遗传性卵巢癌：超出通常的怀疑--爱思唯尔期刊精选全文--爱唯医学网

遗传性卵巢癌：超出通常的怀疑

Hereditary ovarian cancer: Beyond the usual suspects

Kathryn P. Pennington, Elizabeth M. Swisher | 2012/1/20 17:08:00

Gynecologic Oncology | 2012 | Volume 124 Issue 2 | 打印| 推荐给好友

上一篇：复杂表型的分子遗传学研究

下一篇：宫内节育器处于原位的妊娠的预后：一项系统综述

Abstract

In the past, hereditary ovarian carcinoma was attributed almost entirely to mutations in BRCA1 and BRCA2, with a much smaller contribution from mutations in DNA mismatch repair genes. Recently, three new ovarian cancer susceptibility genes have been identified: RAD51C, RAD51D, and BRIP1. In addition, germline mutations in women with ovarian carcinoma have been recently identified in many of the previously identified breast cancer genes in the Fanconi anemia (FA)–BRCA pathway. While mutations in genes other than BRCA1 and BRCA2 are each individually rare, together they make up a significant proportion of cases. With at least 16 genes implicated in hereditary ovarian cancer to date, comprehensive testing for ovarian cancer risk will require assessment of many genes. As the cost of genomic sequencing continues to fall, the practice of evaluating cancer susceptibility one gene at a time is rapidly becoming obsolete. New advances in genomic technologies will likely accelerate the discovery of additional cancer susceptibility genes and increase the feasibility of comprehensive evaluation of multiple genes simultaneously at low cost. Improved recognition of inherited risk will identify individuals who are candidates for targeted prevention. In addition, identifying inherited mutations in a variety of FA–BRCA pathway genes may aid in identifying individuals who will selectively benefit from PARP inhibitors.

Highlights

► Newly identified hereditary ovarian cancer genes include RAD51C, RAD51D, and BRIP1. ► Many genes in the Fanconi anemia–BRCA pathway may increase risk of ovarian cancer. ► New genomic technologies make comprehensive genetic assessment feasible.

Introduction

For the last 15 years, hereditary ovarian carcinoma was attributed almost entirely to mutations in the breast and ovarian cancer susceptibility genes BRCA1 and BRCA2, with a much smaller contribution from mutations in the DNA mismatch repair (MMR) genes. Recently, three additional genes have been associated with hereditary ovarian carcinoma: RAD51C, RAD51D, and BRIP1 [1], [2] and [3]. These three genes share with BRCA1 and BRCA2 key roles in the Fanconi anemia (FA)–BRCA pathway, which is critical for DNA repair by homologous recombination (HR). The FA–BRCA pathway has been broadly implicated in breast cancer, with many genes in this network demonstrating moderate penetrance for familial breast cancer [4]. New data implicate the FA–BRCA pathway as equally important in hereditary ovarian carcinoma. Recognition of inherited risk allows targeted prevention. Ovarian cancer surveillance has not proven effective to date, but risk-reducing salpingo-oophorectomy reduces ovarian, fallopian tube and breast cancer risk [5]. The contribution to ovarian carcinomas of inherited mutations in genes other than BRCA1 and BRCA2 is individually small, but together account for a significant proportion of cases. Therefore, comprehensive genetic assessment for inherited risk of ovarian carcinoma will require interrogating many genes. As the cost of genomic sequencing continues to fall, the practice of evaluating cancer susceptibility one gene at a time is rapidly becoming obsolete. Here, we review the changing genetic landscape of hereditary ovarian carcinoma, discussing implications for genetic testing and risk assessment.

BRCA1 and BRCA2

BRCA1 and BRCA2 (BRCA1/2) mutations are thought to account for the majority of hereditary ovarian carcinomas. Two population studies in North American women demonstrated that 13–15% of women with invasive ovarian carcinoma harbor a germline mutation in BRCA1/2 [6] and [7]. The proportion of ovarian carcinoma that is hereditary in a specific population varies based on the prevalence of founder mutations. In the Ashkenazi population, 2.3% of individuals have one of three founder mutations (BRCA1 187delAG, BRCA1 5358insC, BRCA2 6174delT) [8], [9] and [10]. The high prevalence of founder mutations results in the identification of germline BRCA1/2 mutations in 35–40% of Ashkenazi women with ovarian carcinomas [11], [12] and [13].

