BRCA1 and BRCA2 Mutations in an Asian Clinic-based Population Detected Using a Comprehensive Strategy

Background and objective: Genetic testing for germ line mutations in the BRCA1 and BRCA2 genes for some families at high risk for breast and/or ovarian cancer may yield negative results due to unidentified mutations or mutations with unknown clinical significance. We aimed to accurately determine the prevalence of mutations in these genes in an Asian clinic-based population by using a comprehensive testing strategy.

Materials and Methods: Ninety-four subjects from 90 families were accrued from risk assessment clinics. In addition to conventional mutational screening of BRCA1 and BRCA2, multiplex ligation-dependent probe amplification for the detection of large genomic rearrangements, evaluation of splice site alterations using transcript analysis and SpliceSiteFinder prediction, and analysis of missense mutations of unknown significance by multiple sequence alignment, PolyPhen analysis, and comparison of Protein Data Bank structures were incorporated into our testing strategy.

Results: The prevalence rates for clearly deleterious BRCA1 and BRCA2 mutations were 6.7% (6 of 90) and 8.9% (8 of 90), respectively, or 7.8% (7 of 90) and 11.1% (10 of 90), respectively, by including missense mutations predicted to be deleterious by computational analysis. In contrast to observations from European and American populations, deleterious mutations in BRCA2 (10 families) were more common than for BRCA1 (7 families). Overall, the frequency of mutations was 12.2% (n = 11) by conventional screening. However, by including deleterious mutations detected using multiplex ligation-dependent probe amplification (n = 1), transcript analysis (n = 2), and computational evaluation of missense mutations (n = 3), the frequency increased substantially to 18.9%. This suggests that the comprehensive strategy used is effective for identifying deleterious mutations in Asian individuals at high risk for breast and/or ovarian cancer. (Cancer Epidemiol Biomarkers Prev 2007;16(11):2276–84)

Breast cancer is the most frequent cancer among females in many Caucasian populations and in Singapore. Increased susceptibility for breast cancer is usually associated with deleterious mutations of the breast cancer susceptibility genes BRCA1 and BRCA2, particularly in families with both breast and ovarian cancer. Such germ line mutations in the BRCA1 and BRCA2 genes have been documented in most populations (1-3). Clinical genetic testing for BRCA1 and BRCA2 mutations after genetic counseling is the standard of care in North America and Europe but is not available in many countries in Asia.

Genetic testing of the BRCA1 and BRCA2 genes typically involves the DNA sequencing of all exons and intron-exon junctions. However, some high-risk pedigrees will have “negative” results, possibly due to unidentified mutations or mutations with unknown clinical significance, thus presenting a dilemma in risk assessment and genetic counseling (4).

Large genomic rearrangements, such as exon duplications or deletions found in high frequencies in European populations, are not detected by conventional genetic testing strategies (5). Recently, the multiplex ligation-dependent probe amplification (MLPA) assay has been used by many studies to detect large genomic rearrangements in the BRCA1 and BRCA2 genes (6, 7). In a large study of 300 families with four or more cases of breast or ovarian cancer but who were commercially tested negative for BRCA1 and BRCA2 mutations, MLPA detected genomic rearrangements in 12% of the probands (8), thus highlighting the importance of screening for such aberrations in addition to conventional PCR-sequencing protocols.

Intronic alterations that are located within or near to intron-exon junctions may affect mRNA splicing fidelity (9). These alterations are categorized as unclassified variants unless they have been evaluated by transcript analysis to determine their effect on mRNA splicing (10, 11). Other unclassified variants include missense mutations, which have unclear pathogenicity and which form between a third to half of all genetic variants documented in the Breast Cancer Information Core database. Pathogenicity of these variants may be established from family studies of cosegregation, absence in unaffected controls, or by using biochemical criteria, such as conservation of amino acid across species, severity of amino acid change, and involvement of an amino acid within a functional domain (12-14).

In this study, we aimed to accurately determine the prevalence of mutations in the BRCA1 and BRCA2 genes in a cohort of Singaporean women accrued at risk assessment clinics by using a comprehensive testing strategy. In addition to conventional screening of the coding exons of the BRCA1 and BRCA2 genes and their intron-exon boundaries, MLPA analysis for large genomic rearrangements, RNA analysis, in silico prediction of intronic alterations, and evaluation of missense mutations of unknown clinical significance by computational analyses were done. Although these techniques have been used for studies to evaluate the prevalence of BRCA1/BRCA2 mutations, to our knowledge, there have not been any reports incorporating all of these techniques for a comprehensive testing strategy for BRCA1 and BRCA2 mutation detection.

