Abstract

Background: Tumors with homologous recombination deficiency (HRD), such as BRCA1-associated breast cancers, are not able to reliably repair DNA double-strand breaks (DSBs) and are therefore highly sensitive to both DSB-inducing chemotherapy and poly (ADP-ribose) polymerase inhibitors. We have studied markers that may indicate the presence of HRD in HER2-negative breast cancers and related them to neoadjuvant chemotherapy response.

Patients and methods: Array comparative genomic hybridization (aCGH), BRCA1 promoter methylation, BRCA1 messenger RNA (mRNA) expression and EMSY amplification were assessed in 163 HER2-negative pretreatment biopsies from patients scheduled for neoadjuvant chemotherapy.

Results: Features of BRCA1 dysfunction were frequent in triple-negative (TN) tumors: a BRCA1-like aCGH pattern, promoter methylation and reduced mRNA expression were observed in, respectively, 57%, 25% and 36% of the TN tumors. In ER+ tumors, a BRCA2-like aCGH pattern and the amplification of the BRCA2 inhibiting gene EMSY were frequently observed (43% and 13%, respectively) and this BRCA2-like profile was associated with a better response to neoadjuvant chemotherapy.

Conclusions: Abnormalities associated with BRCA1 inactivation are present in about half of the TN breast cancers but were not predictive of chemotherapy response. In ER+/HER2− tumors, a BRCA2-like aCGH pattern was predictive of chemotherapy response. These findings should be confirmed in independent series.

introduction

Neoadjuvant chemotherapy has become a widely used treatment strategy for patients with early or locally advanced breast cancer. It is equally effective as similar drug therapy following local treatment and it has additional advantages: breast-conserving therapy is more frequently possible as a result of tumor shrinkage and the effect of the drugs on the tumor can be assessed during treatment. The complete disappearance of all tumor cells at microscopic examination [pathologic complete remission (pCR)] correlates well with overall survival [1, 2] and achieving a pCR is considered an appropriate intermediate end point for clinical trials. Current neoadjuvant drug regimens achieve a pCR rate of 5%–10% in luminal type breast cancers and ∼40% in basal-like and in HER2/neu-positive tumors [3, 4].

Bifunctional alkylators and platinating agents cause interstrand DNA cross-linking, which cause DNA double-strand breaks (DSBs) during DNA replication. In normal cells, these DSBs are repaired by a process called homologous recombination. If this process is unavailable or impaired, a situation referred to as ‘homologous recombination deficiency’ (HRD) is present and alternative error-prone DNA repair mechanisms take over, leading to genomic instability. The breast cancer genes BRCA1 and BRCA2 are essential for homologous recombination and tumors of patients carrying germline mutations in these genes show HRD as a result of the loss of the second unmutated allele. BRCA1 and BRCA2 can be inactivated in sporadic cancers as well [5, 6], a phenomenon referred to as ‘BRCA-ness’. Many additional genes are involved in homologous recombination, including the Fanconi anemia genes and the BRCA2 inactivating gene EMSY [7].

Tumors with HRD have been shown to be particularly sensitive to DNA cross-linking agents, such as alkylators and platinum drugs [8–10]. Both classes of drugs are employed in locally advanced breast cancer. Importantly, the novel poly (ADP-ribose) polymerase (PARP) inhibitors are specifically effective in HRD tumors as well and have shown impressive activity in clinical studies recently [11–13]. Unfortunately, no clinical tests exist which can reliably determine HRD in tumor biopsies. Previous studies have focused on genes that have a role in homologous recombination, such as the BRCA1 and -2 genes, FANC genes and EMSY [6]. We and others have previously shown that breast cancers of BRCA1 and BRCA2 mutation carriers have a characteristic pattern of DNA gains and losses in an array comparative genomic hybridization (aCGH) assay [5, 14–18]. In a recent study from our institute, a subgroup of hormone receptor-negative tumors characterized by BRCA1-like aCGH pattern were shown to benefit markedly from intensive platinum-based chemotherapy [Vollebergh MA, Lips EH, Nederlof PM et al. An aCGH classifier derived from BRCA1-mutated breast cancer and benefit of high-dose, platinum-based, chemotherapy in breast cancer patients. Submitted for publication 2010]. Another recent report showed that a subset of triple-negative (TN) tumors might be sensitive to the DNA DSB-inducing drug cisplatin, as a result of low BRCA1 expression levels or BRCA1 promoter methylation [19].

We prospectively determined the frequency in which these HRD-associated features occur in untreated patients with breast cancer. We correlated the findings with response to chemotherapy that causes DNA DSBs. If HRD is indeed confirmed to be the ‘Achilles heel’ of certain sporadic tumors, such tests could eventually serve to individualize drug treatment.

patients and methods

patients

Pretreatment biopsies of primary breast tumors from 163 women with HER2-negative breast cancer were collected. All patients had received neoadjuvant treatment at The Netherlands Cancer Institute from 2004 to 2009 as part of two ongoing clinical trials or were treated off protocol according to the standard arm of one of these studies. Both studies had been approved by the ethical committee and informed consent was obtained from all patients. For eligibility, breast carcinoma with either a primary tumor size of at least 3 cm was required or the presence of fine needle aspiration-proven axillary lymph node metastases. Biopsies were taken using a 14G core needle under ultrasound guidance. After collection, specimens were snap frozen in liquid nitrogen and stored at −70°C. Each patient had two or three biopsies taken to assure that enough tumor material was available for both diagnosis and further study.

