Purpose: A genome-wide scan of 175 hereditary prostate cancer families from the University of Michigan Prostate Cancer Genetics Project provided evidence of prostate cancer linkage to 17q markers near the BRCA1 gene. To examine the possibility that germ-line BRCA1 mutations were associated with hereditary prostate cancer, individuals from 93 families with evidence of linkage to chromosome 17q were screened for germ-line BRCA1 mutations.
Experimental Design: One individual from each of the 93 families, the majority with three or more cases of prostate cancer, were screened for BRCA1 mutations with denaturing high-performance liquid chromatography (HPLC). Fragments exhibiting denaturing HPLC variant patterns were additionally analyzed by direct sequencing.
Results: Sixty-five of the individuals selected for sequencing from 65 unrelated families were determined to have wild-type BRCA1 sequence by denaturing HPLC. One individual from a family with both prostate and ovarian cancer was found to have a truncating BRCA1 mutation (3829delT). An additional 27 germ-line variants were identified, including 15 missense variants.
Conclusions: These sequencing results suggest that BRCA1 truncating mutations do not account for the linkage evidence on chromosome 17 observed in University of Michigan Prostate Cancer Genetics Project families. A recently completed combined genome scan has also detected linkage to 17q22, and studies are ongoing to identify the relevant prostate cancer susceptibility gene in this region.
The search for prostate cancer predisposition genes has been complicated by locus heterogeneity and the high, sporadic rate of disease in the general population. Whole genome approaches have been used to map prostate cancer susceptibility loci, including HPC1 (1), HPCX (2), PCAP (3), CAPB (4), and HPC20 (5); however, many of these loci have not been widely confirmed in replication studies (see reviews; ref. 6, 7). Given the complexity of this disease, analysis of additional multiplex prostate cancer families using genome-wide scans will likely point to additional genomic regions that may harbor prostate cancer susceptibility genes or provide confirmation of regions suggested by previously reports published. Consistent with this supposition, we recently conducted a genome-wide scan on 175 pedigrees, the majority containing three or more individuals diagnosed with prostate cancer, and detected suggestive evidence for linkage on chromosome 17q (LOD = 2.36; ref. 8). The strongest evidence was observed in the subset of pedigrees with four or more affected individuals (LOD = 3.28). The closest genome-wide scan marker to this peak, D17S1868, is located within 5 cM of the breast cancer susceptibility gene BRCA1.
The evidence of a prostate cancer susceptibility gene at 17q is additionally strengthened by new results from a combined genome-wide linkage scan on 426 prostate cancer pedigrees (9), which included the 175 pedigrees from our initial report (8). In the combined analysis, the LOD score at D17S1868 was 2.72 when analyzing all of the pedigrees. However, the peak LOD score (LOD = 3.40) was observed at the neighboring genome-wide scan marker D17S787, which is ∼13 cM distal to BRCA1. Although the results at D17S787 were in large part driven by the 175 pedigrees from Lange et al. (8), there also was supportive evidence of linkage in pedigrees from Johns Hopkins University (LOD = 1.38) and Finland (LOD = 0.68) at this peak location. Evidence for linkage was strongest in pedigrees with early age of diagnosis and in pedigrees with four or more confirmed affected men; both of these subsets of pedigrees are plausibly enriched for an inherited form of prostate cancer.
The BRCA1 gene at 17q21 (OMIM ∗113705) was identified by Miki et al. (10) based on the identification of a number of deleterious mutations in genomic DNA from individuals in families with breast and/or ovarian cancer after mapping of the locus by Hall et al. (11) in 1990. The gene has 24 exons, including a single large exon (exon 11) that includes about 50% of the coding sequence. Overall, the gene encodes a protein product of 1,863 amino acids. Many of the mutations reported in the literature to date describe variants, which result in a truncated protein. Some missense mutations, for example those occurring in the RING-finger domain (amino acids 20 to 68; ref. 12), have also been shown by genetic studies to be pathogenic. Two common BRCA1 founder mutations in individuals of Ashkenazi Jewish descent have been described, namely the 185delAG in exon 2 and the 5385insC in exon 20. Founder mutations elsewhere in the gene have been reported from studies of other populations (13, 14, 15).
