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).

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

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.

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).

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).

TP53 mutations in A gBRCA1 mutants and B gBRCA2 mutants

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).

Heatmap of HR-related gene mutations

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).

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

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).

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