Prostate Cancer Linkage Studies
Hereditary Prostate Cancer 1
Prostate Cancer Predisposing Locus
Hereditary Prostate Cancer X
CAPB
ELAC2/HPC2
HPC20
8p Loci
8q
BRCA1 and BRCA2
KLF6
AMACR
Other Potential Prostate Cancer Genes
Other Regions Identified by Linkage Studies
Genome-wide Association Studies

Like most cancers, prostate cancer is a complex neoplastic disorder in which disease initiation is the result of an interaction between genetic and nongenetic factors. The identification of causative genes for prostate cancer, however, has been elusive in spite of segregation analyses of prostate cancer families that support the existence of one or more hereditary prostate cancer genes.[1-8] Several candidate loci have been identified by performing genome-wide linkage analysis studies in high-risk families, but confirmation of these proposed susceptibility loci from subsequent studies has often been lacking. Further, some prostate cancer susceptibility genes have been characterized by positional cloning, but follow-up studies have not yet demonstrated that any of these loci contribute to a significant number of high-risk prostate cancer families. While the goal of linkage analysis is to identify the chromosomal location of prostate cancer susceptibility genes, none of the putative genes in these regions identified to date have been widely accepted as clinically useful. Examples of loci that have been identified in studies of high-risk families are discussed below and are summarized in Table 2.

The recognition that prostate cancer clusters within families has led many investigators to collect multiplex families with the goal of localizing prostate cancer susceptibility genes through linkage studies. Despite the extensive collection of prostate cancer families and the formation of a collaborative research group (the International Consortium for Prostate Cancer Genetics [ICPCG]), the identification of prostate cancer genes has been exceedingly difficult. A review of eight prostate cancer linkage studies that evaluated a total of 4,600 cases of prostate cancer from 1,293 kindreds found several methodological differences. The authors suggest that differences in populations, enrollment criteria, and underlying genetic models used for each analysis may account for the lack of consistency between linkage studies.[9] The following discussion highlights both the clinical and research issues leading to this complexity.

Linkage studies are typically performed on high-risk extended families in which 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. The statistical power of linkage analysis is affected by:

• Family size and having a sufficient number of family members who volunteer to contribute DNA.
• The number of disease cases in each family.
• Factors related to age at disease onset.
• Gender differences in disease risk.

Because the risk of prostate cancer is influenced by both age at onset in affected relatives and number of relatives affected, the lack of accurate family history information about prostate cancer can limit the overall analysis.

Because a standard definition of hereditary prostate cancer (HPC) has not been accepted, prostate cancer linkage studies have not used consistent criteria for enrollment.[9] One criterion that has been proposed is the Hopkins Criteria that provides a working definition of HPC families.[10] The three criteria are kindreds with prostate cancer in the following:

1. Three or more first-degree relatives (father, brother, son),
2. Three successive generations of either the maternal or paternal lineages, and/or
3. At least two relatives affected at age 55 years or younger.

Families need to fulfill only one of these criteria to be considered to have HPC. Validity of these research criteria has not been confirmed for clinical management and must await identification of specific prostate cancer susceptibility genes. Using these criteria, a study has shown that approximately 5% of men in a large surgical series will be from a family with HPC.[10]

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 1 in 6, it is possible that families under study have men with both inherited and sporadic prostate cancer.[11] 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 no definitive 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 screening as the rates of prostate cancer in families will differ between screened and unscreened families.

In an effort to clarify the inconsistent linkage results, the ICPCG combined genome-wide linkage data from 1,233 families contributed by ten individual research teams. One analytic approach used the entire set of 1,233 families and five regions of suggestive linkage (logarithm of the odd [LOD] scores between 1.87 and 3.30) were identified: 5q12, 8p21, 15q11, 17q21, and 22q12. Therefore, the pooled analysis did not formally confirm any previously identified chromosomal regions of interest (see below). In the hope that targeting more homogenous family subsets might facilitate gene identification, a second analysis focused on subsets of the 1,233 families sharing common features, such as multiple affected family members or younger age at diagnosis. In 269 families with at least five affected members, significant linkage was detected at 22q12 (LOD score 3.57) and suggestive linkage was also observed at 1q25, 8q13, 13q14, 16p13, and 17q21. In 606 families with members aged 65 years or younger at diagnosis, linkage was suggested at 3p24, 5q35, 11q22, and Xq12.[12] These findings may facilitate prioritization of genomic regions for further study.