Depending on the particular study, a treatment regimen was assigned to each patient, which consisted of one of the following: (i) six courses of dose-dense doxorubicin/cyclophosphamide (ddAC) or (ii) six courses of capecitabine/docetaxel (CD) or (iii) if the therapy response was considered unfavorable by magnetic resonance imaging (MRI) evaluation after three courses, ddAC was changed to CD or vice versa. For the current study, we only considered patients who started with ddAC (group 1 and group 3); thus, all patients received at least three courses of ddAC (a DSB-inducing regimen).

pathology and response evaluation

All pretreatment biopsies were reviewed by two pathologists (MJvdV and JW). Estrogen receptor (ER) and progesterone (PR) percentages were determined by immunohistochemistry (IHC), and HER2 was assessed by IHC and chromosome in situ hybridization. For some analysis, ER and PR were dichotomized as percentage <50% or ≥50% (variable names: ER_50, PR_50). Pretreatment lymph node status was assessed at pathology. The response of the primary tumor to chemotherapy was evaluated by contrast-enhanced MRI [20] after three courses of chemotherapy and after completion of chemotherapy by pathologic evaluation of the resection specimen. The primary end point of both studies was a pCR, defined as the complete absence of residual invasive tumor cells seen at microscopy. If only noninvasive tumor (carcinoma in situ) was detected, this was considered a pCR as well. When a small number of scattered tumor cells were seen, the samples were classified as ‘near pathologic complete remission’ (npCR). Because the aim of this study was to determine if HRD was correlated with a higher sensitivity to chemotherapy, tumors with an npCR were included in the group of complete remission for analytical purposes. Patients with larger amounts of residual tumor left were classified as partial and nonresponders (PR+NR).

array comparative genomic hybridization

Tumor DNA and reference DNA were cohybridized using two different CyDyes to a microarray containing 3.5k BAC/PAC-derived DNA segments covering the whole genome with an average spacing of 1 MB and processed as described before [21]. Classification of subtypes was carried out using the aCGH BRCA1 and BRCA2 classifier developed by Joosse et al. [5, 22]. We used the same classifier as described by Vollebergh et al. [Vollebergh MA, Lips EH, Nederlof PM et al. An aCGH classifier derived from BRCA1-mutated breast cancer and benefit of high-dose, platinum-based, chemotherapy in breast cancer patients. Submitted for publication 2010] and we considered a BRCA1 probability score ≥0.63 as a BRCA1-like aCGH pattern. Under this cut-off, a tumor was called sporadic like. The cut-off for a BRCA2-like aCGH pattern was 0.5, as described in the original publication [22].

RT-PCR

Messenger RNA (mRNA) isolation and extraction were carried out using RNA Bee, according to the manufacturer's protocol (Isotex, Friendswood, TX). A 5-μm section halfway through the biopsy was stained for hematoxylin & eosin and analyzed by a pathologist for tumor cell percentage. Only samples that contained at least 60% tumor cells were included in the further analysis. Quantitative real time polymerase chain reaction was carried out using TaqMan Pre-Designed Gene Expression Assay for BRCA1 (#Hs01556193). The standard curve method was used. GAPDH and β-actin were measured for normalization purposes and the average of both gene expression values was used. The cut-off between BRCA1 low and normal gene expression was 0.25. This cut-off was empirically determined (see results section).

multiplex ligation-dependent probe amplification

Hypermethylation of the BRCA1 promoter was determined using a custom Methylation-specific multiplex ligation-dependent probe amplification (MLPA)-set, according to the manufacturers' protocol (MRC-Holland; ME005-custom). When the two BRCA1 markers both showed methylation, we considered the BRCA1 promoter to be methylated. Amplification of EMSY (C11orf30) was determined using a custom MLPA set, containing seven different EMSY probes and nine reference probes (MRC Holland; X025). This EMSY MLPA set was first validated by an EMSY FISH assay (Dako). From the comparison of the EMSY FISH assay and the MLPA, we concluded that an average of the seven probes >1.5 corresponded to EMSY amplification, as detected by at least six copies of the probe at the FISH assay. DNA fragments were analyzed on a 3730 DNA Analyzer (Applied Biosystems, Foster City, CA). Probe sequences for both MLPA kits are available on request (info@mlpa.com). For normalization and analysis, the Coffalyzer program was used (MRC-Holland).

statistical tests

The Fisher's exact test was used to assess association between the dichotomized HRD characteristics, pathological and clinical variables. Logistic regression was carried out to adjust for the following variables: age, T-stage, N-stage, ER percentage and PR percentage. All data analyses were carried out using SPSS version 17.

