Background

Breast cancer is a common malignancy among women, with an estimated annual rate of incidence increasing by 2–3% in China, especially in metropolitan areas [1]. It is known that while the majority of breast cancers are sporadic in origin, an appreciable fraction result from inherited causative variants [2, 3] . Cancer is caused by the cumulative effects of mutations in multiple genes, in combination with environmental factors. It has been suggested that reproductive and hormonal factors, such as nulliparity, increased age at first live birth, and limited breast feeding are associated with a modest increase in the risk of breast cancer in Western countries [4, 5]. Breast cancer susceptibility genes BRCA1 and BRCA2 causative variant account for only 10–20% of breast cancers with a known family history [6]. The prevalence of hereditary breast cancers is approximately 11.8% in China [7], suggesting that other genes may play an important role in increasing the susceptibility to breast cancer, albeit at a markedly lower frequency and penetrance. For example, women with inherited causative variant in the Fanconi anemia genes BRIP1 and PALB2 have a 20–50% lifetime risk of breast cancer [8, 9]. Multiple studies have also demonstrated that genes such as ATM [10,11,12] and CHEK2 [13,14,15,16] are associated with increased breast cancer risk. In addition, inherited causative variants in TP53, PTEN, STK11, and CDH1 are associated with a moderate to very high-risk of developing breast cancer [17,18,19,20].

Although studies have demonstrated the clinical benefit of multiple-gene sequencing for the assessment of patients with high-risk hereditary cancer [21, 22], little information is currently available regarding the value of multiple-gene sequencing for the assessment of the risk of hereditary breast cancer in China. The goal of this study was to identify the variant spectrum for the clinical and genetic characterization of familial breast cancer in a Chinese population. Twenty-seven breast cancer susceptibility genes (Additional file 1: Table S1), selected through a database (HGMD: Human Gene Mutation Database, NCBI ClinVar database) and published research articles, were tested by Next-Generation Sequencing (NGS).

Methods

Patients and samples

In total, 240 participants, including 120 patients with breast cancer and 120 high-risk women with first- or second-degree relative(s) suffering from breast cancer were recruited from Shenzhen Second People’s Hospital of China during a two year period (2014–2016). The rate of susceptibility gene causative variants in East Asian population in 1000 Genomes database was used as a control. The clinical breast cancer diagnosis and classification criteria were in accordance with the World Health Organization criteria. Written informed consent was obtained from patients and healthy high-risk women. The study was approved by a local ethics committee. Two hundred and forty peripheral blood samples were collected and referred for genetic testing to the BGI research Department (Shenzhen, China).

Sample treatment, next-generation sequencing and variants calling

DNA was extracted from participants’ peripheral blood samples using a Qiagen DNA blood mini kit (Qiagen, Hilden, Germany) according to the manufacturer’s recommendations. Qubit Fluorometer (Life Technologies) and agarose gel electrophoresis were used to determine DNA concentration and purity. Genomic DNAs were randomly fragmented to 200-300 bp by Covaris E210 (Massachusetts, USA) and treated as follows: end-repair, A-tailing and adapter ligation, and PCR amplification. PCR products were captured by the same BGI chip in the Blackbird platform. Their frequency was determined by quantitative PCR, and the segments were pooled for sequencing on the Hiseq 2500 (Illumina) according to the manufacturer’s protocols. Over 0.6 GB data was generated per sample with approximately 200X depth and over 99% coverage of the target region. Variants were detected using Small Variant Assembler Methods (http://www.completegenomics.com/documents/Small_Variant_Assembler_Methods.pdf) which is available on the official website of Complete Genomics. Then, variants were filtered according to their read support, assemble quality and reference allele repeat status. Sequences generated by high-throughput sequencing platforms were filtered by SOAPnuke1.5.0 with standard augmentation, and then assembled by BWA 0.7.12 using MEM. Sam Tools 1.2 was used to convert file format into BAM. Base quality was recalibrated by GATK 3.4. Duplications were removed by Picard Mark Duplicates 1.138. Local realignment of reads around insertion/deletion was performed and variants were called by insertion/deletion Realigner and Haplotype Caller in GATK 3.4. Variants were further filtered by quality depth, strand bias, mapping quality and reads position.

Variant classification

In accordance with the American College of Medical Genetics (ACMG) recommendations for the interpretation of sequence, variants were classified into pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign, and benign variant. Variants were classified as pathogenic if they conferred truncations, or initiation codons, affected splicing or if they have been reported in the central mutation database (HGMD, ClinVar), or in published literature, and demonstrated to be causative of the disorder in a particular disease with no conflicting results. Variants were classified as VUS if they fulfilled the following three criteria at the same time: 1) missense, non-frame shift or intronic (exon-intron boundaries ±10 bp) variants, and 2) allele frequency in the 1000 Genomes Study and 101 BGI normal Chinese genomes study are both less than 0.03, and 3) variants were not uniformly identified as benign/likely benign in ClinVar. The rest of variants were identified as benign. In addition, every pathogenic variant detected by next-generation sequencing was confirmed by conventional PCR-Sanger sequencing. Twenty-seven genes examined in this study (Additional file 1: Table S1) were selected through database or published articles about known mutations in hereditary breast cancer.

