This information is produced and provided by the National Cancer Institute (NCI). The information in this topic may have changed since it was written. For the most current information, contact the National Cancer Institute via the Internet web site at http://cancer.gov or call 1-800-4-CANCER.
Many of the medical and scientific terms used in this summary are found in the NCI Dictionary of Genetics Terms. When a linked term is clicked, the definition will appear in a separate window.
Many of the genes described in this summary are found in the Online Mendelian Inheritance in Man (OMIM) database. When OMIM appears after a gene name or the name of a condition, click on OMIM for a link to more information.
The public health burden of prostate cancer is substantial. A total of 241,740 new cases of prostate cancer and 28,170 deaths from the disease are anticipated in the United States in 2012, making it the most frequent nondermatologic cancer among U.S. males. A man's lifetime risk of prostate cancer is one in six. Prostate cancer is the second leading cause of cancer death in men, exceeded only by lung cancer.
Some men with prostate cancer remain asymptomatic and die from unrelated causes rather than as a result of the cancer itself. This may be due to the advanced age of many men at the time of diagnosis, slow tumor growth, or response to therapy. The estimated number of men with latent prostate carcinoma (i.e., prostate cancer that is present in the prostate gland but never detected or diagnosed during a patient's life) is greater than the number of men with clinically detected disease. A better understanding is needed of the genetic and biologic mechanisms that determine why some prostate carcinomas remain clinically silent, while others cause serious, even life-threatening illness.
Prostate cancer exhibits tremendous differences in incidence among populations worldwide; the ratio of countries with high and low rates of prostate cancer ranges from 60-fold to 100-fold. Asian men typically have a very low incidence of prostate cancer, with age-adjusted incidence rates ranging from 2 to 10 per 100,000 men. Higher incidence rates are generally observed in northern European countries. African American men, however, have the highest incidence of prostate cancer in the world; within the United States, African American men have a 60% higher incidence rate than white men.
These differences may be due to the interplay of genetic, environmental, and social influences (such as access to health care), which may affect the development and progression of the disease. Differences in screening practices have also had a substantial influence on prostate cancer incidence, by permitting prostate cancer to be diagnosed in some patients before symptoms develop or before abnormalities on physical examination are detectable. An analysis of population-based data from Sweden suggested that a diagnosis of prostate cancer in one brother leads to an early diagnosis in a second brother using prostate-specific antigen (PSA) screening. This may account for an increase in prostate cancer diagnosed in younger men that was evident in nationwide incidence data. A genetic contribution to prostate cancer risk has been documented, but knowledge of the molecular genetics of prostate cancer is still limited. Malignant transformation of prostate epithelial cells and progression of prostate carcinoma are likely to result from a complex series of initiation and promotional events under both genetic and environmental influences.
Risk Factors for Prostate Cancer
The three most important recognized risk factors for prostate cancer in the United States are:
Age is an important risk factor for prostate cancer. Prostate cancer is rarely seen in men younger than 40 years; the incidence rises rapidly with each decade thereafter. For example, the probability of being diagnosed with prostate cancer is 1 in 8,499 for men younger than 40 years, 1 in 38 for men aged 40 through 59 years, 1 in 15 for men aged 60 through 69 years, and 1 in 8 for men aged 70 years and older, with an overall lifetime risk of developing prostate cancer of 1 in 6.
Endogenous hormones, including both androgens and estrogens, likely influence prostate carcinogenesis. It has been widely reported that eunuchs and other individuals with castrate levels of testosterone prior to puberty do not develop prostate cancer. Some investigators have considered the potential role of genetic variation in androgen biosynthesis and metabolism in prostate cancer risk, including the potential role of the androgen receptor (AR) CAG repeat length in exon 1. This modulates AR activity, which may influence prostate cancer risk. For example, a meta-analysis reported that AR CAG repeat length greater than or equal to 20 repeats conferred a protective effect for prostate cancer in subsets of men.
Some dietary risk factors may be important modulators of prostate cancer risk; these include fat and/or meat consumption, lycopene,[13,14] and dairy products/calcium/vitamin D. Phytochemicals are plant-derived nonnutritive compounds, and it has been proposed that dietary phytoestrogens may play a role in prostate cancer prevention. For example, Southeast Asian men typically consume soy products that contain a significant amount of phytoestrogens; this diet may contribute to the low risk of prostate cancer in the Asian population. There is little evidence that alcohol consumption is associated with the risk of developing prostate cancer; however, data suggest that smoking increases the risk of fatal prostate cancer. Several studies have suggested that vasectomy increases the risk of prostate cancer, but other studies have not confirmed this observation. Obesity has also been associated with increased risk of advanced stage at diagnosis, prostate cancer metastases, and prostate cancer–specific death.[20,21]
Other nutrients have been studied for their potential influence on prostate cancer risk. The effect of selenium and vitamin E in preventing prostate cancer was studied in the Selenium and Vitamin E Cancer Prevention Trial (SELECT). This randomized placebo-controlled trial of selenium and vitamin E among 35,533 healthy men found no evidence of a reduction in prostate cancer risk, although a statistically significant increase (hazard ratio [HR], 1.17; 99% confidence interval [CI], 1.004–1.36; P = .008) in prostate cancer with vitamin E supplementation alone was observed. The absolute increased risk associated with vitamin E supplementation compared to placebo after more than 7 years of follow-up was 1.6 per 1,000 person years.
(Refer to the PDQ summary on Prevention of Prostate Cancer for more information.)
The Surveillance, Epidemiology and End Result (SEER) Cancer Registries has assessed the risk of developing a second primary cancer in 292,029 men diagnosed with prostate cancer between 1973 and 2000. Excluding subsequent prostate cancer and adjusting for the risk of death from other causes, the cumulative incidence of a second primary cancer among all patients was 15.2% at 25 years (95% CI, 5.01–5.4). There was a significant risk of new malignancies (all cancers combined) among men diagnosed prior to age 50 years, no excess or deficit in cancer risk in men aged 50 to 59 years, and a deficit in cancer risk in all older age groups. The authors suggested that this deficit may be attributable to decreased cancer surveillance in an elderly population. Excess risks of second primary cancers included cancers of the small intestine, soft tissue, bladder, thyroid, and thymus, and melanoma. Prostate cancer diagnosed in patients aged 50 years or younger was associated with an excess risk of pancreatic cancer.
The underlying etiology of developing a second primary cancer after prostate cancer may be related to various factors. Some of the observed excess risks could be associated with prior radiation therapy. Radiation therapy as the initial treatment for prostate cancer was found to increase the risk of bladder and rectal cancers and cancer of the soft tissues. More than 50% of the small intestine tumors were carcinoid malignancies, suggesting possible hormonal influences. The excess of pancreatic cancer may be due to mutations in BRCA2, which predisposes to both. The risk of melanoma was most pronounced in the first year of follow-up after diagnosis, raising the possibility that this is the result of increased screening and surveillance.
One Swedish study using the nationwide Swedish Family Cancer Database assessed the role of family history in the risk of a second primary cancer following prostate cancer. Of 18,207 men with prostate cancer, 560 developed a second primary malignancy. Of those, the relative risk (RR) was increased for colorectal, kidney, bladder, and squamous cell skin cancers. Having a paternal family history of prostate cancer was associated with an increased risk of bladder cancer, myeloma, and squamous cell skin cancer. Among prostate cancer probands, those with a family history of colorectal cancer, bladder cancer, or chronic lymphoid leukemia were at increased risk of that specific cancer as a second primary cancer.
Risk of Other Cancers in Multiple-Case Families
Several reports have suggested an elevated risk of various other cancers among relatives within multiple-case prostate cancer families, but none of these associations have been established definitively.[26,27,28]
In a population-based Finnish study of 202 multiple-case prostate cancer families, no excess risk of all cancers combined (other than prostate cancer) was detected in 5,523 family members. Female family members had a marginal excess of gastric cancer (standardized incidence ratio [SIR], 1.9; 95% CI, 1.0–3.2). No difference in familial cancer risk was observed when families affected by clinically aggressive prostate cancers were compared with those having nonaggressive prostate cancer. These data suggest that familial prostate cancer is a cancer site-specific disorder.
Family History as a Risk Factor for Prostate Cancer
As with breast and colon cancer, familial clustering of prostate cancer has been reported frequently.[30,31,32,33,34] From 5% to 10% of prostate cancer cases are believed to be due primarily to high-risk inherited genetic factors or prostate cancer susceptibility genes. Results from several large case-control studies and cohort studies representing various populations suggest that family history is a major risk factor in prostate cancer.[31,35,36] A family history of a brother or father with prostate cancer increases the risk of prostate cancer, and the risk is inversely related to the age of the affected relative.[32,33,34,35,36] However, at least some familial aggregation is due to increased prostate cancer screening in families thought to be at high risk.
Although many of the prostate cancer studies examining risks associated with family history have used hospital-based series, several studies described population-based series. The latter are thought to provide information that is more generalizable. The Massachusetts Male Aging Study of 1,149 Boston-area men found a RR of 3.3 (95% CI, 1.8–5.9) for prostate cancer among men with a family history of the disease. This effect was independent of environmental factors, such as smoking, alcohol use, and physical activity. Further associations between family history and risk of prostate cancer were characterized in an 8-year to 20-year follow-up of 1,557 men aged 40 to 86 years who had been randomly selected as controls for a population-based case-control study conducted in Iowa from 1987 to 1989. At baseline, 4.6% of the cohort reported a family history of prostate cancer in a brother or father, and this was positively associated with prostate cancer risk after adjustment for age (RR, 3.2; 95% CI, 1.8–5.7) or after adjustment for age, alcohol, and dietary factors (RR, 3.7; 95% CI, 1.9–7.2).
A meta-analysis of 33 epidemiologic studies provides more detailed information regarding risk ratios related to family history of prostate cancer. Risk appears to be greater for men with affected brothers (RR, 3.4; 95% CI, 3.0–3.8) than for men with affected fathers (RR. 2.2; 95% CI, 1.9–2.5). Although the reason for this difference in risk is unknown, possible hypotheses include X-linked or recessive inheritance. In addition, risk increased with increasing numbers of affected close relatives: RR was 2.6 (95% CI, 2.3–2.8) for one first-degree relative (FDR) and 5.1 (95% CI, 3.3–7.8) for two or more FDRs, but RR was only 1.7 (95% CI, 1.1–2.6) for an affected second-degree relative. Risk was influenced by age at prostate cancer diagnosis in this meta-analysis: RR was 3.3 (95% CI, 2.6–4.2) for diagnosis before age 65 years, versus a RR of 2.4 (95% CI, 1.7–3.6) for diagnosis at age 65 years or older.