Women with BRCA1/2 mutations have an elevated risk of fallopian tube and peritoneal carcinomas as well as breast and ovarian carcinoma. In contrast to the general population, in which the lifetime risk for developing ovarian carcinoma is 1.6%, women with a BRCA1 mutation have a lifetime risk of 35–60% with an average age of diagnosis of 50 years. The penetrance for ovarian cancer is somewhat lower for BRCA2, which confers a lifetime risk of 12–25% with an average age of diagnosis of 60 years [11], [14], [15], [16] and [17]. BRCA1/2-associated ovarian carcinoma has a distinct clinical phenotype, with the majority having high-grade serous histology and advanced stage [7], [18] and [19]. BRCA1/2-associated ovarian carcinoma can also demonstrate other high grade histologies including endometrioid, clear cell, carcinosarcoma, and undifferentiated carcinoma [7], [18] and [19]. In contrast, BRCA1/2 mutations are typically under-represented amongst mucinous and borderline neoplasms [16], [20] and [21]. In addition, BRCA1/2-associated ovarian carcinomas have been associated with longer overall survival [22] and [23], increased sensitivity to platinum-based chemotherapy [23] and [24], and increased sensitivity to therapy with poly-ADP-ribose polymerase (PARP) inhibitors [25] compared to their sporadic counterparts.

Genetic testing for cancer susceptibility is preferentially initiated in an affected individual after genetic counseling by a cancer genetics professional. Once a mutation is identified, unaffected relatives may be tested for that specific mutation. In the United States, gene patents restrict clinical testing for BRCA1 and BRCA2 to a single company, Myriad Genetics (with the exception of the Ashkenazi founder mutations). The standard BRCA1/2 test by Myriad Genetics includes sequencing of all the coding exons and intron–exon boundaries as well as testing for five gene rearrangements common in northern and western Europeans. However, other gene rearrangements, which account for 8–15% of all BRCA1/2 mutations, are not identified by standard sequencing [26], [27], [28], [29] and [30]. Myriad Genetics offers comprehensive rearrangement testing for BRCA1/2 as a separate test at an additional cost of $650, which is not covered by Medicare and many private insurance companies. Consequently, the majority of women in the U.S. tested for BRCA1/2 receive an incomplete mutation evaluation.

Lynch syndrome or hereditary non-polyposis colorectal cancer (HNPCC)

Lynch syndrome is an autosomal dominant disorder which predisposes to colorectal cancer, endometrial cancer, ovarian, gastric, small bowel, biliary/pancreatic, urothelial, skin, and central nervous system cancers. The most common cancers for women with Lynch syndrome are colorectal and endometrial adenocarcinomas, with lifetime risks of approximately 30–50% and 40–50%, respectively [31], [32] and [33]. The cumulative risk of ovarian cancer is estimated to be 8–10%, with an average age of onset of 42 years [34]. In the largest series of ovarian cancers occurring within Lynch families, a higher proportion (18.3%) of cases were noted to be of endometrioid histology compared to sporadic ovarian carcinoma, and one quarter of cases occurred in the setting of metachronous endometrial cancer [34]. Presumably, some of these cases were actually metastases from endometrial cancer and therefore the lifetime risk of ovarian cancer may be slightly over-stated.

Mutations in MSH2 or MLH1 account for approximately 70% of families with Lynch syndrome, with MSH6 and PMS2 mutations accounting for most of the remainder [35]. Less common genetic explanations for Lynch syndrome include deletions in EPCAM and germline MLH1 promoter methylation, which lead to down-regulation of MSH2 and MLH1, respectively [36]. The common phenotype of Lynch-associated cancers is microsatellite instability (MSI) resulting from defective DNA mismatch repair.

In the genetic evaluation of Lynch syndrome, two main diagnostic criteria assist in determining who to evaluate with genetic testing (Table 1). The modified Amsterdam criteria clinically define a family with possible Lynch syndrome and are meant to be somewhat restrictive. Families that meet Amsterdam criteria may be considered directly for germline DNA testing of the four Lynch genes, with testing preferably initiated in an affected individual. In some cases, tumor testing for protein expression or microsatellite instability can help define which genes to test in such a family. In contrast, the Bethesda guidelines are used to define which subsets of individuals with colon or endometrial cancer should be screened for Lynch syndrome using tumor analyses. The purpose of the Bethesda guidelines is not to define Lynch syndrome, but rather to establish loose criteria for tumor testing with low specificity and high sensitivity.

