Epidemiologic and screening studies performed in the past several decades have raised several important questions about the pathogenesis of this disease, but a definitive cause for prostate cancer has not been established. It has been apparent for several years that the age-adjusted incidence rates as well as death rates from clinical prostate cancer vary dramatically from country to country, even if one allows for differences in and availability of screening programs Figure 1 [5, 6]. For example, Waterhouse and Muir [5] found a 25-fold difference between incidence rates in American black men living in San Francisco and in Japanese men. In 1988, the age-adjusted death rates per 100 000 population were 15.7 for men in the United States and 3.5 for men in Japan [7].

View larger version (22K): [in this window] [in a new window]   Figure 1. Age-adjusted death rates due to prostate cancer per 100 000 population. (From Boring and colleagues [7].).

The clinical effect of prostate cancer is growing. The advent of newer screening procedures such as the prostate-specific antigen blood test and growing public awareness have raised questions concerning the cause of prostate cancer as well as ways to screen for and prevent this disease [8-12]. The ultimate goal of epidemiologic studies is to identify risk factors to guide disease prevention strategies. This review examines the current data on the identification of potential risk factors that may be important in the development of prostate cancer.

Clinical Compared with Histologic Prostate Cancer

The prevalence of both clinical and histologic prostate cancer increases with age. After age 50 years, both incidence and mortality rates from prostate cancer increase at a nearly exponential rate. Prostate cancer increases faster with age than any other major cancer and, with an aging population, the burden of illness from prostate cancer will probably continue to increase in the future [13]. It remains unclear why this cancer increases with age so much more than do other cancers. Although the clinical incidence of prostate cancer varies greatly around the world, the prevalence of histologic cancer is remarkably similar. Autopsy study data from many countries demonstrate that 15% to 30% of men older than 50 years have histologic evidence of prostate cancer [14-37]. The presence of histologic cancer increases with age so that by age 80 approximately 60% to 70% of men have evidence of histologic carcinoma at autopsy [14-37].

It is now generally accepted that the development of a fully malignant cancer cell requires multiple malignant genetic events, including those that initiate cell transformation as well as those that promote or encourage the transformation process [38-40]. If histologic cancer represents a step in the development of clinically evident prostate cancer, then the data suggest that the initiation event of prostate cancer appears to occur at approximately the same rate independent of race or place of birth of the individual [1]. Carter and colleagues [1] have shown that although the age-specific prevalence of histologic prostate cancer is similar in Japan and the United States, there is a marked difference in the age-specific prevalence of clinical prostate cancer between Japanese and American men (Figure 2). These data suggest that the initiation rate of prostate cancer may be the same in both groups but that there appear to be differences in the rate of promotion or progression to clinically evident prostate cancer. This interpretation is further supported by the observation that immigrants who move from low-risk areas to the United States gradually assume the higher risk of the U.S. population [3, 41-46]. Therefore, whereas the presence of histologic cancer appears to be related to age, other risk factors that increase the development of prostate cancer probably affect the "promotion" steps of the transformation pathway.

View larger version (15K): [in this window] [in a new window]   Figure 2. Comparison of the age-specific prevalence of prostate cancer in the United States and Japan. Top panel. Histologic prostate cancer. Bottom panel. Clinical prostate cancer. (From Carter and colleagues [1].).

Risk Factors

Family History

Several studies have suggested that the incidence of prostate cancer in male relatives of patients with prostate cancer is increased [47-59]. Thiessen [47] has reported a higher incidence of prostatic cancer among male relatives of patients with breast cancer. Cannon and colleagues [48] found a familial clustering of prostate cancer in Utah Mormons. Their analysis demonstrated a high kinship coefficient for patients younger than 65 years with their brothers younger than 65 (age-specific relative odds, 5.97). Spitz and colleagues [49] showed an increased risk among men with first-degree relatives with the disease (odds ratio, 2.41). Carter and colleagues [50-53] have done a series of analyses that show that men with a father or brother with prostate cancer are twice as likely to develop prostate cancer as men without affected relatives and that risk increases with an increasing number of affected relatives. These authors have gone on to show through segregation and linkage analyses that early-onset prostate cancer may be inherited in an autosomal-dominant fashion of a rare high-risk allele and suggest that the autosomal-dominant form of prostate cancer accounts for a significant number of early-onset cases. Overall, however, these cases only represent a small proportion of prostate cancer.

