This executive summary reviews the topics covered in this PDQ summary on the genetics of breast and gynecologic cancers, with hyperlinks to detailed sections below that describe the evidence on each topic.
Factors suggestive of a genetic contribution to both breast cancer and gynecologic cancer include 1) an increased incidence of these cancers among individuals with a family history of these cancers; 2) multiple family members affected with these and other cancers; and 3) a pattern of cancers compatible with autosomal dominant inheritance. Both males and females can inherit and transmit an autosomal dominant cancer predisposition gene.
Additional factors coupled with family history—such as reproductive history, oral contraceptive and hormone replacement use, radiation exposure early in life, alcohol consumption, and physical activity —can influence an individual's risk of developing cancer.
Risk assessment models have been developed to clarify an individual's 1) lifetime risk of developing breast and/or gynecologic cancer; 2) likelihood of having a pathogenic variant in BRCA1 or BRCA2; and 3) likelihood of having a pathogenic variant in one of the mismatch repair genes associated with Lynch syndrome.
Breast and ovarian cancer are present in several autosomal dominant cancer syndromes, although they are most strongly associated with highly penetrant germline pathogenic variants in BRCA1 and BRCA2. Other genes, such as PALB2, TP53 (associated with Li-Fraumeni syndrome), PTEN (associated with Cowden syndrome), CDH1 (associated with diffuse gastric and lobular breast cancer syndrome), and STK11 (associated with Peutz-Jeghers syndrome), confer a risk to either or both of these cancers with relatively high penetrance.
Inherited endometrial cancer is most commonly associated with Lynch syndrome, a condition caused by inherited pathogenic variants in the highly penetrant mismatch repair genes MLH1, MSH2, MSH6, PMS2, and EPCAM. Colorectal cancer (and, to a lesser extent, ovarian cancer and stomach cancer) is also associated with Lynch syndrome.
Additional genes, such as CHEK2, BRIP1, RAD51, and ATM, are associated with breast and/or gynecologic cancers with moderate penetrance. Genome-wide searches are showing promise in identifying common, low-penetrance susceptibility alleles for many complex diseases, including breast and gynecologic cancers, but the clinical utility of these findings remains uncertain.
Breast cancer screening strategies, including breast magnetic resonance imaging and mammography, are commonly performed in carriers of BRCA pathogenic variants and in individuals at increased risk of breast cancer. Initiation of screening is generally recommended at earlier ages and at more frequent intervals in individuals with an increased risk due to genetics and family history than in the general population. There is evidence to demonstrate that these strategies have utility in early detection of cancer. In contrast, there is currently no evidence to demonstrate that gynecologic cancer screening using cancer antigen 125 testing and transvaginal ultrasound leads to early detection of cancer.
Risk-reducing surgeries, including risk-reducing mastectomy (RRM) and risk-reducing salpingo-oophorectomy (RRSO), have been shown to significantly reduce the risk of developing breast and/or ovarian cancer and improve overall survival in carriers of BRCA1 and BRCA2 pathogenic variants. Chemoprevention strategies, including the use of tamoxifen and oral contraceptives, have also been examined in this population. Tamoxifen use has been shown to reduce the risk of contralateral breast cancer among carriers of BRCA1 and BRCA2 pathogenic variants after treatment for breast cancer, but there are limited data in the primary cancer prevention setting to suggest that it reduces the risk of breast cancer among healthy female carriers of BRCA2 pathogenic variants. The use of oral contraceptives has been associated with a protective effect on the risk of developing ovarian cancer, including in carriers of BRCA1 and BRCA2 pathogenic variants, with no association of increased risk of breast cancer when using formulations developed after 1975.
Psychosocial factors influence decisions about genetic testing for inherited cancer risk and risk-management strategies. Uptake of genetic testing varies widely across studies. Psychological factors that have been associated with testing uptake include cancer-specific distress and perceived risk of developing breast or ovarian cancer. Studies have shown low levels of distress after genetic testing for both carriers and noncarriers, particularly in the longer term. Uptake of RRM and RRSO also varies across studies and may be influenced by factors such as cancer history, age, family history, recommendations of the health care provider, and pretreatment genetic education and counseling. Patients' communication with their family members about an inherited risk of breast and gynecologic cancer is complex; gender, age, and the degree of relatedness are some elements that affect disclosure of this information. Research is ongoing to better understand and address psychosocial and behavioral issues in high-risk families.
General Information
Many of the medical and scientific terms used in this summary are found in the NCI Dictionary of Genetics Terms. When a linked term is clicked, the definition will appear in a separate window.
A concerted effort is being made within the genetics community to shift terminology used to describe genetic variation. The shift is to use the term "variant" rather than the term "mutation" to describe a genetic difference that exists between the person or group being studied and the reference sequence, particularly for differences that exist in the germline. Variants can then be further classified as benign (harmless), likely benign, of uncertain significance, likely pathogenic, or pathogenic (disease causing). Throughout this summary, we will use the term pathogenic variant to describe a disease-causing mutation. Refer to the Cancer Genetics Overview summary for more information about variant classification.
Many of the genes and conditions described in this summary are found in the Online Mendelian Inheritance in Man (OMIM) catalog. Refer to OMIM for more information.
Among women in the United States, breast cancer is the most commonly diagnosed cancer after nonmelanoma skin cancer, and it is the second leading cause of cancer deaths after lung cancer. In 2022, an estimated 290,560 new cases of breast cancer (including 2,710 cases in men) will be diagnosed, and 43,780 deaths (including 530 deaths in men) will occur.[1] The incidence of breast cancer, particularly for estrogen receptor (ER)–positive cancers occurring after age 50 years, is declining and has declined at a faster rate since 2003; this may be temporally related to a decrease in hormone replacement therapy (HRT) after early reports from the Women's Health Initiative (WHI).[2] An estimated 19,880 new cases of ovarian cancer are expected in the United States in 2022, with an estimated 12,810 deaths. Ovarian cancer is the fifth most deadly cancer in women.[1] An estimated 65,950 new cases of endometrial cancer are expected in the United States in 2022, with an estimated 12,550 deaths.[1] (Refer to the PDQ summaries on Breast Cancer Treatment [Adult]; Ovarian Epithelial, Fallopian Tube, and Primary Peritoneal Cancer Treatment; and Endometrial Cancer Treatment for more information about breast, ovarian, and endometrial cancer rates, diagnosis, and management.)
A possible genetic contribution to both breast and ovarian cancer risk is indicated by the increased incidence of these cancers among women with a family history (refer to the Risk Factors for Breast Cancer, Risk Factors for Ovarian Cancer, and Risk Factors for Endometrial Cancer sections below for more information), and by the observation of some families in which multiple family members are affected with breast and/or ovarian cancer, in a pattern compatible with an inheritance of autosomal dominant cancer susceptibility. Formal studies of families (linkage analysis) have subsequently proven the existence of autosomal dominant predispositions to breast and ovarian cancer and have led to the identification of several highly penetrant genes as the cause of inherited cancer risk in many families. (Refer to the PDQ summary Cancer Genetics Overview for more information about linkage analysis.) Pathogenic variants in these genes are rare in the general population and are estimated to account for no more than 5% to 10% of breast and ovarian cancer cases overall. It is likely that other genetic factors contribute to the etiology of some of these cancers.
Risk Factors for Breast Cancer
Refer to the PDQ summary on Breast Cancer Prevention for information about risk factors for breast cancer in the general population.
Age
The cumulative risk of breast cancer increases with age, with most breast cancers occurring after age 50 years.[3] Breast cancer (and ovarian cancer, to a lesser degree) tends to occur at an earlier age in women with a genetic susceptibility than it does in women with sporadic cases.
Family history including inherited cancer genes
In cross-sectional studies of adult populations, 5% to 10% of women have a mother or sister with breast cancer, and about twice as many have either a first-degree relative (FDR) or a second-degree relative with breast cancer.[4,5,6,7] The risk conferred by a family history of breast cancer has been assessed in case-control and cohort studies, using volunteer and population-based samples, with generally consistent results.[8] In a pooled analysis of 38 studies, the relative risk (RR) of breast cancer conferred by an FDR with breast cancer was 2.1 (95% confidence interval [CI], 2.0–2.2).[8] Risk increases with the number of affected relatives, age at diagnosis, the occurrence of bilateral or multiple ipsilateral breast cancers in a family member, and the number of affected male relatives.[5,6,8,9,10] A large population-based study from the Swedish Family Cancer Database confirmed the finding of a significantly increased risk of breast cancer in women who had a mother or a sister with breast cancer. The hazard ratio (HR) for women with a single breast cancer in the family was 1.8 (95% CI, 1.8–1.9) and was 2.7 (95% CI, 2.6–2.9) for women with a family history of multiple breast cancers. For women who had multiple breast cancers in the family, with one occurring before age 40 years, the HR was 3.8 (95% CI, 3.1–4.8). However, the study also found a significant increase in breast cancer risk if the relative was aged 60 years or older, suggesting that breast cancer at any age in the family carries some increase in risk.[10] Another study in women with unilateral versus contralateral breast cancer (CBC) evaluated breast cancer risk among family members.[11] Results indicated that among women with affected FDRs, CBC risk was 8.1% at 10 years. This risk was higher among relatives diagnosed before age 40 years or with CBC and approached the lower risk estimates among BRCA carriers. (Refer to the Contralateral breast cancer in carriers of BRCA pathogenic variants section in the High-Penetrance Breast and/or Gynecologic Cancer Susceptibility Genes section of this summary for information about cancer risk estimates in that population.) These risk estimates remained unchanged when the analysis was restricted to women who tested negative for a deleterious variant in BRCA1/BRCA2, ATM, CHEK2, and PALB2.
One of the largest studies of twins ever conducted, with 80,309 monozygotic twins and 123,382 dizygotic twins, reported a heritability estimate for breast cancer of 31% (95% CI, 11%–51%).[12] If a monozygotic twin had breast cancer, her twin sister had a 28.1% probability of developing breast cancer (95% CI, 23.9%–32.8%), and if a dizygotic twin had breast cancer, her twin sister had a 19.9% probability of developing breast cancer (95% CI, 17%–23.2%). These estimates suggest a 10% higher risk of breast cancer for monozygotic twins than for dizygotic twins. However, a high rate of discordance even between monozygotic twins suggests that environmental factors also have a role in modifying breast cancer risk.
(Refer to the Penetrance of BRCA pathogenic variants section of this summary for a discussion of familial risk in women from families with BRCA1/BRCA2 pathogenic variants who themselves test negative for the family pathogenic variant.)