Statistics

Statistical tests were carried out using SPSS 20.0 (IBM, Armonk, NY), applying chi-square or Fisher’s exact tests when required to analyze categorical data. A p values less than 0.05 was considered as statistically significant.

Results

Characteristics of the study population

We recruited for this trial 120 patients diagnosed with breast cancer and 120 high-risk women who had first-degree relatives affected by breast cancer. Table 1 summarizes the risk factor data of the study population reflecting the epidemiology of breast cancer. The median age at blood sample collection was 46 years (range from 25 to 81 years) in the breast cancer group and the median age was 37 years in the high-risk group. There were no statistically significant differences in body mass index (BMI), age at menarche, and breast-feeding history. However, there were statistically significant differences between the two groups in parity and abortion rates. In this study 77.5% of patients had no history of childbearing and 41.7% of patients had a history of abortion, which may confer a high-risk of breast cancer in Chinese individuals.

Table 1 Epidemiological characteristics of the study participants
Full size table

Prevalence of panel-gene causative variants in the two groups

In order to explore the presence of predisposing genetic factors for the development of breast cancer, all participants were subjected to a multiple-gene panel sequencing and variant analysis. The presence of 27 causative variants (Additional file 1: Table S1) associated with an increased susceptibility to breast cancer was tested in this panel using NGS. As showed in Table 2, the ratio of variants in the breast cancer group was 19.2% (23/120) and 12.5% (15/120) in the high-risk group. Twelve predisposing causative variants in 27 panel-genes were identified in this study. Three (2.5%) in BRCA1, nine (7.5%) in BRCA2, two (1.7%) each in MUTYH, CHEK and PALB2, one (0.8%) each in ATM, BARD1, NBN, RAD51C, TP53 were identified in the breast cancer group, while seven (5.8%) in MUTYH, three (2.5%) in ATM, two (1.7%) in BRCA2, one (0.8%) each in BARD1, BRIP1 and CDH1 were identified in the high-risk group. There were no causative variants found in other genes examined.

Table 2 Distribution of multiple-gene variants in two groups of 240 participants
Full size table

All germline changes revealed by panel sequencing were termed germ line causative variants by the 5-tier rating system. We have excluded “likely benign”, “benign” variants and VUS in the paper, and have listed “pathogenic”, “likely pathogenic,” changes in Tables 3 and 4. Detailed information regarding causative variants in the breast cancer group and the high-risk group (women with a family history of breast cancer) is listed in Tables 3 and in Table 4. Genetic causative variants identified were heterozygous mutations, and most were frameshift deletions. We did not include healthy women with no known history of familial breast cancer in our study, however frequencies of gene causative variants that we identified were examined in healthy population by surveying available databases: http://www.internationalgenome.org/ and http://www.ncbi.nlm.nih.gov/projects/SNP/. We found that the frequencies of these variants were zero in East Asian population in 1000G_ALL (the frequency of this causative variants in all populations of the human international genome). However, we detected MUTYH gene variants (Intron10, c.892-2A > G) at a rate of 2.77% https://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=77542170 in East Asian healthy individuals.

Table 3 Causative variants identified in the high-risk healthy people
Full size table
Table 4 Causative variants identified in patients with BC
Full size table

Association between genetic causative variants and clinicopathological characteristics

Gene causative variants prevalence was 69.6% (16/23) in patients with invasive ductal carcinoma (IDC), 4.3% (1/23) patients with ductal carcinoma in situ (DCIS) and 26.1% (6/23) with an unknown histological type (Table 5). There was no significant difference in the presence of variants between clinical stages of breast cancer (Pearson’s Chi-squared test p = 0.537). Although some patients were lost to follow-up, our data suggest that similar causative variants were found in patients regardless of tumor size and lymph nodes status.

Table 5 Comparison of patients with and without a pathogenic variant
Full size table

When analyzed, based on the molecular subtype of breast cancer, the genetic causative variant ratio was 43.5% in patients with triple negative breast cancer (TNBC), 39.1% in patients with non-TNBC, and 17.4% in patients with undetermined molecular subtype (p = 0.001) (Table 5). Eight of the 12 BRCA1/2 causative variants were found in the TNBC group. The other two gene variants in the TNBC group were BARD1 and RAD51.

