Human Genetics

, Volume 116, Issue 6, pp 497–506

BRCA1 variants in a family study of African-American and Latina women

Authors

    • Department of Preventive MedicineUniversity of Southern California Keck School of Medicine, Norris Comprehensive Cancer Center
  • Heather Spencer Feigelson
    • Department of Epidemiology and Surveillance Research American Cancer SocietyNational Home Office
  • Lucy Y. Xia
    • Department of Preventive MedicineUniversity of Southern California Keck School of Medicine, Norris Comprehensive Cancer Center
  • Celeste Leigh Pearce
    • Department of Preventive MedicineUniversity of Southern California Keck School of Medicine, Norris Comprehensive Cancer Center
  • Duncan C. Thomas
    • Department of Preventive MedicineUniversity of Southern California Keck School of Medicine, Norris Comprehensive Cancer Center
  • Daniel O. Stram
    • Department of Preventive MedicineUniversity of Southern California Keck School of Medicine, Norris Comprehensive Cancer Center
  • Brian E. Henderson
    • Department of Preventive MedicineUniversity of Southern California Keck School of Medicine, Norris Comprehensive Cancer Center
Original Investigation

DOI: 10.1007/s00439-004-1240-5

Cite this article as:
McKean-Cowdin, R., Spencer Feigelson, H., Xia, L.Y. et al. Hum Genet (2005) 116: 497. doi:10.1007/s00439-004-1240-5

Abstract

We sequenced the entire coding region of BRCA1 to improve our understanding of the frequency and nature of BRCA1 variants in African-American and Latina women identified from a multiethnic cohort in Los Angeles, California. The study included 109 African-American and 140 Latina sibships from families with two or more cases of breast or ovarian cancer among first-degree relatives. BRCA1 was sequenced in 278 breast or ovarian cancer cases and 229 unaffected sisters. The proportion of cases with known disease-causing mutations was low (0.72, 95% confidence interval: 0–1.7%). In total, 33 sequence variants were identified, including two protein truncation mutations, one deletion, and six silent and 24 missense variants. Two novel rare variants were identified that appeared to act as benign polymorphisms. Four rare variants may be unique to women of African descent based on existing literature, and three have been described exclusively in Latina women. The frequency of common variants was similar for cases and controls, but the frequency of common variants for African-American women significantly differed from those previously described for Caucasian women. We believe this to be the largest study of high-risk African-American and Latina women sequenced for variants in the BRCA1 gene to date.

Introduction

Since the identification (Hall et al. 1990; Newman et al. 1988) and cloning of the BRCA1 gene (Futreal et al. 1994; Miki et al. 1994) in the early 1990s, protein truncation mutations in BRCA1 have been recognized to predispose women to early-onset breast and ovarian cancer. Penetrance estimates from family studies and population-based samples suggest that women with a BRCA1 mutation have a lifetime risk of developing breast or ovarian cancer between 26% and 80% depending upon other modifying genetic, reproductive, or lifestyle characteristics (Begg 2002; Burke and Austin 2002; Easton et al. 1993; King et al. 2003). Despite the compilation of BRCA1 data in the past decade, the spectrum of sequence variants that occurs in populations other than those of white European descent and the contribution of non-truncating alleles of BRCA1 genetic variants (e.g., missense or silent) to risk of breast or ovarian cancer are understudied. A small number of missense variants (n=22) are currently identified as being disease-associated in the Breast Cancer Information Core (BIC) database (BIC 2003). For most of these variants, functional work has not been completed to verify or disprove these assumptions. Currently, there are over 730 unique BRCA1 variants reported in BIC, and many are of uncertain relevance to breast or ovarian cancer risk. Identification and analysis of rare and common sequence variants, particularly among families from multiple racial/ethnic groups, may help us understand the degree of allelic variation associated with the BRCA1 gene and the potential relevance to breast cancer risk.