results

overview of samples

We studied the frequency of HRD characteristics in pretreatment biopsies and subsequently related the findings to neoadjuvant chemotherapy response. A total of 60 TN and 103 ER+/HER2− tumors were studied, which all received neoadjuvant chemotherapy with doxorubicin and cyclophosphamide (AC-regimen). Table 1 shows the clinical pathological characteristics of all tumors. The majority of the tumors were T-stage 2 or 3 and lymph node positive. Most patients were treated by 6× ddAC, although some switched to the DC regimen after three courses of AC. TN tumors had a higher percentage of responders (pCR + npCR) than ER+ patients. Table 2 gives the frequencies of the HRD characteristics per tumor group. BRCA1-related abnormalities (aCGH BRCA1-like profile, BRCA1 promoter methylation and low BRCA1 mRNA expression) were predominantly observed in the TN tumors, while BRCA2-associated changes (aCGH BRCA2-like profile, EMSY amplification) were mainly observed in the ER+/HER2− tumors (Table 2). The percentage of aberrations was not different between patients treated with six cycli of AC versus patients treated with three cycles AC followed by three cycles of DC (data not shown). As the pattern of characteristics and also the response rates to chemotherapy are different in hormone receptor positive and negative tumors, we analyzed them separately.

Table 1.

Patient and tumor characteristics

 TN ER+ 
Number of patients 60 103 
Median age (SD) 42 (11.8) 48 (8.9) 
Progesterone receptor, n (%) 
    Positive  70 (68) 
    Negative 60 (100) 32 (31) 
    NA  1 (1) 
T-stage, n (%) 
    T1 3 (5) 12 (12) 
    T2 42 (70) 56 (54) 
    T3 10 (17) 31 (30) 
    T4 5 (8) 4 (4) 
N-stage, n (%) 
    Node negative 23 (38) 16 (16) 
    Node positive 37 (62) 87 (84) 
Chemotherapy, n (%) 
    6 × ddAC 51 (85) 81 (79) 
    3 × ddAC, 3 × DC 9 (15) 22 (21) 
Response, n (%) 
    pCR 21 (35) 12 (12) 
    npCR 10 (17) 12 (12) 
    PR + NR 27 (45) 77 (75) 
    Unknown 2 (3) 2 (2) 
 TN ER+ 
Number of patients 60 103 
Median age (SD) 42 (11.8) 48 (8.9) 
Progesterone receptor, n (%) 
    Positive  70 (68) 
    Negative 60 (100) 32 (31) 
    NA  1 (1) 
T-stage, n (%) 
    T1 3 (5) 12 (12) 
    T2 42 (70) 56 (54) 
    T3 10 (17) 31 (30) 
    T4 5 (8) 4 (4) 
N-stage, n (%) 
    Node negative 23 (38) 16 (16) 
    Node positive 37 (62) 87 (84) 
Chemotherapy, n (%) 
    6 × ddAC 51 (85) 81 (79) 
    3 × ddAC, 3 × DC 9 (15) 22 (21) 
Response, n (%) 
    pCR 21 (35) 12 (12) 
    npCR 10 (17) 12 (12) 
    PR + NR 27 (45) 77 (75) 
    Unknown 2 (3) 2 (2) 

DC, docetaxel, capecitabine; ddAC, dose-dense doxorubixin cyclophosphamide; npCR, near pathological complete remission; pCR, pathologic complete remission; PR + NR, partial and nonresponse; SD, standard deviation; TN, triple negative.

Table 2.

Summary of HRD characteristics

 TN (n = 60), n (%) ER+ (n = 103), n (%) P value 
aCGH BRCA1 like 
    BRCA1 like 34 (57%) 6 (6%)  
    Sporadic like 26 (43%) 97 (94%) <0.001 
aCGH BRCA2 like 
    BRCA2 like 19 (32%) 44 (43%)  
    Sporadic like 41 (68%) 59 (57%) 0.241 
BRCA1 expressiona 
    Low 13 (22%) 2 (2%)  
    Normal/high 23 (38%) 58 (56%) <0.001 
    Not determined 24 (40%) 43 (42%)  
BRCA1 promotor methylationa 
    Methylated 12 (20%) 1 (1%)  
    Unmethylated 37 (62%) 55 (53%) <0.001 
    Not determined 11 (18%) 47 (46%)  
EMSY amplificationa 
    Amplification 2 (3%) 11 (11%)  
    Retention 34 (57%) 72 (70%) 0.339 
    Not determined 24 (40%) 20 (19%)  
 TN (n = 60), n (%) ER+ (n = 103), n (%) P value 
aCGH BRCA1 like 
    BRCA1 like 34 (57%) 6 (6%)  
    Sporadic like 26 (43%) 97 (94%) <0.001 
aCGH BRCA2 like 
    BRCA2 like 19 (32%) 44 (43%)  
    Sporadic like 41 (68%) 59 (57%) 0.241 
BRCA1 expressiona 
    Low 13 (22%) 2 (2%)  
    Normal/high 23 (38%) 58 (56%) <0.001 
    Not determined 24 (40%) 43 (42%)  
BRCA1 promotor methylationa 
    Methylated 12 (20%) 1 (1%)  
    Unmethylated 37 (62%) 55 (53%) <0.001 
    Not determined 11 (18%) 47 (46%)  
EMSY amplificationa 
    Amplification 2 (3%) 11 (11%)  
    Retention 34 (57%) 72 (70%) 0.339 
    Not determined 24 (40%) 20 (19%)  
a

Due to limited biopsy material, methylation, gene expression and EMSY amplification were not carried out on all samples. P values in bold are significant.

aCGH, array comparative genomic hybridization; TN, triple negative.

