METHODS

Human subjects

Genomic DNA was prepared from peripheral blood from breast cancer patients using DNA isolation kits from Puregene (Gentra Systems, Minneapolis, Minnesota, USA) according to the manufacturer’s instructions. Patients were all Nigerian women diagnosed with breast cancer at the age of 40 years or younger (median age 37 years; range 27 to 40), and samples were selected from a consecutive series of newly diagnosed breast cancer cases seen at the University College Hospital, Ibadan, Nigeria. Thirty three of 39 patients (85%) reported no family history of cancer. Of the remaining six (15%), five reported one case of breast cancer in a first degree relative, and one reported cervical cancer in her sister and grandmother. The patient distribution among Nigerian ethnic groups was as follows: 25 Yoruba (64.1%); three Igbo (7.7%); three Urhobo (7.7%); two Edo (5.1%); one Delta (2.6%); one Goemai (2.6%); one Kagoro (2.6%); one Ishan (2.6%); one Owan (2.6%); and one unknown (2.6%). Thus most of the cases were Yorubas.

Seventy four control subjects, also Nigerian women seen at the University College Hospital at Ibadan, were given breast examinations to confirm that no breast cancer was present. These women had a median age of 37 years old (range 22 to 66). Seventy of them (94.6%) reported no family history of breast cancer. The remaining four (5.4%) reported cancers in first degree relatives but did not specify the type of cancer. The control distribution among Nigerian ethnic groups was as follows: 67 Yoruba (90.5%): two Igbo (2.7%); two Edo (2.7%); one Urhobo (1.4%); and one Tbibio (1.4%). Thus, as with the cases, the controls were predominantly Yorubas.

All patients and control subjects gave informed consent after appropriate counselling according to the protocols approved by the Universities of Chicago and Ibadan institutional review boards.

Heteroduplex analysis

We screened the entire coding sequence and all intron/exon junctions using heteroduplex analysis (HA) followed by DNA sequencing as previously published.^23,24 Briefly, PCR products were heated at 95°C for three minutes, then slowly cooled to 37°C over 40 minutes. After mixing with 6× triple dye loading buffer at 1:5 volume ratio, the PCR products were run on 1×MDE gel (FMC BioProducts, Rockland, Massachusetts, USA) with 15% urea at 400 volts at room temperature for 14 to 20 hours. The gels were stained with in SYBR™ Green I nucleic acid (FMC BioProducts) and photographed under ultraviolet light.

Protein truncation test

In addition to heteroduplex analysis, the protein truncation test (PTT) was used to screen genomic DNAs of exon 10 from BRCA2 and exon 11 from both BRCA1 and BRCA2, as previously published.²⁴ The forward primers include additional 5′ sequences containing the T7 promoter and a translation start site. PTT using the TNT-T7 coupled wheat germ extract system (Promega, Madison, Wisconsin, USA) was carried out according to the manufacturer’s instructions, incorporating ³⁵S-methionine or ³⁵S-cysteine. Samples were electrophoresed on 12.5% acrylamide gels and exposed to Kodak x ray film. Candidate positives were sequenced.

DNA sequencing

PCR products showing an electrophoretic variant pattern by HA or PTT were reamplified from the original genomic DNA and both strands were directly sequenced by the fluorometric method with automated sequencing procedures (Dyedeoxy Terminator cycle sequencing kit and ABI377 DNA sequencer; Applied Biosystems, Foster City, California, USA).

Analysis of variants

Data regarding clinical significance and ethnic occurrence of previously reported alleles were taken from the breast cancer information core (BIC) website http://research.nhgri.nih.gov/bic/. Predictions of exonic splicing enhancer (ESE) consensus scores were made using previously described scoring matrices.^25–27 Predictions of which amino acids are not theoretically tolerated at specific positions within BRCA1 and BRCA2 were generated using the “sorting intolerant from tolerant” (SIFT) website http://blocks.fhcrc.org/sift/SIFT.html. This method searches public databases for protein homologs, performs alignments, and then generates a score for the theoretical tolerance of each possible amino acid substitution. Records used for single nucleotide polymorphism (SNP) frequencies are kept in the SNP500Cancer database http://snp500cancer.nci.nih.gov/.