Reproductive history
In general, breast cancer risk increases with early menarche and late menopause and is reduced by early first full-term pregnancy. There may be an increased risk of breast cancer in carriers of BRCA1 and BRCA2 pathogenic variants with pregnancy at a younger age (before age 30 y), with a more significant effect seen for carriers of BRCA1 pathogenic variants.[13,14,15] Likewise, breastfeeding can reduce breast cancer risk in carriers of BRCA1 (but not BRCA2) pathogenic variants.[16] Regarding the effect of pregnancy on breast cancer outcomes, neither diagnosis of breast cancer during pregnancy nor pregnancy after breast cancer seems to be associated with adverse survival outcomes in women who carry a BRCA1 or BRCA2 pathogenic variant.[17] Parity appears to be protective for carriers of BRCA1 and BRCA2 pathogenic variants, with an additional protective effect for live birth before age 40 years.[18]
Reproductive history can also affect the risk of ovarian cancer and endometrial cancer. (Refer to the Reproductive History sections in the Risk Factors for Ovarian Cancer and Risk Factors for Endometrial Cancer sections of this summary for more information.)
Oral contraceptives
Oral contraceptives (OCs) may produce a slight increase in breast cancer risk among long-term users, but this appears to be a short-term effect. In a meta-analysis of data from 54 studies, the risk of breast cancer associated with OC use did not vary in relationship to a family history of breast cancer.[19]
OCs are sometimes recommended for ovarian cancer prevention in carriers of BRCA1 and BRCA2 pathogenic variants. (Refer to the Oral Contraceptives section in the Risk Factors for Ovarian Cancer section of this summary for more information.) Although the data are not entirely consistent, a meta-analysis concluded that there was no significant increased risk of breast cancer with OC use in carriers of BRCA1/BRCA2 pathogenic variants.[20] However, use of OCs formulated before 1975 was associated with an increased risk of breast cancer (summary relative risk [SRR], 1.47; 95% CI, 1.06–2.04).[20] (Refer to the Reproductive factors section in the Clinical Management of Carriers of BRCA Pathogenic Variants section of this summary for more information.)
Hormone replacement therapy
Data exist from both observational and randomized clinical trials regarding the association between postmenopausal HRT and breast cancer. A meta-analysis of data from 51 observational studies indicated a RR of breast cancer of 1.35 (95% CI, 1.21–1.49) for women who had used HRT for 5 or more years after menopause.[21] The WHI (NCT00000611), a randomized controlled trial of about 160,000 postmenopausal women, investigated the risks and benefits of HRT. The estrogen-plus-progestin arm of the study, in which more than 16,000 women were randomly assigned to receive combined HRT or placebo, was halted early because health risks exceeded benefits.[22,23] Adverse outcomes prompting closure included significant increase in both total (245 vs. 185 cases) and invasive (199 vs. 150 cases) breast cancers (RR, 1.24; 95% CI, 1.02–1.5, P < . 001) and increased risks of coronary heart disease, stroke, and pulmonary embolism. Similar findings were seen in the estrogen-progestin arm of the prospective observational Million Women's Study in the United Kingdom.[24] The risk of breast cancer was not elevated, however, in women randomly assigned to estrogen-only versus placebo in the WHI study (RR, 0.77; 95% CI, 0.59–1.01). Eligibility for the estrogen-only arm of this study required hysterectomy, and 40% of these patients also had undergone oophorectomy, which potentially could have impacted breast cancer risk.[25]
The association between HRT and breast cancer risk among women with a family history of breast cancer has not been consistent; some studies suggest risk is particularly elevated among women with a family history, while others have not found evidence for an interaction between these factors.[26,27,28,29,30,21] The increased risk of breast cancer associated with HRT use in the large meta-analysis did not differ significantly between subjects with and without a family history.[30] The WHI study has not reported analyses stratified on breast cancer family history, and subjects have not been systematically tested for BRCA1/BRCA2 pathogenic variants.[23] Short-term use of hormones for treatment of menopausal symptoms appears to confer little or no breast cancer risk.[21,31] The effect of HRT on breast cancer risk among carriers of BRCA1 or BRCA2 pathogenic variants has been studied in the context of bilateral risk-reducing oophorectomy, in which short-term replacement does not appear to reduce the protective effect of oophorectomy on breast cancer risk.[32] (Refer to the Hormone replacement therapy in carriers of BRCA1/BRCA2 pathogenic variants section of this summary for more information.)
Hormone use can also affect the risk of developing endometrial cancer. (Refer to the Hormones section in the Risk Factors for Endometrial Cancer section of this summary for more information.)
Radiation exposure
Observations in survivors of the atomic bombings of Hiroshima and Nagasaki and in women who have received therapeutic radiation treatments to the chest and upper body document increased breast cancer risk as a result of radiation exposure. The significance of this risk factor in women with a genetic susceptibility to breast cancer is unclear.
Preliminary data suggest that increased sensitivity to radiation could be a cause of cancer susceptibility in carriers of BRCA1 or BRCA2 pathogenic variants,[33,34,35,36] and in association with germline ATM and TP53 variants.[37,38]
The possibility that genetic susceptibility to breast cancer occurs via a mechanism of radiation sensitivity raises questions about radiation exposure. It is possible that diagnostic radiation exposure, including mammography, poses more risk in genetically susceptible women than in women of average risk. Therapeutic radiation could also pose a carcinogenic risk. A cohort study of carriers of BRCA1 and BRCA2 pathogenic variants treated with breast-conserving therapy, however, showed no evidence of increased radiation sensitivity or sequelae in the breast, lung, or bone marrow of carriers.[39] This finding was confirmed in a retrospective cohort study of 691 patients with BRCA1/BRCA2-associated breast cancer who were followed up for a median of 8.6 years. No association between receiving adjuvant radiation therapy and increased risk of CBC was observed in the entire cohort, including the subset of patients younger than 40 years at primary breast cancer diagnosis.[40] Conversely, radiation sensitivity could make tumors in women with genetic susceptibility to breast cancer more responsive to radiation treatment. Studies examining the impact of radiation exposure, including, but not limited to, mammography, in carriers of BRCA1 and BRCA2 pathogenic variants have had conflicting results.[41,42,43,44,45,46] A large European study showed a dose-response relationship of increased risk with total radiation exposure, but this was primarily driven by nonmammographic radiation exposure before age 20 years.[45] Subsequently, no significant association was observed between prior mammography exposure and breast cancer risk in a prospective study of 1,844 BRCA1 carriers and 502 BRCA2 carriers without a breast cancer diagnosis at time of study entry; average follow-up time was 5.3 years.[46] (Refer to the Mammography section in the Clinical Management of Carriers of BRCA Pathogenic Variants section of this summary for more information about radiation.)
Alcohol intake
The risk of breast cancer increases by approximately 10% for each 10 g of daily alcohol intake (approximately one drink or less) in the general population.[47,48] Prior studies of carriers of BRCA1/BRCA2 pathogenic variants have found no increased risk associated with alcohol consumption.[49,50,51]
Physical activity and anthropometry
Increased physical activity has been associated with reduced breast cancer risk in most epidemiological studies. The reduction of risk has also been seen in studies of women with BRCA1 or BRCA2 pathogenic variants. For example, one study reported a 38% reduction in premenopausal breast cancer risk from moderate physical activity (odds ratio [OR] for the top quartile of physical activity compared with the lowest level, 0.62; 95% CI, 0.40–0.96).[52] The reduction of breast cancer risk has also been seen across the spectrum of absolute breast cancer risk, including in women who are at higher risk for breast cancer on the basis of their breast cancer family history but who do not have known BRCA1 or BRCA2 pathogenic variants.[53]
Benign breast disease and mammographic density
Benign breast disease (BBD) is a risk factor for breast cancer, independent of the effects of other major risk factors for breast cancer (age, age at menarche, age at first live birth, and family history of breast cancer).[54] There may also be an association between BBD and family history of breast cancer.[55]
An increased risk of breast cancer has also been demonstrated for women who have increased density of breast tissue as assessed by mammogram,[54,56,57] and breast density is likely to have a genetic component in its etiology.[58,59,60]
Other factors
Other risk factors, including those that are only weakly associated with breast cancer and those that have been inconsistently associated with the disease in epidemiologic studies (e.g., cigarette smoking), may be important in women who are in specific genotypically defined subgroups. One study [61] found a reduced risk of breast cancer among carriers of BRCA1/BRCA2 pathogenic variants who smoked, but an expanded follow-up study failed to find an association.[62]
Risk Factors for Ovarian Cancer
Refer to the PDQ summary on Ovarian, Fallopian Tube, and Primary Peritoneal Cancer Prevention for information about risk factors for ovarian cancer in the general population.
Age
Ovarian cancer incidence rises in a linear fashion from age 30 years to age 50 years and continues to increase, though at a slower rate, thereafter. Before age 30 years, the risk of developing epithelial ovarian cancer is remote, even in hereditary cancer families.[63]
Family history including inherited cancer genes
Although reproductive, demographic, and lifestyle factors affect risk of ovarian cancer, the single greatest ovarian cancer risk factor is a family history of the disease. A large meta-analysis of 15 published studies estimated an odds ratio of 3.1 for the risk of ovarian cancer associated with at least one FDR with ovarian cancer.[64]
Reproductive history
Nulliparity is consistently associated with an increased risk of ovarian cancer, including among carriers of BRCA/BRCA2 pathogenic variants, yet a meta-analysis identified a risk reduction only in women with four or more live births.[15] Risk may also be increased among women who have used fertility drugs, especially those who remain nulligravid.[65,66] Several studies have reported a risk reduction in ovarian cancer after OC use in carriers of BRCA/BRCA2 pathogenic variants;[67,68,69] a risk reduction has also been shown after tubal ligation in BRCA1 carriers, with a statistically significant decreased risk of 22% to 80% after the procedure.[69,70] Breastfeeding for more than 12 months may also be associated with a reduction in ovarian cancer among carriers of BRCA1/BRCA2 pathogenic variants.[71] On the other hand, evidence is growing that the use of menopausal HRT is associated with an increased risk of ovarian cancer, particularly in long-time users and users of sequential estrogen-progesterone schedules.[72,73,74,75]
Surgical history
Bilateral tubal ligation and hysterectomy are associated with reduced ovarian cancer risk,[65,76,77] including in carriers of BRCA/BRCA2 pathogenic variants.[78] Ovarian cancer risk is reduced more than 90% in women with documented BRCA1 or BRCA2 pathogenic variants who chose risk-reducing salpingo-oophorectomy (RRSO). In this same population, risk-reducing oophorectomy also resulted in a nearly 50% reduction in the risk of subsequent breast cancer.[79,80] While some studies have shown more benefit for breast cancer reduction in patients with BRCA2 versus BRCA1 pathogenic variants, others have shown no benefit for BRCA1 carriers. Additionally, many of the studies remain underpowered to demonstrate benefit.[81] (Refer to the RRSO section of this summary for more information about these studies.)