The contribution of BRCA1 mutations to breast cancer incidence in African-American women is of particular interest, because these women experience high rates of early-onset breast cancer. The rate of breast cancer in the US for African-American women under 40 years of age is 65.7 per 100,000, which is higher than the rate in US white women (57.2 per 100,000), despite the fact that the overall incidence rate of breast cancer in African-American women in the US is relatively low (332.4 per 100,000 for African-American women vs. 395.5 per 100,000 for white women; Ries et al. 2002). BRCA1 sequence variants have been described in several studies of African or African-American women (Dangel et al. 1999; Futreal et al. 1994; Gao et al. 1997, 2000a, 2000b; Mefford et al. 1999; Miki et al. 1994; Newman et al. 1998; Panguluri et al. 1999; Shen et al. 2000; Stoppa-Lyonnet et al. 1997). In these studies, a few missense variants have been described as being potentially associated with breast cancer risk, because they were identified exclusively in breast cancer cases or because they caused an amino acid change in a region highly homologous with mouse DNA (Newman et al. 1998; Shen et al. 2000). Additional studies of African-American high-risk families, characterized by early-onset disease, multiple cases in first-degree relatives, or breast and ovarian cancer in the same family have reported a frequency of variants between 4% and 56% (Gao et al. 1997, 2000a; Panguluri et al. 1999). However, the sample sizes for these studies are relatively small with the largest study including only 45 breast cancer cases.

Latina women in the US and worldwide have a lower incidence of breast cancer compared with US whites (Howe et al. 2001; Pike et al. 2002). Much of this difference has been attributed to differences in lifestyle, particularly reproductive characteristics, among Latina women (Pike et al. 2002). However, the extent that these differences may reflect an underlying genetic susceptibility has received little attention. No study has specifically addressed the prevalence of BRCA1 variations among Latina or Hispanic women originating from central or south America. However, studies of Spanish (European) families with multiple cases of breast and ovarian cancer have found novel sequence variants (Llort et al. 2002; Osorio et al. 2000).

In the current study, we describe the frequency and nature of BRCA1 variants in a sample of African-American and Latina women identified from a multiethnic cohort in Los Angeles, California based on family history of breast or ovarian cancer. The analysis includes 278 women affected with breast or ovarian cancer and their healthy sisters (n=229) from families with at least two cases of breast or ovarian cancer among first-degree relatives. We believe this to be the largest study of high-risk African-American and Latina women sequenced for variants in the BRCA1 gene to date.

Materials and methods

Study population

Probands for this study were members of the Multi-ethnic Cohort Study (MEC; Kolonel et al. 2000). The cohort was assembled in Los Angeles and Hawaii from 1993 to 1996; it totals 215,251 men and women aged 45–74 years at inception and includes African-Americans, Japanese, Hawaiians, Latinos, non-Latino whites, and small numbers of other racial/ethnic groups. Approximately 75% of the Latino families are of Mexican origin, and the remaining 25% are descended from other central and south American countries. Further details of the cohort study are provided elsewhere (Kolonel et al. 2000). This family study includes African-American and Latino sibships ascertained through the Los Angeles component of the larger cohort study. From a detailed family history questionnaire, we identified 455 families with at least two cases of female breast cancer or at least one female breast and one ovarian cancer in the proband, their full-sisters, or their mother. No restriction was placed on age of cancer diagnosis. From these families, we enrolled 331 families, including 153 families having at least one living affected sister (with breast or ovarian cancer) and at least one living unaffected sister or a sister diagnosed with breast cancer who had survived past the age of diagnosis of the initial case. We attempted to enroll all living sisters into the study, including families with multiple cases of breast or ovarian cancer and in which all sisters were affected or no living unaffected sisters remained to be enrolled in the study. Sisters were contacted first by letter and then by telephone to invite participation in the study. We attempted to contact individuals with up to three letters and nine telephone calls covering different days of the week and hours of the day. For 91 of the 331 families, we collected blood from a breast cancer case but not a healthy sibling control, and for 87 families, the proband agreed to complete the family history questionnaire but chose not to submit a blood sample.

After appropriate informed consent was obtained, all participants were asked to provide a blood or buccal sample and complete a 26-page questionnaire that included information regarding medical history, family cancer history, diet, medication use, physical activity, reproductive history, and use of hormones. Permission to obtain medical records was requested from all cases to verify diagnosis and provide information on tumor characteristics. Blood samples were collected from the majority of participants (94% blood, 6% buccal).

Population controls

In order to provide accurate population frequency estimates of sequence variants identified in these families, we used 48 African-American and 48 Latina females (96 total) who were randomly selected from MEC participants who had not been diagnosed with any type of cancer and who reported no family history of breast or ovarian cancer in their baseline questionnaire.

Laboratory methods

We developed high throughput methods for sample preparation and DNA sequencing consisting of six steps: (1) DNA extraction, (2) polymerase chain reaction (PCR) amplification, (3) PCR product purification, (4) fluorescent dye labeling and extension, (5) extension product purification, and (6) sequencing.

