Mutational Analysis of BRCA1 and BRCA2 and Clinicopathologic Analysis of Ovarian Cancer in 82 Ovarian Cancer Families: Two Common Founder Mutations of BRCA1 in Japanese Population¹

Ovarian cancer is the most lethal disease in gynecological malignancy. Approximately 5–10% of cases are thought to have a hereditary basis (1), and a positive family history of ovarian cancer is one of the strongest and most consistent of the risk factors for the development of the disease. It has been reported that first-degree relatives of ovarian cancer patients were found to be at a 2–4-fold increased risk for developing the disease (2, 3). Familial ovarian cancer occurs as part of two clinically distinct syndromes: site-specific ovarian cancer families and breast-ovarian cancer families (4). Predisposition to ovarian cancer also occurs as part of Lynch type II or hereditary nonpolyposis colorectal cancer syndrome (5). Inherited mutations of BRCA1, BRCA2, and the mismatch-repair genes are known to confer predisposition to ovarian cancer (6, 7, 8). Germ-line mutations of BRCA1 are predicted to be responsible for ∼45% of breast cancer families and 80% of breast-ovarian cancer families (8, 9, 10). Both male and female BRCA2 carriers have a high risk of early onset breast cancer; however, ovarian cancer initially was thought to be a much less prominent feature of these families, but it is now thought that BRCA2 may account for as much as 10–35% of familial ovarian cancers (8, 11).

The characteristics of familial ovarian cancer are not as well documented in larger studies. A high proportion of serous adenocarcinoma and of an advanced stage has been reported for BRCA1-associated tumors in several studies (12, 13, 14). However, the clinical features of BRCA2-associated ovarian cancers are still unknown. Recently, it was reported that germ-line BRCA1 or BRCA2 mutations were found in 43% of 112 ovarian cancer families in the United Kingdom (14, 15). We previously reported seven BRCA1 mutations found in 19 ovarian cancer families (16). No other studies have reported on as many ovarian cancer families in Japan. It is very difficult to recruit a sufficient number of ovarian cancer families in Japan, because lifetime risk of ovarian cancer for women in Japan is 2 or 3 times lower than that in the United States (17), and a registry for ovarian cancer has not been established.

The mutation spectrum is similar in both BRCA1 and BRCA2 genes. Most germ-line mutations are predicted to result in protein truncation caused by frameshift, nonsense, or splice-site alterations, and the mutations are spread along the length of the coding region (The Breast Cancer Information Core). However, it was reported that genetic and epigenetic alterations in familial ovarian cancer in Japan may differ from those in Western countries (18). Founder mutations for BRCA1 and BRCA2 have been described in many racial and ethnic groups (19, 20). In Japan, few reports have been described about a common founder mutation for these genes.

In this study, we have collected information on 82 ovarian cancer families and analyzed their genetic alterations and characteristics of ovarian cancer patients with BRCA1 or BRCA2 mutation in Japan.

We examined the clinical data from hospital records and pathological reports or asked physicians to answer questionnaires or hear from patients in nationwide major centers for gynecological cancer in Japan. Among ∼7900 patients with epithelial ovarian cancer in approximately 80 centers for ∼10 years, we recruited 82 probands and ovarian cancer families in which 198 patients were ascertained. The criterion for a site-specific ovarian cancer family is as follows: Two or more members with well-documented epithelial ovarian cancer in the second-degree relatives and no breast cancer cases in the third-degree relatives. When the family had at least one breast cancer case in a third-degree relative, it was classified as a breast-ovarian cancer family. All of the experiments were performed under informed consent. Population control patients with epithelial ovarian cancer were ascertained from the cancer registry of Niigata in Japan from 1983 to 1996 and a nationwide survey in Japan from 1980 to 1987 (21).