Oral contraceptives
Use of OCs for 4 or more years is associated with an approximately 50% reduction in ovarian cancer risk in the general population.[65,82] A majority of, but not all, studies also support OCs being protective among carriers of BRCA/BRCA2 pathogenic variants.[70,83,84,85,86] A meta-analysis of 18 studies including 13,627 carriers of BRCA pathogenic variants reported a significantly reduced risk of ovarian cancer (SRR, 0.50; 95% CI, 0.33–0.75) associated with OC use.[20] (Refer to the Oral contraceptives section in the Chemoprevention section of this summary for more information.)
Risk Factors for Endometrial Cancer
Refer to the PDQ summary on Endometrial Cancer Prevention for information about risk factors for endometrial cancer in the general population.
Age
Age is an important risk factor for endometrial cancer. Most women with endometrial cancer are diagnosed after menopause. Only 15% of women are diagnosed with endometrial cancer before age 50 years, and fewer than 5% are diagnosed before age 40 years.[87] Women with Lynch syndrome tend to develop endometrial cancer at an earlier age, with the median age at diagnosis of 48 years.[88]
Family history including inherited cancer genes
Although the hyperestrogenic state is the most common predisposing factor for endometrial cancer, family history also plays a significant role in a woman's risk for disease. Approximately 3% to 5% of uterine cancer cases are attributable to a hereditary cause,[89] with the main hereditary endometrial cancer syndrome being Lynch syndrome, an autosomal dominant genetic condition with a population prevalence of 1 in 300 to 1 in 1,000 individuals.[90,91] (Refer to the Lynch Syndrome section in the PDQ summary on Genetics of Colorectal Cancer for more information.)
Non-Lynch syndrome genes may also contribute to endometrial cancer risk. In an unselected endometrial cancer cohort undergoing multigene panel testing, approximately 3% of patients tested positive for a germline pathogenic variant in non-Lynch syndrome genes, including CHEK2, APC, ATM, BARD1, BRCA1, BRCA2, BRIP1, NBN, PTEN, and RAD51C.[92] Notably, patients with pathogenic variants in non-Lynch syndrome genes were more likely to have serous tumor histology than were patients without pathogenic variants. Furthermore, although the overall risk of endometrial cancer after RRSO was not increased among carriers of BRCA1 pathogenic variants, these patients seemed to have an increased risk of serous and serous-like endometrial cancer.[93] These findings were supported by a Dutch multicenter cohort study in women with germline BRCA1 and BRCA2 pathogenic variants. This study concluded that the absolute risk of endometrial cancer was approximately 3%. Because some serous and p53-aberrant endometrial cancers may harbor germline or somatic BRCA1/BRCA2 mutations, poly (ADP-ribose) polymerase (PARP) inhibitor therapy may also be a therapeutic option.[94]
Reproductive history
Reproductive factors such as multiparity, late menarche, and early menopause decrease the risk of endometrial cancer because of the lower cumulative exposure to estrogen and the higher relative exposure to progesterone.[95,96]
Hormones
Hormonal factors that increase the risk of type I endometrial cancer are better understood. All endometrial cancers share a predominance of estrogen relative to progesterone. Prolonged exposure to estrogen or unopposed estrogen increases the risk of endometrial cancer. Endogenous exposure to estrogen can result from obesity, polycystic ovary syndrome, and nulliparity, while exogenous estrogen can result from taking unopposed estrogen or tamoxifen. Unopposed estrogen increases the risk of developing endometrial cancer by twofold to twentyfold, proportional to the duration of use.[97,98] Tamoxifen, a selective estrogen receptor modulator, acts as an estrogen agonist on the endometrium while acting as an estrogen antagonist in breast tissue, and increases the risk of endometrial cancer.[99] In contrast, OCs, the levonorgestrel-releasing intrauterine system, and combination estrogen-progesterone hormone replacement therapy all reduce the risk of endometrial cancer through the antiproliferative effect of progesterone acting on the endometrium.[100,101,102,103]
Autosomal Dominant Inheritance of Breast and Gynecologic Cancer Predisposition
Autosomal dominant inheritance of breast and gynecologic cancers is characterized by transmission of cancer predisposition from generation to generation, through either the mother's or the father's side of the family, with the following characteristics:
Breast and ovarian cancer are components of several autosomal dominant cancer syndromes. The syndromes most strongly associated with both cancers are the syndromes associated with BRCA1 or BRCA2 pathogenic variants. Breast cancer is also a common feature of Li-Fraumeni syndrome due to TP53 pathogenic variants and of Cowden syndrome due to PTEN pathogenic variants.[104] Other genetic syndromes that may include breast cancer as an associated feature include heterozygous carriers of the ataxia telangiectasia gene and Peutz-Jeghers syndrome. Ovarian cancer has also been associated with Lynch syndrome, basal cell nevus (Gorlin) syndrome, and multiple endocrine neoplasia type 1.[104] Lynch syndrome is mainly associated with colorectal cancer and endometrial cancer, although several studies have demonstrated that patients with Lynch syndrome are also at risk of developing transitional cell carcinoma of the ureters and renal pelvis; cancers of the stomach, small intestine, liver and biliary tract, brain, breast, prostate, and adrenal cortex; and sebaceous skin tumors (Muir-Torre syndrome).[105,106,107,108,109,110,111]
Germline pathogenic variants in the genes responsible for these autosomal dominant cancer syndromes produce different clinical phenotypes of characteristic malignancies and, in some instances, associated nonmalignant abnormalities.
The family characteristics that suggest hereditary cancer predisposition include the following:
Figure 1 and Figure 2 depict some of the classic inheritance features of a BRCA1 and BRCA2 pathogenic variant, respectively. Figure 3 depicts a classic family with Lynch syndrome. (Refer to the Standard Pedigree Nomenclature figure in the PDQ summary on Cancer Genetics Risk Assessment and Counseling for definitions of the standard symbols used in these pedigrees.)
Figure 1. BRCA1 pedigree. This pedigree shows some of the classic features of a family with a BRCA1 pathogenic variant across three generations, including affected family members with breast cancer or ovarian cancer and a young age at onset. BRCA1 families may exhibit some or all of these features. As an autosomal dominant syndrome, a BRCA1 pathogenic variant can be transmitted through maternal or paternal lineages, as depicted in the figure.
Figure 2. BRCA2 pedigree. This pedigree shows some of the classic features of a family with a BRCA2 pathogenic variant across three generations, including affected family members with breast (including male breast cancer), ovarian, pancreatic, or prostate cancers and a relatively young age at onset. BRCA2 families may exhibit some or all of these features. As an autosomal dominant syndrome, a BRCA2 pathogenic variant can be transmitted through maternal or paternal lineages, as depicted in the figure.
Figure 3. Lynch syndrome pedigree. This pedigree shows some of the classic features of a family with Lynch syndrome, including affected family members with colon cancer or endometrial cancer, a young age at onset in some individuals, and incomplete penetrance. Lynch syndrome families may exhibit some or all of these features. Lynch syndrome families may also include individuals with other gastrointestinal, gynecologic, and genitourinary cancers, or other extracolonic cancers. As an autosomal dominant syndrome, Lynch syndrome can be transmitted through maternal or paternal lineages, as depicted in the figure. Because the cancer risk is not 100%, individuals who have Lynch syndrome may not develop cancer, such as the mother of the female with colon cancer diagnosed at age 37 years in this pedigree (called incomplete penetrance).
There are no pathognomonic features distinguishing breast and ovarian cancers occurring in carriers of BRCA1 or BRCA2 pathogenic variants from those occurring in noncarriers. Breast cancers occurring in carriers of BRCA1 pathogenic variants are more likely to be ER-negative, progesterone receptor–negative, HER2/neu receptor–negative (i.e., triple-negative breast cancers [TNBC]), and have a basal phenotype. BRCA1-associated ovarian cancers are more likely to be high-grade and of serous histopathology. (Refer to the Pathology of breast cancer and Pathology of ovarian cancer sections of this summary for more information.)
Some pathologic features distinguish carriers of Lynch syndrome–associated pathogenic variants from noncarriers. The hallmark feature of endometrial cancers occurring in Lynch syndrome is mismatch repair (MMR) deficiencies, including the presence of microsatellite instability (MSI), and the absence of specific MMR proteins. In addition to these molecular changes, there are also histologic changes including tumor-infiltrating lymphocytes, peritumoral lymphocytes, undifferentiated tumor histology, lower uterine segment origin, and synchronous tumors.
Considerations in Risk Assessment and in Identifying a Family History of Breast and Ovarian Cancer Risk
The accuracy and completeness of family histories must be considered when they are used to assess risk. A reported family history may be erroneous, or a person may be unaware of relatives affected with cancer. In addition, small family sizes and premature deaths may limit the information obtained from a family history. Breast or ovarian cancer on the paternal side of the family usually involves more distant relatives than does breast or ovarian cancer on the maternal side, so information may be more difficult to obtain. When self-reported information is compared with independently verified cases, the sensitivity of a history of breast cancer is relatively high, at 83% to 97%, but lower for ovarian cancer, at 60%.[112,113] Additional limitations of relying on family histories include adoption; families with a small number of women; limited access to family history information; and incidental removal of the uterus, ovaries, and/or fallopian tubes for noncancer indications. Family histories will evolve; therefore, it is important to update family histories from both parents over time. (Refer to the Accuracy of the family history section in the PDQ summary on Cancer Genetics Risk Assessment and Counseling for more information.)
