Introduction

Germline mutations of breast cancer susceptibility gene BRCA1 and BRCA2 (gBRCA1/2) are associated with an elevated lifetime risk of cancer in multiple organs including breast, ovary, colon, prostate, and pancreas [1,2,3]. Breast cancer in patients with gBRCA1/2 mutations accounts for 1–4% of all breast cancer, but the prevalence increases up to 8–30% in familial or early-onset breast cancer [4,5,6,7]. Especially in the Asian populations, gBRCA1/2-associated breast cancer are known to develop in younger age, with higher incidence of germline BRCA2 (gBRCA2) mutation than germline BRCA1 (gBRCA1) when compared to other ethnicities [6, 8, 9].

BRCA1 and BRCA2 proteins act as tumor suppressors that has distinct role in homologous recombination (HR)-mediated double-strand DNA break repair (DDR) [10, 11]. In response to DNA damage, BRCA1 and BRCA2 proteins interact with a numbers of other proteins including BARD1, PALB2, and RAD51 to maintain genomic integrity [12,13,14,15,16]. Tumors with deficiency or mutations in these genes, known as homologous recombination deficiency (HRD), are considered sensitive to poly(adenosine diphosphate [ADP]–ribose) polymerase (PARP) inhibitors that stall replication fork and lead to synthetic lethality [17].

HRD is identified as a potential prognostic and predictive biomarker across multiple cancer types [18,19,20,21], and currently, multiple methods to measure HRD are developed [22,23,24]. While these methods utilizing whole-genome and whole-exome sequencing techniques are expected to harbor in-depth information about the HR-related gene mutations, its application is still expensive, laborious, and time-consuming for most of the patients in clinical settings [25]. Alternatively, targeted sequencing provides genetic mutations and is useful to identify therapeutic biomarkers. In this study, we analyzed the targeted sequencing data of Korean breast cancer patients with gBRCA1/2 mutations to investigate the alterations in HR-related genes and their clinical implications.

Materials and methods

Study design and subjects

This is a retrospective cohort study of the breast cancer patients with pathogenic gBRCA1/2 mutations from October 2015 to December 2020 at Seoul National University Hospital, Seoul, Korea. The patients with breast cancer of age 20 years or older who had gBRCA1/2 mutation and SNUH FiRST cancer panel, a targeted next-generation sequencing (NGS) platform, were included. Pathologic diagnosis and immunohistochemistry (IHC) on estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status were confirmed by the pathologic reports of surgical and percutaneous biopsies. ER or PR ≥ 1% in IHC were considered hormone receptor-positive. HER2-positive was defined according to the manufacturer’s criteria and ASCO/CAP 2018 guideline.

Genetic sequencing

Genomic DNA was isolated from peripheral blood leukocytes and were tested for pathogenic gBRCA1/2 mutation by direct Sanger sequencing. Patients with equivocal variants and variants of unknown significance were excluded from analyses.

DNA was extracted from breast cancer and was subjected to targeted NGS platform named SNUH FiRST Cancer Panel. SNUH FiRST panel included 214 genes (version 3.0), 215 genes (version 3.1), and 216 genes (version 3.2) including microsatellite status with five microsatellite markers (D2S123, D5S346, D17S250, BAT25, and BAT26).

The NGS data with either mean coverage were less than 100X or the proportion of bases with coverage above 50X were less than 80% were regarded as disqualified and were excluded. Adaptor sequences and low-quality reads of targeted sequencing data were trimmed with fastp [26]. Trimmed reads were aligned with reference genome UCSC hg19, using BWA (version 0.7.17) [27]. Preprocessing was performed using MarkDuplicates, BaseRecalibrator, and ApplyBQSR function of GATK best practices (version 4.1.7.0) [28, 29]. After calling single nucleotide polymorphisms (SNP) and small insertions and deletions (INDEL) using Tumor-only mode Mutect2 in the preprocessed BAM file, the GATK FilterMutectCalls function was performed. For both SNP and INDEL, variants satisfying allele depth ≥ 3, total depth ≥ 10, minimum allele depth in both strands > 1, and variable allele frequency (VAF) > 0.2 remained. The strand bias test was conducted by using a Fisher’s exact test, leaving only variants that met the criteria of p value > 1 × 10–6 for SNP and p value > 1 × 10–20 for INDEL. Functional annotation of the variants was performed using ANNOVAR (version 20,191,024) [30]. Among variants located within the exon or splicing region, variants satisfying minor allele frequency (MAF) ≤ 0.01 in the population databases (Exome aggregation consortium—East Asian, 1000 Genomes project—East Asian, gnomAD—East Asian, NHLBI ESP6500) were used in the downstream analysis [31,32,33,34]. Among single nucleotide polymorphisms, silent mutations were excluded. The TP53 Database R20, July 2019 version was used [35]. R package maftools was used to draw the heatmap of HR-related gene mutations and lollipop plots for TP53 amino acid changes [36].