Together these data show an increased risk for the development of clinical prostate cancer in men with affected relatives. It remains unclear if the current National Cancer Institute prostate cancer screening guideline of a digital rectal examination yearly after age 40 should be altered for men with affected family members. The American Cancer Society screening guidelines for prostate cancer consist of a yearly digital rectal examination and a prostate-specific antigen test starting at age 50. It is suggested that men with a positive family history of prostate cancer initiate screening at an earlier age, but a specific age is not recommended at this time.

Race

Wide variation in the reported incidence of clinical prostate cancer has been reported between different ethnic groups. It is clear from both incidence and mortality data that the incidence of clinical prostate cancer is low in Asian men and higher in Scandinavian men [7]. Incidence rates vary 120-fold between Chinese men and African-American men living in San Francisco [5, 60]. It is unclear if this is based on genetic or environmental factors; however, migration studies demonstrate that men tend to take on the risk for their host countries [3, 40-45]. These data, unfortunately, are confounded by differences in life expectancy, diet, socioeconomic status, and reporting. African-American men have a higher incidence of prostate cancer than do black men in Africa or Asia [3, 5].

A recent study from the National Cancer Institute based on data extracted from the Surveillance, Epidemiology, and End Results Program (SEER) as well as census data reveal that African-American men living in the United States have a higher incidence rate of clinical prostate cancer than do white men of similar education and socioeconomic classes [61]. The age-adjusted invasive prostate cancer incidence rates in the tri-county metropolitan Detroit area, as measured by the SEER database in 1990, showed a 30% greater incidence among African-American men compared with white men (Figure 3). African-American men have a higher incidence of prostate cancer at all ages (Figure 4). Furthermore, African-American men are routinely diagnosed with later-stage disease, and survival rates are uniformly shorter for African-American men, even when corrected for stage. The 5-year survival rates for all stages of prostate cancer are 62% for African-American men and 72% for white men.

View larger version (15K): [in this window] [in a new window]   Figure 3. Prostate cancer incidence rates for the years 1973 to 1990. Rates are shown for the tricounty metropolitan Detroit area as measured by the Surveillance, Epidemiology, and End Results (SEER) program. Rates are per 100000, are age-adjusted to the 1970 U.S. standard population, and are representative of new cancer cases diagnosed in a given year.

View larger version (13K): [in this window] [in a new window]   Figure 4. Incidence rates by age and race for invasive prostate cancer. Rates are shown for metropolitan Detroit tricounty residents in 1989 as reported by the SEER data base. Rates are per 100000 and are age adjusted to the 1970 U.S. standard population.

If the prevalence of histologic prostate cancer is essentially the same in different racial populations [13-36], these data suggest that African-American men are either more susceptible to prostate cancer-promoting events or are exposed to different promoting agents. Interestingly, Whittemore and colleagues [62] note that the transformation rate from histologic to clinically evident cancer is similar for African-American and white men, but that African-American men appear to have a larger volume of latent prostate cancer. These investigators believe that the larger-volume latent cancers are the carcinomas that progress more rapidly to become clinically evident. These data suggest that events that account for racial differences in prostate cancer incidence may occur early in cell transformation [37-39].

Socioeconomic Factors

Differences in socioeconomic status, especially those between African-American and white men, have been suggested as a reason for the differences in prostate cancer incidence between these two groups. This issue has been especially difficult to resolve given that minority populations in many studies come from low socioeconomic groups. Baquet and colleagues [61] at the National Cancer Institute, however, recently investigated the incidence of prostate cancer in African-Americans and whites by correlating cancer incidence with population density, education, and income level. Using incidence data from SEER, these authors found that incidence was generally higher in African-American men than in white men but that no statistically significant association exists between socioeconomic status and prostate cancer incidence [61]. These findings are corroborated by Ernster and colleagues [63], who used data from the Third National Cancer Survey (1969-1971 in Alameda county) and found that prostate cancer incidence was not associated with socioeconomic status in African-American or white men, as did McWhorter and colleagues [64] using the SEER database from 1978-1982.

Previously, several studies had examined the effects of socioeconomic status with both positive and negative results [65-74]. In a study related to socioeconomics, Mishina and colleagues [73] found no relationship between educational level and prostate cancer incidence and, in general, the data support the concept that socioeconomic status is not an important risk factor for the development of prostate cancer.