Models for Prediction of Breast and Gynecologic Cancer Risk
Models to predict an individual's lifetime risk of developing breast and/or gynecologic cancer are available.[114,115,116,117] In addition, models exist to predict an individual's likelihood of having a pathogenic variant in BRCA1, BRCA2, or one of the MMR genes associated with Lynch syndrome. (Refer to the Models for prediction of the likelihood of a BRCA1 or BRCA2 pathogenic variant section of this summary for more information about some of these models.) Not all models can be appropriately applied to all patients. Each model is appropriate only when the patient's characteristics and family history are similar to those of the study population on which the model was based. Different models may provide widely varying risk estimates for the same clinical scenario, and the validation of these estimates has not been performed for many models.[115,118,119]
Breast cancer risk assessment models
In general, breast cancer risk assessment models are designed for two types of populations: 1) women without a pathogenic variant or strong family history of breast or ovarian cancer; and 2) women at higher risk because of a personal or family history of breast cancer or ovarian cancer.[119] Models designed for women of the first type (e.g., the Gail model, which is the basis for the Breast Cancer Risk Assessment Tool [BCRAT]) [120], and the Colditz and Rosner model [121]) require only limited information about family history (e.g., number of FDRs with breast cancer). Models designed for women at higher risk require more detailed information about personal and family cancer history of breast and ovarian cancers, including ages at onset of cancer and/or carrier status of specific breast cancer-susceptibility alleles. The genetic factors used by the latter models differ, with some assuming one risk locus (e.g., the Claus model [122]), others assuming two loci (e.g., the International Breast Cancer Intervention Study [IBIS] model [123] and the BRCAPRO model [124]), and still others assuming an additional polygenic component in addition to multiple loci (e.g., the Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm [BOADICEA] model [125,126,127]). The models also differ in whether they include information about nongenetic risk factors. Three models (Gail/BCRAT, Pfeiffer,[117] and IBIS) include nongenetic risk factors but differ in the risk factors they include (e.g., the Pfeiffer model includes alcohol consumption, whereas the Gail/BCRAT does not). These models have limited ability to discriminate between individuals who are affected and those who are unaffected with cancer; a model with high discrimination would be close to 1, and a model with little discrimination would be close to 0.5; the discrimination of the models currently ranges between 0.56 and 0.63).[128] The existing models generally are more accurate in prospective studies that have assessed how well they predict future cancers.[119,129,130,131] An analysis comparing the 10-year performance of the BOADICEA, BRCAPRO, BCRAT, and IBIS models demonstrated superiority of the models with more detailed pedigree inclusion—specifically, BOADICEA and IBIS.[132]
In the United States, BRCAPRO, the Claus model,[122,133] and the Gail/BCRAT [120] are widely used in clinical counseling. Risk estimates derived from the models differ for an individual patient. Several other models that include more detailed family history information are also in use and are discussed below.
Additional considerations for clinical use of breast cancer risk assessment models
The Gail model is the basis for the BCRAT, a computer program available from the National Cancer Institute (NCI) by calling the Cancer Information Service at 1-800-4-CANCER (1-800-422-6237). This version of the Gail model estimates only the risk of invasive breast cancer. The Gail/BCRAT model has been found to be reasonably accurate at predicting breast cancer risk in large groups of White women who undergo annual screening mammography; however, reliability varies depending on the cohort studied.[134,135,136,137,138,139] Risk can be overestimated in the following populations:
The Gail/BCRAT model is valid for women aged 35 years and older. The model was primarily developed for White women.[138] Extensions of the Gail model for African American women have been subsequently developed to calibrate risk estimates using data from more than 1,600 African American women with invasive breast cancer and more than 1,600 controls.[140] Additionally, extensions of the Gail model have incorporated high-risk single nucleotide variants and pathogenic variants; however, no software exists to calculate risk in these extended models.[141,142] Other risk assessment models incorporating breast density have been developed but are not ready for clinical use.[143,144]
Generally, the Gail/BCRAT model should not be the sole model used for families with one or more of the following characteristics:
Commonly used models that incorporate family history include the IBIS, BOADICEA, and BRCAPRO models. The IBIS/Tyrer-Cuzick model incorporates both genetic and nongenetic factors.[123] A three-generation pedigree is used to estimate the likelihood that an individual carries either a BRCA1/BRCA2 pathogenic variant or a hypothetical low-penetrance gene. In addition, the model incorporates personal risk factors such as parity, body mass index (BMI); height; and age at menarche, first live birth, menopause, and HRT use. Both genetic and nongenetic factors are combined to develop a risk estimate. The BOADICEA model examines family history to estimate breast cancer risk and also incorporates both BRCA1/BRCA2 and non-BRCA1/BRCA2 genetic risk factors.[126] The most important difference between BOADICEA and the other models using information on BRCA1/BRCA2 is that BOADICEA assumes an additional polygenic component in addition to multiple loci,[125,126,127] which is more in line with what is known about the underlying genetics of breast cancer. The BOADICEA model has also been expanded to include additional pathogenic variants, including CHEK2, ATM, and PALB2.[145] However, the discrimination and calibration for these models differ significantly when compared in independent samples;[129] the IBIS and BOADICEA models are more comparable when estimating risk over a shorter fixed time horizon (e.g., 10 years),[129] than when estimating remaining lifetime risk. As all risk assessment models for cancers are typically validated over a shorter time horizon (e.g., 5 or 10 years), fixed time horizon estimates rather than remaining lifetime risk may be more accurate and useful measures to convey in a clinical setting.
In addition, readily available models that provide information about an individual woman's risk in relation to the population-level risk depending on her risk factors may be useful in a clinical setting (e.g., Your Disease Risk). Although this tool was developed using information about average-risk women and does not calculate absolute risk estimates, it still may be useful when counseling women about prevention. Risk assessment models are being developed and validated in large cohorts to integrate genetic and nongenetic data, breast density, and other biomarkers.
Although most breast cancer risk models have been shown to be well calibrated overall, model performance can be different for subgroups of women. In particular, independent, prospective validation of risk models for women who tested negative for BRCA1 or BRCA2 pathogenic variants supported that the most commonly used clinical risk models underpredicted risk for this group of women.[146] The performance also differed on the basis of whether the test results of relatives were known. The models also underpredicted risk by 26.3% to 56.7% in women who tested negative but whose relatives had not been tested.
Ovarian cancer risk assessment models
Two risk prediction models have been developed for ovarian cancer.[116,117] The Rosner model [116] included age at menopause, age at menarche, oral contraception use, and tubal ligation; the concordance statistic was 0.60 (0.57–0.62). The Pfeiffer model [117] included oral contraceptive use, menopausal hormone therapy use, and family history of breast cancer or ovarian cancer, with a similar discriminatory power of 0.59 (0.56–0.62). Although both models were well calibrated, their modest discriminatory power limited their screening potential.
Endometrial cancer risk assessment models
The Pfeiffer model has been used to predict endometrial cancer risk in the general population.[117] For endometrial cancer, the relative risk model included BMI, menopausal hormone therapy use, menopausal status, age at menopause, smoking status, and OC use. The discriminatory power of the model was 0.68 (0.66–0.70); it overestimated observed endometrial cancers in most subgroups but underestimated disease in women with the highest BMI category, in premenopausal women, and in women taking menopausal hormone therapy for 10 years or more.
In contrast, MMRpredict, PREMM5 (PREdiction Model for gene Mutations), and MMRpro are three quantitative predictive models used to identify individuals who may potentially have Lynch syndrome.[147,148,149] MMRpredict incorporates only colorectal cancer patients but does include MSI and immunohistochemistry (IHC) tumor testing results. PREMM5 is an update of PREMM(1,2,6) and includes each of the five genes associated with Lynch syndrome, including PMS2 and EPCAM. It accounts for other Lynch syndrome–associated tumors but does not include tumor testing results.[148] MMRpro incorporates tumor testing and germline testing results, but is more time intensive because it includes affected and unaffected individuals in the risk-quantification process. All three predictive models are comparable to the traditional Amsterdam and Bethesda criteria in identifying individuals with colorectal cancer who carry MMR gene pathogenic variants.[150] However, because these models were developed and validated in colorectal cancer patients, the discriminative abilities of these models to identify Lynch syndrome are lower among individuals with endometrial cancer than among those with colon cancer.[151] In fact, the sensitivity and specificity of MSI and IHC in identifying carriers of pathogenic variants are considerably higher than the prediction models and support the use of molecular tumor testing to screen for Lynch syndrome in women with endometrial cancer.
Table 1 summarizes salient aspects of breast and gynecologic cancer risk assessment models that are commonly used in the clinical setting. These models differ by the extent of family history included, whether nongenetic risk factors are included, and whether carrier status and polygenic risk are included (inputs to the models). The models also differ in the type of risk estimates that are generated (outputs of the models). These factors may be relevant in choosing the model that best applies to a particular individual.
Model | Family History (input) | Pathogenic Variants (input) | Risk Factors (input) | Risk Estimate Generated (output) |
---|---|---|---|---|
BCRAT = Breast Cancer Risk Assessment Tool; BOADICEA = Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm; IBIS = International Breast Cancer Intervention Study; PREMM = PREdiction Model for gene Mutations. | ||||
a High risk is defined as those with a personal or family history of the designated cancer type. | ||||
b Takes into account polygenes as an underlying assumption of the model. | ||||
Breast Cancer Risk Assessment Models | ||||
Models for Average-Risk Women | ||||
Gail/BCRAT | First-degree relatives (breast cancer) | No | Yes | Breast cancer |
Pfeiffer (breast)[117] | First-degree relatives (breast, ovarian cancers) | No | Yes | Breast cancer |
Colditz and Rosner[121] | None | No | Yes | Breast cancer |
Models for High-Risk Womena | ||||
Claus[122] | Multigenerational (breast cancer) | No | No | Breast cancer |
BRCAPRO | Multigenerational (breast, ovarian cancers) | BRCA1/BRCA2 | No | Breast cancer; % risk of carryingBRCA1/BRCA2pathogenic variant |
IBIS | Multigenerational (ovarian cancer) | BRCA1/BRCA2 | Yes | Breast cancer; % risk of carryingBRCA1/BRCA2pathogenic variant |
BOADICEAb | Multigenerational (pancreatic, breast, ovarian cancers) | BRCA1/BRCA2 | No | Breast and ovarian cancer; % risk of carryingBRCA1/BRCA2pathogenic variant |
Ovarian Cancer Risk Assessment Models | ||||
Models for Average-Risk Women | ||||
Rosner[116] | None | No | Yes | Ovarian cancer |
Pfeiffer (ovarian)[117] | First-degree relatives (breast, ovarian cancers) | No | Yes | Breast cancer |
Models for High-Risk Womena | ||||
BOADICEAb | Multigenerational (pancreatic, breast, ovarian cancers) | BRCA1/BRCA2 | No | Breast and ovarian cancer; % risk of carryingBRCA1/BRCA2pathogenic variant |
Endometrial Cancer Risk Assessment Models | ||||
Models for Average-Risk Women | ||||
Pfeiffer (endometrial)[117] | None | No | Yes | Endometrial cancer |
Models for High-Risk Womena | ||||
PREMM5 | Multigenerational (colon, endometrial and other Lynch syndrome–associated cancers and polyps) | No | No | % risk of carryingMLH1,MSH2,MSH6pathogenic variant |
MMRpro | Multigenerational (colon, endometrial cancers) | No | No | % risk of carryingMLH1,MSH2,MSH6pathogenic variant |
MMRpredict[147] | Multigenerational (colon, endometrial cancers) | No | No | % risk of carryingMLH1,MSH2,MSH6pathogenic variant |
Considerations When Conducting Genetic Testing
Indications for genetic testing
Several professional organizations and expert panels— including the American Society of Clinical Oncology,[152] the National Comprehensive Cancer Network (NCCN),[153] the American Society of Human Genetics,[154] the American College of Medical Genetics and Genomics,[155] the National Society of Genetic Counselors,[155] the U.S. Preventive Services Task Force,[156] and the Society of Gynecologic Oncologists —[157] have developed clinical criteria and practice guidelines that can be helpful to health care providers in identifying individuals who may have a BRCA1 or BRCA2 pathogenic variant.