Following list of genes were considered as HR-related genes: ABL1, ATM, ATR, BARD1, BRCA1, BRCA2, CDKN1A, CDKN2A, CHEK1, CHEK2, FANCA, FANCD2, FANCG, FANCI, FANCL, KDR, MUTYH, PALB2, POLE, POLQ, RAD50, RAD51, RAD51D, RAD54L, and TP53 [11, 37]. Mismatch-repair (MMR) genes were defined as MLH1, MSH2, and MSH6.

Statistical analyses

Categorical variables were summarized with the frequencies in number and rates in percentages. Continuous variables were represented with the median values and ranges. Differences were assessed using Mann–Whitney test for continuous variables and Pearson’s χ2 or Fisher’s exact test for categorical variables. Multivariable Cox-proportional hazard models were constructed to find risk factors for survivals and correlation among covariates.

Overall survival (OS) was defined as the time from diagnosis of breast cancer to death of any cause. The OS curves were estimated using the Kaplan–Meier method. If patients survived without death, the survival was censored at the latest date of follow-up when no death was confirmed. Log-rank test p value < 0.05 was considered statistically significant. Data and statistical analyses were performed in R version 4.1.3 and RStudio version 2022.02.0 with R packages survminer and survival [38].

Results

Patient characteristics

Among 109 breast cancer patients with known gBRCA1/2 mutations and targeted NGS data, 7 patients who had silent mutations or variance of unknown significance were excluded. One patient was excluded because NGS was done from uterine adenosarcoma. Forty-six cases failed in quality assurance of NGS. Eventually, 55 patients with pathogenic gBRCA1/2 mutations with qualified NGS data were included in the study (Fig. 1).

Fig. 1
figure 1

Enrollment of the breast cancer patients with pathogenic gBRCA1/2 mutations

Full size image

All patients were of Korean ethnicity, and one male patient was included. The median age of all patients at the diagnosis of breast cancer was 42 years old, and majority of the patients were in premenopausal or perimenopausal state (39 of 55, 70.9%). Family history of breast cancer was found in 34 patients (61.8%). Thirteen patients (23.6%) had bilateral breast cancer, and 4 patients (7.3%) also suffered from ovarian cancer. Half of the patients (28 of 55, 50.9%) underwent prophylactic bilateral salpingo-oophorectomy, and 8 patients (14.5%) had prophylactic mastectomy. Most patients (50 of 55, 90.9%) had early breast cancer, and five de novo stage IV patients were included. The most common subtype was hormone receptor-positive, HER2-negative breast cancer (28 of 55, 50.9%), followed by triple-negative breast cancer (TNBC, 23 of 55, 41.8%), and hormone receptor-positive, HER2-positive breast cancer (2 of 55, 3.6%). Median Ki-67 value was 10.0. The baseline characteristics are shown in Table 1.

Table 1 Patient characteristics
Full size table

Pathogenic gBRCA1/2 mutation

There were 27 patients with pathogenic mutations in gBRCA1, and 28 patients with deleterious gBRCA2 mutations. The only male patient had nonsense mutation in gBRCA2 (p.Ile332PhefsTer17). While TNBC (n = 18, 66.7%) was significantly dominant in the patients with pathogenic gBRCA1 mutations, hormone receptor-positive breast cancer accounted for the 75% of the patients with pathogenic gBRCA2 mutations. In the gBRCA1 mutation group, patients had more TNBC compared to those in gBRCA2 mutation group (66.7% vs. 25.0%, p = 0.006). Median age, family history of breast cancer, prevalence of bilateral breast cancer were similar in both groups (Table 1).