RESULTS

In this cohort of early onset Nigerian breast cancer cases, 29 of 39 (74%) carried at least one genetic variation in BRCA1, BRCA2, or both. BRCA2 variations were found in a surprising 27 cases (69%) (table 1). Thirty four different variants were detected, four in BRCA1 and 30 in BRCA2 (table 2). These included nine intronic and 25 exonic variations. Of these variations, only one (2.5%) is a known deleterious truncating mutation (BRCA2 3034del4). Of the 21 previously reported non-truncating exonic variants,²⁸ 15 are described as “unclassified variants” and only six are described as polymorphisms. The remaining four exonic variants are reported here for the first time. Seven of the nine BRCA2 intronic variations have not been reported previously. It is important to note that the heteroduplex method used here identifies heterozygous genotypes, and the frequencies of the alleles reported does not include the frequencies of potential variant homozygotes.

To test the potential clinical significance of these variants, we examined the frequency of amino acid substitution variants in a population of 74 unaffected Nigerian samples (that is, 148 chromosomes) (table 2). BRCA1 and BRCA2 single base changes not expected to result in amino acid substitutions were not examined. Five of 13 predicted BRCA2 amino acid substitution variants (38%) were not detected in the unaffected population, indicating that they may be rare high risk alleles. However, several of the alleles found in the Nigerian cohort have been described as single nucleotide polymorphisms (SNPs) by the National Cancer Institute cancer genome anatomy project “SNP500Cancer” database (http://snp500cancer.nci.nih.gov/). It is reported that BRCA2 alleles N289H, L929S, N987I, N991D, S2414S, H2440R, and I2944F all exist in reference samples and are just as frequent in patients of African ancestry as in those of other ethnicities. Curiously, the BRCA1 K1183R allele is also described as an SNP found in unaffected populations, but may be less common in individuals of African ancestry. However, the alleles found in our patient set but not in our control population (BRCA2 A248T, Q713L, D1902N, L1904V, and Q2384K) are also not described in the SNP database. While it is possible that these rare variants found in patients may be deleterious, we did not have sufficient power to determine whether their absence from control populations was statistically significant.

To investigate whether exonic base changes found in African BRCA1 and BRCA2 variants might affect correct message splicing, we examined exonic base changes and the surrounding sequence for disruption of high scoring ESE consensus sequences.^{25–27,30–32} ESEs are SR protein binding sequences that may lie within exons of pre-mRNAs that indicate exon identity during splicing. Mutated ESEs could therefore result in exon skipping during splicing. Eight of 24 non-truncating exonic base changes (33%) in the African population fall within high scoring predicted ESE, and the base substitutions associated with five of these alleles are predicted to cause ESE disruption (BRCA2 A248T, Q713L, T1414M, D1902N, and V2171V). Two of these (T1414M and V2171V) are seen in our unaffected population as well as in patients. It is important to note that ESE mediated splicing patterns are complex, and not all disruptions of predicted ESE consensus motifs will necessarily result in high penetrance exon skipping phenotypes. However, such a mutation could cause shifts in relative levels of alternate splicing products, resulting in limited or conditional clinical phenotypes. Finally, we used methods on the SIFT website to predict which amino acid substitutions found in BRCA1 and BRCA2 protein variants in the African population are unlikely to be tolerated at mutant positions (see Methods). Of these variants, all the amino acid substitutions seen in BRCA1 (I379M, K820E, and K1183R) are predicted to contain non-tolerated amino acid substitutions by scoring based on homolog alignment, and are thus potentially clinically important base changes. Interestingly, BRCA1 I379M was also identified as a potentially deleterious amino acid substitution by an “ancestral sequence” method.³³ Of the BRCA2 amino acid substitutions seen in the African cohort, alleles N987I, N991D, and I2944F are predicted to be non-tolerated amino acid substitutions within the limits of confidence in the alignments.