DNA extraction

Genomic DNA was extracted from 300 μl buffy coat or one vial of mouthwash sample (buccal cell) by using the Puregene DNA isolation kit (Gentra Systems) for buffy coat or the QIAamp DNA blood mini kit (QIAGEN) for buccal cells, following the manufacturers’ protocols.

PCR amplification and product purification

Amplification of the entire coding sequence and all intron-exon boundaries of BRCA1 from each of the DNA samples was performed. PCR with 22 pairs of primers was used to amplify exons 2, 3, and 5–24. Exon 1 was not amplified, because the translation start site is in exon 2. Exon 4 was excluded, because it is known to be a variant exon not seen in normal BRCA1 messenger RNA. Primer sequences used were described and are available in BIC.

PCRs for all coding exons, except exon 11, were carried out in a total volume of 25 μl. Each PCR mix contained 30 ng genomic DNA as template, 40 pmoles each exon-specific forward and reverse primer, 100 uM dNTPs, 2 U Taq polymerase (Amersham Pharmecia Biotech) and 1× reaction buffer (0.5 M MgCl2). The PCR amplification consisted of 30 cycles with denaturation at 94°C for 1 min, annealing from 52°C to 58°C for 1 min (depending on the melting temperature of the specific primer pairs), and extension at 72°C for 1 min. An initial denaturation step of 3 min at 94°C and a final extension at 72°C for 10 min were employed. Amplification of exon 11 was performed by using a GeneAmp XL PCR kit (PE Applied Biosystems). The exon 11 PCR was carried out in a total volume of 100 μl. All PCR products were amplified in a 96-well plate format by using a PTC-100 thermal cycler (MJ research).

PCR products were then purified by ultrafiltration on MultiScreen FB filter plates (Millipore Corporation). The purified PCR product was eluted in 50 μl double-distilled H20, and 5 μl was analyzed on a 1% agarose gel containing ethidium bromide to ensure that the product was present for each subject.

Fluorescent dye labeling and extension, purification, and sequencing

Sequencing of the purifed PCR products was performed with an ABI PRISM BigDye Terminator Cycle Sequencing Ready Reaction Kit (PE Applied Biosystems). The labeling and extension reactions were carried out in a total volume of 20 μl containing 5–10 ng purified PCR fragments as template, 6 pmoles exon-specific forward or reverse primer (identical to those used in the original PCR), 1.34 μl Ready Reaction Premix, and 1× reaction buffer. Sequencing reactions were run for 25 cycles with denaturation at 96°C for 10 s, annealing at 50°C for 5 s, and extension at 60°C for 4 min.

To generate high quality DNA sequence data, unincorporated dye terminators were removed from the extension product prior to analysis by capillary electrophoresis by using MultiScreen-HV 96-well Filtration Plates (Millipore Corporation) prepared following the manufacture’s directions. After purification, samples were denatured at 90°C for 2 min and immediately placed on ice for 1 min. Sequencing was performed on an ABI PRISM 3700 DNA Analyzer (PE Applied Biosystems).

Sequence analysis

Data collected from the ABI detection system was processed by using software developed at the University of Washington (Ewing and Green 1998; Ewing et al. 1998; Gordon et al. 1998; Nickerson et al. 1997). All common and rare sequence variants identified by using the forward sequences were subsequently verified by using the reverse sequences. Rare variants were confirmed by repeating the entire PCR-sequencing procedure to be certain that DNA polymerase errors were not responsible for creating the rare variant. All nucleotide calls and sequence traces made by the software program were visually reviewed and verified by our laboratory specialist (L.X.). A database of all sequence variants was created by using an export PERL program designed at the University of Southern California (D.S.).

Statistical analysis

The frequency of sequence variants are described for three sub-groups: sisters with breast or ovarian cancer, sisters without a history of breast or ovarian cancer, and cohort controls without a history of breast or ovarian cancer. Tables include all sisters, even when multiple cases or multiple unaffected sisters were enrolled. Frequency tables were stratified by the two major racial/ethnic groups, by variant frequency (common; rare, <5% of chromosomes), and by variant type (frameshift, missense, silent, insertion, etc). A chi-square test was used to compare observed allele counts for our African-American and Latina cases and controls to expected counts, based on allele frequencies of Caucasians in the literature.