In 2019, the American Society of Breast Surgeons published a recommendation to make genetic testing for "BRCA1/BRCA2, and PALB2, with other genes as appropriate for the clinical scenario and family history" available to all breast cancer patients.[158] This recommendation was based on a study that suggested similar pathogenic variant rates identified through an extended multigene panel in patients with breast cancer who did or did not meet the NCCN guidelines for genetic testing.[159] This study had important methodologic challenges that need to be considered, including exclusion of participants previously tested, uncertain accuracy of the reported risk criteria for study participants, inclusion of genes with uncertain management guidelines, and difference in the specific genes in which pathogenic or likely pathogenic variants were identified across the two groups. For example, there was a statistically significant difference between participants who met and who did not meet NCCN criteria in the detection of BRCA1/BRCA2 variants.
Other studies have also found that the NCCN criteria have good sensitivity when predicting BRCA1/BRCA2 variants; however, less is known about many other genes. For example, one study showed that the NCCN criteria were able to detect 88.9% of the BRCA1/BRCA2 pathogenic variant carriers [160] and others have found that, if more than one NCCN criterion is met, then the positive predictive value does pass the 10% threshold (e.g., 12% for more than two NCCN criteria).[161]
As the cost of genetic testing continues to decrease, there is a need for unbiased evidence to guide indications for testing, including the cost-benefit impact on screening, prevention, and treatment. Efforts to generate less biased evidence include a single institution study of 3,907 unselected women with breast cancer tested for nine breast cancer genes, including BRCA1/BRCA2, ATM, CDH1, CHEK2, NF1, PALB2, PTEN, and TP53.[162] The study assessed the relative performance of NCCN genetic testing criteria as compared with the American Society of Breast Surgeons' recommendation to test all women aged 65 years or younger with breast cancer. The sensitivity of the criteria was defined as the proportion of individuals who met testing criteria and tested positive for a pathogenic or likely pathogenic variant of the total population of pathogenic or likely pathogenic variant carriers in the study, while the specificity was defined as the proportion of individuals who did not meet testing criteria and tested negative for a pathogenic or likely pathogenic variant of the total population of noncarriers in the study. High sensitivity and specificity are both important considerations; however, higher sensitivity leads to lower specificity, so it is important to balance these two factors. Detection of BRCA1/BRCA2 pathogenic or likely pathogenic variants based on NCCN criteria had a sensitivity of 87% with a specificity of 53.5%; when expanded to the nine genes included in the study, sensitivity was 70% and specificity was 53.2%. When including all women diagnosed with breast cancer at age 65 or younger, the sensitivity to detect BRCA1/BRCA2 pathogenic or likely pathogenic variants increased to 98%, while the specificity dropped to 22%. Among those who did not meet NCCN criteria, 0.7% had pathogenic or likely pathogenic BRCA1/BRCA2 variants.
Another study to assess frequency of pathogenic or likely pathogenic variants among breast cancer patients included a nested case-control study conducted through the Women's Health Initiative (WHI) cohort among women with (cases) and without (controls) invasive breast cancer. Participants were tested for pathogenic or likely pathogenic variants in ten breast cancer–associated genes, including BRCA1/BRCA2.[163] The prevalence of pathogenic or likely pathogenic BRCA1/BRCA2 variants among those diagnosed with invasive breast cancer before age 65 years was 2.21%, compared with 1.09% among those diagnosed at age 65 years or older. In comparison, the frequency of pathogenic or likely pathogenic BRCA1/BRCA2 variants was 0.22% in the control group. Current genetic testing criteria detect BRCA pathogenic variants. Although higher sensitivity is always desired, it is at the expense of specificity. Lower specificity leads to higher costs to achieve one positive genetic test.
Benefits of offering genetic testing at the time of cancer diagnosis
At the time of a new cancer diagnosis, genetic testing for inherited cancer predisposition may guide patient care including decisions about surgery, chemotherapy and other biologics, and radiation treatment.[164,165] Among high-risk patients, the option of genetic testing is an important part of the shared decision-making process regarding cancer treatments at the time of diagnosis. Tools are available to facilitate decision making about genetic testing in this context.[166]
Breast cancer diagnosis
Benefits of offering genetic testing at the time of breast cancer diagnosis include, but are not limited to, the following:
Ovarian cancer diagnosis
Benefits of offering genetic testing at the time of ovarian cancer diagnosis include, but are not limited to, the following:
Endometrial cancer diagnosis
Benefits of offering genetic testing at the time of endometrial cancer diagnosis include, but are not limited to, the following:
Multigene (panel) testing
Since the availability of next-generation sequencing and the Supreme Court of the United States ruling that human genes cannot be patented, several clinical laboratories now offer genetic testing through multigene panels at a cost comparable to that of single-gene testing. Even testing for BRCA1 and BRCA2 is a limited panel test of two genes. Approximately 25% of all ovarian/fallopian tube/peritoneal cancers are caused by a heritable genetic condition. Of these, about one-quarter (6% of all ovarian/fallopian tube/peritoneal cancers) are caused by genes other than BRCA1 and BRCA2, including many genes associated with the Fanconi anemia pathway or otherwise involved with homologous recombination.[180] In a population of ovarian cancer patients who test negative for BRCA1 and BRCA2 pathogenic variants, multigene panel testing can reveal actionable pathogenic variants.[181,182,183]
In general, multigene panel testing increases the yield of non-BRCA pathogenic variants across a variety of populations.[165,184,185,186] In an unselected population of breast cancer patients, the prevalence of BRCA1 and BRCA2 pathogenic variants was 6.1%, while the prevalence of pathogenic variants in other breast/ovarian cancer–predisposing genes was 4.6%.[187] In an unselected population of endometrial cancer patients, the prevalence of Lynch syndrome pathogenic variants (MLH1, MSH2, EPCAM-MSH2, MSH6, and PMS2) was 5.8%; the prevalence of pathogenic variants in other actionable genes was 3.4%.[92] Similarly, in a study of 35,409 women with breast cancer tested with the Myriad 25-gene panel, a pathogenic variant was found in 9.3% of women.[188] Among that 9.3%, 48.5% of the women carried a pathogenic variant in BRCA1 or BRCA2. The majority of other breast cancer genes with pathogenic variants identified included CHEK2 (11.7%), ATM (9.7%), and PALB2 (9.3%). The prevalence of pathogenic variants in the other breast cancer genes on the panel ranged from 0.05% to 0.31%. Pathogenic variants in Lynch syndrome genes accounted for 7.0% of variants identified; 3.7% were found in other genes included in the panel. The rate of pathogenic variants was higher in women with TNBC diagnosed before age 40 years. A similar trend of identifying pathogenic variants in non-BRCA susceptibility genes in male breast cancer patients has also been described.[189] In two studies of women who had previously tested negative for BRCA1/BRCA2, reflex testing with a multigene panel identified pathogenic variants in additional genes among 8% to 11% of cases.[190,191] In a study of 77,085 patients with breast cancer and 6,001 patients with ovarian cancer, 24.1% and 30.9% had genetic testing, respectively. Of those tested, pathogenic or likely pathogenic variants were identified in 7.8% of patients with breast cancer and 14.5% of patients with ovarian cancer. Prevalent non-BRCA pathogenic variants identified in patients with breast cancer included CHEK2 (1.6%), PALB2 (1.0%), ATM (0.7%), and NBN (0.4%). In patients with ovarian cancer, non-BRCA pathogenic variants included CHEK2 (1.4%), BRIP1 (0.9%), MSH2 (0.8%), and ATM (0.6%).[192] The potential utility of genetic testing in patients with ovarian tumors of all histologies was suggested in a study using a 32-gene panel that found 13.2% of 4,439 tumors harbored a pathogenic variant. Rates were highest among those with serous ovarian carcinoma (14.7%), although likely pathogenic variants were also seen in those with other histologies (borderline, germ cell, and sex cord stromal tumors), the significance of which is unclear to clinical management or etiology of disease.[193]
Women diagnosed with TNBC are currently recommended to undergo BRCA1/BRCA2 testing to guide treatment decisions at any age.[153] A large study utilizing multigene (panel) testing comprising two separate cohorts reported that, in addition to BRCA1/BRCA2 genes, six other breast cancer susceptibility genes were also related to a higher risk of TNBC. Specifically, pathogenic variants in BARD1, PALB2, and RAD51D, in addition to BRCA1 and BRCA2, were each associated with more than a fivefold increase in breast cancer.[194] Pathogenic variants in three other genes —BRIP1, RAD51C, and TP53— were each associated with an increased TNBC risk of more than twofold. Pathogenic variants in these eight genes were reported in 12% of the TNBC cases (8.3% BRCA1/BRCA2, 3.7% non-BRCA1/BRCA2). The study was conducted in a clinical testing cohort of 140,449 individuals (8,753 TNBC cases) who received genetic testing using a 21-gene panel (sample A). In addition, a second sample (sample B) examined gene frequency rates in a pooled consortium of 2,143 individuals using a 17-gene panel. The overall frequency of pathogenic variants in the 21 genes examined in sample A was 14.4% (8.4% BRCA1/BRCA2, 6.0% non-BRCA1/BRCA2). The two samples had very consistent findings with respect to the risk estimates despite differences in age, race, ethnicity, and family history of cancer with sample A being younger, more racially and ethnically diverse, and more likely to have a family history of cancer. The pathogenic variant frequency detection in these 21 genes was also similar for White individuals (14% overall, 7.8% BRCA1/BRCA2, 6.2% non-BRCA1/BRCA2) and African American individuals (14.6% overall, 9.0% BRCA1/BRCA2, 5.6% non-BRCA1/BRCA2) supporting that the higher rates of TNBC in African American individuals versus White individuals is driven by environmental factors.
