effect of 1,25-D may be restricted to the biologically active free 1,25-D.292, 293 A case–control study reported that men with prostate cancer had significantly lower serum levels of free 1,25-D. In contrast, another report did not find a lower free 1,25-D serum concentration in men with prostate cancer.176

Vitamin E

Vitamin E (α-tocopherol) is an antioxidant that inhibits prostate cancer cell growth in vitro through apoptosis.294 Two epidemiologic studies suggest that vitamin E is protective, at least among cigarette smokers. Two prospective cohort studies found that a low serum vitamin E level in smokers was associated with an increased risk of prostate cancer.278, 295 Recent case–control studies also have found a strong, inverse association between vitamin E intake and prostate cancer risk.231, 251 The Alpha-Tocopherol Beta-Carotene Cancer Prevention Study, a randomized intervention trial of cigarette smoking men in Finland between the ages of 50–69 years,206, 275 found that men who consumed 50 mg of α-tocopherol daily for 5–8 years reduced their incidence of prostate cancer by 32%, and mortality from prostate cancer decreased by 41% compared with men who did not take α-tocopherol.206 In a case–control study, men who received α-tocopherol supplementation had significantly lower serum androstenedione and testosterone concentrations compared with men who received placebo. This suggests that long-term α-tocopherol supplementation decreases serum androgen levels, which, in turn, may reduce the risk of prostate cancer.296 Conversely, another case–control study noted no significant difference in risk according to α-tocopherol intake after adjustment for energy intake.229

Further evidence is now available to support the hypothesis that vitamin E is protective against prostate cancer in smokers. In a 17-year follow-up of the Prospective Basel Study in Switzerland, Eichholzer et al.280 reported a significantly increased risk among smokers with low plasma levels of vitamin E (RR, 3.26; 95% CI, 1.27–8.35). It is believed that the combination of these factors, smoking and low plasma vitamin E, is more detrimental than each factor alone (low vitamin E levels in nonsmokers: RR, 0.76; 95% CI, 0.25–2.36; smokers with normal vitamin E levels: RR, 1.28; 95% CI, 0.49–3.33). However, it is important to note that these RR values may be imprecise due to the small numbers of participants examined (n = 30 participants). The possible prostate cancer protective effects of other forms of vitamin E, namely, γ tocopherol or mixed tocopherols, have not been studied.

3.5.2.1.3 Trace elements

Zinc

The prostate has a higher concentration of zinc than any other organ in the human body,162, 297–299 with highest content in epithelial cells and the lowest content in the stroma.300 Cancer-bearing prostates have lower levels of zinc compared with noncancer-bearing prostates.301 Most zinc is found in the nuclear fraction, but substantial levels are found in the cytosol.302 The physiologic role of zinc in the prostate is unknown. In vitro studies have shown that binding of androgen-AR complexes to cell nuclei is enhanced by zinc,303 and total androgen uptake by prostatic tissue is increased.302 Large amounts of zinc are secreted by the prostate into the seminal fluid, where it most likely plays a role in prolonging the viability of ejaculated sperm.297

Tissue and serum zinc contents are lowest (> 90% reduction) in men who have prostate cancer compared with healthy men or men who have prostatitis and BPH.298, 301, 304–308 In men with prostate cancer, serum zinc declined further after androgen-deprivation therapy.307, 309 Zinc content is lower in older men than in younger men.310 Collectively, these data suggest that zinc homeostasis is disrupted during aging and prostatic carcinogenesis. Alternatively, a reduction in zinc content in tissue and serum may be a risk factor for cancer. Currently, the relation between dietary zinc intake and prostate cancer risk is uncertain. Serum levels of zinc and prostate cancer risk do not appear to be related,162 but a case–control study from Hawaii that adjusted for age and ethnicity222 and a more recent study in Maryland311 found that increased zinc consumption was associated positively with prostate cancer risk. A recent study299 confirmed these findings and suggested that zinc inhibits human prostatic carcinoma cell growth, possibly due to the induction of cell cycle arrest and apoptosis. Two other case–control studies found no meaningful association.225, 229

Zinc metabolism is inhibited by cadmium, and occupational exposure to cadmium is associated with an increased risk of prostate cancer.312 However, the mechanism by which cadmium exerts its effect on zinc availability is unknown.

Selenium

Selenium is an essential trace element that inhibits viral and chemical, carcinogen-induced tumors in animals.313 A cancer chemopreventive role for selenium is plausible biologically, but the evidence in humans is limited.

In a recent randomized controlled trial in which the primary endpoint was the recurrent incidence of skin cancer,208 daily supplementation with 200 ng of selenium in the form of high-selenium yeast significantly reduced prostate cancer incidence (RR, 0.37; 95% CI, 0.18–0.71).314 Previous studies of serum selenium levels had not reported a significant association with prostate cancer risk,266, 315, 316 although there was a study that included only 11 patients with prostate cancer. One case–control study found an inverse association between serum selenium at diagnosis and prostate cancer; however, selenium status may have been altered by the cancer.317 Another case–control study found a trend between dietary selenium and risk that was not significant.225 A nested case–control study found an inverse association for selenium and prostate cancer; however, the association was present mainly in current or past smokers rather than nonsmokers.318 A different case–control study found that individuals with low plasma selenium levels were at an increased risk of prostate cancer. Those authors suggested that supplemental selenium may reduce this risk.319

One particularly interesting study requested that participants send in samples of their toenail clippings for analysis in regard to selenium levels. The selenium level in the toenails varied among the men, with quintile medians ranging from 0.66 to 1.14 μg/g. Higher selenium levels were associated with a reduced risk of advanced prostate cancer (highest quintile vs. lowest quintile: OR, 0.49; 95% CI, 0.25–0.96).320

Calcium

Many studies have noted a positive association between milk consumption and prostate cancer risk221, 226, 231, 250, 270, 321–325; however, no association was observed in other studies.237, 254, 326–328 It is unclear the extent to which the consumption of calcium, compared with dairy fat, contributes to an increased risk. Men who had the highest calcium intake had a relative risk of almost 3.0 for advanced prostate cancer compared with men who had the lowest calcium intake in the Health Professionals Follow-Up Study.283 A possible mechanism is that high calcium intake suppresses the formation of 1,25-D from 25-D.

One study found a potentially important interaction between calcium, phosphorus, and prostate cancer risk. Those investigators were unable to find independent associations for each nutrient, but they found that men who had low calcium intake and high phosphorus intake had an RR of 0.6 (95% CI, 0.3–1.0) compared with men who had low intake of both nutrients.329

3.5.2.1.4 Alcohol

Alcohol intake may increase metabolic clearance of testosterone, suggesting a possible protective effect in prostate carcinogenesis.330 However, almost all studies to date have found no significant association with risk.229, 235, 239, 241, 331–344 Conversely, 2 case–control studies have implicated heavy alcohol consumption: Hayes et al.345 found an elevated risk for men who had 22–56 drinks per week (OR, 1.4; 95% CI, 1.0–1.8) and ≥ 57 drinks per week (OR, 1.9; 95% CI, 1.3–2.7) compared with never-users.

A cohort study in Iowa346 also found an association between alcohol consumption and risk. Risk increased with the amount of alcohol consumed (consumption of < 22 g per week: RR, 1.1; 95% CI, 0.6–2.1; consumption of > 96 g per week: RR, 3.1; 95% CI, 1.5–6.3). A past history of heavy alcohol consumption was associated inversely with prostate cancer risk, although there were relatively few of cases, and the biologic plausibility of the inverse relation was not clear.347 In the Harvard Alumni Health Study, it was found that moderate alcohol consumption had a positive association with prostate cancer risk. In that report, liquor consumption, but not wine or beer consumption, was implicated in an increased risk of developing prostate cancer.348 In contrast, a case–control study in Montreal found that beer consumption had the strongest association with an increased risk of prostate cancer. That study also implicated drinking at an early age in raising the risk, with an OR of 3.8 (95% CI, 1.6–9.3) for those who began consuming alcohol before age 15 years.349

Alcohol has no effect on circulating testosterone levels, perhaps accounting for the absence of evidence for a protective effect. One study found that alcohol intake appeared to reduce serum levels of IGF-I, a growth factor believed to contribute to a higher risk of prostate cancer.350

3.5.2.1.5 Smoking

Cigarette smoke is a source of exposure to cadmium,312 and smoking appears to increase circulating levels of androgens in men.351 There have been numerous case–control studies of cigarette smoking and prostate cancer,117, 229, 235, 242, 270, 332, 334, 339, 340, 343, 344, 352–366 but only seven studies reported either a significant association229, 344, 357, 361, 364, 366 or a “marked” disparity in the proportion of smokers between cases and controls.354 Despite increased risks in current smokers (OR, 1.5; 95% CI, 1.0–2.4) and former smokers (OR, 1.4; 95% CI, 1.0–1.5) of ≥ 40 cigarettes per day, the lack of consistent findings in population subgroups and the lack of a clear dose-response relation argues against the existence of a causal association.361 However, a more recent case–control study366 did find a dose-response relation for men with > 40 pack-years of smoking (OR, 1.6; 95% CI, 1.1–2.2). In that study, it also was found that cessation of smoking resulted in a decline in risk (trend P = 0.02), and the authors concluded that smoking should be added to the list of risk factors for prostate cancer.

Cohort studies of cigarette smoking and prostate cancer incidence have produced conflicting results.263 Five studies237, 238, 346, 367, 368 reported no association, but two studies detected a significant positive relation.335, 337 In the Iowa 65+ Rural Health Study,337 men who smoked ≥ 20 cigarettes per day were at nearly 3-fold increase in risk relative to nonsmokers. The positive association reported by Hiatt et al.335 was limited to men who smoked more than one pack of cigarettes per day.

Cohort studies that were conducted prior to 1990240, 369–371 did not observe an association between cigarette smoking and prostate cancer mortality among former or current smokers compared with never-smokers.369–371 Since 1990, 4 of 5 published cohort studies have reported a positive association.241, 372–374 One report found an 80% increase in mortality risk, but there was no evidence of a trend in effect.241 A significant 20–35% increase in mortality risk was reported in 3 studies.372–374 In one of those studies,372 a trend in effect was noted, with the highest mortality experienced by those who smoked ≥ 40 cigarettes per day (RR, 1.5; 95% CI, 1.2–1.9), and a lesser effect was observed for former smokers (RR, 1.13; 95% CI, 1.03–1.24). In contrast, a 40-year follow-up of nearly 35,000 British male physicians found that prostate cancer mortality rates were virtually identical in current smokers and never smokers.375 Although one study could not indicate a relation between smoking and prostate cancer incidence, the findings did support the idea that recent tobacco use had a substantial impact on the occurrence of fatal prostate cancer.376 A recent study found that smoking resulted in higher disease stage at diagnosis.377

In 1996, participants of an international consensus conference on smoking and prostate cancer unanimously agreed that there was inadequate evidence to associate smoking with prostate cancer incidence, although smoking was associated with mortality.378 It is likely that smoking adversely affects survival in patients with prostate cancer.374, 379

3.5.2.1.6 Other

Diabetes mellitus

One study examined the role of diabetes mellitus as protective against prostate cancer. After conducting a case–control study in New York City, the authors found that diabetics had a lower risk of prostate cancer. However, they did note that this effect was limited to whites and Hispanics; diabetes did not affect prostate cancer risk among blacks. This may be due to the fact that blacks have higher levels of testosterone, and even though diabetes lowers the testosterone level, this is not sufficient in blacks to reduce the risk of cancer. It is also important to note that the use of specific medications for the treatment of diabetes, including insulin, did not affect these associations.380

Energy intake

Differences in energy intake between individuals are determined largely by the level of physical activity, body size, and metabolic efficiency. A positive association between energy intake and risk of prostate cancer was found in a Swedish case–control study.229 The association was stronger for patients who had advanced prostate cancer compared with patients who had localized cancer (fourth quartile vs. first quartile: RR, 1.70; 95% CI, 1.10–2.61). However, a Canadian case–control study noted a significant association with energy intake for preclinical prostate cancer (highest quartile vs. lowest quartile: OR, 2.67). A significant positive association was reported in a case–control study from Canada.227 Animal studies indicated that energy restriction reduced tumor growth in the prostate by inhibiting tumor angiogenesis.381 In a case–control study from Utah, a positive association was restricted to men between ages 68 years and 74 years, particularly those with aggressive cancer. Energy intake was a risk factor, whereas body mass and physical activity were not.225 Conversely, four case–control studies155, 223, 228, 242 and one cohort study237 failed to find an association between energy intake and risk. One of those studies237 assessed energy intake using a 24-hour food-recall assessment, a method with limited usefulness in estimating individual dietary intake.

Fruits and vegetables

With regard to fruits and vegetables, the consumption of cruciferous vegetables is associated with a decreased risk of many cancers,382 but there is no evidence of a protective effect for prostate cancer.234, 241, 383, 384 However, recent case–control studies have found that high consumption of vegetables, particularly cruciferous vegetables, was associated with a reduced risk of prostate cancer.385, 386 Fruit consumption was protective in the Health Professionals Follow-Up Study, in which the association was accounted for by fructose intake.387 High consumption of fruits and vegetables was associated with reduced prostate cancer risk in case–control studies in Uruguay,251 Hawaii,388 and Canada.385 One study found a moderate, inverse association between prostate cancer mortality and cabbage consumption.323

Grains and cereals

With regard to grains and cereals, the consumption of cereals is related inversely to prostate cancer mortality. Hebert et al. found a negative association between cereal energy and prostate cancer. The protection afforded by cereals most likely is related to the consumption of traditional breads, which usually contain whole flaxseeds, rye, and buckwheat flour. Foods such as soy, legumes, cereals, and vegetables are rich in ligands, which, when combined with isoflavonoids, may be protective against prostate cancer.323 A case–control study in China found a reduced risk of prostate cancer with increased consumption of soy foods and isoflavones.389

Theobromine

Theobromine, which is a constituent of tea and cocoa powder, is genotoxic.390 A single case–control study noted an increased risk of prostate cancer in older men with high theobromine intake.334 In contrast, a case–control study in Canada noted a decreased risk of prostate cancer in men who drank > 500 g of tea per day.338 A cohort in Canada found no association between tea intake and prostate cancer.341

Micronutrients

With regard to micronutrients, tomato-based products and lycopene consumption have been associated with reduced prostate cancer risk in several studies.179, 231, 238, 241, 243, 272, 324, 385, 391, 392 A prospective study of approximately 50,000 health professionals revealed a protective effect for frequent consumption (> 10 servings per week vs. < 1.5 servings per week) of tomatoes, tomato sauce, tomato juice, and/or pizza (RR, 0.65; 95% CI, 0.44–0.95) and an inverse association between lycopene intake and risk (RR, 0.79; 95% CI, 0.64–0.99).179 An inverse association (OR, 0.50), particularly among men age < 70 years (OR, 0.35) also was noted in a nested case–control study that examined prediagnostic plasma lycopene levels.241 Intake of tomatoes was related significantly to lower risk in a cohort study of Seventh-Day Adventists,238 but not in a case–control study. A nested case–control study failed to observe any effect for lycopene.269 A recent case–control study of diet and prostate cancer risk that was conducted in Athens, Greece, found that a higher consumption of tomatoes, and specifically cooked tomatoes, reduced the risk of disease. By increasing cooked tomato intake from 2 times per week to 4 times per week, a 15% reduction in risk was predicted (OR, 0.85; 95% CI, 0.75–0.97).231

Lycopene is the most efficient scavenger of the ROS singlet oxygen among the common carotenoids393 and is the predominant carotenoid in the prostate.394 A study using the Fisher F344 rodent model tested the chemopreventive effects of lycopene on 3,2′dimethyl-4-aminobiphenyl (DMAB)-induced rat ventral prostate cancer. The authors concluded that lycopene did not prevent rat prostate carcinogenesis consistently in this model.395

3.5.2.2 Environmental agents

Endocrine-disrupting chemicals

In the category of endocrine-disrupting chemicals (EDCs), several studies have suggested that exposure to estrogen mimics in the environment may be responsible for abnormalities of sexual differentiation, sexual maturation, and adult reproductive processes in both genders.396 Furthermore, environmental or nutritional estrogen agonists may influence carcinogenesis.397 Exposure to a high concentration of estrogen agonists is unusual, but a chronic low-dose exposure to estrogen agonists has been reported for some men, and this exposure may contribute to disease, including prostate cancer.150 There are numerous man-made contaminants that have been shown to possess estrogenic activity, including certain industrial and agricultural chemicals, plastics, detergents,396, 398 and dyes used in the food industry, such as Red Dye No. 3.399 Some of these chemicals, such as dithiothreitol (DDT), are no longer registered for use in the U.S. However, prior usage of those chemicals that are highly lipophilic, persistent, and bioaccumulative and have accumulated in the environment, food chain, and body fat.

Men can be exposed to natural plant substances with estrogenic properties, known as phytoestrogens.400 Plant ligands and isoflavenoids, which are found in soybean products, whole-grain cereals, seeds, nuts, and berries,162 are converted by intestinal bacteria into compounds with weak estrogenic and antioxidative activity.401 The potential anticarcinogenic effect of phytoestrogens has been the subject of considerable interest in breast cancer401 and prostate cancer.402 Tofu consumption was associated with a significant decreased prostate cancer risk in a prospective study of men of Japanese ancestry in Hawaii.237 High urinary excretion of isoflavenoids in Japanese men suggested that the low incidence of prostate cancer mortality in Japan and other Asian countries may be due to an inhibitory effect of dietary phytoestrogens.401, 403 Higher concentrations of the phytoestrogens daidzen and equol were found in plasma and prostatic fluid samples in Asian men compared with the concentrations found in European men.404 Clinical trials have examined the role of dietary supplementation of phytoestrogens in reducing the risk of prostate cancer.404, 405 One case–control study found a possible relation between intake of different phytoestrogens and prostate cancer risk. An inverse association was found between coumestrol and daidzen, whereas genistein showed a protective effect, and positive associations were found for campersterol and stigmasterol.406

The effects of exposure to a number of individual EDCs has been investigated in in vitro and in vivo models.398, 407 Wildlife and humans likely are exposed to a mixture of EDCs, some of which may act cumulatively. EDC exposure can affect hormone levels significantly through a number of different modes of action and, consequently, may affect the process of prostate carcinogenesis.

Prostate and breast tissue share features of hormone-dependent growth and differentiation, and estrogens may be a common etiologic link.408 Most risk factors for breast cancer are tied to a lifetime exposure to estradiol and other hormones,409 and it is tempting to speculate whether the same may apply for the prostate. The role of estrogens in prostate development and disease was postulated > 60 years ago.410 During critical periods in organ differentiation, such as fetal and neonatal life, hormones are involved in “imprinting” genes (permanently turning off genes or turning them on and setting their rate of activity), and this is a time of greater vulnerability to an alteration in hormone level or the function of the endocrine system.

The role of estrogen in prostate development is supported by animal studies. Male mice positioned in utero between 2 female fetuses (so-called 2F males) are exposed to higher concentrations of estrogen due to transport of estradiol from adjacent female fetuses, unlike males positioned in utero between 2 male fetuses (so-called 2M males).411 A higher estrogen exposure during male fetal life (2F males) was associated with significant enlargement of the prostate and a 3-fold “imprinted” increase in prostatic AR levels in adulthood in 2F male mice relative to 2M male mice.412 These findings suggest that exposure to a small increase in estrogen concentration during fetal life may induce permanent enlargement of the prostate and an increase in AR levels. It is interesting to note that this finding contrasts dramatically with findings from numerous prior studies that involved the administration of high doses of diethylstilbestrol (DES), which produced a permanent decrease in prostate size.410

Exposure to estrogenic chemicals may occur during critical stages of fetal life, but the consequences of such exposure typically are not recognized until adulthood, when problems relating to the reproductive system become apparent. Thus, men who are exposed during fetal life to low-dose EDCs or altered physiological levels of endogenous estrogens may be at increased risk of prostate cancer. Exposure to environmental estrogen mimics may affect the estrogen:androgen ratio, which, in turn, may have affects on the prostate. Some investigators believe that prostate cancer initiation is associated with a set of endogenous factors, whereas promotion is influenced more strongly by exogenous factors.413

Exogenous estrogens bind to ERs in target cells of the body and initiate or inhibit estrogen-like actions.398 Thus, estrogen mimics have the potential to alter, either beneficially or harmfully, the growth, development, and function of target tissues. The recent discovery of a second type of ER, ERβ, which is associated specifically with the prostate,414 may help explain the often paradoxical effects of estrogen-like drugs and compounds415 on the prostate. It should be noted that not all EDCs affect estrogen activity. In addition, some EDCs affect androgen (particularly relevant to prostate cancer), retinoic acid, or thyroid hormone activities.

Cadmium

Cadmium (Cd) is a significant environmental contaminant that results from zinc mining and smelting, sewage-sludge disposal, various industrial uses, and combustion of municipal waste and fossil fuels.416, 417 Annual worldwide production of cadmium is > 18,000 tons, and approximately 4,000 tons are used in the U.S., with approximately50% used for plating metals and the other 50% used for miscellaneous purposes, including pigments, batteries, stabilizers in plastics, metallurgy, nuclear reactor rods, the semiconductor industry, and as catalysts. Cadmium contamination in food, soil, air, and water may be high in industrial areas.418, 419 In addition to occupational exposure, Americans most likely are exposed to low doses of cadmium through consumption of contaminated fish, drinking water, contaminated air, and cigarette smoking.418–420 Cadmium is included frequently in the National Priorities List sites, and the National Toxicology Program classified it as a substance that may reasonably be anticipated as a human carcinogen.419 The International Agency for Research on Cancer considers it a Category I (highest level) human carcinogen.

Cadmium exposure has been linked to prostate cancer in some, but not all, epidemiologic studies,418, 420 and it appears to be associated with a modest increase in risk.128, 421, 422 Among 522 Swedish workers who were exposed to cadmium for at least 1 year in a nickel-cadmium battery plant, the mortality rate for prostate cancer increased in a dose-dependent and latency-dependent manner.423 Similarly, in a Utah population-based case–control study, occupational exposure to cadmium was correlated with a small increase in risk.312 Most studies that revealed a positive correlation between cadmium exposure and prostate cancer indicated a weak association.312, 423 An exception was noted, however, for aggressive cancer (OR, 1.7).312 These epidemiologic findings are in accordance with data showing that tissue levels of cadmium are higher in cancer than in benign specimens and BPH,306, 424, 425 with the highest concentration in high-grade cancers.424 A high incidence of prostate cancer also was observed in certain areas of Spain where cadmium was present naturally in the water and food supply because of abnormally high concentrations in stream sediment.426 Cadmium is distributed evenly between the epithelial cells and stroma within the normal prostate.300, 424 In one study, cadmium was associated inversely with testosterone content in BPH.300

Conversely, other studies have reported no association between cadmium exposure and prostate cancer.427 A population-based case–control study in Utah found no association.225 A cohort mortality study of cadmium-exposed workers found no increased risk after 5 years of follow-up.428 Another cohort study of 3025 nickel-cadmium battery workers revealed no association between occupational cadmium exposure and prostate cancer.429 Cadmium tissue content was not higher in cancer samples than in benign tissue.300

In rats, the unique susceptibility of the ventral prostate to cadmium-induced carcinogenesis may result from a lack of substantial expression of MT, which is a Cd-Cu-Zn-binding protein that has putative antioxidant activity.430–433 When cadmium was injected directly into the ventral prostate of rats, there was a high incidence of PIN and cancer within 270 days.434 Cadmium also enhanced the potency of other chemical carcinogens, such as 3,2′-dimethyl-4-aminobiphenyl (DMAB), in the ventral prostate.435 In vitro, cadmium causes malignant transformation of rat ventral prostate epithelial cells436 and increased proliferation of human prostatic epithelial cells437 as well as transformation.438 It is noteworthy that the growth-stimulatory action of cadmium on human epithelial cells is blocked by the antioxidant trace element, selenium, suggesting that the mitogenic action of cadmium may be mediated by induction of a prooxidation state in cells.437 Thus, cadmium-induced carcinogenesis in the rat prostate appears to be lobe-specific, although no clear mechanism has emerged from the limited data.

The carcinogenic potential of cadmium may be modified by zinc.439 Occupational exposure usually includes both zinc and cadmium.312 Administration of zinc potentiated or inhibited the carcinogenic effect of cadmium in the rat ventral prostate in a dose-dependent and route-of-administration-dependent manner. In the human prostate, zinc and cadmium have antagonistic effects.306, 420, 424, 425, 440 Prostate cancer has a much lower tissue zinc:cadmium ratio than benign tissue, suggesting that 1) high prostatic zinc content may confer protection against prostate cancer, 2) cadmium may be a weak prostatic carcinogen that is potentiated in the presence of zinc deficiency, and 3) the inability of the prostate to maintain proper zinc homeostasis may favor transformation induced by cadmium or other metal ions.

Pesticide exposure

With regard to pesticide exposure, a case–control study of prostate cancer in North Carolina332 noted an association with farming (see also 3.5.2.3 Occupation, below). No significant relation was apparent according to pesticide or gasoline exposure. However, that study was small (only 40 cases and 64 controls) and, accordingly, lacked statistical power.

A case–control study from the Netherlands observed that farmers with prostate cancer applied pesticides during significantly more days per year than farmers without prostate cancer.441 A modest, but statistically significant increased risk of prostate cancer was observed in a Swedish cohort of agricultural pesticide applicators.442 A study in North Dakota linked pesticide exposure to the development of prostate adenocarcinoma in relatively young males (age ≤ 50 years).443 However, in an English study, no association was noted between prostate cancer risk and the use of pesticides.326 A California farm workers study444 did not find an association between prostate cancer risk and exposure to pesticides in general; however, those authors did find that risk was increased with exposure to certain chemicals, including simazinew, lindane, and heptachlor. In a retrospective cohort study, Morrison et al.445 found an association between the number of acres sprayed with herbicides and the risk of prostate cancer mortality after 17 years of follow-up. The National Academy of Sciences reviewed the health effects of exposure to herbicides in Vietnam veterans and concluded that there was limited suggestive evidence linking herbicide exposure, including Agent Orange, to prostate cancer. A study in Italy found an increased risk among farmers who were exposed to organochlorine insecticides and acaricides and, more specifically, to DDT (OR, 21; 95% CI, 1.2–3.8) and dicofol (OR, 28; 95% CI, 1.5–5.0).446

A recent prospective cohort study of 55,332 male pesticide applicators asked whether there was an association between exposure to 45 agricultural pesticides and prostate cancer incidence.447 The men were from Iowa and North Carolina and were enrolled during 1993–1997 in the Agricultural Health Study. Pesticide exposure information was determined by questionnaire, and cancer incidence was determined from population-based cancer registries through 1999. The cohort had a 1.14 incidence ratio of prostate cancer compared with the male population in Iowa and North Carolina. In that study, three significant associations were found between pesticide use and prostate cancer risk. A significant association between the application of methyl bromide, an alkylating agent that is considered a potential occupational carcinogen by the National Institute for Occupational Safety and Health, and prostate cancer risk (well differentiated and poorly differentiated tumors) was observed. A significant association was found between the use of chlorinated pesticides, including DDT and heptachlor, only for men age > 50 years. A significant association was found between prior use of several other pesticides, including butylate and coumaphos, among men who had a family history of prostate cancer. These findings suggest that exposure to certain pesticides and interesting age-environment and gene-environment interactions influence prostate cancer risk.447

3.5.2.3 Occupation

Many industries, occupations, and exposures have been studied in relation to prostate cancer risk, but the findings have been inconclusive. Of greatest concern is farming and, to a lesser extent, working in the rubber industry.

Farming

A meta-analysis of prostate cancer incidence and farming found a modest positive association (RR, 1.12; 95% CI, 1.01–1.24).448 The increased risk was restricted to retrospective studies (RR, 1.29; 95% CI, 1.10–1.51), and studies that reported standardized mortality ratios suffered from a healthy worker-effect bias.

The basis for the modest increase in risk in farmers is unclear. One possibility is that the lifestyle of farmers increases their risk. There is evidence that a high-fat diet increases the risk of prostate cancer, and farmers eat a high-fat diet relative to nonfarmers.445 Conversely, farmers are exposed to high levels of sunlight; and, to the extent that ultraviolet radiation is protective for prostate cancer,449 farmers should have a lower risk. To the extent that physical activity is protective,450 farmers, who have physically demanding jobs, would be expected to have a reduced risk.

To our knowledge, relatively few studies of farming and prostate cancer published to date have examined the particular farm exposures that may be responsible for increased risk. A number of studies have implicated exposure to farm animals.219, 451, 452 These associations may reflect exposure to zoonotic viruses or to farm chemicals.448 Another possibility is that farmers who raise animals are more likely to consume them, and the increased consumption of animal fat and red meat may explain the higher risk.

Rubber industry workers

Studies of rubber industry workers have found both positive and negative associations with prostate cancer.355, 453–455 The International Agency for Research on Cancer concluded that, although there was limited evidence of excess risk in rubber workers, the data were inadequate to establish a causal association.

Other occupational exposures

Ambient particulate air pollution456 and employment in the metal plating and photographic film development industries are associated with an increased risk of prostate cancer.457 Exposure to organic dust also was linked to risk.458 Exposure to machine oil in the automotive industry was associated with a significantly increased risk of fatal prostate cancer.459, 460 The risk of prostate cancer also was increased by occupational exposure to certain radionuclides.461

A recent case–control study in Germany462 found an association between diesel fuel exposure and prostate cancer risk (OR, 3.7; 95% CI, 1.4–9.8 for men who were exposed to > 25 dose-years vs. men who were never exposed). Polycyclic aromatic hydrocarbons are recognized carcinogenics and are prevalent in diesel exhaust. A nested case–control study of electric utility workers in North Carolina463 found an association between prostate cancer mortality and large amounts of electromagnetic field exposure (OR, 2.02; 95% CI, 1.34–3.04). The risk of mortality also was increased with exposure to polychlorinated biphenyls (OR, 1.47; 95% CI, 0.97–2.24).

Activities and exposures during leisure time

A population-based case–control study in Montreal, Canada, identified various leisure-time activities that may be related to prostate cancer risk. Home or furniture maintenance (OR, 1.4; 95% CI, 1.0–1.9) and painting, stripping, or varnishing furniture (OR, 2.1; 95% CI, 0.7–6.7) were associated with an increased risk of prostate cancer. Exposures to certain substances, such as metal dust (OR, 3.2; 95% CI, 1.0–9.9), lubricating oils or greases (OR, 2.2; 95% CI, 1.2–3.7), and pesticides or garden sprays (OR, 2.3; 95% CI, 1.3–4.2), also were associated with increased risk.464

Exposure to ultraviolet radiation potentially is protective against prostate cancer according to Bodiwala et al.465 Their recent study in the U.K. confirmed that high levels of cumulative ultraviolet exposure, adult sunbathing, childhood sun burning, and regular holidays in hot climates all were associated independently and significantly with reduced risk.

3.5.2.4 Sexual activity and marital status

The role of sexual activity in the development of prostate cancer is unclear despite extensive study.162 Sexual activity and marital status are presumed surrogate measures of hormonal factors and infectious agents, both of which may influence risk. However, they are poor surrogates, because many cultural factors influence sexual activity and marital status, thus confounding potential associations.

Marital status does not appear to be associated with an increased risk of prostate cancer mortality.221, 241, 466, 467 However, two studies found an increased risk with early age at marriage compared with marriage later in life,117, 270 and two Japanese studies noted higher risk among married men compared with men who were never married.240, 270

Studies have found an increased risk of prostate cancer in men with a higher number of sexual partners467 and early first intercourse.468 A case–control study in California noted that, among black men, prostate cancer risk was associated positively with frequency of sexual intercourse; however, the relation was significant only for intercourse later in life.235 Another study also found an association between prostate cancer risk and increased frequency of sexual activity (for an increase of 3 times per week: RR, 1.2; 95% CI, 1.1–1.3).467 In contrast, the results from a case–control study in Greece340 and from another case–control study in Australia469 suggested that a high levels of sexual activity in early adulthood reduced the risk of prostate cancer. Roman Catholic priests470 and Mormon High Priests471 do not have a significantly lower risk of prostate cancer, arguing against an important role for sexual transmission.