To estimate cancer risk, accounting for multiple case or unaffected sisters serving as controls in the same family, conditional logistic regression with SAS-PHREG was completed for common sequence variants by age of diagnosis (Siegmund et al. 2000). This analysis permitted the inclusion of families in which all sisters were affected with breast or ovarian cancer, by allowing sisters to serve as sibling-controls until the age that they were diagnosed with breast or ovarian cancer. Population controls and 105 cases without a sibling control were not considered in the case-control analyses. To evaluate possible effect modification by other risk factors, we examined whether risk associated with these common sequence variants differed by youngest age of diagnosis in family, age of menarche, parity, oral contraceptive use, number of first-degree relatives with cancer, height, and body mass index.

Results

In total, 331 of 455 eligible families (73%) were enrolled into the study. The primary reasons given for non-participation were that the proband was not in good communication or standing with their sisters, concerns about privacy, or advanced age or illness among sisters. The families enrolled tended to be large, with an average of six total sisters (living or deceased) per family (5.7 in African-American families and 6.1 in Latina families). On average, 2.3 family members (median of two and maximum of five) reported having been diagnosed with breast or ovarian cancer. In eight families, all sisters had been diagnosed with breast cancer. The mean age at diagnosis of breast cancer was 57 years (n=270) and 49 years (n=10) for ovarian cancers. Blood was drawn, on average, 10 years after diagnosis. The unaffected sisters were approximately 3 years younger than their affected sibs at the time of blood sampling; however, most had lived past the age of diagnosis of their case sisters. The Latina women were slightly younger than the African-American women at time of diagnosis and time of blood sampling. Of the families, 100 included a mother or a sister with a history of breast cancer diagnosis before the age of 50 years; 36 of the families included a family member diagnosed with breast cancer before the age of 40 years or breast and ovarian cancer in the same family, or included a male breast cancer. Complete sequencing was performed on all participating sisters from 249 families (75% of families enrolled), with priority being given to samples from families with a case-sibling control pair. In 105 families, only one member (always a breast or ovarian cancer case) provided a blood sample. In the remaining 144 families, up to 7 siblings provided a blood sample for analysis. Table 1 shows the characteristics of the 249 families with complete sequencing information.
Table 1

Characteristics of study population

Subjects

African-American

Latino

Total

Total families sequenced

109

140

249

Affected sistersa

121

157

278

 Breast cases

118

152

270

 Ovarian cases

3

7

10

Unaffected sisters

78

151

229

Median age (range) of affected sister at diagnosisb

58.8 (27–78)

57.9 (19–85)

58.0 (19–85)

Median age (range) of affected sister at blood samplingb

67.7 (50–81)

66.8 (45–90)

67.8 (45–90)

Median age (range) of unaffected sister at blood sampling

67.0 (45–86)

63.1 (41–87)

64.4 (41–87)

aTwo cases were diagnosed with both breast and ovarian cancer

bBirthdate missing for one case

The proportion of cases with a known disease-causing mutation in our African-American and Latina sample was 0.72% (95% confidence interval: 0%–1.7%). In total, we identified 33 sequence variants in the coding regions of BRCA1 (Fig. 1), including two protein truncation mutations (one frameshift and one nonsense mutation), a deletion resulting in the loss of a single amino acid, and six silent and 24 missense variants. Comparison of our findings with BIC (2003) and existing literature (Olopade et al. 2003) indicates that four of the 33 variants (I379M, H476R, R601H, M1783T) may be unique to Africans or African-Americans, and three have been described exclusively in Latina women (A280G, S784L, V1804V). The frameshift mutation, 943Ins10, was found in an African-American women who was diagnosed with breast cancer in her 40s and with a second breast cancer 10 years later. A healthy sister, who participated in the study, did not carry the mutation. We identified a stop codon (R1443X) in a Latina woman who was diagnosed with ovarian cancer in her 40s, and whose siblings died of breast and colon cancers. Her living healthy sister did not carry the mutation. The single amino acid deletion (S616del) was identified in two African-American families, including one breast cancer case from each family and three total healthy sibling controls. The first case was diagnosed in her 60s, and the second case was diagnosed in her 70s; the healthy sisters had all reached their 70s at the time of enrollment.
Fig. 1

Location of 33 BRCA1 sequence variants in relation to exons, BRCA domains, and select protein-binding domains, as described in Welcsh et al. (2000)

Two novel rare variants were identified in our data set (amino acid designations: V1804V and R601H) and appeared to act as benign polymorphisms. Mutation V1804V, located in exon 23, was found in three Latina sibling controls, but no cases. Two of the V1804V carriers were unaffected sisters from the same family; a third sister, who had been diagnosed with breast cancer after age 50, was not a carrier. The third carrier of the V1804V mutation was among a group of sisters of whom five provided a blood sample for sequencing (four healthy controls and one breast cancer case diagnosed in her 20s); the sister diagnosed with breast cancer was not a carrier. Variant R601H, located in exon 11, was identified in three African-American sisters without a history of cancer, but not in their two sisters who had been diagnosed with breast cancer.

