Abstract
Purpose: As estrogen receptor–positive (ER+) breast cancer in BRCA1 mutation carriers arises at an older age with less aggressive tumor characteristics than ER-negative (ER−) BRCA1-mutated breast cancer, it has been suggested that these tumors are “sporadic” and not BRCA1 driven. With the introduction of targeted treatments specific for tumors with a nonfunctioning BRCA1 or BRCA2 gene, the question whether the BRCA genes are impaired in the tumor is highly relevant. Therefore, we performed genomic profiling of BRCA1-mutated ER+ tumors.
Experimental Design: Genomic profiling, BRCA1 promoter methylation assessment, and loss of heterozygosity analysis were done on 16 BRCA1-mutated ER+ tumors. Results were compared with 57 BRCA1-mutated ER− tumors, 36 BRCA2-mutated ER+-associated tumors, and 182 sporadic ER+ tumors.
Results: The genomic profile of BRCA1-mutated ER+ tumors was different from BRCA1-mutated ER− breast tumors, but highly similar to BRCA2-mutated ER+ tumors. In 83% of the BRCA1-mutated ER+ tumors, loss of the wild-type BRCA1 allele was observed. In addition, clinicopathologic variables in BRCA1-mutated ER+ cancer were also more similar to BRCA2-mutated ER+ and sporadic ER+ breast cancer than to BRCA1-mutated ER− cancers.
Conclusions: As BRCA1-mutated ER+ tumors show a BRCAness copy number profile and LOH, it is likely that the loss of a functional BRCA1 protein plays a role in tumorigenesis in BRCA1-mutated ER+ tumors. Therefore, we hypothesize that these tumors are sensitive to drugs targeting the BRCA1 gene defect, providing new targeted treatment modalities for advanced BRCA-deficient, ER+ breast cancer. Clin Cancer Res; 23(5); 1236–41. ©2016 AACR.
With the revised interest in platinum salts and the introduction of PARP inhibitors that specifically target tumors with a nonfunctioning BRCA1 or BRCA2 gene, the question whether the BRCA genes are impaired in the tumor is highly relevant. As estrogen receptor–positive (ER+) tumors in BRCA1 mutation carriers generally arise at a later age and have less aggressive tumor characteristics than ER-negative (ER−) BRCA1-mutated tumors, it is questioned whether BRCA1 is completely lost. In the current study, we performed genomic analysis on ER+ BRCA1-mutated tumors. ER+ BRCA1-mutated tumors showed a BRCAness genomic profile and loss of heterozygosity at the BRCA1 locus, indicating complete loss of BRCA1. Therefore, we suggest that these tumors are indeed sensitive to therapies targeting the BRCA gene deficiency.
Introduction
Most breast cancers arising in BRCA1 mutation carriers are triple negative, but 10% to 20% are estrogen receptor positive (ER+; refs. 1–4). These ER+ BRCA1-mutated breast cancers tend to arise in older patients, have less aggressive tumor characteristics, and are sometime believed to be “sporadic” rather than related to BRCA1 deficiency (3, 5, 1). Others, however, showed that loss of the BRCA1 wild-type allele is equally frequent in BRCA1-mutated ER-negative (ER−) breast cancer as BRCA1-mutated ER+ breast cancer, indicating that BRCA1 is involved in tumorigenesis in these tumors as well (6). Still, others proposed that BRCA1-mutated ER+ tumors are intermediate in pathologic variables between BRCA1-mutated ER− tumors and sporadic ER+ breast cancers (7). With the revived interest in platinum salts and the introduction of targeted treatments specific for tumors with a nonfunctioning BRCA1 or BRCA2 gene (i.e., PARP inhibitors; refs. 8–10), the question whether the BRCA genes are impaired in the tumor is highly relevant.
Loss of the remaining wild-type allele in BRCA1- or BRCA2-mutated tumors results in deficient homologous recombination DNA repair. The normal tissue of the patient does not share this DNA repair defect. This tumor-specific defect can be exploited by PARP inhibitors and double-strand break inducing chemotherapy to induce selective tumor cytotoxicity while sparing normal cells. Indeed, several clinical trials have shown that most evidence for an effect of PARP inhibitors and platinum salts is seen in BRCA1 and BRCA2 mutation carriers (11). Should BRCA1-mutated ER+ tumors be more alike sporadic breast cancer and BRCA1 is not impaired in the tumor, these tumors would not be sensitive to PARP inhibitors.