Occupation

Several studies examining the risk for prostate cancer and occupation and physical activity have had mixed results [78-90]. Checkoway and coworkers [75], in a case–control study, found that 75% of 40 patients with prostate cancer had a history of farming compared with 37.5% of control men with benign prostatic hyperplasia. Other studies, however, have not shown significant differences between urban versus rural dwellers [76, 77]. Other industries or occupations that have been associated with a higher incidence of prostate cancer include mechanics, newspaper workers, plumbers, and men in rubber manufacturing industries, but these reports have not been confirmed conclusively [78-83]. Industries that have received the closest scrutiny as increasing the risk for prostate cancer have been those in which workers are exposed to cadmium.

Cadmium Exposure

Cadmium is a trace mineral found in cigarette smoke and alkaline batteries. People working in the welding and electroplating occupations are exposed to high levels of cadmium. Several studies, with varying results, have been done to determine whether cadmium exposure places an individual at greater risk for prostate cancer [84-89]. Most of these studies tend to support the hypothesis that cadmium exposure weakly increases the risk for prostate cancer. It has been suggested that cadmium increases the risk for prostate cancer by interacting with zinc. Zinc is a necessary trace element in multiple intracellular metabolic pathways, and the prostate contains high amounts of zinc [90]. Several enzymes that are involved in the replication and repair of DNA and RNA, such as the polymerases, require zinc to function properly [90]. The prostate has the highest concentration of zinc of any organ in the body. Prostate glands containing cancer have lower levels of zinc than do noncancerous glands; however, it remains unclear whether zinc is associated with prostate cancer [91-95].

Smoking

A number of studies have found that cigarette smoking may be a risk factor for the development of prostate cancer [96-100]. Hsing and colleagues [96] demonstrated relative risks of 1.8 and 2.1 for cigarette smoking and chewing tobacco, respectively. A case–control study of 382 men by Fincham and colleagues [97], however, did not find a link between smoking and prostate cancer. One study proposes a higher risk in smokers with increased exposure to cadmium [98]. The data for cigarette smoking risk are also complicated by conflicting reports on the effect of cigarette smoking on serum sex hormones [99, 100]. It appears that cigarette smoking adds little, if anything, to the risk for developing prostate cancer.

Infectious Agents

No definitive links between prostate cancer and viruses have been established [75, 101-106]. In a small study, Schuman and colleagues [59] found that men with prostate cancer had higher titers of herpesvirus and cytomegalovirus than did a control population. Centifanto and colleagues [101] reported the presence of herpesvirus particles in the prostate cancer cells of one patient as demonstrated by immunofluorescence staining. To date no studies have been reported using more sensitive techniques such as the polymerase chain reaction. Three studies have not shown an association between sexually transmitted diseases and prostate cancer development, whereas three others have suggested a possible association [48, 75, 102-106]. The relationship between the risk for developing prostate cancer and a history of sexually transmitted disease remains unclear but appears tenuous.

Sexual Behavior

Several investigators have studied the history of men with prostate cancer in an effort to determine if sexual behavior and/or fertility are related to the development of prostate cancer [48, 65, 73, 75, 76, 99, 107-113]. Interpreting these studies is often difficult because marriage is often assumed to be synonymous with increased sexual activity, and fertility is often measured by the number of children a man admits to having fathered. It remains unclear if sexual activity has any relationship to prostate cancer risk.

Vasectomy

It has been suggested that vasectomy may increase the risk for prostate cancer because of the observation that vasectomized men have higher levels of circulating testosterone [98]. The largest study of vasectomized men was performed by Sidney [114], who followed a cohort of 5332 men, each matched with three nonvasectomized comparison controls. Neither the risk for prostate cancer nor benign prostatic hypertrophy was increased by vasectomy; however, only 68 cases of prostate cancer were observed overall. Several other smaller studies have revealed a small increased prostate cancer risk among vasectomized men [98, 115, 116]. Mettlin and colleagues [116], for example, reported a relative risk of 1.7 for reporting a vasectomy at any age and a relative risk of 2.2 for men reporting vasectomy 13 to 18 years before being diagnosed with cancer. This trend in the association of years since vasectomy and risk was also reported by Honda and colleagues [98]. Two large studies by Giovannucci and colleagues [117, 118] have recently been completed that confirm this positive trend between the number of years since vasectomy and prostate cancer risk. The first, a prospective study involving 10 055 vasectomized men and 37 800 nonvasectomized men, showed a relative risk of 1.85 for men reporting vasectomy [117]. The second, a retrospective cohort study that matched 14 607 vasectomized men with 14 607 controls, reported an age-adjusted relative risk of 1.56 associated with vasectomy. This relative risk for prostate cancer increased over time since vasectomy, with men who had a vasectomy 20 or more years in the past demonstrating a relative risk of 1.89. Vasectomy appears to confer an increased risk for the development of prostate cancer.

