© 2000 by Oxford University Press
Journal of the National Cancer Institute Monographs, No. 27, 39-66,
2000
© 2000 Oxford University Press
Chapter 2: The Role of Steroid Hormones in Prostate Carcinogenesis
Correspondence to: Maarten C. Bosland, D.V.Sc., Ph.D., Departments of Environmental Medicine and Urology, New York University School of Medicine, 550 First Ave., New York, NY, 10016 (e-mail: maarten.bosland{at}med.nyu.edu).
 |
ABSTRACT |
|---|
 TopNotes Abstract IntroductionEepidemiologic Evidence for...Prostate Cancer and Steroid...Mechanisms of Hormonal Prostate...Overall ConclusionsFuture Research NeedsReferences |
|---|
Carcinoma of the prostate is the most frequently diagnosed malignancy and the second leading cause of death as a result of cancer in men in the United States and in many other Western countries. Notwithstanding the importance of this malignancy, little is understood about its causes. The epidemiology of prostate cancer strongly suggests that environmental factors, particularly diet and nutrition, are major determinants of risk for this disease, and evidence is mounting that there are important genetic risk factors for prostate cancer. Human prostate carcinomas are often androgen sensitive and react to hormonal therapy by temporary remission, followed by relapse to an androgen-insensitive state. These well-established features of prostate cancer strongly suggest that steroid hormones, particularly androgens, play a major role in human prostatic carcinogenesis, but the precise mechanisms by which androgens affect this process are unknown. In addition, the possible involvement of estrogenic hormones is not entirely clear. The purpose of this overview is to summarize the literature about steroid hormonal factors, androgens and estrogens, and prostate carcinogenesis. From these literature observations, a multifactorial general hypothesis of prostate carcinogenesis emerges with androgens as strong tumor promoters acting via androgen receptor-mediated mechanisms to enhance the carcinogenic activity of strong endogenous genotoxic carcinogens, such as reactive estrogen metabolites and estrogen- and prostatitis-generated reactive oxygen species and possible weak environmental carcinogens of unknown nature. In this hypothesis, all of these processes are modulated by a variety of environmental factors such as diet and by genetic determinants such as hereditary susceptibility and polymorphic genes that encode for steroid hormone receptors and enzymes involved in the metabolism and action of steroid hormones.
|
INTRODUCTION |
|---|
|
TopNotesAbstractIntroductionEepidemiologic Evidence for...Prostate Cancer and Steroid...Mechanisms of Hormonal Prostate...Overall ConclusionsFuture Research NeedsReferences |
|---|
Carcinoma of the prostate is the most frequently diagnosed malignancy and the second leading cause of death as a result of cancer in men in the United States and in many Western countries (not counting nonmelanoma skin cancer) (1). Notwithstanding the importance of this malignancy, little is understood about its causes. Steroid hormones, particularly androgens, are suspected to play a major role in human prostate carcinogenesis, but the precise mechanisms by which androgens affect this process and the possible involvement of estrogenic hormones are not clear. A causal relation between androgens and prostate cancer development is generally considered to be biologically very plausible because the vast majority of human prostate cancers are androgen sensitive and respond to hormonal therapy by temporary remission, later followed by relapse to a hormone-refractory state. The purpose of this overview is to summarize the literature about steroid hormonal factors and prostate carcinogenesis. Although the objective of this overview is not to be comprehensive, an attempt is made to be complete, especially where crucial aspects of this hormonal involvement are concerned.
In contrast, the prostate is a rare site of tumor development in carcinogenesis bioassays in rodents (2,3) and in aging male laboratory rodents, with the exception of ventral prostatic neoplasms in some rat strains (4–9). Prostate cancer is also rare in male farm and companion animals, with the notable exception of the dog, which is the only species besides man that develops this malignancy. As will be discussed in this overview, steroid hormones can induce and can substantially enhance prostate carcinoma development in rodents, and this phenomenon has been exploited to further our knowledge about the involvement of hormonal factors and mechanisms in prostate cancer etiology.
In this overview, the epidemiologic evidence for a role of steroid hormonal factors in prostate carcinogenesis is summarized first, followed by review of experimental data, a discussion of the possible mechanisms whereby steroid hormones, androgens as well as estrogens, may be involved in prostate cancer causation, and overall conclusions and suggestions for future research. As will be demonstrated, there is no lack of hypotheses about the role of steroid hormones in prostate cancer etiology, but the available data are often contradictory and incomplete, and an in-depth overall mechanistic understanding of how steroid hormonal factors are involved in prostate carcinogenesis is very limited.
|
EEPIDEMIOLOGIC EVIDENCE FOR INVOLVEMENT OF STEROID HORMONES |
|---|
|
TopNotesAbstractIntroductionEepidemiologic Evidence for...Prostate Cancer and Steroid...Mechanisms of Hormonal Prostate...Overall ConclusionsFuture Research NeedsReferences |
|---|
The epidemiology of prostatic cancer has been reviewed in depth elsewhere (10–15). Prostate cancer risk factors that are associated with hormonal factors are summarized in subsequent sections, and epidemiologic and other studies related to the metabolism and action of steroid hormones are reviewed in detail. Besides hormonal factors, there are only a few established risk factors for prostate cancer. These risk factors are briefly summarized below to put the relative importance of steroid hormonal factors in perspective.
Prostate cancer incidence and mortality rates have increased in the United States over the few decades preceding the frequent use of prostate-specific antigen (PSA) for early detection (10). Even though incidence rates have increased substantially since the mid-1980s because of the use of PSA "screening" for early detection (1), incidence has declined over the period from 1990 through 1996 by an average of 2% per year and mortality by 1.6% per year (16). Because of the increasing use of PSA for early detection and treatment in this period, it is too early to separately determine changes in prostate cancer rates and the impact of PSA screening on rates. In 1999, prostate cancer was the most frequent malignancy in U.S. males with 179 300 new cases expected, and it was the second most frequent cause of death as a result of cancer with 37 000 deaths expected (17).
Many studies have demonstrated that prostate cancer is more frequent in men with a family history of prostate cancer, as summarized elsewhere (10,11,15,18–20). This familial aggregation appears to be similar in African-American and in European-American men (21,22). However, inherited risk for prostate cancer can only explain a small proportion of prostate cancer cases, less than 10% (20). Besides a variety of genetic alterations identified with varying frequency in human prostate carcinomas, as summarized by Dong et al. (23), a few susceptibility loci linked to inherited prostate cancer risk have been identified on chromosome 1 (24–28) and on the X chromosome (29,30). Breast and prostate cancers cluster in some families, and there is some evidence that BRCA1 and BRCA2 mutations are involved in this clustering (18,31). However, none of these loci have thus far been associated with hormonal factors.
Evidence is limited that a history of venereal disease (11,32–34) and a history of prostatitis (34,35) are risk factors. An association between prostate cancer risk and the prior occurrence of benign prostatic hypertrophy (BPH) is biologically unlikely, even though steroid hormones are also implicated in BPH. Prostate cancer and BPH originate from different parts of the prostate (all BPH is found in the transition zone, and more than 80% of all cancers develop in the peripheral zone), and their epidemiology is dissimilar (11).