Therefore, the question whether ER+ tumors in patients with a germline BRCA1 mutation are deficient in BRCA function is clinically important. To answer this question, we performed a detailed genomic analysis and compared histopathologic features between BRCA1-mutated ER+ tumors, BRCA1-mutated ER− tumors, BRCA2-mutated ER+, and sporadic ER+ breast cancers.
Materials and Methods
Patients
Four breast cancer groups were selected from the Netherlands Cancer Institute's pathology archive: (i) 16 ER+ tumors from BRCA1 mutation carriers; (ii) 57 ER− tumors from BRCA1 mutation carriers; (iii) 36 ER+ tumors from BRCA2 mutation carriers; and (iv) 182 tumors from women with unknown BRCA1 or BRCA2 status. Patients were treated in Dutch hospitals between 1994 and 2014. The first three groups were patients who tested positive for a BRCA1 or BRCA2 mutation by one of the clinical genetics centers in the Netherlands. The last group had not been tested for BRCA1 or BRCA2 mutations and contains cases treated with neoadjuvant chemotherapy, as well as cases selected from the pathology archive without a breast cancer family history (12–14).
Pathology
ER positivity (antibody from Roche Diagnostics cat. no. 5278406001) and progesterone receptor (PR) positivity (Roche Diagnostics cat. no. 5278392001) were defined by staining at least 1% positive neoplastic cells by IHC. A sample was scored as being HER2 positive when either strong membrane staining (3+) could be observed by IHC (Roche Diagnostics cat. no. 5278368001) or if CISH revealed amplification of HER2 in samples with moderate (2+) membrane staining at IHC. IHC, histology, and grading were revised by an experienced breast cancer pathologist (J. Wesseling).
BRCA1/2 germline testing
BRCA1 mutation status was obtained from patient records, obtained through our family cancer clinic. Briefly, germline DNA was isolated from peripheral blood lymphocytes of affected patients. We used mutation-scanning methods to detect pathogenic mutations. The protein truncation test was used for exon 11 of BRCA1 and exon 10 and 11 of BRCA2. The remaining exons were tested using denaturing gradient gel electrophoresis. Confirmation of aberrant samples was done by Sanger sequencing on an ABI 3730 capillary sequencer (Life Technologies/Applied Biosystems; ref. 15). In addition, multiplex ligation-dependent probe amplification (MLPA) was performed using MLPA Kit P087 (BRCA1; MRC-Holland) to detect large genomic deletions or duplications in the genes.
aCGH BRCA-like genomic profile
Array comparative genomic hybridization (aCGH) BRCA-like scores were obtained by two different assays. The older samples (2004–2010) were obtained by a 3.5 k BAC aCGH platform. The DNA segments covered the whole genome with an average spacing of 1 MB. The protocols have been described before (16). The more recent samples were analyzed using NimbleGen 128K oligo arrays. For the NimbleGen arrays labeling, hybridization, imaging, and data analysis have been performed according to the manufacturer's protocol (NimbleGen, Roche). Classification of subtypes was performed using the aCGH BRCA1-like classifier and the BRCA2-like classifier, both developed by Joosse and colleagues. (13, 12). The cutoff to call a tumor BRCA1-like or BRCA2 like is set at 0.5, according to the original publication (13, 12). The BRCA1-like and BRCA2-like classifier tool is available via http://ccb.nki.nl/software/nkiBRCA/. The translation and concordance between the BAC aCGH BRCA1-like and BRCA2-like classifiers and the NimbleGen classifier have been described previously (17). Array data are accessible through GEO (GSE77605 for the 16 BRCA1-mutated ER+ tumors, GSE9021, GSE9114, GSE9114, and GSE16511 for the control samples).
BRCA1 promoter methylation
Hypermethylation of the BRCA1 promoter was determined using methylation-specific MLPA analysis, according to the manufacturers' protocol (ME-001, MRC-Holland). For normalization and analysis, we used an in-house–developed Microsoft Excel macro. We used a cutoff of 20% to call a gene promoter methylated.
Loss of heterozygosity analysis
Loss of heterozygosity (LOH) at the BRCA1 locus for the BRCA1-mutated ER+ tumors was determined by Sanger sequencing of tumor DNA using flanking primers. The incorporation of wild-type versus mutant nucleotide was determined on an ABI sequencer 3730 and used to determine whether there was retention of heterozygosity or LOH and whether the wild-type or mutant allele was underrepresented (lost). For every tumor sample, matched normal DNA, obtained from peripheral blood, was available as a control.
