Background

Women carrying a pathogenic germline mutation in the BRCA1 and BRCA2 genes have an increased lifetime risk of developing breast, ovarian, and several other cancers [1]. The identification of women harboring mutations in these genes is clinically important and has a significant socio-cultural impact. A major challenge faced by physicians is to identify most appropriate candidates for genetic BRCA1/2 testing since the cost of comprehensive genetic testing can be high and only 3 % of all breast cancers are attributed to BRCA1/2 germline mutations.

The decision to offer genetic testing to a breast cancer patient is currently based on family history of breast/ovarian cancer and age of disease onset. Several prediction models, which consider age of onset and family history of cancer, can be used to estimate the prior probability of having a BRCA1 or BRCA2 mutation [2]. In addition, histopathological tumor parameters can be considered to help predict the presence of a mutation.

Triple negative breast cancer (TNBC) is defined by the absence of estrogen receptor (ER), progesterone receptor (PR), human epidermal growth factor receptor 2 (HER2) and accounts for 12–15 % of all invasive breast cancer [3]. It occurs most frequently in young women and African-Americans. In Pakistan, 10-year outcome analysis of 636 breast cancer patients registered at a tertiary-care cancer center (Shaukat Khanum Memorial Cancer Hospital and Research Centre - SKMCH & RC) showed that 30.5 % (194/636) of the cases had TNBC; and majority (56.2 %) had their diagnosis made at less than 40 years of age [4]. Patients with TNBC are known to have unfavorable survival compared to patients with other breast cancer subtypes [5].

A large proportion of tumors in women with BRCA1 mutations are associated with the TNBC phenotype [6]. BRCA1/2 mutations have been identified with frequencies varying from 9.4 to 15.4 % in unselected, 17.4 to 49.1 % in younger age and 11.6 to 62 % in high risk patients with TNBC [7–15]. Studies reporting the frequency of BRCA1/2 mutations in TNBC patients from Asia have had several deficiencies including small population size [16–18], restriction of analysis to BRCA1 gene [19, 20] and evaluation limited to small-range mutations [16, 21, 22]. In order to determine the utility of genetic testing for BRCA1 and BRCA2 germline mutations for women with TNBC in Pakistan, we comprehensively screened both genes for small-range mutations as well as large genomic rearrangements in a group of 523 breast cancer patients who were selected based on early-age of disease onset or family history of breast/ovarian cancer, including 192 patients diagnosed with TNBC.

Methods

Study subjects

Index patients included in this study had a diagnosis of primary invasive breast cancer and were selected based on the following criteria: 1) one female breast cancer diagnosed ≤ 30 years of age; 2) two or more first- or second-degree (through a male) female relatives diagnosed with breast cancer with at least one diagnosed ≤ 50 years of age; or 3) at least one female breast cancer and one ovarian cancer at any age. A total of 573 women recruited at the SKMCH & RC in Lahore, Pakistan, from June 2001 to February 2014 fulfilled these criteria. Blood samples were obtained from all patients for the isolation of genomic DNA. Clinical, histopathologic and risk factor data were collected from all study participants. Fifty patients were excluded from the study. Reasons for exclusion are detailed in Fig. 1.

Fig. 1
figure 1

Description of the study participants. BC, breast cancer; ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; OC, ovarian cancer; PR, progesterone receptor; TNBC, triple-negative breast cancer

Full size image

The study was approved by the Ethical Review Board of the SKMCH & RC. All study participants signed informed written consent.

BRCA1/2 screening for small-range mutations and large genomic rearrangements

Genomic DNA was isolated as previously described [23]. One hundred and twenty-one cases comprehensively screened for BRCA1 (Genbank accession number U14680.1) and BRCA2 (Genbank accession number U43746.1) small-range mutations using protein-truncation test (PTT), single-strand conformational polymorphism analysis (SSCP) and denaturing high-performance liquid chromatography (DHPLC) analysis followed by DNA sequencing of variant fragments, and 26 BRCA1/2 mutations was described in an earlier report [23] (primer sequences are available upon request). When available, a mutation positive control was included in each set of PTT, SSCP and DHPLC analyses. A description of the BRCA1/2 screening methods is given in Supplementary methods (Additional file 1). The remaining 402 cases recruited subsequently were screened for BRCA1/2 small-range mutations using DHPLC and DNA sequencing analyses. Of these, 295 cases were previously described [24]. All patients negative for small-range BRCA1/2 mutations were further screened for large genomic rearrangements. Multiplex ligation-dependent probe amplification (MLPA) analysis was performed using probe mix P003 and P087 for BRCA1 and probe mix P045 for BRCA2 according to the manufacturer’s instructions (MRC Holland, Amsterdam, The Netherlands).