In a single study, the relative risk for men reporting any sexually transmitted disease was 1.86 (95% CI, 1.43–2.42), but syphilis (RR, 0.77; 95% CI, 0.53–1.11) and gonorrhea (RR, 1.22; 95% CI, 0.92–1.62) were not associated significantly with prostate cancer risk.468 In another study, a weak association was found between gonorrhea and the risk of prostate cancer. In a later case–control study, it was found that, compared with asymptomatic men, symptomatic men who reported a history of venereal disease were at increased risk for prostate cancer (age-adjusted OR, 2.11; 95% CI, 1.18–3.80).105 Bacterial prostatitis, particularly in association with gonorrhea, may be a risk factor for prostate cancer.162 However, evidence to support a role for prostatitis in the etiology of prostate cancer is sparse.150

3.5.2.5 Viruses and other infectious agents

The prevalence of human papillomavirus (HPV) has been studied in prostate cancer, BPH, and normal prostatic tissue.472 Although a greater proportion of cancers were positive (32%) than normal tissue (9%), the greatest proportion was observed for BPH (49%). The implications of these findings are unclear. The prostate may serve as a reservoir for HPV, or the growth of HPV may be stimulated in prostate tissue with higher rates of cell proliferation. This may be critical if endemic HPV levels are often below detection levels. It is noteworthy that a Finnish seroepidemiologic study noted a 2.6-fold and 2.4-fold increased risk of prostate cancer for men who had antibodies to HPV type 18 (HPV-18) and HPV-16, respectively, but no increased risk was found with seropositivity to HPV-11, HPV-33, or Chlamydia.473

3.5.2.6 Vasectomy

Results from cohort studies that examined the relation between vasectomy and prostate cancer have been mixed. There were significantly elevated relative risks of 1.7 and 1.6 in prospective474 and retrospective475 cohorts, respectively. Conversely, in another study, no association was found between prostate cancer and vasectomy, a finding that was confirmed in a second report based on additional years of follow-up.476, 477 Two other cohort studies were inconclusive because of insufficient follow-up time and limited power.478, 479

Results from case–control studies regarding risk with vasectomy also have been inconsistent. One group found an elevated risk regardless of whether cases were compared to cancer controls (OR, 3.5; 95% CI, 2.1–6.0) or noncancer controls (OR, 5.3; 95% CI, 2.7–10).480 In a study from China,481 a positive association was reported regardless of whether hospital cancers, hospital noncancers, or neighborhood controls were used. Other studies reported an increased risk of 40%,357 60%,95 and 70%482 with vasectomy. A recent study conducted in Canada found that, compared with asymptomatic men, symptomatic men were at a significantly increased risk of prostate cancer (age-adjusted OR, 1.49; 95% CI, 1.14–1.95).105 Conversely, in a large multiethnic case–control study conducted in the U.S. and Canada,483 a history of vasectomy was not associated with prostate cancer risk (OR, 1.1), nor did risk vary by age at vasectomy or years since vasectomy. A case–control study of blacks and whites in the U.S.484 found no overall effect for vasectomy (OR, 1.1; 95% CI, 0.8–1.77) but noted an increased risk among white men who had undergone vasectomy ≥ 20 years prior to the study (OR, 1.7; 95% CI, 0.9–3.3) or who had undergone vasectomy at age < 35 years (OR, 2.2; 95% CI, 1.0–4.4). One study found no association between prostate cancer and vasectomy (OR, 1.2; 95% CI, 0.6–2.7), except among men who underwent vasectomy before age 40 years (OR, 3.4; 95% CI, 0.8–14).485 In another study, a modest reduction in risk was found (OR, 0.86; 95% CI, 0.57–1.32),486 similar to a study that included only 5 men with prostate cancer who had a history of vasectomy.457

The biologic mechanism that may link vasectomy and prostate cancer is uncertain.487 Another major concern in studies of vasectomy and prostate cancer is detection bias.483 Vasectomized men are more likely to visit a urologist subsequently, which may result in an increased chance of detecting prostate cancer.487 In addition, most studies have used self-reported history of vasectomy; unfortunately, no study to date has validated these reports against medical records.483 Many studies also have used self-reported disease status, which may be inaccurate.488

3.5.2.7 Social factors, including lifestyle, socioeconomic factors, and education

Social factors, such as income and education, do not influence the risk of developing prostate cancer directly but are surrogate measures for other, unmeasured exposures, such as diet. However, social factors may influence prostate cancer mortality directly through access or use of medical care.

Differences in socioeconomic status do not appear to account for the marked differences in prostate cancer risk among racial groups.128, 237, 489 One exception was a study that noted an inverse association between risk and education.238 In one study, an association was found between insurance and employment status and risk for advanced prostate cancer at the time of diagnosis. However, those authors found that socioeconomic factors alone could not explain adequately the difference in risk among racial groups.153 However, socioeconomic factors may influence the likelihood of prostate cancer diagnosis. Access to health care may influence the use of PSA screening or treatment patterns for BPH, which, in turn, may influence cancer detection rates.

3.5.2.8 Physical activity

Physical training may lower both body fat and testosterone levels, thus reducing the risk of prostate cancer for active men.490, 491 There have been reports that physically active men experience decreased,205, 450, 490–494 increased,339, 369, 495, 496 and similar155, 237, 242, 497 risk of prostate cancer compared with inactive men.

Studies from China,491 Finland,205 Turkey,498 Canada,499 and Sweden500 have indicated that individuals who work in sedentary jobs are at a small but not significantly increased risk. Results from the Chinese and Turkish studies were independent of whether physical activity was measured by total energy expenditure or percent of occupational time spent sitting. However, neither of those studies controlled for the potentially confounding effect of diet.

There may be an inverse association between prostate cancer and occupational physical activity.501 Conversely, a study of lifetime occupational physical activity levels among Hawaiian men found that physical activity was associated positively with risk.187 A study in Taiwan also found a significant association between physical activity and prostate cancer risk.339 However, a cohort study in Iowa,346 a case–control study in China,502 and a cohort study in the Netherlands503 detected no association. One article reported an inverse association between physical activity and prostate cancer risk. However, those authors noted that the epidemiologic evidence to date was inconsistent and that the observed risk reduction was minimal.504

3.5.2.9 Anthropometry

Elevated body mass index (BMI) was associated with increased prostate cancer risk in a case–control study in Italy. The OR for men in the heaviest group (BMI ≥ 28.0) was nearly 4.5 times greater than in OR for the reference group (BMI < 23.0).221 A Norwegian study242 reported a > 2-fold increased risk for obese men (BMI > 27.6). Men who were at least 30% over their desired weight had a 2.4-fold increased risk of fatal prostate cancer compared with men who were near their desirable weight in a cohort study of Seventh-Day Adventists. Similarly, a study in France found that obese men had 2.5 times the risk of developing prostate cancer.505 Studies of Japanese (RR, 1.33), Dutch (OR, 1.5), and Seventh-Day Adventists (RR, 1.17) reported increased, although nonsignificant, risks.237, 238, 506 Increased BMI was associated with risk of high-grade cancer in a more recent study. This risk was especially prevalent in men age < 50 years.119 Conversely, a cohort study of > 20,000 men in Hawaii239 found that a high BMI was slightly protective (RR, 0.7; 95% CI, 0.5–1.2). A case–control study in Taiwan found that a low BMI was associated significantly with a higher risk of prostate cancer.339 Other studies found no difference in BMI between cases and controls.222, 234, 235, 346, 352, 383, 507

The positive association between BMI and prostate cancer may reflect the contribution of muscle mass, rather than fat, to risk.508, 509 In one study, it was found that muscle mass, but not fat, of the upper arm was related significantly to prostate cancer risk.509 Increased muscle mass may be a marker for elevated lifetime androgen exposure.127

Obesity may be linked causally to prostate cancer risk due to its affect on sex hormone metabolism or through its association with high-energy intake, high-fat diet, or reduced physical activity. A case–control study in China found that abdominal adiposity (high waist-to-hip ratio) was related to an increased risk of prostate cancer. However, those authors also found that men in the highest quartile of hip circumference (> 97.4 cm) had a reduced risk (OR, 0.46; 95% CI, 0.29–0.74).510 A recent review concluded that there was no association between obesity and prostate cancer risk.511

Height, and in particular leg length, represents a surrogate for dietary and hormonal influences prior to adulthood that may affect the risk of prostate cancer significantly. Results from a population-based case–control study showed that height was related to an increased risk of prostate cancer. Although adult height was not related to the risk of localized prostate cancer, there was a moderate positive association between increasing height and risk of advanced disease (upper quartile vs. lower quartile: RR, 1.62; 95% CI, 0.97–2.73). Height also was associated more strongly with prostate cancer risk in men who had a positive family history of the disease. These findings provide supportive evidence that there are growth-related risk factors for prostate cancer, particularly for advanced and familial forms of the disease.512 A case–control study in Greece attributed this association to IGF-I levels.513 In contrast, in a recent case–control study, it was found that greater height was associated with a greater chance of survival in patients who had prostate cancer.514

Male pattern baldness and prostate cancer risk share some common risk factors, such as aging, androgens, and heredity. One cohort study attempted to find a correlation between male pattern baldness and risk of prostate cancer. The RR for men with baldness was 1.50 (95% CI, 1.12–2.00), and the effect was independent of the severity of the baldness or other risk factors. The authors concluded that male pattern baldness is a risk factor for prostate cancer.515 A recent case–control study evaluated the potential association between weight and length at birth and the risk of prostate cancer, but no significant association was found.516

3.6 Data Gaps

Prostate cancer control is problematic, in part because international prostate cancer incidence and mortality rates vary dramatically. Migrant studies suggest that nongenetic, lifestyle factors play a key role in the etiology of prostate cancer. Unfortunately, with the possible exceptions of animal fat consumption and pesticide exposure/farming, no known, widespread, modifiable risk factors have been identified by the numerous epidemiologic studies of prostate cancer.

One possible reason is that descriptive epidemiology is a relatively crude tool for examining what may prove to be an unusually complex etiology. An understanding of the interplay between genetically determined factors (e.g., testosterone, SHBG, and 5-α-reductase) and environmental factors (e.g., dietary fat, vitamin E, and cigarette smoking) will be necessary before a measure of consistency is achieved across epidemiologic studies. The integration of biomarkers of susceptibility and exposure in epidemiologic studies is needed. Future studies need to consider the potential importance of critical windows of exposure. The inclusion of information on clinical stage should allow for the control of biases that may occur from combining cancers of different clinical significance. Incidence studies also should take into account the basis for prostate cancer diagnosis (screening vs. urologic symptoms) to identify confounding risk factors associated with prostate cancer screening.

Some questions for future studies are: 1) What is the relative importance of particular environmental exposures to prostate cancer risk? Issues of dose, duration and timing of exposure are highly relevant, poorly understood, and badly recapitulated in animal model systems. 2) What are the important interactions between environmental factors (for example, between cadmium and dietary fat exposure) that contribute to prostate cancer risk? 3) Does the “internal environment,” such as oxidative stress and testosterone action, modify prostate cancer risk? In addition, what genetic and external environmental factors significantly modulate oxidative stress or the hormonal milieu within the prostate? The importance of taking a life-stages approach in analyzing the importance of these factors is highlighted by the observation that testosterone likely plays an important role in prostate cancer development, yet prostate cancer incidence is peaking at an age when serum testosterone levels are declining. 4) What are the critical gene-environment interactions that confer susceptibility or resistance to prostate cancer-inducing environmental influences? Alavanja et al.447 identified an interaction between family history (genetics) and pesticide use. It will be critical to identify the particular genetic polymorphisms within the human population that will allow stratification in terms of prostate cancer susceptibility. 5) Is there a signature mutational spectrum (or spectra) in most human prostate cancers?

4.0 ANIMAL AND CELL CULTURE MODELS FOR THE PREDICTION OF HUMAN RISK

4.1 Existing Animal Models

Numerous animal bioassays and resultant cell lines exist for spontaneous, induced, and transplantable prostate cancer (Table 2). Each bioassay has strengths and weaknesses as models for human prostatic carcinogenesis.

Table 2. Human, Rat, and Mouse Cell Lines
Cell line(s)OriginProperties and experimentsReference(s)
  1. SCID: severe combined immunodeficiency; IUdR: 5-iodi-2′-deoxyuridine; uPA: urokinase plasminogen activator; HPV-18: human papilloma virus type 18; PSA: prostate-specific antigen; AR: androgen receptor; EGF: epidermal growth factor; bFGF: basic fibroblast growth factor; TGF: transforming growth factor; BPH: benign prostatic hyperplasia; T: testosterone; testosterone DHT: dihydrotestosterone; ER: estrogen receptor; PAP: prostatic acid phosphatase; TUR: transurethral resection; SV40: simian virus 40; Gy: grays; VP: ventral prostate; NMU: N-methyl-N-nitrosourea; PIN: prostatic intraepithelial neoplasia; PSCA: prostate stem cell antigen; DLP: dorsolateral prostate; TRAMP: transgenic adenocarcinoma of the mouse prostate.

Human species   
 BM1604Prostatic adenocarcinomaEstablished by culture of tissue pieces on extracellular matrix laid down by fibroblasts, these cells have a doubling time of 28 hrs and show somatic mutations.van Helden et al., 19941510
 DuPro-1Prostate adenocarcinomaDerived from a xenograft established from DU5683 cells in an athymic mouse, these cells have a doubling time of 22–24 hrs and exhibit pseudoglandular configuration of matrigel. Cells are near-tetraploid and have 3–4 X chromosomes but no Y chromosome. Cells implanted with matrigels into kidney capsules form tumors that grow equally well in male and female nude mice, suggesting that they are androgen-insensitive.Gingrich et al., 19911511
 PC-3 ML, PC-3 MR, and PC-3 MCSublines derived from invasive PC-3 cellsEstablished by injecting an invasive subline of PC-3 cells into SCID mice, the invasive subline was selected based on its ability to migrate across a reconstituted basement membrane (matrigel) in Boyden chamber chemotactic assays. The cells preferentially metastasize to the lumbar vertebrae (PC-ML), the rib cartilage (PC-3 MR), and the right cheek (PC-3 MC), as demonstrated by IUdR labeling and histology; taxol (50–250 mg/m2/day) blocked the establishment, growth, and long-term survival of PC-3 ML cells in SCID mice.Wang and Stearns, 1991695; Stearns and Wang, 1992696
 1013LPrimary prostate cancer cellsForm tumors in SCID mice with the aid of a gelatin sponge (Spontostan). The latency period for growth in SCID mice is 42–64 days followed by a slow growth phase, and then a faster growth phase. The fast growth phase is characterized by rapid degeneration of tumor tissue. The tumors do not show invasiveness and could be used as a model for studying progression of prostate cancer to a more aggressive phenotype. These cells do not produce type PA (uPA). The 1013L cells have a metacentric chromosome that shows an extra C band. G-banding shows multiple marker chromosomes.Hartley et al., 19891512; Billstrom et al., 1995707
 UCRU-PR-2Undifferentiated prostate cancer lineThis small cell, undifferentiated cancer cell line grows as a small cell carcinoma when implanted into nude mice. Fragments implanted intramuscularly and under the kidney capsule are invasive locally. Xenograft tumors show both glandular and neuroendocrine differentiation features.Jelbart et al., 19891513
 RWPE-1 and RWPE-2Immortalized primary prostate cancer cellsRWPE-1 is an immortalized, nonneoplastic human prostatic epithelial cell line that was established from primary epithelial cells obtained from the prostate of a white male donor and immortalized with HPV-18. RWPE-1 cells were transformed further by v-Ki-ras to establish the RWPE-2 cell line. Both RWPE-1 and RWPE-2 express luminal and basal cell cytokeratins. They show androgen-dependent growth, secrete PSA, and express AR, and they exhibit dose-dependent growth stimulation by EGF and bFGF and exhibit growth inhibition by TGF-β. RPWE-1 cells are not noninvasive, do not grow in soft agar, and do not form tumors in nude mice. In contrast, RPWE-2 cells are invasive, exhibit anchorage-independent growth in soft agar, and form tumors in nude mice.Bello et al., 19971514
 BRF-41TProstatic adenocarcinomaThis cell line was established from cellular outgrowths of prostatic adenocarcinoma explants. It expresses epithelial cell markers, cytokeratins 8 and 18, PSA, AR, H-ras, K-ras, and p53 gene products.Iype et al., 19981515
 BRF-55TImmortalized cells from BPHThis cell line originated from cellular outgrowths of explanted BPH tissues using pSRV-T for immortalization. It exhibits epithelial cell features, secretes PSA, and expresses AR. It grows in serum-free medium and expresses H-ras, K-ras, and p53 gene products.Iype et al., 19981515
 ALVA101Primary cells from a prostatic adenocarcinomaThis cell line exhibits increased cell proliferation rates upon androgen (T or DHT) or EGF challenge. It expresses AR, TGFα, and EGF receptor mRNA. Experimental data suggest that the mitogenic effect of androgen on this cell line may involve TGF-α/EGF receptor signaling.Liu et al., 19931516
 ALVA-41Metastatic cancer to the boneThis cell line exhibits a fast growth rate. It expresses AR and glucocorticoid receptor but not ER. Growth can be enhanced by DHT. Cells do not secrete PSA but secrete PAP in an androgen-dependent manner.Nakhla et al., 19941517
 ALVA-31Biopsy specimens of a primary tumorThese cells have a doubling time of about 26 hrs. Chromosomal numbers vary from 24 to 112, with a modal number of 59. Late-passaged cells have approximately 70 chromosomes, 8–14 marker chromosomes, 2 X chromosomes, but no Y chromosome. The cell line expresses AR, PSA, and PAP. Xenograft tumors grow in intact male, castrate male, and female athymic mice; however, the rate of tumor growth is dependent on serum T levels. This cell line offers a good model to study androgen regulation of prostate tumor cell growth.Loop et al., 1993727
 ND-1Primary prostatic adenocarcinomaThis cell line grows as monolayers in soft agar and produces subcutaneous tumors in nude mice. G-banding showed aneuploid karyotype with a modal chromosome number of 62, multiple marker chromosomes with 25–30% structural abnormalities, and an abnormal Y chromosome. Cells show common features of neoplastic epithelial cells, including microvilli, junctional complexes, abnormal nuclei, nucleoli, and mitochondria. They secrete PSA in very small amounts. The cells have a higher metastatic potential in nude mice.Narayan and Dahiya, 19921518; Dahiya et al., 19921519
 JCA-1Poorly to moderately differentiated prostatic adenocarcinoma.These cells form colonies in soft agar and grow in suspension cultures. Tumors formed are transplantable in nude mice. Tumors are poorly differentiated adenocarcinomas that express PSA and PAP. Karyotype analysis demonstrated aneuploidy with a modal chromosome number of 69, and 6 marker chromosomes have been identified.Muraki et al., 19901520
 TSU-Pr1Lymph node metastatic adenocarcinomaCells in culture show loss of contact inhibition and rapid proliferation rate. Early cell passages show AR protein expression, but the levels decline with subsequent passages; the cells produce a small amount of PAP. Karyotype analyses show aneuploidy with a modal chromosome number of 80, including a Y chromosome and at least 10 marker chromosomes. This cell line forms tumors in athymic nude mice; heterotransplanted tumors do not express AR. The cells express functional type I and II receptors for TGF-β and is growth-stimulatory by this cytokine.Iizumi et al., 19871521; Lamm et al., 19981522
 PPC-1A poorly differentiated adenocarcinoma of the prostateThis cell line exhibits many transformed phenotypes, including relaxed growth factor requirements and anchorage-independent growth. It is highly tumorigenic in nude mice. Cytogenetic analysis demonstrated a macroscopic abnormal karyotype with a modal chromosomal number of 84, multiple marker chromosomes, double minutes, and clonal loss of chromosomes 3, 5, 10, 15, and Y.Brothman et al., 19891523
 PZ-HPV-7 and CA-HPV-10Human prostatectomy specimensThe PZ-HPV-7 cell line was established from normal prostatic tissue, whereas the CA-HPV-10 line was established from cancerous tissues using HPV-18 transformation; both cell lines show expression of cytokeratins 5 and 8 similar to those expressed in the cells of origin. PZ-HPV-7 cells have a modal chromosome number of 46 in early passages but become tetraploid in later passages. CA-HPV-10 cells are aneuploid, and some retain double-minute chromosomes that are present in the cancer cells from which they are derived. Both cell lines are nontumorigenic in nude mice.Weijerman et al., 19941524
 BPH-1TUR tissuesIt is established from TUR specimens using SV40 large-T antigen to immortalize epithelial cells. These cells exhibit a cobblestone appearance in monolayer cultures, and they are nontumorigenic in nude mice. The cell line overexpresses p53 gene products; cytogenetic analysis demonstrates an aneuploid karyotype with a modal chromosome number of 76 and with 6–8 marker chromosomes. The Y chromosome is present; cytokeratin profile is consistent with the profile expressed in luminal epithelial cells. In serum-free cultures, EGF, TGF-α, FGF 1, and FGF-7 (KGF) increase cell proliferation, whereas FGF-2, TGF-β1, and TGF-β2 inhibit proliferative activity. T has no effect on proliferation of BPH-1. These cells do not express AR, PAP, or PSA (at both the protein level and the mRNA level).Hayward et al., 19951525
 PNT1Normal prostatic epithelial cellsNormal prostatic cells immortalized using a SV40 genome with a defective replication origin (SV40 ori−). These cells express low levels of cytokeratins 18 and 19, but not cytokeratin 14. They secrete PAP, PSA, and AR.Cussenot et al., 19911526
 267B1Normal prostatic epithelial cells from fetal prostateEpithelial cells were transformed by multiple exposures to ionizing radiation (X-ray dose, 30 Gy). Immortalized/transformed cells exhibit anchorage-dependent growth in soft agar and tumorigenicity in nude mice. Poorly differentiated adenocarcinomas form in nude mice. These cells express PSA and epithelial cell-specific cytokeratins; no p53 or ras mutations are found in this cell line. Cytogenetic analysis reveals numerous chromosomal defects, among which chromosome 3 and 8 translocations predominate.Kuettel et al., 19961527
 NP-2sNormal prostatic epithelial cellsNormal prostatic epithelial cells were transformed by transfection with a plasmid containing SV40 early genes. Cells express cytokeratins but are nontumorigenic in nude mice. In vitro, it requires bovine pituitary extracts for growth in serum-free medium; the growth of these cells can be stimulated by TGF-β1 and EGF in clonal growth assays.Kaighn et al., 19891528
 PWR-1ENormal prostatic epithelial cells from an adult prostateThe only known immortalized human prostatic cell line using adenovirus 12-SV40 hybrid for immortalization, it expresses cytokeratins 8 and 18 and, thus, represents a well differentiated, secretory, prostatic epithelial line. In response to androgen, it shows up-regulation of AR and PSA. It exhibits strong nuclear p53 staining. Cells of low and high passage numbers are nontumorigenic in nude mice, even when coinjected with matrigel. Cells grow in a serum-free medium and respond to EGF, bFGF, and TGF-β; passage 42 cells show a human male (XY) hyperdiploid karyotype.Webber et al., 19961529
 DuK50Prostatic stromal cells from a normal region of a radical prostatectomy specimenThis cell line exhibits normal fibroblastic characteristics and stains positive for vimentin, fibronectin, and α-actin but is negative for cytokeratins and PSA. Androgens stimulate growth. It is nontumorigenic in nude mice. Karyotypic analysis reveals normal male 46 XY chromosomal content with no numeric or structural abnormalities. It may serve as a wonderful tool for studying stromal-epithelial interactions.Roberson et al., 19951530
 DU-145Prostatic adenocarcinoma metastasized to the brainThis was one of the earliest prostate cancer cell lines established. Cells grow in isolated islands on plastic Petri dishes and form colonies in soft agar suspension culture. It has an aneuploid karyotype with a modal chromosome number of 64; distinctive markers include a translocated Y chromosome, metacentric minute chromosomes, and three large, acrocentric chromosomes are found. It does not express the AR at either the mRNA level or the protein level, and it forms tumors when injected into athymic mice.Stone et al., 19781531
 PC-3Prostatic adenocarcinoma metastasized to the boneCells are anchorage-independent and grow both as monolayers and in soft agar; these cells are tumorigenic in nude mice and have a reduced dependence on serum for growth compared with normal prostatic epithelial cells. Cells do not respond to androgens, glucocorticoids, EGF, or FGF. Karyotype analysis indicates aneuploid with modal chromosomal number in the hyrotriploid range. At least 10 distinctive chromosomal markers were identified originally on this cell lines. Poorly differentiated carcinomas are formed from this cell line.Kaighn et al., 19791532
 LNCaPProstatic adenocarcinoma metastasized to the left supraclavicular lymph nodeThese cells have a low anchorage potential and do not produce smooth, uniform monolayers. The cell line has a doubling time of approximately 60 hrs. Chromosomal analysis shows an aneuploid human male karyotype with a modal chromosome number of 76–91. Both cultures and tumors produce PAP and PSA. The cells express AR and ER. However, the data on ER expression are conflicting. The cells express EGF, TGF-α, and TGF-β, and they are highly responsive to T and DHT; the AR in this cell line has a single point mutation, resulting in an 868 Thr-Ala substitution in the ligand-binding domain that permits it to bind to nonandrogenic sex steroids (estrogens and progestins). It is capable of forming tumors when injected into male and female nude mice with matrigel or bone fibroblasts. Tumors developed in male mice earlier and at a greater frequency than those in female hosts. The rate of tumor growth is dependent on serum androgens but is independent of the gender of the host. Since its establishment, several sublines of LNCaP have been isolated.Horoszewicz et al., 19831533; Pousette et al., 19971534
 LNCaP-rContinuous growth of LNCaP cells in media containing low amounts of androgensThis is an LNCaP derivative that is totally insensitive to sex steroids. It exhibits a doubling time of 14–36 hrs. Karyotype analysis shows a modal chromosomal number in the tetraploid region.Pousette et al., 19971534
 C4-2An androgen-independent derivative of LNCaPThis cell line was established from LNCaP tumors maintained in castrated, athymic mice. It differs from the parental LNCaP cell line in its tumorigenicity and androgen dependence. It forms tumors in castrated and intact hosts. Tumors metastasize to the lymph nodes and bone with an incidence of 11–50%. The incidence of osseous metastases is higher in castrated hosts than in intact hosts, suggesting possible androgen suppression of dissemination. Karyotype analysis shows chromosome numbers between 72 and 90. This cell line shares common marker chromosomes with parental LNCaP cells.Thalmann et al., 19941535
 LNCaP 104S, 104I, 104R, and 104-R2LNCaP 104-SLNCaP 104-S is a clonal subline of LNCaP. Its growth rate is very low in androgen-depleted medium and but can be stimulated by low doses, but not high doses, of androgen. Continued passaging of 104-S gave rise to 104-I, which has a greater response to low-dose androgen. 104-R derives from 104-I, and it proliferates rapidly in the absence of androgen. The up-regulation of AR appears to be a mechanism for these cells to gain responsiveness to low-dose androgen. These cell lines form tumors in castrated, athymic mice. T treatment inhibits cell proliferation and causes tumor regression Estradiol and medroxy progesterone acetate also inhibit their growth in vivo.Umekita et al., 1996728
 ARCaPCancer cells in ascitic fluid from a patient with hormone-refractory and metastatic carcinomaCells express low levels of AR mRNA, PSA mRNA, and protein. Androgen and estrogen repress growth in a dose-dependent manner both in vivo and in vitro. Androgen also represses the expression of PSA. These cells metastasize to the lymph node, lung, liver, pancreas, kidney, and bone and form ascites in athymic hosts. Cells stain positive for EGF receptor, c-erb-2/neu, c-erb-3, bombesin, serotonin, and c-met protooncogene. They secrete gelatinase A and B and stromelysin; therefore, the cell line expresses markers of invasive adenocarcinoma and selective neuroendocrine markers. It has 11 marker chromosomes and has X and Y chromosomes in multiple copies. Chromosomes 20 and 22 are disomic, whereas chromosomes 5, 7, and 13 are present in 4 copies.Zhau et al., 1996729
 WPE1-NA22, WPE1-NB14, WPE1-NB11, and WPE1-NB26RWPE-1 human prostatic cells exposed to MNUThis family of cell lines represents a unique and relevant model that mimics stages of PIN and progression to invasive cancer. WPE1-NA22 cells are the least malignant and form small, well differentiated tumors. WPE1-NB26 cells are the most malignant and form large, poorly differentiated, invasive tumors; they can be used to study carcinogenesis progression, intervention, and chemoprevention.Webber et al., 20011536
 VCaPMetastatic tumor in the vertebrae of a patient with prostate cancerVCaP cells are immortal in vitro and can be passaged serially in vivo. This cell line exhibits many of the characteristics of clinical prostate carcinoma. It expresses PSA, PAP, cytokeratin-18, and AR. The cells are androgen sensitive in vitro and in vivo. This model will be useful for more advanced study of the mechanisms of prostate cancer progression and metastasis.Korenchuk et al., 20011537
 957E/hTERTPrimary tumor from a patient with a family history of prostate cancerThis study is the first documented case of a telomerase-immortalized human prostatic cancer cell line that was established from a patient with a family history of prostate cancer. 957E/hTERT cells show transformed morphology, are grown in serum-free medium, and express CK8, NKX3.1, PSCA, and p16. They do not express AR or PSA. The cells showed growth inhibition when exposed to retinoic acid and TGF-β1. This cell line was near diploid and showed random loss of chromosomes 8, 13, X, and Y. It will be useful for the identification and characterization of prostate cancer susceptibility genes, and it also will provide a means for testing new modalities for prevention and progression of prostate cancer.Yasunaga et al., 20011538
 RC-58T/hTERTPrimary prostate tumorThis is the first documented case of an established human prostate cancer cell line from a primary tumor of a prostate cancer patient with telomerase. The cells exhibit transformed morphology and show anchorage-independent growth as they form colonies in soft agar. They express the androgen-regulated, prostate-specific gene NKX3.1; epithelial-specific cytokeratin 8; PSCA; and p16, but not PSA or AR. They showed growth inhibition when exposed to retinoic acid and TGF-β1. A number of known chromosome alterations were observed, including the loss of chromosomes Y, 3p, 10p, 17p, and 18q and the gain of chromosomes 16 and 20.Yasunaga et al., 20011539
 LNCaPParental cellsThis model closely resembles the progression of human prostate cancer from the androgen-responsive to the hormone-refractory state. This model may provide an opportunity to understand the molecular mechanisms associated with the acquisition of androgen independence during prostate cancer progression.Igawa et al., 20021540
Rat species   
 AXCVP of aged XC ratsCells injected into isogenic males form tumors. This cell line shows responses to androgens, prolactin, and androgen ablation in manners similar to the responses observed for the VP. It secretes physiologic levels of PAP and expresses 5-α reductase activity. Several sublines of AXC cells have been established subsequently.Shain et al., 19841541
 Dunning R-3327 sublinesDunning prostatic tumorDunning prostatic adenocarcinoma cell lines were derived from DLP tumors in Copenhagen rats. Several sublines were established from Dunning R-3327 cells. These sublines include G, AT-1, and AT-2 cells, which exhibit low metastatic potential, and AT-3, Mat-LyLu, and MAT-Lu cells, which have high metastatic potential in isogenic hosts. These cell lines express different amounts of AR, ER, and 5-α reductase, and they exhibit different in vitro morphology and growth rates. AT-3 and Mat-LyLu cells metastasize to lung and lymph nodes. Metastatic activities are blocked by doxorubicin pretreatment.Wenger et al., 19841542; Isaacs et al., 19861543; Bashir et al., 1994595
 EPYP-1Copenhagen Rat prostatic epithelial cellsNormal DLP epithelial cells are immortalized by SV40. Cells grow in monolayers and exhibit epithelial morphology. They express PAP, cytokeratins, and integrins a6 and B1. They are stimulated by androgens. Cytogenetic analysis indicates that these cells are aneuploid. They are nontumorigenic in isogenic hosts at both subcutaneous and intraprostatic sites.Yamazaki and Pienta, 19951544
 PLS10, PLS20 and PLS303,2′-Dimethyl-4 aminobiphenyl and T-induced carcinomas of DLP of male F344 ratsAll 3 cell lines are cytokeratin positive and grow as monolayers. They are androgen-independent, tumorigenic, and metastatic in vivo. PLS10 and PLS30 cells form well differentiated adenocarcinomas, whereas PLS20 cells form poorly differentiated adenocarcinomas in nude mice. PLS tumors do not respond to castration. Tumor growth rates in nude mice are in the following descending order: PLS20 > PLS10 > PLS30. Cell lines responded in vitro to stimulation by insulin/transferrin but not by EGF, bFGF, or dexamethasone. Tumors from all e lines metastasize to the lung, but PLS20 also produces bone metastases. Karyotype analysis indicates that PLS10, PLS30, and PLS20 cells are diploid, hyperdiploid, and hypertetraploid, respectively. These cell lines may be useful in understanding the progression and metastasis of prostatic carcinomas.Nakanishi et al., 1996608
 NRP-152, and RP-154DLP epithelial cells from Wistar rats treated with NMU and T propionateThe NRP-152 cell line does not form tumors in nude mice and retains many of the properties of normal epithelial cells. It produces PAP and has functional AR. Its growth rate is stimulated by EGF, insulin, dexamethasone, DHT, T, and 1,25-dihydroxyvitamin D3 and is inhibited by TGF-β1 and retinoic acid. In contrast, NRP-154 cells are stimulated by dexamethasone and insulin and are inhibited by TGF-β1, but they are not responsive to EGF, DHT, retinoic acid, or vitamin D3.Danielpour et al., 1994537
 RPE-F344Androgen-deprived, involuted prostate after initiation of regeneration by TRPE-F344 cells form a uniform monolayer in culture. They exhibit contact inhibition at confluence, and they do not form colonies in soft agar. The cell line expresses antiapoptotic genes, it is p27kip1 negative, telomerase positive, and it expresses high-molecular-weight cytokeratins that are specific for prostate basal cells. It expresses low levels of AR and PAP, and it can be maintained in vitro without androgen supplementation, but the addition of 15 nM DHT to culture results in a significant but transient enhancement of cellular proliferation.Presnell et al., 19991545
Mouse species   
 C3H cell lineAdult mouse prostateThis cell line expresses PAP, and its growth is dependent on EGF. It is hypertetrapolid and is capable of metabolizing benxo(a)pyrene.Kubota et al., 19811546
 TRAMP-C1, C2, and C3TRAMP mouse tumorThese cell lines were established from tumors of the TRAMP model, a transgenic line of C57BL/6 mice harboring a construct of the rat probasin minimal promoter and SV40 large-T antigen. All 3 lines express cytokeratin, E-cadherin, AR, and wild-type p53. C1 and C2 cells are tumorigenic in isogenic hosts, but C3 is not.Foster et al., 1997665

4.1.1 Rodent models

There are approximately 10 rodent models for study of prostate cancer.517

4.1.1.1 Lobund-Wistar rat

The Lobund-Wistar model has some similarities to human prostate cancer.518 In both, cancer develops spontaneously with older age, metastasizes, and is under hormonal influence. Spontaneous prostate cancer is very unusual in rats other than the Lobund-Wistar strain.519–521 The advantages of the Lobund-Wistar model of prostate cancer are that 1) tumors can be induced in high (90%) incidence after initiation with a single dose of methyl-nitrosourea (MNU) followed by repeated implantations of testosterone propionate522–524; 2) the tumors are adenocarcinoma, the same histologic type found in humans; 3) a high frequency of cancers metastasize to bone, similar to human prostate cancer; 4) 78% of rats with cancer have metastases to the lungs or peritoneal cavity, including the liver, spleen, and kidney; 5) in MNU-initiated, testosterone-promoted rats, early hyperplastic epithelial lesions appear prior to palpable, clinically detectable cancer, and testosterone promotes this process; 6) testosterone promotes cancer development, and estradiol, dihydroxytestosterone, and castration inhibit cancer formation, underscoring the hormonal dependence of this model; 7) the addition of 4-hydroxyphenylretinamide (4-HPR) to the diet inhibits cancer formation, suggesting an antipromotional mode of action523; and 8) transplantable sublines metastasize to bone and cause lytic and blastic responses.525

Spontaneous prostatic adenocarcinoma arose at a 26% incidence in Lobund-Wistar rats at a mean age of 26 months, and the cancer grew into the entire prostate gland; 95% of the tumor-bearing rats had pulmonary metastases.523 The cancer in Lobund-Wistar rats appears to originate in the anterior and dorsolateral lobes of the prostate and seminal vesicle.522, 523 These spontaneous cancers are prevented by moderate (25%) dietary restriction over the lifetime of the rat.523 Dietary genistein can down-regulate the epidermal growth factor (EGF) and ErbB2/Neu receptors in the rat prostate with no adverse toxicity to the host animal and, thus, protect against and treat prostate cancer.526 Dietary soymeal is a promoter of prostate-seminal vesicle cancer in Lobund-Wistar rats.527

Diphosphonate and irradiation prevented the development of bone metastases from the transplantable adenocarcinoma PAIII in rats.528 The benzothiophene antiestrogen, raloxifene, was minimally cytotoxic against PAIII cells in vitro but inhibited metastasis of cells from the tail to the gluteal and iliac lymph nodes and the lungs of Lobund-Wistar rats.529 Thus, raloxifene may be useful in the treatment of hormone-resistant human prostate cancer.529 Thalidomide530 and cyclosporine A531 enhanced pulmonary metastases of subcutaneously implanted PAIII cells in Lobund-Wistar rats. Oral treatment with borpirimine, an aryl pyrimidone that is an immunostimulant, prevented growth of PAIII cells when injected subcutaneously and also caused complete regression of 95% of advanced cancers.