Although, in some studies (34,36,37), a relationship between smoking and risk for prostate cancer has been found, no such relationship has been observed in the vast majority of studies (37–40). In addition, smoking appears to have no effect on circulating levels of testosterone and other hormones that may be involved in prostate carcinogenesis (41,42). Most studies (11,43) addressing alcohol consumption as a potential risk factor for prostate cancer did not find evidence for an association. One notable exception is a study by Hayes et al. (44) that found a positive association in a U.S. case–control study, which was limited to heavy use of alcohol. Possible reasons for the association observed in this study are discussed by Lumey et al. (43). One of these reasons may be that prostate cancer risk is elevated in alcoholics with liver disease (11,43). This risk elevation is possibly related to the impaired clearance of estrogens described in men with liver cirrhosis (11,45,46).
Increased risk has been observed for a variety of occupations in studies of occupational factors and prostate cancer (11,14,47,48), including armed services personnel (11) and workers in the nuclear industry (11,15,48,49). Although prostate cancer risk in survivors of the atomic bomb in Japan appears not to be elevated (50), there is a rather strong international correlation between prostate cancer incidence and indoor radon levels reported (51). Thus, prostate cancer risk may be associated with exposure to ionizing radiation, but the evidence is equivocal. Associations between exposures and prostate cancer risk observed in the rubber industry are limited to one or a few plants (11). The evidence for a positive association between farming and prostate cancer risk is weak to inconclusive (11,14,47,48,52). There is only very weak, if any, evidence for an association of cadmium exposure and prostate cancer risk (11,53,54). Hormonal factors are most likely not involved in any of these (possible) associations between risk and occupational factors.
Risk Factors Associated With Possible Hormonal Mechanisms
The results of a variety of epidemiologic studies have led to suggestions for several risk factors that may be related to a hormonal mechanism. These risks include dietary factors, vasectomy, sexual factors, the level of physical activity, and obesity.
Diet and Nutrition The associations between dietary factors and prostate cancer risk have been extensively summarized elsewhere (10,11,15,55–57). Considerable consistency across studies indicates that a high intake of fat, particularly total fat and saturated fat, is a risk factor for prostate cancer, but the strength of the associations is modest at best (57,58) and may be greater for African-Americans than for European-Americans (59). Results from Hawaiian case–control studies suggest that as much as 25% of prostate cancer in the United States may be attributable to a high saturated-fat intake (60). However, Whittemore et al. (22) estimated that dietary fat intake may account for only 10%–15% of the difference in prostate cancer occurrence between European-Americans and African-Americans or Asians. The mechanism that could underlie an enhancing effect of fat on prostate carcinogenesis is not understood, but several hypotheses, including hormonal mediation, have been discussed elsewhere (11,15,57,61). In addition, a high intake of protein and energy and a low intake of dietary fiber and complex carbohydrates have been found to be associated with the increased risk for prostate cancer in some studies (10,11,15,55).
Associations with prostate cancer risk reported for individual nutrients or foods are not very strong. However, migration from low-risk areas, such as Japan, to high-risk countries, such as the United States, increases risk considerably (10,11). These changes in risk are thought to be due to differences in environment, including lifestyle and particularly dietary habits (10,11). It is, therefore, conceivable that the combined effects of dietary factors on prostate carcinogenesis are more important than the separate effects of any individual dietary factor (62). This idea is supported by the lack of any effect of dietary fat per se on the induction of prostate cancer in animal models, whereas epidemiologic studies (11,15,62) rather consistently show a positive association between prostate cancer risk and intake of dietary fat.
Older studies (63–67) of the effects on hormonal status of dietary changes and of the consumption of vegetarian or health food diets, which have been summarized previously (11), did not separately address the effects of dietary fat. However, they clearly indicate that diet can influence circulating hormone levels by changing androgen production rates and/or the metabolism and clearance of androgens and estrogens. In a study reported by Dorgan et al. (68), controlled changes in fat and fiber were applied to healthy men. The combination of a high-fat, low-fiber diet increased both total testosterone (by 13%) and testosterone bound to sex hormone-binding globulin (SHBG; by 15%) in the plasma as well as urinary testosterone excretion (13%), compared with a low-fat, high-fiber diet. However, urinary excretion of estrone, estradiol, and the 2-hydroxy metabolites of these estrogens was lower. All of these studies indicate that diet can affect steroid hormone status, but no studies have addressed the separate effects of single dietary factors, such as fat intake.
Complete consistency is lacking among epidemiologic studies of prostate cancer risk and intake of dietary vitamin A and 
-carotene (10,11,15,69). It is possible that retinoids and/or carotenes enhance rather than inhibit prostatic carcinogenesis under certain circumstances or in certain populations (69), although animal and in vitro studies suggest a protective effect of retinoids (11). In two more recent experiments on prostate cancer chemoprevention in a rat model, 9-cis-retinoic acid, a major retinol metabolite in mammalian species, strongly inhibited the induction of prostate cancer (70), but N-(4-hydroxyphenyl)retinamide (4-HPR), a synthetic retinoid, did not have any effect (71). 9-cis-Retinoic acid is unique in that it is a panagonist for retinoic acid receptors, binding both retinoic acid receptor (RAR) and retinoid X receptor (RXR) receptors. In vitro, however, both 9-cis-retinoic acid and 4-HPR inhibit the growth and induce apoptosis of the androgen-sensitive human prostate cancer LNCaP cell line, and so does all-trans-retinoic acid, which only binds to RAR (72–74). There are indications that 4-HPR acts via a nonreceptor mechanism (74). The specific mechanism is not known by which retinoids and/or carotenoids may inhibit or enhance prostate carcinogenesis, but inhibition seems biologically more plausible than enhancement, as discussed previously (11). The retinoic acid and androgen receptors both belong to the steroid receptor superfamily (75). This circumstance raises the intriguing possibility that retinoids may be able to bind to and activate mutated forms of the androgen receptor or that the retinoic acid RAR and/or RXR may activate transcription of androgen-regulated genes. Studies on the regulation by sex steroids and retinoic acid of glutathione S-transferase in hamster smooth muscle tumor cells (76) and on androgen-receptor gene expression in human breast cancer cells (77) suggest that such mechanisms may exist.
Vasectomy Vasectomy has been identified as a possible risk factor for prostate cancer in seven case–control studies (34,78–83) and in three cohort studies (84–86). The range of risk ratios in the case–control studies was 1.4 to 5.3. No elevation of risk for prostate cancer following vasectomy was found in six other case–control studies (87–92) and in two retrospective cohort studies (93–95). Although a meta-analysis (96) of 14 studies indicated that there is no causal relation between vasectomy and prostate cancer, further studies, particularly cohort studies, will be required to definitively establish whether or not vasectomy is a true risk factor for prostate cancer (58,97–99).
Three mechanisms by which vasectomy could enhance risk have been proposed: elevation of circulating androgens, immunologic mechanisms involving antisperm antibodies, and reduction of seminal fluid production (34,78,79,85,90,98,100). Most studies (101–105) that investigated the effect of vasectomy on pituitary–gonadal function did not find any effect, but some studies (90,100,106–110) found slight, but statistically significant, changes in circulating levels of certain hormones. Four groups (34,100,108,111) reported slightly elevated circulating testosterone levels, but only in two of these groups (100,108) was the increase statistically significant. Mo et al. (100) also found elevated levels of 5
-dihydrotestosterone (DHT), the active metabolite of testosterone in the prostate, in vasectomized men. John et al. (90) reported a decrease in SHBG, and Honda et al. (34) observed an increase in the ratio of testosterone to SHBG. These results suggest an elevation of circulating free testosterone following vasectomy, which may be a critical factor associated with risk for prostate cancer. A possible specific mechanism whereby vasectomy could influence the hypothalamic–pituitary–gonadal axis is not known.