Statistical analysis
The Fisher exact test was used to assess the association between the dichotomized histologic and clinical variables and BRCAness characteristics. P < 0.05 was considered statistically significant. IBM SPSS Statistics version 22 was used. Comparison of copy number aberrations between tumor groups was performed in Nexus Copy Number 7.5 (BioDiscovery). To call regions significantly different from each other, the minimum difference in percentage was set at 25% and the P value (calculated by Fisher exact test) was below 0.05. As there were multiple tests performed, Q-bound values were calculated, which correct for multiple testing by performing FDR correction.
Results
Description of the cohort
There were 16 BRCA1-mutated ER+ tumors in our database. Average age of patients involved was 45 years. Most tumors were invasive ductal carcinomas, and HER2 negative (Tables 1 and 2). When we applied our previously developed BRCA1-like classifier, developed to identify BRCA1-associated breast cancer (13), only two BRCA1-mutated ER+ tumors showed the characteristic BRCA1-like genomic profile. From previous series, we know that 80% to 90% of all BRCA1-mutated ER− tumors show this BRCA1-like specific genomic profile (18, 19). Interestingly, 10 of 16 BRCA1-mutated ER+ tumors (62.5%) showed a BRCA2-like genomic profile, developed to specifically identify BRCA2-mutated breast tumors (12). BRCA1 promoter methylation was absent in all BRCA1-mutated ER+ tumors. From 12 tumors, the amount of DNA was sufficient for LOH analysis; 10 of 12 ER+BRCA1-mutated tumors showed loss of the wild-type BRCA1 allele, which is significant (binomial test P = 0.04). One sample showed loss of the mutant allele, whereas another sample did not show loss at the BRCA1 locus (Table 1).
Patient and tumor characteristics
Patient characteristics . | . | . | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | . | Germline BRCA1 mutation . | Histopathology and IHC . | Molecular analyses . | ||||||||||||
ID . | Age . | Exon . | Mutation (c.) . | Tumor type . | Grade . | ER% . | PR% . | HER2 . | p53% . | KI67% . | BRCA1-like . | BRCA1 score . | BRCA2-like . | BRCA2 score . | Methylation . | LOH analysis . |
1 | 53 | 11 | c.2210delC | ILC | 1 | 100 | 0 | Neg | 0 | ND | SP-like | 0 | SP-like | 0.28 | No meth | Loss wt allele |
2 | 55 | 11 | c.2210delC | IDC | 3 | 100 | 0 | Neg | 0 | ND | SP-like | 0 | B2-like | 0.94 | No meth | Loss wt allele |
3 | 40 | 13 | c.4186-1632_4357+2031del3835 | IDC | 2 | 100 | 100 | Neg | 100 | ND | SP-like | 0.019 | SP-like | 0.002 | No meth | Loss wt allele |
4 | 39 | 20 | c.5266dupC | Mixed | 3 | 90 | 70 | Neg | <10 | ND | SP-like | 0.002 | B2-like | 0.69 | No meth | Loss wt allele |
5 | 38 | 22 | c.5333-36_5406+400del510 | IDC | ND | 100 | 30 | Neg | 20 | ND | Sp-like | 0.16 | B2-like | 1 | No meth | ND |
6 | 72 | 14 | c.4416_4417delTTinsG | IDC | 1 | 100 | ND | ND | ND | ND | SP-like | 0 | SP-like | 0 | No meth | ROH |
7 | 41 | 11 | c.3820dupG | ILC | ND | 75 | 10 | Neg | 95 | 5 | Sp-like | 0.45 | Sp-like | 0.16 | ND | Loss wt allele |
8 | 66 | 11 | c.2989_2990dupAA | IDC | ND | 100 | 60 | Neg | 1 | 15 | B1-like | 0.76 | B2-like | 0.59 | No meth | Loss wt allele |
9 | 47 | 22 | c.5333-36_5406+400del510 | IDC | ND | 90 | <5 | Neg | 90 | 80 | Sp-like | 0.35 | Sp-like | 0.001 | No meth | Loss wt allele |