Table 2 lists the frequencies of the rare sequence variants by ethnicity. Of the 17 rare or less common variants discovered in the African-American sibships, only three, viz., L1564P (exon 16), M1783T (exon 22), and the frameshift (943Ins10), were found exclusively in African-American cases and not in their unaffected sibling or population controls. However, no testing was performed on any family members of the L1564P carrier, because her only sister had died from breast cancer. The participant had been diagnosed with breast cancer in her 50s. The carrier of the second variant, M1783T, was diagnosed with breast cancer in her 40s from a family that included four sisters diagnosed with breast or ovarian cancer (samples provided by two sisters were sequenced); her sister diagnosed with ovarian cancer was not a carrier of M1783T. Five of the 17 rare variants found in African-American families were also detected in the population controls.
Table 2

BRCA1 rare variant frequencies by chromosomes from African-American and Latina breast/ovarian cancer cases (CA), sibling controls (CO), and MEC population controls (CO) with no history of cancer

Exon

Nucleotide

Designation

African-American

Latina

Cancer CAb % (95% CI)

Sibling COc % (95% CI)

Population COd % (95% CI)

Cancer CAb % (95% CI)

Sibling COc % (95% CI)

Population COd % (95% CI)

Missense

9

C676A

S186Y

0.8 (0.0, 2.0)

2.0 (0.0, 4.2)

0

0

0

0

11

C958G

A280G

0

0

0

0

1.7 (0.2, 3.2)

0

11

A1186G

Q356R

1.7 (0.0, 3.4)

1.3 (0.0, 3.2)

0

3.0 (1.1, 4.9)

2.4 (0.6, 4.2)

3.1 (0.0, 6.6)

11

T1256G

I379M

0.4 (0.0, 1.3)

0

1.0 (0.0, 3.1)

0

0

0

11

A1546G

H476R

0.4 (0.0, 1.3)

0.7 (0.0, 2.0)

0

0

0

0

11

T1831C

I571T

0.9 (0.0, 2.1)

1.3 (0.0, 3.2)

0

3.0 (1.1, 4.9)

3.1 (1.1, 5.1)

0

11a

A1921G

R601H

0

2.0 (0.0, 4.2)

0

0

0

0

11

G2196A

D693N

2.6 (0.5, 4.6)

0

3.1 (0.0, 6.6)

3.9 (1.7, 6.1)

5.0 (2.5, 7.5)

4.2 (0.0, 8.2)

11

C2470T

S784L

0

0

0

0.7 (0.0, 1.6)

0.3 (0.0, 1.0)

0

11

G3143A

M1008I

0

0

0

0

0.3 (0.0, 1.0)

0

11

G3238A

S1040N

0.9 (0.0, 2.0)

0

2.1 (0.0, 4.9)

1.6 (0.2, 3.1)

1.7 (0.2, 3.1)

4.2 (0.2, 8.2)

11a

G3465C

V1116L

0

0

0

0.3 (0.0, 1.0)

0

0

11

A3537G

S1140G

2.1 (0.3, 4.0)

1.4 (0.0, 3.2)

1.0 (0.0, 3.1)

0.3 (0.0, 1.0)

0.3 (0.0, 1.0)

0

11

G3719C

Q1200H

0.4 (0.0, 1.3)

0

1.0 (0.0, 3.1)

0.3 (0.0, 1.0)

0

0

16

C4801T

T1561I

0.4 (0.0, 1.3)

0.7 (0.0, 1.9)

0

0.3 (0.0, 1.0)

0.3 (0.0, 1.0)

0

16

T4810C

L1564P

0.4 (0.0, 1.3)

0

0

0

0

0

19

A5277G

T1720A

0.4 (0.0, 1.3)

0

0

0.7 (0.0, 1.6)

1.4 (0.0, 2.7)

0

22

T5467C

M1783T

0.4 (0.0, 1.3)

0

0

0

0

0

23a

C5531T

V1804V

0.4 (0.0, 1.3)