One way to address the inconsistency between linkage studies is to require inclusion criteria that defines clinically significant disease (e.g., Gleason grade ≥7, PSA ≥20 ng/mL) in an affected man.[13-15] 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.[16,17] This type of approach, however, has not yet led to the identification of consistent linkage signals across datasets.[18,19]

The results of a genome-wide scan of 91 high-risk prostate cancer families meeting the Hopkins criteria from the United States and Sweden suggested the presence of a major prostate cancer susceptibility locus at chromosome 1q24,[20] designated HPC1. Assuming genetic heterogeneity (i.e., that it is likely that only a subset of these 91 families carry an HPC1 mutation), the odds favoring the presence of this gene are nearly 1 million to 1. The genetic evidence supporting the existence of HPC1 was confined to 35% of the 91 families. This subgroup was characterized clinically by having more than five affected family members and an average age at prostate cancer diagnosis younger than 65 years. Further analyses of families that are genetically linked to HPC1 revealed the following characteristics:

• Younger age at diagnosis.
• Higher tumor grade (Gleason score).
• More advanced stage at diagnosis.[21,22]

Despite the strength of the initial results,[20] subsequent studies have often failed to confirm the linkage.[23-26] Nevertheless, confirmatory results were obtained in two studies in the United States that involved 59 and 92 families.[27,28] Linkage evidence in these reports was stronger among families in which prostate cancer was diagnosed earlier in life (<67 years) or that fit the Hopkins definition of HPC. In an analysis of 41 families from Utah, in which the mean number of affected men per family was large (10.7), linkage with 1q24-25 was confirmed.[29] The ICPCG pooled data from 772 families in North America, Australia, Finland, Norway, Sweden, and the United Kingdom, and obtained some evidence of linkage at 1q24.[30] The estimated percentage of familial prostate cancer families explained on the basis of this putative gene locus was 6%. Stronger evidence of linkage was seen among families with a male-to-male pattern of inheritance. Modest evidence for linkage to this region was also identified on a genome-wide scan of 188 families from Johns Hopkins,[31] including 51 kindreds examined in the initial positive linkage study.[20] A study of 33 African American families demonstrated some evidence in support of prostate cancer linkage to markers that map to several HPC candidate regions.[32]

Data suggest that the RNASEL gene at 1q25 may be the molecular basis of the prostate cancer susceptibility locus HPC1. The gene encodes an endoribonuclease that is a member of the interferon-regulated 2-5A system. Deleterious germline RNASEL mutations were detected in two out of eight families with prostate cancer linkage to 1q24-25 markers. Follow-up studies by several groups, however, have not identified a significant number of RNASEL germline variants among families with HPC.[33,34] In a study of Finnish men with prostate cancer, a stop mutation, E265X, was found in 4.3% of the men from HPC families compared with 1.8% of controls.[35] A founder frameshift mutation in RNASEL (471delAAAG) was identified in 4% of Ashkenazi individuals.[36] The frequency of this mutation was higher in men with prostate cancer than in elderly male controls (6.9% vs. 2.4%, odds ratio [OR] = 3.9; 95% confidence interval [CI], 0.6–15.3; P = .17). Significant associations were noted between the common RNASEL polymorphism R462Q and familial prostate cancer.[33] This substitution results in a threefold reduction in RNASEL activity.[37] A Swedish population-based case-control study examined the prevalence of E265X and other variants in the RNASEL gene. There were no differences for the E265X truncating mutation between the 780 controls (1.9%), 1,204 sporadic prostate cancer cases (1.9%), or 350 familial/HPC prostate cancer cases (1.4%).[38] Further, this group did not find significant differences between cases and controls for the R462Q variant. A meta-analysis summarized the data from ten case-control studies that contained data on the RNASEL variants E265X, R462Q and D541E. Only the D541E allele was associated with an increased risk of prostate cancer, although the magnitude of the effect was small.[39] In summary, there is evidence both for and against rare and common RNASEL variants contributing to a proportion of familial prostate cancer cases, though larger studies are required to more carefully delineate both the clinical and biologic implications of germline RNASEL variants.