TN tumors and BRCA1-related abnormalities

The BRCA1-like aCGH profile was predominantly seen in TN tumors (57% in TN versus 6% in ER+ tumors, P < 0.001) (Table 2). Other features of BRCA1 inactivation were assessed by determination of BRCA1 promoter methylation and the level of BRCA1 mRNA expression. These two characteristics were again predominantly observed in TN tumors but were less frequent than a BRCA1-like aCGH pattern: 25% of TN tumors showed BRCA1 promoter methylation and 36% of TN tumors showed a low BRCA1 gene expression.

We subsequently determined the relation between the three BRCA1-related abnormalities. Figure 1A and B show the relation between mRNA expression, methylation and a BRCA1-like aCGH pattern. The cut-off between low and normal BRCA1 gene expression was empirically determined based on methylation status. We assumed that methylated samples would have a low mRNA expression, so the cut-off was set at 0.25 (Figure 1A). All methylated samples therefore have, by definition, a low BRCA1 gene expression. The median mRNA gene expression of methylated samples was 0.156, while unmethylated samples show a value of 0.398. This difference was statistically significant (P < 0.001). We also studied the relation between the BRCA1-like aCGH pattern and the BRCA1 mRNA expression (Figure 1B), as low gene expression could be expected to be associated with a BRCA1-like aCGH pattern. Indeed, most BRCA1-like samples have a low expression of the BRCA1 gene, whereas sporadic-like samples have more frequently a normal mRNA expression level. Samples with a BRCA1-like aCGH profile have a median mRNA expression of 0.226, while sporadic-like samples have a median mRNA expression value of 0.426; however, this difference was not statistically significant. From the 12 tumors with BRCA1 promoter methylation, 8 had a BRCA1-like aCGH pattern and 4 a sporadic-like aCGH pattern. These data show that although there is considerable overlap between the three characteristics, the concordance is not perfect.

Figure 1.

BRCA1 gene expression versus methylation status (P < 0.001) (A) and BRCA1-like aCGH pattern (P = 0.285) (B) in TN tumors.

Figure 1.

BRCA1 gene expression versus methylation status (P < 0.001) (A) and BRCA1-like aCGH pattern (P = 0.285) (B) in TN tumors.

Next, we studied the association between BRCA1 inactivation and clinical and pathological variables and response to chemotherapy with DSB-causing agents. There was no difference in T-stage or N-stage between tumors with and without BRCA1 alterations (Table 3). Patients with tumors showing BRCA1 methylation were younger than those with nonmethylated tumors. Treatment response on A/C was not different between tumors with BRCA1 alterations and without these alterations: 58% versus 48% (P = 0.47) for BRCA1-like versus a sporadic-like aCGH profile; 55% versus 61% (P = 0.70) for methylated versus unmethylated tumors and 54% versus 61% (P = 0.68) for low gene expression versus normal gene expression.

Table 3.

Clinical and pathological characteristics according to BRCA1 alterations in TN tumors

Variable BRCA1-like aCGH
 
BRCA1 methylation
 
BRCA1 gene expression
 
Sporadic like, N (%) BRCA1 like, N (%) P value Unmethylated, N (%) Methylated, N (%) P value Normal mRNA, N (%) Low mRNA, N (%) P value 
T-stage 
    T1/2 20 (77) 25 (74) 0.76 27 (73) 10 (83) 0.47 18 (78) 10 (77) 0.93 
    T3/4 6 (23) 9 (26) 10 (27) 2 (17) 5 (22) 3 (23) 
N-stage 
    LN negative 10 (38) 13 (38) 0.99 16 (43) 4 (33) 0.54 8 (35) 4 (31) 0.81 
    LN positive 16 (62) 21 (62) 21 (57) 8 (67) 15 (65) 9 (69) 
Age 
    ≤40 10 (38) 19 (56) 0.18 15 (41) 11 (92) 0.002 9 (39) 8 (62) 0.2 
    >40 16 (62) 15 (44) 22 (59) 1 (8) 14 (61) 5 (38) 
Response 
    PR + NR 13 (50) 14 (41) 0.47 14 (38) 5 (42) 0.7 9 (39) 6 (46) 0.68 
    pCR + npCR 12 (46) 19 (56) 22 (59) 6 (50) 14 (61) 7 (54) 
    Unknown 1 (4) 1 (3) 1 (3) 1 (8)   
Variable BRCA1-like aCGH
 