Immunohistochemical (IHC) analysis

Formalin-fixed paraffin-embedded (FFPE) blocks were retrieved from the pathology department; blocks were not available for 38 patients (Fig. 1). Tumor grade was assigned using the Nottingham Histologic Score. IHC analysis of ER, PR and HER2 expression was performed using standard methods [25]. Slides were interpreted by a trained breast pathologist who was blinded to BRCA1/2 mutation status. Tumors were considered negative for ER and PR if < 1 % of tumor cells demonstrated positive nuclear staining. Tumors were considered negative for HER2 if IHC score was 0 or 1+. Cases with IHC score 2+ were further subjected to fluorescence in situ hybridization (FISH) using the PathVysion® HER2 DNA probe kit (Abbott Laboratories, Abbott Park, IL). Tumors with a HER2/CEP17 ratio of > 2.2 and tumors with IHC score 3+ were considered positive.

Statistical analysis

The comparison of the distribution of clinical and histopathological characteristics between BRCA1/2 carriers and non-carriers was done using Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test for quantitative variables. Univariate and multiple logistic regression models were used to investigate possible differences in the prevalence of BRCA1/2 mutations according to patient and tumor characteristics. All statistical tests were two sided. Results were deemed statistically significant if the P value was 0.05 or less. All statistical computations were done using StatXact 4 for Windows (Cytel Inc., Cambridge, USA), SAS version 9.3 and R, version 2.1.

Results

Clinical characteristics of the study participants and histopathologic parameters of tumors according to TNBC status

In total 523 unrelated Pakistani women diagnosed with primary invasive breast cancer were included in the study. Of these, 45.3 % were diagnosed at young age (≤ 30 years) and 54.7 % reported a positive family history of breast/ovarian cancer. IHC analysis of ER, PR, and HER2 expression showed that 36.7 % of the patients presented with TNBC. Compared to non-TNBC patients, women with TNBC had an earlier age of diagnosis (31.6 years (range 18–67) and 35.6 years (range 19–73), respectively; P < 0.0001, Wilcoxon rank-sum test), were more often premenopausal (91.1 % vs. 80.7 %, P = 0.002) and of Punjabi ethnicity (79.5 % vs. 69.2 %, P = 0.03). TNBC tumors were observed to have greater propensity for invasive ductal carcinoma compared to non-TNBC (96.4 % vs. 89.4 %, P = 0.004), higher tumor grade 3 (88.8 % vs. 62.9 %, P < 0.0001), and lymph node negativity (53.9 % vs. 32.5 %, P < 0.0001). Selected clinical and histopathologic characteristics of the study participants by TNBC status are shown in Table 1.

Table 1 Selected clinical and pathological characteristics of the 523 Pakistani cases according to TNBC status
Full size table

BRCA1/2 mutation prevalence in patients with TNBC and non-TNBC

The complete coding regions of the BRCA1 and BRCA2 genes were screened for small-range mutations and large genomic rearrangements in all 523 breast cancer patients. Overall, 125 cases with deleterious mutations were identified, of these, 105 occurred in BRCA1 (84 %) and 20 (16 %) in BRCA2 (Table 2). BRCA1 mutations were more frequent in patients with TNBC than in those with non-TNBC (37 % vs. 10.3 %, P < 0.0001). Majority of the mutations in patients with TNBC, 97.3 % (71/73), were detected in BRCA1; 2.7 % (2/73) had mutations in BRCA2 (P < 0.0001) (Additional file 2: Table S1). The corresponding percentage for BRCA1 and BRCA2 mutations in non-TNBC cases was 65.4 % (34/52) and 34.6 % (18/52) (P = 0.04).

Table 2 BRCA1/2 mutation frequencies in patients with TNBC and non-TNBC
Full size table

In this study, patients with TNBC harboring a BRCA1 mutation (n = 71) were older than BRCA1/2 non-carriers (n = 119) with mean age of diagnosis 32.9 years (range 22–53) and 30.9 years (range 18–67), respectively (P = 0.002, Exact Wilcoxon rank-sum test). The mean age for non-TNBC patients was 33.5 years (range 21–72) for BRCA1 carriers (n = 34), 36.8 (range 25–54) for BRCA2 carriers (n = 18) and 35.7 (range 19–73) for non-carriers (n = 279).