0

0

0

1.0 (0.0, 2.1)

0

Silent

3

G233A

L38L

0

0

0

0

1.0 (0.0, 2.1)

0

9

C710T

C197C

0

0

0

0.3 (0.0, 1.0)

0

0

16

A4931G

Q1604Q

0.8 (0.0, 2.0)

0.7 (0.0, 1.9)

0

0

0

0

Deletion

11

1965

S616del

0.9 (0.0, 2.1)

2.0 (0.0, 4.2)

0

0

0

0

Nonsense

13

C4446T

R1443X

0

0

0

0.3 (0.0, 1.0)

0

0

Frameshift

11

943

943Ins10

0.4 (0.0, 1.3)

0

0

0

0

0

aNot in BIC database

bPercent of chromosomes from cancer cases with variant and 95% confidence intervals (CI)

cPercent of chromosomes from sibling controls with variant and 95% CI

dPercent of chromomsomes from MEC controls with variant and 95% CI

Sixteen rare variants were found among Latina women (Table 2). Three were found exclusively in cases (C197C, V1116L, and R1443X). Mutation V1116L in exon 11 was found in one Latina case diagnosed with breast cancer in her late 50s; none of her four healthy sisters with informative sequencing results carried this mutation. Three of the 16 total variants, viz., Q356R, D693N, and S1040N (all in exon 11), were found in Latina population controls.

We identified eight common variants in our African-American and Latina populations (Table 3), none of which appeared to be over-represented in the affected sibs when compared with unaffected sibs or population controls. All had previously been reported to BIC and were not specific to African-American or Latina women. The frequencies of variants (with the exception of C2731T) were significantly less common in our African-American population than previously reported for Caucasian populations (Table 4). In a matched analysis by sibship, we found no association with breast cancer risk and any of these common variants overall or when restricting to the 36 families with a breast cancer diagnosis before age 40, breast and ovarian cancer in the same family, or male breast cancer (data not shown). Further, we found no significant difference in variant effect by reproductive or demographic characteristics (data not shown).
Table 3

BRCA1 common sequence variant frequencies in African-American and Latina breast/ovarian cancer cases (CA), sibling controls (CO), and MEC population controls (CO) with no history of cancer

Exon

Nucleotide

Designation

African-American

Latina

Cancer CAa % (95% CI)

Sibling COb % (95% CI)

Population COc % (95% CI)

Cancer CAa % (95% CI)

Sibling COb % (95% CI)

Population COc % (95% CI)

Missense variants

11

A2577G

K820E

2.1 (0.3, 4.0)

5.3 (1.7, 8.9)

5.2 (0.8, 9.7)

0.3 (0.0, 1.0)

0.7 (0.0, 1.6)

3.1 (0.0, 6.6)

11

C2731T

P871L

67.1 (61.1, 73.1)

64.7 (57.0, 72.3)

79.2 (71.0, 87.3)

39.9 (34.4, 45.4)

37.7 (32.2, 43.1)

39.6 (29.8, 49.4)

11

A3232G

E1038G

15.8 (11.1, 20.5)

18.7 (12.4, 24.9)

12.5 (5.9, 19.1)

33.7 (28.4, 39.0)

31.0 (25.8, 36.2)

33.3 (23.9, 42.8)

11

A3667G

K1183R

22.6 (17.3, 28.0)

26.4 (19.3, 33.4)

27.1 (18.2, 36.0)

35.2 (29.8, 40.6)

33.6 (28.2, 38.9)

36.5 (26.8, 46.1)

16

A4956G

S1613G

22.5 (17.1, 27.8)

26.3 (19.3, 33.3)

29.2 (20.1, 38.3)

34.5 (29.2, 39.8)

32.6 (27.2, 37.9)

37.5 (27.8, 47.2)

Silent variants

11

C2201T

S649S

19.2 (14.2, 24.3)

25.3 (18.4, 32.3)

27.1 (18.2, 36.0)

34.3 (29.0, 39.6)

33.0 (27.7, 38.3)

36.5 (26.8, 46.1)

11

T2430C

L771L

15.8 (11.1, 20.5)

16.0 (10.1, 21.9)

20.8 (12.7, 29.0)

33.7 (28.4, 39.0)

31.7 (26.4, 36.9)

33.3 (23.9, 42.8)

13

T4427C

S1436S

17.8 (12.9, 22.7)

16.4 (10.6, 22.3)

21.9 (13.6, 30.1)