In our study, 27 patients with pathogenic gBRCA1 mutations and 28 patients with gBRCA2 mutations were included. The most common variant was Val1833Ser (4 of 27, 14.8%), Glu1210Arg (3 of 27, 11.1%), and Tyr130Ter (3 of 27, 11.1%). Leu1780Pro, Lys307Ser, Trp1815Ter were also found in 7.4% of the patients, respectively. Among gBRCA2 mutations, Arg2494Ter (7 of 28, 25.0%) and Lys467Ter (4 of 28, 14.3%) were the most common. All mutations found in the patients are listed in Additional file 1: Table 1.

Of the 50 patients who were initially diagnosed as stage I-III, 23 patients (46.0%) experienced relapse. Three patients (6.0%) experienced local recurrence, and eleven patients (22.0%) suffered from recurrent or de novo early breast cancer in contralateral side. Two patients experienced distant metastases after local relapse, and eventually, eleven patients (22.0%) had distant metastases. Death occurred in 7 patients (12.7%). There was no difference of local or distant relapse rates between gBRCA1 and gBRCA2 mutants. Relapse-free survival of the stage I–III patients was not different between those with gBRCA1 and gBRCA2 mutations (median RFS 138 months vs. 112 months, p = 0.89). Overall survival of the patients with gBRCA1 mutation was also not significantly different from those with gBRCA2 mutation (median OS 290 months vs. not reached, p = 0.41) (Additional file 1: Fig. 1).

Targeted NGS and HR-related genes

Tissues were obtained from primary breast lesions, lymph nodes, lung, liver, and soft tissue for the targeted NGS. Most of the samples (45 of 55, 81.8%) were obtained from the breast primary lesion. About half (26 of 55, 47.3%) were obtained after lines of chemotherapy treatments. Median tumor proportion was 70%.

In the targeted NGS of 55 patients, 348 mutations were observed: 269 nonsynonymous single nucleotide variations (SNV), 30 nonsense mutations, 12 non-frameshift insertion or deletion, 4 non-frameshift substitutions, 25 frameshift insertions or deletions, and 8 splicings. There were 29 somatic BRCA2 mutations and 26 somatic BRCA1 mutations, including 3 cases without detected mutations in either BRCA1 or BRCA2 and 3 cases with mutations in both genes.

The most frequently co-mutated gene was TP53 (21 of 55, 38.1%). In the NGS analysis, mutations in TP53 gene included 6 frameshift insertion or deletion, 2 truncating mutations, and 13 nonsynonymous SNVs. Nonsynonymous SNVs mainly occurred in DNA-binding domain, while frameshift indels and stopgains occurred in oligomerization domain which interacts with other HR-related genes. Among the missense SNVs, five codons were located at DNA-binding grooves and two were at zinc binding sites (Fig. 2).

Fig. 2
figure 2

TP53 mutations in A gBRCA1 mutants and B gBRCA2 mutants

Full size image

Among the HR-related genes, the most frequently mutated genes following TP53 were POLE (7 of 55, 12.7%), ABL1 (4 of 55, 7.3%), and FA-related genes, including FANCA, FANCD2, and FANCI (4 of 55, 7.3%, respectively). ATM was found exclusively in 3 patients with gBRCA2. PALB2 was observed in one patient with gBRCA1 mutation. At least one missense mutation in HR-related genes was observed in 44 cases (80.0%). One patient with gBRCA1 mutation had somatic mutation in BER-related gene, MUTYH. Somatic mutations in the mismatch-repair genes (MLH1, MSH2, MSH6) were also observed in 5–7% of the patients with gBRCA1/2 mutations (Fig. 3).