Table 1. The revised Amsterdam and Bethesda guidelines.

Amsterdam II criteria “3–2–1 rule”
Purpose: Establish clinical criteria for identifying Lynch syndrome in a family
• Three or more relatives with Lynch syndrome-associated cancers*
– One must be first-degree relative of other two
• Lynch syndrome-associated cancers involving at least two generations
• At least one cancer diagnosed before age 50
*Familial adenomatous polyposis should be excluded in the colorectal cancer cases.

Revised Bethesda guidelines
Purpose: Identify individuals with colon cancer to consider for further testing for possible Lynch syndrome. Similar criteria have been applied to endometrial cancer cases.
• Colorectal cancer diagnosed in patient < 50 years of age
• Presence of synchronous, metachronous colorectal or other HNPCC-associated tumors*, regardless of age
• Colorectal cancer with the microsatellite instability-high-like histology^† diagnosed in a patient < 60 years of age
• Colorectal cancer diagnosed in a patient with one or more first-degree relatives with an HNPCC-related tumor, with one of the cancers diagnosed under age 50
• Colorectal cancer diagnosed in a patient with two or more first- or second-degree relatives with HNPCC-related tumors, regardless of age
*Colorectal, endometrial, gastric, ovarian, pancreatic, ureter and renal pelvis, biliary tract, brain tumors (usually glioblastoma as seen in Turcot syndrome), sebaceous gland adenomas and keratoacanthomas in Muir–Torre syndrome, and small bowel carcinoma.
^†Presence of tumor infiltrating lymphocytes. Crohn's-like lymphocytic reaction, mucinous/signet-ring differentiation, or medullary growth pattern.

It is important to remember that tumor testing is not equivalent to germline DNA testing. Tumor testing evaluates protein expression of the 4 MMR genes by immunohistochemistry and/or identifies microsatellite instability (MSI). While some tumor testing may be pathognomonic for Lynch syndrome (such as loss of MSH2 or MSH6 protein), unaffected family members cannot be tested until the actual germline mutation is defined. The use of tumor testing is a lower cost alternative to evaluate for Lynch syndrome when the pre-test probability is low or to decrease the cost of genetic testing by defining the likely gene. The use of tumor testing in the algorithm of the evaluation for Lynch syndrome will likely become less important as comprehensive testing for all Lynch or all colon cancer genes becomes available at lower cost.

Susceptibility pathways common to Fanconi anemia and breast cancer

The study of Fanconi anemia (FA), a rare autosomal or X-linked recessive disorder of chromosomal instability, has led to improved understanding of DNA repair and additional candidate genes for cancer risk. Clinical features of FA include aplastic anemia in childhood, congenital anomalies, increased susceptibility to leukemia and other malignancies, and cellular hypersensitivity to interstrand DNA cross-linking agents such as cisplatin and other alkylators [37] and [38]. There are 14 genes involved in FA, and the proteins encoded by these genes work together in the recognition and repair of damaged DNA [39], [40] and [41].

One of the FA genes, FANCD1, is the breast and ovarian cancer susceptibility gene, BRCA2[42]; monoallelic mutations cause hereditary breast and ovarian cancer while biallelic mutations result in FA. Biallelic mutations have not been described in BRCA1. Therefore, BRCA1 is not a FA gene but is a key component of the FA pathway. The FA–BRCA pathway functions in homologous recombination (HR), the main DNA repair pathway that mends double strand DNA breaks with high fidelity. When HR is deficient, cells rely on non-homologous end-joining (NHEJ) to repair double strand DNA breaks, an alternate and more error prone repair mechanism. PARP inhibitors exploit defects in HR, leading to synthetic lethality in cells lacking BRCA1/2 [25] and [43]. PARP inhibitors may also demonstrate selective killing in cells deficient in other HR genes, which has been recently demonstrated for RAD51D, NBN, ATM, and CHEK2, and theoretically, would apply to many genes in the FA–BRCA pathway [1] and [44]. Thus, the identification of HR genes that confer susceptibility to cancer may have significant therapeutic implications.