33.2 (28.0, 38.5)

30.9 (25.6, 36.1)

32.3 (22.9, 41.6)

aPercent of chromosomes from cancer cases with variant and 95% confidence intervals (CI)

bPercent of chromosomes from sibling controls with variant and 95% CI

cPercent of chromomsomes from MEC controls with variant and 95% CI

Table 4

Comparison of common BRCA1 variants in MEC African-Americans and Latinas

Variant

Subject type 

Allele frequencies

African-Americana

Latinaa

Durocher et al. (1996)

Dunning et al. (1997)

Friedman et al. (1995)

C2201T

Cases

19.2

34.3

31.5b

33b

Controls

27.1

36.5

24.4c

  

T2430C

Cases

15.8

33.7

31.9b

33b

Controls

20.8

33.3

29.8

  

A2577G

Cases

2.1

0.3

Controls

5.2

3.1

   

C2731T

Cases

67.1

39.9

42.1b

34.2b

Controls

79.2

39.6

27.9b

31.6b

 

A3232G

Cases

15.8

33.7

34.2b

34.2b

33b

Controls

12.5

33.3

23.8b

31.6b

 

A3667G

Cases

22.6

35.2

31.3b

33b

Controls

27.1

36.5

32.1

  

T4427C

Cases

18.0

33.2

30.3b

Controls

21.9

32.3

33.1b

  

A4956G

Cases

22.5

34.5

31.7b

34.2b

33b

Controls

29.2

37.5

31.1

31.6

 

aOwn data

bP-value<0.05 compared with African-Americans

cP-value<0.05 compared with Latinas

Discussion

Our study of 109 African-American and 140 Latina sibships is the largest series of African-American families and the first study, to our knowledge, of BRCA1 in Latina families of central and south American descent with a family history of breast or ovarian cancer. The majority (26/33) of the sequence variants identified in our sample are not unique to African-American or Latina women as they have been described previously in multiple racial/ethnic groups (BIC 2003; Olopade et al. 2003); the six variants described exclusively in African, African-American (I379M, H476R, R601H, M1783T), or Latina women (A280G, S784L, V1804V) are rare. We have found a low frequency of protein-truncating mutations in this population of post-menopausal women, despite a family history of breast and ovarian cancer in these families. The frequencies of common variants in our African-American population are significantly lower than those reported previously for Caucasian populations (Dunning et al. 1997; Durocher et al. 1996a; Friedman et al. 1995), whereas the frequencies of common variants in our Latina population are consistent with those reported in the literature. We have identified two novel variants (V1804V and R601H) in African-American and Latina families, respectively. None of the missense variants appear to be associated with increased risk of cancer in the case-control analysis when matching sibship, overall, or by reproductive factors or degree of family history.

We identified two BRCA1 protein-truncating mutations in our population. The frameshift mutation, 943Ins10, is a founder mutation in African-Americans and has been described in US and west African populations (Arena et al. 1996, 1997, 1998; Mefford et al. 1999; Panguluri et al. 1999). An African-American woman in our population with this mutation was diagnosed with breast cancer in her 40s. The single nonsense mutation in our study population (R1443X) was identified in a Latina ovarian cancer case, but not her healthy sister or population controls. This mutation has previously been described in Canadian and French women diagnosed with ovarian or breast cancer (Risch et al. 2001; Stoppa-Lyonnet et al. 1997) and many times in BIC (2003) among families with histories of both breast and ovarian cancer.

Of the six variants identified exclusively in cases (943Ins10, V1116L, L1564P, M1783T, C197C, and R1443X), two (M1783T and C197C) are conserved in many species, and the missense is located in a region known to interact with other key regulating proteins (Fig. 1). M1783T is located in a region that will complex with AR, pRB, p53, and BRCA2 (Welcsh and King 2001). The silent variant, C197C, could influence cancer risk by altering splicing or message stability or through some other mechanisms. Our study and two previous studies have identified variant L1564P exclusively in cases (Durocher et al. 1996a, 1996b; Panguluri et al. 1999); however, an assay of the variant by using site-directed mutagenesis indicates no loss of function (Hayes et al. 2000).

The single amino acid deletion that we have found (S616del) occurs in a highly conserved region of exon 11, one base outside of a bipartite nuclear localization signal (Bennett et al. 1999) in an area that interacts with several regulating proteins (Fig. 1; Welcsh and King 2001). Thus, its location suggests that it may occur in a critical genomic region. Although we have identified the S616del variant in two African-American families, it occurs in both affected and unaffected siblings (two cases diagnosed after age 60 and three healthy siblings aged 70 and above), suggesting that it is not associated with breast cancer, or that the variant is not highly penetrant.