A flow chart summarizing the methodology of the study, from patient accrual to validation of mutations, is shown in Fig. 1.

All subjects were recruited from clinics at the National Cancer Center and KK Women's and Children's Hospital in Singapore under the Risk Evaluation and Prevention program. Detailed pedigree risk assessment was done with a combination of counselor assessment and the BRCAPRO model (15). Family histories of all cancers for three generations were recorded, including ages of diagnoses of all cancers. Subjects were eligible if there is a personal or family history of breast and ovarian cancer, if there are two primaries, if there is a personal and family history of breast (at least one premenopausal) or ovarian cancer(s) in a close relative from the same side of the family, or if the subjects have breast or ovarian cancer and were ages 39 years and below. One unaffected subject was tested as her sister had recent breast cancer at age 35 and was being evaluated for suspected ovarian cancer. Informed consent was obtained, and the research protocol was approved by the ethics review committee at both hospitals.

Between March 2002 and April 2006, 95 eligible subjects were accrued from 90 families. All subjects were of Asian descent. Subjects were predominantly Chinese (75.8%), followed by Malays (11.6%), Indians (4.2%), and others (8.4%), reflecting the ethnic composition of Singapore. The median age of participants at entry was 36 years, with a range from 19 to 72 years. Half of the subjects were below 40 years old.

Genomic DNA was isolated using standard techniques as previously reported (16). Purified DNA was amplified as described using published primers (17, 18) and primers described at the Breast Cancer Information Core Web site⁶

(16). Direct sequencing of the amplified products was done using the CEQ Dye Termination Cycle Sequencing quick start kit (Beckman Coulter), and the products were analyzed using the CEQ 8000 System (Beckman Coulter) according to the manufacturer's recommendations. The PCR products of exon 11 in BRCA1 and exons 10 and 11 in BRCA2 were in vitro transcribed/translated using the protein truncation test (TNT®T7 Quick for PCR DNA, Promega). The translation products were electrophoresed on 10% SDS-polyacrylamide gels. Any mutations detected were confirmed by separate PCR amplifications done at least once.

MLPA was done with the MLPA P002-BRCA1 and P0045-BRCA2 test kits and the MLPA P087 confirmation kit (MRC Holland) according to the manufacturer's recommendations, using alternative protocol 2 for multiplex PCR. Fragment analysis was done on the Beckman CEQ 8000 System (Beckman Coulter). Peak profiles for each sample were compared with a normal control and with other samples within the same experimental batch.

Total RNA was isolated from peripheral blood lymphocytes using Trizol reagent (Invitrogen). cDNA was then reverse transcribed from 150 ng of total RNA using the AMV reverse transcription system (Promega). cDNA (25 ng) was amplified using the following primers by standard procedures with HotStarTaq (Qiagen Gmbh). For amplification of exons 5 to 9 of BRCA1, the primers 5′-CTGAAACTTCTCAACCAGAAGAAA-3′ (E6-8F) and 5′-GTAACAATTCTTGATCTCCCACAC-3′ (E6-8R) were used. For amplification of exons 14 to 16 of BRCA2, the primers 5′-TCATGTTTCTTTAGAGCCGATTAC-3′ (2E14-16F) and 5′-ATTTTAGTTGAAGAAGCACCCTTT-3′ (2E14-16R) were used.

Amplification consisted of 35 cycles, each of 1 min, at 94°C, 56°C, and 72°C. PCR products were electrophoresed on a 3% NuSieve GTG agarose gel (Cambrex; for exons 5-9 of BRCA1) or 2% agarose gel (for exons 14-16 of BRCA2), and PCR fragments were purified by gel extraction with the PureLink Quick Gel extraction kit (Invitrogen). Direct sequencing was done as described above.

SpliceSiteFinder⁷

was used to determine the presence and relative efficiencies of donor, acceptor, and branch point sites.

BRCA1 and BRCA2 sequences from various organisms that contain the region that encompasses the site of missense mutation were obtained from Swiss-Prot. Multiple sequence alignment was then conducted using ClustalX 1.83. The Swiss PDB-Viewer was used for rendering and viewing of protein database structures of mutation sites that were available for analysis. When a mutation occurred at the vicinity of putative phosphorylation sites as reported in the literature, NetPhos 2.0 was used to predict possible sites of phosphorylation (19). Comparison of relevant protein structures deposited at the Protein Data Bank⁸

was done. These Protein Data Bank structures are protein structures which are derived by X-ray crystallography or by nuclear magnetic resonance. Deductions from these analyses were compared with the predictions of the effects of the mutation made by PolyPhen (20). PolyPhen (polymorphism phenotyping) combines information on sequence features, structural variables, and contacts to characterize nucleotide substitutions (20).