BRCA1 methylation
 
BRCA1 gene expression
 
Sporadic like, N (%) BRCA1 like, N (%) P value Unmethylated, N (%) Methylated, N (%) P value Normal mRNA, N (%) Low mRNA, N (%) P value 
T-stage 
    T1/2 20 (77) 25 (74) 0.76 27 (73) 10 (83) 0.47 18 (78) 10 (77) 0.93 
    T3/4 6 (23) 9 (26) 10 (27) 2 (17) 5 (22) 3 (23) 
N-stage 
    LN negative 10 (38) 13 (38) 0.99 16 (43) 4 (33) 0.54 8 (35) 4 (31) 0.81 
    LN positive 16 (62) 21 (62) 21 (57) 8 (67) 15 (65) 9 (69) 
Age 
    ≤40 10 (38) 19 (56) 0.18 15 (41) 11 (92) 0.002 9 (39) 8 (62) 0.2 
    >40 16 (62) 15 (44) 22 (59) 1 (8) 14 (61) 5 (38) 
Response 
    PR + NR 13 (50) 14 (41) 0.47 14 (38) 5 (42) 0.7 9 (39) 6 (46) 0.68 
    pCR + npCR 12 (46) 19 (56) 22 (59) 6 (50) 14 (61) 7 (54) 
    Unknown 1 (4) 1 (3) 1 (3) 1 (8)   

P values in bold are significant. aCGH, array comparative genomic hybridization; npCR, near pathologic complete remission; pCR, pathologic complete remission; LN, lymph node; TN, triple negative.

ER + tumors and BRCA2-like profile and EMSY amplification

In the ER+/HER2− tumors, almost exclusively BRCA2-related characteristics were observed: a BRCA2-like aCGH profile or amplification of the BRCA2-inhibiting gene EMSY. Interestingly, those two aberrations were nearly mutually exclusive as a BRCA2-like aCGH pattern and EMSY amplification only occurred in one tumor sample together.

Table 4 shows the association between a BRCA2-like aCGH pattern, EMSY amplification and clinical pathological factors. The BRCA2-like aCGH pattern was significantly associated with a lower PR percentage. EMSY amplification was not associated with any of the clinical pathological variables. A BRCA2-like aCGH profile was significantly associated with a higher pCR + npCR rate to neoadjuvant chemotherapy (35% versus 14%, P = 0.014). To control for possible confounders, T-stage, N-stage, ER percentage, PR percentage and age were included in a multivariate analysis. We were not able to include histological grade in the analysis, as it cannot reliably be assessed in the small pretreatment needle biopsies. Table 5 shows that only PR percentage remains significant in multivariate analysis. However, the BRCA2-like aCGH profile is more significant than established prognostic factors, such as T-stage, N-stage and ER percentage.

Table 4.

Clinical and pathological characteristics according to BRCA2 alterations in ER+ tumors

Variable BRCA2-like aCGH
 
EMSY2
 
Sporadic like, N (%) BRCA2 like, N (%) P value Retention, N (%) Amplification, N (%) P value 
T-stage 
    T1/2 41 (69) 27 (61) 0.39 46 (64) 8 (73) 0.57 
    T3/4 18 (31) 17 (39) 26 (36) 3 (27) 
N-stage 
    LN negative 9 (15) 7 (16) 0.93 8 (11) 2 (18) 0.5 
    LN positive 50 (85) 37 (84) 64 (89) 9 (82) 
ER_50 
    ER < 50% 2 (3) 3 (7) 0.42 3 (4)  0.49 
    ER ≥ 50 57 (97) 41 (93) 69 (96) 11 (100) 
PR_50 
    PR < 50% 26 (44) 28 (64) 0.04 37 (51) 6 (55) 0.88 
    PR ≥ 50% 33 (56) 15 (34) 34 (47) 5 (45) 
    Unknown  1 (2) 1 (1)   
Age 
    ≤40 11 (19) 10 (23) 0.61 16 (22) 2 (18) 0.76 
    >40 48 (81) 34 (77) 56 (78) 9 (82) 
Response 
    PR + NR 49 (83) 28 (64) 0.014 56 (78) 8(73) 0.647 
    pCR + npCR 8 (14) 15 (34) 15 (21) 3 (27) 
    Unknown 2 (3) 1 (2) 1 (1)   
Variable BRCA2-like aCGH
 