Among the many data sources included in this meta-analysis, those from the Swedish population-based Family Cancer Database warrant special comment, as they are derived from a resource that contains 10.2 million individuals, among whom there are 182,000 fathers and 3,700 sons with medically verified prostate cancer. The size of this data set, with its near complete ascertainment of the entire Swedish population and objective verification of cancer diagnoses, should yield risk estimates that are both accurate and free of bias. The familial SIRs for prostate cancer were 2.4 (95% CI, 2.2–2.6), 3.8 (95% CI, 2.7–5.0), and 9.4 (95% CI, 5.8–14.0) for men with prostate cancer in their fathers only, brothers only, and both father and brother, respectively. The SIRs were even higher if the affected relative was diagnosed with prostate cancer before age 55 years. A separate analysis of this Swedish database reported that the cumulative (absolute) risks of prostate cancer among men in families with two or more affected cases were 5%, 15%, and 30% by ages 60, 70, and 80 years, respectively, compared with 0.45%, 3%, and 10% at the same ages in the general population. The risks were higher still if the affected father was diagnosed before age 70 years. The corresponding familial population attributable fractions (PAFs) were 8.9%, 1.8%, and 1.0% for the same three groups, respectively, yielding a total PAF of 11.6%; approximately 11.6% of all prostate cancer in Sweden can be accounted for on the basis of familial history of the disease.
Using data from the Nationwide Swedish Family-Cancer Database, age-specific HRs for prostate cancer diagnosis and mortality were computed. The analysis was stratified by whether the father and/or brother(s) of affected men also had prostate cancer and by their age at diagnosis. The HRs increased with decreasing age at diagnosis for both fathers and male siblings. As expected, the HR for prostate cancer diagnosis was high in men with a father and two brothers with prostate cancer (HR, 10.86; 95% CI, 7.08–16.66) or with three brothers with prostate cancer (HR, 24.35; 95% CI, 16.18–36.64).
The risk of prostate cancer may also increase in men who have a family history of breast cancer. Approximately 9.6% of the Iowa cohort had a family history of breast and/or ovarian cancer in a mother or sister at baseline, and this was positively associated with prostate cancer risk (age-adjusted RR, 1.7; 95% CI, 1.0–3.0; multivariate RR, 1.7; 95% CI, 0.9–3.2). Men with a family history of both prostate and breast/ovarian cancer were also at increased risk of prostate cancer (RR, 5.8; 95% CI, 2.4–14.0). Other studies, however, did not find an association between family history of female breast cancer and risk of prostate cancer.[38,44] A family history of prostate cancer also increases the risk of breast cancer among female relatives. The association between prostate cancer and breast cancer in the same family may be explained, in part, by the increased risk of prostate cancer among men with BRCA1/BRCA2mutations in the setting of hereditary breast/ovarian cancer or early-onset prostate cancer.[46,47,48,49] (Refer to the BRCA1 and BRCA2 section of this summary for more information.)
Family history has been shown to be a risk factor for men of different races and ethnicities. In a population-based case-control study of prostate cancer among African Americans, whites, and Asian Americans in the United States (Los Angeles, San Francisco, and Hawaii) and Canada (Vancouver and Toronto), 5% of controls and 13% of all cases reported a father, brother, or son with prostate cancer. These prevalence estimates were somewhat lower among Asian Americans than among African Americans or whites. A positive family history was associated with a twofold to threefold increase in RR in each of the three ethnic groups. The overall odds ratio associated with a family history of prostate cancer was 2.5 (95% CI, 1.9–3.3) with adjustment for age and ethnicity.
Evidence for inherited forms of prostate cancer can be found in several U.S. and international studies.[31,35,51,52,53,54] It was first noted in 1956 that men with prostate cancer reported a higher frequency of the disease among relatives than did controls. Shortly thereafter, it was reported that deaths from prostate cancer were increased among fathers and brothers of men who died of prostate cancer versus controls who died of other causes.
(Refer to the PDQ summary on Prevention of Prostate Cancer for more information about risk factors for prostate cancer in the general population.)
Inheritance of Prostate Cancer Risk
Many types of epidemiologic studies (case-control, cohort, twin, family) strongly suggest that prostate cancer susceptibility genes exist in the population. An analysis of monozygotic and dizygotic twin pairs in Scandinavia concluded that 42% (95% CI, 29–50) of prostate cancer risk may be accounted for by heritable factors. This is in agreement with a previous U.S. study that showed a concordance of 7.1% between dizygotic twin pairs and a 27% concordance between monozygotic twin pairs. The first segregation analysis was performed in 1992 using families from 740 consecutive probands who had radical prostatectomies between 1982 and 1989. The study results suggested that familial clustering of disease among men with early-onset prostate cancer was best explained by the presence of a rare (frequency of 0.003) autosomal dominant, highly penetrant allele(s). Hereditary prostate cancer susceptibility genes were predicted to account for almost half of early-onset disease (age 55 years or younger). In addition, early-onset disease has been further supported to have a strong genetic component from the study of common variants associated with disease onset before age 55 years.
Subsequent segregation analyses generally agreed with the conclusions but differed in the details regarding frequency, penetrance, and mode of inheritance.[60,61,62] A study of 4,288 men who underwent radical prostatectomy between 1966 and 1995 found that the best fitting genetic model of inheritance was the presence of a rare, autosomal dominant susceptibility gene (frequency of 0.06). In this study, the lifetime risk in carriers was estimated to be 89% by age 85 years and 3.9% for noncarriers. This study also suggested the presence of genetic heterogeneity, as the model did not reliably predict prostate cancer risk in FDRs of probands who were diagnosed at age 70 years or older. More recent segregation analyses have concluded that there are multiple genes associated with prostate cancer [63,64,65,66] in a pattern similar to other adult-onset hereditary cancer syndromes, such as those involving the breast, ovary, colorectum, kidney, and melanoma. In addition, a segregation analysis of 1,546 families from Finland found evidence for Mendelian recessive inheritance. Results showed that individuals carrying the risk allele were diagnosed with prostate cancer at younger ages (<66 years) than noncarriers. This is the first segregation analysis to show a recessive mode of inheritance.
While genetic testing for prostate cancer is not yet standard clinical practice, research from selected cohorts has reported that prostate cancer risk is elevated in men with mutations in BRCA1, BRCA2, and on a smaller scale, in mismatch repair (MMR) genes. Since clinical genetic testing is available for these genes, information about risk of prostate cancer based on alterations in these genes is included in this section. In addition, mutations in HOXB13 were reported to account for a proportion of hereditary prostate cancer. Although clinical testing is not yet available for HOXB13 alterations, it is expected that this gene will have clinical relevance in the future and therefore is also included in this section. The genetic alterations described in this section require further study and are not to be used in routine clinical practice at this time.
Studies of male BRCA1 and BRCA2 mutation carriers demonstrate that these individuals have a higher risk of prostate cancer and other cancers.
BRCAmutation–associated prostate cancer risk
The risk of prostate cancer in BRCA mutation carriers compared with noncarriers has been studied in various settings.
Among male BRCA1 mutation carriers from hereditary breast and ovarian cancer kindreds studied by the Breast Cancer Linkage Consortium (BCLC) family set, the risk of prostate cancer was not elevated overall ([relative risk] RR, 1.1; 95% CI, 0.8–1.5); however, the risk was modestly increased among men younger than 65 years (RR, 1.8; 95% CI, 1.0–3.3).
A similar study of male BRCA2 mutation carriers in hereditary breast and ovarian cancer kindreds from the BCLC demonstrated that the risk of prostate cancer associated with BRCA2 mutations was increased overall (RR, 4.7; 95% CI, 3.5–6.2). The incidence was also markedly increased among men younger than 65 years at diagnosis (RR, 7.3; 95% CI, 4.7–11.5). Another report from the BCLC suggests that prostate cancer risk may be lower among men with a mutation in the central region of the BRCA2 gene, known as the ovarian cancer cluster region (RR, 0.5; 95% CI, 0.2–1.0).
In an effort to clarify the relationship between BRCA1/BRCA2 and prostate cancer risk, 215 BRCA1 mutation–positive families and 188 BRCA2 mutation–positive families were studied. One hundred fifty-eight of these men were diagnosed with prostate cancer, eight of whom were known to carry their family's BRCA1 mutation, and 20 of whom were known to carry a BRCA2 mutation. Archival pathology material (paraffin blocks) was retrievable from four men with a BRCA1 mutation and 14 men with a BRCA2 mutation. LOH was observed at the BRCA2 locus in 10 of 14 BRCA2-related prostate cancers versus 0 of 4 BRCA1-related prostate cancers (P = .02). BRCA2 mutation carriers were estimated to have a 3.5-fold increased prostate cancer risk, while BRCA1 mutation carriers did not appear to be at increased risk. These observations are consistent with the hypothesis that BRCA2, but not BRCA1, is a tumor-suppressor gene related to prostate cancer risk. The absence of mutation information on the 130 unstudied cases limits the value of this observation. A review of the relationship between germline mutations in BRCA2 and prostate cancer risk supports the view that this gene confers a significant increase in risk among male members of hereditary breast and ovarian cancer families but that it likely plays only a small role, if any, in site-specific, multiple-case prostate cancer families. To further assess this conclusion, a U.K. study screened 1,832 prostate cancer cases for mutations in BRCA2 using multiplex fluorescence heteroduplex detection. Cases consisted of 1,589 men diagnosed before age 65 years and 243 men diagnosed after age 65 years who reported one or more first-degree relatives with prostate cancer. The prevalence of BRCA2 deleterious mutations in men diagnosed before age 65 years was 1.2%, conferring an absolute risk of prostate cancer of 15% by age 65 years. No mutations were identified in men diagnosed after age 65 years, suggesting that BRCA2 mutations may be associated with earlier-onset prostate cancer.
Prevalence ofBRCAfounder mutations in men with prostate cancer
Several studies in Israel and in North America have analyzed the frequency of BRCAfounder mutations among Ashkenazi men with prostate cancer.[8,9,10] Two specific BRCA1 mutations (185delAG and 5382insC) and one BRCA2 mutation (6174delT) are common in individuals of Ashkenazi Jewish (AJ) ancestry. Carrier frequencies for these mutations in the general Jewish population are 0.9% (95% CI, 0.7–1.1) for the 185delAG mutation, 0.3% (95% CI, 0.2–0.4) for the 5382insC mutation, and 1.3% (95% CI, 1.0–1.5) for the BRCA2 6174delT mutation.[11,12,13,14] (Refer to the High-Penetrance Breast and/or Ovarian Cancer Susceptibility Genes section of the PDQ summary on Genetics of Breast and Ovarian Cancer for more information about BRCA1 and BRCA2 genes.) In these studies, the RRs were commonly greater than 1, but only a few have been statistically significant. Many of these studies were not sufficiently powered to rule out a lower, but clinically significant, risk of prostate cancer in carriers of Ashkenazi BRCA founder mutations.
In the Washington Ashkenazi Study (WAS), a kin-cohort analytic approach was used to estimate the cumulative risk of prostate cancer among more than 5,000 American AJ male volunteers from the Washington, District of Columbia area who carried one of the BRCA Ashkenazi founder mutations. The cumulative risk to age 70 years was estimated to be 16% (95% CI, 4–30) among carriers and 3.8% among noncarriers (95% CI, 3.3–4.4). This fourfold increase in prostate cancer risk was equal (in absolute terms) to the cumulative risk of ovarian cancer among female mutation carriers at the same age (16% by the age of 70 years; 95% CI, 6–28). The risk of prostate cancer in male mutation carriers in the WAS cohort was elevated by age 50 years, was statistically significantly elevated by age 67 years, and increased thereafter with age, suggesting both an overall excess in prostate cancer risk and an earlier age at diagnosis among carriers of Ashkenazi founder mutations. Prostate cancer risk differed depending on the gene, with BRCA1 mutations associated with increasing risk after age 55 to 60 years, reaching 25% by age 70 years, and 41% by age 80 years. In contrast, prostate cancer risk associated with the BRCA2 mutation began to rise at later ages, reaching 5% by age 70 years and 36% by age 80 years (numeric values were provided by the author [written communication, April 2005]).