Implantation of testosterone in depot form by use of silastic membranes induced prostate adenocarcinoma in 32% of rats.532 The addition of 15% corn oil to the diet increased the yield of cancer, suggesting that a high-fat diet is involved in prostatic carcinogenesis.520 In aging conventional Lobund-Wistar rats, a 30% reduction in intake of dietary fat caused a 76% reduction in the incidence of spontaneous cancers, increased the latency period from 31 months to 34 months, and reduced the incidence of prostatitis by 76%.533 In germ-free Lobund-Wistar rats, the incidence of spontaneous cancer was only 20% that of conventional Lobund-Wistar rats when both were fed unrestricted diets. This low incidence in the germ-free Lobund-Wistar rat was increased two-fold by a moderately restricted diet.533

Treatment of MNU-treated rats with estradiol cyprionate, dihydrotestosterone, 4-HPR, or surgical castration inhibited the formation of cancer.523 When 4-HPR was administered in the diet, the incidence of metastasis was inhibited by 80% in male rats and by 23% in female rats.534 Rhodamine-123 was cytotoxic to these cancers.535 Tamoxifen, N-(4-hydroxyphenyl) retinamide, and a vitamin D analog, Ro24-5531, prevented MNU-initiated testosterone propionate-promoted cancer. Linomide, a quinoline-3-carboxamide, exerted antiangiogenic effects and inhibited the induction of cancer by MNU and testosterone by 50%.536 The nontumorigenic cell line from the Lobund-Wistar model, NP-152, did not form tumors in nude mice and retained many of the biologic properties of normal prostatic epithelial cells, including production of prostatic acid phosphatase (PAP), AR levels, and a requirement for several growth factors, including EGF.537 In contrast, the tumorigenic cell line, NP-154, lacked AR mRNA, had less stringent growth factor requirements, and was not stimulated significantly by EGF.

MNU acts as an initiator, and testosterone acts as a tumor promoter, likely by inducing cell division.538 The longer the period between MNU treatment and testosterone treatment, the higher the cancer incidence and the shorter the latent period.539

These cancers had high levels of procoagulant activity, and metastatic tumors had even higher activity, indicating that procoagulant activity may be a marker for malignancy in this model.540 The procoagulant activity was considered to be a complex between tissue factor and factor VIIa.540 Arachidonic acid levels were decreased 21%, indicating the remodeling of fatty acids.541

Carcinoma in situ was present in 1 of 8 rats that were killed at age 16 months, and early invasive cancer without metastasis was present in 2 of 39 rats at age 20 months.542 Significantly, there was no evidence of dysplastic or neoplastic changes in the seminal vesicles or in other portions of the prostate.542

There is some controversy regarding the site of origin of cancer in the Lobund-Wistar rat. In a single study, it was found that 73% of cancers occurred in the seminal vesicles, 22% occurred in other locations of the prostate complex and seminal vesicles, and 5% occurred in the coagulating gland (the anterior prostate).543 All hormone-induced cancers were adenocarcinoma, occurred in a large percentage (71%) of the animals, and were metastatic.543 Other studies also observed tumors in the seminal vesicles and the anterior prostate.544, 545 Ablation of the seminal vesicles resulted in a marked reduction in the incidence of cancer. A classification system for histologic scoring and grading of MNU and testosterone-induced prostate and seminal vesicle carcinoma in Lobund-Wistar rats has been reported.546

In summary, the Lobund-Wistar model appears to be useful for the study of mechanisms of age-related, spontaneous prostate cancer. This model also is valuable for evaluating dietary constituents that influence carcinogenesis and for screening phytochemicals that may reduce the risk of prostate cancer associated with certain foods.521

4.1.1.2 ACI/Seg rat

The aging ACI/Seg male rat is susceptible to spontaneous induction of cancer of the ventral prostate, with approximately a 17% incidence.547–549 This strain was derived by crossing an August (AUG) strain with an inbred Copenhagen (COP) rat with an Irish marker, hence the name, ACI.549 Subsequently, a colony of A X C rats was maintained separately by Dr. Segaloff, hence the designation ACI/Seg. This strain develops microscopic and macroscopic cancer of the ventral prostate as a function of age.547–549 The Copenhagen rat has an incidence of microscopic prostate cancer of only 10% by the end of life and an incidence of macroscopic cancer of < 1%. In contrast, the ACI/Seg rat has an incidence of microscopic cancer > 80% and an incidence of macroscopic cancer of 16%. Testosterone and estrogen levels were up to 5 times higher in ACI/Seg rats that developed macroscopic cancer compared with rats that did not develop cancer. Exogenous testosterone did not result in significant progression of microscopic or macroscopic cancer, suggesting that testosterone was not sufficient for such progression.549

Rats treated with dietary ethinyl estradiol and subcutaneous DMAB had a slight but not significant increase in cancer incidence compared with controls (8.1% vs. 3.7%, respectively).550 All tumors were microscopic. Atypical hyperplasia of the prostate dramatically increased in cases compared with controls (23% vs. 3.7%, respectively), as did simple hyperplasia (34% vs. 11%, respectively). All cancers were located in the ventral lobe.

The frequency of atypical hyperplasia increased from 20% to 73% when offspring and pregnant ACI/Seg rats were fed a high-fat diet, and prostate cancer also increased from 0% to 20%.551 Treatment with the 5-α-reductase inhibitor, FK143, inhibited the development of spontaneous macroscopic prostate cancer, suggesting that this inhibitor affected tumor progression.

4.1.1.3 Dunning rat

The R-3327 Dunning rat prostatic adenocarcinoma was derived from a spontaneous papillary adenocarcinoma of the dorsal prostate of an aging Copenhagen rat.552 A number of sublines have been derived from the R-3327 Dunning cancer, including the G and H sublines, and these have given rise to sub-sublines. There are a wide variety of sublines that vary from well differentiated, diploid, androgen-sensitive, and nonmetastatic to anaplastic, aneuploid, androgen-independent, and metastatic.521, 553 The mRNA is the same size (10 kB) in hormonally responsive and hormone-independent tumors,553 although hormonally independent tumors had lower levels of AR mRNA and protein.553

Point mutations in the first and second nucleotides of codon 12 of the Ki-ras gene are present in the Mat-LyLu and AT-3 cell lines, which metastasize to lymph nodes and lungs, suggesting that this mutation correlates with invasion and metastatic potential.554 The overproduction of urokinase is found in Dunning R-3327 Mat-LyLu cell lines, which have been selected for metastatic efficiency, resulting in increased frequency of metastases to the skeleton.555 Treatment of rats with a synthetic inhibitor of urokinase-type plasminogen activator (uPA), B-428, inhibited growth and metastasis of the primary tumor derived from the Dunning R-3327 Mat-LyLu cell line to the lungs and lymph nodes of Copenhagen rats.556

The Dunning R-3327 rat prostate tumor contains receptors for androgens and LHRH,553 similar to human cancer.557 The LHRH antagonist, SB-75, and a somatostatin analog, RC-160, when administered together, are effective against the transplanted Dunning R-3327H rat prostate tumor.558 Tamoxifen inhibits the growth of transplanted Dunning R-3327 cells in Copenhagen-Fischer rats, presumably by indirectly lowering levels of circulating testosterone.559 Farnesyl diphosphate synthase (FPPS) is an androgen-response gene in the rat ventral prostate. In situ hybridization studies have shown that FPPS expression is localized to prostatic epithelial cells. FPPS may play an important role in androgen regulation of prostatic epithelial cell apoptosis, proliferation, and/or differentiation.560

Both the Dunning cancer and human cancer are responsive to androgens, and they regress transiently when androgens are removed, have a similar histology, and possess 5-α-reductase activity.561, 562 Flutamide and castration inhibit growth of the Dunning R-3327 cancer and human cancer. A new, potent 5-α-reductase inhibitor, turosteride, inhibits growth of the Dunning R-3327 cancer by up to 45%.563

The Dunning cancer subline, AT2.1, secretes an autocrine motility factor that may favor metastasis.564 Thymosin β-15 also is up-regulated in highly motile and metastatic Dunning cancer cell lines.565

Orchiectomy or treatment with the antiandrogens cyproterone acetate and flutamide reduced the growth of transplanted R-3327H cells in Copenhagen rats.566 This tumor has aberrant staining for the extracellular matrix protein tenascin that correlates with the stromal content of the tumor, with high amounts of tenascin at sites where the cancer invades neighboring tissues and near-necrotic areas.566

Voltage-activated sodium channels were found in the highly metastatic Dunning R-3327 Mat-LyLu cell line, but they were not found in a minimally metastatic AT-2 cell line.567 Similarly, the highly invasive human cell line, PC-3, had the α subunit of the voltage-activated sodium channel, but the LNCaP cell line, which is less invasive in vitro, did not, suggesting that the presence of voltage-activated sodium channels in rat and human prostate cancer correlates with invasive and metastatic ability.568 Decreased expression of E-cadherin occurs in invasive lines and sublines of the Dunning R-3327 G and H tumors.569 Dunning R-3327 tumor cell sublines overproduce transforming growth factor β 1 (TGF-β-1),570, 571 and overproduction in the R-3327-Mat-LyLu subline causes more rapid growth and more metastases to lungs and lymph nodes.570 TGF-β is a growth stimulant of osteoblasts, and its presence in one metastatic subline may account for the formation of osteoblastic bony metastases.572

The origin of the R-3327H Dunning prostatic adenocarcinoma has been questioned.573, 574 The original cancer was classified as a papillary adenocarcinoma with glandular structures similar to those found in the rat dorsal prostate.573 The Dunning H tumor lacks proteins found in the dorsal prostate, including DP I, DP II, and secretory transglutaminase. A number of breast-specific proteins are present in the Dunning H tumor, including milk-fat globule membrane protein, suggesting that it may be of breast origin.573 This tumor has been passaged many times through animals and likely has undergone significant dedifferentiation, population evolution, selection, and alterations in gene expression, but it retains androgen responsiveness.

Human prostatic inhibin suppresses growth in vivo and clonogenic survival of the androgen-responsive Dunning R-3327 G and highly metastatic Mat-LyLu cell lines.575, 576 It also blocks follicle-stimulating hormone stimulation of the Dunning R-3327 rat prostate cancer cell line, Mat-LyLu, and the human prostate cancer cell line, PC-3. Antagonists of bombesin/gastrin-releasing peptide combined with agonists of LHRH exert strong inhibitory activity against the Dunning R-33237H cancer.577 Treatment of cancer-bearing rats with an aryl pyrimidone, bropirimine, and prostatectomy decreased residual cancer, decreased the expression of proliferating cell nuclear antigen (PCNA) and TGF-β-1, and prolonged survival.578 Bropirimine may play a role in the treatment of human prostate cancer as adjuvant therapy in combination with prostatectomy.579

Difluoromethylornithine (DFMO), which is an inhibitor of ornithine decarboxylase (ODC), inhibits the growth of Dunning R-3327 cancer and Dunning sublines in vitro and in vivo.580 There is a positive correlation between growth rate and ODC activity in Dunning cancers.580 Growth of the Dunning R-3327 Mat-LyLu cancer is inhibited by the negatively charged polysaccharide pentosan polysulfate581 and the quinoline 3-carboxamide linomide, possibly by their ability to inhibit tumor angiogenesis.582 Linomide and surgical castration inhibit growth of the Dunning R-3327 PAP and G sublines in Copenhagen-Fischer F1 hybrid rats.583 Colchicine is cytotoxic to the Dunning Mat-LyLu cancer and to the human cell line, PC-3, perhaps because it inhibits motility and growth.584 There is an association between high levels of glyceraldehyde-3-phosphate, cell motility, and metastatic potential in sublines of the Dunning cancer.585 In addition, there is expression of high-mobility group protein I(Y) in highly metastatic Dunning tumor sublines.586

A combination of the partial antiestrogen zindoxifene and cisplatin are effective against the Dunning R-3327G cancer. The addition of O-6-benzylguanine to 1,3-bis(2-chloro-ethyl)-1-nitrosourea (BCNU) further inhibited growth of the Dunning cancer and reduced the systemic toxicity of BCNU by allowing lower concentrations of this agent to be used, thus sparing the bone marrow.587 Treatment with 12(S)-hydroxyeicosatetraenoic acid increased the motility and invasiveness of the R-3327 cell line AT2.1, which is a variant with low metastatic ability, and these properties were increased by thymelea toxin, which is a selective activator of protein kinase α.588 The overexpression of protein kinase C-ζ inhibited invasiveness and metastasis of the Dunning R-3327 Mat-LyLu cancer line in Copenhagen rats.589 A derivative of citrus pectin inhibited spontaneous metastasis of the Dunning Mat-LyLu cancer to the lungs of rats.590 Microcell-mediated chromosome transfer assays reveal that human chromosome 8p21-p12 suppresses metastasis of the Dunning cancer AT6.2 cell subline.591, 592 The human CD44 gene is a metastasis suppressor gene for rat prostatic cancer, and its expression is down-regulated in highly metastatic Dunning R-3327 sublines, such as AT6.1.593

The cytotoxic effect of the combination of taxol and radiation therapy was more effective than either agent alone against Dunning R-3327 G cancer in vivo and in human prostatic adenocarcinoma cell lines in vitro.594 Combination therapy may be of particular utility, because recent reports indicate that treatment of the Dunning R-3327 cell lines with doxorubicin results in multidrug resistant cell lines.595

In cell lines that were derived from the androgen-sensitive rat prostate carcinoma Dunning G, transfection of bcl-2 renders them more resistant to the apoptotic effects of the cancer chemotherapeutic drugs, doxorubicin and suramin.596 Oral cyclophosphamide suppressed the Mat-LyLu subline when it was transplanted into rats.597 Similarly, the combination of the cancer chemotherapeutic drugs vinblastine and tamoxifen suppressed the growth of Mat-LyLu tumors by 74% in male Copenhagen rats.590

The antiandrogen bicalutamide reduced the growth of the Dunning R-3327H transplantable cancer in rats.598, 599 Amiloride, an inhibitor of uPA, inhibited the growth of highly metastatic AT-3 tumors and inhibited lymph node metastasis.600 Studies utilizing the Mat-LyLu subline showed that transfection of a gene encoding human granulocyte–macrophage-colony stimulating factor, when irradiated and added back to cancer-bearing animals, increased their lifespan.32 In summary, the Dunning cancer may be a good model for spontaneous prostate cancer in rats that has relevance to spontaneous prostate cancer in humans.

4.1.1.4 Fisher F344 rat

Intermittent injection of 3,2′dimethyl-4-aminobiphenyl (DMAB) and high doses of testosterone propionate in Fisher F344 rats induced a high incidence adenocarcinoma in the dorsolateral and anterior lobes of the prostate and in accessory sex organs.538, 601 This model is not overly dependent on the strain of rat, because both Fisher F344 rats and out-bred Wistar WU rats respond to this treatment regimen.538 Approximately 50% of adenocarcinomas induced by DMAB are microscopic, allowing determination of the anatomic origin. Treatment of intact and castrated rats for short periods with testosterone propionate and N-nitrosobis-(2-oxopropyl)-amine induced carcinoma in < 5% of rats.602 The combination of DMAB and ethinyl estradiol, with or without methyltestosterone, resulted in an 85% incidence of high-grade PIN in the ventral lobe of Fisher F344 rats.538, 603 Administration of DMAB in combination with exogenous testosterone produced palpable and metastatic tumors in portions of the accessory sex organs but did not produce carcinoma in the ventral prostate of Fisher F344 rats.604 A diet containing beef tallow (high fat) promoted carcinogenesis in this model.605

These cancers have low androgen dependence at the early stages of carcinogenesis, and androgen ablation lowered the incidence of atypical hyperplasia but not carcinoma.606 Intermittent administration of testosterone propionate and/or DES was promotional only slightly for DMAB-initiated carcinogenesis.607 Cell lines have been derived from androgen-independent cancers that metastasized. The chemically induced cancer in this system is invariably androgen-independent and metastatic.608 Epristeride (5α-reductase inhibitor) and casodex (pure antiandrogen) showed a dose-dependent promoting effect on rat ventral prostate carcinogenesis.604

Finasteride and bicalutamide prevented cancer in the F344 rat prostate609 induced by DMAB and testosterone propionate.601 However, these drugs did not prevent cancer at the microscopic level; the authors concluded that they prevent progression to macroscopic, life-threatening carcinoma.610

One of the heterocyclic amines found in cooked fish and meat, PhiP, has been evaluated as a carcinogen in the rat prostate.611 This compound is a carcinogen in the mammary gland of female F344 rats and in the colon of male rats. Administration of PhiP to F344 rats for 52 weeks induced DNA adducts throughout the rat prostate and seminal vesicles as well as in the colon, pancreas, and liver. Atypical hyperplasia was induced in the ventral and anterior prostate and in seminal vesicles at incidence rates of 82%, 30%, and 89%, respectively, and carcinoma was induced in the ventral prostate in 67%.611 These observations suggest that dietary intake of PhiP in cooked fish and meat may contribute to the high incidence of prostate cancer in Western men.611

4.1.1.5 Noble rat

The Noble rat is moderate in size, hardy, extremely docile, and uniquely susceptible to estrogen-induced prostate cancer612–616 and breast cancer.617, 618 The animal also has a relatively short life span.

Adenocarcinoma of the dorsolateral prostate develops spontaneously between ages 12 months and 23 months in 0.5% of untreated Noble rats.615 Treatment with testosterone for 54–59 months induced tumors in 20%.613 Coadministration of estrone (a weak estrogen) significantly shortened the latency period for tumor development to 43 weeks. Simultaneous treatment of rats with testosterone and estradiol-17β (a strong estrogen) increased tumor incidence to 3% after 4 months,615, 619, 620 to 73% after 18 months, and to nearly 100% in a life-time study.616 All carcinomas were small, microscopic, well differentiated, and located principally in the periurethral ducts of the dorsal, lateral, and anterior prostate,616 sparing the seminal vesicles, ventral prostate, and urethra. Inoculation of MNU and subcutaneous implants of testosterone propionate resulted in early-stage adenocarcinoma in the seminal vesicles and anterior prostate of 41% of rats. The tumors were distinct histologically, either glandular or relatively gland-free, depending on the site from which they were derived. No tumors developed in the dorsolateral or ventral lobes of the prostate after this treatment.621

Dysplasia, similar to human PIN, developed in all rats that were treated with testosterone and estradiol for 16 weeks.619, 620 The onset of dysplasia was accompanied by a sevenfold increase in epithelial mitotic activity,620 overexpression of the ras protooncogene,622 and activation of the TGF-α/EGF receptor (EGFR) signaling pathway.623 Collectively, these changes suggested a resurgence of proliferation in the dorsolateral prostatic epithelium, which normally exhibits a low proliferative rate. Dual hormone treatment also triggered a hydroxyl free radical-related genotoxic response. An increase in DNA single-strand breaks210 and the appearance of a unique DNA adduct211 were observed. Catechol estrogen formation was negligible,624 suggesting that metabolic conversion of estrogens to genotoxic metabolites was not responsible for the genomic damage. In contrast, the amount of lipid peroxidation fluorescence products doubled,210 and the expression levels of MT and intracellular antioxidants were increased markedly. These findings suggest that increased oxidative stress may contribute to free radical-related genomic damage in this model system. Evidence for apoptotic dysfunction also was noted in association with dysplasia.625 Expression of the apoptosis-associated genes, TRPM-2/clusterine625 and bcl-2,626 was disrupted focally in dysplastic foci, yet signs of apoptosis were not present. Coadministration of bromocryptine to treated rats partially suppressed induction of dysplasia,624 whereas cotreatment with the pure ER antagonist ICI 182,980 completely blocked the development dysplasia.626 These findings indicate involvement of pituitary prolactin and tissue ER in the development of dysplasia in this rat model. Using tissue microdissection, ER was present principally in the prostatic epithelium, whereas ER was found exclusively in the stroma. Dual hormone treatment did not alter the levels of either ER-β or ER-α, but it did elevate progesterone receptor expression focally in dysplasia, suggesting a greater estrogen response in dysplastic than in benign acini.627

4.1.1.6 Wistar rat

Twenty-five percent of Wistar rats developed adenocarcinoma in the dorsolateral prostate when they were treated with the combination of cyproterone acetate, testosterone, and N-methyl-nitrosourea. All tumors were differentiated moderately or poorly, and 40% of tumors metastasized. Epithelial hyperplasia and carcinoma in situ also were found occasionally.628 Other types of cancers arose in other organs. Treatment of Wistar rats with DES from birth for 30 days followed by castration induced squamous cell carcinoma in the dorsolateral prostate and coagulating gland.629, 630

A single injection of cadmium chloride in rats induced dysplasia in the ventral prostate of Wistar rats when a low dose of cadmium was used to minimize testicular toxicity, if the cadmium was injected intramuscularly, or if an amount of zinc was added simultaneously.418, 430, 439 The expression level of MT is low in the ventral lobe of the prostate, and this may correlate with the susceptibility to cadmium-induced carcinogenesis.418 MT was more highly inducible in the dorsolateral prostate, a site that is resistant to cadmium-induced carcinogenesis.432, 631 There was minimal basal activity and a lack of cadmium-inducible activation of expression of the MT gene. These results indicate that intact testicular function is necessary for cadmium-induced prostatic carcinogenesis and that androgens may act as a promoter in this system.632

Cancer developed commonly in Wistar rats that were treated by sequential chemical castration, stimulation, and MNU exposure. MNU-induced prostate/accessory sex gland adenocarcinoma or poorly differentiated carcinoma contained activating mutations in codon 12 of the Ki-ras oncogene in 80% of primary tumors and in 66% of metastases.633

4.1.1.7 Other rat models

Other agents induce cancer in a variety of rodent and murine models. N-nitrosobis (2-oxopropyl)-amine and N-nitrosobis (2-hydroxypropyl)-amine induced prostatic cancer in MRC rats.634 Irradiation induced prostate cancer in ICR/−JCL mice.635 Pelvic irradiation induced prostate carcinoma in 33% of Sprague-Dawley rats, but only after castration and androgen-replacement therapy, suggesting that testosterone was required for cancer development.632, 636

In Sprague-Dawley rats, the administration of a high-fat diet increased the weight of the ventral prostate. This may provide a useful model for studying the molecular mechanism by which dietary fat influences the normal prostate.637 Exposure to dietary genistein resulted in down-regulated AR and ER in the dorsolateral prostate. Genistein also reduced mRNA expression of AR and ER. Down-regulated sex steroid receptor expression may explain the lower incidence of prostate cancer in populations with diets that contain high levels of phytoestrogens.638

4.1.2 Murine models

4.1.2.1 Transgenic models

Transgenic mouse models are employed commonly now to study prostate cancer. The utility of these models is that the investigator can make a construct containing a promoter of a gene that is expressed specifically in the prostate and a gene of interest, such as an oncogene, and can study the phenotypic effects of expression of that oncogene in the prostate. Constructs have been developed that function to disrupt tumor suppressor genes, including the SV40 large-T antigen, which binds to both the retinoblastoma and p53 tumor suppressor genes.632

The early region of the SV40 genome, including the large-T and small-t, antigen-encoding genes, under control of the long terminal repeat (LTR) sequences of the mouse mammary tumor virus (MMTV), induced carcinoma of the prostate in transgenic mice, but the tumors were not characterized well.639, 640 The int-2 gene, which is a member of the fibroblast growth factor (FGF) family, was placed into a construct under the transcriptional control of the LTR of MMTV. Transgenic mice bearing this construct showed high levels of expression of the int-2 gene in the prostate, with resultant hyperplastic growth of the epithelium. Hence, overexpression of the int-2 gene in the prostate may be a good model for the induction of BPH.

bcl-2 is a protooncogene that has been implicated in human prostate carcinogenesis. One study examined the impact of bcl-2 expression on the pathogenesis and progression of prostate cancer in a transgenic model. Bcl-2 transgenic mice had a decrease in epithelial apoptosis within the prostate.641

The transgene that contains the gene encoding TGF-α was placed under the control of the enhancer-promoter region of the MT-1 gene. Transgenic mice containing this construct exhibited high levels of TGF-α mRNA and showed epithelial hyperplasia and dysplasia, similar to carcinoma in situ in the ventral prostate (coagulating gland), and had dysplastic areas in mesenchymal cells.642, 643 Human prostatic adenocarcinoma cells in culture also synthesize high levels of TGR-α mRNA, and they grow in response to exogenously added TGF-α.643, 644 Therefore, hyperplasia of the prostate in transgenic mice that bear a construct containing the TGF-α gene may provide support for the contributing role of TGF-α in human prostate cancer.643 Male transgenic mice that carry a construct containing the 5′ flanking sequences of the gp91-phox gene and the early region of the SV40 genome developed neuroblastoma of the prostate gland, apparently due to construction of a novel transcription signal during synthesis of the construct.643, 645, 646

The rat C3(1) gene also has been targeted for expression in the prostate. The rat C3(1) gene encodes 1 polypeptide of a protein called prostatein647 or steroid-binding protein,648 which is synthesized and secreted at high levels by rat ventral epithelial cells.645, 649–651 The expression of this gene is restricted to the prostate of the rat, and its expression represents 10% of the total mRNA synthesized.650, 651 Transgenic mice containing the C3(1) transgene exhibited specific expression in the prostate.645, 652

Transgenic mice containing a C3(1)-SV40 large-T antigen-encoding gene construct developed carcinoma of the prostate within age 1 year.653 This transgenic mouse line was utilized to study the progression from benign epithelium to PIN and carcinoma.521, 654 Given the long interval of time between PIN and invasive carcinoma, this model is unique for comparing genetic alterations between PIN and prostate carcinoma and may be an ideal model for the study of chemoprevention.521 There was an increase in SV40 large-T antigen and p53 protein in the ventral and dorsolateral prostate with progression. Two of 18 foci of PIN, 1 low-grade PIN and 1 high-grade PIN, had mutations in the c-Ha-ras gene, and 33% of adenocarcinomas had mutations in the c-Ki-ras gene.654 Similarly, ras gene mutations were found in 18% of PINs and 33% of adenocarcinomas. A cell line that was developed from one of these adenocarcinomas was tumorigenic in transgenic mice and athymic mice but not in syngenic FVB/N mice: This cell line may be useful to delineate factors underlying the immune rejection of prostate tumor cells.655

The transgenic adenocarcinoma of the mouse prostate (TRAMP) model utilizes a construct that consists of the prostate-specific probasin promoter from the rat driving expression of the SV40 large-T antigen-encoding gene in the background of the C57/Bl6 mouse.656 The rat probasin gene is expressed specifically in the dorsolateral epithelium of the prostate. In this model, mice that hade this transgene and that expressed high levels of transgene mRNA showed specific expression in the dorsal and ventral lobes. TRAMP mice showed a predictable progression from epithelial hyperplasia to cancer within 18 weeks. The tumors showed high levels of expression of SV40 large-T antigen, elevated levels of p53 protein, and decreased AR levels.656–658 The administration of flutamide significantly decreased the incidence of prostate cancer and increased the latency period in the TRAMP model.659 Egr1 deficiency significantly delayed progression from PIN to invasive carcinoma in this model.660 Unique advantages of this system are that the transgene is expressed specifically in epithelial cells of the prostate: The tumor histologically resembles human prostate cancer; it has a short latency; it arises in 100% of animals; and mice show progression from normal prostate to primary adenocarcinoma and, ultimately, metastases to lymph nodes, lungs, bone, kidney, and adrenal glands.661–663 Another important advantage to the TRAMP model is that tumor development initially is androgen-dependent; however, once they are castrated, only 20–30% of TRAMP mice will demonstrate durable responses and remain tumor free, whereas 70–80% will develop androgen-independent and poorly differentiated prostate cancer. This supports the hypothesis that androdysgenesis is a stochastic event in TRAMP mice.664 Cell lines established from an adenocarcinoma from TRAMP mice have been used to study events that mediate tumor progression in this transgenic mouse model system for prostate cancer665 and to determine whether manipulating the T-cell costimulatory pathway holds promise for the immunotherapy of prostate cancer.666 In one study, it was observed that phenotypic variability in tumor and pathologic progression in this model occurred as a function of genetic background.663

A third system utilizing the SV40 large-T antigen to develop transgenic mice that develop prostate cancer also has been developed.667 In this system, a construct comprised of the human fetal G-γ-globin 5′ promoter was linked to the SV40 large-T antigen-encoding gene and then used to generate transgenic mice.667 In 1 of these transgenic lines, the G-γ/T-15 transgene line, 50% of mice developed mixed adenocarcinoma-neuroendocrine prostatic carcinoma by 7 months.668 Adult tissues, but not fetal tissues, expressed SV40 large-T antigen. These prostate cancers originated in the dorsal or ventral lobes and proceeded through PIN, to cancer, and then metastasis to lymph nodes and other sites.643, 662

Two prostate-specific genes, ECO:RI and c-fos, have been implicated in the induction of genomic instability. The progressive presence of regions of mild-to-severe hyperplasia, low-grade and high-grade PIN, and well differentiated carcinoma was observed in the ECO:RI transgenic mice, but no significant pathology was found in c-fos transgenics. Levels of PCNA and Ki-67 were elevated and correlated with the severity of prostatic lesions. Preneoplastic and neoplastic stages in the ECO:RI model were similar to the early stages of human prostate cancer. This model may be useful for studying the mechanisms involved in the early stages of prostate carcinogenesis and the nature of subsequent events that are necessary for progression to advanced disease.669

Endostatin, which is an antiangiogenic protein, delays the development of spontaneous tumors in the transgenic model.670 DFMO inhibits ODC, which is a protein found in high levels in the dorsolateral prostate of the TRAMP model. ODC may be a promising target for the screening of novel drugs and chemopreventive regimens.671

Genistein, which is the primary isoflavone of soy, has been under investigation for its ability to protect against prostate cancer. A transgenic mouse model was used to test this hypothesis. Those investigators found that genistein in the diet reduced the incidence of poorly differentiated prostatic adenocarcinoma. It also down-regulated AR, ER-α, progesterone receptor, EGFR, IGF-I, and extracellular signal-regulated kinase 1. They concluded that dietary genistein protects against prostate cancer by regulating specific sex steroid receptors and growth factor signaling pathways.672 In another study, it was found that dietary genistein significantly reduced the incidence of prostate lesions in TRAMP mice.673 In summary, transgenic models provide the ability to dissect the carcinogenesis process by manipulating the expression of specific oncogenes or tumor suppressor genes with the prostate.