10 | 42 | 22 | c.5333-36_5406+400del510 | IDC | ND | 5 | 0 | Neg | 80 | 90 | Sp-like | 0.41 | Sp-like | 0 | No meth | Loss wt allele |
11 | 62 | 20 | c.5266dupC | IDC | 3 | 20 | 0 | Neg | 1 | 25 | B1-like | 1 | B2-like | 1 | No meth | Loss mut allele |
12 | 58 | 17 | c.4987-?_5074+?del | IDC | 3 | 100 | 5 | Pos | ND | 30 | Sp-like | 0.01 | B2-like | 0.997 | ND | ND |
13 | 68 | 5 | c.181T>G | IDC | 3 | 100 | 1 | Neg | ND | 40 | Sp-like | 0.01 | B2-like | 0.998 | ND | Loss wt allele |
14 | 39 | 22 | c.5333-36_5406+400del510 | IDC | ND | 90 | 80 | Neg | 0 | 10 | SP-like | 0.003 | B2-like | 0.67 | No meth | ND |
15 | 43 | 5 | c.181T>G | IDC | 2 | 100 | 0 | Neg | ND | ND | SP-like | 0 | B2-like | 1 | ND | ND |
16 | 43 | 22 | c.5333-36_5406+400del510 | IDC | 2 | 100 | 100 | Neg | 100 | ND | SP-like | 0.04 | B2-like | 0.68 | ND | Loss wt allele |
Patient characteristics . | . | . | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | . | Germline BRCA1 mutation . | Histopathology and IHC . | Molecular analyses . | ||||||||||||
ID . | Age . | Exon . | Mutation (c.) . | Tumor type . | Grade . | ER% . | PR% . | HER2 . | p53% . | KI67% . | BRCA1-like . | BRCA1 score . | BRCA2-like . | BRCA2 score . | Methylation . | LOH analysis . |
1 | 53 | 11 | c.2210delC | ILC | 1 | 100 | 0 | Neg | 0 | ND | SP-like | 0 | SP-like | 0.28 | No meth | Loss wt allele |
2 | 55 | 11 | c.2210delC | IDC | 3 | 100 | 0 | Neg | 0 | ND | SP-like | 0 | B2-like | 0.94 | No meth | Loss wt allele |
3 | 40 | 13 | c.4186-1632_4357+2031del3835 | IDC | 2 | 100 | 100 | Neg | 100 | ND | SP-like | 0.019 | SP-like | 0.002 | No meth | Loss wt allele |
4 | 39 | 20 | c.5266dupC | Mixed | 3 | 90 | 70 | Neg | <10 | ND | SP-like | 0.002 | B2-like | 0.69 | No meth | Loss wt allele |
5 | 38 | 22 | c.5333-36_5406+400del510 | IDC | ND | 100 | 30 | Neg | 20 | ND | Sp-like | 0.16 | B2-like | 1 | No meth | ND |
6 | 72 | 14 | c.4416_4417delTTinsG | IDC | 1 | 100 | ND | ND | ND | ND | SP-like | 0 | SP-like | 0 | No meth | ROH |
7 | 41 | 11 | c.3820dupG | ILC | ND | 75 | 10 | Neg | 95 | 5 | Sp-like | 0.45 | Sp-like | 0.16 | ND | Loss wt allele |
8 | 66 | 11 | c.2989_2990dupAA | IDC | ND | 100 | 60 | Neg | 1 | 15 | B1-like | 0.76 | B2-like | 0.59 | No meth | Loss wt allele |
9 | 47 | 22 | c.5333-36_5406+400del510 | IDC | ND | 90 | <5 | Neg | 90 | 80 | Sp-like | 0.35 | Sp-like | 0.001 | No meth | Loss wt allele |
10 | 42 | 22 | c.5333-36_5406+400del510 | IDC | ND | 5 | 0 | Neg | 80 | 90 | Sp-like | 0.41 | Sp-like | 0 | No meth | Loss wt allele |
11 | 62 | 20 | c.5266dupC | IDC | 3 | 20 | 0 | Neg | 1 | 25 | B1-like | 1 | B2-like | 1 | No meth | Loss mut allele |
12 | 58 | 17 | c.4987-?_5074+?del | IDC | 3 | 100 | 5 | Pos | ND | 30 | Sp-like | 0.01 | B2-like | 0.997 | ND | ND |
13 | 68 | 5 | c.181T>G | IDC | 3 | 100 | 1 | Neg | ND | 40 | Sp-like | 0.01 | B2-like | 0.998 | ND | Loss wt allele |
14 | 39 | 22 | c.5333-36_5406+400del510 | IDC | ND | 90 | 80 | Neg | 0 | 10 | SP-like | 0.003 | B2-like | 0.67 | No meth | ND |
15 | 43 | 5 | c.181T>G | IDC | 2 | 100 | 0 | Neg | ND | ND | SP-like | 0 | B2-like | 1 | ND | ND |
16 | 43 | 22 | c.5333-36_5406+400del510 | IDC | 2 | 100 | 100 | Neg | 100 | ND | SP-like | 0.04 | B2-like | 0.68 | ND | Loss wt allele |
Abbreviations: IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; meth, methylation; mut, mutant; ND, not determined; ROH, retention of heterozygosity; wt, wild type.