There are caveats of multigene testing. Genes identified as part of multigene panel testing can be associated with varied breast cancer risk or confer no known risk.[183] There is also the possibility of finding a variant of uncertain significance (VUS). Even within a given gene, there may be differential risks on the basis of specific pathogenic variants.[195] A large population-based retrospective study using Surveillance, Epidemiology, and End Results (SEER) program data from Georgia and Los Angeles, California, found that multigene testing led to a twofold increase in the detection of pathogenic variants compared with BRCA-only testing in women with breast cancer.[196] VUS rates, however, were tenfold higher in the multigene panels, especially in African American women (44.5%) and Asian women (50.9%). Many centers now offer a multigene panel test instead of just BRCA1 and BRCA2 testing if there is a concerning family history of syndromes other than hereditary breast and ovarian cancer, or more importantly, to gain as much genetic information as possible with one test, particularly if there may be insurance limitations.
(Refer to the Multigene [panel] testing section in the PDQ summary on Cancer Genetics Risk Assessment and Counseling for more information about multigene testing, including genetic education and counseling considerations and research examining the use of multigene testing.)
References:
The proportion of individuals carrying a pathogenic variant who will manifest a certain disease is referred to as penetrance. In general, common genetic variants that are associated with cancer susceptibility have a lower penetrance than rare genetic variants. This is depicted in Figure 4. For adult-onset diseases, penetrance is usually described by the individual carrier's age, sex, and organ site. For example, the penetrance for breast cancer in female carriers of BRCA1 pathogenic variants is often quoted by age 50 years and by age 70 years. Of the numerous methods for estimating penetrance, none are without potential biases, and determining an individual carrier's risk of cancer involves some level of imprecision.
Figure 4. Genetic architecture of cancer risk. This graph depicts the general finding of a low relative risk associated with common, low-penetrance genetic variants, such as single-nucleotide polymorphisms identified in genome-wide association studies, and a higher relative risk associated with rare, high-penetrance genetic variants, such as pathogenic variants in the BRCA1/BRCA2 genes associated with hereditary breast and ovarian cancer and the mismatch repair genes associated with Lynch syndrome.
Throughout this summary, we discuss studies that report on relative and absolute risks. These are two important but different concepts. Relative risk (RR) refers to an estimate of risk relative to another group (e.g., risk of an outcome like breast cancer for women who are exposed to a risk factor relative to the risk of breast cancer for women who are unexposed to the same risk factor). RR measures that are greater than 1 mean that the risk for those captured in the numerator (i.e., the exposed) is higher than the risk for those captured in the denominator (i.e., the unexposed). RR measures that are less than 1 mean that the risk for those captured in the numerator (i.e., the exposed) is lower than the risk for those captured in the denominator (i.e., the unexposed). Measures with similar relative interpretations include the odds ratio (OR), hazard ratio, and risk ratio.
Absolute risk measures consider the number of people who have a particular outcome, the number of people in a population who could have the outcome, and person-time (the period of time during which an individual was at risk of having the outcome). Absolute risk measures also reflect the absolute burden of an outcome in a population. Absolute measures include risks and rates and can be expressed over a specific time frame (e.g., 1 year, 5 years) or overall lifetime. Cumulative risk is a measure of risk that occurs over a defined time period. For example, overall lifetime risk is a type of cumulative risk that is usually calculated on the basis of a given life expectancy (e.g., 80 or 90 years). Cumulative risk can also be presented over other time frames (e.g., up to age 50 years).
Large relative risk measures do not mean that there will be large effects in the actual number of individuals at a population level because the disease outcome may be quite rare. For example, the relative risk for smoking is much higher for lung cancer than for heart disease, but the absolute difference between smokers and nonsmokers is greater for heart disease, the more-common outcome, than for lung cancer, the more-rare outcome.
Therefore, in evaluating the effect of exposures and biological markers on disease prevention across the continuum, it is important to recognize the differences between relative and absolute effects in weighing the overall impact of a given risk factor. For example, the magnitude is in the range of 30% (e.g., ORs or RRs of 1.3) for many breast cancer risk factors, which means that women with a risk factor (e.g., alcohol consumption, late age at first birth, oral contraceptive use, postmenopausal body size) have a 30% relative increase in breast cancer in comparison with what they would have if they did not have that risk factor. But the absolute increase in risk is based on the underlying absolute risk of disease. Figure 5 and Table 2 show the impact of a relative risk factor in the range of 1.3 on absolute risk. (Refer to the Standard Pedigree Nomenclature figure in the PDQ summary on Cancer Genetics Risk Assessment and Counseling for definitions of the standard symbols used in these pedigrees.) As shown, women with a family history of breast cancer have a much higher benefit from risk factor reduction on an absolute scale.[1]
Figure 5. These five pedigrees depict probands with varying degrees of family history. Table 2 accompanies this figure.
Family History | Lifetime Risk (%) | Lifetime Risk After Risk Factor Modification (%) | Absolute Risk Difference (%) | Relative Risk |
---|---|---|---|---|
a Refer to Figure 5, which accompanies this table. | ||||
Low (Family 1) | 10.9 | 8.4 | 2.50 | 1.29 (29% increased risk) |
Moderate (Family 2) | 21.6 | 16.8 | 4.80 | 1.28 (28% increased risk) |
Moderate/high (Family 3) | 27.1 | 21.3 | 5.80 | 1.27 (27% increased risk) |
High (Family 4) | 32.0 | 25.3 | 6.70 | 1.26 (26% increased risk) |
BRCA1pathogenic variant (Family 5) | 53.7 | 44.2 | 9.50 | 1.21 (21% increased risk) |
With the increasing use of multigene panel tests, a framework for cancer risk management among individuals with pathogenic variants detected in novel genes has been described [2] that incorporates data on age-specific, lifetime, and absolute cancer risks. The framework suggests initiating screening in these individuals at the age when their 5-year cancer risk approaches that at which screening is routinely initiated for women in the general population (approximately 1% for breast cancer in the United States). As a result, the age at which to begin screening will vary depending on the gene. (Refer to the Multigene [panel] testing section in the Introduction section of this summary for more information on multigene panel tests.)
References:
Several genes are found to be associated with the development of breast and/or gynecologic cancers. These genes are categorized as high-penetrance, moderate-penetrance, and low-penetrance in this summary. The high- and moderate-penetrance genes are summarized in Table 3. Low-penetrance genes and loci primarily include polymorphisms that have been associated with cancer susceptibility. (Refer to the High-Penetrance Breast and/or Gynecologic Cancer Susceptibility Genes, Moderate-Penetrance Genes Associated With Breast and/or Gynecologic Cancer, and Single Nucleotide Variant–Associated Cancer Risks sections of this summary for more information.)
Cancer Susceptibilitya | Moderate-Penetrance Genesb | High-Penetrance Genes |
---|---|---|
a Other cancers may be associated with the genes in this table. | ||
b Other genes discussed in the Moderate-Penetrance Genes Associated With Breast and/or Gynecologic Cancersection of this summary but for which penetrance is unknown includeCASP8,TGFB1,Abraxas,RECQL, andSMARCA4. | ||
Breast cancer | ATM,BRIP1,CHEK2,FANCD2,RAD51C | BRCA1,BRCA2,CDH1,PALB2,PTEN,STK11,TP53 |
Ovarian cancer | ATM,BRIP1,EPCAM,MLH1,MSH2,MSH6,RAD51C | BRCA1,BRCA2 |
Endometrial cancer | EPCAM,MLH1,MSH2,MSH6,PMS2,PTEN |
BRCA1andBRCA2
Introduction
Epidemiologic studies have clearly established the role of family history as an important risk factor for both breast and ovarian cancers. After gender and age, a positive family history is the strongest known predictive risk factor for breast cancer. However, it has long been recognized that in some families, there is hereditary breast cancer, which is characterized by an early age of onset, bilaterality, and the presence of breast cancer in multiple generations in an apparent autosomal dominant pattern of transmission (through either the maternal or the paternal lineage), sometimes including tumors of other organs, particularly the ovary and prostate gland.[1,2] It is now known that the cancer in some of these families can be explained by specific pathogenic variants in single cancer susceptibility genes. The isolation of several of these genes, which when altered are associated with a significantly increased risk of breast/ovarian cancer, makes it possible to identify individuals at risk. Although such cancer susceptibility genes are very important, highly penetrant germline pathogenic variants are estimated to account for only 5% to 10% of breast cancers overall.
A 1988 study reported the first quantitative evidence that breast cancer segregated as an autosomal dominant trait in some families.[3] The search for genes associated with hereditary susceptibility to breast cancer has been facilitated by studies of large kindreds with multiple affected individuals and has led to the identification of several susceptibility genes, including BRCA1, BRCA2, TP53, PTEN/MMAC1, and STK11. Other genes, such as the mismatch repair genes MLH1, MSH2, MSH6, and PMS2, have been associated with an increased risk of ovarian cancer, but have not been consistently associated with breast cancer.
BRCA1
In 1990, a susceptibility gene for breast cancer was mapped by genetic linkage to the long arm of chromosome 17, in the interval 17q12-21.[4] The linkage between breast cancer and genetic markers on chromosome 17q was soon confirmed by others, and evidence for the coincident transmission of both breast and ovarian cancer susceptibility in linked families was observed.[5] The BRCA1 gene was subsequently identified by positional cloning methods and has been found to contain 24 exons that encode a protein of 1,863 amino acids. Germline pathogenic variants in BRCA1 are associated with early-onset breast cancer, ovarian cancer, and fallopian tube cancer. (Refer to the Penetrance of BRCA pathogenic variants section of this summary for more information.) Male breast cancer, pancreatic cancer, testicular cancer, and early-onset prostate cancer may also be associated with pathogenic variants in BRCA1;[6,7,8,9] however, male breast cancer, pancreatic cancer, and prostate cancer are more strongly associated with pathogenic variants in BRCA2.
BRCA2
A second breast cancer susceptibility gene, BRCA2, was localized to the long arm of chromosome 13 through linkage studies of 15 families with multiple cases of breast cancer that were not linked to BRCA1. Pathogenic variants in BRCA2 are associated with multiple cases of breast cancer in families, and are also associated with male breast cancer, ovarian cancer, prostate cancer, melanoma, and pancreatic cancer.[8,9,10,11,12,13,14] (Refer to the Penetrance of BRCA pathogenic variants section of this summary for more information.) BRCA2 is a large gene with 27 exons that encode a protein of 3,418 amino acids.[15] While not homologous genes, both BRCA1 and BRCA2 have an unusually large exon 11 and translational start sites in exon 2. Like BRCA1, BRCA2 appears to behave like a tumor suppressor gene. In tumors associated with both BRCA1 and BRCA2 pathogenic variants, there is often loss of the wild-type allele.