Neither ESE nor SIFT scoring mechanisms constitute definitive evidence of the clinically deleterious nature of the base changes observed. However, they suggest a rational and testable basis for anticipating that many of the functionally uncharacterised mutations described here could be deleterious alleles, perhaps some with limited penetrance. Unfortunately, we do not have additional biological materials to carry out further tests on the patients. Nonetheless, we suggest that the numerous variations in BRCA1 and especially BRCA2 in African breast cancer patients may contribute to disease occurrence and should be characterised functionally.

DISCUSSION

We previously investigated the frequency of BRCA1 and BRCA2 protein truncating mutations in a cohort of 70 young breast cancer patients ascertained in the surgical oncology unit at the University College Hospital Ibadan, Nigeria.²² We reported 4% of the population carried deleterious mutations (two truncating mutations in BRCA1 and one in BRCA2), well below the frequencies of deleterious mutations found in other clinic or population based cohorts.³⁴ However, we also reported a preliminary observation of 20 non-truncating BRCA2 variants in 18 individuals (23%). In the present study we re-examined a subset of the original population to better define the frequency of BRCA1 and BRCA2 non-truncating alleles and assess whether these may also be a source of breast cancer susceptibility in young African breast cancer patients. Here we report a single truncating BRCA2 mutation and no additional truncating mutations in the 39 patient cohort (2.6%), but 29 (74%) carried non-truncating variants in BRCA2. These observations are expand our previous observations and are consistent with observations by Wagner and colleagues, who found 45 different BRCA2 sequence variants in 21 unrelated unaffected African individuals, with 33 of 45 (73.3%) being unique to African populations (reported in BIC or for the first time in 1998).²⁹ Other studies using different mutation ascertainment methods have shown high levels of BRCA2 sequence variability and a distinct spectrum of variants in patients of African ancestry.^29,35 This observation is consistent with reports of an exceptionally high level of overall genetic diversity in people of African ancestry.^29,36

Newman and colleagues reported a lower frequency of deleterious BRCA1 mutations in black than in white women (3.3% v 0%) in a population of unrelated patients diagnosed with invasive breast cancer, but not selected for early age onset or family history.³⁷ Other studies indicate that frequencies of deleterious BRCA1 and BRCA2 mutations are just as common in African American high risk patients as in other ethnic groups.^11,38,39,40 However, these studies used clinic based subjects who were referred for genetic testing based on family history, whereas our consecutively ascertained clinic based subjects were selected only for early onset breast cancer. Indeed, it has been shown that rigorous family history criteria predict deleterious BRCA1 and BRCA2 mutations just as strongly in African Americans as other populations.^{11,35,38–40} However, these findings are unlikely to reflect the distribution of BRCA1 and BRCA2 mutations in general populations, which may explain the similarity between our results and those of Newman et al.³⁷

Other studies of genetic contributions to breast cancer in different ethnic groups have revealed population specific patterns of BRCA1 and BRCA2 mutation frequencies and spectra associated with breast cancer. Patient selection criteria and mutation detection methods vary widely in these studies, so it is difficult to compare mutation frequencies among published investigations. However, several reveal that both the frequency of variants and deleterious mutations is higher in BRCA2 than BRCA1. This has been seen in populations from Mexico,⁴⁰ Spain,⁴¹ Sardinia,⁴² and Japan.⁴³ One study that examined a population of Indian breast cancer patients showed similar frequencies of variants per se in BRCA1 and BRCA2, but all the BRCA1 variants were deleterious, while all the BRCA2 variants were clinically unclassified.⁴⁴ Finally, several studies have described populations in which the frequency of unclassified variants and deleterious mutations in BRCA1 are both more frequent than in BRCA2. These include cohorts from the USA,^45,46 Canada,⁴⁷ Germany,^48,49 and China.⁵⁰ The observation that our African cohort showed a much greater frequency of variability in BRCA2 than BRCA1 is surprising, as breast cancers in BRCA1 mutation carriers have more clinical similarity to breast cancers in patients of African descent. It remains possible that some of the patients carry distant promoter mutations, splice disrupting mutations deep within intronic regions, or large scale rearrangements of BRCA1 not detected by our methods. Additionally, it is possible that BRCA1 mutation carriers simply have breast cancer characteristics in common with other early onset breast cancers, and such characteristics will probably occur with a high frequency in disproportionately young populations. Germline BRCA2 mutations are associated with early onset breast cancers that are otherwise phenotypically similar to sporadic breast cancers.⁵¹ Because of the frequency of the unclassified variants in our Nigerian population, we suggest that some of these variants may be deleterious to some extent. It is interesting to note that, in a clinic based cohort of African Americans with family histories of breast cancer, who are thus more likely to carry BRCA1/2 mutations, the frequency of BRCA2 single base changes is even higher than in the age selected Nigerian cohort (Fackenthal JD and Olopade OI, unpublished observations).