A genome-wide scan using 49 high-risk prostate cancer families of German and French origin resulted in evidence of a prostate cancer predisposition locus on chromosome 1q42.[24] This is believed to be a separate gene from the HPC1 locus at 1q24.[20] Prostate cancer linkage to this locus, which has been designated PCAP, was described in a second set of European prostate cancer families [40] and families with evidence of linkage had an earlier average age at diagnosis (<65 years). PCAP linkage has not been observed in several studies of U.S. and international HPC families.[9,17,31,41-48]

A prostate cancer susceptibility locus (designated HPCX) has been mapped to the X chromosome by using a set of high-risk prostate cancer families from the United States, Finland, and Sweden.[49] In this initial report, linkage to a hypothesized gene located at Xq27-28 was predicted to account for 16% of prostate cancer among the 360 families that were analyzed. Analytic epidemiology studies have shown a higher relative risk (RR) of prostate cancer among men with an affected brother versus men with an affected father, a finding that supports the possibility of a prostate cancer susceptibility locus on the X chromosome;[50] however, this pattern is also consistent with an autosomal recessive mode of inheritance or environmental factors. Follow-up HPCX linkage studies have shown some evidence in support of the existence of this locus,[44,51-53] and an ICPCG meta-analysis is in process. Using linkage disequilibrium analysis, a specific haplotype in the Xq27-28 region of HPCX was found to be significantly associated with X-linked prostate cancer in Finnish families.[54] This finding was confirmed in a case control training set (292 cases and controls) and replicated in independent test subjects (215 cases and controls). The Xq27 haplotype extended from rs5907859 to rs1493189, and was associated with prostate cancer (OR = 3.41; 95% CI, 1.04–11.17; P = 0.034).[55]

Many cancer susceptibility genes increase the risk for more than one type of malignancy. For example, BRCA1 mutations increase a woman’s chance of developing both breast and ovarian cancer. In this regard, a set of prostate cancer families who have one or more cases of primary brain cancer was identified.[56] In this set of 12 families, prostate cancer linkage to 1p36 markers was observed. This hypothetical gene locus has been named CAPB. Loss of heterozygosity (LOH) of this same genetic region was previously observed in sporadic brain cancers, suggesting that there is a tumor suppressor gene in this genomic interval. Other groups have not consistently confirmed prostate cancer linkage to CAPB in families with both brain and prostate cancers.[42,57] Further, there is evidence for linkage to 1p36 in one study of 207 prostate cancer families, considering as affected only those individuals with prostate cancer. This was particularly evident in families with early-onset disease in which the prostate cancer was diagnosed before age 59 years.[57] This raises the possibility that CAPB mutations may contribute to prostate cancer in a site-specific manner.

The ELAC2/HPC2 prostate cancer predisposition gene on chromosome 17p was cloned after a genome-wide scan of high-risk families from Utah (Table 3).[58] Two segregating germline mutations were identified among these multiplex prostate cancer families. Neither linkage evidence to 17p11 markers nor rare ELAC2/HPC2 variants were found in other sets of multiplex families.[59] The ELAC2/HPC2 gene from 300 men from 150 prostate cancer families (with three or more cases of prostate cancer) was sequenced and identified only one stop codon and five additional missense mutations.[60]

Two common variants in ELAC2/HPC2 have been extensively studied for their potential contribution to prostate cancer susceptibility. In a clinic-based study of 350 prostate cancer cases and 266 age-matched and race-matched controls, it was reported that men who carry both of two common polymorphisms in the ELAC2/HPC2 gene experience a modest increase in risk of prostate cancer (OR = 2.4; 95% CI, 1.1–5.3).[61] Many additional studies have been reported, six of which have been pooled in a meta-analysis.[62] The authors suggest that the use of unscreened controls in case-control studies results in the inclusion of a significant number of men with prostate cancer cases among subjects who are classified as controls. This misclassification error will bias association studies toward the null. In the ELAC2/HPC2 meta-analysis, if exclusion of data from association studies in which prostate cancer screening was performed in controls resulted in a positive association between the Thr541 substitution and prostate cancer risk (OR = 1.8; 95% CI, 1.2–2.7; P = .0029), then to the extent that misclassification bias is operating in this series, the reported OR may underestimate the strength of the observed association. Studies using population-based sampling might be expected to clarify the potential role of common ELAC2/HPC2 polymorphisms in prostate cancer. An Australian study found no significant association between ELAC2/HPC2 and prostate cancer.[63] Furthermore, these authors pooled their new data with those from seven published studies; their meta-analysis strengthened the conclusion that no association exists.