Subgroup analysis by family phenotype, age of diagnosis, and ethnicity

Prevalence of BRCA1/2 mutations in patients with TNBC and non-TNBC distributed by family phenotype, age of diagnosis, and ethnicity is detailed in Table 3. Among patients with TNBC, BRCA1 mutations were identified in 14.4 % of patients with early-onset disease (≤ 30 years), 48.7 % of patients with familial breast cancer and in 80.8 % of patients with familial breast and ovarian cancer. These frequencies were higher than the corresponding frequencies of 5.4, 12.0 and 28.0 % observed in non-TNBC patients (P = 0.03, P < 0.0001 and P = 0.0005, respectively). Mutations in BRCA2 were detected in 2.6 % of patients with familial breast cancer and were absent in those with early-onset disease and familial breast and ovarian cancer. Similar frequencies were observed in non-TNBC patients; 4.1 % in patients with early-onset disease, 6.9 % in those with familial breast cancer, and 4.0 % in patients with familial breast and ovarian cancer (P = 0.09, P = 0.73 and P = 1.0, respectively).

Table 3 BRCA1/2 mutation frequencies in patients with TNBC and non-TNBC, by age-at-diagnosis, family phenotype and ethnicity
Full size table

In this study, age appeared to have a marked influence on the BRCA1 mutation frequency in familial breast/ovarian cancer patients diagnosed with TNBC. In patients over 50 years of age, the frequency was 11.1 %. For younger patients the frequency was 58.5 % for those ≤ age 30, 68.8 % for those between 31 and 40 years, and 55 % for those between 41 and 50 years. The BRCA1 mutation frequency in the age subgroups 31–40 and 41–50 years were higher in patients with TNBC than those with non-TNBC (68.8 % vs. 15.2 %, P < 0.0001 and 55 % vs. 7 %, P < 0.0001). Higher BRCA1 mutation frequency was observed in early-onset breast cancer patients regardless of family history of breast/ovarian cancer (28.2 % vs. 11 %, P = 0.0003).

In this study, analysis by ethnicity showed that BRCA1 mutation frequency in patients with TNBC belonging to the ethnic group of Punjabis, Pathans and other minor ethnic groups was higher than observed in non-TNBC patients (37.1 % vs. 12.2 %, P < 0.0001; 31.6 % vs. 5.6 %, P = 0.01 and 40 % vs. 6.2 %, P = 0.003), respectively.

Results from logistic regression analysis

BRCA1/2 mutation carriers and non-carriers were diagnosed with breast cancer at similar age. Each additional year at diagnosis translated into a 1 % lower risk of carrying BRCA1 mutations and a 1 % higher risk of harboring BRCA2 mutations, but differences did not reach statistical significance (Ratio of the probability of carrying BRCA1/2 mutations (RP) = 0.99; 95 % CI 0.97–1.01; P = 0.34 and RP = 1.01; 95 % CI 0.97–1.05; P = 0.56), respectively (Table 4). Patients with a family history of breast cancer, and in particular patients with a family history of breast and ovarian cancer, showed 237 and 1172 % increased risk of carrying BRCA1 mutations, respectively, compared to women affected by early-onset breast cancer (Global P < 0.0001) (corresponding RP = 3.37; 95 % CI 1.96–5.80; and RP = 12.72; 95 % CI 6.22–26.0, respectively). Patients presenting with breast tumor histology of other than invasive ductal carcinoma showed a 73 % decreased risk of BRCA1 mutations (RP = 0.27; 95 % CI 0.08–0.89; P = 0.03). The prevalence of BRCA1 mutations also varied with tumor grade; women affected by grade 3 tumors showed the highest risk of carrying a BRCA1 mutation (Global P < 0.0001). In comparison with patients diagnosed with non-TNBC, patients affected by TNBC showed a 390 % higher risk of BRCA1 mutations (RP = 4.90; 95 % CI 3.09–7.77; P < 0.0001).

Table 4 Ratio of the probability of carrying BRCA1/2 mutations in the investigated patients collective based on univariate logistic regression models
Full size table

The association between TNBC and prevalent BRCA1 mutations was independent of the simultaneous consideration of family history, tumor histology and tumor grade in a multiple logistic regression model (Table 5). After adjustment for family history, tumor histology and tumor grade, patients affected by TNBC showed a 323 % higher risk of BRCA1 mutations than non-TNBC patients (RP = 4.23; 95 % CI 2.50–7.14; P < 0.0001).

