Background

Hereditary breast cancers account for approximately 10% of all breast cancers, and approximately 23% of all ovarian cancers are considered hereditary [1, 2]. According to Plakhins et al., BRCA1 pathogenic founder mutations (c.4035delA, c.5266dupC) contribute to 3.77% of all consecutive primary breast cancers and 9.9% of all consecutive primary ovarian cancers [3]. BRCA1 and BRCA2 pathogenic founder mutation analysis is a relatively straightforward and cost-effective screening strategy to identify mutation carriers [4]. In Latvia, all consecutive breast and ovarian cancer cases are eligible for BRCA1 pathogenic founder mutations (c.181 T > G, c.4035delA, c.5266dupC) screening [5], and the costs of the test are covered by the public health care system. However, according to recent studies, non-founder BRCA1 and BRCA2 pathogenic mutations account for up to 21.6% of all BRCA1 and BRCA2 pathogenic mutations in the Aschkenazi Jewish population [6, 7]. There is little information about pathogenic BRCA1/2 non-founder mutations in Latvia. In a study published by Berzina et al., pathogenic non-founder mutations in BRCA1 and BRCA2 were identified in 4 out of 30 high-risk breast/ovarian cancer families from the Latvian population [8]. In another study published by Tihomirova et al., non-founder pathogenic mutations in BRCA1 and BRCA2 were detected in 9 out of 160 patients with breast and ovarian cancer [5]. These findings suggest that the proportion of pathogenic BRCA1/2 non-founder mutations is small and that family cancer history alone is of limited value to find subgroups of individuals, where expensive complete BRCA1/2 testing is indicated.

The remaining hereditary breast and ovarian cancer cases are associated with mutations in other breast and ovarian cancer susceptibility genes, such as BRCA1/2, TP53, PTEN, CDH1, STK11, MLH1, MSH2, MSH6, PMS2, PALB2, CHEK2, ATM, RAD51C, RAD51D, BRIP1 and other [9]. Patients and their relatives harbouring mutations in hereditary cancer predisposing genes could benefit prevention and screening strategies or novel therapeutic approaches [10, 11]. Advances in next-generation sequencing allowed the implementation of low-cost multi-gene panel testing in clinical practice to detect pathogenic mutations in hereditary cancer predisposing genes [12].

Therefore, knowledge of the frequency and phenotypical features of pathogenic mutations beyond BRCA1 pathogenic founder mutations in breast and ovarian cancer susceptibility genes is essential for determining the role of second-line testing with multi-gene panels in counselling unsolved high-risk breast and ovarian cancer patients.

The aim of the study was to evaluate the rate of pathogenic mutations in the 26 breast and ovarian cancer susceptibility genes in patients who meet the criteria for BRCA1/2 testing and to compare the accuracy of different selection criteria for second-line testing in a founder population.

Methods

Patient group

Sixteen sequential patients with primary breast and/or ovarian cancer who met all inclusion criteria were included in the study between October 2016 and August 2017. The inclusion criteria were as follows: 1) fulfil at least one of the National Comprehensive Cancer network (NCCN) BRCA1/2 testing criteria (Table 1) (www.nccn.org); 2) previously tested negative for BRCA1 pathogenic founder mutations (c.181 T > G, c.4035delA, c.5266dupC); 3) able to cover the cost of the 26 multi-gene tests.

Table 1 NCCN selection criteria for screening of mutations in BRCA1and BRCA2
Full size table

The following clinical information was obtained: age at testing, personal cancer history, age at cancer diagnosis, breast and/or ovarian cancer pathology, BRCA1/2 testing history, a family cancer history that covers a 3-generation pedigree according to probands information. The median patient age was 45.6 years (33–63 years). Fifteen out of 16 (93.75%) patients were females, and 1 out of 16 (6.25%) patients was male. Thirteen patients had unilateral breast cancer, 1 patient had bilateral breast cancer, 1 patient had ovarian cancer, and in 1 patient had both breast and ovarian cancer. Four out of 16 (25%) breast cancers were luminal-like HER2 negative, 2 out of 16 (12.5%) breast cancers were luminal B HER2 positive, 8 out of 16 (50%) breast cancers were triple-negative, and 1 out of 16 (6.25%) breast cancers was HER2 positive. The patient characteristics are summarized in Table 2.