Fig. 3
figure 3

Heatmap of HR-related gene mutations

Full size image

A total number of mutated genes were significantly higher in the tumors with TP53 mutations (mean 7.38 vs. 5.35, p = 0.003) (Fig. 4). In further analysis, only the number of the mutations in HR-related genes was significantly different (3.14 vs. 1.94, p < 0.001), but not that of the non-HR-related genes (4.24 vs. 3.41, p = 0.135) (Fig. 4, Additional file 1: Fig. 2). MMR genes were also not affected by the TP53 mutations (0.190 vs. 0.206, p = 0.901). Both gBRCA1- and gBRCA2-mutated tumors showed higher prevalence HR-related gene mutations in TP53 mutants compared to TP53 wild types (3.08 vs. 1.67, p = 0.003 in gBRCA1, 3.22 vs. 2.16, p = 0.041 in gBRCA2, respectively).

Fig. 4
figure 4

Number of mutated genes by TP53 mutation status A of all genes and B of HR-related genes

Full size image

Clinical significance of TP53 co-mutation

The patients who had both TP53 mutation and gBRCA1/2 mutation were significantly associated with TNBC (p = 0.028) and higher Ki-67 (p = 0.001). Among the patients who had gBRCA1, the median age at the diagnosis of breast cancer was 39.5 years in those with concurrent mutation in TP53, compared to 45 years without TP53. The patients also had significantly more premenopausal status and higher Ki-67 values (p = 0.044, 0.006, respectively). Contrastingly, in the patients who had gBRCA2 and TP53 co-mutations, the median age was 45 years compared to 39 years without mutations in TP53. In these patients, the incidence of bilateral breast cancer and node-negative diseases was significantly higher than that with wild type TP53 (p = 0.035 and 0.006, respectively) (Table 1).

Interestingly, these gBRCA2-related patients with TP53 co-mutation also showed superior overall survival to those without TP53 mutations (p = 0.011) (Fig. 5). After the exclusion of de novo stage IV breast cancer in the gBRCA2 group, the relapse-free survival was numerically longer in those with TP53 co-mutation compared to TP53 wild types. Mutation status of TP53 did not affect the survival of patients with gBRCA1 mutations (Additional file 1: Fig. 3). Unfortunately, due to small number of deaths and total cases, none of other factors including age, family history of breast cancer, TNM staging, PR-positivity, or Ki-67 was significant in Cox-proportional hazard models (data not shown).

Fig. 5
figure 5

Survivals of patients with gBRCA2mt by TP53 mutation status A overall survival and B relapse-free survival

Full size image

Discussion

Clinical and genetic characteristics of breast cancer patients with pathogenic gBRCA1/2 mutations were investigated in the present study. In consistence with previous reports, about half of our patients were diagnosed before age 40, and majority were premenopausal [4, 39]. These patients also presented with other high risk features, such as family history of breast cancer, co-occurrence of contralateral breast or ovarian cancer, and high Ki-67 index [5, 7, 40]. Carriers of gBRCA1 mutations had higher prevalence of triple-negative breast cancers (TNBC), while majority of gBRCA2 mutations were associated with hormone receptor-positive breast cancers [41, 42]. However, TNM staging, histologic grade of the tumor, and lymph node metastases were not statistically different between gBRCA1-mutated and gBRCA2-mutated patients.

Previously noted gBRCA1/2 mutations in Korean ethnicity were consistently detected in our patients. The most common mutation of BRCA2, p.Arg2494X, was also detected in 12–15% of Koreans with gBRCA1/2 mutations in other studies and suggested as a founder mutation [43, 44]. Among BRCA1 variants, p.Val1833Serfs, p.Tyr130Ter, Glu1210Argfs, and p.Trp1815Ter had been frequently observed also in Asian populations [44,45,46]. Moreover, there were two cases of BRCA1 p.Leu1780Pro, which was also recently identified as a novel pathogenic variant [47]. None of the known or suspected founder mutations of Ashkenazi Jews [48], Caucasians [49], North African [50], Hispanic [51], or Mexican populations [52] were observed. Interestingly, the Greek founder mutation BRCA1, p.Val1833Met and one of our common variants p.Val1833Serfs had different amino acid change in same location [53]. These results highlight the ethnical differences among gBRCA1/2 mutations.