4.1.2.2 Mouse prostate reconstitution (MPR systems as models of prostate cancer

The MPR model consists of dissected mouse urogenital sinus tissue that is modified genetically in vitro, reconstituted with mesenchymal tissue, implanted under the renal capsule of mice, and allowed to differentiate.632, 643, 645, 662, 674, 675 MPR grafts grow in the mouse into which they are grafted and organize ductal structures similar to ducts of the normal mouse prostate. This technique has revealed that mesenchyme has the ability to influence the development of genital organ systems in mice.632, 643, 645, 662, 674–676

The elements of MPR were cultured, the embryonic prostate cells were transduced with recombinant retroviruses that expressed activated v-Ha-ras and/or v-gag-myc (MC29) oncogenes, and the transduced cells were then implanted under the renal capsule of an adult mouse. After completion of retrovirus-mediated transfer of activated v-Ha-ras gene into fetal urogenital sinus (containing both mesenchymal cells and epithelial cells), the reconstituted prostate developed dysplastic mesenchyme and focally hyperplastic epithelium.677 Transduction of the v-gag-myc oncogene alone resulted in focal hyperplasia, whereas transduction of both the v-Ha-ras gene and the v-gag-myc gene on the same vector resulted primarily in adenocarcinoma.643, 662, 677

In the MPR model, the urogenital sinus epithelium can be separated and transduced alone, and the urogenital sinus mesenchyme also can be isolated and transduced separately. In the C57-BL/6 background, when only the urogenital epithelium was infected with the activated v-Ha-ras gene and an activated v-gag-myc gene, primarily, focal hyperplasia developed, and carcinoma rarely developed. Primarily, carcinoma formed when both the urogenital sinus epithelium and the urogenital sinus mesenchyme were transduced with v-Ha-ras and v-gag-myc, then reconstituted, and implanted.632, 643, 662, 677 The epithelial cells and the mesenchymal cells both must be transduced to result in a high yield of carcinomas, implying cooperation between the two tissue types. The change occurring in the mesenchyme that allows prostate carcinoma development643 may depend on activation and/or overexpression of the TGF-β1 and TGF-β3 genes, because high, steady-state levels of transcripts of these genes are found in carcinoma.643, 678, 679

Some of the experimental results obtained with the MPR system in mice are strain-specific. After transduction with activated ras and myc genes, urogenital sinus tissue derived from C57-BL/6 mice yielded poorly differentiated adenocarcinoma. In contrast, transduction of urogenital sinus tissue from BALB/c mice only formed hyperplastic glands.643 Genetic differences between these 2 mouse strains may render the C57-BL/6 strain more susceptible to carcinogenesis by transduced oncogenes.

The MPR system has been employed in knockout mice. Using urogenital sinus tissue from p53 knockout mice resulted in the rapid progression of cancer in prostate tissue transduced with activated ras and myc oncogenes.632, 680–690 The MPR system is most useful in studies of genetic characteristics, biochemical properties, biologic behavior, histology, and cancer progression.521

4.1.2.3 SCID mouse models

Human prostate cancer or cell lines can be implanted into mice with SCID syndrome. These SCID mice have an abnormal recombinase system. They do not assemble antigen receptor genes efficiently in developing lymphocytes.691 In addition, they have lost the ability to rearrange both immunoglobulin (IG) and T-cell receptor (TCR) gene segments, and, hence, have an absence of functional B cells and T cells.692 The are deficient in all major classes of IGs693 and are very susceptible to tumor formation from implants. SCID mice are more conducive to the study of metastasis than athymic (nude) mice, because athymic mice have a high natural killer cell activity that inhibits tumor cell metastasis. Prior to l991, it was difficult to grow primary human prostatic adenocarcinomas > 3 months in vivo or in vitro,694 but it is feasible now to grow human cancer in nude mice and in SCID mice.

A combination of in vitro selection with Boyden chambers and in vivo selection in SCID mice allowed selection of sublines from PC-3 human prostatic adenocarcinoma cells that exhibited preferential metastasis to specific organs.695 Experiments using different tumor cell implantation sites suggested the regulation of organ-preferential metastases by the host. Taxol significantly inhibited the growth of PC-3 ML cells and their metastases to the lumbar vertebrae in SCID mice.696 Injection of taxol into SCID mice followed by injection of the human PC-3 ML human prostatic adenocarcinoma cell line caused a stabilization of the microtubules in these cells, reduced the number of circulating tumor cells, and prevented metastases to the bone in SCID mice that were injected with these human prostatic adenocarcinoma cells.697 Both liarozole fumarate (a novel benzimidiazole derivative) and with 13-cis-retinoic acid inhibited the formation of subcutaneous tumor growth and bone metastasis.698 Remarkably, neither of these drugs reduced the proliferation or survival of this metastatic human prostatic adenocarcinoma cell line in vitro.

In a recent study, SCID mice were engrafted with human adult bone and lung tissue to investigate the mechanistic basis of the apparent tropism of prostate cancer cells for bone. Eight weeks after engraftment of the human tissue, LNCaP and PC-3 cells had metastasized to human adult bone in 35% and 65% of the mice, respectively. The LNCaP tumors were osteoblastic, whereas PC-3 tumors induced osteolytic lesions without any surrounding osteogenic response. The cancer cells exhibited little or no metastases to the implanted human adult lung, mouse bone, or native mouse bone. This model may provide a useful tool for identifying and analyzing aspects of prostate cancer metastasis that cannot be addressed in conventional models.699

The metastasis suppressor gene for prostate cancer, KAI1 on human chromosome 11p11.2, suppressed metastasis of AT6.1 rat prostate cancer cells in SCID mice.700 The expression of this gene was reduced dramatically in human cell lines that were derived from metastatic prostate cancer, including PC-3, LNCaP, TSU-Pr1, and DU-145. This gene is a member of the transmembrane 4 (TM4) gene family.700

The DU-145, PC-3, and 431P human cell lines have various amounts of integrin subunits, and the form multiple microscopic cancers and macroscopic peritoneal cancers in SCID mice. The LNCaP tumor cell line does not express these integrin subunits but does express α-v β-3 subunits, which are of lower molecular weight and may be inactivated, because these cells did not bind to vitronectin and did not form cancer in SCID mice. It is noteworthy that LNCaP cells incubated with extracellular matrix components formed cancer in nude mice.701 Three tumorigenic cell lines expressed the β-1 integrin subunit and 2 other integrin subunits that may be related to tumorigenicity.702

The metalloproteinase gene, which encodes matrilysin, was transfected into DU-145 cells, causing invasion of the diaphragm in 66% of SCID mice, compared with 11% of mock vector-transfected mice.703 Orthotopic and ectopic injection of DU-145 cells in SCID mice revealed similar levels of expression of four genes related to invasion and metastasis (matrilysin, stromelysin, tissue inhibitor of matrix metalloproteinase 1 [TIMP-1], and TIMP-2).704 Furthermore, DU-145 cells were invasive equally whether they were injected intraperitoneally or into the dorsolateral lobe of the prostate. Intraprostatic inoculation of LNCaP cells resulted in cancer in 89% of SCID mice; 100% of those mice developed retroperitoneal lymph node metastases, and 40% developed microscopic pulmonary metastases. Subcutaneously injected LNCaP cells yielded cancer in 100% of cases but did not metastasize. The tumor yield was lower in nude mice (60% for both intraprostatic and subcutaneous injection).

The Kirsten-ras-revertant 1 (Krev-1) tumor suppressor gene was transfected into the human prostate cancer cell lines PC-3, TSU-Pr1, and DU-145. Only the Krev-1 transfectants of the PC-3 and TSU-Pr1 cell lines had significantly reduced ability to grow in soft agar and formed significantly smaller tumors when injected into SCID mice.705

The human prostatic adenocarcinoma cell lines TSU and PC-3 were grown in SCID mice that had been treated with the antiangiogenic agent linomide, resulting in apoptotic cell death in the cell lines. Linomide decreased the microvessel density and enhanced the areas of necrosis and apoptosis in the resultant tumors.582 Androgen ablation by castration potentiated the antiangiogenic ability of linomide,706 and the apparent mechanism of this effect was down-regulation of vascular endothelial growth factor (VEGF).

A xenograft established from the cell line 1013 L in SCID mice707 formed tumors in SCID mice when it was coinjected with a sterile gelatin sponge. This cell line had barely detectable levels of uPA in vitro or in vivo. The DU-145 cell line grew well in SCID mice, and the urokinase inhibitor p-aminobenzamidine inhibited growth and reduced levels of the uPA in the membranes of the tumors in SCID mice. Various inhibitors of urokinase, including p-aminobenzamidine, amiloride, PAI-1, and its mutant forms, reduced the growth of xenografts of human prostate cancer in SCID mice.708

The percentage of SCID mice with cancer growth varied from 33% to 100%, depending on the strain of SCID mouse.709 The highest cancer yield was obtained using intact cancer tissue; cell suspensions did not grow in the SCID mice. BPH tissue also grew efficiently in the SCID mice, and human interleukin 2 (IL-2)-activated tumor-infiltrating lymphocytes also were grown.

Hpg/hpg scid/SCID mice with the mutation “hypogonadal,” have very low levels of circulating steroid hormones, including testosterone. Transplantation of prostate cancer tissue into these mice allows for the selection of androgen-independent cancers, which may mimic the development of androgen-independent cancer from initially androgen-sensitive tumors. This model may have significant utility for testing cancer chemotherapeutic agents.

A high-efficiency method has been developed to label cells of the Dunning prostate cancer cell lines with the enzyme β-galactosidase utilizing a replication-defective retrovirus that contains a β-galactosidase expression cassette.710 This system is useful for marking Dunning cells injected into SCID mice, and it allows the detection of metastatic cells within target organs. Identification of such cells may allow identification of genes that promote or suppress metastasis.

4.1.3 Canine model

The dog is the only nonhuman species in which spontaneous prostate cancer occurs frequently.521, 711, 712 Prostate cancer in dogs is aggressive clinically, with frequent metastases to regional lymph nodes, lung, and bone.713, 714 Many cases occur in castrated dogs, suggesting that some tumors are independent of testicular androgens.713 Dogs with spontaneous prostate cancer provide novel sources for the isolation of endogenous angiogenesis inhibitors and new cellular model systems to define further the mechanisms by which tumor cells generate angiogenesis inhibitors.715 One study indicated that two basal cell populations exist in the canine prostate. The age-related expansion of proliferative acinar basal cell populations, mediated by sex steroids, is important in the pathogenesis of canine prostatic hyperplasia. Leav et al. suggested that prostatic carcinoma in dogs arises from ductal cells. Canine acinar basal cells and ductal epithelium may have separate susceptibilities to factors that promote hyperplastic and neoplastic development.716

The identification of high-grade PIN in the prostates of dogs with carcinoma provided the first evidence of spontaneously occurring PIN in animals.717 High-grade PIN, the most likely precursor of human prostate cancer, is present in the majority (66%) of prostates of dogs with prostate cancer, similar to the 85% prevalence of PIN reported in men.718 High-grade PIN in the dog prostate was similar histologically to human PIN. These similarities include proliferation and heaping up of the luminal secretory cells that line preexisting ducts, ductules, and acini; cytologic abnormalities included nuclear crowding, variation in nuclear size and shape, and nucleolar enlargement. The prevalence of basal cell layer disruption in human and canine acini with PIN was similar. Basal cell layer disruption was present in 56% of human acini with high-grade PIN718 and was identified in 72% of canine acini with high-grade PIN. The proliferative index of canine PIN was greater than that of benign epithelium but less than that of carcinoma. The proliferative index for canine prostate cancer was considerably greater than that reported for human prostate cancer.718 Microvessel density, a quantitative measure of angiogenesis, was greater in PIN and carcinoma in dogs than in benign prostatic epithelium. Canine models of PIN may provide an opportunity to evaluate rapidly the efficacy of promising chemopreventive agents.

A recent study evaluated the potential of adenovirus-mediated gene transfer to the prostate of normal laboratory beagles. Seven days after injection, expression of the transgene was observed exclusively within prostate epithelial cells. That study showed the capacity of human adenovirus to infect canine prostate cells and constituted the first evaluation of adenovirus-mediated gene transfer in dog prostate. It will provide a basis for the gene therapy treatment of prostate cancer patients.712

Using the in vivo canine model of prostatic carcinogenesis, investigators evaluated whether dietary supplementation with selenium influenced surrogates of prostatic carcinogenesis, including the extent of DNA damage (alkaline comet assay) and epithelial cell apoptosis (terminal deoxyuridine triphosphate nick-end labeling [TUNEL] assay). The study evaluated the effect of 7 months of supranutritional selenium on the prostate of sexually intact beagle dogs that were physiologically equivalent to men ages 62–69 years and free of prostate cancer. Dogs were assigned randomly to an untreated control group or to a treatment group that received selenium at 3 μg/kg or 6 μg/kg body weight per day as either selenomethionine or high-selenium yeast. The extent of DNA damage in prostate cells was reduced significantly in selenium-supplemented dogs compared with controls (P < 0.0001). Selenium-supplemented dogs had a 2-fold increase in the median number of apoptotic prostatic epithelial cells compared with controls (P = 0.04). That study provided in vivo evidence that dietary supplementation with selenium can exert beneficial effects on DNA damage and on epithelial cell apoptosis within the aging prostate.

There are two main advantages to using canine models for experiments related to prostate cancer: 1) They provide a spontaneously occurring, large animal model that enables certain imaging and therapeutic interventions, and 2) their compressed life span enhances the practicality of trials that test potential chemopreventive agents.521 An animal model, such as the dog, that is vulnerable to spontaneous prostate cancer can provide valuable information regarding whether agents may have a beneficial influence on cellular and molecular processes within the prostate that are important in the pathogenesis and or progression of human prostate cancer. Such animal studies compliment ongoing human trials in pursuit of developing practical strategies for prostate cancer prevention in terms of selecting an agent, dose, or combination of chemopreventive agents.

4.2 Xenograft Models

Prostatic cancer cells grown in culture fail to maintain their in vivo phenotype, primarily due to loss of cell polarity, cell-cell interaction, tissue architecture, and systemic influences.719 In recent years, considerable effort has been expended in developing human prostatic carcinoma xenografts. In early experiments, xenografts (Tables 3–6) were established by injecting established human prostatic cancer cells into nude mice. In recent attempts, fragments of primary cell cultures from prostate carcinomas obtained from biopsies, surgery, or intraperitoneal fluid were used (Tables 3–6). The success rate often was low,720 but it could be enhanced by coinjection with bone fibroblasts or matrigel.701, 721 The success rate in immunodeficient athymic (nude) mice that lacked T-cell-mediated immunity was slightly lower than in SCID mice722 that lacked both T-cell-mediated and B-cell-mediated immunity.723 Research involving xenograft models has focused in four areas: 1) characterizing the steps of cancer progression, including the emergence of androgen-independence (Table 3); 2) assessing the efficacy of anticancer therapies (Table 4); 3) identifying novel prostate cancer markers or phenotypes (Table 5); and 4) evaluating the metastatic potential of cell lines or carcinoma from patients (Table 6).724, 725

Table 3. Xenografts in Prostate Cancer Progression
Topic of investigationXenograft systemFindingsReference(s)
  1. VEGF: vascular endothelial growth factor; IGF-I: insulin-like growth factor I; IGF-IR: insulin-like growth factor I receptor; PTHrP: parathyroid hormone-related protein; PMA: phorbol myrstate acetate; SV40: simian virus 40; Tag: T antigen; AR: androgen receptor; HGPIN: high-grade prostatic intraepithelial neoplasia; TRAMP: transgenic adenocarcinoma of the mouse prostate.

Whole-body, low-dose radiation combined with VEGFPC-3 (nude mice)Higher tumor incidence (100%) and a more rapid tumor progression were found. This may involve leukocyte modulation. The regimen can be used to increase tumor incidence.Gridley et al., 19971210
Progression to androgen independenceAdvance human prostate cancer surgical specimens (SCID)Six of 8 patient samples grew as tumors in hosts. Establishment of the LACP series. One grew in an androgen-dependent manner, and another grew independent of the hormone. Some propagated > 6 passages; micrometastases were seen in 50% of the animals.Klein et al., 19971472
Evolution of androgen independence and chromosomal aberrationsLNCaP in castrated athymic miceGenetic aberrations identified by comparative genomic hybridization in androgen-dependent LNCaP xenografts were shared by 6 derivative sublines. Additional genomic alterations were noted in the derivatives.Hyytinen et al., 19971473
Response to androgen ablationLuCaP 23.1 (nude mice)Androgen ablation induced tumor regression that was followed by an androgen-independent tumor recurrence.Bladou et al., 19961474
Evolution of androgen-independent growthCWR22, first human prostate cancer xenograft established from primary cancer fragment grown in vivoTumor regressed bur recurred in < 5 weeks after androgen withdrawal.Nagabhushan et al., 19961475; Pretlow et al., 1993694
Influence of pertussis toxine on progression and metastasisPC-3 (nude mice)Pertussis toxine significantly reduced local tumor growth and metastasis to locoregional lymph nodes.Bex et al., 19991476
Interaction between prostate cancer cells and bone during growth and invasion at a distant sitePC-3, DU-145, and LNCaP in the tibia of congenitally athymic miceSeven of 9 PC-3-inoculated mice and 9 of 9 DU-145-inoculated mice developed tumors in the injected limb. The PC-3 tumors invaded the bone marrow cavity, cortical bone, and surrounding soft tissue. The DU-145 tumors were confined to the bone marrow cavity. Inoculation of LNCaP cells failed to produce tumors.Fisher et al., 20021477
Androgen-independence and osseous metastasisLNCaPThe LNCaP progression model shows many similarities with human prostate cancer. This model will help with understanding the mechanisms of androgen-independence and osseous metastasis.Thalmann et al., 20001478
Progression to androgen independence and expression of IGF-I and IGF-IRLNCaP, LAC-9, and LAPC-4Progression to androgen independence resulted in an increase in expression of IGF-I in LAPC-9 and LCNaP xenografts. IGF-IR levels also were increased in LAPC-9 and LNCaP models. These results suggest that deregulation of the expression of genes related to critical receptor tyrosine kinase regulatory systems may confer androgen independence.Nickerson et al., 20011479
Influence of linomide on local tumor growth and metastasisPC-3 in nude miceLinomide had no effect on net tumor growth and metastasis in this xenograft model.Bex et al., 19991480
Influence of PTHrP on prostate carcinoma growthMat-LyLu, LNCaPPTHrP overexpression did not affect proliferation of the Mat-LyLu cells in vitro. In vivo, primary tumor growth and tumor size were enhanced significantly. In the LNCaP model, expression of PTHrP was protective against PMA-induced apoptosis. The evidence that PTHrP is expressed in many cancer cell lines and that it can influence tumor growth in vivo and apoptotic rates in vitro supports the hypothesis that PTHrP plays an important role in prostate cancer progression.Doughterty et al., 19991481
Effect of neuroendocrine cells in carcinogenesis and progression of prostate cancerTransgenic mouse model with the rat prostate-specific large probasin promoter linked to the SV40 large TagThe neuroendocrine carcinomas occurred in the dorsolateral and ventral lobes and generally were AR-negative. Metastases occurred in regional lymph nodes, liver, and lung and showed histologic features of neuroendocrine differentiation. In the 12T-10 large probasin-promoter Tag, HGPIN developed progressively greater neuroendocrine differentiation and progressed to invasive carcinoma and neuroendocrine carcinoma with a high percentage of metastases. This model will provide for further testing of therapeutic interventions and further delineation of the role of neuroendocrine cells in prostate cancer progression.Masumori et al., 20011482
Relation between somatic mutation in the AR and the emergence of androgen-independent cancerTRAMPFifteen unique somatic mutations in the AR were identified in prostate tumors from 8 TRAMP mice. The mutations were single base substitutions, and the majority (78%) identified in the androgen-independent tumors were colocalized to the AR transactivation domain. The AR variants demonstrated promoter-specific, cell-specific, and cofactor-specific activities in response to various hormones, maintained strong sensitivity for androgens, and four of the variants demonstrated increased activities in the absence of ligand. This study indicates that somatic mutations in the AR gene occur spontaneously in TRAMP tumors and also that changes in the hormonal environment may drive the selection of spontaneous somatic mutations that provide a growth advantage.Han et al., 20011483
Conversion from a paracrine to an autocrine mechanism of androgen-stimulated growth in prostatic epithelial cellsHuman PC-82, LNCaP, LAPC-4, and rat R-33276The androgen-stimulated growth of prostate cancer cells occurred in both AR-null and AR wild type nude male mice. A direct autocrine mechanism is responsible for androgen-stimulated growth of malignant prostatic epithelial cells. Therefore, a change in the mechanism for androgen-stimulated growth does occur during the transformation from normal to malignant prostatic epithelial cells.Gao et al., 20011484
Effect of diet and exercise on prostate cancer growthLNCaPA low-fat, high-fiber diet and exercise intervention caused serum changes that significantly reduced the growth of androgen-responsive prostate cancer cells in vivo.Tymchuk et al., 20011485
Table 4. Xenografts and Antiprostate Cancer Therapy
Antitumor therapiesXenograft systemFindingsReference(s)
  • LHRH: luteinizing hormone-releasing hormone; ATP: adenosine triphosphate; uPA: urokinase plasminogen activator; SCID: severe combined immunodeficiency; PSA: prostate-specific antigen; VEGF: vascular endothelial growth factor; TRAMP: transgenic adenocarcinoma of the mouse prostate; IGF: insulin-like growth factor; RA: retinoic acid; 4-HPR: N-(4-hydroxyphenyl) retinamide.

  • a

    See Srkalovic et al., 1990.557

Cetrorelix (LHRH antagonist),a LHRH (agonist), bombesin antagonists RC-3940-II and RC3950-IIDU-145 (nude mice)Bombesin antagonists and Cetrorelix induced tumor volume reduction,a LHRH caused little tumor growth inhibitionJungwirth et al., 19971486
Ethionine combined with methionine starvationPC-3 (nude mice)Reduction in tumor ATP synthesis and induction of cell cycle arrest and apoptosisPoirson-Bichat et al., 19971487
uPA inhibitorsSCIDA mutated form of plasminogen activator inhibitor type 1 as well as p-aminobenzamidine and amiloride inhibited tumor growthJankun et al., 1997708
188-Relabeled somatostatin analogue (188Re-RC-160)DU-145, PC-3 (nude mice)Reduction in tumor mass and survival benefits for animals bearing PC-3 xenograftsZamora et al., 19961488
Androgen supplementationLNCaP 104-R2, an androgen-repressed cell line (castrated nude mice)Testosterone caused tumor regression and finasteride induced regrowth regrowth in xenograft tumors under androgen inhibitionUmekita et al., 1996728
Suramin, an antitrypanosomal agentC4-2 (nude mice)Suramin did not induce tumor growth in this hormone-refractory tumor but significantly reduced PSAThalmann et al., 1996726
Qinoline-3-carboxamide (linomide)TSU, PC-3 (SCID mice)Inhibition of tumor angiogenesis, induced tumor apoptosis and necrosisVukanovic and Issacs, 19951489
Tea epigallocatechin gallatePC-3 and LNCaP 104-R (nude mice)(−) Epigallocatechin-3-gallate, but not structurally related catechins, inhibited tumor growthLiao et al., 19951490
Bombesin receptor antagonist (RC-3095), LHRH agonist,a somatostatin analogue (RC-160)PC-82 (nude mice)Synergism between LHRH agonist and somatostatin analog, bombesin antagonist inhibits tumor growthMilovanovic et al., 19921491
NS398, cyclooxygenase-2 inhibitorPC-3 (nude mice)NS398 had no effect of proliferation, but it did induce apoptosis and decreased angiogenesis. VEGF expression also was down-regulated in NS398-treated tumors; cyclooxygenase-2 inhibitor suppressed PC-3 tumor cell growthLiu et al., 20001492
Phytosterol supplementationPC-3 (SCID mice)Phytosterols inhibited the growth and metastasis of PC-3 cells; β-sitosterol was more effective than campesterol in offering this protectionAwad et al., 20011493
Green tea polyphenolsTRAMP miceOral infusion of green tea polyphenols resulted in a lower incidence of prostate cancer by inducing apoptosis of various human carcinoma cells without affecting the normal cells; green tea polyphenols also resulted in an increase of tumor-free survival and a longer life expectancyGupta et al., 20011494
Black tea polyphenolsDu-145Black tea polyphenols inhibit the IGF-signal transduction pathway, which has been linked to increased prostate cancer incidence; this case provides further support for the potential of black tea polyphenols to prevent prostate cancerKlein et al., 20021495
Flavonoids (genistein, apigenin, luteolin and quercetin)LNCaPThe flavonoids affected the cell cycle progression of LNCaP cells in different ways; however, taken together, the results indicate that flavonoids are potent regulators of cyclin B and p21 for cell cycle progression that may play a role in the prevention of carcinogenesisKobayshi et al., 20021496
HypericinLNCaPCells treated with hypericin showed no signs of cytotoxicity; hypericin was distributed broadly in the tissues studied, including LNCaP tumor xenograft tissue; tumor tissue eliminated hypericin at a slower rate than any of the other tissues examined; and, in combination with photoirradiation, hypericin inhibits tumor growth and the elevation of PSAXie et al., 20011497
HypericinLNCaPCells treated with hypericin showed no signs of cytotoxicity; hypericinXie et al., 20011497
All-trans RA and 4-HPRRWPE-1, WPE1-NB14, WPE1-NB11RA and 4-HPR inhibited anchorage-dependent growth of all cell lines and anchorage-independent growth of WPE1-NB14 and WPE1-NB11 cells; however, 10 times more RA than 4-HPR was required to produce the same effect; overall, 4-HPR was more effective than RA in inhibiting growth and invasion of the cell linesQuader et al., 20011498
DexamethasoneDU-145, PC-3, LNCaPDexamethasone decreased glucocorticoid receptor levels and inhibited the growth of DU-145 and PC-3 cells, but not LNCaP cellsNishimura et al., 20011499
DibenzoylmethaneLNCaP, DU-145, PC-3Dibenzoylmethane inhibited the growth of the LNCaP, DU-145, and PC-3 cell lines in vitroJackson et al., 20021500
Table 5. Xenografts Systems for Identification and Characterization of Prostate Cancer-Specific Molecules and Phenotypes
Topic of investigation (phenotypes and molecules)Xenograft systemFindingsReference(s)
  1. PSCA: prostate stem cell antigen; SCID: severe combined immunodeficiency; GPI: glycosyl phosphatidyl inositol; PSA: prostate-specific antigen; TGF-α: transforming growth factor α; uPA: urokinase plasminogen activator; AR: androgen receptor; PAP: prostatic acid phosphatase; ACTH: corticotropic hormone.

PSCALAPC-4 (SCID)A modified subtractive hybridization; representational difference analysis was used to isolate a putative prostate stem cell marker (PSCA); the protein belongs to a family of GPI-anchored proteins; clinical studies indicate that 88% of cancer patients overexpress PSCA; the gene is localized to chromosome 8q24, a region of allelic gain in > 80% of prostate cancersReiter et al., 19981501
To identify PSA-specific small peptidesPC-82 (nude mice)HSSKLQ was identified showed high specificity for PSA; it may serve as a carrier to peptide-coupled prodrugDenmeade et al., 19971502
ErbB family ligand expressionA xenograftTGF-α expressed in cancer cells and Neu differentiation factor expressed in nontransformed epithelial cellsGrasso et al., 19971503
uPA inhibitorsSCIDIdentified an important binding site for small molecule inhibitors of uPAJankun et al., 1997708
AR mutationCWR22 (nude mice)A somatic mutation of AR at codon H874Y found in CWR22 exhibited sensitivity for dehydroepiandrosteroneTan et al., 19971504
Resistance to alkylator therapyLuCaP23 (nude mice)Glutathione content and glutathione S-transferase activity in an LuCaP23 xenograft were comparable to values found in prostate cancer tissue and primary cell cultures from patients with prostate cancer. In contrast, levels in established cell lines were markedly different from xenograft valueCanada et al., 1996719
Androgen-repressed phenotypeARCaP (nude mice)ARCaP, an androgen-repressed human prostate cancer cell line, was derived from ascites fluid of a patient with advanced metastatic disease; xenograft expressed high levels of epidermal growth factor receptor, c-erbB2/neu, c-erbB3, bombesin, serotonin, neuron-specific enolase, gelatinase A and B, stromelysin, and c-met; androgen repressed its growth and PSA secretionZhau et al., 1996729
Recurrence after androgen deprivationPrimary prostatic carcinoma fragments (nude mice)Seven permanent, transplantable human prostate cancer xenograft models were established, mostly diploid, expressed AR, secreted PAP and PSA, androgen-dependent and independent models, PC-346 regressed after androgen ablation but recurred subsequentlyvan Weerden et al., 19961505
Establishment of the first series of human prostate cancer xenograft lines, including an androgen-responsive modelCells from primary prostatic carcinomas (testosterone-supplemented nude mice)Cells from 6 of 20 primary prostatic carcinomas established xenografts, and 4 were transplantable serially (CWR21, CWR31, CWR91, and CWR22); CWR22 exhibited a clonal cytogenetic aberration, was responsive to androgen deprivation, and caused high PSA elevation in hostsPretlow et al., 1993694; Wainstein et al., 19941506
Hormonal dependencyALVA-31 (nude mice)ALVA-31 is a primary prostatic cancer cell line with androgen sensitivity; xenografts grew in intact male, castrate male, and female mice; a dependency on serum testosterone was demonstratedLoop et al., 1993727
Androgen dependencyLNCaP (nude mice)Xenografts secreted PSA and were androgen-responsive; tumor volume was correlated with PSA concentration in hosts; castration of host led to involution of tumor and stabilization of serum PSA levelLim et al., 1993721
Small cell carcinoma phenotype and originUCRU-PR-2 (nude mice)Xenografts shared common morphologic and ultrastructural features of small cell, undifferentiated carcinoma (including neuroendocrine characteristics) but also exhibited epithelial membrane antigen and carcinoembryonic antigen; xenografts were androgen and estrogen receptor negative, expressed neuron-specific enolase, secreted ACTH, β-endorphin, somatostatin, and PAP; phenotypes were stable on several passages in hostsJelbart et al., 1988730; Pittman et al., 1987731; van Haaften-Day et al., 19871507
Table 6. Xenograft Systems and Studies of Metastases of Prostate Cancer
Modulators of metastasisXenograft systemFindingsReference
  1. SCID: severe combined immunodeficiency; METs: metastases; SC: subcutaneous; RT-PCR: reverse transcriptase-polymerase chain reaction; PSA: prostate-specific antigen; MMP: matrix metalloproteinase.