Benign Prostatic Hyperplasia

It is difficult to determine whether benign prostatic hyperplasia is a risk factor for prostate cancer because both are common diseases in men as they age. Although both diseases appear to be androgen dependent for growth, benign prostatic hyperplasia arises most often in the central or transitional zone of the prostate, and cancer is most often found in the peripheral zone of the gland. Nevertheless, Mishina and colleagues [73] found that history of benign prostatic hyperplasia carried a relative risk of 13.5. Armenian and colleagues [119] found a relative risk of 5.1 for prostate cancer in patients identified earlier as having benign prostatic hyperplasia; however, not all cases in this study were histologically confirmed. In both retrospective and prospective studies, these authors found a higher death rate from prostate cancer in men with a history of benign prostatic hyperplasia; however, the reasons for these findings remain unclear [119]. Kolonel and coworkers [107] found that for men younger than 70 years, more men with prostate cancer had a history of benign prostatic hyperplasia than did the control population. Greenwald and coworkers [120], however, found no association between benign prostatic hyperplasia and prostate cancer, and it remains unclear whether this entity is associated with a higher risk for prostate cancer.

Dietary Fat

Epidemiologic research and migrant studies have associated diet with prostate cancer [5, 40-45]. This may explain why histologic cancers occur at a similar frequency around the world but clinical incidence is higher in "Western" societies [1, 13-35, 41]. The idea that the risk for prostate cancer is increased by high intake of fat has been the subject of many studies and much heated debate; however, dietary fat appears to stand out as an important risk factor [121-127].

Hutchinson [128] hypothesized that dietary patterns could alter the production of sexual hormones and affect the risk for cancer within the prostate gland. This hypothesis has implications not only on the intake of dietary fat but also on the fat-soluble vitamins, such as vitamin A, as well as associated trace nutrients, such as zinc. Studies have revealed differences between the fat content of diets in high- and low-risk areas [13, 129-131]. The diet of Japanese men, for example, has much lower fat content than that of U.S. men [131]. Armstrong and Doll [130] found that prostate cancer deaths from 32 countries were highly correlated with total fat consumption, a finding similar to that for breast cancer. Rose and colleagues [131] confirmed these earlier data and determined that this relationship was limited to animal fat and not vegetable fat intake (Figures 5 and 6). This finding is supported by several studies that appear to show that men with diets high in fiber have lower incidences of prostate cancer [126, 132-137]. This finding is also supported by the data of Hirayama [138], who in his prospective study of 122 261 men found a lower mortality rate from prostate cancer in those men who consumed green and yellow vegetables each day. Several other studies have confirmed this relationship between high dietary fat and increased prostate cancer risk [59, 107, 132-137, 139]. Other studies, including a large cohort study by Severson and colleagues [127], have not demonstrated an association between dietary fat and prostate cancer development [140]. Overall, a high intake of dietary fat seems to be associated with a higher risk for developing prostate cancer, with relative risks reported of 1.6 to 1.9 and odds ratios reported as high as 3.6 for high unsaturated fat intake [104, 122]. How dietary fat is related to this higher risk is unclear.

View larger version (21K): [in this window] [in a new window]   Figure 5. The positive correlation between breast and prostate cancer by country. The incidence of both breast and prostate cancer is higher in more developed countries. These data suggest that environmental influences may be important in the promotion of these diseases (see text). (From Rose and colleagues [131].).

View larger version (22K): [in this window] [in a new window]   Figure 6. The positive correlation between the mortality rate for prostate cancer and animal fat consumption by country. These data suggest that high fat intake may be related to the risk for developing prostate cancer (see text). (From Rose and colleagues [131].).