The studies summarized in Table 2 used similar case/control methods to examine the prevalence of Ashkenazi founder mutations among Jewish men with prostate cancer and found an overall positive association between founder mutation status and prostate cancer risk.
These studies support the hypothesis that prostate cancer occurs excessively among carriers of AJ founder mutations and suggest that the risk may be greater among men with the BRCA2 founder mutation (6174delT) than among those with one of the BRCA1 founder mutations (185delAG; 5382insC). The magnitude of the BRCA2-associated risks differ somewhat, undoubtedly because of interstudy differences related to participant ascertainment, calendar time differences in diagnosis, and analytic methods. Some data suggest that BRCA-related prostate cancer has a significantly worse prognosis than prostate cancer that occurs among noncarriers.
Three Polish BRCA1 founder mutations (C16G, 4153delA, and 5382insC) were studied in 1,793 Polish prostate cancer cases and 4,570 controls. Overall, the prevalence of the three mutations combined was identical in cases and controls. However, the most common mutation, 5382insC, occurred in 0.06% of cases versus 0.37% of controls, suggesting that this specific variant is not likely to be associated with increased prostate cancer risk. Furthermore, the presence of either of the other two mutations (C16G and 4153delA) was associated with a 3.6-fold increase in prostate cancer risk (P = .045) and an even greater risk (OR, 12; P = .0004) of familial prostate cancer. These data suggest that prostate cancer risk in BRCA1 mutation carriers varies with the location of the mutation (i.e., there is a correlation between genotype and phenotype). This observation might explain some of the inconsistencies encountered in prior studies of this association, since populations may have varied relative to the proportion of persons with specific pathogenic BRCA1 mutations.
Two hundred sixty-three men with prostate cancer diagnosed in the United Kingdom before age 56 years underwent testing for BRCA2 mutations. Screening of all coding regions resulted in the identification of six men (2.3%) with protein-truncating BRCA2 mutations and an additional 22 men harboring variants of undetermined significance. Three of the men with deleterious mutations had no family history of prostate, breast, or ovarian cancers. Using estimates of the frequency of BRCA2 mutations in the general U.K. population of 0.14% and 0.12%, the investigators estimated a 23-fold increase in the RR of early-onset prostate cancer attributable to BRCA2 mutations (95% CI, 9–57). In a similar study conducted in a U.S. population, 290 men (11% African American and 87% white) diagnosed with prostate cancer prior to age 55 years and unselected for family history were screened for BRCA2 mutations. Two protein-truncating BRCA2 mutations were identified for a prevalence of 0.69% (95% CI, 0.08–2.49). Both mutations were found in white cases, for a prevalence in whites of 0.78% (95% CI, 0.09–2.81) and a RR of 7.8 (95% CI, 1.8–9.4) for prostate cancer in white BRCA2 mutation carriers. Of the two individuals with a protein-truncating mutation, neither reported a family history of breast cancer or ovarian cancer. This study confirms that on rare occasions, germline mutations in BRCA2 account for some cases of early-onset prostate cancer, although this is estimated to be less than 1% of early-onset prostate cancers in the United States.
Genomic DNA of 266 subjects from 194 HPC families was screened for BRCA2 mutations using sequence analysis focusing on exonic and preserved regulatory regions. Although a number (n = 31) of nonsynonymous variations were identified, no truncating or deleterious mutations were detected. These investigators concluded that BRCA2 mutations did not significantly contribute to hereditary prostate cancer. A genome-wide scan for HPC using 175 families from the UM-PCGP found evidence for linkage to chromosome 17q markers. The maximum LOD score in all families was 2.36, and the LOD score increased to 3.27 when only those families with four or more confirmed affected men were analyzed. The linkage peak was centered over the BRCA1 gene. In follow-up, these investigators screened the entire BRCA1 gene for mutations using DNA from one individual from each of 93 pedigrees with evidence of prostate cancer linkage to 17q markers. Sixty-five of the individuals screened had wild-type BRCA1 sequence, and only one individual from a family with prostate and ovarian cancers was found to have a truncating mutation (3829delT). The remainder of the individuals harbored one or more germline BRCA1 variants, including 15 missense variants of uncertain clinical significance. The conclusion from these two reports is that there is evidence for a prostate cancer susceptibility gene on chromosome 17q near BRCA1; however, large deleterious inactivating mutations in BRCA1 are not likely to be associated with prostate cancer risk in chromosome 17-linked families.
In another study from the UM-PCGP, common genetic variation in BRCA1 was examined. Conditional logistic regression analysis and family-based association tests were performed in 323 familial and early-onset families, which included 817 men with and without prostate cancer, to investigate the association of SNPs tagging common haplotype variation in a 200-kilobase (kb) region surrounding and including BRCA1. Three SNPs in BRCA1 (rs1799950, rs3737559, and rs799923) were found to be associated with prostate cancer. The strongest association was observed for SNP rs1799950 (OR, 2.25; 95% CI, 1.21–4.20), which leads to a glutamine-to-arginine substitution at codon 356 (Gln356Arg) of exon 11 of BRCA1. Furthermore, SNP rs1799950 was found to contribute to the linkage signal on chromosome 17q21 originally reported by the UM-PCGP.
Prostate cancer aggressiveness inBRCAmutation carriers
A founder mutation in BRCA2 (999del5 in exon 9), which was originally described in male and female breast cancer families in Iceland, has been reported to be associated with aggressive prostate cancer in multiple small studies.[26,27,28,29,30,31] A population-based case-control study of BRCA2 999del5 mutation carriers and noncarriers (all of whom had a prostate cancer diagnosis) from the Icelandic Cancer Registry was conducted. Of 596 prostate cancer patients from Iceland with prostate tissues available for pathology review, 527 had genetic analysis performed. Thirty patients carrying this BRCA2 mutation were identified and matched to 59 noncarriers by year of diagnosis and year of birth. The results showed that mutation carriers had lower mean ages of prostate cancer diagnosis, advanced tumor stage, higher tumor grade, and shorter median survival than noncarriers. Carrying the BRCA2 999del5 mutation was associated with a higher risk of death from prostate cancer (hazard ratio [HR], 3.42; 95% CI, 2.12–5.51), which remained after adjustment for stage and grade (HR, 2.35; 95% CI, 1.08–5.11). These investigators concluded that the Icelandic BRCA2 999del5 founder mutation was associated with aggressive prostate cancer. Their observations differ from similar analyses of BRCA-related prostate cancer in other population groups and may be specific for the Icelandic founder mutation.
The relationship between the AJ founder mutations in BRCA1 and BRCA2 and prostate cancer aggressiveness was also evaluated in 979 prostate cancer cases and 1,251 controls. A significant increase in the risk of prostate cancer was observed in men with high-grade (Gleason score ≥7) prostate cancers with both BRCA2-6174delT (OR, 3.18; 95% CI, 1.37–7.34) and BRCA1-185delAG (OR, 3.54; 95% CI, 1.22–10.31) mutations. These findings suggest that the previously reported relationship between the AJ founder mutations and prostate cancer risk may be accounted for by high-grade prostate cancers. These observations require confirmation in additional studies because the design of the current report (nationwide volunteers recruited through the mail) leaves open the possibility of ascertainment bias.
BRCA1/BRCA2and survival outcomes
Analysis of prostate cancer cases in families with known BRCA1 or BRCA2 mutations have been examined for survival. In an unadjusted analysis, median survival was 4 years in 183 men with prostate cancer with a BRCA2 mutation and 8 years in 119 men with a BRCA1 mutation. The study suggests that BRCA2 mutation carriers have a poorer survival than BRCA1 mutation carriers. To further assess this observation, two cohorts of men with prostate cancer were studied: cohort 1 included 263 men with prostate cancer diagnosed before age 55 years, of which six (2.3%) were found to have a BRCA2 mutation, and a control group of 1,587 prostate cancer cases from a single clinic matched for age and stage; cohort 2 included BRCA2 carriers diagnosed with prostate cancer at any age, as ascertained through a genetics clinic. The median overall survival of all BRCA2 mutation–positive prostate cancer cases from both cohorts was 4.8 years, in contrast to 8.5 years in non-mutation carriers (HR, 2.14; 95% CI, 1.28–3.56; P = .003). When each cohort was analyzed separately, median survival in cohort 1 was 3.6 years (HR, 3.36; 95% CI, 1.50–7.50; P = .002); median survival in cohort 2 was 5 years. In both univariate and multivariate analyses, germline BRCA2 mutation status was an independent prognostic factor.
One study examined BRCA founder mutation prevalence in 832 AJ men with prostate cancer and 454 controls. Among the cases, 26 mutation carriers and 806 noncarriers were identified. BRCA2-related prostate cancers were significantly more likely to present with a Gleason score of at least 7 (85% vs. 57%, P = .0002). After adjusting for disease stage, PSA, Gleason score, and therapy received, mutation carriers had significantly greater risk of recurrence (BRCA1: HR, 2.41; 95% CI, 1.23-4.75 and BRCA2: HR, 4.32; 95% CI, 1.31–13.62) and prostate cancer–specific death (BRCA1: HR, 5.16; 95% CI, 1.09–24.53 and BRCA2: HR, 5.48; 95% CI, 2.03–14.79) than noncarriers.
The effect of BRCA mutation status on prostate cancer survival was evaluated in a study of 148 men from 130 families at a high risk of breast cancer. Eligible men were either known mutation-positive or known mutation-negative cases from mutation-positive families. There were too few BRCA1 carriers available to permit their being analyzed. BRCA2 carriers were shown to have an increased risk of death (all causes combined) and prostate-specific cancer mortality (HR, 4.5; 95% CI, 2.12–9.52; P = 8.9 × 10-5) than noncarrier controls. The BRCA2-related prostate cancer cases presented with higher Gleason scores and more advanced tumor stage than did controls. There were no differences in age at diagnosis between carrier and noncarrier cases. This is the largest retrospective study of confirmed BRCA2 mutation carriers with proven prostate cancer in a population that is generally not undergoing routine PSA-based prostate cancer screening. Of further interest, the noncarrier cases from these mutation-positive families had a significantly worse prognosis than did cases from the general population, a novel finding that requires confirmation. The authors concluded that all men (both mutation carriers and noncarriers) from BRCA mutation–positive families are at risk of developing clinically aggressive prostate cancer, a finding that warrants discussion with family members as they undergo cancer risk assessment and genetic counseling.