BRCA1 has been investigated for its potential role in prostate cancer susceptibility because of the observation that male carriers of BRCA1 mutations in families with hereditary breast/ovarian cancer have an increased risk of prostate cancer. Specifically, in a large study of families with known BRCA1 mutations identified from centers across Europe and North America, the risk of prostate cancer was slightly elevated in male mutation carriers under age 65 [relative risk 1.82, 95% confidence interval (CI) 1.01–3.29] but not in those ≥65 years of age (ref. 16; see Discussion). Other investigators have examined men with early onset and/or familial prostate cancer for the presence of BRCA1 mutations (17, 18, 19, 20). Most of these studies have found little evidence for an association between prostate cancer susceptibility and the inheritance of BRCA1 germ-line mutations. However, these reports have generally been limited in size and scope and often focused on families without younger mean ages at diagnosis. Indeed, most of the families studied have fallen short of the now well-established criteria for hereditary prostate cancer families proposed by Carter et al. (21).
Given the provocative findings of chromosome 17q linkage in our prostate cancer genome-wide scans, we set out to fully scan the entire BRCA1 coding region together with intron/exon boundaries for germ-line BRCA1 variants in a cohort of men from families whose disease could be attributed to a locus at 17q. This analysis represents the most comprehensive study of germ-line BRCA1 variants in hereditary prostate cancer done to date.
MATERIALS AND METHODS
A genome-wide mode-of-inheritance-free linkage scan, using 405 genetic markers, was conducted on 175 families participating in the University of Michigan Prostate Cancer Genetics Project. One hundred seventy pedigrees had three or more affected individuals diagnosed with prostate cancer, and the remaining five families were selected based on the occurrence of two cases of prostate cancer diagnosed before age 55 years (8). Ninety-five of the 175 University of Michigan Prostate Cancer Genetics Project families genes were determined to have a nonparametric linkage score >0 at 60.1 cM on chromosome 17. This distance was selected because it represents the linkage peak for all of the 175 families analyzed together. Ninety-three of these families are included in this report. DNA was isolated from nucleated blood cells by use of the Puregene Kit (Gentra Systems, Minneapolis, MN). All of the participants provided written informed consent, and all of the research protocols and consent forms were approved by the Institutional Review Board at the University of Michigan.
Mutation Screening Using Transgenomic Denaturing High-Performance Liquid Chromatography (HPLC).
Using 25 ng of genomic DNA, we amplified coding regions of BRCA1 in 35 separate amplicons for denaturing HPLC, as described previously in Wagner et al. (22). PCR reactions were carried out in a total of 32.5 μL with 1.35 μmol/L concentration of each primer, 2.5 or 1.5 mmol/L MgC12, 0.2 mmol/L deoxynucleoside triphosphates, 1× PCR reaction buffer, and 0.5 units Taq DNA Polymerase (Roche Diagnostics, Basel, Switzerland). After PCR amplification, products were heteroduplexed by slow cooling. Heteroduplexed products were then analyzed using denaturing HPLC and WAVEMAKER version 4.1.44 software. Ten microliters of PCR product were used in each analysis. Analysis temperatures (50°C to 62°C) were predicted by the software and selected to allow maximal separation of different species within each analysis. Fragments 2, 11AB, 11S, 11TU, 15, 16, and 21 were each analyzed at two separate temperatures corresponding to the melting of two separate domains within each fragment. Any denaturing HPLC variant pattern that was seen in <5% of the study population was additionally analyzed by direct sequencing (23). Less than 1% of the 3255 amplicons were not assessable because of technical reasons.
Direct Sequence Analysis.