Comparison of clinicopathologic characteristics between B1mutER+, B1mutER−, B2mutER+, and sporadic ER+ tumors
. | . | B1mutER+ (n = 16) . | B1mutER− (n = 57) . | . | B2mutER+ (n = 36) . | . | SporER+ (n = 182) . | . |
---|---|---|---|---|---|---|---|---|
. | . | n (%) . | n (%) . | Pa . | n (%) . | Pa . | n (%) . | Pa . |
BRCA1-like | BRCA1-like | 2 (12.5%) | 49 (86%) 2 | 2 (5.6%) | 4 (2.6%) | |||
Sporadic-like | 14 (87.5%) | 8 (14%) | <0.001 | 34 (94.4%) | 0.39 | 147 (97.4%) | 0.04 | |
Not tested | 0 | 0 | 0 | 31 | ||||
BRCA2-like | BRCA2-like | 10 (62.5%) | 31 (60.8%) | 30 (83.3%) | 49 (32.5%) | |||
Sporadic-like | 6 (37.5%) | 20 (39.2%) | 0.90 | 6 (16.7%) | 0.10 | 102 (67.5%) | 0.02 | |
Not tested | 0 | 6 | 0 | 31 | ||||
BRCA1 Meth | Methylated | 1 (2%) | ||||||
Nonmethylated | 10 (100%) | 44 (100%) | ND | 0 90%) | ND | 49 (98%) | 0.65 | |
Not done | 6 | 13 | 36 | 132 | ||||
PR IHC | Negative | 8 (57.1%) | 48 (100%) | 21 (60%) | 51 (30.9%) | |||
Positive | 6 (42.9%) | 0 (0%) | <0.001 | 14 (40%) | 0.85 | 114 (69.1%) | 0.05 | |
Unknown | 2 | 9 | 1 | 17 | ||||
HER2 IHC | Negative | 14 (93.3%) | 46 (100%) | 31 (93.9%) | 131 (80.4%) | |||
Positive | 1 (6.7%) | 0 (0%) | 0.08 | 2 (6.1%) | 0.94 | 32 (19.6%) | 0.22 | |
Unknown | 1 | 11 | 3 | 19 | ||||
Histology | IDC | 10 (83.3%) | 24 (100%) | 35 (100%) | 139 (87.4%) | |||
ILC | 2 (16.7%) | 0 (0%) | 0.04 | 0 (0%) | 0.01 | 20 (12.6%) | 0.68 | |
Unknown | 4 | 33 | 1 | 23 | ||||
Grade | Grade 1/2 | 4 (50%) | 3 (13.6%) | 22 (62.9%) | 71 (65.7%) | |||
Grade 3 | 4 (50%) | 19 (86.4%) | 0.04 | 13 (37.1%) | 0.50 | 37 (34.3%) | 0.37 | |
Unknown | 8 | 35 | 1 | 74 | ||||
Age (median ± SD) | 45 ± 11.7 | 43 ± 9.4 | 0.04 | 41 ± 11.7 | 0.04 | 48 ± 8.6 | 0.65 |
. | . | B1mutER+ (n = 16) . | B1mutER− (n = 57) . | . | B2mutER+ (n = 36) . | . | SporER+ (n = 182) . | . |
---|---|---|---|---|---|---|---|---|
. | . | n (%) . | n (%) . | Pa . | n (%) . | Pa . | n (%) . | Pa . |
BRCA1-like | BRCA1-like | 2 (12.5%) | 49 (86%) 2 | 2 (5.6%) | 4 (2.6%) | |||
Sporadic-like | 14 (87.5%) | 8 (14%) | <0.001 | 34 (94.4%) | 0.39 | 147 (97.4%) | 0.04 | |
Not tested | 0 | 0 | 0 | 31 | ||||
BRCA2-like | BRCA2-like | 10 (62.5%) | 31 (60.8%) | 30 (83.3%) | 49 (32.5%) | |||
Sporadic-like | 6 (37.5%) | 20 (39.2%) | 0.90 | 6 (16.7%) | 0.10 | 102 (67.5%) | 0.02 | |
Not tested | 0 | 6 | 0 | 31 | ||||
BRCA1 Meth | Methylated | 1 (2%) | ||||||
Nonmethylated | 10 (100%) | 44 (100%) | ND | 0 90%) | ND | 49 (98%) | 0.65 | |
Not done | 6 | 13 | 36 | 132 | ||||
PR IHC | Negative | 8 (57.1%) | 48 (100%) | 21 (60%) | 51 (30.9%) | |||
Positive | 6 (42.9%) | 0 (0%) | <0.001 | 14 (40%) | 0.85 | 114 (69.1%) | 0.05 | |
Unknown | 2 | 9 | 1 | 17 | ||||
HER2 IHC | Negative | 14 (93.3%) | 46 (100%) | 31 (93.9%) | 131 (80.4%) | |||
Positive | 1 (6.7%) | 0 (0%) | 0.08 | 2 (6.1%) | 0.94 | 32 (19.6%) | 0.22 | |
Unknown | 1 | 11 | 3 | 19 | ||||