Among the most clinically useful observations from BRCA1 and BRCA2 genetic analyses in individual ethnic populations is the existence of population specific founder mutations. For example, among Ashkenazi Jews, 20% of patients diagnosed before the age of 40 carry the BRCA1 185delAG allele, whereas only 10% of a general white population with early onset breast cancer may carry BRCA1 mutations of any kind.^52,53 Additionally, the BRCA2 999del5 allele has been identified as a founder mutation in the Icelandic population and has been detected in 0.6% of the total population, 7.7% of all female breast cancers, and 40% of all male breast cancers.^54–56 In populations of African descent, such founder mutations have been difficult to identify, possibly because of the diversity in the African diaspora. Several reports have shown that a single allele, BRCA1 943ins10, is common to patients of African ancestry detected in Washington DC, Florida, South Carolina, Bahamas, and the Ivory Coast.^58,59,60,61 This allele does not occur often enough to provide expediency in clinical diagnostics, but it does suggest that the mutational spectrum of BRCA1/2 may be distinctive in Africans and African Americans.

The mutations identified in BRCA2 in our Nigerian cohort were distributed throughout the coding region and did not fall predominantly into known functional domains or cluster into potential novel domains. Several amino acid substitutions occurred more than once, and some of these have been reported only in African or African American individuals, or are reported here for the first time (table 2). While these variants may contribute to a unique spectrum of BRCA2 mutations in subSaharan African populations, we cannot conclude from currently available data that these alleles are unique to individuals of African ancestry.

However, the unique spectrum of BRCA2 alleles seen in these African patients may reflect BRCA2 involvement in early onset breast cancer in African populations. Several observations suggest that some of these mutations may represent deleterious or incompletely penetrant alleles. First, five of 13 amino acid substitutions (38%) that were detected in the breast cancer population were not detected in the control population. While our sample size is too small for definitive statistical analysis, this observation suggests these alleles could be rare high risk alleles. Additionally, eight of 14 BRCA2 exonic single base substitutions (57%) fell within high scoring ESE motifs, and the majority of these are motif disrupting alleles that could result in exon skipping. ESE disruption alleles that result in exon skipping have been reported in numerous cancer related genes, including BRCA1 and BRCA2.^25,31,60,61 Finally, all three BRCA1 amino acid substitutions and three of 13 BRCA2 amino acid substitutions (23%) are predicted to be disruptive to protein structure based on scoring from alignment of homologs. Many DNA sequence variants that result in an alteration of protein structure, protein binding ability, or aberrant splicing patterns are likely to have limited penetrance or be dependent upon genetic or environmental backgrounds for expressivity. To address the potentially subtle nature of such hypothetical alleles, it will be necessary to construct haplotypes containing variants of interest in linkage disequilibrium with other markers. Such haplotypes are potentially more likely to cosegregate with disease than any single marker component. Taken together, these observations suggest that the high variability of BRCA1 and BRCA2 sequences could contribute to early onset breast cancers seen in patients of African descent. We suggest that functional characterisation of non-truncating alleles like the ones reported here will be necessary in the future in order to understand the structural requirements for BRCA1/2 tumour suppressor functions and provide informed genetic counselling for patients from ethnically diverse backgrounds.