PCR was used to amplify 40 ng of genomic DNA with primer sequences from Wagner et al. (22), although some modifications were required. PCR primers were purchased from Invitrogen Life Technologies, Inc. (Carlsbad, CA) and are available on request. With the exception of exons 8 and 11K, each reaction contained 4 μL of 10× PCR buffer (Invitrogen Life Technologies, Inc.), 1 μL 50 mmol/L MgC12, 1 μL 10 mmol/L deoxynucleoside triphosphates, 5 μL each of the two PCR primers at 5 μmol/L concentration, 2.0 μL template DNA at 20 ng/μl, and 0.5 μL of Taq (Platinum Taq Polymerase, Invitrogen) in 30.5 μL of double-distilled H2O, for a total reaction volume of 50 μL. Details for PCR conditions for exon 8 and 11K are available on request. PCR products were cleaned with Montage PCR Centrifugal Filter Devices (Millipore, Billerica, MA) and sequenced using an ABI Prism 3100 Genetic Analyzer using Big Dye Terminator v1.1 chemistries (Applied Biosystems, Foster City, CA). Forward and reverse strand sequences were obtained for all individuals noted to have a possible germ-line variant denaturing HPLC. Sequences were screened with Mutation Surveyor Software (SoftGenetics, State College, PA). If a missense (or possibly deleterious) variant was identified in one family member, all of the members of that particular family who had donated blood samples to the University of Michigan Prostate Cancer Genetics Project were also sequenced to test for potential cosegregation of the mutation with disease.
Data analysis to describe the demographic and clinical characteristics of the eligible pedigrees was done using SAS v.8.2 (Cary, NC, 2001). Frequency distributions and/or means with SD and ranges were produced depending on the nature of available variables.
The clinical characteristics of the 93 University of Michigan Prostate Cancer Genetics Project families with a nonparametric linkage score >0 at 60.1 cM on chromosome 17 are presented in Table 1. The majority of families were Caucasian, and the average number of confirmed cases of prostate cancer was 4.1 per family (range 2 to 14). Pedigrees were studied to identify other types of cancers that may be present in chromosome 17-linked prostate cancer families. Only cases of cancer occurring in family members in a first- and/or second-degree relationship to a man with confirmed prostate cancer were considered. Thirty-five of the 93 families (38%) had one or more cases of breast cancer in the pedigree. Eight families (9%) had one or more cases of ovarian cancer, and five families had cases of both breast and ovarian cancer. Other types of cancers were observed in these families, including cases of colon, stomach, lung, and thyroid cancers. Seventeen of the 93 families (18%) had only prostate cancer reported.
One individual was selected for BRCA1 mutation screening from each of the 93 University of Michigan Prostate Cancer Genetics Project families with evidence of linkage to chromosome 17 using the following approach, which was designed to select the individual in the pedigree that was most likely to harbor a deleterious BRCA1 mutation. In 85 families, the youngest man with prostate cancer in each pedigree for whom we had a DNA sample was used for mutation screening. In three families, an older family member with prostate cancer was selected for testing. In these families, the older affected man had at least one first- and/or second-degree relative with breast cancer, whereas the youngest affected man in these pedigrees was more distantly related to the woman with breast cancer. In the remaining five families, DNA from a woman with breast cancer was used for BRCA1 mutation screening. In each case, the woman was in a first-degree relationship (e.g., a sister) to a man with prostate cancer in the pedigree. The average age of prostate cancer diagnosis was 55.7 +/− 7.8 years (range 37 to 74 years) in the men whose DNA was screened for BRCA1 mutations. Similarly, the average age at diagnosis of breast cancer in the five women whose DNA was screened for BRCA1 mutations was 58.2 +/− 15.6 years (range 43 to 77 years).
Sixty-five of the family members from 65 unrelated families were determined to have wild-type BRCA1 sequence by denaturing HPLC. A total of 28 different sequence variants were identified among University of Michigan Prostate Cancer Genetics Project family members (Table 2, Fig. 1); some individuals harbored more than one germ-line variant. Five of the variants identified herein have not been reported previously in the Breast Cancer Information Core Database (http://research.nhgri.nih.gov/projects/bic/): 667–79T→C, H448Y, 5194–53C→T, 5271 + 85delT, and I1858T. If a missense variant or insertion/deletion was identified in one family member, all of the available family members were tested for the same variant. Because we used direct sequencing to confirm the presence of germ-line variants in family members, we had the opportunity to identify additional sequence variants in selected individuals who were not initially selected for whole-gene analysis. Three of the reported variants, namely E1038G, K1183R, and E1214K, were present in affected family members but not in the individual selected for denaturing HPLC mutation screening.