The clinicopathologic features of the breast and ovarian tumors are summarized in Table 1. As expected, BRCA1 or BRCA2 mutations were more likely to be present in patients from hereditary breast and ovarian cancer (HBOC) families than in those with no family history. All of the patients with BRCA1 or BRCA2 mutations were Her2 negative. All but one BRCA1 mutation carrier had estrogen, progesterone, and Her2-negative breast cancer.

Deleterious mutations are indicated in bold in Table 2. All frameshift and nonsense mutations generating premature termination codons were classified as deleterious. Missense mutations that were absent in at least 50 normal individuals and which involved a highly conserved amino acid were classified as deleterious. An exon 13 duplication in BRCA1 (21) and the IVS7-15del10 in BRCA1 were deemed deleterious, as there was linkage with breast and/or ovarian cancer within the respective families.

Fifteen novel genetic alterations were detected in the BRCA1 and BRCA2 genes, with 11 classified as deleterious mutations and four as polymorphisms (Tables 2 and 3).

Frameshift, nonsense, and intronic mutations and an exon duplication were identified in 16 of 95 subjects (16.8%) or in 14 of 90 families (15.6%). The 90 families composed of 26 HBOC families, 34 breast cancer only families, one ovarian cancer only family, and 29 unrelated subjects with early-onset breast cancer and no family history (Table 1). The prevalence of clearly deleterious BRCA mutations in these groups were 35% (9 of 26) in HBOC families, 8.8% (3 of 34) in breast cancer only families, and 6.9% (2 of 29) in subjects with early-onset breast cancer and no family history (Table 2). The prevalence rates for clearly deleterious BRCA1 and BRCA2 mutations were 6.7% (6 of 90) and 8.9% (8 of 90), respectively, or 7.8% (7 of 90) and 11.1% (10 of 90), respectively, by including probably deleterious missense mutations. Deleterious mutations in BRCA1 and BRCA2 were observed in 7 and 10 families, respectively.

By ethnicity, these mutations were detected in 12 of 67 Chinese families (17.9%), 1 of 11 Malay families (9.1%), and 1 of 4 Indian families (25.0%; Table 2). The mutation in the Indian family was the 185delAG mutation frequently found in Ashkenazi Jews and in studies from India (22-28). The Malay founder mutation previously identified in Singapore was not identified in this cohort (16, 29).

Six different BRCA1 deleterious mutations were identified in six families (Table 2). Five of these have been previously reported either in the Breast Cancer Information Core or in the literature and comprised three frameshift mutations, one nonsense mutation, and an intronic deletion within intron 7, causing a frameshift with a predicted stop at codon 182 (Fig. 2; ref. 30). A novel genomic rearrangement resulting in the duplication of exon 13 of BRCA1 was identified by MLPA and confirmed by DNA sequencing of the breakpoint (21).

Of the eight deleterious mutations in BRCA2, six were novel (Table 2). There were seven frameshift mutations and a GT to AT 5′ splice site mutation in intron 15, resulting in skipping of exon 15 detected in one subject (Fig. 2; ref. 31).

Eight missense mutations of unknown significance were detected in both BRCA1 and BRCA2, and these were evaluated by computational analyses to determine if they were deleterious (Tables 2 and 4).

None of the mutations listed in Table 2 were detected in 50 normal individuals, matched for ethnicity. Several polymorphisms were detected and are listed in Table 3.

All cases with BRCA1 and BRCA2 intronic splice site alterations were subjected to RNA analysis. Of the 15 intronic alterations detected, two variants, IVS7-15del10 in BRCA1 and IVS15+1G>A in BRCA2, showed aberrant splicing and were classified as deleterious (Table 2; Fig. 2). The remaining 13 variants with normal transcripts were classified as benign polymorphic alterations (Table 3).