Mismatch Repair Genes
There are four genes implicated in mismatch repair (MMR), namely MLH1, MSH2, MSH6, and PMS2. Germline mutations in four of the genes implicated in MMR have been associated with Lynch syndrome, which manifests by cases of nonpolyposis colorectal cancer and a constellation of other cancers in the families, including endometrial, ovarian, and duodenal cancers and transitional cell cancers of the ureter and renal pelvis. Scattered case reports have suggested that prostate cancer may be observed in men harboring an MMR gene mutation. The first quantitative study described nine cases of prostate cancer occurring in a population-based cohort of 106 Norwegian male MMR mutation carriers or obligate carriers. The expected number of cases among these 106 men was 1.52 (P < .01); the men were younger at the time of diagnosis (60.4 years vs. 66.6 years, P = .006) and had more evidence of Gleason score of 8 to 10 (P < .00001) than the cases from the Norwegian Cancer Registry. Kaplan Meier analysis revealed that the cumulative risk of prostate cancer diagnosis by age 70 years was 30% in MMR gene mutation carriers and 8% in the general population. This finding awaits confirmation in additional populations. A population-based case-control study examined haplotype-tagging SNPs in three MMR genes (MLH1, MSH2, and PMS2). This study provided some evidence supporting the contribution of genetic variation in MLH1 and overall risk of prostate cancer. To assess the contribution of prostate cancer as a feature of Lynch Syndrome, one study performed microsatellite instability (MSI) testing on prostate cancer tissue blocks from families enrolled in a prostate cancer family registry who also reported a history of colon cancer. Among 35 tissue blocks from 31 distinct families, two tumors from MMR mutation–positive families were found to be MSI-high. The authors conclude that MSI is rare in hereditary prostate cancer.
Linkage to 17q21-22 was initially reported by the University of Michigan Prostate Cancer Genetics Project (UM-PCGP) from 175 pedigrees of families with hereditary prostate cancer. Fine-mapping of this region provided strong evidence of linkage (LOD score = 5.49) and a narrow candidate interval (15.5 Mb) for a putative susceptibility gene among 147 families with four or more affected men and average age at diagnosis of 65 years or younger. The exons of 200 genes in the 17q21-22 region were sequenced in DNA from 94 unrelated patients from hereditary prostate cancer families (from the UM-PCGP and Johns Hopkins). Probands from four families were discovered to have a recurrent mutation (G84E) in HOXB13, and 18 men with prostate cancer from these four families carried the mutation. The mutation status was determined in 5,083 additional case subjects and 2,662 control subjects. Carrier frequencies and odds ratios for prostate cancer risk were as follows:
Additional rare variants in HOXB13 were also observed. Penetrance estimates of the G84E mutation in HOXB13 are under study. HOXB13 plays a role in prostate development and binds to the androgen receptor; however, the mechanism by which it contributes to the pathogenesis of prostate cancer remains unknown. This is the first gene proven to account for a fraction of hereditary prostate cancer, particularly early-onset prostate cancer, but the clinical utility of testing for this mutation has not yet been defined.
Various research methods have been employed to uncover the landscape of genetic variation associated with prostate cancer. Specific methodologies inform of unique phenotypes or inheritance patterns. The sections below describe prostate cancer research utilizing various methods to highlight their role in uncovering the genetic basis of prostate cancer. In an effort to identify disease susceptibility genes, linkage studies are typically performed on high-risk extended families in which multiple cases of a particular disease have occurred. Typically, gene mutations identified through linkage analyses are rare in the population, highly penetrant in families, and have large effect sizes. The clinical role of mutations that are identified in linkage studies is a clearer one, establishing precedent for genetic testing for cancer with genes such as BRCA1 and BRCA2. Genome-wide association studies (GWAS) are another methodology used to identify candidate loci associated with prostate cancer. Genetic variants identified from GWAS typically are common in the population and have modest effect sizes for prostate cancer risk. The clinical role of markers identified from GWAS is an active area of investigation. Case-control studies are useful in validating the findings of linkage studies and GWAS as well as for studying candidate gene alterations for association with prostate cancer risk, although the clinical role of findings from case-control studies needs to be further defined.
Introduction to linkage analyses
The recognition that prostate cancer clusters within families has led many investigators to collect multiple-case families with the goal of localizing prostate cancer susceptibility genes through linkage studies.
Linkage studies are typically performed on high-risk kindreds in whom multiple cases of a particular disease have occurred in an effort to identify disease susceptibility genes. Linkage analysis statistically compares the genotypes between affected and unaffected individuals and looks for evidence that known genetic markers are inherited along with the disease trait. If such evidence is found (linkage), it provides statistical data that the chromosomal region near the marker also harbors a disease susceptibility gene. Once a genomic region of interest has been identified through linkage analysis, additional studies are required to prove that there truly is a susceptibility gene at that position. Linkage analysis is affected by the following:
Furthermore, because a standard definition of hereditary prostate cancer (HPC) has not been accepted, prostate cancer linkage studies have not used consistent criteria for enrollment. One criterion that has been proposed is the Hopkins Criteria, which provides a working definition of HPC families. Using the Hopkins Criteria, kindreds with prostate cancer need to fulfill only one of following criteria to be considered to have HPC:
Using these criteria, surgical series have reported that approximately 3% to 5% of men will be from a family with HPC.[2,3]
An additional issue in linkage studies is the high background rate of sporadic prostate cancer in the context of family studies. As a man's lifetime risk of prostate cancer is one in six, it is possible that families under study have men with both inherited and sporadic prostate cancer. Thus, men who do not inherit the prostate cancer susceptibility gene that is segregating in their family may still develop prostate cancer. Currently there are no clinical or pathological features of prostate cancer that will allow differentiation between inherited and sporadic forms of the disease. Similarly, there are limited data regarding the clinical phenotype or natural history of prostate cancer associated with specific candidate loci. Measurement of the serum prostate-specific antigen (PSA) has been used inconsistently in evaluating families used in linkage analysis studies of prostate cancer. In linkage studies, the definition of an affected man can be biased by the use of serum PSA screening as the rates of prostate cancer in families will differ between screened and unscreened families.
One way to address inconsistencies between linkage studies is to require inclusion criteria that defines clinically significant disease (e.g., Gleason score ≥7, PSA ≥20 ng/mL) in an affected man.[5,6,7] This approach attempts to define a homogeneous set of cases/families to increase the likelihood of identifying a linkage signal. It also prevents the inclusion of cases that may be considered clinically insignificant that were identified by screening in families.
Investigators have also incorporated clinical parameters into linkage analyses with the goal of identifying genes that may influence disease severity.[8,9] This type of approach, however, has not yet led to the identification of consistent linkage signals across datasets.[10,11]
Candidate genes and susceptibility loci identified in linkage analyses
Refer to the HOXB13 section in the Genes with Potential Clinical Relevance in Prostate Cancer Risk section of this summary.
Additional prostate cancer susceptibility loci identified in linkage analyses
Table 3 summarizes the proposed prostate cancer susceptibility loci identified in families with multiple prostate cancer–affected individuals. Conflicting evidence exists regarding the linkage to some of the loci described above. Data on the proposed phenotype associated with each locus are also limited, and the strength of repeated studies is needed to firmly establish these associations. Evidence suggests that many of these prostate cancer loci account for disease in a small subset of families, which is consistent with the concept that prostate cancer exhibits locus heterogeneity.
Other regions identified by linkage studies
Genome-wide linkage studies of families with prostate cancer have identified several other loci that may harbor prostate cancer susceptibility genes, emphasizing the underlying complexity and genetic heterogeneity of this cancer. The chromosomal regions with modest-to-strong statistical significance (LOD score ≥2) include the following chromosomes:
Linkage analyses in population subgroups
Linkage studies have also been performed in specific populations or with specific clinical parameters to identify population-specific susceptibility genes or genes influencing disease phenotypes.
Linkage analysis in African American families
The African American Hereditary Prostate Cancer study conducted a genome-wide linkage study of 77 families with four or more affected men. Multipoint heterogeneity LOD (hLOD) scores of 1.3 to less than 2.0 were observed using markers that map to 11q22, 17p11, and Xq21. Analysis of the 16 families with more than six men with prostate cancer provided evidence for two additional loci: 2p21 (multipoint hLOD score = 1.08) and 22q12 (multipoint hLOD score = 0.91).[91,97] A smaller linkage study that included 15 African American hereditary prostate cancer families from the southeastern and southcentral Louisiana region identified suggestive linkage for prostate cancer at 2p16 (hLOD = 1.97) and 12q24 (hLOD = 2.21) using a 6,000 SNP platform. Further study including a larger number of African American families is needed to confirm these findings.
Linkage analysis in families with aggressive prostate cancer
In an effort to identify loci contributing to prostate cancer aggressiveness, linkage analysis was performed in families with one or more of the following: Gleason grade 7 or higher, PSA of 20 ng/mL or higher, regional or distant cancer stage at diagnosis, or death from metastatic prostate cancer < before age 65 years. One hundred twenty-three families with two or more affected family members with aggressive prostate cancer were studied. Suggestive linkage was found at chromosome 22q11 (hLOD score = 2.18) and 22q12.3-q13.1 (hLOD score = 1.90). These findings suggest that using a clinically defined phenotype may facilitate finding prostate cancer susceptibility genes. A fine-mapping study of 14 extended high-risk prostate cancer families has subsequently narrowed the genomic region of interest to an 880-kb region at 22q12.3. An analysis of high-risk pedigrees from Utah provides an overview of this strategy. A linkage analysis utilizing a higher resolution marker set of 6,000 SNPs was performed among 348 families from the International Consortium for Prostate Cancer Genetics with aggressive prostate cancer. Aggressive disease was defined as Gleason score 7 or higher, invasion into seminal vesicles or extracapsular extension, pretreatment PSA level of 20 ng/ml or higher, or death from prostate cancer. The region with strongest evidence of linkage among aggressive prostate cancer families was 8q24 with LOD scores of 3.09–3.17. Additional regions of linkage included with LOD scores of 2 or higher included 1q43, 2q35, and 12q24.31. No candidate genes have been identified.
Linkage analysis in families with multiple cancers
In light of the multiple prostate cancer susceptibility loci and disease heterogeneity, another approach has been to stratify families based on other cancers, given that many cancer susceptibility genes are pleiotropic. A genome-wide linkage study was conducted to identify a susceptibility locus that may account for both prostate cancer and kidney cancer in families. Analysis of 15 families with evidence of HPC and one or more cases of kidney cancer (pathologically confirmed) in a man with prostate cancer or in a first-degree relative of a man with prostate cancer revealed suggestive linkage with markers that mapped to an 8 cM region of chromosome 11p11.2-q12.2. This observation awaits confirmation. Another genome-wide linkage study was conducted in 96 HPC families with one or more first-degree relatives with colon cancer. Evidence for linkage in all families was found in several regions, including 11q25, 15q14, and 18q21. In families with two or more cases of colon cancer, linkage was also observed at 1q31, 11q14, and 15q11-14.
Summary of prostate cancer linkage studies
Linkage to chromosome 17q21-22 and subsequent fine-mapping and exome sequencing have identified recurrent mutations in the HOXB13 gene to account for a fraction of hereditary prostate cancer, particularly early-onset prostate cancer. The clinical utility of testing for HOXB13 mutations has not yet been defined. Furthermore, many linkage studies have mapped several prostate cancer susceptibility loci (Table ), although the genetic alterations contributing to hereditary prostate cancer from these loci have not been consistently reproduced. With the evolution of high-throughput sequencing technologies, there will likely be additional highly penetrant genetic mutations identified to account for subsets of hereditary prostate cancer families.