Sexual Factors Attempts have been made in several case–control studies (11,32–34,112–114) to investigate the possibility that sexual factors play a role in prostate cancer etiology. The results of these studies suggest that prostate cancer risk may be associated with the level of sexual activity, but no conclusive evidence exists for such a relation (11). Tsitouras et al. (115) reported a significant positive association between the level of sexual activity (intercourse and masturbation) and circulating total testosterone levels in men between the ages of 60 and 79 years as well as an absence of a decrease in testosterone levels with aging in sexually active men. These findings suggest that a hormonal mechanism may underlie a possible association between prostate cancer risk and sexual activity suggested by the aforementioned case–control studies.
Physical Activity and Anthropometric Correlates of Risk Contradictory indications are found that the level of physical activity may be a possible risk factor for prostate cancer, but the evidence for such an association is inconclusive at present (15,62,116). Sports exercise may decrease, as well as increase, circulating androgen concentrations or have no effect, depending on such factors as the time of blood sampling in relation to the exercise, the level of exercise, and the training protocol followed (117,118). Therefore, it is possible that the type and extent of physical activity influence circulating androgen concentrations and, thereby, perhaps prostate cancer risk. At present evidence is contradictory that obesity or an increased body mass index is a risk factor for prostate cancer (15,62,119). Severson et al. (120) observed a significant increase in prostate cancer risk with increasing upper-arm circumference and upper-arm muscle area but not fat area. A positive association between prostate cancer risk and muscle mass, but not fat mass, may suggest exposure to endogenous or exogenous androgenic hormones or other anabolic factors (120,121). Indeed, evidence is available that body mass index is inversely associated with plasma testosterone and SHBG levels and positively associated with estradiol levels (119,122,123), as discussed elsewhere (11,42).
Epidemiologic Studies of Endogenous Hormones and Hormone Metabolism
As indicated earlier, a causal relation between androgens and prostate cancer development is generally considered biologically plausible because this malignancy develops in an androgen-dependent epithelium and is usually androgen sensitive. In addition, a few case reports (124–129) are available of prostate cancer in men who used androgenic steroids as anabolic agents or for medical purposes, suggestive of a causal relationship.
Studies (11,15,130) comparing the endocrine status of human prostate cancer patients with that of control subjects are probably not very informative about the endocrine status prior to the onset of the disease and are, therefore, not meaningful to explore this relationship; in addition, the presence of the malignancy may by itself alter hormonal status. Indeed, the results of such studies do not provide a consistent pattern as summarized by Andersson et al. (130), which is confirmed in other case–control studies (119,131–133). These types of studies will, therefore, not be discussed here.
Nested case–control studies of steroid hormonal factors in ongoing cohort studies, as well as studies comparing healthy males in populations that are at high risk for prostate cancer with populations at lower risk, are likely to be more meaningful. These studies are summarized in the following sections. The two major hypotheses for these studies were that increased risk for prostate cancer would be associated with either an increased testicular production of testosterone or an increased conversion of testosterone to DHT because of an increased 5-reductase activity (134–136). Studies have focused on the notion that functional genetic polymorphisms in the 5-reductase gene or in genes involved in testosterone biosynthesis (the CYP17 gene) or DHT catabolism (the 3-hydroxysteroid dehydrogenase gene) could be responsible for increased testosterone production or increased DHT levels (136). In addition, polymorphisms have been discovered in the androgen receptor gene that can have functional significance for androgen receptor activity (137–139). Such polymorphisms have been postulated to be critical determinants of prostate cancer risk at the population or individual level by affecting intraprostatic DHT concentrations and androgen receptor transactivation (18,136).
Steroid Hormonal Factors in Populations That Differ in Risk For Prostate Cancer
Circulating of steroid hormone levels.
A summary of the results of studies that compared circulating levels of steroid hormones in very high-risk African-Americans with those in high-risk European-Americans, lower-risk Asian-Americans, and very low-risk Asians living in Asia or African black men is provided in Table 1
. The details of each study are summarized in the following paragraphs.
|
Ahluwalia et al. (140) studied 170 African-Americans and 55 black-Nigerian men who were matched control subjects in a case–control study of prostate cancer and were older than 50 years. Plasma levels of testosterone and estrone were significantly higher in the African-American men than in the Nigerian men, whereas levels of DHT and estradiol were not different. Similar differences were found for the prostate cancer case patients.
Hill et al. (63–65) compared the hormonal status of small groups (n = 11–20) of 40- to 55-year-old African-American, European-American, and black (rural) South African men consuming their customary diets (the effects of diet changes were also studied in these men; see earlier section on "Diet and Nutrition"). In a separate study (141), African-American, European-American, and black South African boys (ages 12–15 years) and young African-American and black South African men (ages 18–21 years) were compared. In the older men, plasma levels of the testosterone processor dehydroepiandrosterone (DHEA) were significantly lower in the two groups of black men than in the white men, whereas estrone levels were higher. Plasma levels of the testosterone processor androstenedione and estradiol were significantly higher in the African men than in the two American groups, whereas no differences were noted among these groups in testosterone levels. In the study with the 12- to 15-year-old boys and young men, similar findings were obtained for testosterone and DHEA. However, androstenedione levels were significantly lower (not higher) in the African than in the American subjects, and estradiol was lower in young black boys (12–14 years old) than in white boys but higher in older black boys (12–14 years old) and young black men than in white boys and men. These data suggest a complex interaction between ethnic background and environmental differences that change over the years of sexual maturation. In these studies by Hill and colleagues among South African black men, the 18- to 21-year-old men were different from those in 40- to 55-year-old men for androstenedione and DHEA. This divergence suggests that it is probably important for the interpretation of hormonal profiles to separately consider younger and older men.
Ross et al. (134) compared 50 healthy young African-American men (at very high risk for prostate cancer) and 50 young European-American males (at half the risk of the black men). Total circulating testosterone was 19% higher, and free testosterone was 21% higher in the group of black subjects than in the group of white subjects. Serum estrone concentrations were also significantly higher (16%) in the black than in the white group. No significant differences were seen between the groups in circulating estradiol and SHBG levels. The authors estimated that the 19%–21% difference in circulating testosterone is sufficiently large to explain the twofold difference in prostate cancer risk between white and black men in the United States. This study suggests an association between prostate cancer risk and high concentrations of circulating androgens and, possibly, estrogens.
Henderson et al. (142) compared circulating hormone levels in 20 pregnant African-American with 20 European-American women in their first trimester. Serum testosterone levels were 47% higher in black women than in white women, and estradiol levels were 37% higher. No significant differences were observed in circulating SHBG and human chorionic gonadotrophin, or in relevant pregnancy characteristics, such as the sex ratio of the offspring. These findings suggest that African-American males are exposed to higher androgen concentrations than European-American males even before birth.