Many additional genes in the FA–BRCA pathway have been identified which increase breast cancer susceptibility (Fig. 1). ATM is an important cell cycle checkpoint kinase within the FA–BRCA pathway that phosphorylates BRCA1. Biallelic mutations in ATM are uniquely responsible for the autosomal recessive disease ataxia telangiectasia, which is characterized by progressive cerebellar ataxia, apraxia, immunodeficiency, increased risk of lymphomas and leukemias, and hypersensitivity to ionizing radiation. As individuals with ataxia telangiectasia are now living longer, other cancers have been observed including ovarian cancer, breast cancer, and gastric cancer [45]. Monoallelic mutations in ATM have been associated with moderate penetrance for breast cancer [46], but to date have not been implicated in ovarian cancer risk.

Fig. 1. Schematic of FA–BRCA pathway and its role in breast cancer susceptibility. Germline mutations in all blue and green genes except ATM have also been identified in women with ovarian carcinomas [19].

View thumbnail images

Monoallelic mutations in other genes in the FA–BRCA pathway also confer moderate penetrance for breast cancer (2–3 fold increased risk) including CHEK2, BARD1, MRE11A, NBN (NBS1), RAD50, RAD51C, and two other FA genes, BRIP1 (FANCJ) and PALB2 (FANCN) [2], [47], [48], [49], [50] and [51]. Until recently, these genes had only been evaluated comprehensively in high risk breast cancer families, though certain founder mutations such as CHEK2 1100delC have been studied in large population series of breast cancer [50]. Previously, most hereditary breast and ovarian cancer risk was thought to stem from BRCA1 and BRCA2, while breast but not ovarian cancer risk was associated with other FA–BRCA genes. Current data, however, contradict those assumptions, implicating these genes in ovarian as well as breast cancer risk.

Evaluation of HR genes identifies additional genes involved in hereditary ovarian cancer

PALB2

PALB2, also known as FANCN, is a FA gene which confers breast cancer susceptibility [52]. PALB2 may also play a role in ovarian cancer susceptibility. Casadei et al. sequenced PALB2 in high risk breast cancer families, identifying PALB2 mutations in 33 of 972 families (3.4%) [49]. Of note, 18 of these 33 families (55%) had a family member with ovarian cancer, who were confirmed to carry the familial PALB2 mutation. Notably, these families had a similar phenotype to BRCA2, with an increased incidence of pancreatic as well as breast and ovarian cancer. While ovarian cancer was more common among relatives of PALB2 mutation carriers, these results were not statistically significant, and therefore the penetrance of PALB2 for ovarian cancer is not certain. In another study of 339 unselected ovarian cancer subjects, a Polish PALB2 founder mutation was detected in 0.6% of subjects, but in only 1 of 1310 (0.08%) of controls [53].

RAD51C

RAD51C is an essential gene in HR, and a biallelic missense mutation causes a FA-like phenotype [54]. RAD51C was investigated as a possible breast and ovarian cancer susceptibility gene in 1100 high risk families negative for BRCA1/2 mutations [2]. Pathogenic RAD51C germline mutations were identified in 6 of 480 families (1.3%) with both breast and ovarian cancer, but in none of 620 families with breast cancer only or in 2912 healthy controls. Within the RAD51C mutation-associated pedigrees, the mean age at cancer diagnosis was 60 years for ovarian (range 50–81) and 53 years (range 33–78) for breast cancer.

RAD51D

Investigators recently sequenced RAD51D in 911 BRCA1/2 negative breast-ovarian cancer families as well as 1060 population controls [1]. Inactivating mutations were identified in 8 of 911 breast and ovarian cancer families (0.9%), 0 of 737 breast cancer only families, and 1 of 1060 controls (0.09%). There was a higher prevalence of mutations present in families with more cases of ovarian cancer: 4 of the mutations were detected in the 235 families with 2 or more cases of ovarian cancer (1.7%), and 3 of the mutations were detected in the 59 families with 3 or more cases of ovarian cancer (5.1%). The relative risk of ovarian cancer for carriers of deleterious RAD51D mutations was estimated at 6.3 (95% CI 2.9–13.9), whereas the relative risk of breast cancer was non-significantly increased at 1.3 (95% CI 0.6–3.0). Interestingly, these authors confirmed that cells deficient in RAD51D show selective sensitivity to PARP inhibitors.