A case-control study involves evaluating factors of interest for association to a condition. The design involves investigation of cases with a condition of interest, such as a specific disease or gene mutation, compared with a control sample without that condition, but often with other similar characteristics (i.e., age, gender, and ethnicity). Limitations of case-control design with regard to identifying genetic factors include the following:[112,113]
Additionally, identified associations may not always be valid, but they could represent a random association and, therefore, warrant validation studies.[112,113]
Refer to the BRCA1 and BRCA2 section in the Genes with Potential Clinical Relevance in Prostate Cancer Risk section of this summary.
The EMSY gene is located on chromosomal locus 11q13.5 and is in an area of both linkage and association by genome-wide association studies (GWAS).[41,115] This gene has also been shown to interact with and inhibit the activity of BRCA2. A study from Finland screened the EMSY gene for sequence variants and evaluated their association with prostate cancer risk in a population-based case-control study including 923 controls, 184 familial cases, and 2,301 unselected prostate cancer cases. An intronic variant (IVS6-43A>G) was associated with increased risk of prostate cancer in familial cases (OR, 7.5; 95% CI, 1.3–45.5; P = .02). This variant was also associated with increased risk of aggressive prostate cancer (PSA ≥20 or Gleason score ≥7) in cases unselected for family history (OR, 6.5; 95% CI, 1.5–28.4; P = .002). Validation of this finding with association to other measures of disease aggressiveness (e.g., prostate cancer–specific mortality) is needed. The functional consequence of this intronic variant also needs to be explored for insight into the role of this gene in susceptibility to aggressive disease.
Mismatch repair genes
Refer to the Mismatch Repair Genes section in the Genes with Potential Clinical Relevance in Prostate Cancer Risk section of this summary.
The tumor suppressor gene Kruppel-like factor 6 (KLF6), located on chromosome 10p15, is a zinc finger transcription factor potentially associated with prostate cancer risk. Somatic mutations and allelic loss of KLF6 have been found in tumors of several primary neoplasms, including prostate cancer. A germline mutation in KLF6 (IVS1-27G>A) appears to have a novel mechanism of gene inactivation: generation of alternatively spliced products that antagonize wild-type gene function. However, data are inconsistent regarding the association of germline mutations in KLF6 and hereditary prostate cancer. A Finnish study of 69 prostate cancer families did not identify an association between KLF6 mutations and prostate cancer susceptibility. The germline KLF6 SNP described above, IVS1-27G>A, was found to increase the RR of prostate cancer in a U.S. study of 3,411 men (RR, 1.61; 95% CI, 1.20–2.16; P = .01). However, the prostate cancer risk associated with the IVS1-27G>A SNP was not detected in a study of 300 Jewish prostate cancer families. In fact, the A allele, which was previously shown to be more common in U.S. men with prostate cancer and associated with the creation of splice variants, was significantly less common among cases than among controls in the Israeli study (49 of 804 alleles in cases and 55 of 600 control alleles; P = .030).
The alpha methylacyl-CoA racemase (AMACR) gene, located at 5p13.3, encodes a protein that is localized to peroxisomes and mitochondria and plays an important role in the metabolism of branch-chained fatty acids. The protein has been shown to be overexpressed in many cancers including prostate cancer. AMACR resequencing experiments using DNA from probands in HPC families were conducted. From the 17 sequence variants identified, 11 SNPs were selected for genotyping in 159 HPC probands, 245 sporadic prostate cancer cases, and 211 controls. Several variants (including M9V, G1157D, S291L, and K277E) were shown to be associated with HPC (but not sporadic prostate cancer). A haplotype-tagging strategy was used to test for association between genetic variation in AMACR and prostate cancer in a set of siblings discordant for prostate cancer who participated in a research study focused on early-onset prostate cancer and/or HPC. An association was found for SNP rs3195676 (M9V) with an OR of 0.58 (95% CI, 0.38–0.90, P = .01 for a recessive model). The reported magnitude and direction of the association observed for this SNP were similar among this study and previously mentioned AMACR resequencing experiments. A nested case-control study was conducted using samples from the screening arm of the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO-1) to test for potential association between seven AMACR SNPs , including M9V, and prostate cancer. No association was detected between any of the SNPs and prostate cancer. The prostate cancer cases in the PLCO study are all older than 55 years and not specifically enriched for family history. In the same study, risk of prostate cancer was reduced in men who reported using ibuprofen who also had specific alleles in four SNPs (M9V, D175G, S201L, and K77E) or a specific six-SNP haplotype. Ibuprofen mediates its anti-inflammatory effect through COX2 inhibition; AMACR contributes to the conversion of the COX-inactive to the COX-active form of ibuprofen. This observation suggests that these AMACR SNPs may alter enzyme function, although experiments have not been conducted to directly test this hypothesis.
Other potential prostate cancer genes
Individuals who were heterozygous for one of the Nijmegen breakage syndrome (NBS) founder mutations identified in Poland may be at increased risk of prostate cancer. NBS is a rare autosomal-recessive cancer susceptibility disorder of childhood that is characterized by growth retardation, facial dysmorphism, immunodeficiency, and a predisposition to lymphoma and leukemia in patients with germline biallelic (e.g., homozygous) mutations. NBS1, located on chromosome 8q, has an important role in DNA repair and is part of the ataxia telangiectasia pathway. Recent observations have suggested that there may be an increased risk of cancer among heterozygous carriers of mutations in a number of genes involved in response to DNA damage, such as xeroderma pigmentosum  and ataxia telangiectasia.[127,128] Polish investigators analyzed the prevalence of an NBS1 founder mutation in a sample of 56 men with familial prostate cancer, 305 men with sporadic cancer, and control subjects who included men, women, and newborns. Cases with a positive family history were 16 times more likely to be mutation carriers than were controls (P < .0001). LOH was commonly observed in mutation-associated prostate cancers, with preferential loss of the wild-type allele. A collaborative report from five groups participating in the International Consortium for Prostate Cancer Genetics demonstrated a carrier frequency of 0.22% (2 of 909) for probands with familial prostate cancer and 0.25% (3 of 1,218) for men with sporadic cancer for the founder 657del5 mutation. Although this mutation was not detected in any of the 293 unaffected family members, the low frequency of the founder mutation suggests that NBS1 mutations do not contribute to a significant proportion of prostate cancer cases.
In the first report of possible germline CHEK2 variants in men with prostate cancer, mutations were identified in 4.8% of 578 prostate cancer patients and in none of 423 unaffected men. Nine of 149 multiplex prostate cancer families were also found to have germline CHEK2 mutations. The I157T substitution was detected in equal numbers of cases and controls and thus was reported to likely represent a polymorphism. Functional studies of additional identified variants revealed substantial reductions in CHEK2 protein levels and/or other functional changes that suggest CHEK2 mutations contribute to prostate carcinogenesis.[130,131] Subsequently, Polish investigators sequenced the CHEK2 gene in 140 patients with prostate cancer and then analyzed the three detected variants in a larger series of prostate cancer cases and controls. CHEK2 truncating mutations were identified in 9 of 1,921 controls (0.5%) and in 11 of 690 (1.6%) unselected patients with prostate cancer (OR, 3.4; P = .004). These same mutations were also found in 4 of 98 familial prostate cases (OR, 9.0; P = .0002). The I157T missense variant was also more frequent in men with prostate cancer (7.8%) than in controls (4.8%) (OR, 1.7; P = .03) and was identified in 16% of men with familial prostate cancer (OR, 3.8; P = .00002). LOH was not observed in any of the five men with truncating CHEK2 mutations. A follow-up to this study has been reported from Poland with 1,864 prostate cancer patients and 5,496 controls. All three founder mutations and a large germline deletion of exons 9 and 10 (5395-bp deletion) were genotyped. The truncating mutation 1100delC was identified in 14 of 1,864 (0.8%) unselected prostate cancer cases and 3 of 249 (1.2%) familial cases (OR, 3.5; P = .002 and OR, 5.6; P = .02, respectively). A significant association with another truncating mutation (IVS2+1G→A) was identified in 5 of 249 (2.0%) familial cases that had the mutation (OR, 5.1; P = .002). The missense mutation I157T was detected in 142 of 1,864 (7.6%) unselected prostate cancer cases and in 30 of 249 (12%) familial cases (OR, 1.6; P < .001 and OR, 2.7; P < .001, respectively). The large deletion in exons 9 and 10 accounted for 4 of 249 (1.6%) familial cases (OR, 3.7; P = .03). Overall, it appears that there are at least four founder mutations in the CHEK2 gene, which account for an estimated 7% of patients with prostate cancer in the Polish population. The most common missense mutation is I157T, and the most common truncating mutation is 5395-bp deletion. These reports suggest that truncating and missense mutations in CHEK2 may play a role in prostate cancer susceptibility. However, a recent molecular analysis designed specifically to assess the role of seven different CHEK2 coding variants (including 1100delC) in AJ men with prostate cancer, suggested that germline mutations in this gene have a minor role, if any role at all, in modifying the risk of prostate cancer in AJ men. This conclusion is limited by the relatively small number of individuals in whom CHEK2 sequencing was performed.
Table 4 summarizes the candidate genes for prostate cancer susceptibility, their chromosomal location, and availability of clinical testing.
To summarize, studies to date have mapped site-specific prostate cancer susceptibility loci to chromosomes 1q25 (HPC1), 1q42.2–43 (PCAP), 1p36 (CAPB), Xq27–2 (HPCX), 20q13 (HPC20), 17p (ELAC2/HPC2), and 8p. Other studies have suggested that prostate cancer may be part of the cancer spectrum of syndromes that include a more diverse set of malignancies, such as seems to be the case for BRCA2 and, perhaps, BRCA1. Both linkage and candidate gene studies have been complicated by the later-onset nature of the disease and the high background rate of prostate cancer in the general population. In addition, there is likely to be real, extensive locus heterogeneity for HPC, as suggested by both segregation and linkage studies. In this respect, HPC resembles a number of the other major adult-onset hereditary cancer syndromes, in which more than one gene can produce the same or very similar clinical phenotype (e.g., hereditary breast/ovarian cancer, Lynch syndrome, hereditary melanoma, and hereditary renal cancer).
Linkage studies may be used to evaluate the possibility that an HPC gene might exist in a particular family, but this analytic approach is currently being done only in the research setting. Until the specific genes and mutations involved are identified with their associated phenotype defined, it is difficult to establish the analytical validity of this approach. Without a validated laboratory test, clinical validity and clinical utility cannot be measured. At present, clinical germline mutation testing for most HPC susceptibility loci is not available. In addition, the clinical validity and utility of BRCA testing solely based on evidence for HPC susceptibility has not been established.