The U.S. black and white men from the study by Ross et al. (134) were compared with 54 Japanese men of the same age (mean age, 19–23 years) in a follow-up study (135). The serum testosterone levels of Japanese men were not lower than those of the U.S. whites and blacks but were intermediate between these two groups, whereas their SHBG levels were significantly lower. This finding may suggest a higher percentage of free testosterone in the Japanese (at very low risk) than in the U.S. men (at high risk), but free testosterone was not measured. Compared with the Japanese men, the two U.S. groups had significantly higher circulating levels of the conjugated androgen metabolites androsterone glucuronide (41%–50% higher) and 3,17-androstanediol glucuronide (25%–31% higher). This finding suggests that, in comparison with the high-risk U.S. groups, the low-risk Japanese population has a lower testosterone metabolism, most likely a lower activity of the enzyme 5-reductase that converts testosterone to DHT and the testosterone precursor androstenedione to androsterone. However, the markedly higher levels of androsterone glucuronide in U.S. men could also be indicative of a higher testosterone production in comparison with Japanese men.
Lookingbill et al. (143) reported a similar observation, comparing 53 normal healthy U.S. Caucasians and 57 Chinese males in Hong Kong between the ages of 24 and 26 years. The Caucasian men had 67% higher serum levels of androsterone glucuronide and 76% higher levels of 3,17-androstanediol glucuronide than the Chinese men did. Circulating levels of testosterone, free testosterone, or DHT were not significantly different, but Caucasian men had 46% higher serum levels of the androgen precursor DHEA sulfate and 32% higher levels of androstenedione. These data are also suggestive of a higher 5-reductase activity in high-risk Caucasians than in low-risk Chinese men, and they suggest an increased production of androgen precursors in the Caucasians.
In contrast to the observations of Ross et al. (134,135) and those of Lookingbill et al. (143), De Jong et al. (144) found 71% higher circulating total testosterone levels in 123 Caucasian-Dutch men (high risk) than in 91 Japanese men (low risk). The men in these studies were considerably older (50–79 years) than those studied by the previous two other groups. DHT levels were not different, but the ratio of DHT to testosterone was 10% lower in Dutch men than in Japanese men, possibly reflecting lower 5-reductase activity; no data were presented on androgen metabolites. Serum levels of estradiol were 15% higher (significant) in the Dutch men than in the Japanese men. SHBG levels were not different, but the ratio of testosterone to SHBG concentrations was 34% higher in Dutch men than in Japanese men, which suggests higher amounts of free testosterone in Dutch men, but this parameter was not measured separately.
Ellis and Nyborg (145) studied 4462 U.S. Army Vietnam veterans, ages 31–50 years, and compared serum testosterone levels in 3654 non-Hispanic white men (mean 6.37 ng/mL) with those in 525 African-Americans (6.58), 200 Hispanics (6.33), 34 Asian/Pacific Islanders (6.89), and 49 Native Americans (6.31). The serum testosterone levels in the African-American men were significantly higher than those in the non-Hispanic white men, but the differences among the other groups were not significant. The serum testosterone difference between black and white men was larger in men between 31 and 35 years of age (6.6%) than for men ages 35–40 years (3.7%) or ages 40–50 years (0.5%). No other hormones were measured in this study.
Wu et al. (146) conducted a population-based study, comparing circulating hormone levels in 1127 healthy men: 325 African-American men, 411 European-American men, 126 Chinese-Americans, and 275 Japanese-Americans with a median age of 69.6 years (range, 35–89 years), 8.2% of whom were 60 years or younger. Serum levels of total testosterone were slightly, but significantly, higher (9%–11%) in Asian-Americans than in European-American men, whereas they were intermediate and not significantly different from the two other groups in African-Americans. The same pattern was found for serum levels of bioavailable testosterone (not bound to SHBG) and the percentage of free testosterone (not bound to either SHBG or albumin), but only the 11%–12% difference between Chinese-American and European-American men was significant. SHBG levels were not different among the four groups. In comparison with European-American men, DHT levels were 7% higher (significant) in high-risk African-Americans and low-risk Japanese-Americans, but similar in Chinese-Americans. The ratio of DHT to testosterone was 10% lower (significant) in Chinese-Americans than in European-Americans, but not significantly different in African-Americans and Japanese-Americans who had slightly higher and lower ratios, respectively, than European-Americans. These data do not appear to provide clear support for the notions of a relation between increased 5-reductase activity or testosterone production and prostate cancer risk, but this study did not include more direct indicators, such as androsterone glucuronide and 3,17-androstanediol glucuronide.
DHEA sulfate was measured by Corder et al. (147) in stored serum samples of 90 African-American and 91 European-American men with prostate cancer and equal numbers of matched controls who were identified in a nested case–control study in a cohort of men in the Kaiser Permanente Medical Care Program in Northern California. Regardless of age, no significant differences were found between the two groups in DHEA sulfate levels, which were lower in men 57 years and older than in younger men.
Santner et al. (148) conducted the only study to date in which androgen production and metabolism by 5-reductase were determined in a direct fashion in populations with different risk for prostate cancer. A radioisotope method involving intravenous administration of tritiated testosterone was used to measure the conversion of testosterone to DHT in healthy European-Americans (ages 22–27 years), Chinese-Americans (ages 20–37 years), and Chinese men living in Beijing, China (ages 24–39 years). No differences in conversion of testosterone to DHT were found among these three groups. Circulating testosterone and SHBG levels were lower in the Beijing Chinese than in the two U.S. groups, and the differences with the U.S. Chinese subjects were significant, whereas no differences were found in free testosterone. There was a nonsignificant trend toward lower calculated metabolic conversion rates of testosterone comparing European-Americans with the Chinese groups and U.S. Chinese with Beijing Chinese. Calculated testosterone production rates were lower in Beijing Chinese than in the two U.S. groups, the difference with American Chinese being significant. The ratios of urinary 5- to 5-reduced steroids, which are an indicator of overall 5-reductase activity, were also not different in 20 European-American male students compared with 20 Chinese students living in Hong Kong. Urinary excretion of androsterone, etiocholanolone, and total ketosteroids was lower in the Chinese than in the U.S. students, which was significant when the data were combined with those of 20 female students from each of the two populations. Taken together, these data indicate that 5-reductase activity is not different in Asian and Caucasian men and is not affected by the environment in which Asian men live. However, these results suggest that the living environment influences testosterone production in Asian men.
Polymorphisms in genes involved in steroid hormone metabolism and action.
Studies have addressed the hypothesis (137–139) that functional polymorphisms in the 5-reductase gene, in genes involved in testosterone biosynthesis or DHT catabolism, and in the androgen receptor gene could be associated with the differences in prostate cancer risk among various populations. These studies are summarized in the following paragraphs.
The SRD5A2 gene, which encodes for human type II 5-reductase enzyme, is expressed in the prostate and is located on chromosome 2p23 (149,150) and contains polymorphic TA dinucleotide repeats in its transcribed 3' untranslated region (151). Reichardt et al. (152) demonstrated that TA(0) [87 base pairs (bp)] is the most common allele and was homozygous in 81% of non-Hispanic, white Americans (n = 68), 78% of Asian-Americans (n = 37), and 67% of African-Americans (n = 94). The next most common allele TA(9) (103–105 bp) is heterozygous with the TA(0) allele and occurred in 19% of the non-Hispanic, white American men, 22% of the Asian-Americans, and 15% of the African-Americans. The TA(18) allele (212–131 bp) was only found in African-Americans (18%) as heterozygous with the TA(0) allele in all except one who was homozygous. Thus, the longer alleles are unique to African-Americans and may be related to their extremely high risk for prostate cancer. However, the functional significance of these polymorphisms is not yet known.