The protein truncating mutation 3829delT BRCA1 mutation was identified in a family that is bilineal with respect to prostate cancer. There is a history of ovarian cancer on the maternal side of the family. The proband and his unaffected maternal uncle were shown to carry the 3829delT mutation.
The finding of prostate cancer linkage to chromosome 17q21 markers in University of Michigan Prostate Cancer Genetics Project families led us to consider more fully the possibility that germ-line BRCA1 mutations may be contributing to hereditary prostate cancer. Because suggestive linkage was observed in our genome-wide scans, we targeted 93 families with individual nonparametric linkage scores >0 and selected the single individual most likely to carry a mutation for denaturing HPLC analysis. By this approach, only 1 of 93 families was discovered to have a deleterious mutation; this pedigree did not have strong evidence for prostate cancer linkage to chromosome 17 markers (nonparametric linkage = 0.22). Although a number of additional germ-line variants, particularly missense changes, were identified in our University of Michigan Prostate Cancer Genetics Project families, we found no evidence to suggest that these variants in BRCA1 were associated with increased evidence of linkage (data not shown). It is also difficult to draw conclusions about the frequency of missense variants in this study population given the absence of an appropriate control group. To more fully test whether variants in BRCA1 explain the evidence for prostate cancer linkage at 17q, ideally one would like to have BRCA1 sequence data from all family members from all prostate cancer pedigrees, regardless of nonparametric linkage score. Given the considerable expense of screening for BRCA1 mutations, such a test was beyond the scope of this study. Interestingly, BRCA1 is located near the outer limits of the 1-LOD drop linkage support interval presented in the new combined linkage scan on 17q (9). Taken together with our BRCA1 mutation results, we conclude that BRCA1 is unlikely to be the hereditary prostate cancer gene indicated by our combined linkage results on 17q.
The precise role of BRCA1 mutations in prostate cancer has been a subject of debate over the last decade. The most recent data regarding prostate cancer risk in BRCA1 mutation carriers comes from the Breast Cancer Linkage Consortium in which a cohort of 11,847 individuals from 699 families harboring BRCA1 mutations were studied. These investigators observed an increased risk of prostate cancer in male mutation carriers under age 65 (relative risk 1.82, 95% CI 1.01–3.29) but not in men ≥65 years of age (relative risk 0.84, 95% CI 0.53–1.22; ref. 16). By comparison, similar data from the Breast Cancer Linkage Consortium showed that the risk of prostate cancer in BRCA2 (OMIM ∗600185) mutation carriers is higher than for BRCA1 mutation carriers. The relative risk for prostate cancer in men with BRCA2 mutations who are less than 65 years of age was 7.3 (95% CI = 4.7–11.5), and the overall relative risk was 4.7 for men of all ages (95% CI = 3.5–6.2). Male BRCA2 mutation carriers are also at an increased risk for the development of breast cancer (24).
Additional studies have attempted to determine the role of BRCA1 mutations specifically in hereditary prostate cancer. Langston et al. (20) sequenced the entire BRCA1 gene using genomic DNA from 49 men with early onset prostate cancer diagnosed before age 65 years derived from a population-based, case-control study of middle aged men and prostate cancer. In the Langston et al. (20) study, the Ashkenazi 185delAG founder mutation was identified in one subject diagnosed with prostate cancer in his early 60s who reported no female relatives with breast and/or ovarian cancer but several male relatives with prostate cancer. Six additional rare sequence variants were also identified. No germ-line BRCA1 mutations were identified in affected family members from hereditary prostate cancer families studied by Sinclair et al. (17) and Gayther et al. (19). The total number of families in these reports was 51, and in contrast to our report, chromosome 17 linkage data were not used to select families for analysis.