Other Regions Identified by Admixture Studies
Admixture mapping is a method used to identify genetic variants associated with traits and/or diseases that controls for population composition associated with geographically distinct ancestral groups. This approach is used when admixture occurred two or more generations ago. It is based on the availability of population-specific genetic markers associated with ancestry, and on the number of generations since admixture.[163,164] The advantage to this approach is that recent mixtures of distinct ancestral populations may have longer-range linkage disequilibrium between susceptibility alleles and genetic markers when compared with other populations. In that scenario, fewer markers are needed to search for genetic variants associated with specific diseases, such as prostate cancer. Admixture studies have identified the following chromosomal regions associated with prostate cancer:
Genome-wide Association Studies (GWAS)
Introduction to GWAS
Genome-wide searches are showing great promise in identifying common low-penetrance susceptibility alleles for many complex diseases, including prostate cancer. This approach can be contrasted with linkage analysis, which searches for genetic risk variants cosegregating within families that have a high prevalence of disease. While linkage analyses are designed to uncover rare, highly penetrant variants that segregate in predictable heritance patterns (e.g., autosomal dominant, autosomal recessive, X-linked, and mitochondrial), GWAS are best suited to identify multiple, common, low-penetrance genetic polymorphisms. GWAS are conducted under the assumption that the genetic underpinnings of complex phenotypes, such as prostate cancer, are governed by many alleles, each conferring modest risk. Most genetic polymorphisms genotyped in GWAS are common, with minor allele frequencies greater than 1% to 5% within a given population (e.g., men of European ancestry). GWAS capture a large portion of common variation across the genome.[168,169] The strong correlation between many alleles located close to one another on a given chromosome (called linkage disequilibrium) allows one to "scan" the genome without having to test all 10 million known single nucleotide polymorphisms (SNPs). GWAS can test 500,000 to 1,000,000 SNPs and ascertain almost all common inherited variants in the genome.
In a GWAS, allele frequency is compared for each SNP between cases and controls. Promising signals–in which allele frequencies deviate significantly in case and control populations–are validated in replication cohorts. In order to have adequate statistical power to identify variants associated with a phenotype, large numbers of cases and controls, typically thousands of each, are studied. Since up to 1 million SNPs are evaluated in a GWAS, false-positive findings are expected to occur frequently when using standard statistical thresholds. Therefore, stringent statistical rules are used to declare a positive finding, usually using a threshold of P < 1 × 10-7.[170,171,172]
To date, approximately 40 variants associated with prostate cancer have been identified by well-powered GWAS and validated in independent cohorts (see Table 5). These studies have revealed convincing associations between specific inherited variants and prostate cancer risk. However, the findings should be qualified with a few important considerations:
The implications of these points are discussed in greater detail below. Additional detail can be found elsewhere.
Candidate genes and susceptibility loci identified in GWAS
In 2006, two genome-wide studies seeking associations with prostate cancer risk converged on the same chromosomal locus, 8q24. Using a technique called admixture mapping, a 3.8 megabase (Mb) region emerged as significantly involved with risk in African American men. In another study, linkage analysis of 323 Icelandic prostate cancer cases also revealed an 8q24 risk locus. Detailed genotyping of this region and an association study for prostate cancer risk in three case-control populations in Sweden, Iceland, and the United States revealed specific 8q24 risk markers: a SNP, rs1447295, and a microsatellite polymorphism, allele-8 at marker DG8S737. The population-attributable risk of prostate cancer from these alleles was 8%. The results were replicated in an African American case-control population, and the population attributable risk was 16%. These results were confirmed in several large, independent cohorts.[69,70,71,72,79,80,81,82,175] Subsequent GWAS independently converged on another risk variant at 8q24, rs6983267.[72,73,74] Fine mapping, genotyping a large number of variants densely packed within a region of interest in many cases and controls, was performed across 8q24 targeting the variants most significantly associated with prostate cancer risk. Across multiple ethnic populations, three distinct 8q24 risk loci were described: region 1 (containing rs1447295) at 128.54–128.62 Mb, region 2 at 128.14–128.28 Mb, and region 3 (containing rs6983267) at 128.47–128.54 Mb. Variants within each of these three regions independently confer disease risk with ORs ranging from 1.11 to 1.66. In 2009, two separate GWAS uncovered two additional risk regions at 8q24. In all, approximately nine genetic polymorphisms, all independently associated with disease, reside within five distinct 8q24 risk regions.[85,86]
Since the discovery of prostate cancer risk loci at 8q24, other chromosomal risk loci similarly have been identified by multistage GWAS comprised of thousands of cases and controls and validated in independent cohorts. The most convincing associations reported to date for the European-American population are included in Table 5.
Clinical application of GWAS findings
Since the variants discovered by GWAS are markers of risk, there has been great interest in using genotype as a screening tool to predict the development of prostate cancer. In an attempt to determine the potential clinical value of risk SNP genotype, cases of prostate cancer (n = 2,893) were identified from four cancer registries in Sweden. Controls (n = 1,781) were randomly selected from the Swedish Population Registry and were matched to cases by age and geographic region. Known risk SNPs from 8q24, 17q12, and 17q24.3 were analyzed (rs4430796 at 17q12, rs1859962 at 17q24.3, rs16901979 at 8q24 [region 2], rs6983267 at 8q24 [region 3], and rs1447295 at 8q24 [region 1]). ORs for prostate cancer for men carrying any combination of one, two, three, or four or more genotypes associated with prostate cancer were estimated by comparing them with men carrying none of the associated genotypes using logistic regression analysis. Men who carried one to five risk alleles had an increasing likelihood of having prostate cancer compared with men carrying none of the alleles (P = 6.75 × 10-27). After controlling for age, geographic location, and family history of prostate cancer, men carrying four or more of these alleles had a significant elevation in risk of prostate cancer (OR, 4.47; 95% CI, 2.93–6.80; P = 1.20 × 10-13). When family history was added as a risk factor, men with five or more factors (five SNPs plus family history) had an even stronger risk of prostate cancer (OR, 9.46; 95% CI, 3.62–24.72; P = 1.29 × 10-8). The population-attributable risks (PARs) for these five SNPs were estimated to account for 4% to 21% of prostate cancer cases in Sweden, and the joint PAR for prostate cancer of the five SNPs plus family history was 46%.
A second study assessed prostate cancer risk associated with a family history of prostate cancer in combination with various numbers of 27 risk alleles identified through four prior GWAS. Two case-control populations were studied, the Prostate, Lung, Colon, and Ovarian Cancer Screening Trial (PLCO) in the United States (1,172 cases and 1,157 controls) and the Cancer of the Prostate in Sweden (CAPS) study (2,899 cases and 1,722 controls). The highest risk of prostate cancer from the CAPS population was observed in men with a positive family history and greater than 14 risk alleles (OR, 4.92; 95% CI, 3.64–6.64). Repeating this analysis in the PLCO population revealed similar findings (OR, 3.88; 95% CI, 2.83–5.33).
However, the proportion of men carrying large numbers of the risk alleles was low. While ORs were impressively high for this subset, they do not reflect the utility of genotyping the overall population. Receiver operating characteristic curves were constructed in these studies to measure the sensitivity and specificity of certain risk profiles. The area under the curve (AUC) was 0.61 when age, geographic region, and family history were used to assess risk. When genotype of the five risk SNPs at chromosomes 8 and 17 were introduced, a very modest AUC improvement to 0.63 was detected. The addition of more recently discovered SNPs to the model has not appreciably improved these results. While genotype may inform risk status for the small minority of men carrying multiple risk alleles, testing of the known panel of prostate cancer SNPs is currently of questionable clinical utility. This is expected to change as more risk alleles are discovered, particularly rarer alleles with higher ORs.
GWAS and insight into the mechanism of prostate cancer risk
Notably, almost all reported prostate cancer risk alleles reside in nonprotein coding regions of the genome, and the underlying biological mechanism of disease susceptibility remains unclear. Hypotheses explaining the mechanism of inherited risk include the following:
The 8q24 risk locus, which contains multiple prostate cancer risk alleles and risk alleles for other cancers, has been the focus of intense study. c-MYC, a known oncogene, is the closest known gene to the 8q24 risk regions, although it is located hundreds of kb away. Given this significant distance, SNPs within c-MYC are not in linkage disequilibrium with the 8q24 prostate cancer risk variants. One study examined whether 8q24 prostate cancer risk SNPs are in fact located in areas of previously unannotated transcription, and no transcriptional activity was uncovered at the risk loci. Attention turned to the idea of distal gene regulation. Interrogation of the epigenetic landscape at the 8q24 risk loci revealed that the risk variants are located in areas that bear the marks of genetic enhancers, elements that influence gene activity from a distance.[188,189,190] To identify a prostate cancer risk gene, germline DNA from 280 men undergoing prostatectomy for prostate cancer was genotyped for all known 8q24 risk SNPs. Genotypes were tested for association with the normal prostate and prostate tumor RNA expression levels of genes located within one Mb of the risk SNPs. No association was detected between expression of any of the genes, including c-MYC, and risk allele status in either normal epithelium or tumor tissue. Another study, using normal prostate tissue from 59 patients, detected an association between an 8q24 risk allele and the gene PVT1, downstream from c-MYC. Nonetheless, c-MYC, with its substantial involvement in many cancers, remains a prime candidate. A series of experiments in prostate cancer cell lines demonstrated that chromatin is configured in such a way that the 8q24 risk variants lie in close proximity to c-MYC, even though they are quite distant in linear space. These data implicate c-MYC despite the absence of expression data.[189,191] Further work at 8q24 and similar analyses at other prostate cancer risk loci are ongoing.
GWAS in non-European populations
Most prostate cancer GWAS data generated to date have been derived from populations of European descent. This shortcoming is profound, considering that linkage disequilibrium structure, SNP frequencies, and incidence of disease differ across ancestral groups. To provide meaningful genetic data to all patients, well-designed, adequately powered GWAS must be aimed at specific ethnic groups. Most work in this regard has focused on African American and Japanese men.
The African American population is of particular interest since American men with African ancestry are at higher risk of prostate cancer than any other group. In addition, inherited variation at the 8q24 risk locus appears to contribute to differences in African American and European American incidence of disease. A handful of studies have sought to determine whether GWAS findings in men of European ancestry are applicable to men of African ancestry. One study interrogated 28 known prostate cancer risk loci via fine mapping in 3,425 African American cases and 3,290 African American controls. On average, risk allele frequencies were 0.05 greater in African Americans than in European Americans. Of the 37 known risk SNPs analyzed, 18 replicated in the African American population were significantly associated with prostate cancer at P ≤ .05 (the study was underpowered to properly assess nine of the remaining 19 SNPs). For seven risk regions (2p24, 2p15, 3q21, 6q22, 8q21, 11q13, and 19q13), fine mapping identified SNPs in the African American population more strongly associated with risk than the index SNPs reported in the original European-based GWAS. Fine mapping of the 8q24 region revealed four SNPs associated with disease that are substantially more common in African Americans. The SNP most strongly correlated with disease among African Americans (rs6987409) is not strongly correlated with a European risk allele and may account for a portion of increased risk in the African American population. In all, the risk SNPs identified in this study are estimated to represent 11% of total inherited risk.
