Borderline ovarian tumors (BOTs) represent a heterogeneous group of noninvasive tumors of uncertain malignant potential. They have characteristic histology and usually occur in younger women, before 40–45 years of age. The majority of BOT is diagnosed at the early stage and has a favorable prognosis [1,2,3]. However, the symptomatic recurrence rate varies between 3% [26] and 34% [27] and is associated with decreased survival [28]. The molecular changes in BOTs indicate linkage of this disease to type I ovarian tumors (low-grade ovarian carcinomas) [2, 3, 29, 30]. Recently, we found that pathogenic founder/recurrent germline mutations in 4 genes (BRCA1, BRCA2, RAD51C, PALB2) are responsible for 12.5% of ovarian cancer cases among unselected patients in the Polish population [5]. In the present study, we aimed to estimate the prevalence of 21 founder/recurrent germline mutations in 5 genes (BRCA1, BRCA2, RAD51C, PALB2, and CHEK2) among Polish patients with BOTs and to assess an association of these mutations with the risk of development of BOTs. Because of morphological similarities of borderline tumors and low-grade ovarian cancer [2, 3], we also included in the analysis 167 patients with ovarian cancer G1. Pathogenic mutations were detected in 12.74% of BOT (13 out of 102 cases), in 12.57% of ovarian cancer cases (21 out of 167 cases), and in 6.71% controls (117 out of 1743 women).

In the group of BOT patients, the most common was the missense mutation (c.470T>C) in the CHEK2 gene. The frequency of this mutation was higher in BOTs than in controls (10.8% vs. 5.6%), which corresponded to a statistically significant 2-times elevated risk of BOT (OR = 2.05, p = 0.03). The prevalence of other tested recurrent mutations in BRCA1, BRCA2, RAD51C, and PALB2 and two protein-truncating mutations in CHEK2 among BOT patients was very low and was insufficient to perform a reliable association analysis (Table 3).

In a group of ovarian cancer G1 patients, statistically significant association with cancer risk was detected for mutations in BRCA1 (OR = 8.53, p = 0.005) and PALB2 (OR = 7.03, p = 0.03) genes; however, it should be noted that the number of mutation carriers was very limited (4 cases with BRCA1 and 2 cases with PALB2). The CHEK2 mutations, including missense and two protein-truncating, were not significantly associated with cancer risk (Table 3).

The CHEK2 missense mutation (c.470T>C) was previously analyzed in another study of 539 ovarian cystadenomas, 122borderline ovarian tumors, and 447 ovarian cancer cases, including 88 low-grade (G1) tumors [15]. In that study, the significant association of CHEK2 missense mutation (c.470T>C) with the risk of non-invasive tumors (OR = 1.7, p = 0.005 for ovarian cystadenoma and OR= 2.6, p = 0.002 for BOTs) and borderline significant correlation with low-grade ovarian cancer (OR=2.1, p = 0.04) were reported. In our study, the association of CHEK2 missense mutation (c.470T>C) was statistically significant for BOT risk, but not for the risk of ovarian cancer G1 (Table 3). The possible explanation of the discrepancy in the above results may be the difference in a number of tested low-grade ovarian cancer cases, which was almost 2-times larger in our study (167 cases) than in previous analysis (88 cases). The mutations in the CHEK2 gene were already reported to correlate with the risk of prostate and breast cancer [20, 21, 31], but not with ovarian cancer [5, 21, 32]. The results of this study seem to support this observation and may extend it, suggesting a possible association of CHEK2 missense (c.470T>C) mutation with the 2-times increased risk of BOT.

Mutations in BRCA1, BRCA2, RAD51C, and PALB2 genes were already shown to correlate with the risk of ovarian cancer [5,6,7,8,9], but not with BOTs [10,11,12,13,14, 33]. The two studies conducted on the Jewish population (including 117 BOT and 161 ovarian cancer cases, as well as 46 BOT and 59 ovarian cancer cases, respectively) showed a much lower incidence of thfoundersnder BRCA1/2 mutations in BOT patients than in invasive early-stage ovarian carcinoma patients—the prevalence varied between 2.2 and 4.3% in BOT cases and 24.2–32% in ovarian cancer cases [10, 11]. In other studies, from Norway (190 BOT and 478 ovarian cancer patients) and Canada (134 BOT and 515 ovarian cancer patients), the BRCA1/2 mutations were detected in 4% and 11.7% invasive cases, respectively, but none of the patients with BOT [12, 13]. There were performed several other smaller studies, in which only single cases of BRCA1/2 mutations among BOT patients were detected, and the cumulative prevalence in all tested patients was 1.3% for BRCA1 and 0.2% for BRCA2 genes [33]. In contrary to the reports mentioned above, in a study of 1333 Czech ovarian cancer patients and 152 borderline ovarian tumor cases recruited from seven centers, the prevalence of BRCA1/2 mutations was similar in high-grade ovarian cancers and BOT cases (30.9% and 28.9%) [14]. There is no obvious explanation for such a high frequency of BRCA1/2 mutations among patients with BOT. It may be the consequence of several factors, such as differences in sample size or population-related differences. It should be also considered that pathological differentiation of BOTs from invasive tumors is not easy and may differ, especially when evaluation is done by pathologists from different centers [34, 35]. In our study, the 102 patients with histologically confirmed BOTs were diagnosed in one hospital, and the histology was evaluated and reviewed by 3 independent pathologists, which indicates a high-quality assurance of pathology diagnosis.

In our analysis, the mean age at diagnosis was lower among BOT patients than among low-grade ovarian cancer patients (47.76 vs. 54.25; Table 1), which is consistent with reported data [3, 36, 37]. The occurrence of any detected mutation in BRCA1, BRCA2, RAD51C, PALB2, and CHEK2 genes was in general associated with earlier age of diagnosis of BOT, when compared to non-carriers (45 years vs. 49 years) and was 10 years younger in carriers of CHEK2 missense mutation (c.470T>C) than in BOT patients without mutation (38 years vs. 49 years). This observation is consistent with a previous study which also reported a correlation of CHEK2 missense (c.470T>C) mutation with earlier age of diagnosis among BOT patients [15].

The all-cause survival, as it was expected, was better among BOT patients than ovarian cancer cases (5-year survival rate: 90.48 vs. 71.32% and 10-year survival rate: 88.89 vs. 59.36%), which is consistent with other studies reporting favorable prognosis for patients with borderline ovarian tumors [4, 28, 38,39,40]. There was no significant association of mutation status among BOT patients and death. However, the 10-year survival rate was lower for mutation carriers (80%) than for non-carriers (89.8%). Among patients with ovarian cancer G1, the all-cause survival was significantly associated with a BRCA1 mutation (HR=4.73, 95%CI 1.45–15.43, p=0.01). This result is different than in our previously reported published analysis of 2270 ovarian cancer patients, in which we did not find an association of BRCA1 mutation with overall survival [5]. However, it should be noted that low-grade ovarian cancers are rare and in that study, 7.3% of all cases (167 out of 2270) were accounted and only 4 cases carried BRCA1 mutation.

Our study has several limitations. It is based on a relatively small number of cases, and obtained results should be considered as preliminary. In this study, we analyzed the 21 most common Polish founder mutations in 5 genes associated with breast/ovarian cancer risk. Therefore, we omitted other non-founder mutations which might be detected by full sequencing of these genes. In our study, the cases and controls were not matched. The time of recruitment and mean age of cases and control were similar. It is unlikely but it might be that other unknown genetic or lifestyle factors could cause a study bias.