Direct sequencing was performed with genomic DNA obtained from one individual with ovarian cancer from each family initially. When the individual was positive with a mutation and more than one material was available in the same family, we examined using direct sequencing as to whether other affected individuals were carrying the same mutation or not. Genomic DNA was prepared from lymphocytes and paraffin-embedded blocks using the standard phenol-chloroform methods. The entire exons, 23 exons in BRCA1 and 26 exons in BRCA2, and the intronic boundary regions were sequenced in both forward and reverse directions for detecting germ-line mutations. The noncoding intronic regions that were analyzed did not extend more than 20-bp proximal to the 5′ end and 10-bp distal to the 3′ end of each exon. These regions were amplified by PCR respectively from 100 ng of genomic DNA (35 reactions for BRCA1 and 47 reactions for BRCA2). The PCR products were sequenced by the dideoxy method using an Autocycle sequencing kit (Pharmacia Biotech, Tokyo, Japan) and an end-labeled primer by Cy5. PCR products were electrophoresed in 6% polyacrylamide gel and analyzed with an automatic sequencer, ALF express (Pharmacia Biotech).

Clinical characteristics among ovarian cancer patients were tested by unpaired t test, χ² analysis, and Fisher’s exact test.

The existence of germ-line mutations of BRCA1 or BRCA2 was analyzed on 82 ovarian cancer families (Table 1), of which 55 were site-specific ovarian cancer families and 27 were breast-ovarian cancer families. In the 55 site-specific ovarian cancer families, 22 families were carrying germ-line mutations in BRCA1 (22/55, 40.0%); however, in 27 breast-ovarian cancer families, 18 families were positive with the mutation (18/27, 66.7%). A very small proportion of the families were carrying germ-line mutations in BRCA2, 2 families among the 55 site specific ovarian cancer families (2/55, 3.6%) and 3 families among the 27 breast ovarian cancer families (3/27, 11.1%). Thirty-seven families were negative for germ-line mutations of BRCA1 or BRCA2 (37/82, 45.1%). BRCA1 mutations were 8 times more common than BRCA2 mutations (40 versus 5).

Germ-line mutations in BRCA1 or BRCA2 are listed in Table 2. Twenty-four independent BRCA1 mutations were identified in 40 of the 82 families (40/82, 48.8%). Thirty-six of the 40 mutations are either frameshift or nonsense mutations (36/40, 90.0%) that would be predicted to result in premature truncation of the BRCA1 protein. Three mutations were missense mutations that were presumably predicted to disrupt the start codon of exon 2, to lose a zinc-binding motif of exon 5, or to induce splice aberration by occurring at the end of exon 21. These mutations were not found in healthy women in this family or in a substantial number of healthy volunteers who had no family history of ovarian and/or breast cancer, indicating that these genetic variants could be diagnosed as pathogenic mutations but not polymorphisms (data not shown). One mutation, IVS14–2A>G, consisting of a nucleotide substitution in a noncoding intervening sequence was expected to prevent mRNA processing. We identified five recurrent mutations and, among these recurrent mutations, a substitution of T to A at nucleotide 307 (L63X) was detected in 7 independent families (7/40, 17.5%), and a substitution of C to T at nucleotide 2919 (Q934X) was detected in 8 families (8/40, 20.0%). Five independent BRCA2 mutations were identified (5/82, 6.1%), and all of the mutations are predicted to result in premature truncation of the BRCA2 protein. Sixteen of 24 BRCA1 and three of five BRCA2-independent mutations detected in this study were not found in any other reports. In 38 of 45 families carrying mutation in BRCA1 or BRCA2, multiple affected members including patients with breast cancer shared a same mutation (7 families with >3 members and 31 families with 2 members). In 7 other families, no available DNA was obtained from any affected individual except the proband.

Genetic variants of uncertain significance and common polymorphisms in BRCA1 or BRCA2 are listed in Table 3. Two genetic variants and eight common polymorphisms were observed in BRCA1, and three variants and eight polymorphisms were observed in BRCA2.