Table 2 The baseline characteristics of patient group
Full size table

DNA testing

Informed consent for genetic testing was obtained for all patients. All patients underwent DNA testing with a 26-gene panel (myBRCA HiRisk Hereditary Breast and Ovarian Cancer screening Test, VeritasGenetics, USA) that is a targeted next-generation sequencing assay for the detection of mutations in 26 breast and ovarian cancer susceptibility genes. The genes included high-penetrance breast-ovarian genes (BRCA1, BRCA2, PTEN, TP53, CDH1, STK11, PALB2), moderate-penetrance breast and/or ovarian genes (CHEK2, BRIP1, ATM), and additional genes (BARD1, BLM, EPCAM, RAD50, RAD51C, RAD51D, MEN1, MRE11A, MUTYH MSH2, MLH1, NBN, MSH6, PMS2, FAM175A, XRCC2). In all patients, the test was performed using saliva. The specificity and sensitivity of the assay are 99.9% for point mutations and small insertions/deletions in the 24 sequenced genes and 99.9% for structural variations in BRCA1 and BRCA2.

Statistical analysis

The specificity, sensitivity and accuracy of the NCCN criteria, Manchester scoring system and Swedish Breast cancer group criteria for the prediction of pathogenic non-founder mutations were evaluated. The Manchester score of 15 points threshold was used to assess the likelihood of BRCA1/2 pathogenic mutation [13]. The specificity, sensitivity and accuracy of different selection criteria for BRCA1/2 testing in our cohort were calculated using MedCalc Statistical Software version 17.9.

Results

In seven out of sixteen (44%) patients included, pathogenic non-founder BRCA1/2 mutations were identified. Six patients carried pathogenic variants of BRCA1 and one of BRCA2. In four patients, variants of uncertain significance of BRCA2, RAD50, MRE11A and CDH1 were found. Detailed results are shown in Table 3. The NCCN criteria showed a high sensitivity (100%) with low specificity (50%) for the prediction of non-founder pathogenic BRCA1/2 mutations. The Swedish Breast cancer group criteria showed a low sensitivity (57.1%) with three false negative results. The Manchester scoring system showed a high accuracy (87.5%) for the prediction of pathogenic non-founder BRCA1/2 mutations with high sensitivity (85.7%) and specificity (88.9%). The sensitivity, specificity and accuracy of different criteria/scoring systems for the detection of probability of BRCA1/2 pathogenic mutations in our cohort are compared in Table 4.

Table 3 Results
Full size table
Table 4 Comparison of different selection criteria for BRCA1/2 testing in our cohort
Full size table

Discussion

Our study is the first report on the use of a 26 gene panel in to examine breast and ovarian cancer susceptibility genes in patients in Latvia. We demonstrated a high frequency of pathogenic non-founder germline mutations in BRCA1 and BRCA2 genes. In seven out of sixteen (44%) primary breast and ovarian cancer patients matching the criteria for BRCA1/2 testing pathogenic non-founder BRCA1/2 mutations were identified. All 7 pathogenic mutations, including 2 large deletions, are novel in populations of Latvia [5, 8]. These results may suggest that the present practice of testing only the 3 most frequent BRCA1 pathogenic founder mutations is insufficient and fails to detect a considerable number of pathogenic mutations in BRCA1/2. However, our study comprises a relatively small cohort of selected patients. In a study published by Frank et al., 21.6% of patients with Ashkenazi ancestry pathogenic non-founder BRCA1 and BRCA2 mutations were identified [6]. In contrast, in the Finnish population of high-risk individuals tested negative for 28 BRCA1/2 pathogenic founder mutations, additional pathogenic mutations in BRCA1 and BRCA2 accounted for just 1.2% [12]. Much larger numbers are necessary to assess the real proportion of pathogenic non-founder mutations in the population of Latvia.