Discussion

In this clinical study, we examined 27 genes associated with an increased susceptibility to breast cancer (Tables 2, 3 and 4) in patients with breast cancer and in high-risk participants with a family history of breast cancer. In addition to BRCA1/2, genes with an established role in breast cancer, other predisposing genes such as CHEK and PALB2 were evaluated for a possible association with the risk of breast cancer, although their frequency and penetrance was significantly lower. We found causative variants in 12 of the 27 genes examined in the participants (Table 2).

There appeared to be considerable discrepancies in the causative variant rates of BRCA1 and BRCA2 in breast cancer patients in different areas of China. Song [23] reported that the variant ratio of BRCA1 and BRCA2 in Shanghai was 11.4% and 2.9%, respectively, whereas in our study the variant ratio of BRCA1 and BRCA2 in breast cancer patients were 2.5% and 7.5%, respectively (Table 2). The main reason for lower causative variant rates of BRCA1 and higher variant rates of BRCA2 in our study may be the different detection methods used in the studies. PCR-SSCP analysis, examining only four “hot areas” in BRCA1/2 was used in the Song’s study, while whole exon NGS of BRCAs was used in our study. In addition, geographical differences are likely to contribute to discrepancies between results. The participants in the Song’s study mainly were recruited from Eastern and Northern China, while the subjects in our study were largely from Southern and Central China.

We found a relatively high variant rate (4.2%, 5/120) of MUTYH c.892-2A > G in the high-risk group, but lower rate (0.8%, 1/120) in the breast cancer group (Table 2). According to the 5-tier rating system in ACMG, this variant is likely pathogenic. However, a correlation between MUTYH variants and breast cancer remains unclear. For example, two other studies suggested a significantly increased breast cancer risk among carriers of the bi-allelic MUTYH variants [24, 25], while other studies showed that germline MUTYH variants are not associated with carcinomas of the breast [26, 27] . In our study, the variant ratio of MUTYH c.892-2A > G in high-risk women with a family history of breast cancer is over 2.77% https://www.ncbi.nlm.nih.gov/projects/SNP/snp_ref.cgi?rs=77542170, the frequency of MUTYH c.892-2A > G in East Asians in 1000G_ALL, but the rate in the breast cancer group is lower. The variant MUTYH c.892-2A > G identified in our study is a heterozygous mutation (Tables 3 and 4). Further, two family pedigrees suggests segregation of this variant (Fig. 1) - the proband did not carry the variant, while their relatives with no BC carried it. Therefore, it is possible that MUTYH c.892-2A > G is a benign variant in the development of BC in East Asians, however we need to enlarge the sample size to confirm this result.

Fig. 1
figure 1

Pedigree maps of two families. stands for MUTYH c.892-2A > G variants. stands for that MUTYH c.892-2A > G variant was not tested. stands for man. stands for woman. and stand for non-cancerous death. stands for patient with breast cancer. The black arrow indicates the proband. (BC: Breast cancer)

Full size image

To explore the relationship between gene variants associated with hereditary predisposition and tumor characteristics, we analyzed the association between available pathological and clinical data in breast cancer patients and the presence of gene causative variants. Our results show no statistically significant differences between the presence of gene variants in breast cancer patients and differences in tumor histology, size, clinical stage and lymph node status, however; we found a statistically significant difference in the variant rate in patients with tumors of different molecular type (Table 5). Ten of 22 patients with TNBC were found to harbor gene causative variants. Furthermore, most of TNBC patients (8/10) were found to have BRCA1/2 causative variants. It has been reported that TNBC is common in BRCAs variant carriers [28,29,30,31]. Indeed, the incidence of TNBC is around 70% in BRCA1 mutation carriers [32, 33]. Our data are consistent with this observation, however we need to enlarge the sample size to further confirm this association.

As for the clinical significance of the presence of predisposing variants, different advice may be given to specific groups of patients. Patients carrying these pathogenic variants are considered to be at a high-risk in developing tumor recurrence or secondary cancer according to the NCCN guidelines [9, 34]. However, contralateral mastectomy or oophorectomy for these patients is currently not recommended in China, and asymptomatic women carrying pathogenic variants usually prefer not to undergo preventive surgery. In light of this situation, we suggest that patients with a high-risk of developing breast cancer have a comprehensive physical exam every six months, and we advise them to focus on breast self-examination and maintain a healthy life style.

Conclusion

As the incidence of breast cancer is increasing, it is necessary to carry out more studies to identify susceptibility genes of breast cancer and to establish their frequency. Our results enrich our knowledge of predisposing variants in the population of Southern and Central China, and provide some experimental data for the identification of alternative susceptibility genes, and for the establishment of a clinical model of genetic screening.

However, our study also has some limitations. We did not analyze the relationship between clinicopathological characteristics and gene VUS. More than two hundred VUS were identified in this study, but we have not analyzed them to date. In addition, some patients were lost due to follow-up, which made it difficult to draw conclusions between the association of genetic causative variants and clinicopathological characteristics of patients.