The clinical and pathological characteristics of familial and sporadic patients with epithelial ovarian cancer are summarized in Table 4. There are the characteristics of 110 patients with BRCA1 mutation, 10 patients with BRCA2 mutation, 78 patients with no mutation in BRCA1 or BRCA2, 1299 patients with epithelial ovarian cancer from the cancer registry of Niigata in Japan as control 1 and 1185 patients with epithelial ovarian cancer from a nationwide survey in Japan as control 2 (21). The mean age at diagnosis of tumors with no mutation (49.7 years of age) was significantly younger than control cases, 54.2 years of age (P = 0.0076). However, there were no significant differences of mean age at diagnosis between patients with BRCA1 or BRCA2 mutation, 52.1 and 58.4 years of age, respectively, and those of control cases. The major histological type for BRCA1- or BRCA2-associated tumors was serous adenocarcinoma in 79.8% and 88.9% of tumors, respectively. There were significant differences in the proportion of tumors with serous adenocarcinoma between these two groups and the control groups; however, no difference was found between the nonmutation groups and the control groups. No tumor with mucinous or clear cell adenocarcinoma occurred in the BRCA1-related cases. In regard to the clinical stage at diagnosis, there was a significantly higher proportion of stage III or IV tumors in the BRCA1 or BRCA2 cases than in the control cases. On the other hand, no difference was seen in the stage distribution between the tumors with no mutation and those of the control 1 (P = 0.25).

Table 5 represents the characteristics of the cases with L63X or Q934X and other mutations in BRCA1. The mean age at diagnosis for cases with either mutation did not significantly differ from that for the others. In addition, no difference was seen in the histological subtypes between each mutation status. On the other hand, there was a significantly lower proportion of stage III or IV tumors in the cases with L63X than in the others (P = 0.016).

The geographic distribution of ovarian cancer families with these two mutations is shown in Fig. 1 and Table 6. We divided the Japanese Islands into east and west based on the Fossa Magna. The Fossa Magna, named by German geologist, means a great chasm in the earth in Latin. All of the families with L63X were in the eastern part of Japan, and the proportion of families with L63X in the east was significantly higher than that in the west (P = 0.035).

The clinical and genetic characteristics of familial ovarian cancer are not as well documented as those of familial breast cancer. Recently, Gayther et al. (15) reported that germ-line BRCA1 or BRCA2 mutations were found in 43% of 112 ovarian cancer families in the United Kingdom, and BRCA1 mutations were ∼4 times more common than BRCA2 mutations. In this study, we identified germ-line mutations of BRCA1 or BRCA2 in slightly more than one-half (55%) of 82 families and found that BRCA1 mutations were 8 times more common than BRCA2 mutations in Japanese ovarian cancer families. In our study, most of the BRCA1 mutations were predicted to result in protein truncation similar to observations in many other reports (14, 15), suggesting that this is one of the unique characteristics of the BRCA1 gene. Contrary to our findings, in Japanese breast-ovarian cancer families, a much lower frequency of the mutation of BRCA1 in which missense mutations tended to predominate was observed (18). This contradictory result could be related to technical problems of mutation analysis, e.g. a direct sequencing on the whole coding region versus a screening by single strand conformation polymorphism analysis.

Currently, various BRCA1 mutations have been identified, though a few have been detected recurrently. One such mutation, 185delAG, is observed in ∼1% of Ashkenazi Jews (22). These studies demonstrate that some mutations of BRCA1 are segregated geographically and racially. On the basis of our findings, the L63X and the Q934X were seen in seven and eight independent families, respectively. These two mutations were reported in breast cancer families in Japan (eight families with L63X and three families with Q934X, respectively) but not in other countries (23, 24, 25, 26). Haplotyping analysis showed that the cases with these two mutations were likely derived from common ancestors (data not shown). These results indicate that these kinds of mutations are common founder mutations in the Japanese population. The geographic distribution of ovarian cancer patients with these two mutations seems to be skewed. Japanese are a relatively uniform population ethnically, so this difference of geographic prevalence based on mutation status may be caused by migration and marriage patterns in the population structure in Japan. The Japanese islands are divided into east and west based on the Fossa Magna according to culture. In fact, people in the eastern part of Japan traditionally eat square rice cakes during New Year’s ceremony. On the other hand, people in the west eat round ones.