Several groups have also examined the frequency of the BRCA1 founder mutations among Ashkenazi men with prostate cancer. For example, in a study of 83 men with prostate cancer, Nastiuk et al. (25) identified only one individual who was a carrier of the185delAG BRCA1 mutation. This is not substantially different from the frequency reported in the general Ashkenazi population. However, a recent report provides a more accurate measure of the degree of risk elevation associated with the BRCA1 Ashkenazi founder mutations. Giusti et al. (26) studied 940 Ashkenazi men diagnosed with prostate cancer in Israel and found an increased risk of prostate cancer carriers of the 185delAG mutation compared with Ashkenazi men over 50 years of age with no history of prostate cancer. The controls in this study came from two independent studies of Ashkenazi men in the United States and Israel, and the results were statistically significant (odds ratio 2.5; 95% CI 1.1–6.0). In the Giusti et al. (26) study, the percentage of cases with a family history of disease was not reported, but the cases were not specifically enriched for men with hereditary prostate cancer. By comparison, Wilkens et al. (18) were unable to detect either BRCA1 founder mutations in 47 family members from 18 Ashkenazi prostate cancer families from the United States. Thus, whereas the Ashkenazi BRCA1 founder mutations may increase the risk of prostate cancer, other more penetrant mutations may account for the clustering of prostate cancer within Ashkenazi families.
In our study of hereditary prostate cancer families, we identified one deleterious BRCA1 mutation in a family with both prostate and ovarian cancer. However, we also identified 16 missense variants whose precise significance is unknown. Some of the missense changes, for example D693N in exon 11, were observed in five unrelated families in our study and have been described as polymorphisms of unknown or unlikely significance in Breast Cancer Information Core Database. Other changes such as H448Y and I1858T have not been reported previously. In our study, the H448Y substitution was identified in a bilineal family in which some unaffected males were determined to carry the variant, and the I1858T substitution was shared among both members of a brother pair with early onset prostate cancer (data not shown). It is also important to note that both families with novel BRCA1 missense substitutions are African American. Studies of breast and ovarian cancer families have shown that the pattern of genetic variation in the BRCA1 gene differs between African Americans compared with other racial groups (27). Because relatively few African American families have undergone clinical BRCA1 testing, the degree of known polymorphic variation in this population is less clear, and our report adds to this developing dataset.
If the BRCA1 gene is acting as a tumor suppressor gene in prostate cancer, somatic inactivation of one (or both) allele(s) may be observed, particularly in men harboring germ-line mutations. Gao et al. (28) observed a high rate of loss of heterozygosity in sporadic prostate cancers using chromosome 17q markers. In the Gao et al. (28) report, 11 of 21 informative tumors showed allelic loss at one or more 17q markers, and 8 of 18 (44%) informative cases had loss of heterozygosity using the intragenic BRCA1 marker D17S85. More recent mapping work by Dai et al. (29), however, places the common region of 17q deletion in sporadic prostate cancers distal to BRCA1. This study used P1 probes and fluorescent in situ hybridization to define an 85kB common region of deletion in prostate cancer located 470 kB distal to BRCA1. This data suggests that there is a tumor suppressor gene distal to BRCA1 that may be important in prostate carcinogenesis.
In conclusion, epidemiologic data derived primarily from studies of breast and ovarian cancer families suggest that BRCA1 mutations result in a small but detectable increase in risk for prostate cancer, especially for early onset disease. Our recent observation of prostate cancer linkage to chromosome 17q marker (8) motivated this comprehensive mutational analysis of BRCA1 in hereditary prostate cancer families. We were unable to identify deleterious mutations in a sufficient number of University of Michigan Prostate Cancer Genetics Project families to account for our reported linkage evidence. This suggests the existence of a hereditary prostate cancer gene near BRCA1 on chromosome 17, and efforts are ongoing toward additionally localizing and characterizing this putative locus.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Kathleen A. Cooney, 7310 Cancer Center Geriatric Center, 1500 East Medical Center Drive, Ann Arbor, MI 48109-0946. Phone: (734) 764-2248; Fax: (734) 615-2719; E-mail: email@example.com