Table 5 Ratio of the probability of carrying BRCA1/2 mutations in the investigated patients collective relying on multiple logistic regression model
Full size table

Discussion

We report a comprehensive analysis of the prevalence of BRCA1 and BRCA2 germline mutations in Pakistani patients with TNBC and non-TNBC selected for age of onset or family history of breast/ovarian cancer. Results from our analysis showed that 97 % of all BRCA1/2 mutations in patients with TNBC were found in the BRCA1 gene. The BRCA1 mutation frequency in patients diagnosed at early age who did not report a family history of breast/ovarian cancer, patients diagnosed at early age irrespective of family history, patients with a family history of breast cancer, and patients with a family history of breast and ovarian cancer were approximately 3 to 4 times higher than those observed in non-TNBC patients. The diagnosis of TNBC independently increased the risk of carrying a BRCA1/2 mutation. Several studies have demonstrated the relevance of TNBC status as a criterion for genetic BRCA testing [7, 13, 15, 22, 26–28]; our study confirms this observation for patients with TNBC in an Asian population from Pakistan.

Pakistani women were diagnosed with TNBC at a younger age and with higher grade tumors than non-TNBC. These findings confirm those from previous studies conducted among Asian [4, 29], North-American [30, 31] and African-American patients [31, 32]. A lower rate of lymph node involvement was observed in Pakistani patients with TNBC than non-TNBC, which is in line with previous data from other Asian [29, 33, 34] and North-American studies [35]. A higher rate was observed in one study among North-Americans [30], while no difference was observed in a European study [36]. The discrepant data may be explained by differences in the study design or the IHC cut-off values for ER and PR negativity. While data from the study of Dent and colleagues were based on unselected cases and a cut-off for ER/PR negativity of < 10 % of tumor cells staining positive, the Pakistani study participants were selected for young age or family history of breast/ovarian cancer and the threshold for negative ER/PR result was < 1 % of tumor cells staining positive.

In Pakistan, 42 distinct mutations including 40 in BRCA1 and two in BRCA2 were identified in patients with TNBC. Of these mutations, 17 mutations (including five mutations previously identified in Pakistani breast/ovarian cancer patients) are population-specific as they were not identified in other populations [23, 37]. Twenty-five recurrent BRCA1/2 mutations (including 18 mutations previously reported in Pakistani breast/ovarian cancer patients) have also been described elsewhere in the world, indicating that majority of mutations found in the current study did not differ from those previously reported in Pakistan or elsewhere.

In most Western studies, the mean age of diagnosis of TNBC in BRCA1 mutation carriers was significantly lower than in non-carriers [7, 13, 15]. No difference in the age of TNBC diagnosis between BRCA1 carriers and non-carriers was detected in some studies on early-onset or familial cases from the US [6] and Singapore [18]. In contrast, Pakistani BRCA1 carriers were two years older at TNBC diagnosis than non-carriers implying that other environmental or genetic factors may be operant in TNBC in this group of women. It is also possible that the diverse results are due to differences in study design, selection criteria or ethnicity.

Pakistani women are usually diagnosed with breast cancer below 40 years of age [38] and often present with advanced disease [39]. In the current study, BRCA1 mutations were identified in 14.4 % of early-onset patients with TNBC, who had no family history of breast/ovarian cancer. Lower BRCA1 mutations frequencies of 4.3, 7.4 and 8.7 % were observed in other studies conducted in China, Italy, and the UK, respectively [9, 22, 40]. However, given the small number of patients with TNBC diagnosed < 30 years of age investigated in these studies (n = ≤ 30), the percentages may not be truly representative.

In the present study BRCA1/2 mutations were identified in 58.8 % of patients with TNBC, who reported a family history of breast/ovarian cancer. In other Asian studies performed in China, Malaysia, and Korea and also in Caucasian studies conducted in Australia, Europe, and the United States, the mutation frequencies were similar or lower ranging from 20.8 to 59.5 % [17, 22, 26, 41, 42] and 11.6 to 62 % [7, 8, 10–13, 15], respectively. The varying mutation frequencies obtained in these studies may be explained by differences in sample size, mutation detection assays used, or ethnic origin of study participants.