Intraprostatic inoculation vs. subcutaneous injectionLNCaP (SCID vs. nude mice)Subcutaneous tumor take was higher in SCID mice vs. in nude mice; no lymph node or distant METs were found in SC tumors, whereas intraprostatic tumors yielded 100% lymph node METs and 40% pulmonary METsSato et al., 1997723
Progression of metastatic cancer to androgen-independencePrimary prostate cancer fragments (SCID)Six of 8 primary prostate cancer tissues from patients were established as xenografts in SCID; both androgen-dependent and androgen-independent cell lines released metastatic cells into host circulation (50% detection rate by RT-PCR of PSA-positive cells)Klein et al., 19971472
Orthotopic vs. ectopic microenvironment on tumor metastatic potentialsPC-3 (nude mice)Xenografts established with intraprostatic inoculation of PC-3 cells resulted in 100% paraaortic lymph node METs, whereas METs were present in only 22% of mice with SC tumors; subserosal implantation of cells into urinary bladder and stomach wall also yielded (100%) METs to regional lymph nodesWaters et al., 19951508
Role of prostate-derived growth factors on new bone formation in vivoCanine prostate tissue implanted in nude miceThe prostate tissue induced abundant, new, woven bone formation; the new bone stained intensely with calcein, which demonstrated mineralization of the bone matrixLeRoy et al., 20021509
Role of MMP activity in prostate cancer METs to bonePC-3MMP inhibition reduced the number of osteoclasts per millimeter in PC-3-injected implants and reduced proliferating tumor cells without affecting angiogenesis or; apoptosis; MMP inhibition had no toxic effect on PC-3 cells; MMP activity plays an important role in bone matrix turnover when prostate cancer cells are present; both bone matrix turnover and metastatic tumor growth are involved in a mutually supportive cycle that is disrupted by MMP inhibitionNemeth et al., 2002849

Established xenograft models include the androgen-responsive LNCaP cell line721 and its androgen-insensitive derivative LNCaP-C4-2,726 a primary prostate tumor cell line ALVA-31,727 and a number of androgen-insensitive lines, such as PC-3, PC-133, DU-145, and LuCaP 23.1 (Tables 2–5). In addition, 2 models with androgen-repressive phenotypes have been produced using LNCaP-104R2 cells728 and ARCaP cells.729 Xenograft models also have been established from a small cell carcinoma cell line, UCRU-PR-2.730, 731 Usually, cancer is monitored by tumor size and the levels of serum PSA and PAP. The androgen dependency of tumor growth is studied in castrated mice. In many cases, the xenografts retain the cytogenetic, biologic, and molecular features of the original cancer even after multiple passages. Mouse-derived endothelial cells have been found recently between the intravascular space and human tumor cells,732 suggesting contribution of murine tissues to the vasculature of the tumor.

A recent study collected cells from the peripheral blood of patients with metastatic prostate cancer and injected them subcutaneously into nude mice. Prostate cancer from 2 of 11 patients was found growing as metastases in the lungs of the mice. This approach may have important applications for the development of models of human cancer and the sampling of cancer from specific patients for novel molecular and therapeutic approaches.733 One study attempted to develop methods for the generation of short-term primary tissue xenografts from benign and tumor-derived human prostate tissue. Primary human prostate xenografts were established in athymic mice. These xenografts maintained tissue architecture and expression of key prostatic markers. This model system allows investigators to conduct multiple (successive) analyses of human prostate tissue over time using a tissue sample derived from a single patient.734

4.3 Existing Cell Culture Models

Advances in cell culture technologies have enabled researchers to establish numerous cell lines from human and rodent prostatic neoplasms (Table 2). The 3 most widely used prostate carcinoma cell lines, DU-145, PC-3, and LNCaP, were established between 1977 and 1980 from metastatic prostate cancers of the brain, bone, and lymph nodes, respectively. DU-145 and PC-3 cells express little or no AR and do not rely on androgens for growth. LNCaP cells exhibit androgen dependency, secrete PSA, and express an AR with a point mutation. Since the development of those cell lines, many other cell lines have been established from metastatic and primary site tumors, including the TSU-Pr1, JCA-1, ND-1, ALVA-31, and ALVA-41 cell lines. In addition, several sublines have been derived from earlier cell lines (e.g., C4-2 from LNCaP). These derivatives (PC-3 ML, PC-3 MR, and PC-3 MC) exhibit unique properties while retaining the parental phenotype; therefore, they offer novel opportunities for comparative studies.

Apart from cancer cell lines, noncancerous prostatic cell lines derived from the immortalization of normal or hyperplastic prostatic cells with HPVs or SV40 are of special interest. Several of these immortalized epithelial cell lines (RWPE-1, BRF-55T, BPH-1, and PNT1) that exhibit phenotypes of normal epithelial cells, including growth responses to androgen and growth factors; expression of AR, PSA, or PAP secretory activities; and anchorage dependency. Step-wise transformations to malignancy in these cell lines have been achieved by genetic-engineered expression of oncogene products. These findings have extended our understanding of the mechanisms of multistep carcinogenesis.

The epithelial or stromal nature of immortalized cells usually is confirmed by immunostaining for cell type-specific markers. The expression of epithelial cell-specific cytokeratins and the lack of expression of desmin, vimentin, or factor VIII have been recognized as phenotypes of epithelial cells; whereas positive immunostaining for vimentin, fibronectin and α-actin and negative staining for cytokeratins are accepted as stromal cell markers. Immortalized, noncancerous cell lines usually retain “parental” phenotypes and can be recognized easily as epithelial or stromal. However, cancer cell lines often express both epithelial and stromal markers, creating confusion with regard to their tissue of origin.

Epithelial cancer cell lines derived from rodents produce prostatic adenocarcinoma readily in isogenic hosts at subcutaneous sites (AXC, Mat-LyLu, and Dunning R-3327 cell sublines). In contrast, human cell lines may require matrigel, fibroblasts, or a prostatic environment for in vivo tumor formation in nude/SCID mice. Nevertheless, several human lines (PC-3 ML, PC-3 MR, PC-3 MC, and C4-2) form tumors in nude mice and exhibit metastatic potential. Therefore, they may be used as model systems for antitumor or antimetastatic drug development.

4.4 Data Gaps: Animal and Cell Culture Models for Prediction of Human Risk

The first data gap in this area is our lack of knowledge of the causes of the majority of human prostate cancers. It is important to identify the putative factors that regulate prostatic growth, development, and carcinogenesis, so that these factors then can be evaluated in animal model systems. Human prostate cancers likely are composed of a number of classes, including 1) those that arise spontaneously as a function of age, likely due to spontaneous mutations in protooncogenes and in tumor suppressor genes; 2) those induced by testosterone promotion of prostate cells bearing spontaneous mutations; 3) those induced by food-derived mutagens, such as PhiP; 4) those induced by as yet unknown, environmentally derived carcinogens, possibly endocrine disrupters; and 5) those induced by lower concentrations of PhiP, as well as other similar food-derived or environmentally derived carcinogens, plus promotion by testosterone.

The second data gap is a lack of knowledge regarding the genes that are targets for spontaneous or chemically induced mutation and the types of mutations that occur in these specific genes in human prostate cancers. Specifically, which protooncogenes are activated to oncogenes, what are the specific activating mutations that occur, and at which codons do they occur? Similarly, it is important to determine the role of specific known and novel tumor suppressor genes. In most animal models of prostate cancer, it is not known which oncogenes are activated and which tumor suppressor genes are inactivated. Before conducting investigations to address such a data gap, investigators should decide which are the best spontaneous and chemically induced models of prostate cancer from the animal systems studied to date in terms of their ability to mimic human prostate cancer most closely. Then, experiments should be conducted applying those factors that cause human prostate cancer to the animal model systems of prostate cancer. The most appropriate animal systems should be coupled with appropriate exposure conditions relevant to humans to yield the animal prostate tumors that are most relevant to human prostate cancer induction. Then, these tumors can be analyzed to fill another data gap, namely, to determine which oncogenes have become activated and which tumor suppressor genes have been inactivated in animal prostate tumors that arose as a result of treatment of animals under conditions mimicking human prostate cancer induction.

A third data gap is our lack of knowledge of how aberrant animal tumors would be if they were induced under conditions of exposure that mimic human prostate cancer induction. Recent studies indicate that as many as 300 genes may be overexpressed and underexpressed in human gastrointestinal tumors. Such studies also need to be conducted on human prostate tumors as soon as possible. Methods like RNA differential display should be applied, with both human prostate tumors and with animal prostate tumors induced under conditions that induce human prostate tumors. This should yield the identification of known and novel genes that are overexpressed or underexpressed in prostate tumors. Such knowledge may lead to the development of novel therapies for the treatment of prostate cancer. In this way, relevant animal models of human prostate cancer can lead to insight into the genesis of human prostate cancer and may lead to clinical trials directed toward therapy for patients with prostate cancer.

A further data gap is our lack of information on the phenotypic and molecular steps that occur in animals and the order in which these steps occur when animals are exposed to conditions that result in the induction of human prostate cancer. To fill this data gap, investigators first must identify conditions that lead to the majority of human prostate cancers, whether they are induced spontaneously, by excess testosterone, by specific food mutagens (such as PhiP), or by combinations of food mutagens (such as PhiP, other food-derived mutagens, or unidentified environmental mutagens or carcinogens) with or without testosterone. Such knowledge would have strong mechanistic and therapeutic importance. Investigators need to identify specific genes that are both altered and important with great frequency in the majority of human prostate cancer. Then, transgenic mice need to be constructed using these genes targeted to expression in the prostate tissues of transgenic mice. These mice could then be exposed to conditions that generate the majority of human prostate cancer.

Another data gap is our lack of knowledge concerning whether the genesis of animal prostate tumors mimics the genesis of human prostate cancers and whether animal prostate tumors metastasize like human prostate tumors and to the same sites. Some work already has been done in these areas with the retinoblastoma (Rb) and p53 genes. More work needs to be done with other oncogenes, tumor suppressor genes, and with combinations of these genes when they have been identified as altered with great frequency in human prostate tumors.

The MPR system is an elegant model in which much important work has been done to understand the biology of prostate cancer. This model can be used to address another data gap—our lack of knowledge concerning the behavior of mouse urogenital sinus tissue that has been transfected with activated oncogenes and inactivated tumor suppressor genes. This transfected tissue can be reconstituted with urogenital mesenchymal tissue, then treated either with testosterone or with high fat in vitro, and reconstituted and implanted orthotopically into mice; or it can be reconstituted and implanted into mice, and the mice can be treated with high-fat diets and/or testosterone. Does this process result in prostate tumors with the same metastatic capability of human prostate tumors, and would they also metastasize to bone, for instance? Can dissected urogenital epithelium treated with PhiP, then reconstituted with urogenital mesenchymal tissue, and implanted orthotopically cause prostate tumors when implanted into adult mice? Would these tumors metastasize to bone and sites similar to the sites where human prostate tumors metastasize?

A large body of work also has been conducted utilizing the transplantation of human prostate tumors or human prostate tumor cell lines into SCID mice. This system also can be used to address a number of remaining data gaps. First, the data gap concerning whether human primary prostate tumors would be influenced further may be addressed by feeding the animals a high-fat diet or by treating them with testosterone. Do these regimens cause the tumors to evolve faster and to metastasize more efficiently or more rapidly? Is there a further effect of food-derived mutagens, such as PhiP, on the evolution of the phenotype of these tumors and on their ability to metastasize, and does this metastasis mimic the metastatic spread of human tumors? Are there environmental carcinogens, such as endocrine disruptors, that affect the progression and metastasis of these human tumors in SCID mice?

Another interesting data gap is our lack of knowledge of the behavior of benign human prostate tumors in SCID mice that are treated with testosterone, high-fat diets, and PhiP either alone or in combination. A further data gap is our lack of knowledge of whether implanting human prostate tumors ectotopically, rather than orthotopically, and treating the animals with high-fat diets, testosterone, PhiP, or endocrine disrupters causes the orthotopically implanted tumors to behave like human prostate cancer and whether the ectotopically implanted tumors behave the same or differently than the orthotopically implanted prostate tumors.

Finally, the use of a single cell line or animal model system will have limited utility for extrapolating to the human disease. Specifically, a single model system can be used only to model a specific biologic pathway that leads to prostate cancer. Parallel studies in multiple systems will provide a better understanding of the variability observed in human populations. The sensitivity of these models may make them particularly useful for investigating low-susceptibility and high-susceptibility factors and/or low-penetrance and high-penetrance genes. The development of in vitro human epithelial prostate cell culture systems that can be induced reproducibly and conveniently to undergo neoplastic transformation upon treatment with carcinogens, such as PhiP, and promoters, such as testosterone, also would help answer basic questions regarding the process of human prostate cell transformation and carcinogenesis. Clearly, it is important to focus this research along conditions of exposure in which humans develop prostate cancer to make it relevant for understanding the induction of human prostate cancer.

How should animal models be used to answer relevant questions regarding environmental factors and prostate cancer risk? How do laboratory exposures to specific agents mimic exposures that humans encounter in terms of dose, duration, and age at exposure? Of particular importance, a logical approach is needed to integrate data collected from human epidemiologic studies, animal and cellular models, and human interventional studies to generate a meaningful assessment of prostate cancer risk and the potential benefits of interventions.

What dietary factors confer increased risk or protection against prostate cancer development? Particular attention must be focused on conditions of specific nutrient deficiency, adequacy, or supranutritional states. The concept of threshold effects of specific nutrients on prostate cancer risk deserves further study. There undoubtedly are complex gene-diet interactions that will need to be understood before strong nutritional guidelines can be made to reduce the risk of prostate cancer.

5.0 BIOMARKERS OF DISEASE, EXPOSURE, AND EFFECT

5.1 Morphometric Markers

Morphometric markers provide useful predictive information in prostate cancer but still are considered investigational. Morphometric studies should employ objective, quantitative techniques that preferably are computer-assisted. The College of American Pathologists735, 736 and a conjoint Mayo Clinic-National Cancer Institute-World Health Organization Consensus Panel recognized that there are no accepted standards for morphometric studies, and this is considered an important and significant area for investigation.

Quantitative digital-image analysis is beset by a variety of potential problems, including 1) reproducibility and high intraobserver and interobserver variability; 2) the high cost of equipment and personnel; 3) user-dependence and subjectivity in selection of cases, cells to analyze, etc.; 4) lack of standardized methods; 5) discrepant results with prostate cancer; and 6) unproven clinical utility. Variables that should be controlled to minimize variance in digital image analysis include 1) complete, nonoverlapping, and well focused nuclei, preferably > 150 per case; 2) consistent tissue fixation, processing, cutting, and staining; 3) internal age-matched and procedure-matched controls; 4) rigorous definitions for classifying results; and 5) sufficiently large population groups to perform valid statistical analyses. These variables rarely are controlled in published studies of morphometry in prostate cancer; therefore, reported findings should be interpreted with caution.737

The most popular morphometric markers are nuclear size,738 nuclear volume,739 nuclear shape, nuclear roundness, chromatin texture, size and number of nucleoli, and number of apoptotic bodies. Prostates that harbor preneoplastic and neoplastic lesions show morphologic nuclear abnormalities that are not seen by the human eyes but that can be detected with image analysis (malignancy-associated changes).740–742

5.1.1 Nuclear roundness

Roundness refers to the degree to which a nucleus approximates a circle, and deviations from roundness indicate more aggressive cancer. Many of the initial favorable reports that predicted a correlation between survival and nuclear roundness were from a single institution, were limited by a small sample size (< 30 patients), required an expenditure of 4 hours per patient, used the same patient cohort in multiple publications, failed to describe the morphologic variations and nuclear roundness extremes, and were beset with patient selection bias.743

Significant problems of reproducibility in nuclear roundness measurement have been described, and the results with different digitizing instruments are not comparable. A comparative study of digital image analysis and video planimetry with manual nuclear tracing found a lower level of intraobserver and interobserver variability with planimetry.744 Blom et al. were unable to reproduce the positive results from an earlier study from another institution despite their use of the same equipment and examination of the same patients.745 Those authors found that intraobserver variation was low, but interobserver variation was high. There was no correlation between nuclear roundness and patient survival. Nuclear roundness measurements were not reliable in needle biopsy specimens.746 Furthermore, multivariate studies that used image analysis to correlated nuclear morphometry and nuclear grade with pathologic stage at the time of radical prostatectomy demonstrated little additional predictive value for nuclear features compared with Gleason grade.747–749 Virtually all measures of nuclear abnormality by computer-based image analysis reveal the similarity of high-grade PIN and cancer compared with normal and hyperplastic epithelium.750, 751 The mean nuclear volume showed a wide range of values in prostate cancer, varying from 81 μm3 to 782 μm3,752 and it may752 or may not753 correlate with cancer grade and clinical stage. Nuclear morphometry predicts patient outcome after radiation therapy.754

5.1.1.1 Nuclear chromatin texture

Alterations in chromatin texture reflect quantitative or qualitative changes in DNA content and structure.755 Digital image analysis provides a variety of measures of chromatin texture, including chromatin intensity, heterogeneity, condensation, margination, run-length nonuniformity, long-run emphasis, gray-level nonuniformity, and inertia. Each of these measures, as well as nuclear area and roundness, was successful in separating prostate cancer from BPH.756, 757 Chromatin texture was superior to nuclear volume and shape in discriminating hyperplasia and cancer in a study that used confocal laser-scanning microscopy.758 Four chromatin textural features discriminated androgen-responsive and unresponsive cancer in patients with metastases.759, 760 Greater chromatin pattern heterogeneity was correlated with shorter recurrence-free survival in patients who underwent surgery.761

5.1.1.2 Nucleolar organizer regions

Nucleolar organizer regions (NORs) are loops of DNA that transcribe to ribosomal RNA. A simple histochemical argyrophilic method stains the proteins associated with NORs in routine tissue sections, and the number of NORs can be counted manually or digitally by light microscopy. The number of argyrophilic NORs (AgNORs) correlates with prostate cancer grade762–765 and stage776; furthermore, that number is predictive of patient survival in most,764, 767 but not all), studies.766 In a single multicenter study, there were 2–3 AgNORs per nucleus in well differentiated cancer and 3 AgNORs per nucleus in higher grade cancer.762 The range of NOR numbers per nucleus was 1.6–2.5 in BPH, 1.7–5.6 in atypical adenomatous hyperplasia, 3.3–7.4 in high-grade PIN, 2.6–7.9 in low-grade carcinoma, and 3.1–11.4 in high-grade carcinoma.764 Standard procedures for staining and counting NORs have been accepted by the European Committee on AgNOR Quantification.

5.1.1.3 Nucleolar size

Nucleolar size is a significant diagnostic factor in prostate cancer and is a major determinant of nuclear grade.768 “Prominent” nucleoli in cancer have been defined variously as a greatest tumor dimension > 1.0–3.0 μm.769 The presence of two or more nucleoli in an individual cell usually indicates carcinoma.764, 770 In routine practice, nucleolar size is not measured objectively, and differences in tissue preparation and staining limit the comparability of this factor in morphometric studies.

5.2 DNA Ploidy Analysis

A good correlation exists between DNA ploidy and both nuclear grade771, 772 and histologic grade, and DNA ploidy adds clinically useful predictive information for some patients.773–777 Aberrant DNA ploidy indicates genomic instability and is one of the most widely used markers, regardless of whether it is determined by flow cytometry or image analysis. One of the hallmarks of cancer is genetic instability, which seems to generate new mutations by deletion and rearrangement, and this instability appears to produce the excess DNA ploidy of cancer cells.

The mean proliferative index and the proportion of aneuploid cells in high-grade PIN are similar to those in cancer, and they are much greater compared with BPH and low-grade PIN.778 The incidence of aneuploidy in high-grade PIN varies from 32% to 68%, and it is somewhat lower than the incidence in carcinoma, which shows aneuploidy in 55–62% of tumors.779, 780 There is a high level of concordance between DNA content of PIN and cancer. Approximately 70% of aneuploid PIN is associated with aneuploid carcinoma; conversely, only 29% of aneuploid cancer is associated with aneuploid PIN.781

DNA ploidy analysis of prostate cancer provides important predictive information that supplements histopathologic examination. Patients who have diploid tumors have a more favorable outcome compared with patients who have aneuploid tumors. Among patients who have lymph node metastases treated with radical prostatectomy and androgen-deprivation therapy, men with diploid tumors may survive ≥ 20 years, whereas men with aneuploid tumors die within 5 years.782 However, the ploidy pattern of prostate cancer often is heterogeneous, which creates potential problems with sampling error. Most studies use matched benign or hyperplastic prostatic tissue as controls. Seminal vesicle tissue is unsuitable as a control because of the low but significant level of aneuploidy in the epithelial cells.783

Flow cytometry is the most common method of DNA ploidy analysis, but it is limited by the need for a large number of cells. The minimum amount of needle-core tissue necessary to yield satisfactory results with flow cytometry is a 0.2-cm length of cancer sample, corresponding to approximately 2500–5000 nuclei.784 Digital image analysis overcomes this limitation and is gaining popularity despite a lack of standards.785 Cytogenetic methods, such as fluorescence in situ hybridization (FISH), can assess DNA ploidy for individual chromosomes. FISH requires less tissue, but it is labor-intensive, expensive, and not widely available.

5.2.1 Flow cytometry

DNA ploidy pattern by flow cytometry correlates with cancer grade,773 volume, and stage.786 Most low-stage tumors are diploid, and high-stage tumors are nondiploid, but numerous exceptions occur.787 The 5-year cancer-specific survival rate is approximately 95% for diploid tumors, 70% for tetraploid tumors, and 25% for aneuploid tumors.776, 788 Patients who had diploid lymph node metastases treated by androgen-deprivation therapy alone had longer progression-free survival and overall survival compared with patients who had aneuploid metastases.789 An International DNA Cytometry Consensus Conference reviewed the literature in 1993 and concluded that the clinical significance and biologic basis of DNA ploidy needed further investigation.775 DNA by flow cytometry is not an independent predictor of patient outcome after radical prostatectomy790 or after androgen-deprivation therapy,791 although this has been refuted in some reports.792–795

5.2.2 Digital image analysis

Digital image analysis appears to have a high level of concordance (approximately 85%) with radical prostatectomy specimens that are evaluated by flow cytometry.784 Digital image analysis usually is performed in routinely processed tissue sections stained by the Feulgen method. The Feulgen stain stoichiometrically binds to hydrolyzed nucleic acids, coloring nuclei blue according to the amount of DNA present. The concentration of DNA is proportional directly to the absorption at 620 nanometers. DNA ploidy analysis of fine-needle aspirates is useful prognostically, particularly when combined with the proliferative fraction796 or with bcl-2 expression.771 Multiple reports have demonstrated the utility of DNA ploidy in predicting outcome after radical prostatectomy.797 However, the value of DNA ploidy in predicting cancer recurrence has been refuted in patients who underwent radical prostatectomy70, 798 and androgen-deprivation therapy.791 In addition, DNA ploidy in needle biopsies was unable to predict prostatectomy cancer volume.76

5.2.3 Cytogenetics

FISH analysis of interphase cells with centromere-specific and region-specific probes is useful for the detection of numeric chromosomal anomalies in solid tumors, including prostatic carcinoma, which often is difficult for conventional cytogenetic analysis. When it is applied to histologic sections, this method allows the study of multiple foci of normal epithelium, PIN, and carcinoma within a single specimen, and it makes the evaluation of matched metastatic sites possible. FISH was superior to immunohistochemistry for detecting HER-2/neu amplification in prostate cancer (6–53% vs. 0–2%, respectively),799–801 and expression of HER-2/neu increases with cancer progression to androgen independence.802

Genetic alterations are present in 9% of cases of atypical adenomatous hyperplasia according to a FISH ploidy study that used centromere-specific probes for chromosomes 7, 8, 10, 12, and Y.803 The overall frequency of numeric chromosomal anomalies in PIN and carcinoma foci is remarkably similar (50% and 51%, respectively), suggesting that they have a similar pathogenesis.804, 805 Overall, the mean number of abnormal chromosomes increased in PIN to carcinoma foci, and malignant foci contained more anomalies than paired PIN foci. These findings suggest that PIN is a precursor of carcinoma.780, 804, 806, 807

Gain of chromosome 8 is the most frequent numeric anomaly in PIN and prostatic carcinoma,804 suggesting that alterations of this chromosome and/or a tumor suppressor gene(s) on the short arm may be important for the initiation or early progression of prostate cancer. FISH studies with centromere-specific probes for chromosomes 7, 8, 11, and 12 showed that gains of chromosomes 7 and 8 were markers of tumor aggressiveness and prognosis.806, 808 There was c-myc gene amplification in 22% of metastatic foci, much more frequent compared with primary cancer (9%), suggesting that the 8q arm harbors a gene(s), the amplification and overexpression of which is involved in the progression and evolution of prostatic carcinoma.809–811 Loss of 8p22 was associated with a poor prognosis after radical prostatectomy.812, 813 A decline in the number of cancer cells with gains of chromosomes 7, 8, and 12 after combined external beam radiation therapy and brachytherapy may predict clinical outcome.814 The number of FISH-aneuploid cells declined with androgen-deprivation therapy.800

5.3 Differentiation Markers

Normal cell organization, which characterizes many differentiated secretory cells, includes tissue-specific secretions, cell polarity, cell-cell adhesion, and cell-basement membrane adhesion. The progressive loss of many of these features is characteristic of cancer progression.815 At the ultrastructural level, cytoplasmic organelles associated with the secretory pathway, the endoplasmic reticulum, the Golgi complex, and secretion granules are visible markers of androgen-dependent secretory function in prostatic epithelium.816 Such features are valuable indictors of differentiation, but not all cancers lose these features, and biochemical markers often are more appropriate indicators. Altered gene expression that results from cellular transformation also leads to changes in cell-cell interactions and abnormal growth control. Consequently, cell organization often is altered, giving rise to a diverse range of histologic abnormalities, including PIN and cancer.

5.3.1 Cytoplasmic differentiation markers, including enzymes and other secretory proteins

5.3.1.1 Prostatic acid phosphatase

Human prostatic acid phosphatase (PAP) is a neutral protein-tyrosine phosphatase that is involved in growth regulation. The immunohistochemical localization and distribution of PAP in normal, hyperplastic, and cancerous human prostate has been used as a prostate-specific marker for many years.817–821 In the normal and hyperplastic prostate, PAP was present uniformly at the apical portion of the glandular epithelium of secretory cells. There was more intense and uniform staining of tumor cells in the glandular epithelium of well differentiated adenocarcinoma, whereas less intense and more variable staining was seen in moderately and poorly differentiated adenocarcinomas. In recent years, however, examination of PAP expression has been replaced by examination of other tissue markers, including PSA and other human glandular kallikreins.822–825 Serum PAP remains useful for the detection of bone metastases, despite concerns regarding the stability of enzyme activity.826

5.3.1.2 Prostate-specific antigen and human kallikreins

Prostate-specific antigen (PSA) is a 34-kilodalton (kD), single-chain glycoprotein of 237 amino acids that is produced almost exclusively by prostatic epithelial cells. PSA is a serine protease, is a member of the kallikrein gene family, and has a high sequence homology with human glandular kallikrein 2. It has chymotrypsin-like, trypsin-like, and esterase-like activity. In the serum, PSA is present mainly as a complex with α 1-antichymotrypsin. It is secreted in the seminal plasma and is responsible for gel dissolution in freshly ejaculated semen by proteolysis of the major gel-forming proteins, semenogelin I and II, and fibronectin. A small amount of PSA in semen is complexed. The free, noncomplexed form of PSA constitutes a minor fraction of the serum PSA. Production of PSA appears to be under the control of circulating androgens acting through the ARs. Serum levels of PSA may be elevated by conditions other than cancer, including prostatitis, PIN, acute urinary retention, and renal failure. The value of PSA as a screening tool has been questioned by some investigators because of the overlap of PSA concentration between BPH and prostate cancer. PSA is particularly sensitive and accurate in the detection of residual cancer, recurrent cancer, and cancer progression after treatment, irrespective of the treatment modality. PSA accurately predicts cancer status and can detect recurrence several months before detection by any other method. PSA is also a sensitive and specific immunohistochemical marker for tumors of prostatic origin.820, 827–829

The human kallikrein family consists of 15 members, including hK2 and hK3 (PSA).830 The mRNA for hK2 and PSA is located predominantly in prostatic epithelium and is regulated by androgens.831 In addition, hK2 has 78% amino-acid homology with PSA and is expressed predominantly in the prostate, suggesting that it may be a clinically useful marker for the diagnosis832 and monitoring of prostate cancer,833 although some reports indicate that its value is marginal.834 The intensity and extent of hK2 expression was greater in cancer compared with those in PIN, which, in turn, were greater compared with those in benign epithelium. Gleason primary Grade 4 and 5 cancer showed hK2 staining in almost every cell, whereas there was greater heterogeneity of staining in lower grades of cancer. In marked contrast to hK2, PSA and PAP immunoreactivity was most intense in benign epithelium and stained to a lesser extent in PIN and carcinoma. The number of immunoreactive cells for hK2 and PSA was not predictive of cancer recurrence. Tissue expression of hK2 appears to be regulated independently of PSA and PAP.

5.3.1.3 Extracellular matrix proteases

Localized degradation of the extracellular matrix is required for tissue remodeling, cell migration through the basal lamina, and metastasis. This degradation is accomplished through the action of proteases. The proteolytic enzymes comprise two main classes, including the metalloproteinases (such as collagenases and matrilysin) and serine proteases (such as the kallikreins and uPA).

uPA converts plasminogen into plasmin and, thus, mediates pericellular proteolysis during cell migration and tissue remodeling. uPA is secreted as an enzymatically inactive proenzyme by cancer and stromal cells. Active uPA converts plasminogen to plasmin, which, in turn, degrades components of the cancer stroma, such as fibrin, fibronectin, proteoglycans, and laminin; it also may activate procollagenase Type IV, which degrades collagen Type IV, a major part of the basement membrane. Primary cancer and metastases contain elevated concentrations of uPA compared with benign tissues.835–840 The urokinase-derived peptide A6 reduced lymph node metastases and prolonged survival of prostate cancer-bearing nude mice.841 Antibodies directed against uPA receptor decreased cancer volume by the induction of apoptosis and was able to detect the presence of microscopic metastases in a mouse model.842 Elevated serum uPA and uPA receptor levels were predictive of decreased survival in patients with prostate cancer.843 The expression of Type IV collagenase was minimal in benign tissue but consistently was strong in PIN and in cancer of all Gleason grades.844

A proteolytic cascade may occur during initiation and invasion of cancer cells.838, 845, 846 The actions of proteases, such as matrix metalloproteinases (MMPs), often are confined to specific areas by secreted protease inhibitors, such as the TIMPs. MMP activity was greater in cancer than in BPH. Pro-MMP-9, in its 92-kD form, was expressed exclusively by cancers, particularly those with aggressive and metastatic phenotypes.847 Many isoforms of these proteases and their inhibitors are present in the prostate.848 MMP activity was correlated with bone matrix turnover in the presence of prostate cancer cells.849 The expression of TIMP-1, TIMP-2, MMP-2,850 and MMP-9851 appear to be independent predictors of outcome for patients with prostate cancer. The inhibition of MMP-2, MMP-9, and membrane type 1-MMP prolonged survival in a rat model of prostate cancer.852

Matrilysin (MAT or MMP-7), another member of the MMP family, is involved in tissue remodeling. It is expressed in epithelial cells of BPH and cancer, in contrast with the majority of MMPs, which are produced by the stroma.853, 854 Hepatocyte growth factor (HGF)/scatter factor induced matrilysin disruption of the E-cadherin/β-catenin complex, resulting greater cancer aggressiveness.855

5.3.1.4 5-α-Reductase

Various androgen-metabolizing enzymes are present in the prostate, including the dominant testosterone, 5-α-reductase.856 The genes encoding the 2 5 α-reductase isozymes, designated type 1 and type 2, have been identified.857–859 The secretory epithelium of normal and hyperplastic glands showed strong nuclear 5-α-reductase 1 reactivity. The androgen-independent basal cell layer variably expressed type 1 and 2 isoenzymes in nuclear and cytoplasmic compartments. Increased 5-α-reductase reactivity was detected in prostate cancer, particularly in high-grade and androgen-insensitive cancer.860–862 Polymorphisms of the 5-α-reductase gene (SRD5A2) predict a poor prognosis.863

5.3.1.5 Telomerase

The enzyme telomerase adds a hexanucleotide sequence to the end of each chromosome during cell division. Senescence may be caused by a failure to maintain the length of the telomeres as a result of a deficiency in telomerase. Telomerase activity was present in up to 93% of prostate cancers, unlike benign tissue,864 and appeared to be independent of grade, stage, or DNA ploidy.865–869 Higher grade tumors have maximally activated telomerase and may be most responsive to antitelomerase therapy.869, 870

5.3.1.6 Antioxidant enzymes

Prostate cancer expresses lower levels of antioxidant enzymes compared with benign tissue. These enzymes include catalase, manganese-containing superoxide dismutase, and copper-zinc-containing superoxide dismutase.871, 872 Similar results have been observed in PIN.872, 873 Oxidative stress, which arises as a consequence of an imbalance in cellular antioxidant status, has been implicated in aging. Other markers associated with loss of differentiation and invasive potential in prostate cancer include elevated levels of 12-lipoxygenase activity874–876 and hemeoxygenase isozymes, members of the heat-shock family of proteins.877–879

5.3.1.7 Activin and inhibin

Activin, which is a member of the TGF-β family, has growth-inhibitory effects on select human prostate tumor cell lines.880 These growth and differentiation factors and their binding proteins, follistatins, are synthesized in high-grade cancer. The mRNA and protein for the activin β A subunit, the activin β B subunit, and follistatin are expressed in poorly differentiated cancer. In benign tissue, activin β A and β B subunit mRNA and proteins are located predominantly in the epithelium, whereas follistatin mRNA is expressed in basal cells and in stromal fibroblasts.881 In progression to cancer, resistance to the growth-inhibitory effects of activin may be conferred by follistatins. A related growth factor, inhibin, also is expressed in the prostate.882 In prostate cancer, the inhibin α subunit gene is down-regulated, and this is associated with loss of heterozygosity at the gene locus and methylation of the promoter.880

5.3.1.8 Relaxin

Relaxin is a peptide hormone with A and B chains derived by posttranslational cleavage from a single, 185-amino-acid preprorelaxin. Relaxin is synthesized by the human prostate, and there are marked tissue differences in the relative amounts of expression of the H1 and H2 relaxin mRNA forms in different tissues. The role of relaxin in human male reproductive physiology is uncertain.883, 884

5.3.2 Cytoskeletal proteins

The structural framework of the cell, the cytoskeleton, consists of three distinct structures: microtubules, microfilaments, and intermediate filaments. The cytoskeleton plays an active role in coordinating many cellular functions, including gene expression. A large complex of adhesion molecules, receptors, and associated intracellular proteins is involved intimately in cell membrane-extracellular matrix interactions. Signals from the extracellular matrix may exert as much control over the behavior of cells as hormones and other soluble mediators.885 Alterations of the extracellular matrix-cytoskeleton-nuclear matrix axis may alter gene regulation, cell function, and growth control.886 The microfilament system changes dramatically with cancer; focal attachments of actin bundles in the cell membrane disappear and alter binding to the stroma, allowing these cells to migrate and invade.