Evidence for yet another prostate cancer susceptibility locus on chromosome 20, which has been termed HPC20, has been reported.[44,64] In stratified analyses, the group of patients with the strongest evidence of linkage to this locus were the families with fewer than five family members affected with prostate cancer, a later average age at diagnosis, and no male-to-male transmission, a pattern distinctly different from that reported for HPC1. Some evidence of prostate cancer linkage to HPC20 has been observed in two independent sets of families,[65,66] though the candidate genomic interval remains large; however, a combined linkage analysis of 1,234 pedigrees performed by the ICPCG failed to replicate linkage of hereditary prostate cancer to 20q13 markers.[67] In this report, the original 158 Mayo families that were used to identify HPC20 had a maximum heterogeneity LOD score under a recessive model of 2.78 whereas the remaining 1,076 families has a maximum heterogeneity LOD score of 0.06 using the same model. These data suggest that if HPC20 truly exists, it may only account for a small fraction of all hereditary prostate cancers.

Chromosome 8p is commonly deleted in prostate cancer; consequently, many groups have focused on using deletion mapping in an attempt to localize one or more tumor suppressor genes in this region. Several genome-wide scans have provided modest evidence of prostate cancer linkage to markers that map to 8p.[31,45,68] Evidence has been reported that both rare and common variants in the macrophage scavenger receptor 1 gene (MSR1) at 8p22 are associated with prostate cancer susceptibility (Table 3).[69,70] Case-control studies examining an association between these alleles and prostate cancer, however, did not show significant findings, including a meta-analysis.[71-74] Germline variants of the LZTS1 gene, also at 8p22, have been reported to be associated with sporadic prostate cancer.[75]

A combined analysis of somatic deletions of chromosome 8p in prostate cancer tumor tissue and fine-mapping linkage in multiple-case families has identified two additional regions (8p23.1 and 8p21.3) that are associated with prostate cancer risk.[76]

A linkage peak at chromosome 8q24 was reported in 323 Icelandic prostate cancer families with a peak LOD score of 2.11. Detailed genotyping of this region revealed an association in three case-control populations in Sweden, Iceland, and the United States with allele -8 at marker DG8S737. The population attributable risk for prostate cancer from this allele was 8%. The results were replicated in an African American case-control population, in which the population attributable risk was 16%.[77] Support for the existence of a prostate cancer susceptibility gene at 8q24, specifically in African American men, was also observed using admixture mapping.[78]

A series of studies confirming the association between prostate cancer risk and single nucleotide polymorphism (SNP) rs1447295 has been published.[79-84] Three additional studies evaluating the 8q24 locus have identified a second SNP, rs6983267, which is close to but distinct from rs1447295.[82,85,86] Furthermore, a multiethnic analysis identified five new variants all within this same region, each of which appears to be independently associated with prostate cancer risk. A number of these variants are much more common than rs1447295, suggesting that the proportion of all prostate cancers that may be explained on the basis of genetic variation in this region could be quite large. The well-known differences in prostate cancer risk among diverse population groups also may be related to these findings.[86] These observations are also notable because they occur in a region without known protein-encoding genes, which makes it very difficult to know what the underlying biological mechanism of susceptibility is likely to be. This is likely a recurring situation with genome-wide association studies in which statistically convincing associations are detected, but the truly causal variant and biological mechanism will be difficult to determine, requiring biochemical and other functional studies. These susceptibility alleles are generally associated with OR of 2 or lower and are not immediately clinically relevant.

Refer to the Polymorphisms and Prostate Cancer Susceptibility section of this summary for more information on polygenic factors in 8q.