EMSY2
 
Sporadic like, N (%) BRCA2 like, N (%) P value Retention, N (%) Amplification, N (%) P value 
T-stage 
    T1/2 41 (69) 27 (61) 0.39 46 (64) 8 (73) 0.57 
    T3/4 18 (31) 17 (39) 26 (36) 3 (27) 
N-stage 
    LN negative 9 (15) 7 (16) 0.93 8 (11) 2 (18) 0.5 
    LN positive 50 (85) 37 (84) 64 (89) 9 (82) 
ER_50 
    ER < 50% 2 (3) 3 (7) 0.42 3 (4)  0.49 
    ER ≥ 50 57 (97) 41 (93) 69 (96) 11 (100) 
PR_50 
    PR < 50% 26 (44) 28 (64) 0.04 37 (51) 6 (55) 0.88 
    PR ≥ 50% 33 (56) 15 (34) 34 (47) 5 (45) 
    Unknown  1 (2) 1 (1)   
Age 
    ≤40 11 (19) 10 (23) 0.61 16 (22) 2 (18) 0.76 
    >40 48 (81) 34 (77) 56 (78) 9 (82) 
Response 
    PR + NR 49 (83) 28 (64) 0.014 56 (78) 8(73) 0.647 
    pCR + npCR 8 (14) 15 (34) 15 (21) 3 (27) 
    Unknown 2 (3) 1 (2) 1 (1)   

P values in bold are significant. aCGH, array comparative genomic hybridization; pCR, pathologic complete remission; npCR, near pathologic complete remission.

Table 5.

Multivariate analysis for response in ER+ patients treated with A/C chemotherapy

Variablesa  OR (95% CI) P value 
Ageb Per year 0.95 (0.90–1.01) 0.08 
T-stageb  1.224 (0.59–2.53) 0.59 
Lymph node Positive versus negative 0.609 (0.16–2.30) 0.46 
ER percentageb Per percent 0.994 (0.97–1.02) 0.61 
PR percentageb Per percent 0.981 (0.97–0.997) 0.019 
BRCA2-aCGH profile BRCA2-like versus sporadic like 2.398 (0.82–6.98) 0.11 
Variablesa  OR (95% CI) P value 
Ageb Per year 0.95 (0.90–1.01) 0.08 
T-stageb  1.224 (0.59–2.53) 0.59 
Lymph node Positive versus negative 0.609 (0.16–2.30) 0.46 
ER percentageb Per percent 0.994 (0.97–1.02) 0.61 
PR percentageb Per percent 0.981 (0.97–0.997) 0.019 
BRCA2-aCGH profile BRCA2-like versus sporadic like 2.398 (0.82–6.98) 0.11 
a

Grade could not be reliable assessed in tumor biopsies.

b

Age, T-stage, ER% and PR% were included as continuous variables. P values in bold are significant.

aCGH, array comparative genomic hybridization; CI, confidence interval; OR, odds ratio.

discussion

In the series of patients described in this paper, the frequency of certain features associated with HRD was studied in untreated breast cancers and possible relationships with neoadjuvant treatment response were explored. This study was restricted to HER2-negative tumors, as we wanted to study the effect of DNA DSB-inducing agents unperturbed by the effect of targeted therapy such as Traztuzumab. In TN tumors, we found mainly BRCA1-related abnormalities, whereas in ER+ tumors, BRCA2 dysfunction-associated characteristics were the predominant features. A significantly higher response rate to DSB-causing chemotherapy was observed in ER+ tumors with the BRCA2-like aCGH profile than in ER+ tumors with a sporadic-like aCGH profile. If these findings are validated in an independent study, there could be important implications for breast cancer chemotherapy selection for luminal tumors.

The higher response percentage observed in patients with tumors with a BRCA2-like aCGH pattern could be due to the underlying biology of these tumors. A genomic pattern like that seen in BRCA2-mutated cancer may be due to HRD. In the absence of homologous recombination, adequate DNA DSB repair is impossible, and DSB-inducing chemotherapeutic agents such as cyclcophosphamide and doxorubicin may target the Achilles heel of the tumors. In the current study, standard systemic therapy with DNA damaging agents was employed, but the novel class of the PARP inhibitors, which specifically target the DNA repair enzyme PARP, should be particularly effective in this tumor type. Recent studies have shown promising results in BRCA1- and BRCA2-associated breast cancer treated with PARP inhibitors [13]. Based on results of our findings, we hypothesize that tumors with a BRCA2-like aCGH pattern could benefit from this new class of drugs as well.

In TN tumors, no difference in response rates was observed between patients with BRCA1-like aCGH tumors and tumors with a sporadic-like aCGH pattern. In a recent study of our institute, the BRCA1-like aCGH pattern was shown to be associated with an important survival benefit of intensive treatment with platinum-based chemotherapy for high-risk primary breast cancer [M. A. Vollebergh, E. H. Lips, P. M. Nederlof et al., unpublished data]. It is possible that any hypersensitivity to DSB-inducing agents only shows at higher doses, while the lower standard dose causes increased genomic instability rather than cell death.

Our finding that tumors with a BRCA2-like aCGH profile are more sensitive to a combination of an alkylator and an anthracycline compared with tumors with a sporadic-like aCGH pattern is consistent with the results of a study in metastatic breast cancer. In a recent report by Kriege et al. [23], it was shown that BRCA2 hereditary breast cancers were more sensitive to chemotherapy with anthracyclines or combination chemotherapy with cyclophosphamide, methotrexate and fluorouracil than sporadic breast cancers. For BRCA1 hereditary breast cancer, there was no significant difference in sensitivity. The authors explain the difference in outcome between BRCA1- and BRCA2-mutated tumors by different tumor characteristics, including higher grade, triple negativity and a higher incidence of p53 mutations. Our finding that aberrations in BRCA1 are characteristic for TN tumors and aberrations of BRCA2 for ER + tumors is in line with this. BRCA1-mutated tumors are usually basal like or triple negative, while BRCA2-mutated tumors comprise the complete spectrum of subtypes seen in sporadic tumors. The divergent results we found in ER + and TN tumors underline the need for subtype-specific analysis.