This analysis was followed by a GWAS to discover risk variants not previously identified in GWAS performed in other ethnicities. The GWAS was conducted in a standard multistage fashion in which 3,621 African American cases and 3,502 controls were genotyped for approximately 1 million SNPs. SNPs meeting proscribed statistical thresholds were selected for a second stage in 1,396 cases and 2,383 controls (known prostate cancer risk SNPs were excluded, as they had been rigorously analyzed, as described above). One marker–rs7210100 at chromosome 17q21–emerged and remained significant when tested in a third stage with 3,471 cases and 904 controls. When combining cases and controls from all three stages, prostate cancer risk in heterozygote and homozygote carriers of the rs7210100 risk allele was 1.49 and 2.73, respectively (P = 3.4 × 10-13). The risk allele is uncommon in African Americans (4%–7% frequency) but is virtually nonexistent in men of European ancestry. The SNP may therefore account for some ethnic difference in risk. It resides in intron 1 on the gene ZNF652. Co-expression of ZNF652 and the androgen receptor in prostate tumors has been associated with a decrease in relapse-free survival, which may suggest a mechanism of action if this variant influences expression.
Similar work has been accomplished in the Japanese population. Twenty-three candidate SNPs related to prostate cancer risk in two GWAS studies of European populations were evaluated in a relatively small population of Japanese cases (n = 311) and controls (n = 1,035). Seven of these SNPs (from five genetic loci) were associated with prostate cancer risk (OR, 1.35–1.82). Men with six or more risk alleles (27% of cases and 11% of controls) had a sixfold greater prostate cancer risk than those with two or fewer risk alleles (7% of cases and 20% of controls [OR, 6.22; P = 1.5 × 10-12]). To further assess susceptibility loci in a Japanese population, a two-stage GWAS was conducted using a total of 4,584 Japanese men with prostate cancer and 8,801 controls. The study resulted in the identification of five SNPs from five separate loci not previously associated with prostate cancer: rs13385191 at 2p24 (OR, 1.15); rs12653946 at 5p15 (OR, 1.26); rs1983891 at 6p21 (OR, 1.15); rs339331 at 6q22 (OR, 1.22); and rs9600079 at 13q22 (OR, 1.18) [data after combining cohorts from both stages of the study]. A set of nine SNPs that were nominally associated with disease risk in the initial GWAS were subsequently analyzed in other large Japanese cohorts and then united with the original cases and controls in a meta-analysis (7,141 prostate cancer cases and 11,804 controls). This study revealed three new prostate cancer risk loci in this ancestral population: rs1938781 at 11q12 (OR, 1.16); rs2252004 at 10q26 (OR, 1.16); and rs2055109 at 3p11.2 (OR, 1.20).
These results confirm the importance of evaluating SNP associations in different ethnic populations. Considerable effort is still needed to fully annotate genetic risk in these and other populations.
Although the statistical evidence for an association between genetic variation at these loci and prostate cancer risk is overwhelming, the clinical relevance of the variants and the mechanism(s) by which they lead to increased risk are unclear and will require further characterization. Additionally, these loci are associated with very modest risk estimates and explain only a fraction of overall inherited risk. Further work will include genome-wide analysis of rarer alleles catalogued via sequencing efforts, such as the 1000 Genomes Project. Disease-associated alleles with frequencies of less than 1% in the population may prove to be more highly penetrant and clinically useful. In addition, further work is needed to describe the landscape of genetic risk in non-European populations. Finally, until the individual and collective influences of genetic risk alleles are evaluated prospectively, their clinical utility will remain difficult to fully assess.
Genetic Modifiers of Prostate Cancer Aggressiveness
Due to the screening-related debate over risk of identifying clinically insignificant prostate cancers and the potential for overtreatment, studies characterizing genetic variants in subsets of patients with aggressive disease (e.g., Gleason score ≥8) are now being reported.
One study evaluated the association between the CASP8 D302H polymorphism and aggressive prostate cancer in a pooled analysis from three studies including 796 aggressive prostate cancer cases and 2,060 controls. Aggressive disease was defined as having androgen ablation therapy for prostate cancer, a PSA level greater than 50 ng/mL, radiographic evidence of metastases, or a Gleason score of 8 to 10. The H allele was associated with a protective effect for aggressive prostate cancer (OR per allele, 0.67; 95% CI, 0.54–0.83, P = .0003). The results were similar for European Americans and African Americans. The protective effect was observed only for aggressive disease, not for prostate cancer risk overall or for indolent prostate cancer, implying potential utility in identifying patients at risk of clinically significant disease.
Twenty prostate cancer risk SNPs identified in GWAS and fine-mapping follow-up studies were evaluated in 5,895 prostate cancer patients in search of SNP associations with prostate cancer aggressiveness. The risk-associated alleles of two SNPs (rs2735839 in KLK3 and rs10993994 in MSMB) were significantly associated with less aggressive prostate cancer; no significant associations were observed for the other 18 candidate SNPs. The two SNPs are known to be associated with PSA levels in normal men without prostate cancer. The authors concluded that the observed associations may be driven by over-representation within their case series of PSA screen-detected low-grade/low-stage disease and that none of these risk-related SNPs appear to hold the potential for identifying men at increased genetic risk of more aggressive prostate cancer.
One study evaluated the risk of metastatic prostate cancer (470 incident metastatic prostate cancer cases and 1,945 controls) and prostate cancer recurrence after prostatectomy for localized disease (1,412 localized prostate cancer cases, 328 of which had recurrence) with 12 SNPs previously found to be associated with prostate cancer risk. The T allele of rs10993994 in MSMB was associated with increased metastatic prostate cancer risk (RR, 1.24; 95% CI, 1.05–1.48; P = .012). The authors hypothesize that this SNP could be associated with primary carcinogenesis because metastatic prostate cancer at the time of diagnosis is less likely to be associated with PSA screen–detected disease. The other significant finding was the association in 8q24 of the A allele of rs4242382 (RR, 1.40; 95% CI, 1.13–1.75) and inverse association of the T allele of rs6938267 (RR, 0.67; 95% CI, 0.50–0.89) with metastatic prostate cancer. None of the SNPs studied were associated with risk of recurrence. These findings were not consistent with results of similar retrospective series.[201,202]
The association between prostate cancer–specific mortality (PCSM) and 846 SNPs was studied in a population-based prostate cancer cohort of 1,309 individuals in Seattle. Twenty-two SNPs found to be significantly associated with PCSM were then studied in a validation cohort of 2,875 prostate cancer cases from Sweden, of which five SNPs were significantly associated with PCSM. Hazard ratios in the Swedish validation cohort after adjusting for age at diagnosis, Gleason score, stage, PSA at diagnosis, and treatment for three of the SNPs were as follows: rs1137100 (LEPR) (HR, 0.82; 95% CI, 0.67–1.00; P = .027); rs2070874 (IL4) (HR, 1.27; 95% CI, 1.04–1.56, P = .011); and rs10778534 (CRY1) (HR, 1.23; 95% CI, 1.00–1.51, P = .022). Two of the SNPs were validated after adjusting for age at diagnosis alone: rs627839 (RNASEL) (HR, 1.22; 95% CI, 1.00–1.50, P = .024) and rs5993891 (ARVCF) (HR, 0.72; 95% CI, 0.52–1.01, P = .024). Compared with patients with zero to two at-risk genotypes, there was an increase in risk observed in patients with a greater number of at-risk genotypes after adjusting for age at diagnosis, Gleason score, stage, PSA at diagnosis, and treatment as follows: three at-risk genotypes (HR, 1.05; 95% CI 0.81–1.37); four at-risk genotypes (HR, 1.51; 95% CI, 1.16–1.97); and five at-risk genotypes (HR, 1.46 95% CI, 0.97–2.19). These results need validation for informing patient risk assessment and management.
A single institution study evaluated 36 SNPs for association with disease aggressiveness and prostate cancer–specific mortality in a prostate cancer cohort including 3,945 cases (predominantly European ancestry) and 580 prostate cancer–specific deaths. Two SNPs were associated with prostate cancer–specific survival (rs2735839 at 19q13, P = 7 × 10-4 and rs7679673 at 4q24, P = .014). Twelve SNPs were associated (P < .05) with other measures of prostate cancer aggressiveness including age at diagnosis, PSA level at diagnosis, Gleason score, and D'Amico criteria. These results need confirmation, as adjustment for multiple testing was not performed and ascertainment bias from single institution referral and screening patterns may have influenced the findings.
To definitively identify the inherited variants associated with prostate cancer aggressiveness, well-powered GWAS focusing on prostate cancer subjects with poor disease-related outcomes are needed. The control arm of such a study could be comprised of age-matched controls with no evidence of the disease or men with low-grade, indolent disease. One underpowered study genotyped 202 aggressive cases and 100 men matched by PSA and age who had not developed the disease using a SNP panel of 387,384 polymorphisms. Results were validated in a cohort of 527 aggressive cases, 595 less-aggressive cases, and 1,167 controls. The GWAS produced one SNP, rs6497287 at chromosome 15q13, as associated with aggressive disease. These results require further validation but point to the potential for GWAS focusing on this important phenotype.
The advent of large-scale high-throughput genotyping capabilities has resulted in an explosion of association studies between particular genes or genomic regions and prostate cancer risk. It is difficult to assess the import of any individual study. Accordingly, this PDQ Genetics of Prostate Cancer information summary will not attempt to provide an encyclopedic review of all such studies. Rather it will focus on studies that meet one or more of the following criteria: (1) Biological plausibility for the gene that is implicated; (2) Study designed with sufficient power to detect an odds ratio of an appropriate magnitude; (3) Multiple reports demonstrating the same association in the same direction; (4) Similar associations identified in studies of different design; (5) Evidence that the polymorphism is of functional significance; or (6) Existence of a prior hypothesis. However, individual studies may be cited by way of illustrating a specific theoretical point and do not imply that the association is definitive.
While many research teams have collected multiplex prostate cancer families with the goal of identifying rare, highly penetrant prostate cancer genes, other investigators have studied the potential roles of more common genetic variants as modifiers of prostate cancer risk. While these polymorphisms may not be associated with a large increase in relative risk (RR), these variants may have a high population-attributable risk because they are common. For example, if the population-attributable risk of prostate cancer associated with a genetic variant was 10% among carriers, that would imply that 10% of prostate cancer could be explained by the presence of this variant among carriers. For a rare variant, the proportion of cancer in the population attributed to the variant would be much less than 10%. Thus, a small increase in the RR of prostate cancer associated with a genetic variant that occurs frequently in the general population might, theoretically, account for a larger proportion of all prostate cancers than would the effects of a rare mutation in a gene, such as HPC1. This fact has provided much of the stimulus for studying the role of common genetic variants in the pathogenesis of prostate cancer and other cancers.
Concerns have been raised that differences in ethnic composition (population stratification) may confound the results of some prostate cancer association studies because the incidence of prostate cancer varies according to ethnicity. If a polymorphism also exhibits different frequencies according to race, it may appear to be associated with the disease in the absence of a true causal relationship. This issue was explored in a study in which the CYP3A4-V allele appeared to be statistically associated with increased prostate cancer risk in African Americans (P = .007) and European Americans (P = .02), but not in Nigerians. However, when the investigators added ten markers at other chromosomal regions, the significance for CYP3A4-V in African American men was lost. When the P value above was corrected for the observed population stratification, it was no longer significant. Thus, population admixture and stratification can create false associations (and obscure true associations) between genetic polymorphisms and disease risk.