BRIP1

Recently, two frameshift BRIP1 mutations were also found to significantly increase the risk of ovarian cancer [3]. Evaluating sequence variants identified through the Icelandic whole-genomic sequencing project, researchers identified a rare (0.41% allelic frequency) frameshift mutation (c.2040_2041insTT) in the BRIP1 gene which confers an increased risk of ovarian cancer (OR 8.1, 95% CI 4.7–14.0), but no increased risk of breast cancer. The authors then entirely sequenced BRIP1 in ovarian cancer cases or controls from Iceland, the Netherlands, and Spain. In Spanish individuals, another rare BRIP1 mutation (c.1702_1703del) was identified (allelic frequency 0.03% in 1782 controls) and associated with a markedly elevated risk of ovarian cancer (OR = 25, 95% CI 1.8–340) as well as a significant risk of breast cancer (OR = 12, 95% CI 1.9–70). In contrast, common missense variants in BRIP1 were found at similar frequencies in cancer cases and controls, or were too rare to determine an association. In addition, Rafnar et al. confirmed low penetrance for three of four common variants reported to associate with ovarian cancer in previous genome-wide association studies with odds ratios of 1.1–1.28 [55], [56] and [57].

Genome-wide association studies

A genome-wide association study (GWAS) is a study design which compares the genotype frequencies between cases and controls of hundreds of thousands of common variants distributed throughout the genome, in an effort to identify alleles associated with disease risk. This approach has identified common genetic variants that confer low-penetrance susceptibility to a variety of cancers. To date, genetic variants that show the strongest association with ovarian cancer risk include single nucleotide polymorphisms (SNPs) on chromosomes 19p13, 2q31, 9p22.2, and 8q24 [55], [56] and [57]. However, these effects on risk are modest, with odds ratios of 1.13, 1.20, 0.82 and 0.76, respectively, and the causality of these variants is uncertain. Potential candidate genes at or near these susceptibility loci include BNC2, HOXD1, MYC, TIPARP, SKAP, and the BRCA1-interacting gene C19orf62. Prior to GWAS, a popular hypothesis was that much of the familial cancer risk that was not accounted for by known highly penetrant cancer genes would be explained by common genetic variants of low penetrance. Interestingly, GWAS in cancers and other diseases have identified only a small fraction of unexplained familial risk [58]. One current explanation is that there are many private or very rare mutations with moderate or even high penetrance in a variety of genes that account for most of the remaining familial risk [59]. Given their small effect on ovarian cancer risk, GWAS variants have little current clinical utility in cancer risk assessment for the individual patient.

BROCA

With the advent of targeted capture and massively parallel sequencing, it is now feasible to sequence many genes simultaneously with high coverage, allowing improved mutation detection sensitivity. We have applied this technique to interrogate breast and ovarian cancer genes in a test we call BROCA in honor of Paul Broca, the 19thC neurosurgeon, oncologist, anatomist, and evolutionist, who elegantly described inherited breast and ovarian cancer [60]. We applied BROCA to test the contribution of germline mutations in 21 tumor suppressor genes in 360 women with ovarian, fallopian tube, and peritoneal carcinoma, unselected for age or family history [19]. Of note, BROCA identifies all classes of mutations, including single base substitutions, small insertion or deletions, and larger gene rearrangements (copy number variations) that would be missed by standard Sanger sequencing. The 21 tumor suppressor genes were: BRCA1, BRCA2, PALB2, NBN, BRIP1, RAD50, RAD51C, CHEK2, ATM, MRE11A, BARD1, TP53, PTEN, CDH1, STK11, MUTYH, MSH2, MLH1, MSH6, PMS2, and PMS1.

Twenty-three percent of cases had a loss-of-function germline mutation: 18% in BRCA1 or BRCA2, and an additional 6% in one of 10 other genes. Only clearly deleterious mutations were included. We identified mutations in the newly identified ovarian cancer susceptibility genes RAD51C and BRIP1. (Of note, at the time of the study design, RAD51D had not yet been identified as an ovarian cancer susceptibility gene and was therefore not included in the BROCA design.) Additional mutations were also identified in every other gene in the FA–BRCA pathway tested except ATM, including BARD1, CHEK2, MRE11A, NBN, PALB2, and RAD50. All of these genes were previously known to confer a moderate risk of breast cancer, but had not been reported in unselected ovarian cancers.

Mutations were also identified in MSH6 in 2 endometrioid ovarian cancers occurring in women that did not meet criteria for Lynch syndrome (0.05% of cases). Germline mutations were identified in TP53 in 3 cases, none of whom met the criteria for Li–Fraumeni syndrome. As comprehensive genetic testing is increasingly being undertaken for individuals not selected for established syndromic phenotypes, a wider range of expressivity associated with germline mutations of cancer susceptibility genes may become increasingly apparent.