Other Dietary Factors

The relationship between other dietary factors and prostate cancer is complex. It is very difficult to separate the effect of a given nutrient from other parts of the diet and to identify an association with a given cancer [121, 126, 127, 131, 137, 139-143]. Mills and colleagues [137], in their cohort study of 14 000 Adventist men, found that increasing consumption of beans, lentils, peas, tomatoes, raisins, dates, and dried fruit significantly decreased risk for prostate cancer. Dietary factors, including vitamin C, vitamin B1, vitamin B2, selenium, zinc, protein, carbohydrate, and fiber have not been shown to be associated with prostate cancer risk [139-143]. Ohno and coworkers [140] found no association between prostate cancer risk and consumption of protein; carbohydrate; water; fiber; ash; retinol; vitamins B1, B2, C, and niacin; as well as calcium, potassium, sodium, phosphorus, and iron. In general, the results from dietary intake studies support the concept that a high-fiber, low-fat diet may protect men against the development of prostate cancer.

Vitamin A

Vitamin A or retinol is a fat-soluble vitamin that is essential for normal differentiation of epithelial cells, physiologic growth, visual function, and reproduction [144, 145]. Vitamin A deficiency has been related to the development of several different tumors in experimental model systems; supplementary retinoids have decreased the number of experimentally induced prostate cancers in animal studies [144]. Several studies have reported an increased risk for prostate cancer with increased vitamin A intake [107, 119, 121, 137, 139, 140], whereas other studies have contradicted these results [138, 140]. These findings may be explained by the observation that vitamin A as found in plants (ß-carotene) reduces risk, whereas vitamin A intake from animal sources increases risk [136, 147]. In Japan and other low-risk areas, the primary source of vitamin A is from vegetables; in high-risk areas such as the United States, however, the major source of vitamin A appears to be from animal fat. Therefore, the risk for prostate cancer associated with vitamin A intake may actually reflect the higher risk associated with high animal fat intake.

Hormones

The effect of dietary fat may be mediated through endogenous hormones. The interaction of steroid hormones with the development of prostate cancer is poorly understood; however, a low-fat, high-fiber diet has been shown to affect male sex hormone metabolism by decreasing circulating testosterone [148-150]. Testosterone is necessary for normal prostate epithelium to grow, and early prostate cancer has been shown to be endocrine dependent [151]. Noble [152] induced prostatic adenocarcinoma in rats by the prolonged administration of testosterone, and the ablation of androgens has formed the basis for first-line therapy of metastatic prostate cancer [153]. It has been suggested that altered hormone metabolism may play a role in the progression of prostate cancer from histologic to clinically significant forms, and it has been noted that the incidence of prostate cancer is very low in eunuchs and castrated men [154-156]. However, higher circulating levels of testosterone in patients with prostate cancer have not been consistently observed [58, 157-159]. Other hormones, especially prolactin and estrogen, may play an undefined role in prostate metabolism [160-162]. For example, patients with cirrhosis of the liver have low plasma testosterone levels but also have hyperestrogenism [163, 164]. Estrogens are also a successful form of hormone therapy for metastatic prostate cancer [151].

Consistent alterations in plasma androgens and estrogens in patients with prostate cancer have not been documented, although plasma levels of these hormones do not necessarily reflect intraprostatic hormone levels [155, 165-173]. Ross and colleagues [171], however, demonstrated that young African-American men had serum testosterone levels that were approximately 15% higher than their white counterparts and have suggested that this difference is enough to explain the increased risk for prostate cancer in African-American men. In a related study comparing American and Japanese men, these investigators found that American men had higher levels of sex hormone binding globulin than did their Japanese counterparts [172]. Similarly, Meikle and colleagues [173] demonstrated a higher sex hormone binding globulin level in men with prostate cancer compared with controls and also noted a higher testosterone conversion rate in patients with prostate cancer. Furthermore, both African-American and white men in the United States had higher levels of 3-,17-ß-androstenediol glucuronide and androsterone glucuronide, two androgens that are indices of 5--reductase activity, the enzyme in the prostate that converts testosterone to its active metabolite, dihydrotestosterone. Because dihydrotestosterone is responsible for the mitotic activity of the prostate, this lower enzyme level may contribute to the lower incidence of clinical prostate cancer observed in Japanese men [172].