Chromosome 8q24 risk variants have also been characterized in families with HPC. Twelve 8q24 variants from Regions 1, 2, and 3 and one variant from the c-MYC gene were genotyped in 168 probands from HPC families: 1,404 prostate cancer patients from non-HPC families undergoing radical prostatectomy, and 560 control subjects undergoing prostate cancer screening at Johns Hopkins University.[87] Risk alleles from five SNPs in Region 1 (rs1447295, rs4242382, rs7017300, rs10090154, and rs7837688) and alleles from two SNPs in Region 2 (rs6983561 and rs16901979) were significantly more frequent in HPC probands compared with controls. Genotype risk for HPC was also higher for these seven SNPs. Family-based transmission tests found that risk alleles of two SNPs in Region 2 were significantly over-transmitted to affected men in these HPC families. No evidence for linkage to 8q24 was found in these HPC families. Another report from the University of Michigan Prostate Cancer Genetics Project (UM-PCGP) evaluated five SNPs and one microsatellite marker found previously to be associated with prostate cancer in 403 non-Hispanic white families with discordant sibling pairs.[88] Using a family-based association test, the minor allele of rs6983561 and the major allele of rs6983267 were found to be preferentially transmitted to affected men. Furthermore, rs6983561 was significantly associated with prostate cancer among men diagnosed before age 50 years and rs6983267 was significantly associated with clinically aggressive disease. These data provide further support for modification of familial prostate cancer risk by variants in 8q24, particularly variants from Region 2.

Table 2 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 listed above. Data are also limited on the proposed phenotype associated with each loci, 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.

Studies of male BRCA1 [89] and BRCA2 mutation carriers demonstrate that these individuals have a higher risk of prostate cancer, as well as other cancers.[90]

Among male BRCA1 mutation carriers from hereditary breast ovarian cancer kindreds studied by the Breast Cancer Linkage Consortium (BCLC) family set, the risk of prostate cancer was not elevated overall (RR = 1.1; 95% CI, 0.8–1.5); however, the risk was modestly increased (RR = 1.8; 95% CI, 1.0–3.3) among men younger than 65 years.[89]

In contrast, a similar study of male BRCA2 mutation carriers in hereditary breast 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).[91] 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).[92]

Several small case series in Israel and in North America have analyzed the frequency of BRCA founder mutations among Ashkenazi men with prostate cancer.[93-95] Two specific BRCA1 mutations (185delAG and 5382insC) and one BRCA2 mutation (6174delT) are common in individuals of Ashkenazi (Eastern European) Jewish 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.[96-99] (Refer to the Major Genes section of the PDQ summary on Genetics of Breast and Ovarian Cancer for more information on the BRCA1 and BRCA2 genes.) In these studies, the point estimates of risk 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 a study of more than 5,000 American Ashkenazi Jewish volunteers from the Washington D.C. area (the Washington Ashkenazi Study [WAS]), a kin-cohort analytic approach was used to estimate the cumulative risk of prostate cancer among men 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%) compared with 3.8% among noncarriers (95% CI, 3.3%–4.4%).[99] 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]).

Two studies using similar methods examined 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. The first study [100] analyzed 979 consecutive Ashkenazi men with prostate cancer diagnosed in a large region of Israel, and compared the prevalence of founder mutations with age-matched controls from two different sources, the WAS and the Molecular Epidemiology of Colorectal Cancer (MECC) study from Israel. Overall, there was a twofold, statistically significant increase in the risk of prostate cancer among all carriers of founder mutations (OR = 2.1; 95% CI, 1.2–3.6). The magnitude of this risk was similar for BRCA1 and BRCA2 founder mutations, but only the BRCA2 association was statistically significant, when considered separately. This study did not find that mutation carriers developed prostate cancer at an earlier-than-usual age. Further, there was no evidence of unique or specific histopathology findings within the mutation-associated prostate cancers.

The second study [101] tested genomic DNA from 251 Ashkenazi men diagnosed with prostate cancer at their institution for the three common BRCA1/2 founder mutations. Using the control data from the WAS study described above, and after adjusting for age, all founder mutation carriers had a significantly increased risk of prostate cancer (OR = 3.4; 95% CI, 1.6–7.1). When evaluating BRCA1 versus BRCA2 founder mutations separately, no significantly increased risk of prostate cancer was detected for BRCA1 mutation carriers, while the risk among BRCA2 mutation carriers was increased substantially (OR = 4.8; 95% CI, 1.9–12.2).

These two studies support the hypothesis that prostate cancer occurs excessively among carriers of Ashkenazi Jewish founder mutations, and both suggest that the risk may be greater among men with the BRCA2 founder mutation (6174delT) than 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.