Histology | IDC | 10 (83.3%) | 24 (100%) | 35 (100%) | 139 (87.4%) | |||
ILC | 2 (16.7%) | 0 (0%) | 0.04 | 0 (0%) | 0.01 | 20 (12.6%) | 0.68 | |
Unknown | 4 | 33 | 1 | 23 | ||||
Grade | Grade 1/2 | 4 (50%) | 3 (13.6%) | 22 (62.9%) | 71 (65.7%) | |||
Grade 3 | 4 (50%) | 19 (86.4%) | 0.04 | 13 (37.1%) | 0.50 | 37 (34.3%) | 0.37 | |
Unknown | 8 | 35 | 1 | 74 | ||||
Age (median ± SD) | 45 ± 11.7 | 43 ± 9.4 | 0.04 | 41 ± 11.7 | 0.04 | 48 ± 8.6 | 0.65 |
NOTE: Patients with unknown characteristics were not included in P value calculations; therefore, percentages for these groups are not depicted in the table.
Abbreviations: B1mutER+, BRCA1-mutated ER+; B1mutER−, BRCA1-mutated ER−; B2mutER+, BRCA2-mutated ER+; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; ND, not determined; SporER+, sporadic ER+.
aP value for comparison of BRCA1-mutated ER+ tumors with BRCA1-mutated ER− tumors, BRCA2-mutated ER+ tumors, and sporadic ER+ tumors, respectively.
Comparison of copy number profiles
As we observed that BRCA1-mutated ER+ tumors more often showed a BRCA2-like genomic profile than a BRCA1-like profile, we looked at the mean genomic profile of the 16 tumors, with special focus on the BRCA1-like and BRCA2-like specific genomic regions. For comparison purposes, we also show the aggregate copy number profiles of 57 BRCA1-mutated ER− tumors, 36 BRCA2-mutated ER+ tumors, and 182 sporadic ER+ tumors (Fig. 1). Figure 1A highlights the characteristic BRCA1-like genomic features, such as 3q gain, 5q loss, and others. Those regions are used in our BRCA1-like classifier. They are clearly present in the mean BRCA1-mutated ER− profile, but absent in the BRCA1-mutated ER+ profile as well as in the BRCA2-mutated tumors or in the sporadic ER+ tumors.
This figure compares the genomic profiles of the BRCA1-mutated ER+ (B1mutER+), BRCA1-mutated ER− (B1mutER−), BRCA2-mutated ER+ (B2mutER+), and sporadic ER+ (SporER+) tumor samples, respectively. Top, regions from the BRCA1-like signature (black squares); bottom, regions from the BRCA2-like signature. Top, we can see that BRCA1-mutated ER+ does not show the characteristic BRCA1-like features, as seen in the BRCA1-mutated ER− tumors. Bottom, we can see that several BRCA2-like features are clearly observed in the BRCA1-mutated ER+, like 3p loss, 6q loss, 11q loss, 16q loss, and 17q gain.
This figure compares the genomic profiles of the BRCA1-mutated ER+ (B1mutER+), BRCA1-mutated ER− (B1mutER−), BRCA2-mutated ER+ (B2mutER+), and sporadic ER+ (SporER+) tumor samples, respectively. Top, regions from the BRCA1-like signature (black squares); bottom, regions from the BRCA2-like signature. Top, we can see that BRCA1-mutated ER+ does not show the characteristic BRCA1-like features, as seen in the BRCA1-mutated ER− tumors. Bottom, we can see that several BRCA2-like features are clearly observed in the BRCA1-mutated ER+, like 3p loss, 6q loss, 11q loss, 16q loss, and 17q gain.