Makridakis et al. (153) identified another polymorphism in the SRD5A2 gene, the presence of a valine to leucine mutation at codon 89. If this mutation occurs in a homozygous state, it confers 28% lower 5-reductase activity as measured in Asian men with this genotype compared with heterozygous men and men without the mutation. These researchers observed that the frequency of the 89 valine–valine genotype was 59% in African-American men (n = 95), 57% in non-Hispanic white Americans (n = 49), 48% in Latino Americans (n = 40), and 29% in Asian-Americans (n = 102). The 89 valine–leucine genotype occurred in 37%–39% of African-Americans, non-Hispanic white Americans, and Latino Americans, and in 49% of Asian-Americans. The frequency of the 89 leucine–leucine genotype (lower 5-reductase activity) was 3%–4% in African-American and non-Hispanic white Americans, 15% in Latino Americans, and 22% in Asian-Americans. A recent report from another, larger study by Lunn et al. (154) is essentially consistent with these findings. In this study, the frequency of the 89 valine–valine genotype was 65% in African-American men (n = 118), 41% in European-Americans (n = 176), and 15% in Asians (Taiwanese) (n = 108). The 89 valine–leucine genotype occurred in 32% of African-Americans, 50% of European-Americans, and in 57% of Asians. The frequency of the 89 leucine–leucine genotype (lower 5-reductase activity) was 2.5% in African-Americans, 8.5% in European-Americans, and 28% in Asian-Americans. The higher frequency of the 89 leucine–leucine genotype in Latino American men and particularly Asians may be related with the lower risk for prostate cancer found in these two ethnic groups, and the low frequency of 86 leucine alleles in African-Americans may be related to their extremely high risk. However, there appears not to be a relation between plasma concentrations of 3-androstanediol-glucuronide as an indicator of 5-reductase activity and the three different SRD5A2 gene codon 89 genotypes (155).
Makridakis et al. (156) also identified another polymorphism in the SRD5A2 gene, a mis-sense alanine to threonine mutation at codon 89. An in vitro construction of the mutant enzyme displayed a substantial increase in activity (Vmax). The frequency of the mutation was very low, 1.0% and 2.3%, in healthy, high-risk African-Americans and lower-risk Hispanic men, respectively. Although no data were presented on other ethnic/racial groups, it seems unlikely that this mutation is responsible for the large ethnic/racial variations in prostate cancer risk.
The CYP17 gene, which encodes for the cytochrome P450C17 enzyme that has both 17-hydroxylase and 17,20-lyase activity in the adrenal and testicular biosynthesis of androgens, is located on chromosome 10q24.3 (157). This gene is polymorphic with two common alleles, the wild-type CYP17A1 allele and the CYP17A2 allele containing a single base pair mutation in the untranscribed 5' region of exon 1 (157). This mutation creates an additional Sp1 site in the promoter region, suggestive of increased expression potential (157). The functional significance of this polymorphism in men is not known, but premenopausal and postmenopausal women with the A2 allele have been reported to have higher circulating estradiol and progesterone levels than women homozygous for the A1 allele (158,159). Circulating levels of DHEA and androstenedione, but not testosterone, were increased in postmenopausal women (159). Lunn et al. (154) recently reported that the frequencies of the A1/A1 and A1/A2 genotype were between 40% and 44%, and the frequency of the A2/A2 genotype was 16%–17% in both African-American men (n = 115) and European-Americans (n = 115), accounting for an A2 allele frequency of 0.36–0.38. In Asians (Taiwanese; n = 110), however, the A1/A1 genotype occurred in 24%, the A1/A2 genotype in 49%, and the A2/A2 genotype in 27%, with an A2 allele frequency of 0.52. The frequency differences between the Asians and the two American groups were statistically significant and are perhaps related to the low risk for prostate cancer in Asian men.
Verreault et al. (160) reported complex dinucleotide polymorphisms in the 3rd intron of the human HSD3B2 gene, located on chromosome 1p13, which encodes type II 3-hydroxysteroid dehydrogenase, which is expressed in the adrenals and testes, and catabolizes DHT (161). Devgan et al. (162) reported that the frequency of HSD3B2 alleles differs between African-American, European-American, and Asian men. One minor allele is unique for African-American men (6% allele frequency), whereas the most common allele is more frequent in European-Americans (52%) than among African-American or Asian men (34%–37%). The second most common allele is more frequent in African-Americans (25%) than in either Asians (15%) or European-American men (11%). As with the TA dinucleotide polymorphisms in the SRD5A2 gene, the functional significance of these HSD3B2 gene polymorphisms is not known.
The human androgen receptor gene, which is located on the X chromosome, also contains polymorphisms that are found as 8–31 CAG and 8–17 GGC (or GGN) microsatellite repeats in exon 1 encoding for the N-terminal domain of the protein where transactivation activity resides (139). The CAG repeat length has been demonstrated to determine transactivation activity of the androgen receptor, with 40 or more repeats being associated with human androgen insensitivity syndromes, such as spinal and bulbar muscular atrophy, and reduction of repeat length leading to increased transactivation activity in vitro (137–139). The functional significance of the GGC repeat length is not clear. Irvine et al. (163) reported that 75% of African-Americans (n = 44) had CAG repeat lengths of less than 22, whereas 62% of European-Americans (n = 39) and 49% of Asian-Americans (n = 39) had such shorter alleles. Very short alleles (<17 CAG repeats) occurred almost exclusively in African-Americans. The most common GGC allele (16 repeats) was found in 70% of Asian-Americans, 57% of European-Americans, and only 20% of African-Americans. The frequency of short GGC repeats (<16) was 61% in African-Americans, 27% in Asian-Americans, and 11% in European-Americans. GGC repeats longer than 16 were rare in the Asian-American men (3%) but more frequent in African-Americans (20%) and European-Americans (32%). Sartor et al. (164) essentially confirmed the findings on CAG repeats in a sample of African-Americans (n = 65) and European-American men (n = 130). Mean and median number of CAG repeats was 19 in African-Americans and 21 in European-Americans, and 57% of the African-American men had less than 20 repeats, whereas only 28% of European-American men had such short repeats. Ekman et al. (165), however, did not find significant differences in the distribution of CAG repeats comparing Swedish and Japanese men with BPH but without cancer (n = 38 and 33, respectively), but Swedish men with prostate cancer (n = 118) had somewhat shorter CAG repeats (mean, 15.9; median, 15) than Japanese prostate cancer patients (n = 34; mean, 17.5; median, 17). In conclusion, in two studies short CAG repeat alleles in the androgen receptor gene, which are probably associated with greater androgen receptor transactivation activity, were most frequent in the highest-risk population (African-Americans) and least frequent in the lowest-risk group (Asian-Americans), whereas the frequency was intermediate in intermediate-risk European-Americans. The high frequency of short GGC repeats found in African-Americans may also be related with their extremely high risk for prostate cancer, but the functional significance of this polymorphism is not yet known.
Summary and conclusions.