These results suggest that more than one in five ovarian carcinomas is associated with a germline mutation in a tumor suppressor gene, and inherited mutations are distributed in a larger number of genes than previously thought (Table 2). Importantly, our germline mutation rate was a minimum rate as we did not include missense mutations without known functional consequence, and some damaging missense mutations were likely not included. Of the FA–BRCA genes targeted by BROCA, we identified clearly deleterious mutations in all genes except ATM. For many of these genes, we know neither the penetrance nor the average age of ovarian cancer onset. Therefore, large case control studies will be needed to identify lifetime risk estimates to appropriately counsel women about the risk, benefit, and timing of ovarian cancer prevention strategies. The low cost and high through put of BROCA (or a similar genomic sequencing strategy) should facilitate such studies. It is noteworthy that 30% of women with inherited mutations had no prior family history of breast or ovarian cancer, and 37% were diagnosed after age 60. These findings confirm previous studies that suggest that selecting patients with ovarian cancer for genetic testing based on family history or young age would miss one-third of mutation carriers [7]. Therefore, we should consider offering comprehensive genetic testing to all women with invasive ovarian carcinoma, regardless of age or family history.

Table 2. To date, at least 16 genes have been associated with hereditary ovarian cancer.

Genes implicated in hereditary ovarian carcinoma
FA–BRCA pathway genes
• BRCA1
• BRCA2
• RAD51C
• RAD51D
• BRIP1
• BARD1
• CHEK2
• MRE11A
• NBN
• PALB2
• RAD50
Mismatch repair genes
• MLH1
• MSH2
• MSH6
• PMS2
Other genes
• TP53

How should we counsel breast cancer-only families about their risk of developing ovarian cancer?

Individuals with a BRCA1 or BRCA2 mutation from hereditary breast cancer families have an elevated risk of ovarian cancer. It is less clear whether individuals from hereditary breast cancer only families who do not carry a BRCA1 or BRCA2 mutation are at increased risk of ovarian cancer. A large Swedish study demonstrated that a strong family history of breast cancer increases the subsequent risk of ovarian cancer in breast cancer survivors, and this effect is greater with younger age of breast cancer diagnosis and increased number of relatives with breast cancer [61]. However, this study did not include genetic analyses and presumably, much of this increased risk was accounted for by those individuals who had BRCA1/2 mutations.

One prospective study specifically addressed the question of ovarian cancer risk in breast cancer-only families that have tested negative for BRCA1/2; Kauff et al. followed 165 probands from BRCA1/2 mutation-negative breast cancer-only families over a mean follow-up of 40.6 months (2534 women-years of follow-up), [62]. Only one case of ovarian cancer was identified in a proband or family member compared to 0.66 expected cases (SIR 1.52, 95% CI 0.02–8.46, P = 0.48). The authors concluded that women from BRCA1/2 mutation-negative, site-specific breast cancer families may not be at an increased risk for ovarian cancer. Clinicians should interpret these results with caution for several reasons. First, the relatively small number of families and expected ovarian cancers creates a wide confidence interval in the estimate of risk and includes ranges of increased ovarian cancer risk that are clinically significant (up to 8 fold). Second, two-thirds of their study cohort consisted of Ashkenazi Jewish women, and it is possible that BRCA1/2 mutation testing in this group more effectively excludes the possibility of hereditary ovarian cancer risk compared to non-Ashkenazim.

The identification of many more genes in addition to BRCA1/2 that increase the risk of breast and ovarian cancer calls into question whether we can accurately counsel women with apparent hereditary breast cancer that they have a normal risk of ovarian cancer based only on BRCA1/2 testing. Some highly penetrant genes do increase the risk of breast cancer but not ovarian cancer such as CDH1, PTEN, and STK11. However, each of these genes is associated with a clear clinical phenotype that is usually recognizable. In apparent autosomal dominant breast cancer-only families without such a phenotype, there are a number of candidate FA–BRCA genes that could explain the cancer pedigree and would increase the risk of both breast and ovarian cancer. At this time, we should be cautious about assuming these families are explained by “breast cancer only” genes.