Pathogenic variants in BRCA1 and BRCA2 appear to be responsible for disease in 45% of families with multiple cases of breast cancer and in up to 90% of families with both breast and ovarian cancer.[16]
BRCA1andBRCA2function
Most BRCA1 and BRCA2 pathogenic variants are predicted to produce a truncated protein product, and thus loss of protein function, although some missense pathogenic variants cause loss of function without truncation. Because inherited breast/ovarian cancer is an autosomal dominant condition, individuals with a BRCA1 or BRCA2 pathogenic variant on one copy of chromosome 17 or 13 also carry a normal allele on the other paired chromosome. In most breast and ovarian cancers that have been studied from carriers of pathogenic variants, deletion of the normal allele results in loss of all function, leading to the classification of BRCA1 and BRCA2 as tumor suppressor genes. In addition to their roles as tumor suppressor genes, BRCA1 and BRCA2 are involved in myriad functions within cells, including homologous DNA repair, genomic stability, transcriptional regulation, protein ubiquitination, chromatin remodeling, and cell cycle control.[17,18]
Pathogenic variants inBRCA1andBRCA2
Nearly 2,000 distinct variants and sequence variations in BRCA1 and BRCA2 have already been described.[19] Approximately 1 in 400 to 800 individuals in the general population may carry a germline pathogenic variant in BRCA1 or BRCA2.[20,21] The variants that have been associated with increased cancer risk result in missing or nonfunctional proteins, supporting the hypothesis that BRCA1 and BRCA2 are tumor suppressor genes. While a small number of these pathogenic variants have been found repeatedly in unrelated families, most have not been reported in more than a few families.
Variant-screening methods vary in their sensitivity. Methods widely used in research laboratories, such as single-stranded conformational polymorphism analysis and conformation-sensitive gel electrophoresis, miss nearly a third of the variants that are detected by DNA sequencing.[22] In addition, large genomic alterations such as translocations, inversions, deletions, or insertions are missed by most of the techniques, including direct DNA sequencing; however, testing for these alterations is commercially available. Such rearrangements are believed to be responsible for 12% to 18% of BRCA1 inactivating variants but are seen less frequently in the BRCA2 gene and in individuals of Ashkenazi Jewish (AJ) descent.[23,24,25,26,27,28,29] Furthermore, studies have suggested that these rearrangements may be more frequently seen in Hispanic and Caribbean populations.[27,29,30]
Variants of uncertain significance
Germline pathogenic variants in the BRCA1/BRCA2 genes are associated with an approximately 60% lifetime risk of breast cancer and a 15% to 40% lifetime risk of ovarian cancer. There are no definitive functional tests for BRCA1 or BRCA2; therefore, the classification of nucleotide changes to predict their functional impact as deleterious or benign relies on imperfect data. The majority of accepted pathogenic variants result in protein truncation and/or loss of important functional domains. However, 10% to 15% of all individuals undergoing genetic testing with full sequencing of BRCA1 and BRCA2 will not have a clearly pathogenic variant detected; instead, they will have a variant of uncertain (or unknown) significance (VUS). VUS may cause substantial challenges in genetic counseling, particularly in terms of cancer risk estimates and risk management. Clinical management of such patients needs to be highly individualized and must take into consideration factors such as the patient's personal and family cancer history, in addition to sources of information to help characterize the VUS as benign or deleterious. Thus an improved classification and reporting system may be of clinical utility.[31]
A comprehensive analysis of 7,461 consecutive full gene sequence analyses performed by Myriad Genetic Laboratories, Inc., described the frequency of VUS found over a 3-year period.[32] Among subjects who did not have a clearly pathogenic variant, 13% had VUS defined as "missense mutations and mutations that occur in analyzed intronic regions whose clinical significance has not yet been determined, chain-terminating mutations that truncate BRCA1 and BRCA2 distal to amino acid positions 1853 and 3308, respectively, and mutations that eliminate the normal stop codons for these proteins." The classification of a sequence variant as a VUS is a moving target. An additional 6.8% of subjects without clear pathogenic variants had sequence alterations that were once considered VUS but were reclassified as a polymorphisms, or occasionally as pathogenic variants.
The frequency of VUS varies by ethnicity within the U.S. population. African American individuals appear to have the highest rate of VUS.[33] In a 2009 study of data from Myriad, 16.5% of individuals of African ancestry had VUS, the highest rate among all ethnicities. The frequency of VUS in Asian, Middle Eastern, and Hispanic populations clusters between 10% and 14%, although these numbers are based on limited sample sizes. Over time, the rate of changes classified as VUS has decreased in all ethnicities, largely the result of improved variant classification algorithms.[34] VUS continue to be reclassified as additional information is curated and interpreted.[35,36] Such information may impact the continuing care of affected individuals.
A number of methods for discriminating deleterious from neutral VUS exist and others are in development [37,38,39,40] including integrated methods (see below).[41] Interpretation of VUS is greatly aided by efforts to determine if there is cosegregation of a VUS with the cancers seen in a family. In general, a VUS observed in individuals who also have a pathogenic variant, especially when the same VUS has been identified in conjunction with different pathogenic variants, is less likely to be deleterious, although there are rare exceptions. As an adjunct to the clinical information, models have been deployed to interpret VUS. These models analyze VUS based on sequence conservation, biochemical properties of amino acid changes,[37,42,43,44,45,46] incorporation of information on pathologic characteristics of BRCA1- and BRCA2-related tumors (e.g., BRCA1-related breast cancers are usually estrogen receptor [ER]–negative),[47] and functional studies to measure the influence of specific sequence variations on the activity of BRCA1 or BRCA2 proteins.[48,49] When attempting to interpret a VUS, all available information should be examined.
Population estimates of the likelihood of having aBRCA1orBRCA2pathogenic variant
Statistics regarding the percentage of individuals found to be carriers of BRCA pathogenic variants among samples of women and men with a variety of personal cancer histories regardless of family history are provided below. These data can help determine who might best benefit from a referral for cancer genetic counseling and consideration of genetic testing but cannot replace a personalized risk assessment, which might indicate a higher or lower pathogenic variant likelihood based on additional personal and family history characteristics.
In some cases, the same pathogenic variant has been found in multiple unrelated families. This observation is consistent with a founder effect, wherein a pathogenic variant identified in a contemporary population can be traced to a small group of founders isolated by geographic, cultural, or other factors. Most notably, two specific BRCA1 pathogenic variants (185delAG and 5382insC) and a BRCA2 pathogenic variant (6174delT) have been reported to be common in AJs. However, other founder pathogenic variants have been identified in African American and Hispanic individuals.[30,50,51] The presence of these founder pathogenic variants has practical implications for genetic testing. Many laboratories offer directed testing specifically for ethnic-specific alleles. This greatly simplifies the technical aspects of the test but is not without limitations. Nonfounder BRCA pathogenic variants in the AJ population have been reported to be between 3% and 15%.[32,52,53]
Among the general population, the likelihood of having any BRCA pathogenic variant is as follows:
Among AJ individuals, the likelihood of having any BRCA pathogenic variant is as follows:
Two large U.S. population-based studies of breast cancer patients younger than 65 years examined the prevalence of BRCA1[57,73] and BRCA2[57] pathogenic variants in various ethnic groups. The prevalence of BRCA1 pathogenic variants in breast cancer patients by ethnic group was 3.5% in Hispanic individuals, 1.3% to 1.4% in African American individuals, 0.5% in Asian American individuals, 2.2% to 2.9% in non-AJ White individuals, and 8.3% to 10.2% in AJ individuals.[57,73] The prevalence of BRCA2 pathogenic variants by ethnic group was 2.6% in African American individuals and 2.1% in White individuals.[57]
A study of Hispanic patients with a personal or family history of breast cancer and/or ovarian cancer, who were enrolled through multiple clinics in the southwestern United States, examined the prevalence of BRCA1 and BRCA2 pathogenic variants. BRCA pathogenic variants were identified in 189 of 746 patients (25%) (124 BRCA1, 65 BRCA2);[74] 21 of the 189 (11%) BRCA pathogenic variants identified were large rearrangements, of which 13 (62%) were the BRCA1 exon 9–12 deletion. An unselected cohort of 810 women of Mexican ancestry with breast cancer were tested; 4.3% had a BRCA pathogenic variant. Eight of the 35 pathogenic variants identified also were the BRCA1 exon 9–12 deletion.[75] In another population-based cohort of 492 Hispanic women with breast cancer, the BRCA1 exon 9–12 deletion was found in three patients, suggesting that this variant may be a Mexican founder pathogenic variant and may represent 10% to 12% of all BRCA1 pathogenic variants in similar clinic- and population-based cohorts in the United States. Within the clinic-based cohort, there were nine recurrent pathogenic variants, which accounted for 53% of all variants observed in this cohort, suggesting the existence of additional founder pathogenic variants in this population.
A retrospective review of 29 AJ patients with primary fallopian tube tumors identified germline BRCA pathogenic variants in 17% of patients.[72] Another study of 108 women with fallopian tube cancer identified pathogenic variants in 55.6% of the Jewish women and 26.4% of non-Jewish women (30.6% overall).[76] Estimates of the frequency of fallopian tube cancer in carriers of BRCA pathogenic variants are limited by the lack of precision in the assignment of site of origin for high-grade, metastatic, serous carcinomas at initial presentation.[6,72,76,77]
Population screening
Population screening has identified carriers in a number of AJ populations who would not have met criteria for family-based testing.[64,78,79,80] This could potentially expand the number of individuals who could benefit from preventive strategies. A study has suggested that population screening (compared with personal/family history–based testing) for AJ founder variants is cost-effective on the basis of data from the United States and the United Kingdom.[81] The authors used a decision-analytic model that estimated lifetime costs and the effects of genetic testing to assess cost-effectiveness; the model included costs of pretest genetic counseling and genetic testing and the anticipated risk of cardiovascular outcomes. Additional analyses conducted by the same group also suggested cost-effectiveness when testing was expanded to include all pathogenic variants in BRCA1, BRCA2, RAD51C, RAD51D, and PALB2.[82] These studies are based on various assumptions, some of which are imprecise (e.g., population prevalence estimates for some genes). Furthermore, as acknowledged by the authors, these types of efforts would require implementation of clinical support across the care continuum, in order for patients identified with pathogenic variants to benefit from this information. Consequently, there remain significant resource implications as population screening efforts are considered, which are the focus of ongoing research efforts. Because the detection rate is highly dependent on the prevalence of pathogenic variants in a population, it is not clear how applicable this approach would be for other populations, including other founder pathogenic variant populations. Another unanswered question is whether adequate genetic counseling can be provided for whole populations.