When examining Table 1
, few clear or convincing patterns emerge about associations between circulating hormone concentrations and prostate cancer risk at the population level. Two studies examined levels of androsterone glucuronide and 3,17-androstenediol glucuronide, which are considered (166–168) indicators of 5-reductase activity, particularly 3,17-androstenediol glucuronide, which is a direct metabolite of DHT. In both studies, the levels of these 5-reduced androgen metabolites were lower in low-risk Asian populations than in high-risk European-Americans (135,143). These findings suggest lower 5-reductase activity in the Asians and consequently reduced formation of DHT and androgenic stimulation of the prostate. This notion is supported by the reported higher frequency in Asians than in European-Americans or African-Americans of a polymorphism in the 5-reductase (SRD5A2) gene that appears to be associated with lower 5-reductase activity (a valine to leucine mutation at codon 89) (136,153). However, no differences were found between Asians and European-Americans in a study in which overall conversion of testosterone into DHT was directly measured (148). Furthermore, androsterone glucuronide and 3,17,-androstenediol glucuronide levels were not higher in very high-risk African-Americans than in intermediate-risk European-Americans, and circulating levels of DHT and the ratio of DHT to testosterone were not different in ethnic populations (Asian, black, and white) that differ in prostate cancer risk (143,144,146). Thus, the relation between 5-reductase activity and prostate cancer risk at the population level remains unclear at present.
Circulating levels of testosterone and/or free testosterone were slightly higher in African-Americans than in European-Americans in five of six studies that examined this question, but this finding is statistically significant in only one study (134). Furthermore, lower as well as higher testosterone concentrations have been found in lower-risk Asian or African men compared with European-Americans or African-Americans, although testosterone levels were lower in Asians living in Asia than in American populations regardless of ethnicity in two of three studies. Thus, these studies in ethnic/racial groups provide at present no substantive evidence in support of the hypothesis of a causal positive relation between elevated 5-reductase activity and prostate cancer risk at the population level and only very limited evidence for elevated (free) testosterone levels being associated with prostate cancer risk.
The only other patterns appearing in the data in Table 1 are that levels of estrogens are slightly higher (in five of five studies) and those of DHEA (sulfate) lower (in three of three studies) in black Africans and African-Americans than in men of European descent—hardly any data are available on Asians in this regard (63–65,134,140,141,144). The biologic significance of these observations is unclear, but they may be related to the high susceptibility of black men to prostate cancer when they live in the American environment. However, the above summarized endocrine differences between very high-risk African-Americans and high-risk European-Americans were not consistent in younger and older men, and they were not similar to the differences observed between the high-risk U.S. populations and the low-risk African black men (63–65,140,141). These inconsistencies raise the possibility that the factors and endocrine mechanisms that determine the difference in risk between African black men and African-Americans are dissimilar from those that determine the risk difference between African-Americans and European-Americans (11).
Finally, CAG repeat length polymorphism in the androgen receptor gene was found to be associated with prostate cancer risk in two studies. Short CAG repeat alleles are probably associated with greater androgen receptor transactivation activity. Such short CAG repeat alleles were most frequent in African-Americans (very high-risk), least frequent in Asian-Americans (low risk), and intermediate in European-Americans (intermediate risk). Another androgen receptor polymorphism in GGC repeats may also be related with risk for prostate cancer, but the functional significance of this polymorphism is unknown.
Association of Steroid Hormonal Factors With Prostate Cancer Risk in Population-Based Case–Control Studies
Circulating of steroid hormone levels in nested case–control studies.
A summary of the results of population-based, nested case–control studies that examined the association between circulating levels of steroid hormones and risk for prostate cancer is provided in Table 2. The details of each study are summarized in the following paragraphs. One study by Carter et al. (169) concerned only 16 case subjects and contained considerable bias because of storage effects on hormone measurements, which were recognized but not controlled for. This study is, therefore, not further discussed here.
|
Nomura et al. (170) compared 98 prostate cancer case patients with matched control subjects from a cohort of 6860 Hawaiian-Japanese men, which were followed for an average of 14 years. No significant differences were found between case patients and control subjects or associations with risk for serum levels of testosterone, DHT, estrone, estradiol, and SHBG, measured once at the start of the cohort study (free testosterone levels were not determined). An elevation in risk was only observed for an increasing ratio of testosterone to DHT, which was borderline significant. This latter observation perhaps suggests an inverse relation between (peripheral) 5
-reductase activity and prostate cancer risk. Barrett-Connor et al. (171) followed a Californian cohort of 1008 white, upper-middle class men between the ages of 40 and 79 years for a period of 14 years, during which time 57 cases of prostate cancer occurred (26 deaths and 31 incident cases). No significant relation was found between the risk for prostate cancer and baseline serum concentrations of testosterone, estrone, and SHBG. However, RR increased linearly with an increasing serum level of androstenedione, a testosterone precursor. RR also increased linearly with an increasing serum level of estradiol, but this finding was not statistically significant.
Hsing and Comstock (172) and Comstock et al. (173) reported results of a population-based, nested case–control study in a cohort of 25 620 men (98% European-American) in Maryland. Blood samples were obtained in 1974, and 98 cases of prostate cancer were identified in the first 13 years of follow-up (81 cases in 12 years of follow-up for DHEA and DHEA sulfate). Men 70 years and older as well as men younger than 70 years were studied separately (except for DHEA and DHEA sulfate). No significant differences were found between case patients and control subjects or associations with risk for baseline serum testosterone, DHT, DHEA, DHEA-sulfate, estrone, or estradiol. The ratio of testosterone to DHT was higher in case patients than in control subjects of all ages, and, for men younger than 70 years but not for older men, risk for prostate cancer increased with an increasing testosterone/DHT ratio; both findings were borderline significant (0.05<P<0.1). This latter observation could suggest an inverse relation between (peripheral) 5-reductase activity and prostate cancer risk.
Nomura et al. (174) reported a follow-up of their 1988 study, including 141 case patients and 141 matched control subjects from their cohort of 6860 Hawaiian-Japanese men followed for an average of more than 20 years. In this population-based, nested case–control study, there were no significant differences between case patients and control subjects or associations with risk for baseline serum testosterone, free testosterone, DHT, ratio of testosterone to DHT, androsterone-glucuronide, 3-androstanediol-glucuronide, and androstenedione.
DHEA sulfate was measured by Corder et al. (147) in stored serum samples of 181 men with prostate cancer (90 African-Americans and 91 European-Americans) and equal numbers of matched control subjects who were identified in a nested case–control study in a cohort of men in the Kaiser Permanente Medical Care Program in Northern California. For men younger than 57 years or for older men there were no significant differences between case subjects and control subjects in DHEA sulfate levels, which were also not associated with risk.
Gann et al. (175) conducted a prospective, nested case–control study on 222 case patients with prostate cancer and 390 matched control subjects obtained from the Physician's Health Study (a randomized intervention trial with aspirin and -carotene in 22 071 U.S. male physicians, probably largely white), with a mean follow-up of approximately 6 years. There were no significant differences between case patients and control subjects for plasma testosterone, SHBG, DHT, ratio of testosterone to DHT, 3-androstanediol-glucuronide, or estradiol. Several highly significant associations were found between plasma levels of SHBG and the steroid hormones studied. Therefore, odd ratios were calculated after simultaneous adjustment for all these endocrine factors for 222 matched case–control sets. This approach resulted in significant positive associations with risk for testosterone and the ratio of testosterone to DHT and inverse associations with risk for SHBG and estradiol. A positive association was also found with risk for 3-androstanediol-glucuronide, which was borderline significant, but there was no association with risk for DHT. These observations support a relationship between elevated testosterone and prostate cancer risk, but they are contradictory regarding a relation between (peripheral) 5-reductase activity and prostate cancer risk.