New genomic technologies will change clinical genetic testing

With the development of next generation sequencing technologies, there has been a shift away from Sanger sequencing for genome analysis. The major advantages of next-generation sequencing techniques are higher throughput, lower cost, and the potential for deeper coverage, which refers to an increased number of reads at each base sequenced [63]. Massively parallel sequencing technology has been applied both to whole-genome, exome, and more targeted approaches. Whole-genome sequencing is the most comprehensive approach, but remains costly and time-consuming, and identifies a huge number of variants of uncertain significance. A targeted-sequencing approach increases sequence depth-of-coverage at regions of interest, with lower cost and higher throughput per sample compared to whole genome sequencing. Any subset of the genome can be targeted. One popular approach is exome sequencing, which focuses on the 1% of the genome that consists of protein-coding exons, and does not require a-priori knowledge as to which genes are important. This approach was used by The Cancer Genome Atlas Research Network (TCGA), which performed exome sequencing in paired germline and neoplastic DNA from 316 women with serous ovarian carcinomas [64].

In contrast, we used BROCA to target selected breast and ovarian cancer genes, including exons, introns, untranslated regions, and flanking sequences [19]. It is interesting to compare our results with the TCGA exome sequencing in women with serous ovarian cancer. TCGA reported 44 germline BRCA1/2 mutations in 316 women [64]: 14%, after subtracting three reported occurrences of BRCA2 p.K3326X, a benign polymorphism found in 1% of the general population [65]. TCGA also fully sequenced BRCA1/2 using Sanger sequencing, but the fraction of BRCA1/2 mutations identified by Sanger sequencing versus exome sequencing was not reported. They reported no germline mutations in other genes. In contrast, among the 242 women with serous carcinoma evaluated by BROCA, mutations were identified in 57 (24%, P = 0.004), including 6% with loss of function germline mutations in genes other than BRCA1/2. Explanations for this discrepancy include: 1) low depth of coverage by the TCGA analysis; and 2) virtually no intronic coverage, precluding TCGA assessment of gene rearrangements, which comprised 7% of damaging events in our series. While exome sequencing is very well-suited for the identification of new cancer susceptibility genes, the overall much higher read depth of hybridization to a smaller number of target genes, following by multiplexed sequencing, confers greater sensitivity at lower cost, making it currently more suitable as a clinical test.

Current clinical practice relies upon the identification of high-risk individuals based upon personal and family history, and if their estimated risk exceeds a certain threshold, they are candidates for genetic counseling and testing. Although BRCA1 and BRCA2 still account for the majority of hereditary risk of ovarian cancer, deleterious mutations in other ovarian cancer genes together account for a significant proportion of cases. Comprehensive evaluation of hereditary ovarian cancer risk will require evaluating many genes, and it is expensive and time consuming to test one gene at a time. Next generation sequencing technologies are already sensitive and cheap enough to evaluate many genes simultaneously at a fraction of the cost of standard Sanger sequencing. For example, in our new version of BROCA, we can target 30–40 genes and identify all class of mutations (obviating the need for a separate test to evaluate gene rearrangements) with a current reagent cost of about $150. Clearly, there is an increasing discrepancy between the falling cost of genomic sequencing and the high cost of clinical genetic testing when done on a gene by gene basis. Targeted capture and massively parallel sequencing is already being used in clinical practice for cancer susceptibility testing. At the University of Washington Medical Center, this technology is used for Coloseq, which evaluates 7 colon cancer genes at a list price of $2650. Unfortunately, comprehensive evaluation of all breast and ovarian cancer genes in one panel in the United States is currently limited by the gene patents on BRCA1 and BRCA2.

Conclusion

The genetics of hereditary ovarian cancer is rapidly evolving. Although BRCA1/2 mutations remain the most common cause of hereditary ovarian cancer, recent studies have identified mutations in several new genes. In addition to the identification of RAD51C, RAD51D, and BRIP1 as ovarian cancer susceptibility genes, many of the previously identified breast cancer genes in the FA–BRCA pathway may also increase risk of ovarian cancer. While mutations in these genes are each individually rare, together they make up a significant proportion of ovarian cancers. New advances in genomic technologies will likely accelerate the discovery of additional cancer susceptibility genes and increase the feasibility of comprehensive evaluation of multiple genes simultaneously at low cost. Theoretically, as recognition of genetic risk improves, most inherited risk of ovarian carcinoma could be identified before cancer onset and the proportion of ovarian carcinoma associated with inherited risk could be effectively prevented. In addition, identifying inherited mutations in a variety of FA-genes may aid in identifying individuals who will selectively benefit from PARP inhibitors.

Conflict of interest statement

The authors of this article declare no conflicts of interest.