All of our common missense variants have been described previously and appear to act as benign polymorphisms. One previous study observed a statistically significant positive association between BRCA1 common variants and family history of breast or ovarian cancer (Durocher et al. 1996a) and another reported an inverse association (Smith et al. 2001). Whereas none of the individual missense variants in our dataset appear to be associated with risk, relevant mutations may exist outside the amplicons that we have analyzed, or we may have missed large rearrangements of the BRCA1 gene (Eng et al. 2001; Gad et al. 2001; Unger et al. 2000). Previous studies have suggested that reproductive- and estrogen-modulating factors, including use of oral contraceptives, parity, and age of first birth, modify the risk of breast cancer among BRCA1 and BRCA2 mutation carriers (Jernstrom et al. 1999; King et al. 2003; Narod et al. 2002; Rebbeck et al. 2001; Ursin et al. 1997). Our analysis of common variants did not reveal any reproductive variables that dramatically modified risk. Since we identified so few carriers in our population-based study, we have limited the statistical power to evaluate such interactions; our study does not have the power to detect interaction effects below 2.0.

Despite a family history of breast or ovarian cancer in our families, we found a low frequency of BRCA1 protein truncation mutations in our population (0.72%). This frequency is similar to those reported for population-based (Newman et al. 1998) and hospital-based (Shen et al. 2000) samples of African-American breast cancer cases selected irrespective of family history or age of onset of disease (0% and 1.9%, respectively). The frequency of BRCA1 mutations in populations with strong family histories of breast or ovarian cancer worldwide range from 9% to 79% (Szabo and King 1995). Other studies of African or African-American cases selected based on family history or early age of diagnosis have reported mutation frequencies between 2.9% and 5.0% (Gao et al. 1997, 2000a; Panguluri et al. 1999); however, a high frequency of mutations (56%) was observed in a series identified from a high-risk cancer clinic (Gao et al. 1997, 2000b). This population had a similar family history of disease to ours (mean of 2.7 immediate family members with history of breast or ovarian cancer), but the median age of diagnosis (46 years) was nearly 10 years younger than that of the current study.

A weakness of our study is that families carrying more severe mutations were likely to have been missed if the BRCA1 mutations were associated with decreased survival. The cohort from which this sample was identified was designed to enroll men and women aged 45 years and older, and thus, we had no opportunity to identify early-onset cases of breast or ovarian cancer that were rapidly fatal. The median age of all breast and ovarian cancer cases occurring in the eligible families that we enrolled versus eligible families excluded because all cases died prior to the proband completing the family questionnaire was approximately the same, viz., 55 years (range 28–87 years); however, 35% of cases from the excluded families had missing age of diagnosis. The mean and median age of the proband at entry into the cohort was slightly higher (2 years) for eligible families in which all cases had died before being invited into the study than for those families that were enrolled. In our eligible families, the median age of diagnosis for deceased cases was 52, whereas the median age of diagnosis among living sisters was 56.

The finding that many of the women had a family history of disease without evidence of protein-truncating mutations or an obvious disease-associated missense mutation indicates that mutations in genes other than BRCA1 contribute to the increased risk of cancer in these families. There may also be important splice variants in the intron-exon boundaries that we have not yet analyzed. We are currently investigating the role of BRCA2 in these same families; however, current evidence suggests that only 15%–20% of excess familial risk is attributable to BRCA1 and BRCA2 (Anglian Breast Cancer Study Group 2000). Other genes, including BRCA2, p53, STK11/LKB1 protein kinase, or PTEN (Venkitaraman 2002), may together explain another small fraction of familial risk. Additional research is necessary to identify the complement of genes responsible for cancer risk in African-American and Latina women with a family history of breast or ovarian cancer.

Acknowledgements

The authors thank the families participating in this study. They also gratefully acknowledge the contribution of Stacy Clark, Marina Herrera, Maria Torres, Mike Turin, and Katherine De Lellis to data collection and management, Hank Huang and Kristine R. Monroe for questionnaire and database design, Sheila Mangune for laboratory findings, and Kim Siegmund for consultions regarding statistical analyses. This work was supported by NCI grant nos. CA63464, CA54281 and CA77571.

Copyright information

© Springer-Verlag 2005