To minimize confounding by population stratification, family-based association methods can be used. An inverse association has been identified between a single nucleotide polymorphism (SNP) in the CYP17 gene and prostate cancer risk using a set of 461 discordant sibling pairs. Since the siblings are genetically related, population stratification cannot bias this finding. A study of 1,461 Swedish men with prostate cancer in an ethnically homogenous population and 796 control men confirmed an inverse association between a CYP17 variant and prostate cancer risk (P = .04).
In an effort to more comprehensively evaluate the relationship between genetic variants in a particular gene and the risk of a specific cancer, single SNP association studies are augmented by a haplotype -based analytical strategy, in which a series of closely linked SNPs is selected to represent the entire gene. The Multiethnic Cohort Study (MEC) investigators provide a example of this approach as it applies to prostate cancer. Twenty-nine SNPs were used to define four haplotypes spanning the IGF1 gene. The investigators observed modest statistically significant elevations in RR (ranging from 1.19–1.25) for each of the four haplotypes. They concluded that inherited variation in IGF1 may play a role in the risk of prostate cancer.
In addition to the specific examples cited above, there have been additional candidate genes examined for their potential roles in genetic susceptibility to prostate cancer. These include both systematic literature reviews [5,6,7] and formal meta-analyses evaluating specific candidate genes [8,9] on this complicated and evolving subject. Due to the cross-sectional nature of these studies and the inconsistent results among reports targeting the same gene, these findings currently have no role in clinical decision making. The results of large, adequately powered, prospective analyses of these associations will be required.
Androgen Receptor Gene
Androgen receptor (AR) gene variants have been examined in relation to both prostate cancer risk and disease progression. The AR is expressed during all stages of prostate carcinogenesis. One study demonstrated that men with hereditary prostate cancer who underwent radical prostatectomy had a higher percentage of prostate cancer cells exhibiting expression of the AR and a lower percentage of cancer cells expressing estrogen receptor alpha than did men with sporadic prostate cancer. The authors suggest that a specific pattern of hormone receptor expression may be associated with hereditary predisposition to prostate cancer.
Altered activity of the AR caused by inherited variants of the AR gene may influence risk of prostate cancer. The length of the polymorphic trinucleotide CAG and GGN microsatellite repeats in exon 1 of the AR gene (located on the X chromosome) have been associated with the risk of prostate cancer.[12,13] Some studies have suggested an inverse association between CAG repeat length and prostate cancer risk, and a direct association between GGN repeat length and risk of prostate cancer; however, the evidence is inconsistent.[10,12,13,14,15,16,17,18,19,20,21,22] A meta-analysis of 19 case-control studies demonstrated a statistically significant association between both short CAG length (odds ratio [OR], 1.2; 95% confidence interval [CI], 1.1–1.3) and short GGN length (OR, 1.3; 95% CI, 1.1–1.6) and prostate cancer; however, the absolute difference in number of repeats between cases and controls is less than one, leading the investigators to question whether these small, statistically significant differences are biologically meaningful. Subsequently, the large MEC of 2,036 incident prostate cancer cases and 2,160 ethnically matched controls failed to confirm a statistically significant association (OR, 1.02; P = .11) between CAG repeat size and prostate cancer. A study of 1,461 Swedish men with prostate cancer and 796 control men reported an association between AR alleles with more than 22 CAG repeats and prostate cancer (OR, 1.35; 95% CI, 1.08–1.69; P = .03).
An analysis of androgen receptor CAG and CGN repeat length polymorphisms targeted African American men from the Flint Men's Health Study in an effort to identify a genetic modifier that might help explain the increased risk of prostate cancer in black versus white males in the United States. This population-based study of 131 African American prostate cancer patients and 340 screened-negative African American controls showed no evidence of an association between shorter AR repeat length and prostate cancer risk. These results, together with data from three prior, smaller studies,[24,26,27] indicate that short AR repeat variants do not contribute significantly to the risk of prostate cancer in African American men.
Germline mutations in the AR gene (located on the X chromosome) have been rarely reported. The R726L mutation has been identified as a possible contributor to about 2% of both sporadic and familial prostate cancer in Finland. This mutation, which alters the transactivational specificity of the AR protein, was found in 8 of 418 (1.91%) consecutive sporadic prostate cancer cases, 2 of 106 (1.89%) familial cases, and 3 of 900 (0.33%) normal blood donors, yielding a significantly increased prostate cancer OR of 5.8 for both case groups. A subsequent Finnish study of 38 early-onset prostate cancer cases and 36 multiple-case prostate cancer families with no evidence of male-to-male transmission revealed one additional R726L mutation in one of the familial cases and no new germline mutations in the AR gene. These investigators concluded that germline AR mutations explain only a small fraction of familial and early-onset cases in Finland.
A study of genomic DNA from 60 multiple-case African-American (n = 30) and white (n = 30) families identified a novel missense germline AR mutation, T559S, in three affected members of a black sibship and none in the white families. No functional data were presented to indicate that this mutation was clearly deleterious. This was reported as a suggestive finding, in need of additional data.
5-Alpha-Reductase Gene (SRD5A2)
Molecular epidemiology studies have also examined genetic polymorphisms of the 5-alpha-reductase type II gene, which is also involved in the androgen metabolism cascade. Two isozymes of 5-alpha-reductase exist. The gene that codes for 5-alpha-reductase type II (SRD5A2) is located on chromosome 2. It is expressed in the prostate, where testosterone is converted irreversibly to dihydroxytestosterone (DHT) by 5-alpha-reductase type II. Evidence suggests that 5-alpha-reductase type II activity is reduced in populations at lower risk of prostate cancer, including Chinese and Japanese men.[32,33]
A polymorphism in the untranslated region of the SRD5A2 gene may also be associated with prostate cancer risk. Ten alleles fall into three families that differ in the number of TA dinucleotide repeats.[31,35] Although no clinical significance for these polymorphisms has yet been determined, some TA repeat alleles may promote an elevation of enzyme activity, which may in turn increase the level of DHT in the prostate.[10,31] A subsequent meta-analysis failed to detect a statistically significant association between prostate cancer risk and the TA repeat polymorphism, although a relationship could not be definitively excluded. This meta-analysis also examined the potential roles of two coding variants: A49T and V89L. An association with V89L was excluded, and the role for A49T was found to have at most a modest effect on prostate cancer susceptibility. Bias or chance could account for the latter observation. A study of 1,461 Swedish men with prostate cancer and 796 control men reported an association between two variants in SRD5A2 and prostate cancer risk (OR, 1.45; 95% CI, 1.01–2.08; OR, 1.49; 95% CI, 1.03–2.15). Another meta-analysis of 25 case-control studies, including 8,615 cases and 9,089 controls, found no overall association between the V89L polymorphism and prostate cancer risk. In a subgroup analysis, men younger than 65 years (323 cases and 677 controls) who carried the LL genotype had a modest association with prostate cancer (LL vs. VV, OR, 1.70; 95% CI, 1.09–2.66 and LL vs. VV+VL, OR, 1.75; 95% CI, 1.14–2.68). A subsequent systematic review and meta-analysis including 27 non-familial case-control studies found no statistically significant association between either the V89L or A49T polymorphisms and prostate cancer risk.
Polymorphisms in several genes involved in the biosynthesis, activation, metabolism, and degradation of androgens (CYP17, CYP3A4, CYP19A1, and SRD5A2) and the stimulation of mitogenic and antiapoptotic activities (IGF-1 and IGFBP-3) of normal prostate cells were examined for association with prostate cancer in 131 African American cases and 342 controls. While allele frequencies did not differ between cases and controls regarding three SNPs in the CYP17 gene (rs6163, rs6162, and rs743572), heterozygous genotypes of these SNPs were found to be associated with a reduced risk (OR, 0.56; 95% CI, 0.35–0.88; OR, 0.57; 95% CI, 0.36–0.90; OR, 0.55; 95% CI, 0.35–0.88, respectively). Evidence suggestive of an association between SNP rs5742657 in intron 2 of IGF-1 was also found (OR, 1.57; 95% CI, 0.94–2.63). Additional studies are needed to confirm these findings.
Other investigators have explored the potential contribution of the variation in genes involved in the estrogen pathway. A Swedish population study of 1,415 prostate cancer cases and 801 age-matched controls examined the association of SNPs in the estrogen receptor-beta (ER-beta) gene and prostate cancer. One SNP in the promoter region of ER-beta, rs2987983, was associated with an overall prostate cancer risk of 1.23 and 1.35 for localized disease. This study awaits replication.
E-cadherin is a tumor suppressor gene in which germline mutations cause a hereditary form of gastric carcinoma. A SNP designated -160→A, located in the promoter region of E-cadherin, has been found to alter the transcriptional activity of this gene. Because somatic mutations in E-cadherin have been implicated in development of invasive malignancy in a number of different cancers, various investigators have searched for evidence that this functionally significant promoter might be a modifier of cancer risk. A meta-analysis of 26 case-control studies evaluated this genetic variant as a candidate susceptibility allele for seven different cancers. Eight of these studies (~2,600 cases and 2,600 controls) evaluated the risk of prostate cancer. Overall, carriers of the -160→A allele were at 30% increased risk of prostate cancer (95% CI, 1.1–1.6) compared with controls. A second meta-analysis  of E-cadherin associations with prostate cancer, reported findings that were quite similar to those noted above, although the overall association between the -160→A allele was not statistically significant (OR, 1.21; 95% CI, 0.97–1.51). The second study was based on a set of individual studies that largely, but not completely, overlapped with those in the earlier report; it was the exclusion of a study judged to have inappropriate controls  that accounts for this difference. The overlap in individual studies included in these two meta-analyses is large; therefore, the second meta-analysis does not represent confirmation of the first. Further studies are required to determine whether this finding is reproducible and biologically/clinically important.
Toll-like Receptor Genes
There is a great deal of interest in the possibility that chronic inflammation may represent an important risk factor in prostate carcinogenesis. The family of toll-like receptors has been recognized as a critical component of the intrinsic immune system, one which recognizes ligands from exogenous microbes and a variety of endogenous substrates. This family of genes has been studied most extensively in the context of autoimmune disease, but there also have been a series of studies that have analyzed genetic variants in various members of this pathway as potential prostate cancer risk modifiers.[45,46,47,48,49] The results have been inconsistent, ranging from decreased risk, to null association, to increased risk.
One study was based upon 1,414 incident prostate cancer cases and 1,414 age-matched controls from the American Cancer Society Cancer Prevention Study II Nutrition Cohort. These investigators genotyped 28 SNPs in a region on chromosome 4p14 that includes TLR-10, TLR-1, and TLR-6, three members of the toll-like receptor gene cluster. Two TLR-10 SNPs and four TLR-1 SNPs were associated with significant reductions in prostate cancer risk, ranging from 29% to 38% for the homozygous variant genotype. A more detailed analysis demonstrated these six SNPs were not independent in their effect, but rather represented a single strong association with reduced risk (OR, 0.55; 95% CI, 0.33–0.90). There were no significant differences in this association when covariates such as Gleason s