In the prostatic epithelium, three main cell types are distinguished according to location, morphology, degree of differentiation, and cell-specific markers: luminal secretory cells, basal cells, and endocrine-paracrine cells.887, 888 The luminal and basal cells have characteristic patterns of cytokeratin expression.889, 890 The phenotypic plasticity of basal cells suggests that they house the stem cell population that gives rise to all epithelial cells.888

According to the stem cell model, there are at least three other cell types, including stem cells, amplifying cells, and transit cells. Anitkeratin immunoreactivity reveals at least three subpopulations of cells, one putatively representing amplifying cells.887 The candidate stem cell population may be concentrated in the proximal portion of prostatic ducts891 and appears to be absent in prostatic carcinoma. Amplifying cells are defined in the stem cell model as precursors of transit (luminal) cells in the hierarchical pathway of prostatic epithelium differentiation, so the keratin expression profile led to the concept that this subpopulation mat be the target of neoplastic transformation.892

Prostate stem cell antigen (PSCA), a glycosylphosphatidyinositol-anchored cell-surface protein, also appears to be a marker for amplifying cells.893 PSCA expression is greater in cancer than in the benign epithelium, and it correlates with grade and stage. Monoclonal antibodies directed against PSCA have antitumor activity in xenograft models and preclinical models.894, 895

There is differential expression of cytokeratin polypeptides in normal, hyperplastic, and malignant prostatic epithelial cells.896 Assays of fragments of cytokeratins 8 and 18 identify patients with high-grade and metastatic cancer.897

5.3.3 Cell adhesion markers

There are a multitude of cell-adhesion molecules (CAMs), including the cadherins, the catenins, and the CD family of glycoproteins. The cadherins are a large family of phenotypical markers responsible for calcium-dependent cell-cell adhesion.898 The three main cadherins are E-cadherins (associated with many types of epithelial cells), N-cadherins (nerve and muscle cells), and P-cadherins (cells in the placenta and epidermis). All are found transiently in other tissues during development. In the absence of calcium, the cadherins undergo conformational changes and are degraded rapidly by proteolytic enzymes. E-cadherins are concentrated in the adhesion belts (zonula adherens) in mature epithelial cells, where they connect to the actin cytoskeleton through intracellular catenins.899, 900

Dysfunction of the cadherin pathway is involved in cancer invasion and progression. In a single report, high (normal) E-cadherin expression was seen in 87% of benign tissue, 80% of high-grade PIN, 82% of prostate carcinoma, and 90% of hormone-refractory human prostate carcinoma samples.901 E-cadherin expression in prostate cancer inversely correlates with grade, stage, metastasis, recurrence, and survival.902–906 The severity of methylation of E-cadherin correlates with cancer progression.907 Matrilysin mediates the cleavage of E-cadherin under the control of HGF/scatter factor.855 Polymorphisms of the E-cadherin gene alter prostate cancer risk.908, 909 N-cadherin expression is up-regulated in cancer and is not present in normal prostate tissue),910 and the N cadherein-catenin complex is linked by a signal-transduction cascade to the antiapoptotic protein bcl-2.911 P-cadherin shows decreased or absent expression in prostatic cancer, most likely reflecting loss of the basal cell layer rather than transcriptional down-regulation.912 P-cadherin expression was limited to the basal cells, and P-cadherin-immunoreactive cells were negative for PSA. Prostatic cancer usually was negative for P-cadherin, but some tumors had focal positive areas, which frequently were located close to ejaculatory ducts that were negative for PSA. The mutually exclusive expression of P-cadherin and PSA suggests that these proteins are involved in differential mechanisms of cell regulation.913

Catenins, particularly α-catenin, also play an important role in the dysfunction of the cell adhesion complex.914 Catenins are down-regulated with prostate cancer, similar to E-cadherin.914 Dysregulation of β-catenin, either by direct mutation or by defects in interacting pathways/regulators, may result in accumulation within cells. In the nucleus, β-catenin forms a transcriptional complex capable of up-regulating target genes, many of which encode proliferative factors; presumptive activating mutations were observed in 5% of prostate cancers.915 The down-regulation of α-catenin and β-catenin has been correlated with cancer grade.916

Another family of transmembrane glycoproteins, the cluster of differentiation (CD) proteins, display extracellular matrix adhesion properties. One of these proteins, CD44, and its isoforms may be involved in malignant progression of prostate cancer,903, 917, 918 and both quantitation of CD44v6 and quantitation of CD44s are univariate predictors of disease recurrence after radical prostatectomy.919, 920 The methylation status of CD44 has been correlated with CD44 expression and cancer progression.921, 922 Epithelial cell populations were separated by flow cytometry using antibodies to differentially expressed CD44 (basal cells) and CD57 (secretory cells).923 PSA expression by CD57-positive cells was abolished after prostate tissue was dispersed by collagenase into single cells. Expression of PSA was restored when CD57-positive cells were reconstituted with stromal cells. Both cell types expressed a novel prostate marker, CD38.924 Complete loss of CD38 expression was found in BPH and cancer.

The CAM CEACAM1 acts as a tumor suppressor in prostate cancer, and this may result in part from its suppression of angiogenesis.925 Decreased expression of CEACAM1 was correlated inversely with cancer grade and proliferation rate.926

5.3.4 Enzymes and other secretory proteins

Numerous other enzymes and secretory proteins that have not been described elsewhere in this report have been identified in human and animal prostatic tissues and in cell cultures, including transglutaminase,927 glucuronosyltransferase,928 drug-metabolizing enzymes,929, 930 lipogenic enzymes,931 and ODC.932 Transglutaminase was more accurate that the TUNEL assay for measuring apoptosis after androgen-deprivation therapy,933 and its expression was lower in prostate cancer compared with benign epithelium.934 ODC catalyzes the rate-limiting step in polyamine biosynthesis, and green tea contains inhibitors of ODC that have been implicated in cancer prevention and therapy.935, 936 ODC activity and protein expression are greater in prostate cancer than in matched benign tissue.937

5.3.5 Stromal proteins

The potential role of gap junction-mediated intercellular communications and stromal-epithelial interactions in the onset and progression of human prostate cancer remains poorly defined. Laminin is a stromal protein, the localization of which is intense and uniform in the basement membranes of acini, blood vessels, smooth muscle, and nerve fibers in normal prostate, BPH, and well differentiated carcinoma. Laminin 5 is an extracellular matrix protein integral to the formation of the hemidesmosomes that attach normal basal cells to the underlying basal lamina. These hemidesmosomal complexes are lost in carcinoma, possibly allowing malignant cells to detach from the anchoring structures, invade, and migrate through the adjacent tissue. Failure of hemidesmosome formation results in a less stable epithelial-stromal junction, which may allow malignant cells to invade and spread.938 The basement membrane of poorly differentiated carcinoma showed an absence of laminin reactivity, although laminin activity was retained in the cytoplasm, on the surface, and in secretory material.939, 940 Altered translation in PIN and cancer of β3 or γ2 laminin 5 mRNAs into functional proteins contributes to the failure of anchoring filaments and hemidesmosomal formation, most likely destabilizing the epithelial-stromal junction and increasing invasive potential and migration of cancer cells as well as disrupting normal integrin-signaling pathways.941, 942 The 67-kD laminin receptor was not a significant predictor of outcome after prostatectomy.943

Cell adhesion and migration are important features of tumor invasion and are mediated in part by integrins. The integrins are a large family of homologous linker proteins that bind to a variety of extracellular matrix components. These receptor proteins bind to and respond to the extracellular matrix. They are composed of two noncovalently associated transmembrane glycoprotein subunits called A and β, both of which contribute to the binding to matrix proteins. Integrin binding is dependent on divalent cations. Many extracellular matrix components, such as laminin and fibronectin, are recognized and bind to multiple forms of integrins heterodimers. Using parallel immunohistochemical analysis, there was reciprocal expression of E-cadherin and β1 integrin in high-grade prostate cancer.944 Integrin expression significantly decreased in human prostate cancer, although an exception is α6 integrin (laminin receptor), which persists during prostate tumor progression.945 The progression of prostate carcinoma may be influenced by the biochemical nature of the basal lamina surrounding the primary carcinoma cells.946 The normal basal cells form focal adhesions and hemidesmosomal-like structures. The normal basal cells exhibit a polarized distribution of hemidesmosomal-associated proteins, including BP180, BP230, HD1, plectin, laminin-γ 2, collagen Type VII, and the corresponding integrin-laminin receptors α6 β1 and α6 β4. The expression and distribution pattern of these proteins are retained in PIN. In contrast, carcinoma often lacked hemidesmosomal structures, the integrin α6 β4, BP180, laminin-γ-2, and Type collagen VII but expressed BP230, plectin, HD1, and the integrin-laminin receptors α3 β1 and α6 β1.947 These results suggest that, although a detectable basal lamina is present in carcinoma, its composition and cellular attachments are abnormal. The production of α6 β1 and laminin in cancer may contribute to the invasive phenotype.948, 949 In part, α(v) β(3) facilitates prostate cancer metastasis to bone by mediating prostate cancer cell adhesion to and migration on osteopontin and vitronectin, which are common proteins in the bone microenvironment.950 The differential expression of integrins may be useful clinically. The α4 subunit of integrins is expressed only on nontumorigenic cells,951 whereas prostate cancer expresses αIIb β3 integrin.952 The 1C integrin is expressed in benign prostate but is undetectable in cancer, consistent with its in vitro growth-inhibitory properties.953 Selective cleavage of the β4 integrin by matrilysin may explain its loss in cancer.954

Prostate cancer also contains a cell surface receptor for entactin, a glycoprotein found in basement membranes in complex with laminin. The heterodimeric receptor was identified as the integrin α3 β1.955 Purified entactin promotes attachment and spreading of cells.

Stromal elements in the primary and metastatic target organs are important mediators of tumor cell intravasation, chemoattraction, adhesion to target organ microvascular endothelium, extravasation, and growth at the metastatic site. Collagen is an important component of the extracellular matrix. More than 19 types of collagen have been identified and are encoded by 33 genes. Collagen Types I, II, III, V, and XI constitute the fibrillar collagens, whereas Types IV, VI–X, and XII–XIX represent the structurally diverse, nonfibrillar members. In addition to their role in basement membrane-stromal interactions, the pronounced vascular association with collagen Type XIX suggests involvement of this and related collagen types with angiogenesis and cancer.956 Type IV collagen, a major component of basement membranes, is organized in a network responsible for mechanical resistance. It also plays a key role in epithelial cell adhesion to basement membranes. The distribution of Type IV collagen α-chains in benign prostate, PIN, and malignant basement membranes revealed specific loss of the α5 (IV) and α6 (IV) chains in cancer, whereas the classic α1 (IV) and α2 (IV) chains were retained consistently. In addition, Type VII collagen colocalized with α5 (IV) collagen chains.957 The increase of collagen fibers accompanied the decrease of smooth muscle cells as cancer grade increased. With increasing Gleason grade of the cancer, collagen content at the focus decreased, whereas that of surrounding unaffected tissue from the same prostate increased to levels significantly above the levels from controls with no cancer. Markers of collagen synthesis in prostate biopsy material were increased significantly in the presence of prostate cancer.958 The characteristics of stromal components and their amounts in the normal prostate appear to correlate with a distinct predilection for cancer occurrence in the peripheral zone and with a weak stromal reaction in prostate cancers.959 Reactive stroma appears to be initiated during PIN and evolves with cancer progression to effectively displace the normal fibromuscular stroma.960 Other components of the extracellular matrix that have been examined in prostatic tissue include fibronectin, which demonstrated higher levels in cancer patients961; perlecan, a multidomain heparan sulfate proteoglycan962; and proteoglycans, macromolecules that contain bound glycosaminoglycans.

Tenascin is a hexameric glycoprotein component in the extracellular matrix of stromal tissue. High expression of tenascin was found in embryonic development and during carcinogenesis of almost all organs and in PIN963 and prostate cancer.904 Tenascin is distributed preferentially in the peripheral zone during postnatal development.964

5.3.6 Neuroendocrine markers

Neuroendocrine cells are ubiquitous but uncommon in benign and neoplastic prostate epithelium, and they are considered important for regulating cell growth and differentiation. These cells are part of the widely dispersed, diffuse neuroendocrine-regulatory system, also known as endocrine-paracrine cells. In the human prostate, subpopulations of neuroendocrine cells have been identified based on morphology and secretory products.965 Most neuroendocrine cells of the prostate contain serotonin,966 chromogranin A,967 and other neuroendocrine markers that are not expressed consistently. Based on their function in other organs, neuroendocrine cells in the prostate most likely are involved in the regulation of growth, differentiation, and secretory function. The absence of chromogranin A-immunoreactive neuroendocrine cells in the rat, guinea pig, cat, and dog prostate challenges the validity of these animal models for physiologic studies of neuroendocrine cells in the human prostate.968 In humans, prostatic carcinoma exhibits neuroendocrine differentiation as 1) infrequent small cell neuroendocrine carcinoma, 2) rare carcinoid-like cancer, and 3) conventional prostatic cancer with focal neuroendocrine differentiation.969

The significance of neuroendocrine differentiation in prostate cancer is uncertain, but the bulk of evidence argues against its predictive value.970 Proponents for a role of neuroendocrine differentiation in prostate cancer have shown that the neuroendocrine cell population enlarges with cancer progression: It appears to correlate with Gleason sum and cancer stage, it may be a prognostic factor for progression,971, 972 and elevated serum levels of chromogranin A may detect prostate cancer in patients without elevated levels of PSA.973 Conversely, the detractors for a role of neuroendocrine differentiation in prostate cancer have shown that the neuroendocrine cell population does not exhibit a relation with cell proliferation; is not correlated with pathologic stage or Gleason sum974 or with other measures of progression,974 such as increasing serum PSA levels;975 and is not correlated with cancer-specific survival,970, 976 and, thus, has no independent prognostic significance for prostate cancer.974, 976, 977 This controversy was summarized recently with a review of the literature.970 Neuroendocrine cells lack ARs,969 so there is no alteration in the extent of neuroendocrine cell differentiation after hormonal manipulation,969 although this finding has been refuted.978

5.4 Proliferation and Apoptotic Markers

5.4.1 Proliferating cell nuclear antigen

Epithelial cells within the normal prostate and in organ-confined cancer typically exhibit a low rate of proliferation. Proliferating cell nuclear antigen (PCNA) is an auxiliary protein for DNA polymerase-α, which reaches maximal expression during the S-phase of the cell cycle. Hence, PCNA has been used widely as an index of the proliferative activity of cancers. The PCNA labeling index is lowest in benign epithelium and increases progressively from well differentiated to poorly differentiated prostate cancer,979 but there is wide variance.980 Many studies have shown that the PCNA index is related to cancer stage,980–982 but that finding also has been refuted.983 Hence, a high PCNA labeling index may indicate progression of prostate cancer984, 985 and may be an independent prognostic indicator.986 Nonetheless, this factor is not recommended currently by the College of American Pathologists and the American Joint Committee on Cancer for routine clinical use.735

5.4.2 Ki-67 and MIB-1

The Ki-67 antigen is a nuclear nonhistone protein that is expressed by non-G0 proliferating cells.987 Because Ki-67 antibodies exhibit diminished immunoreactivity in fixed, paraffin embedded tissues, the MIB-1 antibody was developed against recombinant parts of the Ki-67 antigen for use on embedded tissues.987 Antibodies to either Ki-67 or MIB-1 are used interchangeably, depending on the form of tissue preparation. The Ki-67 and MIB-1 labeling indices correlate with cancer.988, 989 However, a high proliferation index appears to add little predictive information for patient outcome above the traditional indicators of Gleason score, pathologic stage, and DNA ploidy.983, 990 The Ki-67 labeling index may discriminate between organ-confined and metastatic cancer.982 Hence, increases in the proliferation indices of Ki-67 or MIB-1 appear to reflect progression.991, 992 This is reflected further in an association between expression of Ki-67 and epidermal growth factor receptor,993 mutant p53,994, 995 chromosomal aberrations,996 and perineural invasion.997 Taken together, these findings suggest that Ki-67 expression may be a weak prognostic indicator of disease recurrence,988, 998–1001 disease progression,70, 991, 992, 1002 and survival.995, 998, 1003, 1004 Biopsy and radical prostatectomy concordance for MIB-1 in cancer is low.918

5.4.3 Mitotic figures

Mitotic figures are found rarely in tissue sections in normal or hyperplastic prostatic epithelium. Therefore, S-phase markers are used commonly as surrogate markers for estimating proliferation rates. Cells in the cell cycle incorporate bromodeoxyuridine (BRdU), a thymidine analogue, into newly synthesized DNA1005 and express PCNA1006 and Ki-67 antigen.1007 One study of BRdU, PCNA, and Ki-67 labeling in prostatic tissues demonstrated that the three methods are correlated strongly with each other.982

In most studies, the number of mitotic figures progressively increased from benign epithelium to PIN and cancer.982, 1008–1010 Adenocarcinomas with cribriform growth patterns and adenocarcinomas comprised of solid areas of undifferentiated tumor cells had the most mitotic figures.1011 In addition, the number of mitotic figures was correlated with cancer stage and grade.982, 1009, 1010, 1012, 1013 In one study, these parameters were correlated with disease progression and with progression-free survival.1010 Androgen-deprivation therapy1009 results in a dramatic decline in the number of mitotic figures in prostate cancer,1009 whereas the normal prostatic epithelium undergoes apoptosis-mediated involution.

5.4.4 Apoptosis and apoptotic bodies

Apoptotic cell death is the final fate for terminally differentiated cells and is an effective, intrinsic anticancer mechanism through which a tissue discards transformed cells. The apoptotic process produces strand breaks in cellular DNA that can be labeled with TdT. Thus, suppression of apoptosis, specifically, the development of resistance to inductive signals, may favor accumulation of initiated cells and contribute to the establishment of early neoplastic lesions. Furthermore, as cancer progresses, it often exhibits additional phenotypes of apoptotic deregulation. An imbalance in cell proliferation and death rates contributes to clonal expansion and augments the net cancer growth rate. The acquisition of independence of tissue-specific survival factors and/or apoptotic regulators allows metastatic cancer cells to survive and grow at sites distinct from their tissue of origin. Androgen-dependent and androgen-independent prostatic epithelial cells undergo apoptosis after androgen ablation.57, 1014, 1015

Apoptotic bodies are present throughout the normal prostatic epithelium and in acinar lumens in all cases. They usually are present in intercellular spaces and occasionally are present within the cytoplasm of epithelial cells, with the latter observed more often in PIN and carcinoma than in benign epithelium.1016 There is a progressive increase in the number of apoptotic bodies from BPH through PIN to adenocarcinoma, and the greatest frequency is invariably in the basal cell layer (or, in the case of carcinoma, in cells at the periphery of the malignant glands adjacent to the stroma); these trends virtually are identical to those seen with PCNA immunoreactivity.1016 The percentage of apoptotic bodies was progressively higher in low-grade PIN, high-grade PIN, and adenocarcinoma compared with the percentage in BPH (0.68%, 0.75%, and 0.92–2.10% vs. 0.26%, respectively). There also was a positive correlation between the number of apoptotic bodies and increasing Gleason score.1017, 1018 There was no apparent correlation between the numbers of mitotic figures and apoptotic bodies. Larger, cathepsin D-positive granules are markers of apoptotic bodies.1019

Derangement of apoptosis appears to play a critical role in the pathogenesis of prostate cancer.1014, 1020 There is down-regulation of apoptotic rates and up-regulation of proliferative indices in localized prostatic cancer compared with adjacent normal tissues.1021 In localized cancer, the expression of bcl-2, which is an intracellular antagonist of apoptosis, was elevated markedly, suggesting that inhibition of apoptosis may be an early event in transformation. Conversely, metastatic prostate cancer exhibited significantly higher bcl-2 expression, with a concomitant, marked elevation in the proliferative index compared with bcl-2 expression in primary cancer. Cell proliferation rates were higher than apoptotic rates, resulting in a net gain in cell number in localized and metastatic carcinoma. The apoptotic rate in prostatic carcinoma provided more prognostic information than the proliferation rate. There was a significant elevation in the apoptotic count in high-grade cancers and in cancers with aggressive progression profiles.1022–1025 Cancer with perineural invasion had a lower apoptotic rate than that with invasion (median, 4.4 vs. 7.2),1026 perhaps accounting for the frequent tumor spread through perineural spaces.1027 The molecular mechanisms underlying the lower apoptotic rate in carcinoma with perineural invasion did not appear to be due to overexpression of bcl-21028; rather, it was due to an unidentified nerve-derived factor(s), because the apoptotic rate was correlated negatively with the diameter of the nerve. This finding suggests that the microenvironment influences apoptotic activity and, hence, the growth potential of carcinoma, which, in turn, controls tumor progression and metastasis. Androgen-deprivation therapy increased the apoptotic rate in prostate cancer.1009, 1024, 1029

5.4.5 Apoptosis-suppressing oncoprotein bcl-2

Overexpression of several apoptosis-suppressing oncoproteins results in defective apoptotic signaling and is common in prostate cancer. The inhibition of apoptosis or the disruption of cell death signaling promotes carcinogenesis and may play a role in cancer initiation.1030

Early studies of the bcl-2 or “B-cell non-Hodgkin lymphoma-2” gene revealed an intricate relation between cell death disruption and carcinogenesis.1031, 1032 This genetic alteration causes overexpression of bcl-2 and concomitant down-regulation of cell death. Consequently, the initiated cells enjoy a selective survival advantage and achieve clonal expansion, resulting in cancer formation. Since the discovery of bcl-2, other mammalian homologues have been identified, including Bax, bcl-X, Mcl-1, A1, Bad, and Bak,1033 and additional forms arise through alternative splicing.1034 These exist as intracellular proteins, which are found predominantly in the outer mitochondrial membrane, the nuclear envelope, and the endoplasmic reticulum.1035 The formation of homodimers and heterodimers largely depends on the concentrations and ratios of the various proteins. These dimers function as endogenous inducers or suppressors of apoptosis, and their relative ratios determine cellular susceptibility to exogenous regulators.

It is believed widely that bcl-2 is an apoptosis suppressor gene. Overexpression of the protein in cancer cells may block or delay the onset of apoptosis, selecting and maintaining long-living cells, and arresting cells in the G0-phase of the cell cycle.1036, 1037 Forced expression of bcl-2 impairs cell death induced by tumor necrosis factor (TNF)-mediated cytotoxicity, IL-1-converting enzyme, peroxides and ROS, growth factor withdrawal, ultraviolet radiation, p53, c-myc, calcium, doxorubicin, hormone and growth factor withdrawal, and anti-CD3 receptor clustering.1036, 1037 In addition, bcl-2 protects against a wide variety of cell death signals and does so by controlling signaling and preventing ROS formation.1038, 1039 It may play a critical role in regulating the ceramide-signaling and the arachidonate/HETE-signaling pathways, two opposing bioactive lipid second-messenger systems involved in apoptotic cell death.1040 It also mat serve as the guardian of microtubule integrity and, thus, genomic stability by phosphorylation.

The expression of bcl-2 normally is restricted to the basal cell layer of the normal and hyperplastic prostatic epithelium.1041 However, overexpression of bcl-2 is present is PIN1021, 1041 and in prostatitis.1042 In cancer, the prevalence and expression pattern of bcl-2 is controversial. One study found moderate, heterogeneous bcl-2 overexpression in localized cancer1041 that was correlated inversely with Gleason grade. Another report described a significant elevation of bcl-2 in 45% of cases of primary cancer that was heterogeneous but did not correlate with grade. It is noteworthy that the area of cancer with high bcl-2 expression was devoid of apoptotic cells. Metastatic cancer in lymph nodes was negative. In another study, it was found that > 70% of prostate carcinomas were bcl-2 negative, 18% had weak expression, and 11% exhibited strong expression.1043 Expression of bcl-2 was correlated with high stage, metastases, and high grade. Androgen-deprivation therapy decreased bcl-2 expression in cancer, suggesting that these cells develop resistance to apoptotic signals.57, 1020, 1024, 1041 In hormone-refractory prostate cancer, heterogeneous staining was observed in most specimens, with expression retained in clusters of cancer cells. In bone marrow metastases after androgen deprivation, approximately 30% of cancers expressed bcl-2, but expression did not correlate with survival.1044 In summary, bcl-2 expression varies greatly in localized cancer, limiting its utility in tissue microarrays,1045 but it is expressed consistently in cancer cells after androgen deprivation and in tumors that are hormone-refractory. These data support the concept that bcl-2 is linked causally to apoptotic resistance in prostatic cancer cells.

bcl-2 expression also correlates with outcome for some (but not all) patient groups. Patients with bcl-2-negative cancer who were followed expectantly had favorable overall survival,1046 similar to patients who were treated by brachytherapy.1047 Conversely, bcl-2 expression did not predict outcome in patients who underwent radical prostatectomy.1048

Direct evidence in support of bcl-2 as a prostatic apoptotic suppressor is derived mainly from cell transfection studies. Overexpression of bcl-2 in the androgen-dependent cell line LNCaP protects these cells from androgen deprivation-induced apoptosis.1049 This approach also protects androgen-dependent and androgen-independent cells against radiation-induced apoptosis.1050 The suppression of bcl-2 expression in LNCaP variants induces apoptosis through a sequence-specific ribozyme.1051 Antisense bcl-2 oligodeoxynucleotides enhanced DES-induced cytotoxicity in hormone-refractory prostate cancer cells through the apoptotic pathway independent of augmented ROS generation, whereas glutathione depletion augmented cytotoxicity and ROS generation.1052

Several other apoptosis-suppressing gene products have been implicated in the pathogenesis of prostate cancer. The mitogen-activated protein kinase phosphatase-1 (MKP-1), a putative apoptotic inhibitor, is overexpressed in PIN but is down-regulated in cancer.1053 TGF-β1 is overexpressed in cancer cells compared with benign epithelium.1021 Animal studies revealed opposite actions for TGF-β1; it may be a potential mediator of apoptosis or an enhancer of tumorigenicity and metastasis.570 Thus, TGF-β1 may or may not affect apoptotic signaling during cancer progression.

5.5 Growth Factors/Receptors and Related Oncogenes

The progression of prostate cancer is accompanied by modifications in the expression of growth factors and their receptors (Table 7). A characteristic of the transformation process is that the cancer cells exhibit multiple and concurrent modifications in the expression of growth factors, receptors, oncogenes, and tumor suppressor genes. Hence, many of the modifications that occur in cancer cells may be interdependent and/or may be caused by one or more common processes of transformation. Multiple mechanisms of AR activation, some of which involve growth factor receptor signaling, have been demonstrated in prostate cancer models, and it is likely that a number of autocrine and paracrine growth factor ligand-receptor interactions, such as those of EGFs, FGFs, and IGFs, contribute to the androgen-independent phenotype by promoting cell proliferation and survival.1054

5.5.1 Epidermal growth factor family of peptides and receptors

Table 7. Growth Factors, Receptors, and Binding Proteins as Potential Risk Factors in Prostate Cancer
  1. EGF: epidermal growth factor; TGF: transforming growth factor; FGF: fibroblast growth factor; NGF: nerve growth factor; BDNF: brain-derived neurotrophic factor; NT: neurotrophin; PDGF: platelet-derived growth factor; VEGF: vascular endothelial growth factor; BMP: bone morphogenic protein; EGFR: epidermal growth factor receptor; NRT: neurotrophin receptor; TGFβ-RI and TGFβ-RII: TGFβ receptors I and II.