The risk of ovarian cancer is not believed to be the same for all of the BRCA1 mutations and varies according to the position of the mutation along the gene (27). In particular, several studies suggested that the risk of ovarian cancer in females with 185delAG mutation was higher than that with 5382 insC mutation (28, 29). Furthermore, Hedenfalk et al. (30) suggested that a heritable mutation influenced the gene expression profile of the cancer. Therefore, we examined the differences of the clinical features between cases with L63X and with Q934X. As a result, there was a significantly lower proportion of advanced tumors in the cases with L63X than in other mutations. These results suggest that the position of the germ-line mutation along the BRCA1 gene may influence the process of carcinogenesis and progression of the disease. The question of whether different mutations result in a different prognosis in this population will require additional study.

The results of our clinical analysis revealed that there was a significantly higher proportion of tumors of an advanced stage in the BRCA1 or BRCA2 cases than in the control cases; however, we previously reported that 13 patients treated with stage III disease with mutations of BRCA1 showed more favorable outcomes compared with sporadic cases (31). In regard to the mean age at diagnosis, no difference was seen between patients with BRCA1 or BRCA2 mutation and those of the control cases. Boyd et al. (32) reported that ovarian cancers with BRCA1 mutation in the United States were diagnosed more than 8 years earlier compared with those of the sporadic cases. The later onset of BRCA1-associated ovarian cancer in Japan might be attributed to the sensitiveness of subjects to environmental differences (e.g., lower parity, lower use of talc, lower population of obese women, lower serum level of gonadotropin, and lower intakes of fat or milk in Japan) as well as to the pathogenesis of sporadic ovarian cancer, because the tumor cell was caused by both initial germ cell mutation and proceeding somatic cell mutation.

We cannot exclude the possibility that our screening results presented were likely to be an underestimate of the contribution of these two genes to familial ovarian cancer, because, for example, any mutation altering a primer site would have been missed. Nevertheless, about one-half of the ovarian cancer families were not associated with mutation of BRCA1 or BRCA2. The clinical features in relation to the younger mean age at diagnosis of tumors with no mutation other than that of sporadic cases suggested that these patients were influenced with the hereditary genetic instability succeeded from the ancestral patients. Consistent with the previous reports (14), these findings raise the possibility that additional novel susceptibility genes for familial ovarian cancer might exist (33). Therefore, it seems urgent to perform genome-wide linkage analysis and an association study in ovarian cancer families for additional investigation to identify novel susceptibility genes for familial ovarian cancer.

We thank Tsutomu Araki (Nippon Medical School), Toshihiko Iida (Utsunomiya Saiseikai Hospital), Chikashi Ishioka (Tohoku University), Masamichi Kashimura (University of Occupational and Environmental Health), Atsushi Takano (Aomori Kosei Hospital), Tetsuya Chidori (Toyama City Hospital), Ryuichiro Tsunematsu (National Cancer Center), Hideki Mizunuma (Gunma University), Makoto Murakami (Sasebo City General Hospital), Ichiro Yamadori (National Okayama Medical Center), Shigeru Arai (Saiseikai Niigata Daini Hospital), Masami Kato (Nagaoka Chuo Hospital), Noriyasu Saito (Shonai Hospital), Norihito Sudo (Nagaoka Red Cross Hospital), Takeshi Takahashi (Niigata Cancer Center Hospital), Hiroaki Takahashi (Shibata Hospital), Kohei Tanaka (Akita Red Cross Hospital), Akiteru Tokunaga (Niigata City General Hospital), Yuichi Torii (Seirei-hamamatsu General Hospital), Minoru Nakamura (Saiseikai Sanjo Hospital), and Toshio Nishiyama (Kamo Hospital) for their invaluable help and contributions to this work. We are grateful to the patients and their families for participating in this study. We also thank Toshikazu Ushijima at Carcinogenesis Division, National Cancer Center Research Institute for providing primers of BRCA2, and Noriko Araki and Yukari Sato for excellent technical assistance.