Despite the drawbacks of such a small study group, the initial results raised some observations.

Interestingly, probands that carried a pathogenic non-founder mutation had some common features. All six breast cancer patients in our study with proven pathogenic non-founder BRCA1/2 mutations had a triple-negative phenotype. It is well established that approximately 80% of all BRCA1/2– related tumours have a triple-negative phenotype [14,15,16,17,18]. The prevalence of pathogenic germline BRCA1/2 mutations in the selected triple-negative breast cancer patients ranged from 9.2 to 34.4% [19,20,21,22]. Additional analyses of cDNA microarray data from van’t Veer showed that BRCA1-related tumours have a sporadic basal-like breast cancer gene expression profile [23]. Additionally, according to Richardson et al., loss of BRCA1 function could play a role in the development of basal-like breast cancers [24]. Couch et al. identified BRCA1/2 pathogenic mutations in 11.2% of triple-negative breast cancer patients and other breast-ovarian cancer predisposing gene mutations in 3.7% of triple-negative breast cancer patients [25].

In our study we used the NCCN criteria for screening pathogenic mutations in BRCA1 and BRCA2, where triple-negative breast cancer is used as a criterion together with an age limit < 60. Only one out of six breast cancer patients in our study who carried a pathogenic BRCA1/2 non-founder mutation was older than 60 years of age, but in this case, family cancer history was positive in the study published by Couch et al., 3.1% of triple-negative breast cancer patients older than 60 years and only 1.4% with no family history of breast or ovarian cancer were diagnosed with BRCA1/2 pathogenic mutation [25]. Therefore, our study results support the current NCCN guidelines for screening all triple-negative breast cancer patients younger than 60 years of age.

In contrast, the application of the upper age limit for triple-negative breast cancer patients of 40 years (Swedish Breast cancer group criteria for screening for mutations in BRCA1 and BRCA2) would miss several BRCA-positive cases in our cohort [26].

Our small study showed the high accuracy of the Manchester scoring system for the prediction of pathogenic non-founder BRCA1/2 mutations in founder mutation-negative patients. Our finding is supported by several other studies performed on the validation of the Manchester scoring system in populations of UK, Germany and South East Asia [13, 27, 28]. However, larger numbers of cases are needed for comprehensive validation of these criteria in the population of Latvia.

Additionally, three out of eight patients tested negative for 26 breast and ovarian cancer susceptibility genes were HER2 positive. According to a recently published study, only 9% of BRCA1-related breast tumours and 13% of BRCA2-related breast tumours were HER2 positive [29]. HER2 positivity is also included in the Manchester scoring system as a BRCA1/2 probability decreasing factor [13].

Ovarian cancer in a personal or family history was documented in three out of seven patients who carried a pathogenic BRCA1/2 non-founder mutation. Additionally, in one case, unknown gynaecological cancer was reported in a paternal aunt. According to recent studies, the presence of ovarian cancer in personal or family history of pathogenic BRCA1 founder-negative breast cancer patients increases the possibility of carrying previously undetected pathogenic BRCA1/2 non-founder mutations [30, 31]. Recently, in a study published by Couch et al., ovarian cancer in family history was documented only in 1 of 54 pathogenic non-BRCA1/2 mutation carriers with triple-negative breast cancer [25].

In our study, no pathogenic mutations were detected in another 24 genes included in the panel. Some previously published studies demonstrated that the rate of pathogenic mutations in non-BRCA1/2 genes ranged from 2.9 to 9.3% [32,33,34,35].

Four of the 16 (25%) patients were identified to have a variant of unknown significance (VUS) in BRCA2, RAD50, CDH1 and MRE11. Unfortunately, due to an insufficient sample size in our study, we cannot elaborate upon those results.

Conclusion

A relatively high incidence of pathogenic non-founder BRCA1/2 mutations was observed among patients with triple-negative familial breast cancer in a founder population. The Manchester scoring system predicted the probability of non-founder pathogenic mutations with high accuracy.