The role of EMSY in HRD remains questionable. In vitro assays have shown that the EMSY protein can bind BRCA2 protein and inactivate its function [24]. An increase in chromosomal instability was observed after EMSY overexpression in vitro. However, it is not clear if EMSY amplification affects the role of BRCA2 in the maintenance of genomic instability in vivo. By counting the number of genomic breakpoints, it was clear that tumors with EMSY amplification displayed a lesser degree of genomic instability than tumors with normal EMSY expression. Furthermore, EMSY and a BRCA2-like aCGH pattern appeared to be almost mutually exclusive events, as only in a single patient's tumor, both characteristics were observed. As EMSY was not related to genomic instability or chemotherapy response, its role in HRD remains to be determined.

A limitation of this study was that we could not assess the association between histological grade, HRD characteristics and chemotherapy response. High tumor grade is associated with decreased overall survival [25], but it also predicts increased response to neoadjuvant chemotherapy [26]. However, assessment of grade on small needle biopsy material is unreliable and was therefore not included in our analyses. It could not be excluded that the associations found between the BRCA2-like aCGH profile and treatment response will be confounded by histological grade.

In conclusion, we showed that a BRCA2-like aCGH profile is a strong predictor for chemotherapy response in ER + tumors. In TN tumors, BRCA-ness occurred in about half of all cases but did not predict a better treatment response to standard dose chemotherapy with AC. It is certainly possible that conventional doses of cisplatin or carboplatin would be highly effective in this subgroup, as suggested in the literature [19]. In contrast, the BRCA2-like aCGH pattern may be useful for treatment selection as it predicts sensitivity to DNA DSB-inducing chemotherapy and possibly to PARP inhibitors. However, validation on an independent set of samples finding will be important. Tests such as these may represent a further step toward truly personalized medicine in breast cancer.

funding

CTMM, Center for Translational Molecular Medicine (project Breast CARE grant 030-104).

disclosure

The authors declare no conflicts of interest.

This study was carried out within the framework of CTMM, the Center for Translational Molecular Medicine (www.ctmm.nl), project Breast CARE.

References

1.
Rastogi
P
Anderson
SJ
Bear
HD
, et al.  . 
Preoperative chemotherapy: updates of National Surgical Adjuvant Breast and Bowel Project Protocols B-18 and B-27
J Clin Oncol
 , 
2008
, vol. 
26
 (pg. 
778
-
785
)
2.
van der Hage
JA
van de Velde
CJ
Julien
JP
, et al.  . 
Preoperative chemotherapy in primary operable breast cancer: results from the European Organization for Research and Treatment of Cancer trial 10902
J Clin Oncol
 , 
2001
, vol. 
19
 (pg. 
4224
-
4237
)
3.
Gianni
L
Baselga
J
Eiermann
W
, et al.  . 
Feasibility and tolerability of sequential doxorubicin/paclitaxel followed by cyclophosphamide, methotrexate, and fluorouracil and its effects on tumor response as preoperative therapy
Clin Cancer Res
 , 
2005
, vol. 
11
 (pg. 
8715
-
8721
)
4.
Sachelarie
I
Grossbard
ML
Chadha
M
, et al.  . 
Primary systemic therapy of breast cancer
Oncologist
 , 
2006
, vol. 
11
 (pg. 
574
-
589
)
5.
Joosse
SA
van Beers
EH
Tielen
IH
, et al.  . 
Prediction of BRCA1-association in hereditary non-BRCA1/2 breast carcinomas with array-CGH
Breast Cancer Res Treat
 , 
2009
, vol. 
116
 (pg. 
479
-
489
)
6.
Turner
N
Tutt
A
Ashworth
A
Hallmarks of 'BRCAness' in sporadic cancers
Nat Rev Cancer
 , 
2004
, vol. 
4
 (pg. 
814
-
819
)
7.
Hughes-Davies
L
Huntsman
D
Ruas
M
, et al.  . 
EMSY links the BRCA2 pathway to sporadic breast and ovarian cancer
Cell
 , 
2003
, vol. 
115
 (pg. 
523
-
535
)
8.
Kennedy
RD
Quinn
JE
Mullan
PB
, et al.  . 
The role of BRCA1 in the cellular response to chemotherapy
J Natl Cancer Inst
 , 
2004
, vol. 
96
 (pg. 
1659
-
1668
)
9.
Rottenberg
S
Nygren
AO
Pajic
M
, et al.  . 
Selective induction of chemotherapy resistance of mammary tumors in a conditional mouse model for hereditary breast cancer
Proc Natl Acad Sci U S A
 , 
2007
, vol. 
104
 (pg. 
12117
-
12122
)
10.
Rottenberg
S
Jaspers
JE
Kersbergen
A
, et al.  . 
High sensitivity of BRCA1-deficient mammary tumors to the PARP inhibitor AZD2281 alone and in combination with platinum drugs
Proc Natl Acad Sci U S A
 , 
2008
, vol. 
105
 (pg. 
17079
-
17084
)
11.
Ratnam
K
Low
JA
Current development of clinical inhibitors of poly(ADP-ribose) polymerase in oncology
Clin Cancer Res
 , 
2007
, vol. 
13
 (pg. 
1383
-
1388
)
12.
O'Shaughnessy
J
Osborne
C
Pippen
J
, et al.  . 
Efficacy of BSI-201, a poly (ADP-ribose) polymerase-1 (PARP1) inhibitor, in combination with gemcitabine/carboplatin (G/C) in patients with metastatic triple-negative breast cancer (TNBC): results of a randomized phase II trial
J Clin Oncol
 , 
2009
, vol. 
27
  