Two hundred sixty-three men with prostate cancer diagnosed in the United Kingdom (U.K.) before the age of 56 years underwent testing for BRCA2 mutations.[102] Screening of all coding regions resulted in the identification of six men (2.3%) with protein-truncating BRCA2 mutations, as well as 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 cancer. 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 RR of early-onset prostate cancer attributable to BRCA2 mutations (95% CI, 9–57). In a similar study conducted in a U.S. population,[103] 290 men (11% African American and 87% Caucasian) diagnosed with prostate cancer prior to age 55 years, 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 Caucasian cases for a prevalence in Caucasians of 0.78% (95%CI, 0.09–2.81%) and a 7.8 (95%CI, 1.8–9.4) RR of prostate cancer in Caucasian BRCA2 mutation carriers. Of the two individuals with a protein truncating mutation, neither reported a family history of breast or ovarian cancer.[103] 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 U.S.

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.[104-109] A recent population-based case-control study between BRCA2 999del5 mutation carriers and noncarriers (all of whom had a prostate cancer diagnosis) from the Icelandic Cancer Registry was conducted.[110] Five hundred and twenty-seven out of 596 prostate cancer patients from Iceland with prostate tissues available for pathology review 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 age 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.

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.[111] A genome-wide scan for HPC using 175 families from the University of Michigan Prostate Cancer Genetics Project (UM-PCGP) found evidence for linkage to chromosome 17q markers.[46] 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.[112] Sixty-five of the individuals screened had wild-type BRCA1 sequence, and only one individual from a family with prostate and ovarian cancer 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.[113] 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-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.[46] These findings support further investigation of BRCA1 variants in prostate cancer risk.

In an effort to clarify the relationship between BRCA1/BRCA2 and prostate cancer risk, 215 BRCA1-positive and 188 BRCA2-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 who 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 = 0.02). 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.[114] The absence of mutation information on the 130 unstudied cases limits the value of this observation.

Three Polish BRCA1 founder mutations (C16G, 4153delA, 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, 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 one of the other two mutations was associated with a 3.6-fold increase in prostate cancer risk (P = 0.045), and an even greater risk (OR = 12; P = 0.0004) 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.[115] 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.

Thus, the literature suggests that there may be a modest increase in prostate cancer risk among men with one of the Ashkenazi founder mutations, and a more substantial increase in risk among BRCA2 carriers in general; the risk is unclear among BRCA1 mutation carriers. These observations may comprise one of many factors that a man contemplating BRCA mutation testing might consider. Uncertainties regarding screening and management of men at increased risk of prostate cancer make it difficult to encourage BRCA mutation testing solely for prostate cancer risk management. (Refer to the Mutations in BRCA1 and BRCA2 section of the PDQ summary on Genetics of Breast and Ovarian Cancer and the Screening section of this summary for more information about testing for BRCA1 and BRCA2.)

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.[116] 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.[117] 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.[118] 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, P = .01; 95% CI, 1.20–2.16).[117] However, the prostate cancer risk associated with the IVS1-27G>A SNP was not detected in a study of 300 Jewish prostate cancer families.[119] 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 this Israeli study (49/804 alleles in cases compared with 55/600 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.[120] 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 are participating in a research study focused on early-onset and/or HPC.[121] The strongest evidence for association was for SNP rs3195676 (M9V) with an OR of 0.58 (95% CI, 0.38 – 0.90, P = 0.01 for a recessive model). The reported magnitude and direction of the association observed for this SNP was similar between this study and previously mentioned AMACR resequencing experiments.[120] A nested case-control study was conducted using samples from the screening arm of the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening trial to test for potential association between 7 AMACR SNPs and prostate cancer including M9V.[122] No association was detected between any of the SNPs and prostate cancer. Note that the prostate cancer cases in the PLCO study are all older than 55 years and not specifically enriched for family history. Risk for prostate cancer was reduced, however, in men who reported using ibuprofen who also had specific alleles and 4 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.

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.[123] 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 [124] and ataxia telangiectasia.[125,126] 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.[123] A collaborative report from five groups participating in the ICPCG 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.[127]

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.[128] 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 felt 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.[128,129] 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.[130] 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 as well as 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.[131]

Table 3 summarizes the candidate genes for prostate cancer susceptibility, their chromosomal location, and availability of clinical testing.