IVS7-15del10 in BRCA1, 10-bp deletion in intron 7 of BRCA1, was identified in two subjects who were sisters (Table 2; Fig. 2). Reverse transcription–PCR of the region spanning exons 5 to 9 amplified two fragments, a fragment of the expected size of 472 bp and another fragment of 531 bp of reduced intensity possibly due to nonsense-mediated mRNA decay, which occurs if an alternative transcript has a premature termination codon >55 nucleotides upstream of the last exon-exon junction (32). Sequencing of both fragments revealed that the aberrant fragment had a 59-bp insertion between exons 7 and 8 in the mRNA sequence. Based on this, a possible mechanism for the partial inclusion of intron 7 is shown in Fig. 2. The deletion of the original branch site within the 10-bp intronic deletion causes the utilization of a new branch site upstream with the insertion of the 59-bp intronic sequence, causing a frameshift and a predicted stop at codon 182. SpliceSiteFinder predicted that the strength of the new 3′ splice acceptor site (score, 64.4) and new branch point (maximum score, 100) are at least comparable with or higher than the original scores of 67.8 and 83.9, respectively. Both the putative normal and new branch points are based on the consensus sequence YTRAY (33).

The BRCA2 IVS15+1G>A mutation is located in the consensus sequence of the 5′ donor splice site and resulted in the generation of an altered transcript by skipping of exon 15.

Table 4 shows the amino acid sequence comparison for sites of missense mutations used to establish the degree of conservation and the results of computational analyses for assessing the significance of the missense mutations. Of the eight missense mutations, three are predicted to be damaging mutations (BRCA1 P58A, BRCA2 C161W, and BRCA2 G2544D), two are predicted to be benign (BRCA1 M1652I and BRCA2 A2351G), and the remaining three are of unclear status (BRCA1 V191I, BRCA1 K1690Q, and BRCA2 K2729N; Table 4).

Codon 58 of BRCA1 lies within the RING domain, which is a highly conserved region among the nine mammalian sequences examined. The domain is implicated in interactions with at least five different proteins (34, 35).

Codon 161 of BRCA2 has polar residues conserved in all species represented at that site. A mutation to the large hydrophobic Tryptophan residue is likely to be damaging, as it may remove possible covalent or polar bonding.

The G2544D mutation involves a conserved glycine that lies in a highly conserved region, which plays a structural role in forming a loop turn and is likely to affect the binding affinity of BRCA2 with DSS1. The residues involved in the BRCA2-DSS1 interaction are highly conserved, and two DSS1-interacting BRCA2 residues (Ala2564 and Arg2580) were previously found to be mutated in cancer (36). This BRCA2-DSS1 complex directly functions in RAD51-mediated recombination double-stranded break repair (36). However, PolyPhen, which did not examine the Protein Data Bank structure 1mje, classified this mutation as benign.

Codon 1652 of BRCA1 lies in the hydrophobic core of the protein, and its hydrophobicity is shown to be conserved in the alignment (Table 4). Because the mutation to isoleucine is a conservative mutation, it is most probably a benign mutation. PolyPhen, however, has not evaluated the similarity between the leucine found in rhesus monkey in the alignment and isoleucine, and predicted the mutation to be probably damaging (Table 4).

The A2351G mutation of BRCA2 is a conservative mutation in a nonconserved region and hence is likely to be a benign mutation.

Codon 191 of BRCA1 is within a putative site of interaction with oncogene and cell cycle regulators, such as c-myc and estrogen receptors (37). Although V191I is conserved in hydrophobicity and PolyPhen predicted this mutation to be benign, it is still possible that the V191I mutation may affect the phosphorylation state of BRCA1 because NetPhos 2.0 predicted its neighbor T190 to be a likely site of phosphorylation.

The K1690Q mutation in BRCA1 is predicted to be benign by PolyPhen and unclear by our analysis (Table 4). The Protein Data Bank structure 1jnx of BRCA1's BRCT domain showed a salt-bridge interaction between the lysine residue and the glutamic acid residue at codon 1661 (38). A mutation from lysine to glutamine will replace the salt-bridge interaction with a hydrogen bond between the mutated glutamine residue at codon 1690 with the glutamic acid residue at codon 1661. However, the effect of this mutation on the structure and dynamics of the protein is unclear.

The K2729N mutation of BRCA2 is a relatively conservative mutation in a conserved region. Hence, the effect of the mutation on the structure and function of the protein remains unclear.

This is the first comprehensive study of both the BRCA1 and BRCA2 genes, involving complete mutation screening, evaluation for large genomic rearrangements and splice site variants, and analysis of missense mutations by computational analysis, in an Asian population. In our cohort of 90 families, the combined frequency of frameshift/nonsense mutations in the BRCA1 and BRCA2 genes was 12.2% by conventional mutation screening. However, by including deleterious mutations detected using MLPA analysis (n = 1), RNA analysis (n = 2), and computational evaluation of missense mutations (n = 3), the overall frequency of deleterious mutations increased substantially to 18.9%. The comprehensive mutation testing strategy used was most successful in families with both breast and ovarian cancer.