Figure 1B highlights specific BRCA2-associated genomic regions, like 3p loss, 6q loss, and other regions. They are clearly present in the BRCA2-mutated ER+ tumors. Interestingly, most genomic regions specific for BRCA2-mutated tumors are also observed in the BRCA1-mutated ER+ tumors and, to a somewhat lesser extent, in BRCA1-mutated ER− tumors. We have to make clear that in the original classifier, the combination of multiple loci could discriminate BRCA2 mutated from sporadic tumors; however (12), in later research, it appears that the profile was also observed in other genomically instable tumors, such as BRCA1-mutated tumors. However the percentage of these changes is lower in non-BRCA2–mutated cancers. In the sporadic tumors, most specific BRCA2-like copy number changes are absent or seen at a low frequency, for example, 3p loss, 11q loss, 13q loss, etc. From these observations, we can conclude that the copy number profiles of BRCA1-mutated ER+ tumors are different from the copy number profiles of BRCA1-mutated ER− tumors, but that they are more alike BRCA2-mutated tumors. From the comparison with sporadic ER+ tumors, we can conclude that the genomic profile of BRCA1-mutated ER+ tumors is like the specific BRCA2-related profile, and not a standard sporadic luminal copy number profile (Supplementary Fig. S1; Supplementary Tables S1 and S2).
Comparison of clinicopathologic and molecular markers
We compared genomic and histopathologic markers of BRCA1-mutated ER+ tumors with BRCA1-mutated ER− tumors, BRCA2-mutated ER+ tumors, and sporadic ER+ breast cancer (Table 2). As expected, a BRCA1-like profile was frequently found in BRCA1-mutated ER− tumors, but not in BRCA1-mutated ER+ tumors, BRCA2-mutated ER+ tumors, or sporadic ER+ tumors (86%, 12.5%, 5.6%, and 2.6%, respectively, had a BRCA1-like profile). A BRCA2-like profile was frequently observed in BRCA2-mutated tumors (83.3%). It was also frequently observed in BRCA1-mutated ER+ tumors (62.5%), as well as in BRCA1-mutated ER− tumors (60.8%), indicating that it is less specific than the BRCA1-like pattern and also depicts tumors with a high amount of genomic instability.
BRCA1 methylation was not observed in BRCA1-mutated tumors, neither in BRCA1-mutated ER+ tumors nor in BRCA1-mutated ER− tumors. Previously, it has been shown that a mutation in BRCA1 and BRCA1 promoter methylation are two mutually exclusive events in ER− breast cancer (18), this is apparently also true for ER+ tumors. BRCA1 promoter methylation was not observed in BRCA2-mutated tumors. Only one of the sporadic ER+ tumors showed BRCA1 promoter methylation. As BRCA1 methylation is frequently observed in non-BRCA–mutated triple-negative tumors, it appears mainly related to this breast cancer subtype (18).
BRCA1-mutated ER+ tumors more often showed positive PR staining, HER2 amplification, invasive lobular histology, and grade 1–2 tumors and occurred at a higher age than BRCA1-mutated ER− tumors. Pathologic characteristics were similar between BRCA1-mutated ER+ tumors and BRCA2-mutated ER+ tumors, except for histology. Age was also higher in BRCA1-mutated ER+ tumors compared with BRCA2-mutated ER+ (45 years vs. 41 years, respectively; P = 0.04). Sporadic breast cancer cases more often had positive immune staining for PR than BRCA1-mutated ER+ tumors.
Discussion
BRCA1-mutated ER+ tumors have been suggested to be “sporadic” tumors and not related to BRCA1 deficiency. In the current study, we showed that most BRCA1-mutated ER+ tumors have a BRCAness copy number profile and LOH, indicating that the loss of a functional BRCA1 protein plays a role in their tumorigenesis. Interestingly, the genomic profile of these tumors is highly similar to BRCA2-mutated ER+ tumors, but not to BRCA1-mutated ER− tumors. Therefore, we hypothesize that these tumors are sensitive to drugs targeting the BRCA1 gene defect, such as PARP inhibitors.
In a study in 2007, Melchor and colleagues concluded that ER status is an important marker of changes in the tumor genome and independent of the underlying mutation status (20). Interestingly, both the BRCA1-mutated ER+ tumors and BRCA2-mutated ER+ tumors in our series showed a similar specific copy number profile, quite distinct from the ER+ sporadic breast cancer profile. So, although the specific type of the mutation does not determine the specific profile, tumors with a hereditary component have a clearly different genomic appearance than sporadic tumors, representing BRCAness. This is also demonstrated by the work from Nik-Zainal and colleagues, showing that BRCA1- and BRCA2-associated cancers have clearly distinct mutational and rearrangement signatures (21, 22).