The low frequency of BRCA2 mutations detected in our study is in keeping with prior reports and suggests that BRCA2 may not play an important role in the development of early-onset TNBC. With the exception of one small German study that included 30 patients with TNBC [11], BRCA2 mutations were less common than BRCA1 mutations in several studies among patients of European or North-American origin [9, 14, 15, 28] and patients from Asia [22, 26] including the present one. These data indicate the tendency for BRCA1 carriers to primarily develop TNBC compared to BRCA2 carriers, which most commonly develop ER positive breast tumors [43].

Recommendations for genetic BRCA1/2 testing for patients with TNBC are not universally accepted and vary between professional societies [13] and studies [15, 22, 26, 28]. The National Comprehensive Cancer Network (NCCN) guidelines recommend that women with TNBC diagnosed before or at age 60 should be considered for genetic BRCA1/2 testing (NCCN Guidelines), while the guidelines of the European Society of Medical Oncology [44] and the Cancer Institute New South Wales (https://www.eviq.org.au) recommend testing if TNBC is diagnosed under the ages of 50 and 40 years, respectively. Moreover, testing was suggested to Mexican patients affected by disease below age 60 [28], below or at age 50 to patients from the UK [27], China [22] and Malaysia [26], and irrespective of age to Polish and Australian patients [15]. The high frequency of BRCA1 mutations in Pakistani patients with a family history of breast/ovarian cancer diagnosed with TNBC below or at age 50 and in early-onset patients diagnosed before or at age 30 irrespective of family history suggest that genetic testing should be considered for these groups of women. Testing women with TNBC diagnosed below age 50 has previously been shown to be a cost-effective strategy [45]. Given the financial burden these considerations are of particular importance for developing countries like Pakistan.

Recently, deleterious mutations in 14 known breast cancer susceptibility genes including BRCA1, BRCA2, and RAD51C were identified at a frequency of 3.7 % in a large series of 1,824 patients with TNBC unselected for family history of breast cancer [7]. As in the study reported by Couch and colleagues, no mutations in the CHEK2 and TP53 genes were observed in two Pakistani studies among 374 (including 103 with TNBC) [46] and 105 (including 47 with TNBC) breast/ovarian cancer patients [47], respectively. Recently, a deleterious mutation (c.5101C > T) in the FANCM gene was identified in BRCA1/2-negative familial patients with TNBC from Finland [48]. This mutation was not detected in a Pakistani study that included 117 patients with TNBC [49].

There are several limitations of our study. First, we have screened only patients with TNBC, who were selected for early-age of onset (≤ 30 years) or family history of breast/ovarian cancer. Hence the selection of high-risk patients may explain the higher BRCA1/2 mutation frequency observed in our study compared to those that evaluated unselected TNBC patients. Secondly, we did not use BRCA1/2 prediction models. However, given the previously observed inaccuracy of these algorithms in predicting risk precisely in Asian populations, limits the usefulness of these algorithms and warrants further investigation [50, 51]. Strengths of the present study include the sample size (N = 523) comprising sufficiently larger number of early-onset breast cancer (≤ 30 years) women (n = 303) with TNBC (n = 131) or non-TNBC (n = 172) compared to studies reported from Asia previously. Additionally, our study evaluated the complete coding regions of the BRCA1 and BRCA2 genes that were comprehensively screened for both, small-range mutations and large genomic rearrangements. Screening for both types of mutations has only been reported in few studies performed previously [10, 26]. Yet another strength was that all data were generated at a single institution, therefore no variability was introduced by using different methods for tumor grading and IHC analysis and evaluation and the pathologist, who evaluated the ER, PR, and HER2 status, was blinded to the mutation status. Finally, the majority of study participants (73.4 %) were recruited within one year of disease presentation, which minimizes the likelihood of survival bias.

Conclusions

We found high prevalence and predominance of BRCA1 germline mutations in Pakistani women with TNBC compared to patients with non-TNBC presenting before or at age 30 irrespective of family history of breast/ovarian cancer and before or at age 50 with familial breast cancer or familial breast and ovarian cancer. The association between TNBC status and presence of BRCA1 mutations was independent of the simultaneous consideration of family phenotype, tumor histology, and tumor grade in a multiple logistic regression model. Our data suggest that TNBC status should be incorporated as a criterion for genetic BRCA1 testing in Pakistan. Identification of individuals with BRCA1 germline mutations will enable physicians to optimize cancer management for this high risk phenotype.