The prevalence rates for BRCA1 and BRCA2 mutations that were observed in this study were 6.7% (6 of 90) and 8.9% (8 of 90), respectively, or 7.8% (7 of 90) and 11.1% (10 of 90), respectively, by including missense mutations predicted to be deleterious by computational analysis. These frequencies are comparable with data from other studies on Chinese women, the majority of which have been small studies with fewer than 50 subjects. In women with early-onset breast cancer, prevalence rates ranging from 8.0% to 9.5% in BRCA1 (39-42) and 2.4% in BRCA2 (40) have been observed, although one study reported no mutations in BRCA1 among 35 Singaporean-Chinese women (43). It should be noted, however, that the age limits set for each of these studies differ and range from <35 to <45 years, which may account for differences in prevalence rates between studies. In addition, a cancer genetics clinic with risk assessment allows risk stratification for genetic testing of a higher risk cohort.

In Chinese cases with a family history of breast or ovarian cancer, mutations in BRCA1 were detected between 8.1% and 12.5% (30, 40, 44, 45) and between 2.7% and 16.7% for BRCA2 (30, 40, 44, 45). In one study on 25 HBOC families, 40% of the women had mutations in BRCA1 (46).

The majority of the BRCA2 frameshift mutations detected were novel mutations. This may be because there is a paucity of information on BRCA2 mutations among the Chinese, Malay, and Indian ethnic groups. Furthermore, we found that in our cohort of Singaporean women, deleterious mutations in BRCA2 (10 families) were more common than for BRCA1 (seven families). This contrasts with observations from European and American populations, wherein the frequency of BRCA1 mutations is higher than that of BRCA2 (1, 3). In a large U.S. study of 1948 families, BRCA1 and BRCA2 mutations were detected in 14.6% and 7.4% of the families, respectively (3). However, in Chinese women from Shanghai, a 1:1 ratio for BRCA1/BRCA2 mutations was observed (44). Thus, our findings underscore the importance of screening for BRCA2 mutations in individuals at high risk of developing breast and/or ovarian cancer.

Although there have been hospital-based genetic testing studies for BRCA1/BRCA2, there have been no Asian risk assessment clinic-based reports with predominantly Asian individuals seeking counseling and genetic testing (22, 47, 48). The advantages of clinic-based ascertainment is, firstly, the use of risk stratification to allow for genetic testing of a higher risk cohort and, secondly, the exclusion of breast cancer families related to other predisposition genes. We excluded one family with Cowden's syndrome and another with Li-Fraumeni syndrome (data not shown). Additionally, we screened for the CHEK2*1100delC mutation, which also predisposes to early-onset breast cancer, and all our subjects were negative (unpublished data).

The technology to sequence BRCA1 and BRCA2 is increasingly available in Asian countries, and although genetic testing is often done as part of research studies, the set up of formal assessment clinics could increase the scope of hereditary cancers detected.

To ascertain the pathologic effect of the intronic alterations on mRNA splicing fidelity and expression, RNA analysis using reverse transcription–PCR and sequencing and in silico splice prediction methods were used (9, 10). This current study identified two deleterious intronic aberrations, IVS7-15del10 in BRCA1 and IVS15+1G>A in BRCA2. The occurrence of these splicing aberrations emphasizes the importance of studying mutations both at the genomic DNA and RNA levels to determine the pathogenic effect of the mutations (49) which would effect on genetic counseling. However, if RNA is not available, then another option would be to use theoretical splicing prediction approaches, such as SpliceSiteFinder, to determine if the BRCA1 and BRCA2 splice site variants identified from genomic sequence would cause aberrant splicing (10).

Missense mutations of unknown clinical significance or unclassified variants pose a challenge in genetic counseling. By using bioinformatic tools to assess such mutations, deleterious mutations may be identified in high-risk individuals, providing information for risk reduction decision making. Such computational tools provide a feasible alternative to the evaluation of missense mutations by segregation analysis, which is particularly difficult in small families or in families for which pedigree information is unknown or limited.

This study has shown that intronic alterations, large genomic rearrangements, and missense mutations in combination contribute significantly to deleterious mutations in our study population. Improved mutation detection by a comprehensive screening approach for the BRCA1 and BRCA2 genes, as shown here, may be effective for the identification of deleterious mutations, with implications for genetic counseling and risk assessment of individuals at high-risk for breast and/or ovarian cancer.

We thank Geraldine Low Hui Min and Dr. Karen P.L. Yap for excellent technical assistance, the referring doctors, patients, and their families who have participated in this study, and WHO Immunology Center, Singapore for providing control samples for this study.