Guess et al. (176) reported on a population-based, case–control study from a cohort of more than 125 000 European-American men in the Kaiser Permanente Medical Care Program. They compared 106 case patients and matched control subjects selected from men, with a median follow-up of 14 years (range, 5–23 years). No significant differences were found between case patients and control subjects or associations with risk for baseline serum testosterone, free testosterone, DHT, androsterone-glucuronide, or 3-androstanediol-glucuronide.
Vatten et al. (177) conducted a population-based, nested case–control study of 59 case patients with prostate cancer and 180 matched control subjects from a cohort of approximately 28 000 Norwegian men, with a mean follow-up of 10 years (range, 1–19 years). There were no significant differences between case patients and control subjects or associations with risk for baseline serum testosterone, DHT, ratio of testosterone to DHT, or 3-androstanediol-glucuronide.
Dorgan et al. (178) reported results from a population-based, nested case–control study of 116 case patients with prostate cancer and 231 matched control subjects from a cohort of 29 133 Finnish men from the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study of cigarette smokers with a follow-up of 5–8 years. No significant differences were found between case patients and control subjects or associations with risk for baseline serum testosterone, free testosterone, SHBG, DHT, DHEA-sulfate, 3-androstanediol-glucuronide, androstenedione, estrone, or estradiol. There was a nonsignificant trend toward a higher ratio of testosterone to DHT in case patients than in control subjects and a positive association with risk for this ratio. This finding may suggest an inverse relation between (peripheral) 5-reductase activity and prostate cancer risk.
Heikkilä et al. (179) reported on the results of a population-based, nested case–control study in a Finnish cohort study in which serum was collected and stored, and a cohort of 16 481 men was followed for up to 24 years. During this period, 166 case patients with prostate cancer were identified and were matched to 300 control subjects. Serum levels of testosterone, SHBG, and androstenedione were similar in case patients and control subjects, and they were not associated with prostate cancer risk. When case patients identified in the first 8 years of follow-up were excluded, there was a borderline significant (P = .06) trend for increasing risk with higher levels of testosterone but not with SHBG or androstenedione. This finding supports the notion of a relationship between elevated androgen levels and prostate cancer risk.
Polymorphisms in genes involved in steroid hormone metabolism and action in poulation-based case–control studies.
Several case–control studies have addressed the hypothesis (137–139) that functional polymorphisms in the 5-reductase gene, in genes involved in testosterone biosynthesis or DHT catabolism, and in the androgen receptor gene could be associated with differences in prostate cancer risk. These studies are summarized in the following paragraphs.
Kantoff et al. (180) studied the association between prostate cancer risk and the polymorphisms in TA dinucleotide repeats in the transcribed 3' untranslated region of the human SRD5A2 gene encoding for type II 5-reductase enzyme; the functional significance of these polymorphisms is not known, as indicated earlier. These investigators conducted a nested case–control study with the use of the Physician's Health Study cohort with 590 men with prostate cancer and 802 age-matched control subjects. They observed that the frequency of the most common genotype, TA(0)/TA(0), was 76.4% in case patients and 75.4% in control subjects, the frequency of the next most common genotypes, TA(0)/TA(9) and TA(0)/TA(18), was 22.4% in case patients and 22.1% in control subjects, and the frequency of genotypes TA(9)/TA(9) and TA(18)/TA(18) was 1.2% in case patients and 2.5% in control subjects; only two control subjects had a TA(9)/TA(18) genotype. Men that were homozygous for long repeats, TA(9)/TA(9) or TA(18)/TA(18), were at lower risk for prostate cancer than men with the predominant TA(0)/TA(0) genotype, with a borderline significant (P = .08) odds ratio (OR) of 0.47. These findings sharply contrast with the aforementioned observation that the long TA(18) alleles are unique to African-Americans, who are at very high risk for prostate cancer (152).
As indicated earlier, Makridakis et al. (153) identified another polymorphism in the SRD5A2 gene, a mis-sense alanine to threonine mutation at codon 89, apparently associated with an increase in 5-reductase activity. In a nested case–control study conducted in the Hawaii–Los Angeles Multiethnic Cohort Study of Diet and Cancer, the frequency of the mutation appeared to be low but was responsible for 8%–9% of cases in African-American (203 case patients and 257 unmatched control subjects) and Hispanic men (160 case patients and 193 control subjects); no data were presented on other ethnic/racial groups (153). The RR (age-adjusted) of prostate cancer for possessing a mutated allele was 3.28 (95% confidence interval [CI] = 1.09–11.87) in African-American men and 2.50 (95% CI = 0.90–7.40) in Hispanics. For advanced prostate cancer, the RRs for possessing a mutated allele were more significant: 7.22 (95% CI = 2.17–27.91) in African-American men and 3.60 (95% CI = 1.09–12.27) in Hispanics. Although the results of this study support the notion that increased 5-reductase activity may be related to prostate cancer risk, it seems unlikely that the alanine to threonine mutation at codon 89 in the SRD5A2 gene is involved in a substantial proportion of prostate cancer cases.
The relation between prostate cancer risk and the occurrence of the aforementioned valine to leucine mutation at codon 89 in the SRD5A2 gene, a polymorphism that is associated with reduced 5-reductase activity, was examined by Febbo et al. (155) and Lunn et al. (154). Febbo et al. (155) conducted a nested case–control study with the use of the Physician's Health Study cohort with 584 men with prostate cancer and 799 matched control subjects. The valine–leucine and leucine–leucine genotypes were found in 50% of case patients and 51% of control subjects and were not associated with elevated prostate cancer risk as compared with the valine–valine genotype (OR = 0.84–0.96). Lunn et al. (154) confirmed these findings in a case–control study that employed 108 prostate cancer patients from urology clinics at the University of North Carolina and nearby Duke University. Control subjects (n = 156) were drawn from BPH and impotence patients at the same clinics and not matched to case patients. The groups were predominantly European-American (5%–11% were African-American). The valine–leucine and leucine–leucine genotypes were found in 56% of case patients and 49% of control subjects and were not associated with prostate cancer risk as compared with the valine–valine genotype (OR = 1.3; 95% CI = 0.8–2.2). These observations are consistent with the absence of a relation between plasma concentrations of 3,17-androstanediol glucuronide and the three different SRD5A2 gene codon 89 genotypes reported by Febbo et al. (155).