Strong proof for the involvement of BRCA1 loss in the origin of these tumors was obtained by LOH analysis. An assumption is that LOH of the wild-type allele is the second hit in BRCA1-mutated breast cancer and responsible for complete silencing of the BRCA1 gene. In the vast majority of BRCA1-mutated ER+ tumors in this series, LOH analysis indeed showed loss of the wild-type allele. In line with published literature, only two of 12 tumors did not show LOH of the wild-type allele. Tung and colleagues (7) reported a percentage of 81% LOH with loss of the wild-type allele in BRCA1-mutated ER+ tumors, comparable with our series and comparable with BRCA1-mutated ER− tumors. BRCA1 promoter methylation is another mechanism for silencing of the BRCA1 gene. However, BRCA1 promoter methylation was not observed in our series of BRCA1-mutated ER+ tumors. This observation is similar to the lack of BRCA1 promoter methylation in BRCA1-mutated ER− tumors (18).
Histopathologic features and age at diagnosis were different between BRCA1-mutated ER+ and ER− tumors. BRCA1-mutated ER+ tumors were more similar to BRCA2-mutated ER+ tumors and to sporadic ER+ tumors. Tung and colleagues compared histopathologic features between BRCA1-mutated ER+ and ER− tumors and concluded that ER+ cancers appear pathologically intermediate between BRCA1-mutated ER− tumors and sporadic ER+ breast cancers (7). This would suggest that either some BRCA1-mutated ER+ tumors are incidental or that there is a unique mechanism by which these cancers develop. In our series, we observed that patients with BRCA1-mutated ER+ tumors had a higher age at diagnosis, even compared with patients with BRCA2-mutated ER+ tumors. One of the patients without loss of the wild-type allele was 72 years old and did not show a BRCAness copy number profile, suggesting that her tumor was indeed a “sporadic” one.
PARP inhibitors are now studied in multiple clinical trials (11). Most support for their use is for BRCA1- and BRCA2-mutated tumors. Therefore, it is essential to know whether PARP inhibitors are effective irrespective of the hormonal status of a tumor. The loss of the wild-type BRCA1 allele as well as a BRCAness copy number profile was shown in a vast majority of the cases, indicating complete homologous recombination deficiency. Therefore, these data suggest that BRCA1-mutated ER+ tumors are sensitive to drugs targeting this DNA repair defect, like PARP inhibitors. In the ongoing clinical trials on PARP inhibitors in BRCA1 mutation carriers, we suggest to include ER+ BRCA1-mutated tumors and to specifically monitor their treatment response. LOH analysis and genomic profiling could be additional biomarkers to select patients for treatment with drugs targeting homologous recombination deficiency. In addition to PARP inhibitors, BRCA1-mutated ER+ tumors would also be good candidates for treatment with other DNA-damaging drugs, such as platinum salts or alkylating agents.
A limitation of the current study is that the number of ER+BRCA1-mutated tumors is small and that treatment data are unavailable. Therefore, we can only speculate how these tumors would respond. Another limitation is that the samples are obtained from our pathology archive and are not a well-described cohort, but rather a mixture of patient groups.
In conclusion, our genetic and genomic analysis confirms a role for BRCA1 in tumorigenesis BRCA1-mutated ER+ tumors and suggests homologous recombination deficiency in these tumors. Our findings are clinically highly relevant, as targeting the DNA damage repair defect, for example, by PARP inhibitors, is an effective and promising strategy to treat homologous recombination–deficient tumors successfully.
Disclosure of Potential Conflicts of Interest
G.S. Sonke reports receiving teaching activity fees from AstraZeneca. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: E.H. Lips, F.B.L. Hogervorst, P.M. Nederlof
Development of methodology: E.H. Lips, G.S. Sonke, F.B.L. Hogervorst, P.M. Nederlof
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): E.H. Lips, R.D. Debipersad, L. Mulder, L.E. van der Kolk, J. Wesseling, F.B.L. Hogervorst, P.M. Nederlof
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E.H. Lips, R.D. Debipersad, L. Mulder, G.S. Sonke, F.B.L. Hogervorst, P.M. Nederlof
Writing, review, and/or revision of the manuscript: E.H. Lips, R.D. Debipersad, L. Mulder, G.S. Sonke, L.E. van der Kolk, J. Wesseling, F.B.L. Hogervorst, P.M. Nederlof
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): E.H. Lips, R.D. Debipersad, C.E. Scheerman, L. Mulder, F.B.L. Hogervorst
Study supervision: E.H. Lips, J. Wesseling, F.B.L. Hogervorst, P.M. Nederlof
Acknowledgments
We would like to acknowledge the NKI-AVL Core Facility Molecular Pathology & Biobanking (CFMPB) for supplying NKI-AVL Biobank material and laboratory support.
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.