(Abstr 3)
13.
Fong
PC
Boss
DS
Yap
TA
, et al.  . 
Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers
N Engl J Med
 , 
2009
, vol. 
361
 (pg. 
123
-
134
)
14.
Waddell
N
Arnold
J
Cocciardi
S
, et al.  . 
Subtypes of familial breast tumours revealed by expression and copy number profiling
Breast Cancer Res Treat
  
2009 Dec 4 [Epub ahead of print]
15.
Tirkkonen
M
Johannsson
O
Agnarsson
BA
, et al.  . 
Distinct somatic genetic changes associated with tumor progression in carriers of BRCA1 and BRCA2 germ-line mutations
Cancer Res
 , 
1997
, vol. 
57
 (pg. 
1222
-
1227
)
16.
Stefansson
OA
Jonasson
JG
Johannsson
OT
, et al.  . 
Genomic profiling of breast tumours in relation to BRCA abnormalities and phenotypes
Breast Cancer Res
 , 
2009
, vol. 
11
 pg. 
R47
 
17.
Jonsson
G
Naylor
TL
Vallon-Christersson
J
, et al.  . 
Distinct genomic profiles in hereditary breast tumors identified by array-based comparative genomic hybridization
Cancer Res
 , 
2005
, vol. 
65
 (pg. 
7612
-
7621
)
18.
Wessels
LF
van Welsem
T
Hart
AA
, et al.  . 
Molecular classification of breast carcinomas by comparative genomic hybridization: a specific somatic genetic profile for BRCA1 tumors
Cancer Res
 , 
2002
, vol. 
62
 (pg. 
7110
-
7117
)
19.
Silver
DP
Richardson
AL
Eklund
AC
, et al.  . 
Efficacy of neoadjuvant cisplatin in triple-negative breast cancer
J Clin Oncol
 , 
2010
, vol. 
28
 (pg. 
1145
-
1153
)
20.
Loo
CE
Teertstra
HJ
Rodenhuis
S
, et al.  . 
Dynamic contrast-enhanced MRI for prediction of breast cancer response to neoadjuvant chemotherapy: initial results
AJR Am J Roentgenol
 , 
2008
, vol. 
191
 (pg. 
1331
-
1338
)
21.
Joosse
SA
van Beers
EH
Nederlof
PM
Automated array-CGH optimized for archival formalin-fixed, paraffin-embedded tumor material
BMC Cancer
 , 
2007
, vol. 
7
 pg. 
43
 
22.
Joosse
SA
Brandwijk
KI
Devilee
P
, et al.  . 
Prediction of BRCA2-association in hereditary breast carcinomas using array-CGH
Breast Cancer Res Treat
  
2010 Jul 8 [Epub ahead of print]
23.
Kriege
M
Seynaeve
C
Meijers-Heijboer
H
, et al.  . 
Sensitivity to first-line chemotherapy for metastatic breast cancer in BRCA1 and BRCA2 mutation carriers
J Clin Oncol
 , 
2009
, vol. 
27
 (pg. 
3764
-
3771
)
24.
Raouf
A
Brown
L
Vrcelj
N
, et al.  . 
Genomic instability of human mammary epithelial cells overexpressing a truncated form of EMSY
J Natl Cancer Inst
 , 
2005
, vol. 
97
 (pg. 
1302
-
1306
)
25.
Trudeau
ME
Pritchard
KI
Chapman
JA
, et al.  . 
Prognostic factors affecting the natural history of node-negative breast cancer
Breast Cancer Res Treat
 , 
2005
, vol. 
89
 (pg. 
35
-
45
)
26.
Fisher
ER
Wang
J
Bryant
J
, et al.  . 
Pathobiology of preoperative chemotherapy: findings from the National Surgical Adjuvant Breast and Bowel (NSABP) protocol B-18
Cancer
 , 
2002
, vol. 
95
 (pg. 
681
-
695
)