Hill and colleagues [154] measured urinary steroid levels of North American African-American and white men compared with those of black South African men and discovered lower levels of urinary androsterone and testosterone in the black South African men. African-American and white men in North America, however, had similar urinary androgen levels. These urinary hormone levels were found to be diet dependent. Black South African men fed a Western diet demonstrated increased excretion of both androgens and estrogens, whereas the opposite trend was true for black North American men fed a vegetarian diet. These cross-national data suggest that dietary factors may modify testicular hormone activity [46, 154]. Hill and Wynder [167] have demonstrated that a vegetarian diet decreased plasma testosterone levels. In support of these data, Howie and Schultz [174] found that nonvegetarian men had higher levels of plasma estradiol and testosterone, but not 5--dihydrotestosterone, than did vegetarian men and suggested that these alterations in blood steroids may be related to the amount of fiber in the diets of these groups [174]. Goldin and colleagues [175] suggested that dietary fiber may influence the metabolism of plasma estrogens by altering the enterohepatic circulation. Following preliminary metabolism in the liver, these steroid metabolites are excreted into the bile and enter the intestinal tract where they can be resorbed and returned to the liver or excreted in the feces [174]. Vegetarians with high dietary fiber intake may excrete more of the sex steroids into the feces, thereby lowering the plasma levels of these hormones [174, 177]. This possibility was investigated by Ross and colleagues [177], who documented the higher fecal fiber content in vegetarian men compared with nonvegetarian controls. These fibers were shown to bind estrogens and testosterone in vitro [178]. This would explain, at least in part, how a high-fiber, low-fat diet could, theoretically, lower the risk for prostate cancer.

Discussion

Top Discussion Author & Article Info References

Prostate cancer, as most cancers, develops as the result of an interplay of genetic and epigenetic events, both of which may be affected by environmental risk factors [1, 37-39]. Understanding how these factors interact with the prostate will allow the investigation of potential sites of therapeutic intervention as well as present opportunities for prevention (Figure 7). This understanding is especially important in the prostate where, to date, the involvement of specific oncogenes in the development of human prostate cancer has not been clearly elucidated. Given the similar prevalence rates for histologic cancer at autopsy, it appears that although genetics may play a role in prostate cancer etiology, environmental factors appear to have a major role in prostate cancer biology, perhaps acting as prostate cancer promoters [1, 37-39].

View larger version (37K): [in this window] [in a new window]   Figure 7. Potential sites of intervention for prostate cancer treatment and prevention. A better understanding of the interplay of multiple hormones as well as other factors with the prostate will allow the potential design of new therapeutic strategies to intervene in the process of prostate carcinogenesis. Epidemiologic studies provide important clues in understanding the factors important in the development of prostate cancer. (Adapted from Coffey and Pienta [13]). ACTH = adrenocorticotropin releasing hormone; CNS = central nervous system; CRH= corticotropin releasing hormone; DHT = dihydrotestosterone; FSH = follicle stimulating hormone; FSH-RH = follicle stimulating hormone releasing hormone; LH = luteinizing hormone; LH-RH = luteinizing hormone releasing hormone; S = seminiferous; T = testosterone.

Several intriguing reports, especially those by Ross and colleagues [171, 172], suggest that the endogenous level of androgenic hormones, either testosterone, or dihydrotestosterone, may play a pivotal role in the cause of prostate cancer. The involvement of these hormones would explain, at least in part, why discerning risk factors for prostate cancer has been so difficult because androgen levels in an individual would result from the interplay of several different genes, including those for testosterone, 5--reductase, sex hormone binding globulin, and estrogen, as well as the environmental influence on these gene products, such as zinc intake, smoking, vitamin A, and dietary fat. Further epidemiologic studies targeted at better understanding how dietary fat and hormones interact at the level of the prostate may provide important clues to understanding the cause of prostate cancer.

Several large prostate screening studies are underway or are about to be undertaken in the United States, including the National Cancer Institute's Prostate, Lung, Colon, Ovary (PLCO) trial, in which thousands of men will be screened prospectively for prostate cancer. These efforts will lend valuable insight into the magnitude of the problem of prostate cancer as well as begin to clarify the relationship between histologic and clinically evident prostate cancer. These screening efforts, furthermore, will allow the identification of a large number of men to target for analysis of prostate cancer risk factors.