Acquisition/up-regulation of autocrine growth factors
 Proliferation (EGF, TGFα, FGF-2, FGF-8, NGF, BDNF, NT-4, PDGF-A)
 Immunosuppression (TGFβ-1, TGFβ-2)
 Angiogenesis (FGF-2, VEGF)
 Metastasis (EGF, FGF-2, BMP-6)
Up-regulation of growth factor binding proteins
 Follistatin
Up-regulation of growth factor receptor protooncogene products
 EGFR, p185erB-2, p160erB-3, PDGF, c-met
Down-regulation of tumor suppressor protooncogene products
 p75NTR, TGFβ-RI, TGFβ-RII

Normal prostatic epithelial cells exhibit autocrine dependence on epithelial cell-derived growth factors and paracrine dependence on stromal cell-derived growth factors. Malignant cells have an enhanced capacity for autocrine growth factor expression that, in some cases, circumvents dependence on stromal cell-derived growth factors. This concept appears to apply to epidermal growth factor (EGF) and a related family member, TGF-α, both of which play fundamental roles in the growth and metastasis of cancer.1055 EGF is secreted by both stromal cells and epithelial cells.1056, 1057 Moreover, EGF and TGF-α, both signaling through the same EGFR, vary in expression during malignant transformation.1058 Evidence that EGF signaling through the EGFR is important in prostate cancer cell proliferation includes the following findings: 1) Prostate cancer cells produce EGF,1056, 1057 2) prostate cancer cells express EGFR in vitro1059 and in vivo,1060 3) the addition of EGF to cultures of prostate cancer cells stimulates growth,1061 4) the addition of anti-EGFR antibody inhibits the proliferation of prostate cancer cells,1062 5) inhibition of the EGFR block signal-transduction pathway is implicated in prostate cancer growth,1063 and 6) levels of EGFR are higher in prostate cancer than in benign prostate.1064 Hence, the autocrine expression of EGF and TGF-α1056, 1057 signaling through EGFR1065 may contribute to the autonomous growth of prostate cancer.1058, 1065 EGF stimulates VEGF gene expression, which enhances angiogenesis and vascular permeability.1066

A second major function of EGF is the stimulation of invasiveness of prostate cancer.1067–1069 This has been demonstrated in Boyden chamber assays, in which 1) EGF promotes the chemomigration of prostate cancer cells,1067, 1068 2) antagonism of EGFR function prevents EGF-stimulated chemomigratory activity of cancer cells,1069 and 3) clones of transfected cells expressing elevated levels of EGFR invade across membranes to a greater extent than parental cells.1070 Taken together, these observations suggest that EGF and TGF-α contribute to the autonomous growth of cancer cells and that EGF has a pleiotrophic effect on prostate cancer cells promoting both growth and invasiveness. AR expression confers a less aggressive phenotype by interfering with the interaction and signaling of the α6 β4-EGFR.1071

Members of the EGFR family of related oncogenes HER-2/neu, HER-3, and HER-4 also are expressed differentially in the prostate. The HER-2/neu gene product p185erb B-2 is not expressed in normal secretory epithelial cells, and it may be expressed in basal cells1072 but not in luminal cells of BPH tissues.1072, 1073 However, p185erb B-2 is expressed in the majority of epithelial cells in PIN and cancer cells.1072, 1073 Moreover, HER-2/neu gene amplification and, to a lesser extent, p185erb B-2 protein expression have been correlated with increasing cancer grade1074, 1075 and androgen independence.1076 Experimental overexpression of p185erb B-2 in normal epithelial cells produces a phenotype with an increased rate of proliferation and enhanced capacity for metastasis.1077 A similar pattern of expression has been observed for the HER-3 gene product p160erbB-3 in BPH, PIN, and cancer.1072 Conversely, the HER-4 receptor protein is expressed strongly in normal epithelial cells but not in prostate cancer. The heregulin ligand, also called neu differentiation factor, which binds the HER-3 and HER-4 gene products preferentially, is expressed in all stromal cells and basal cells and in ≈ 50% of luminal epithelial cells in normal and BPH tissue but largely is absent in cancer. Hence, heregulin may be a paracrine factor that stimulates growth in vitro. It appears that enhanced expression of p185erb B-2 and p160erb B-3 occurs with disease progression and that these oncogenes infer a phenotype with an increased capacity for proliferation and metastasis similar to that exhibited by EGFR and its ligands EGF and TGF-α. EGFR expression has been correlated with cancer progression after radical prostatectomy and hormone-refractory disease.1078

5.5.2 The TGF family of peptides and receptors

The TGF family is comprised of 1) TGF-β isoforms and more distantly related bone-morphogenic proteins (BMPs); 2) the activins and inhibins, all of which are expressed differentially in the adult prostate; and 3) the developmentally expressed mullerian-inhibitory substance. In vitro, mammalian TGF-α isoforms 1–3 inhibit the proliferation of normal epithelial cells1079 and cancer cells whereas in vivo, TGF-β1 enhances cancer growth and metastasis.1080 This paradoxical role of TGF-β in the regulation of cancer growth results from modified expression of TGF-β receptors and the response of the host to TGF-β. Normal prostatic epithelial cells express TGF-β1–TGF-β3 to differing degrees,1081 and TGF-β1–TGF-β3 are overexpressed in cancer.1081, 1082 Consequently, both urinary TGF-1 levels and plasma TGF-β2 levels are elevated in cancer.1083

Androgens negatively regulate the expression of TGF-β and TGF-β receptors, including TGF-β1, c-Fos, and Egr-1 in the LNCaP cell line. Ligand-bound ARs inhibit TGF-β through selective repression of binding of Smad3 (but not Smad2 or Smad4) to the Smad-binding element.1084, 1085 The effect of antiandrogens on TGF-β-induced apoptosis correlates with increased expression of the cell cycle regulator, p21, and the apoptotic executioner, procaspase-1, with a parallel down-regulation of the antiapoptotic protein, bcl-2.1086

TGF-β binding to the TGF-β type II receptor (RII) initiates recruitment of the type I receptor (RI), forming a heterotetrameric complex that undergoes transphosphorylation (from RII to RI). The phosphorylated RI binds adaptor molecules that participate in the signal-transduction cascade. This signaling pathway appears to be down-regulated in prostate cancer. In this context, the TGF-β RI and RII proteins, which are expressed abundantly in normal prostate epithelial cells, exhibit progressive reduction of expression in primary cancer and lymph nodes.960, 1087 Hence, even though cancer cells exhibit an up-regulation of TGF-β1 and TGF-β2 expression,1081 the down-regulation of TGF-β RI and RII expression1087 appears to offset the autocrine growth-inhibitory effect of the TGF-βs. Cancer cells that exhibit up-regulation of TGF-βs and down-regulation of their receptors appear to exhibit host effects that facilitate growth. Caspase-1 activation is required for TGF-β-induced apoptosis and is absent from most cancers.172 Cancer cells exhibit an immunosuppressive effect on lymphocyte action1088 and promote angiogenesis, extracellular matrix deposition, and metastases.1089, 1090

Preoperative plasma TGF-β1 concentration is a strong multivariate predictor of PSA failure after radical prostatectomy (together with Gleason score and surgical margin status), presumably because of an association with occult metastatic disease present at the time of radical prostatectomy.1091 Prostate-derived factor is a member of the TGF-β superfamily proteins involved in differentiation of the prostate epithelium. It is believed that proprotein convertases, such as furin, mediate the processing of TGF-β.1092

The TGF-β-binding protein, endoglin, typically is expressed by newly formed endothelial cells and regulates cell adhesion, motility, and invasion in the prostate.1093 Endoglin expression by immunohistochemistry correlates with cancer grade, stage, metastases, proliferation index, and disease-specific survival.1094

Members of the family of BMPs induce bone morphogenesis in vivo and are implicated in skeletal metastases of advanced prostate cancer.1095 BMP-2, BMP-3, and BMP-4 are present in normal epithelial cells and cancer in vitro.1096 However, in vivo, BMP-6 expression is higher in organ-confined cancer than in adjacent normal, benign epithelial cells.1097 BMP-6 expression is correlated with Gleason score1097 and with pathologic disease stage,1098, 1099 and it is implicated as a mediator in osteoblastic metastases.1100 The BMP receptors (BMPR) BMPR-IA, BMPR-IB, and BMPR-II are present on prostatic epithelial cells.1101, 1102 BMPR-IB, but not BMPR-IA or BMPR-II, is up-regulated by androgens in the LNCaP cell line.1101, 1102 Moreover, BMPR-IA stimulates growth, whereas BMPR-IB inhibits growth in response to BMP-2.1102 Immunohistochemical expression of bone sialoprotein, BMP6, and thymidine phosphorylase predicted outcome in patients who underwent radical prostatectomy.1103

The activin βA and activin βB subunits are expressed in normal epithelial cells and in cancer,881 whereas the inhibin α subunit appears to be absent.1104 Because the inhibin α subunit may function as a tumor suppressor, its absence may be involved with the development of prostate cancer, because activin cannot be opposed by inhibins.880, 1105 However, activins are expressed in prostate cancer,881 and they inhibit growth1106, 1107 by the induction of apoptosis.1108, 1109 Moreover, the activin-binding protein follistatin inhibits activin.1108 In the normal prostate, follistatin is expressed by the stromal cells and basal cells.1108 However, cancer acquires the expression of follistatin.881, 1108, 1110 Hence, colocalization in cancer in vivo suggests that resistance to the growth-inhibitory effects of activin may be conferred by follistatin.881

5.5.3 The fibroblast growth factor family of peptides and receptors

The fibroblast growth factor (FGF) family of peptides currently is comprised of more than a dozen members, many of which are expressed in the prostate. In the adult prostate, FGF-1 (acidic FGF) is expressed at low levels or is undetectable, whereas FGF-2 (basic FGF) is abundantly expressed.1111 FGF receptor-1 in benign prostatic epithelial cells activates protein-kinase signaling pathways in an age-dependent manner through phospholipase Cγ-interactive phosphotyrosine 766.1112 Syndecan-1 contains heparan sulfate motifs that bind to FGF receptor-1.1113 Extracellular-regulated kinase-microtubule-associated protein (MAP) kinase and transcription factor STAT3 are important components of FGF-1-mediated signaling that induce promatrilysin expression in cultured cells.1114

FGF-2 is produced by stromal cells1115 and acts as a mitogen on the normal epithelium.1116 However, cancer cells acquire autocrine expression of FGF-2,1111, 1117, 1118 which may stimulate cancer cell proliferation,1119, 1120 and elevate the titer of FGF-2 in the serum of prostate cancer patients.1118, 1120 FGF-binding proteins, including FGF-BP and HPp17, are secreted proteins that mobilize and activate FGF-2 from the extracellular matrix.1121 In addition to the mitogenic action of FGF-2 on cancer cells, FGF-2 also enhances cell motility,1122 which may reflect an acquired capacity for metastasis. The ability of cancer cells to invade and metastasize is a reflection of FGF-2 regulation of the turnover of extracellular matrix by modulating the expression of proteases and promoting the synthesis of collagen, fibronectin, and proteoglycans. FGF-2 expression in PIN and in cancer also contributes to angiogenesis in primary1123, 1124 and metastatic cancer,82 thereby circumventing growth limitations of diffusion in the supply of nutrients and removal of waste products.

FGF-2 maintains survival of androgen-sensitive cancer cells through positive control of bcl-2 and down-modulation of AR protein expression, thereby allowing escape of select clones from androgen regulation.1125 FGF receptor 2 (IIIb) mRNA is down-regulated in androgen-dependent and androgen-independent cancer, whereas FGF receptor (IIIc) is down-regulated only in androgen-dependent cancer.1126

The FGFs FGF-3, FGF-4, FG-5, and FGF-6 are oncogene products. FGF-3 (int-2 gene product) most likely is not involved in localized prostate cancer.1127 Conversely, expression of FGF-3 and FGF-5 correlates with progression to cancer in the Dunning rat, and FGF-3 and FGF-5 are overexpressed in human cancer.1128 Hence, the role of these oncogene products in human prostate cancer remains unclear.

FGF-7, also known as keratinocyte growth factor, has been investigated extensively in rodent models, with limited confirmatory studies in the human prostate.1129 FGF-7 is an androgen-regulated peptide1130 that is expressed predominantly in stromal cells of the rat1130 and human prostate,1131, 1132 but it also is localized to epithelial cells in the human fetal prostate, normal adult prostate, and prostatic cancer.1133 FGF-7 stimulates the proliferation of prostate epithelial cells1116, 1130 but not the LNCaP cancer cell line.1119 However, rodent Dunning cancer cells exhibit a reduced response to FGF-7 that has been attributed to exon switching in FGF receptor (FGFR) isoforms. In this context, epithelial cells of the Dunning model appear to normally express the FGFR2 (IIIb) isoform,1117 which can bind FGF-7 as a mitogen. However, cancer cells express the alternatively spliced FGFR2 (IIIc) isoform,1117 which preferentially binds FGF-2 over FGF-7. Both FGFR2 isoforms are present in human prostate cancer.1132 The ability of autocrine FGF-2 to substitute for paracrine FGF-7 in cancer cells that exhibit the FGFR (IIIc) isoform facilitates cancer growth.

FGF-8, also known as androgen-induced growth factor, is present in epithelial cells, PIN, and cancer as multiple isoforms.1134, 1135 It is expressed in the LNCaP,1136, 1137 DU-145, and PC-3 cancer cell lines,1136, 1137 and it enhances the growth of LNCaP cells.1136 FGF-8b protein expression correlates with ARs1138 and VEGF in prostate cancer tissue.1139 Overexpression of FGF-8 occurs in cancer, and the level of expression appears to be related to disease progression.

5.5.4 The insulin-like growth factor family of peptides, receptors, and binding proteins

The insulin-like growth factor (IGF) system is characterized by complex interactions between the IGFs, their receptors, high-affinity binding proteins, receptors for these binding proteins, and proteases. Epithelial cells cultured from benign prostate, BPH, and cancer do not secrete significant amounts of IGF-I or IGF-II.1140 However, contradictory reports regarding several prostate cancer cell lines derived from metastases indicate that these cells do not secrete IGF-I1141 or that they are capable of expressing IGF-I1142 and IGF-II.1143 Conversely, stromal cells secrete IGF-II1144 and possibly express IGF-I. Irrespective of paracrine or autocrine origin, both IGF-I and IGF-II stimulate the growth of epithelial cells derived from primary cultures1140 and cancer cell lines,1141 perhaps by interfering with apoptotic signaling1145–1147 or by down-regulating synthesis of SHBH.511 Furthermore, IGFs appear to interact with the EGF autocrine loop.1148 Plasma IGF-I declines with age,1149 and an elevated serum IGF-I is a risk factor for the development of both BPH1150 and prostate cancer,832, 1151–1157 especially advanced-stage cancer.1158–1160 Free IGF-I concentrations in prostatic fluid were not found to be predictive of prostate cancer.1161

The type I IGF receptor (IGF-IR), which binds IGF-I preferentially, is expressed in epithelial cells from benign prostate, BPH, and cancer1140, 1162 as well as prostate cancer cell lines derived from metastases1141, 1148 and stromal cells.1144, 1163 Antagonism of IGF-IR function with IGF-I analogs inhibits the growth of prostate cancer cell lines,1142 similar to neutralizing antibodies.1164 The IGF-IR mediates signal transduction upon binding either IGF-I or IGF-II through its intrinsic tyrosine kinase activity.1165 The IGF-IIR does not contain intrinsic kinase activity, suggesting an alternate role for IGF-IIR. IGF-IIR bears a mannose 6 phosphate-binding site with an opposing effect on cancer cell growth from the IGF-II binding site.1166, 1167 Hence, IGF-IIR may function as a negative growth-regulatory molecule, as suggested for breast cancer.

The ability of the IGFs to mediate growth through their receptors can be altered by interaction with a variety of IGF-binding proteins (IGFBPs). Receptors for some of the IGFBPs have been described, but their role in the prostate remains poorly understood. Epithelial cells cultured from normal prostate, BPH, or cancer express IGFBP-2, IGFBP-4,1140 and possibly IGFBP-3,1168 but not IGFBP-1.1140 IGFBP-1 may be involved in neuroendocrine differentiation in prostate cancer.1169 Elevated levels of IGFBP-2 and decreased levels of IGFBP-3 are present in the serum from patients with prostate cancer1170, 1171; however, it may172, 1154 or may not1172, 1173 be predictive of cancer risk. IGFBP-2 correlates with serum PSA1174 and with cancer burden.150 IGFBP-3 stimulates apoptosis and has an apparent cancer-suppressive effect.1175 IGFBP-5 accelerates androgen-independent prostate cancer upon stimulation by castration.1176

Proteolytic cleavage of IGFBPs reduces the affinity for IGFs,1177 facilitating greater interaction of IGF with its receptor and increased mitogenic activity. PSA is a serine protease that preferentially cleaves IGFBP-31178 and IGFBP-5,1179 thereby stimulating the growth of prostatic epithelial cells1144 and reducing the levels of IGFBP-3 in the serum.1180 The nerve growth factor (NGF)-γ subunit, which shares a high sequence homology with PSA, also cleaves IGFBP-3 at a 3-fold lower concentration compared with PSA.1179 The NGF-γ subunit also displays potent proteolytic activity against IGFBP-4 and IGFBP-6.1179 Hence, cleavage of IGFBP-3, IGFBP-4, and IGFBP-6 by PSA and NGF-related mechanisms1178, 1179 most likely contributes to the bioavailability of IGF-I in the serum, consistent with the elevated serum levels of IGF-I in prostate cancer.1151

5.5.5 The nerve growth factor family of peptides and receptors

Nerve growth factor (NGF) is the prototypic member of the neurotrophin family of growth factors that, in man, includes brain-derived neurotrophic factor (BDNF), neurotrophin-3 (NT-3), and NT-4. NGF is expressed abundantly in the prostate,39, 1181, 1182 and immunoreactive protein is present in the stroma from benign tissue, BPH, and cancer,39 although finding this has been refuted.1183, 1184 Radical prostatectomy specimens were analyzed immunohistochemically for neurotrophins (NGF, BDNF, NT-3, and NT-4/5) and their receptors (tyrosine kinase A [TrkA], TrkB, TrkC, and the p75 neurotrophin receptor [p75NTR]), with expression in benign and malignant epithelial cells but not in stromal cells.1183 Smooth muscle stromal cells express the genes for NGF and BDNF.1185 NGF-immunoreactive protein produced by stromal cells in vitro stimulates the growth of the cancer cell line TSU-pr1 in a paracrine manner.1181 Furthermore, exogenous NGF-β stimulates the anchorage-independent growth of the LNCaP cell line.1119 All androgen-refractory cancer cell lines derived from metastases (DU-145, PC-3, and TSU-pr1) express NGF in an autocrine manner,1181, 1182, 1185 whereas the androgen-responsive LNCaP cell line does not express the NGF gene.1182, 1185 In addition, some androgen-refractory cell lines express the genes for BDNF and NT-4 but not the gene for NT-3.1185 Hence, it appears that the normal prostate expresses stromal cell-derived NGF and BDNF for paracrine regulation of epithelial cell growth and that, after prostatic carcinogenesis, androgen-refractory cancer cells acquire autocrine expression of NGF, BDNF, and NT-4. Thus, cancer cells that produce autocrine NGF, BDNF, and NT-4 are able to escape paracrine dependence on stromal cell-derived NGF and BDNF. Cancer cell migration within the prostate is often by direct extension around prostatic nerves; thus, the up-regulation of autocrine neurotrophin expression in cancer may be associated with invasion along the perineural space and metastasis.1185

The low-affinity p75NTR, a glycoprotein member of the TNF receptor (TNF-R) family, is expressed in varying degrees in epithelial cells,39, 1182 with progressive loss during cancer progression.1186 Immunoblotting,1187 immunofluorescent,39 and immunohistochemical1187 studies have shown that expression of the p75NTR protein declines in prostate cancer. Loss of expression of p75NTR is correlated with cancer grade.1188 Benign epithelium and PIN retain full expression of the p75NTR, whereas well differentiated cancer has much less p75NTR.1188 Moderately and poorly differentiated prostate cancers exhibit even fewer cells with p75NTR expression.1188 Moreover, this protein is absent in four cancer cell lines derived from metastases.1187 Loss of expression of the p75NTR in prostate cancer may be related to its role in the induction of programmed cell death. This is supported by studies in which the TSU-pr1 prostate cancer cell line, which has lost expression of the p75NTR protein,1187 was transfected stably with a p75NTR expression vector.1189 The parental TSU-pr1 cell line was responsive to NGF stimulation of growth, but the TSU-pr1 cells stably transfected with the p75NTR expression vector exhibited a reduction in NGF-mediated growth in proportion to the degree of p75NTR protein expression.1189 Hence, p75NTR appeared to inhibit the growth of transfected TSU-pr1 prostate epithelial cells. Transient transfection studies showed that TSU-pr1 cells expressing the p75NTR protein underwent a 4-fold increase in the rate of programmed cell death.1189 This suggests that p75NTR inhibits the growth of prostate epithelium at least in part by induction of programmed cell death. Hence, the loss of p75NTR expression in prostate cancer cells appears to eliminate a potential apoptotic pathway in these cells, thereby facilitating the immortalization of these epithelial cells during carcinogenesis.1188 Consequently, the p75NTR may be a candidate tumor suppressor gene in the human prostate+ by enhancing cell cycle quiescence, reducing proliferation, and increasing apoptosis.1186, 1190 This is supported by several studies, in which tumor suppressor gene activity was observed on human chromosome 17q,1191, 1192 which is the site of p75NTR.1193 Moreover, this tumor suppressor activity in the human prostate cannot be accounted for fully by the BRCA11192 and the NM23 suppressor genes,1191 which also localize to 17q. NGF treatment induces the reexpression of ARs and p75NTR, and this effect is blocked by androgen antagonists.1194 Expression of the p75NTR is decreased or lost in cancer; thus, neurotrophin-mediated growth of cancer must be mediated by another cell-surface receptor. The Trk family of receptors may serve this role by binding neurotrophins with high affinity.1185, 1195 Trk receptors are expressed in PIN, in cancer,1195 and in cancer cell lines derived from metastases.1185, 1195 TrkA immunoreactivity is restricted to the basal epithelial cells in some acini (37%), a pattern that is unchanged or extends to the whole acini in BPH, and varies widely in prostate cancer; in normal tissue and BPH, TrkC-IR is detected exclusively in the stroma and increases progressively in the epithelial cells of well-to-moderately differentiated prostate carcinoma; whereas, in stromal cells, there are no substantial changes. TrkB-IR is absent in all samples.1196 Trk receptors appear to transduce the proximate signal for neurotrophin-mediated growth in cancer. Because Trk mediates the NGF-stimulated growth response in cancer, antagonism of the Trk receptor may inhibit cancer growth.1184, 1197 In this context, members of the K252 family of kinase inhibitors, the indolocarbazoles, selectively inhibit activity of the Trk receptors at nanomolar concentrations1198 and inhibit NGF-stimulated Trk phosphorylation in the TSU-pr1 cancer cell line.1197 Concurrently, K252a inhibits the growth of cancer cell lines in vitro,1197 further supporting a role of the Trk receptors in the neurotrophin-mediated growth of prostate cancer. The Trk receptor originally was isolated as an oncogene from colon cancer,1199 but similar Trk oncogenes within the prostate have not been identified to date.

5.5.6 Vascular endothelial growth factor and receptors

Vascular endothelial growth factor (VEGF) promotes angiogenesis in a wide variety of normal and neoplastic tissues and is a potent mitogen for endothelial cells. In vivo, studies have reported a complete lack of VEGF in BPH and epithelial cells,1200 or, conversely, 2 isoforms of the protein (VEGF165 and VEGF189) in stromal cells of BPH.1201 In immunohistochemical studies, VEGF has been demonstrated in BPH, in PIN, and in prostate cancer epithelial cells1202–1204 and stroma1139, 1205, 1206 with a significant reduction after castration1207 and androgen-deprivation therapy.1208, 1209 Numerous reports have described VEGF expression in organ-confined cancer,1200, 1201 in cancer cell lines derived from metastases, and in xenografts of prostate cancer.706 Exogenous VEGF appears to promote the growth of xenograft tumors,1210 and androgen deprivation inhibits VEGF expression.706 Anti-VEGF antibodies inhibit androgen-independent prostate cancer growth in xenografts by the induction of endothelial cell apoptosis.1211, 1212 The tumor suppressor gene PTEN appears to modulate angiogenesis by regulating VEGF expression.1213 The expression of VEGF has been correlated with the expression of mRNAs that encode other angiogenic cytokines (angiopoietin-1 and angiopoietin-2), endothelial cell receptor tyrosine kinases (Flt-1, KDR, and Tie-1), and endothelial CAMs (VE-cadherin and PECAM-1).1214 Taken together, these observations suggest that cancer cells express VEGF for angiogenesis of developing cancer, thereby circumventing oxygen diffusion as a rate-limiting step in the growth of prostate cancer. Hypoxia and VEGF expression both show a strong, positive correlation.1215

Plasma VEGF concentrations are elevated significantly in patients who have androgen-insensitive prostate cancer compared with the concentrations in patients who have localized prostate cancer, and VEGF levels are correlated inversely with survival.1216 The VEGF 1 receptor Flt-1 (c-fms-like Trk receptor) is present in prostatic endothelial cells and in most epithelial cells from BPH, with increased1217 or decreased expression in PIN and cancer.1218, 1219

5.5.7 Platelet-derived growth factors and receptors

Platelet-derived growth factors (PDGF) exist as dimers formed from A and B chains that bind with differing affinities to the PDGFα receptor and the PDGFβ receptor. PDGFs signal through their receptors to elicit a diversity of cellular responses in vitro, including cell proliferation, survival, transformation, and chemotaxis.1220 In vivo, contradictory studies have reported a lack of either form of ligand or receptor in BPH,1221 whereas others have observed limited expression of the PDGF-β receptor in BPH.1222 Conversely, epithelial and stromal cells in PIN and in cancer express PDGF-A and the PDGF-α receptor,1221, 1223 whereas PDGF-B and the PDGF-β receptor either are absent or are expressed weakly.1221, 1223 Hence, PDGF-A ligand and its PDGF receptor may modulate autocrine growth in PIN1223 and in cancer,1221 thereby playing a role in malignant transformation.

Platelet-derived endothelial cell growth factor is a potent angiogenic factor that is expressed in benign acini and stromal cells but is not expressed in cancer cells. It has a strong, positive correlation with microvessel density.1224, 1225

5.5.8 Other growth factors

HGF, also called scatter factor, is expressed in the human prostate exclusively by the stroma1226, 1227 and stimulates the proliferation1119, 1228 and motility1228 of cancer cells.1229, 1230 HGF induces motility and scattering of cells from many organs, apparently by inducing matrilysin-mediated cleavage to the extracellular domain of E-cadherin, resulting in its dissociation from the cadherin-catenin complex.855 The correlation of HGF and c-met expression with malignant progression of cancer cells suggests a role in metastases.

HGF binds to the c-met protooncogene product, which is located exclusively in intermediate epithelial cells of PIN,1231, 1232 BPH,1233 and cancer.1228, 1231, 1233–1235 The proportions of tissue samples that express c-met increase progressively from BPH,1233 to PIN,1231 to primary cancer,1228, 1233 and to metastases.1228, 1233 Serum concentrations of HGF are increased in men with metastatic prostate cancer independent of serum PSA level or patient age.1236

Cytokines exert cytostatic and immunomodulatory effects on cancer cells, and polymorphisms of cytokine genes influence prostate cancer development through regulation of the antitumor immune response and angiogenesis.1237 Concurrently, there is limited evidence to suggest that IL-1, IL-2, or the interferons (IFNs) (IFN-α, IFN-β' and IFN-γ) are expressed during prostatic carcinogenesis. Conversely, IL-6 is secreted in a paracrine manner from stromal cells1238 and in an autocrine manner from cancer cells.1239–1241 IL-6 signals IL-6 receptors on cancer cells in vitro1240 and in tissue to regulate growth and apoptosis,1238, 1242, 1243 apparently by abolishing growth control by Rb protein and activation of the MAP kinase1244 and STAT3 signaling pathways.1245 IL-6 and, to a greater extent, IL-10 up-regulate the expression of TIMP-1,1246, 1247 consistent with an overall inhibitory effect on cancer. IL-6 activates AR in the absence of androgens and, thus, is implicated in the transition of cancer from an androgen-dependent phenotype to an androgen-independent phenotype.1243, 1248 Nuclear factor κB (NF-κB)-mediated IL-6 synthesis is required for IL-1β-induced promatrilysin expression.1249 Serum concentrations of IL-6 and its receptor are independent predictors of PSA failure after surgery.1250

IL-11, its receptor, and the activated form of STAT3 were present in BPH and were up-regulated in prostate cancer.1251 Proinflammatory IL-15 stimulates T-cell growth, and its production is driven by IFN-γ (a T-cell product), thus creating a paracrine loop. IL-15 and its receptor are present in all prostatic cell types.1252 IL-17 protein also is expressed in benign and malignant prostate.1253

Another cytokine that is expressed more strongly in cancer compared with benign prostatic tissue, TNF-α,1082, 1254 inhibits chemotaxis1242 and proliferation of cancer cell lines.1242, 1255 Genotype changes in − 308 and 488 of TNF-α are strong risk factors for prostate cancer.1256 Moreover, TNF-α is cytotoxic for cancer cell lines1257 by inducing bcl-2-mediated1258 programmed cell death,1258–1261 activating NF-κB,1262, 1263 stimulating JFC1 expression,1264, 1265 and inducing cyclooxygenase-2 expression.1266 The effects of TNF-α are mediated by TNF-RI (55 kD) and TNF-RII (75 kD), both of which are expressed by prostate cancer cell lines.1267 TNF-related, apoptosis-inducing ligand (TRAIL) increases apoptosis significantly in both benign prostatic epithelial cells1268 and prostate cancer,1269 requiring capsase 8 for function.1270, 1271 Decreases in TRAIL expression may result from decreased activity of fork-head transcription factors FKHRL1 and FKHR when PTEN is lost.1272, 1273

Zinc sensitizes cells to TNF-α, reducing expression of the NF-κB-controlled antiapoptosis protein c-IAP2 and activating c-Jun kinases.1274 Androgen-independent prostate cancer xenografts have greater constitutive NF-κB binding activity compared with their androgen-dependent counterparts.1275 Soy protein supplementation appears to protect cells from oxidative stress by inhibiting NF-κB and decreasing DNA adducts.942, 1276

5.6 Loss of Heterozygosity, Nongrowth Factor/Receptor Oncogenes, and Other Chromosomal Changes

The loss of chromosomal heterozygosity places at risk the matching chromosome to mutational events that may contribute to the formation of cancer cells. In general, mutational activation of protooncogenes and inactivation of tumor suppressor genes have been implicated in the formation of prostate cancer. In addition, other chromosomal changes that affect specific genes may confer a growth advantage on cancer cells.

5.6.1 Chromosome gains and losses, including loss of heterozygosity

Prostate carcinogenesis apparently involves multiple genetic changes, including loss of specific genomic sequences that may be associated with inactivation of tumor suppressor genes and gain of some specific chromosome regions that may be associated with activation of oncogenes. The most common genetic alterations in PIN and in carcinoma are gain of chromosome 7, particularly 7q31; loss of 8p and gain of 8q; and loss of 10q, 16q and 18q.1277

5.6.1.1 Gain and loss of chromosome 7

Gain, deletion, and translocation of 7q22-q31 are common in prostate adenocarcinoma.1278, 1279 FISH studies showed that aneusomy of chromosome 7 is frequent in prostate cancer and is associated with higher cancer grade, higher pathologic stage, and early patient death from prostate cancer.806, 1280–1282 Polymerase chain reaction (PCR) analysis of microsatellite markers identified frequent imbalance of alleles mapped to 7q31 in prostate cancer.1283–1288 Allelic imbalance of 7q31 was correlated strongly with tumor aggressiveness, disease progression, and cancer-specific death.806 These findings suggest that genetic alterations of the 7q arm play an important role in the development of prostate cancer. However, it was assumed that allelic imbalance detected by PCR analysis was a result of chromosomal loss or deletion,1285, 1288 while FISH studies uniformly have observed gain of the chromosome 7 centromere.803, 806, 1280–1282 Resolving this discrepancy will be critically important for the definition of 7q31 anomalies in prostate cancer.

Unexpectedly, overrepresentation (gain) of 7q31 was more common than deletion of 7q31 and was correlated strongly with Gleason score.1289 FISH analysis of metaphases from aphidicolin-induced, chromosome 7 only, somatic cell hybrids1289, 1290 revealed that the DNA probe for D7S522 spanned the common fragile site FRA7G at 7q31. The coincidence of FRA7G with the region that showed the greatest allelic imbalance suggests that the instability in this fragile site may be responsible for both the gains and the losses of this region. These observations suggest that overrepresentation of 7q31 and/or genes in these regions may be important for the development and/or progression of a significant proportion of prostate cancer. Conversely, it also is possible that prostate cancer progression is associated with increased chromosomal fragility, perhaps as a result of other genetic alterations. The frequency of trisomy 7 was much greater in carcinoma than in PIN, and allelic imbalance at 7q31 was slightly more frequent in prostate cancer than in PIN (30% vs. 17%).1291 A recent study showed that aggressive and late-onset prostate cancer is linked to chromosome 7q31–33 in Germans.1292

5.6.1.2 Loss of 8p and gain of 8q

The chromosome 8p arm is 1 of the most frequently deleted regions in prostate cancer.804, 1293–1296 The rate of 8p22 loss ranged from 29% to 50% in PIN, from 32% to 69% in primary cancer, and from 65% to 100% in metastatic cancer.1294–1296 Other frequently deleted 8p regions include 8p21 and 8p12.804, 1293 Emmert-Buck et al. found loss of 8p12–21 in 63% of PIN foci and in 91% of cancer foci using microdissected frozen tissue.804 Bostwick et al. detected loss of 8p21–12 in 37% of PIN foci and in 46% of cancer foci. In a recent study, 7 potentially important mutations in the macrophage scavenger receptor 1 gene (MSR1), located at 8p22, were observed in families affected with prostate cancer, and an indication of cosegregation between these mutations and prostate cancer was reported.1297 These findings suggest that more than 1 tumor suppressor gene may be located on 8p, and inactivation of these tumor suppressor genes may be important for the initiation of prostate cancer.

In addition to loss of the 8p arm, gain of the 8q arm has been reported in prostate cancer.809, 1294, 1298, 1299 Bova et al. found gain of 8q in 11% of primary cancers and in 40% of lymph node metastases.1296 Amplification of 8q DNA sequences is found in 75% of cancers metastatic to lymph nodes. Similarly, Visakorpi et al. found gain of 8q far more frequently in locally recurrent cancer than in primary cancer.1299 Cher et al. also detected frequent gain of 8q in metastatic and androgen-independent prostate cancer.1298 Using FISH, Qian et al. observed that gain of chromosome 8 was the most frequent chromosomal anomaly in metastatic foci, and the frequency was much greater compared with that in PIN and carcinoma.803 Jenkins et al. identified c-myc gene amplification in 22% of metastatic foci that was much more frequent than in primary cancer (9%), suggesting that the 8q arm may harbor a gene(s) in which amplification and overexpression plays a key role in the progression and evolution of prostatic carcinoma.809 Gain in chromosome 8q was correlated with early progression in prostate cancer after androgen-deprivation therapy.1300 It is noteworthy that gain of the chromosome 8 centromere or of the 8q arm occurred simultaneously with loss of portions of the 8p arm in PIN and in carcinoma.809, 1298, 1299, 1301, 1302 One simple genetic mechanism that may explain these prior observations is the presence of multiple copies of isochromosome 8q in cancer cells. Simultaneous FISH studies with probes specific for 8p, 8q, and the chromosome 8 centromere support this explanation.1277

5.6.1.3 Gain and loss of other chromosomal regions

There is a high frequency of allelic imbalance at 10p and 10q in prostate cancer.1303 The most commonly deleted region on the 10q arm includes bands 10q23–24, and allelic loss of this region may inactive the MXI-1 gene. Recently, the PTEN candidate gene in this region has been cloned.1304 All four prostate cancer cell lines had mutations of this gene.1304, 1305 Trybus et al. mapped the common region of 10p deletion to 10p11.2.1306 Chromosome 16 also had frequent allelic imbalance in prostate cancer. Allelic imbalance at 16q was present in approximately 30% of clinically localized prostate cancers,1307 and there was a high frequency of allelic imbalance at 16q23-q24.1308, 1309 The most commonly deleted region was located at 16q24.1-q24.2, and this deletion was associated significantly with cancer progression.1308, 1309 Patients with tumors that showed both 8p22 and 16q24 deletions had a significantly greater frequency of lymph node metastases compared with nonmetastases. A recent study showed that patients with tumors that showed both 8p22 and 16q24 deletions had a significantly greater frequency of lymph node metastases.1310 The frequency of loss of 18q22.1 varied from 20% to 40%. It was reported recently that chromosome 19 harbors a gene for tumor aggressiveness.1311 Recently, Cunningham et al. detected an allelic imbalance at 21q22.2–22.3 in 23% of prostate cancers and at 3p25–26 in 20%.1312 Other regions that demonstrated frequent allelic imbalance included 5q12–23, 6q, 13q, and 17p31.1.1312, 1313 Loss of 10q, 16q, and 18q also has been reported in PIN.1291, 1314 Bostwick et al. found that loss of 18q12.2–12.3 was greater in prostate cancer than in PIN (52% vs. 19%; P = 0.01), indicating that this region may harbor a gene in which inactivation may be important for the progression of PIN to carcinoma.1291

5.6.2 Nongrowth factor/receptor oncogenes and tumor suppressor genes

Point mutations in members of the Ha-ras, Ki-ras, and N-ras gene family transform normal cellular ras genes into activated oncogenes. Activation of p21ras inhibits prostatic acinar morphogenesis1315 and enhances farnesylation-mediated proliferation,1316 consistent with a role in carcinogenesis. The frequency of ras mutations is prostate cancer reportedly differs between Japanese men and American men.1317 In Japanese men, 24% of cancers contain ras mutations.686 Such mutations were not detected in benign tissue or in BPH, whereas the frequency increased with advanced stage of cancer, higher Gleason score,686 and lymph node metastases.1318 Conversely, men in North America and Europe have a low frequency of ras mutations in prostate cancer.1307, 1319 Ras mutations in prostate cancers among white men appear to be late-stage events,1320 suggesting that they may not be significant in early development.685 The greater frequency of ras mutations in prostate cancers among Japanese men has been attributed to advanced disease stage.686, 1317, 1321 Hence, the risk of ras mutations in prostate cancer appears low in white men, although, when it does occur, the prognosis may be relatively poor due to advanced disease stage and tumor grade.