In the same case–control study, Lunn et al. (154) also examined the association between prostate cancer risk and the aforementioned single base-pair mutation polymorphism in CYP17 gene (152c), encoding for the 17-hydroxylase and 17,20-lyase activity. The CYP17A2 allele containing a single base-pair mutation was found in 69% of case patients and 57% of control subjects and was associated with prostate cancer risk with an OR of 1.7 (95% CI = 1.0–3.0). The association appeared to be limited to men younger than 65 years, with an increased OR of 2.30 (95% CI = 1.0–4.8). Contradictory findings of this association between prostate cancer risk and the presence of the CYP17A2 allele were reported from a Swedish case–control study by Wadelius et al. (181). The frequency of the A1/A2 or A2/A2 genotype was 61% in prostate cancer cases (n = 178) and 71% in population controls (n = 160). The OR of having the A1/A1 genotype versus the A1/A2 or A2/A2 genotype was 1.61 (95% CI = 1.02–2.53). This latter finding is consistent with a preliminary report of higher circulating testosterone levels found in men homozygous for the A1 allele than in men with an A2 allele of the CYP17 gene (181).
Six population-based, case–control studies with substantive numbers of cases examine the association between the aforementioned CAG and GGC (or GGN) repeat polymorphisms in the human androgen receptor gene (163,182–186). The results of these studies are summarized in Tables 3–5. As shown in Table 3, these studies indicate that CAG repeat lengths shorter than 22 may be associated with slightly increased risk for prostate cancer with elevated ORs or RRs found for <22 repeats (versus 
22) in four studies (163,182–184), and for 20–21 repeats (versus <20 or 22) in one study (185). However, only in the study by Giovannucci et al. (182), but not in four other studies (163,183–185), the tendency of increased risk with decreasing repeat length was statistically significant. In addition to these six population-based, case–control studies, a study by Hakimi et al. (186) compared the CAG repeat lengths in the androgen receptor of 59 prostate cancer patients with published data of the general population (n = 370). Short repeats (
17) were more frequent in cases than in the general population (OR = 3.7; 95% CI = 1.3–10.5; P = .02). Thus, these findings consistently indicate the possibility that prostate cancer risk is slightly increased with shorter CAG repeat alleles. This possibility may be related to a greater androgen receptor transactivation activity associated with shorter CAG repeat alleles, as indicated earlier.
|
|
|
Two of these six case–control studies and a seventh population-based, case–control study examined GGN or GGC repeats in the androgen receptor gene, the functional significance of which is not known at present (163,184,187). As shown in Table 4
, these studies indicate that GGN or GGC repeat lengths may influence risk, but the results do not agree with one another. Hakimi et al. (186) also compared androgen receptor GGC repeat lengths in 54 prostate cancer patients with published population data (n = 110). Short repeats (14) were more frequent in case patients than in the general population (OR = 4.6; 95% CI = 1.3–16.1; P = .02). However, Correa-Cerro et al. (185) did not find such an association in a French–German population. Thus, even though two studies consisted of substantial numbers of case patients and control subjects (184,187), the influence of GGN or GGC repeat length on prostate cancer risk is presently not clear. The combined effects of CAG and GGN repeat length were also examined in the three case–control studies, and they were greater than those either polymorphism separately in all three, as indicated in Table 5. CAG repeats of less than 22 or 21 appeared to be the consistent factor in this interaction in all three studies and were associated with (borderline) significantly increased risk. However, the functional significance of this combined effect is not clear.
Summary and conclusions.
When examining Table 2, no clear or convincing patterns emerge about associations between circulating hormone concentrations and prostate cancer risk, with few exceptions. In most studies, an association was found between increased risk and increased ratios of testosterone to DHT, which reached statistical significance in three of six studies. Although this finding suggests a relation between reduced 5-reductase activity and prostate cancer risk, no associations were found between risk and the levels of androsterone glucuronide and 3,17-androstenediol glucuronide, indicators of 5-reductase activity (166–168)—a borderline significant trend for such an association was found in only one (175) of five studies (174–178). Significant associations between prostate cancer risk and elevated levels of testosterone and androstenedione or decreased levels of SHBG and estradiol were found each in only a single study [(175) for testosterone, SHBG, and estradiol; (171) for androstenedione], and they were not observed in eight (testosterone), four (SHBG and estradiol), or three (androstenedione) other studies. It is possible that relevant associations may have been missed in most studies, because the data for each individual hormone were not adjusted for concentrations of other hormones studied, even though there are many intercorrelations between circulating levels of these hormones. Only in the study by Gann et al. (175) were these types of adjustments applied, after which risk was significantly associated with increased circulating testosterone levels and testosterone/DHT ratio, as well as decreased concentrations of SHBG and estradiol, and, in men older than 61 years, DHT. A meta-analysis study by Eaton et al. (188) used most but not all studies included in this overview, as well as a study that was discounted here (169) and some unpublished data. They found no significant differences for the ratios of mean hormone levels between case patients and control subjects, with the exception of slightly elevated levels of 3,17-androstenediol glucuronide. This analysis is essentially in agreement with the analysis of this overview, with the only consistent finding slightly elevated ratios between case patients and control subjects of 3,17-androstenediol glucuronide in five of five studies (Table 2). However, Eaton et al. (188) did not take into account the risk estimates produced by these studies, which seriously limits its conclusions.
The results of three nested case–control studies on the relation between prostate cancer risk and two different polymorphisms in the human type II 5-reductase enzyme gene (SRD5A2) do not support the notion of an association between risk and increased 5-reductase activity (154,155,180). However, an infrequent polymorphism associated with increased 5-reductase activity was more common in case patients than in control subjects in one case–control study (156), which indicates that associations between polymorphisms in the SRD5A2 gene and prostate cancer risk cannot be discounted. Data on a relation between prostate cancer risk and a polymorphism in the CYP17 gene, which encodes for the 17-hydroxylase and 17,20-lyase activity involved in androgen biosynthesis, are contradictory (154,181). Furthermore, the functional significance of this polymorphism in males is not yet known. In four of five similar nested case–control studies of polymorphisms in trinucleotide repeats in the promoter region of the androgen receptor gene, an association was found between risk and short CAG repeat alleles—short CAG repeat lengths are associated with greater androgen receptor transactivation activity. However, this association was weak and significant in only one study. An association between risk and polymorphisms in androgen receptor GGC or GGN repeat lengths is not clear because results of three studies were inconsistent, and the functional significance of these polymorphisms is not known. There is possibly an interaction between CAG and GGC/GGN repeat length in relation to prostate cancer risk, but results of the three studies examining this possible interaction were inconsistent. Short CAG repeats were also correlated with advanced disease and/or early onset of prostate cancer (182–186,189,190).
Epidemiologic Evidence for Involvement of Steroid Hormones: Summary and Conclusions
Taken together, the results of the above summarized studies do not provide unequivocal or strong evidence for any particular association between prostate cancer risk and circulating levels of hormones or polymorphisms in genes that encode for proteins involved in steroid hormone action or metabolism. Only a few associations with prostate cancer risk have been observed consistently (in at least three studies), and they are weak at best: 1) slightly, but mostly not significantly, higher circulating testosterone and estrogen levels and lower DHEA (sulfate) levels in high-risk African-American men as compared with lower-risk European-American men, and 2) a CAG repeat length polymorphism in the androgen receptor gene with short repeat lengths associated with increased risk and increased receptor transactivation activity. The evidence for involvement of activity of the enzyme 5-reductase, which is critical in androgen action in the prostate, is inconsistent and contradictory.
Difficulties in Interpretation Several important points should be considered in interpreting these observations: First, there are numerous potential problems with most studies that measure circulating hormone levels, such as the usually large intra- and interassay variability in the immunoassays used (122,191–194). Typically, only single blood samples are avail