While BRCA genes are known as the strongest drivers of the breast cancers, 35 of the 93 known driver mutations were also observed in these patients [54]. The most frequently co-mutated gene was TP53, which usually undergo missense mutations in DNA-binding domains and nonsense or deletions in other domains [40, 55]. In our study, only one-third of the TP53 mutations were found in major grooves or zinc binding site of the DNA-binding domain [56]. Of all, only ten cases were included in 73 codon hotspots defined by Walker et al. [57, 58], and two cases of TP53 p.R175H were observed in our patients among six well-known hotspot codons (R175, R213, G245, R248, R273, and R282) that account for a quarter of all TP53 mutations. All in all, the codon distribution and types of TP53 mutations of gBRCA1/2 mutants had discrepancy from those in known hotspots of sporadic breast cancers [59].

TP53 acts as a tumor suppressor gene, and its mutations were strongly associated with increased chromosomal instability and higher HRD score [60]. In patients with TP53 mutations in addition to gBRCA1/2 mutations, the total number of mutated genes, especially of HR-related genes, increased. In gBRCA1 mutants, mutation of non-HR-related genes also increased, probably due to its tendency toward more frequent structural rearrangements than gBRCA2 [61].

TP53 mutations are associated with breast cancer with younger age, higher grade, advanced stages, hormone receptor negativity, enrichment of mutational signature 3, and with a high HRD index [54, 62]. In our study, the tumors with TP53 mutation were also more likely to be TNBC and have high Ki-67 values. On the other hand, age and staging at initial diagnosis of the patients with TP53 mutations were not significantly different from those with TP53 wild types. High rate of TP53 mutations observed from younger ages and early stages of the patients with gBRCA1/2 mutations implied the effect of DNA-repair deficiency and its selective pressure on tumor suppressor genes [57, 63].

The role of TP53 variants in the breast patients with pathogenic gBRCA1/2 mutation had been controversial. In our study, the improvement of overall survival was observed only in the patients with gBRCA2 mutations, despite the similar pathological complete remission rates after neoadjuvant chemotherapies, relapse rates, and de novo stage IV diseases. Those with wild-type TP53 and gBRCA2 mutation had numerically higher rate of distant metastases and deaths. As for subtypes of breast cancer, TP53 mutation has been reported to be more prevalent in basal-like subtypes [63,64,65]. Consistently, tumors with TP53 mutants (44.4%) were more likely to be ER-negative than those with wild-type TP53 (15.8%) in the present study. Previous study showed that, Asians were more likely to have TP53 mutations among ER-positive breast cancers than Caucasians, and TP53 mutations were associated with poor survival in ER-positive breast cancer [63]. Another study has shown that the neoadjuvant chemotherapies, regardless of the pathological complete remission rates, were more effective in the patients with TP53 mutations than wild types [66]. Taken together, one possible explanation is that initially luminal-like gBRCA2-mutated breast cancer are affected by TP53 co-mutation to become closer to basal-like entities, and more sensitive to neoadjuvant and adjuvant chemotherapy that led to less distant metastases and deaths.

Our study has limitations that arise from the retrospective design utilizing the targeted NGS for clinical purpose. Genetic mutations not targeted in the panels were difficult to be observed, including large deletions, rearrangements, chromosomal abnormalities, and methylations. Moreover, targeted NGS was done for clinical purposes and the paired biopsy with non-neoplastic tissues was difficult to be done. Small number of the patients due to the rarity of gBRCA1/2 mutations were another limitations in obtaining statistical significance. Still, our study holds its value in delineating the rare breast cancer entity with gBRCA1/2 and concomitant somatic mutations, especially in the Asian populations.

Conclusion

Our study showed genetic heterogeneity of pathogenic mutations in gBRCA1 and gBRCA2 in the Korean populations. Patterns of TP53 mutations in concomitant gBRCA1/2 mutations were distinct from those in sporadic breast cancers. Co-mutation of gBRCA1/2 and TP53 genes was associated with TNBC, high Ki-67, and higher number of mutated genes related to HR pathways. Further studies are needed to clarify the association between genetic diversity among different ethnic groups and clinical circumstances to develop treatment strategies that could lead to better survivals.