The p53 tumor suppressor gene functions as a regulator of the cell cycle, and p53 abnormalities are associated with many human malignancies. However, there is a much lower incidence of p53 genetic alterations in primary prostate cancer compared with the incidence in bladder, colon, lung, and breast cancers.1322 Nuclear accumulation of p53 protein is associated strongly with missense p53 mutations.689 Most immunohistochemical studies concluded that mutant p53 expression is a late event in localized prostate cancer1323, 1324 that usually is present in higher Gleason score (≥ 7) cancer.689, 1325, 1326 Moreover, the frequency of mutant p53 expression is elevated in untreated metastatic cancer,995,1322,1327 hormone-refractory cancer,689, 1323, 1327 and recurrent cancer.995 Consequently, p53 gene inactivation does not appear to be essential for the development of metastases1328 and appears to be of limited prognostic value in patients with primary or metastatic cancer.1328

The WAF1/CIP1 gene encodes a p21 cyclin-dependent kinase (Cdk) inhibitor that plays a role in the regulation of the cell cycle. On induction by p53, p21WAF1/CIP1 binds to Cdk2, resulting in the down-regulation of Cdk2 activity and G1 growth arrest. Prostatic mutations in the WAF1/CIP1 gene abrogate this apparent tumor suppressor gene activity,875 thereby facilitating escape of G1/S-checkpoint control with propagation into S-phase and maintenance of malignant potential. There is an increase in WAF1/CIP1 polymorphisms in prostate cancer,1329 but no correlation exists between WAF1/CIP1 expression and grade, stage, or cancer progression.1330 This may be related indirectly to the frequency of p53 mutations, which tend to occur in high stage cancer1328 and would not affect the normal activity of p21WAF1/CIP1 in low-stage, organ-confined cancer.

The Cdk inhibitor (p27Kip1) also regulates cell proliferation negatively by mediating cell cycle arrest in G1. p27Kip1 expression decreases with higher Gleason score and seminal vesicle involvement by cancer.1331 Furthermore, p27Kip1 expression was an independent predictor of treatment failure among patients with lymph node-negative cancer after they underwent radical prostatectomy.1331

In many cell systems, overexpression of the c-myc protooncogene stimulates cell cycle progression and malignant transformation. Some studies have failed to identify c-myc gene amplification1127 or an association with grade1332 and stage,1333 but most reports suggest that it plays a role in the regulation of prostate growth and carcinogenesis. Despite early studies demonstrating elevated expression of c-myc in all grades of cancer1334 or predominantly in tumors with a Gleason score > 5,683 current belief is that there is substantial amplification with increasing grade of cancer, particularly in metastases.809 Moreover, transfection with c-myc and ras induces carcinogenesis of the murine prostate.677 Myc expression also correlates with growth of androgen-responsive prostate epithelium and growth factor/receptor expression in cancer cells. For example, increased c-myc expression is associated with elevated EGFR expression and increased binding of IGF-I to prostate cancer cells.1335 Conversely, inhibitors of prostate cancer growth, including TGF-α1,1336 TNF-α,1337 and c-myc antisense oligonucleotides,1338 inhibit c-myc expression. Furthermore, elevated expression of c-myc is associated with the acquisition of resistance to anticancer drugs.1339 Hence, elevated expression of c-myc expression appears to be associated with the development of prostate cancer and may be a risk factor for disease progression. Furthermore, the c-myc gene may be a potential target for genetic therapy in patients with prostate cancer.1340

The Rb protein binds to E2F transcription factors, thereby inhibiting transactivation of genes that promote progression through the G1-S transition of the cell cycle. Reintroduction of wild-type Rb into prostate cancer cells that express mutated Rb suppresses tumorigenicity,1341, 1342 consistent with its role as a tumor suppressor gene product. Loss of heterozygosity of the Rb gene occurs with great frequency in low-stage, low-grade prostate cancer and in more advanced cancer.1343–1347 Hence, early inactivation of the Rb gene appears to be an important event in prostatic carcinogenesis1348 and was a prognostic marker of disease progression after surgery in some studies,1349–1351 but not all studies.1352 Androgen independence is achieved through deregulation of the androgen to Rb signal,1353 and androgen-deprivation therapy lowers the frequency of Rb abnormalities.1354 Rb immunoreactivity is greater in PIN and cancer than in benign prostatic epithelium, and a reduction in the Rb protein level occurs through Rb degradation by the ubiquitin/proteasome pathway preceded by selective Rb phosphorylation by cyclin A/Cdk2 and cyclin B/Cdk1.1355

The deleted-in-colorectal-cancer (DCC) tumor suppressor gene exhibits a low frequency of loss of heterozygosity and loss of expression in human prostate cancer.875 However, genetic alterations in the DCC gene occur only in high-grade cancer and metastases,1356 suggesting that it is a late event in transformation. The adenomatous polyposis coli tumor suppressor gene also exhibits loss of heterozygosity,875 mutation,875, 1357 and loss of expression875 in prostate cancer, albeit at a low frequency.1358

5.6.3 Metastasis suppressor genes

The molecular mechanisms of metastasis are understood poorly. In the past decade, numerous putative metastasis suppressor genes have been identified, including those that apparently do not affect primary prostate cancer growth: nm23-H1, KAI1, CD44, and MAPK kinase 4.1359

5.6.3.1 Metastasis suppressor gene nm23

The nm23 gene family currently is comprised of nm23-H1, nm23-H2,1360 DR-nm23,1361 and nm23-H4,1362 all of which are expressed in the human prostate. These encode nucleosidase diphosphate kinases, which may function as metastasis suppressor genes. Microcell transfer of a truncated human chromosome 17 containing an nm23 gene (and many other genes) into prostate cancer cells suppresses metastatic ability.1191 The expression of nm23-H1 correlates with the localization of PCNA in cancer,1363 correlates negatively with stage and grade,1090 and is higher during proliferation.1363 Early studies suggested that nm23-H11360 and nm23-H21360, 1364 were correlated inversely with progression of prostate cancer to a metastatic phenotype, but subsequent studies revealed that an increase in nm23-H1 expression was associated with the metastatic phenotype.1365, 1366 Results correlating nm23-H2 expression with cancer grade have not been confirmed.1364, 1365 Hence, the role of nm23 in prostate cancer metastasis remains to be clarified.

5.6.3.2 KAI/CD821 metastasis suppressor gene

Microcell mediated transfer of human chromosome 11p11.2–p13 into rat prostate carcinoma cells suppressed metastatic activity independent of tumorigenicity or cancer growth.1191 A metastasis suppressor gene from the region of chromosome 11p11.2, designated KAI1/CD82, was reduced in human prostate cancer cell lines derived from metastases.700 Localized expression of KAI1 to the plasma membrane1367 was high in benign and hyperplastic epithelial cells1367 and declined with increasing Gleason score.1368–1371 KAI1/CD82 expression is lowest in metastases1367 and recurrent cancer after androgen-deprivation therapy.1367–1369 KAI1 expression is stronger in BPH associated with cancer compared with its expression in cancer-free BPH.1372 Disease progression in patients with Stage T1b cancer has been correlated with lower expression of KAI1.1373

To our knowledge, the mechanism for loss of KAI1 expression remains to be clarified.1359 In an early report, it was determined that the down-regulation of KAI1 expression only infrequently involved mutation or allelic loss.1367 However, other reports showed a high frequency of loss of heterozygosity or allelic imbalance at the centromeric region of 11p that contains the KAI1 gene.1374 The down-regulation of KAI1 appears to be at the posttranscription level and is related to loss of p53 function.1375

5.6.4 Hypermethylation of the glutathione-S-transferase gene GST-P1

Human cells are challenged constantly by various endogenous and exogenous toxins, many of which have potent oncogenic properties. Most cells are equipped amply with detoxification mechanisms for the removal or neutralization of these harmful substances. The glutathione-S-transferases (GSTs) belong to the Phase II detoxification enzymes. Their primary function is to conjugate toxins (reactive electrophiles) with glutathione, a universal intracellular reductant. Reduced activity of GSTs lowers the detoxification capability and likely increases susceptibility to cancer-causing xenobiotics. Conversely, GST overexpression in cancer is one mechanism of anticancer drug resistance and often is associated with a poor outcome after chemotherapy.

GST-P1 enzyme activity varies significantly among different cancer types and often is associated with patient survival. Greater than 90% of prostate cancers exhibit extensive deoxycytidine methylation of the GST-P1 promoter, leading to transcriptional inactivation of the gene.1376, 1377 The methylation-induced inactivation of GST-P1 is restricted to cancer and is not found in BPH.1378 At the protein level, intense GST-P1 staining is found in basal cells of BPH.1379 The acinar epithelium showed weak immunoreactivity in all benign specimens, and GST-P1 immunoreactivity is detected only in a small number of cancers.1380, 1381 High-grade PIN also has occasional positive staining.741, 1380, 1382 The near-absence of GST-P1 in PIN and in incidental cancer suggests that down-regulation of this enzyme is an early event in carcinogenesis.1383 It has been shown that the α and π class isoforms of GSTs inhibit adduction of activated PhIP metabolites to DNA in cell-free systems. In humans, silencing of GST-P1 through CpG island hypermethylation is found in nearly all prostate carcinomas and is believed to be an early event in prostate carcinogenesis.1384 It is easy to speculate that lack of GST-P1 activity may render prostatic epithelial cells susceptible to xenobiotic transformation. The marked difference between cancer and benign tissue in GST-P1 promoter methylation status may offer a unique opportunity for the development of a cancer-specific detection assay.1383–1385 Men who are homozygous for polymorphisms of GST-P1 have a 76% decreased risk of prostate cancer compared with matched control patients.1386 Furthermore, this polymorphism elevates the risk for early-onset prostate cancer1387 and response to androgen-deprivation therapy.1388 Quantitation of GST-P1 hypermethylation accurately detects the presence of cancer in needle biopsies with 73% sensitivity, 100% specificity, 100% positive predictive value, and 78% negative predictive value.1389

5.6.5 AR gene mutations

The AR belongs to a superfamily of ligand-dependent nuclear transcription factors that include the sex steroid hormone receptors, the thyroid hormone receptor, retinoic acid receptors, and a number of orphan receptors.1390–1396 The gene is located on the long arm of the X-chromosome at Xq11–12, contains 8 exons, and spans a length of approximately 90 kB of DNA. The minimal promoter region is located at − 74 to + 87, flanking the transcription start site. Transcription occurs from 1 of the 2 initiation sites, 11 base pairs apart, and approximately 1100 base pairs upstream of the translation-initiation methionine. Two AR transcripts with different 3′-untranslated sequence lengths are expressed in most target tissues. The major transcript is approximately10 kB in length, whereas the minor transcript is approximately7 kB long. The primary amino-acid sequence is approximately 910–919 amino acids in length, with a calculated molecular weight of 98 kDa. Similar to other steroid receptor proteins, the full-length AR protein contains four functional domains: the amino terminus, a DNA-binding domain, a hinge region, and the ligand-binding domain.

The N terminus, which is encoded by exon 1, comprises nearly half of the AR molecule and harbors 1 of 2 transcriptional activation sites. It has several homopolymeric amino-acid stretches, including a polyglutamine repeat, a polyglycine repeat, and a polyproline repeat. Exons 2 and 3 encode the DNA-binding domain. This region contains two zinc finger motifs. The N-terminus zinc finger interacts with androgen-responsive elements of androgen-regulated genes, whereas the C-terminus finger stabilizes the DNA-receptor interaction and mediates dimerization. The hinge region, which is encoded by the proximal region of exon 4, has a nuclear localization signal and most likely directs the AR to the cell nucleus. The ligand-binding domain is encoded by exons 4–8. It binds to androgens and undergoes conformational changes that lead to dimerization and release of the DNA-binding domain from inhibition. At the C-terminus end of the AR, a second transcriptional-activation domain is located between residues 360–528. Serine phosphorylation appears to be an important event in AR activation and stabilization.1359

The function of AR largely is dependent on protein levels and on the structural integrity of the protein and other transcription-activation factors.1397, 1398 AR gene mutations are present in prostate cancer prior to hormonal therapy and in hormone-refractory cancer. Dysregulation of AR function in prostate cancer results in an abnormal profile of regulated genes that includes cell cycle regulators, transcription factors, and proteins responsible for cell survival, lipogenesis, and secretion.1399 The activation of mutant AR by estrogens and weak androgens may confer on cancer cells an ability to survive testicular androgen deprivation by allowing activation of the AR by adrenal androgens or exogenous estrogens. Such mutations may confer a growth advantage, even without androgen deprivation, because prostate cancer has lower levels of 5-α-reductase and dihydrotestosterone than normal tissue.1400, 1401 CAG repeat length polymorphisms of the AR gene is associated with an increased risk of prostate cancer in some patients,1402, 1403 but not in others.1404–1406 Progression to hormone-refractory growth of prostate cancer may be mediated by AR gene alterations.1407 AR expression has been correlated with progression-free survival after hormonal therapy.1408

In early studies, AR expression was measured by radioligand-binding assays using 5α-dihydrotestosterone or [17-methyl]-methyltrienolone (R1881). Results suggest that there are higher levels of AR in prostate cancer than in BPH. Immunohistochemical results revealed that AR is present in most prostate cancers regardless of stage.1390, 1392, 1409–1414 Immunoreactivity is heterogeneous and is restricted to a limited number of cells within each cancer.991, 1409, 1415–1417 There was no apparent correlation in most studies between staining intensity or pattern and patient prognosis,1390, 1392, 1416 although this finding has been refuted.1408 However, some reports noted that an increase in AR heterogeneity or a decrease in AR immunoreactivity in cancer is associated with higher grade and poorer prognosis.991, 1408, 1409, 1415, 1416, 1418, 1419 A sensitive reverse transcriptase-PCR protocol for detecting the AR transcript revealed that expression was present in the majority of primary cancers but absent in approximately one-third of cases after androgen-deprivation therapy.1420 Expression was most consistent in high-stage and high-grade cancer prior to treatment.

Mutations in the AR gene may cause down-regulation or loss of receptor expression, production of hypoactive or hyperactive receptors, and alteration of receptor ligand specificity, and such mutations are present in > 50% of prostate cancers.1407 However, in prostate cancer, AR mutations consistently retain ligand-dependent transcriptional activity.1393 Amino acid substitutions (44%) are the predominant phenotype of most AR mutations.1421, 1422 Thus, many mutations may confer a growth advantage on cancer cells during progression.

In the androgen-insensitivity syndrome and in spinal and bulbar muscular atrophy, mutational changes of the AR gene almost always lead to receptor inactivation and androgen insensitivity.1394 In contrast, AR mutations in prostate cancer usually are associated with hypersensitivity, promiscuous usage of ligands, and gain in function. For example, AR mutations with 715 (Val-Met) substitution are transactivated by lower concentrations of adrenal androgens and progesterone than the wild-type receptor. AR with mutation at the 877 (Thre-Ser) or 874 (His-Tyr) codon, commonly found in metastatic cancers (including those found in bone marrow), can be activated by estrogens and progesterone.1427, 1428 These ARs also can be transactivated by antiandrogens, such as hydroxyflutamide and nilutamide, but not bicalutamide.1429 A germline mutation at 726 (Arg-Leu) can be activated by estrogen. These mutational changes, in general, permit transactivation of the receptor either by lower concentrations of androgen or by other nonandrogenic ligands. Prostate cancer cells endowed with these hypersensitized receptors, therefore, are expected to be able to survive in an androgen-deprived environment. Some of these mutations also may explain, in part, the “antiandrogen withdrawal syndrome.”1390, 1391, 1430 In comparison, mutations that cause inactivation of AR are rare. One study, however, revealed that 22% of latent carcinomas in Japanese men contained inactivating mutations of the AR gene, whereas these mutations were not found in latent cancers in American men. Inactivating mutations may function by preventing the progression of early-stage cancer in Japanese men.1431

Mutational changes are not found commonly in X-linked genes in prostate cancer. One X-linked gene that has been studied in parallel with AR is the hypoxanthine guanosine phosphoribosyl transferase (HPRT) gene.1418 No mutation has been found in HPRT in cancer with AR mutations, suggesting that genetic instability, by mechanisms such as impairment of DNA mismatch repair, may not be responsible for the wide range of mutations observed in the AR gene.1418 This possibility is supported by the finding that microsatellite instability is an uncommon phenomenon in first-degree relatives of patients with prostatic cancer.1432, 1433 Linkage of microsatellite instability and AR mutations raises the possibility of a cause-and-effect relation.

AR amplification is found rarely in primary cancer, but it is common in recurrent, therapy-resistant cancer (26–30%)1299, 1434 and in bone metastases.1435 In addition, AR amplification is frequent in cancer that has a prolonged response (> 12 months) to androgen-deprivation therapy.1434, 1436 Thus, the androgen-deprived environment may favor clonal expansion of cancer cells that express higher levels of AR due to gene amplification. Clones that express high levels of AR initially may have a growth advantage and survive in an environment with declining androgenic stimulation.

The polyglutamine region, which is encoded by CAG repeats in the N-terminus of the human AR, is polymorphic. CAG repeat length is correlated inversely with transactivation of the receptor.1437 The majority of prostate cancers are AR positive; thus, polymorphic differences or mutational changes in CAG repeat length may affect cancer risk and clinical progression. The median length of CAG repeats is 22 for Asians, 21 for whites, and 18 for African-American males.1438 Thus, AR with shorter CAG repeat lengths are more prevalent in racial groups with higher prostate cancer risk,1439–1441 although this has been disputed.1403–1406 Increased risk with polymorphic CAP repeat lengths also applies to Hispanic men, who traditionally have a relatively low risk,1402 and to Chinese men1442 and Japanese men.1443 Familial clustering of prostate cancer is not attributable to genetic polymorphisms in the CAG repeat or in the GGC repeat.1444 There is a significant correlation between CAG repeat length and patient age at the onset of cancer.1445 Shorter CAG repeat lengths may be associated with an earlier onset of cancer, suggesting that carcinogenesis is dependent on a more active AR. Germline shortening of the CAG repeat length has been correlated with advanced clinical stage and a poor prognosis.1438, 1446, 1447

In summary, the progression of prostate cancer likely requires wild-type AR. ARs with greater transactivating potential are promotional for this process. Thus, probable mechanisms for cancer progression include polymorphic differences or mutation-induced contraction of the polyglutamine repeat in exon 1, mutations that lead to hypersensitivity or gain in function of the receptor, and gene amplification. Androgen-deprivation therapy introduces new selection criteria for clones that express different AR phenotypes and modifies the AR status of the cancer cells.

5.6.6 Other cancer-related genes

Several other prostatic cancer-related genes have been identified that are useful for the diagnosis and/or prognosis of prostate cancer, including HEPSIN, racemase, and enhancer of zeste homolog 2 (EZH2). Expression of HEPSIN, which is a type II transmembrane serine protease in prostate cancer, has been highlighted by several studies.1448, 1449 HEPSIN not only shows increased expression in prostate cancer compared with benign prostatic tissue, but it also is expressed highly in poorly differentiated carcinoma. Overexpression of HEPSIN was correlated significantly with metastatic, hormone-refractory prostate cancer.1450 Racemase (P504S) is a prostate cancer-related gene that encodes a protein involved in the β-oxidation of branched-chain fatty acids.1451 Immunohistochemical detection of the P504S gene product is a sensitive and specific marker of prostatic carcinoma in formalin fixed, paraffin embedded tissues, including those treated by hormones and radiation.1452–1455

The polycomb group protein EZH2 is overexpressed in hormone-refractory, metastatic prostate cancer.1456 The high expression level of the Met receptor tyrosine kinase in bone metastasis renders Met a promising target for nuclear imaging and treatment of metastatic prostate cancer.1457 Differential expression of the mismatch-repair gene hMSH2 in malignant prostate tissue has been associated with cancer recurrence.1458 The genes encoding the specific granule protein (SGP28), low-density lipoprotein (LDL)-phospholipase A2, and the antiapoptotic gene PYCR1 reportedly were overexpressed in prostate cancer.943, 1094, 1459–1462 The radiosensitivity gene ATDC and the genes that encode the DNA-binding protein inhibitor ID1 and the phospholipase inhibitor uteroglobin were down-regulated significantly in cancer samples.1459

A prostatic tumor-inducing gene 1 (PTI-1), which was identified by differential display, was cloned from a cyclic DNA library of the LNCaP prostate cancer cell line.1463 The PTI-1 gene is homologous to a truncated and mutated form of human elongation factor 1 α,1464 suggesting a role in protein translation.1463 The PTI-1 gene is not expressed in normal prostate or BPH,1463 but it is expressed in prostate cancer,1463 in cancer cell lines,1463, 1465 and in sera from cancer patients.1465 Lipoprotein (LDL)-phospholipase A2, and the antiapoptotic gene PYCR1 reportedly were overexpressed in prostate cancer.943, 1094, 1459–1462 The radiosensitivity gene ATDC and the genes that encode the DNA-binding protein inhibitor ID1 and the phospholipase inhibitor uteroglobin were down-regulated significantly in cancer samples.1459

5.7 Comparative Genomic Hybridization

Comparative genomic hybridization (CGH) is a molecular cytogenetic technique used to screen cancer DNA alterations. The significant advantage of CGH is that all chromosome regions can be screened for gains and losses at one time.1287, 1298 However, CGH analyses are limited by the relatively low resolution (the method can only detect deletions on the order of 10 MB in size). In addition, CGH data interpretation is difficult and requires relatively expensive microscope and computer equipment.1298, 1299, 1301, 1302 Applying PCR, FISH, and CGH to the same specimens can be useful to clarify the discrepancies observed with a single technique alone.1284, 1285, 1287, 1298, 1299, 1301, 1302, 1431

5.8 Data Gaps: Biomarkers

To date, many studies of individual biomarkers in experimental animal models and human populations have been undertaken. Biomarkers are of a number of classes, including indicators of exposure or dose, indicators of effect (such as altered structure, function, or disease), and indicators of possible susceptibility to early detection of disease. The major challenge in the development of biomarkers is to determine whether there is a predictive link between a biomarker and a disease. The validation of any biomarker-effect link requires parallel experimental and human epidemiologic studies. The emerging technologies and their rapid development in genomics and epidemiology likely will lead to more precise exposure and effects assessments as well as to the development of effective intervention strategies.

One issue in the development of biomarker studies is a common lack of rigorous statistical methods. There are only scattered multivariate studies, often with different clinical or pathologic endpoints, and most reports are performed in isolation of other potentially useful biomarkers. Statistical requirements for adequate study include sufficiently large and homogeneous populations that are free of selection bias and ascertainment bias. Clinical trials provide many of the patient cohorts for such studies.

A second issue is in regard to establishing a linkage between the biomarker and the outcome. Is the biomarker an effect of cancer or causally related? This inference needs exploration for each biomarker under study. For example, it is well recognized that the utility of surrogate endpoint biomarkers of risk, such as PSA, has not been demonstrated.

A third issue is the use of biomarkers examined in late stages of cancer and their applicability for early prostate cancer and risk. Cancers acquire greater genetic instability as they grow, so the number of biomarker changes increases, creating higher “signal-to-noise,” which may mask the association of cancer and risk.

A fourth issue is the confounding of linked factors. It is recommended that genetic phenomena and epiphenomena be studied together to avoid this problem.

A fifth data gap is the need for tissue banks with well characterized human prostate cancer specimens (tissue, serum, and other fluids) matched with clinical follow-up. Currently, there are initiatives by the National Cancer Institute to address this concern.

6.0 MECHANISMS OF ACTION

6.1 Genotoxic

Although specific chemical carcinogens, such as MNU plus testosterone, induce prostate cancer in some strains of rats and mice,632 currently, there is no strong evidence that chemical carcinogens play a role in the induction or progression of human prostate cancer. Prostate cancer in humans arises as a function of age. The possibility exists that small amounts of dietary carcinogens, such as PhiP in cooked fish and meat, induce cancer-causing mutations in prostate tissue over a lifetime.611 PhiP is mutagenic, forms DNA adducts, and is a carcinogen in the rat prostate. If this finding is confirmed in humans, then it would suggest that consumption of foods containing PhiP over a lifetime could result in the consumption of a substantial prostate carcinogen.

Another possibility is that continuous cell division, driven by hormones such as testosterone, leads to spontaneous mutations during cell division, resulting in activation of oncogenes and inactivation of tumor suppressor genes. PhiP also may function jointly as a carcinogen and a tumor initiator, and androgen stimulation functions as a mitogen and tumor promoter. Molecular cloning of activated oncogenes from human prostate cancer and sequencing mutations may provide information on whether specific mutational spectra occur in specific genes and whether this can be related to mutations induced by specific mutagenic carcinogens.

6.2 Nongenotoxic

Testosterone plays a significant role in the development and progression of human prostate cancer by acting as a stimulus for prostate cell growth. It may function as a mitogen or a tumor promoter. Testosterone stimulates cell division; and, over a lifetime, the large number of cell divisions may lead to spontaneous mutations in prostate cells. Cancer may arise from the division of cells with spontaneous mutations in protooncogenes, activating them to oncogenes, and from further spontaneous mutations in tumor suppressor genes, inactivating them, in an iterative fashion. Spontaneous genetic variability, evolution of the cancer genotype and phenotype, and selection for specific cancer cells capable of surviving in the host may lead to disease progression. This is a parsimonious and likely hypothesis, similar to the hypotheses proposed for the development and progression of breast cancer. More data are needed on isolation and sequencing genes that are altered in human prostate cancer before any conclusions can be drawn regarding whether the key mutations in human prostate cancers are spontaneous or chemically induced.

6.3 Research Needs

In assessing the currently available data, it is clear that the exact mechanisms of prostate cancer development and progression in humans are unknown. It is likely that the process of aging allows spontaneous mutations to occur in prostate cells as a result of errors in DNA replication. This, on its own, may result in some fraction of human prostate tumors. The fraction of human prostate tumors that are due to spontaneous tumors, of course, is unknown. Second, it also is possible that mutagens in cooked food, such as PhiP and similar mutagens and carcinogens produced during the cooking of foods, are responsible for the induction of mutations in human prostate cells and can lead, in part or in whole, to carcinogenesis in human prostate and to a certain fraction of human prostate cancers. Finally, it is likely that the influence of testosterone on prostate cells bearing spontaneous mutations or mutations induced by carcinogens, such as PhiP, serves to induce tumor promotion and to provoke the final tumors. These three hypotheses cover three essential data gaps, and all need to be tested separately with regard to human prostate cancer.

Thus, research needs to include studying a large number of human prostate cancers to determine which specific protooncogenes are activated to oncogenes and the specific mutations that cause this activation. Next, these tumors could be studied to determine which specific tumor suppressor genes are inactivated and to determine the inactivating mutations. From such a data base, mutational spectra for frequently activated oncogenes and frequently inactivated tumor suppressor genes in human prostate tumors could be developed. These data could be used to determine whether the mutations that occur are due to spontaneous mutations, such as those that occur during DNA replication, or whether they are due to known mutagens, such as PhiP or related food-derived mutagens. Finally, it may be possible to determine from these mutational spectra whether some of these mutations are due to novel mutagens.

There is a lack of information regarding the molecular mechanisms of neoplastic transformation of human prostate cells. Research needs in this area are derivation of a convenient, reproducible model cell culture system of human prostate epithelial cells followed by determination of whether these cells could be induced to undergo neoplastic transformation upon treatment with PhiP, PhiP plus testosterone, testosterone alone, or spontaneously. Finally, determining the molecular mechanisms of this neoplastic transformation at the level of oncogene activation and tumor suppressor gene inactivation would provide important information regarding the process of carcinogenesis.

It is becoming increasingly apparent that some growth factors may induce receptor cross-talk in tumor cells. The extent to which receptor cross-talk may potentiate tumor progression is a data gap that needs to be addressed by further studies. This is particularly relevant in cases in which extracellular stimuli with previously characterized effects may have unforeseen consequences in the progression of cancer. Many of the molecular mechanisms of neoplastic cell transformation that lead to aberrant growth factor/receptor expression are understood incompletely. This link between changes in the expression of growth factors and/or their receptors in tumor cells and the effect these changes have on the phenotype of prostate tumor cells is an important area for future research.

It is necessary to determine more concerning the cross-species comparisons for prostate carcinogenesis with the human. Research needs include human prostate cells and human prostate organ-culture studies. The data obtained from the studies could then be applied back into the animal models to focus the animal studies more clearly on specific carcinogens, tumor promoters, or combinations to mimic human prostate cancer as closely as possible.

Most, if not all, of the animal models of prostate cancer rely on a limited number of mutational events that induce the transformation process. Thus, some of these animal models exhibit a limited number of transformation events in common with the pathologic progression of human prostate cancer, except for the production of a prostate cancer as an endpoint. Animal models that more closely mimic the malignant progression of human prostate cancer by undergoing an accumulation of alterations in growth factor receptors, oncogenes, and tumor suppressor genes that have demonstrated relevance in human prostate cancer need to be identified and/or developed.

Crucial questions that future research could address include the following: 1) How strong is the role of testosterone in induction of human prostate cancer? 2) Is testosterone a cocarcinogen, or a tumor promoter, or merely a growth stimulant to cells that bear multiple mutations, either spontaneous or induced? 3) Is PhiP an important mutagen, tumor initiator, or complete carcinogen for human prostate? 4) Are there other carcinogens in the diet or the environment that can induce or contribute to induction of human prostate cancer? 5) Can knowledge gained from the human studies listed above be translated back into animal model systems or human prostate cell culture systems to gain more insight into the molecular mechanisms of induction of human prostate cancer?

Finally, information is needed with regard to whether other hormones, such as estrogens, may affect prostatic carcinogenesis. Natural estrogens, through catechol estrogen formation and redox cycling, produce DNA-reactive molecules, which, in turn, cause DNA damage. It has been shown that these events occur in breast cancer. A key question that remains to be addressed is whether similar activation pathways exist in the prostate. Furthermore, estrogens, when supported by androgens, are potent mitogens for prostatic epithelial cells. Hence, they may play a tumor-promoting role. Information on whether circulating and/or tissue estrogen levels influence prostate cancer risk is needed. The relation between mitogenic response in the prostate and tissue sensitivity, as related to receptor type and levels and their coactivator/repressors, is an important area for future research. In addition, environmental or pharmaceutical estrogens, such as DES, bisphenol A, methoxychlor, dioxin, and certain polychlorinated biphenyls, also may exert direct or indirect genotoxic/nongenotoxic actions on the prostate.

Acknowledgements

The authors gratefully acknowledge the efforts of Thomas Crisp, Ph.D., formerly with the U.S. EPA, for initiating the project and serving as a Technical Monitor for the original contract. They thank EPA scientists Karl Baetcke, Ralph Cooper, Joyce Donahue, Richard Hill, Jennifer Seed, Peter Grevatt, and Jerome Puskin for their critical review of the original draft.