Clinical criteria and models for prediction of the likelihood of aBRCA1orBRCA2pathogenic variant
Several studies have assessed the frequency of BRCA1 or BRCA2 pathogenic variants in women with breast or ovarian cancer.[57,58,73,83,84,85,86,87,88,89,90,91] Personal characteristics associated with an increased likelihood of a BRCA1 and/or BRCA2 pathogenic variant include the following:
Studies have shown that women with metastatic breast cancer have a higher frequency of inherited pathogenic or likely pathogenic variants in breast cancer risk genes than women with nonmetastatic breast cancer. Mutations were seen in BRCA1, BRCA2, and other breast cancer risk genes. Hence, providers may want to consider genetic testing in women with metastatic breast cancer.[96,97]
Family history characteristics associated with an increased likelihood of carrying a BRCA1 and/or BRCA2 pathogenic variant include the following:
Clinical criteria and practice guidelines for identifying individuals who may have aBRCA1orBRCA2pathogenic variant
Several professional organizations and expert panels, including the American Society of Clinical Oncology,[98] the National Comprehensive Cancer Network (NCCN),[99] the American Society of Human Genetics,[100] the American College of Medical Genetics and Genomics,[101] the National Society of Genetic Counselors,[101] the U.S. Preventive Services Task Force,[102] and the Society of Gynecologic Oncologists,[103] have developed clinical criteria and practice guidelines that can be helpful to health care providers in identifying individuals who may have a BRCA1 or BRCA2 pathogenic variant.
Models for prediction of the likelihood of aBRCA1orBRCA2pathogenic variant
Many models have been developed to predict the probability of identifying germline BRCA1/BRCA2 pathogenic variants in individuals or families. These models include those using logistic regression,[32,83,84,86,89,104,105] genetic models using Bayesian analysis (BRCAPRO and Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm [BOADICEA]),[89,106] and empiric observations.[54,57,60,107,108,109]
In addition to BOADICEA, BRCAPRO is commonly used for genetic counseling in the clinical setting. BRCAPRO and BOADICEA predict the probability of being a carrier and produce estimates of breast cancer risk (refer to Table 4). The discrimination and accuracy (factors used to evaluate the performance of prediction models) of these models are much higher for their ability to report on carrier status than for their ability to predict fixed or remaining lifetime risk.
BOADICEA is a polygenetic model that uses complex segregation analysis to examine both breast cancer risk and the probability of having a BRCA1 or BRCA2 pathogenic variant.[106] Even among experienced providers, the use of prediction models has been shown to increase the power to discriminate which patients are most likely to be carriers of BRCA1/BRCA2 pathogenic variants.[110,111] Most models do not include other cancers seen in the BRCA1 and BRCA2 spectrum, such as pancreatic cancer and prostate cancer. Interventions that decrease the likelihood that an individual will develop cancer (such as oophorectomy and mastectomy) may influence the ability to predict BRCA1 and BRCA2 pathogenic variant status.[112] One study has shown that the prediction models for genetic risk are sensitive to the amount of family history data available and do not perform as well with limited family information.[113] BOADICEA is being expanded to incorporate additional risk variants (genome-wide association study [GWAS] single nucleotide variants [SNVs]) to better predict pathogenic variant status and to improve the accuracy of breast cancer and ovarian cancer risk estimates.[114]
The performance of the models can vary in specific ethnic groups. The BRCAPRO model appeared to best fit a series of French Canadian families.[115] There have been variable results in the performance of the BRCAPRO model among Hispanic individuals,[116,117] and both the BRCAPRO model and Myriad tables underestimated the proportion of carriers of pathogenic variants in an Asian American population.[118] BOADICEA was developed and validated in British women. Thus, the major models used for both overall risk (Table 1) and genetic risk (Table 4) have not been developed or validated in large populations of racially and ethnically diverse women. Of the commonly used clinical models for assessing genetic risk, only the Tyrer-Cuzick model contains nongenetic risk factors.
The power of several of the models has been compared in different studies.[119,120,121,122] Four breast cancer genetic-risk models, BOADICEA, BRCAPRO, IBIS, and eCLAUS, were evaluated for their diagnostic accuracy in predicting BRCA1/BRCA2 pathogenic variants in a cohort of 7,352 German families.[123] The family member with the highest likelihood of carrying a pathogenic variant from each family was screened for BRCA1/BRCA2 pathogenic variants. Carrier probabilities from each model were calculated and compared with the actual variants detected. BRCAPRO and BOADICEA had significantly higher diagnostic accuracy than IBIS or eCLAUS. Accuracy for the BOADICEA model was further improved when statuses of the tumor markers ER, progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2/neu) were included in the model. The inclusion of these biomarkers has been shown to improve the performance of BRCAPRO.[124,125]
| Myriad Prevalence Tables[86] | BRCAPRO[89,112] | BOADICEA[89,106] | Tyrer-Cuzick[126] |
---|---|---|---|---|
AJ = Ashkenazi Jewish; BOADICEA = Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm; FDR = first-degree relatives; SDR = second-degree relatives. | ||||
Method | Empiric data from Myriad Genetics based on personal and family history reported on requisition forms | Statistical model, assumes autosomal dominant inheritance | Statistical model, assumes polygenic risk | Statistical model, assumes autosomal dominant inheritance |
Features of the model | Probandmay or may not have breast or ovarian cancer | Proband may or may not have breast or ovarian cancer | Proband may or may not have breast or ovarian cancer | Proband must beunaffected |
Considers age of breast cancer diagnosis as <50 y, >50 y | Considers exact age at breast and ovarian cancer diagnosis | Considers exact age at breast and ovarian cancer diagnosis | Also includes reproductive factors and body mass index to estimate breast cancer risk | |
Considers breast cancer in ≥1 affected relative only if diagnosed <50 y | Considers prior genetic testing in family (i.e.,BRCA1/BRCA2 pathogenic variant–negative relatives) | Includes allFDRandSDRwith and without cancer | ||
Considers ovarian cancer in ≥1 relative at any age | Considers oophorectomy status | Includes AJ ancestry | ||
Includes AJ ancestry | Includes all FDR and SDR with and without cancer | |||
Very easy to use | Includes AJ ancestry | |||
Limitations | Simplified/limited consideration of family structure | Requires computer software and time-consuming data entry | Requires computer software and time-consuming data entry | Designed for individuals unaffected with breast cancer |
Incorporates only FDR and SDR; may need to change proband to best capture risk and to account for disease in the paternal lineage | ||||
May overestimate risk in bilateral breast cancer[127] | ||||
Early age of breast cancer onset | May perform better in White populations than in racial and ethnic minority populations[117,128] | Incorporates only FDR and SDR; may need to change proband to best capture risk | ||
May underestimate risk ofBRCApathogenic variant in high-grade serous ovarian cancers but overestimate the risk for other histologies[129] |
Genetic testing for BRCA1 and BRCA2 pathogenic variants has been available to the public since 1996. As more individuals have undergone testing, risk assessment models have improved. This, in turn, gives providers better data to estimate an individual patient's risk of carrying a pathogenic variant, but risk assessment continues to be an art. There are factors that might limit the ability to provide an accurate risk assessment (i.e., small family size, paucity of women, or ethnicity) including the specific circumstances of the individual patient (such as history of disease or risk-reducing surgeries).
Penetrance ofBRCApathogenic variants
The proportion of individuals carrying a pathogenic variant who will manifest the disease is referred to as penetrance. (Refer to the Penetrance of Inherited Susceptibility to Hereditary Breast and/or Gynecologic Cancers section of this summary for more information.)
Numerous studies have estimated breast and ovarian cancer penetrance in carriers of BRCA1 and BRCA2 pathogenic variants. Risk of both breast and ovarian cancer is consistently estimated to be higher in carriers of BRCA1 pathogenic variants than in carriers of BRCA2 pathogenic variants. Results from two large meta-analyses are shown in Table 5.[130,131] One study [130] analyzed pooled pedigree data from 22 studies involving 289 BRCA1 and 221 BRCA2 pathogenic variant–positive individuals. Index cases from these studies had female breast cancer, male breast cancer, or ovarian cancer but were unselected for family history. A subsequent study [131] combined penetrance estimates from the previous study and nine others that included an additional 734 BRCA1 and 400 BRCA2 pathogenic variant–positive families. The estimated cumulative risks of breast cancer by age 70 years in these two meta-analyses were 55% to 65% for carriers of BRCA1 pathogenic variants and 45% to 47% for carriers of BRCA2 pathogenic variants. Ovarian cancer risks were 39% for carriers of BRCA1 pathogenic variants and 11% to 17% for carriers of BRCA2 pathogenic variants.
Study | Breast Cancer Risk (%) (95% CI) | Ovarian Cancer Risk (%) (95% CI) | ||
---|---|---|---|---|
CI = confidence interval. | ||||
a Risk estimate calculated up to age 70 years. | ||||
b Risk estimate calculated up to age 80 years. | ||||
BRCA1 | BRCA2 | BRCA1 | BRCA2 | |
Antoniou et al. (2003)[130] | 65 (44–78)a | 45 (31–56)a | 39 (18–54)a | 11 (2.4–19)a |
Chen et al. (2007)[131] | 55 (50–59)a | 47 (42–51)a | 39 (34–45)a | 17 (13–21)a |
Kuchenbaecker et al. (2017)[132] | 72 (65–79)b | 69 (61–77)b | 44 (36–53)b | 17 (11–25)b |
While the cumulative risks of developing cancer by age 70 years are higher for carriers of BRCA1 pathogenic variants than for BRCA2 pathogenic variants, the relative risks (RRs) of breast cancer decline more with age in carriers of BRCA1 pathogenic variants.[130] Studies of penetrance for carriers of specific individual variants are not usually large enough to provide stable estimates, but numerous studies of the Ashkenazi founder pathogenic variants have been conducted. One group of researchers analyzed the subset of families with one of the Ashkenazi founder pathogenic variants from their larger meta-analyses and found that the estimated penetrance for the individual pathogenic variants was very similar to the corresponding estimates among all carriers.[133] A later study of 4,649 women with BRCA pathogenic variants reported significantly lower RRs of breast cancer in those with the BRCA2 6174delT variant than in those with other BRCA2 variants (hazard ratio [HR], 0.35; confidence interval [CI], 0.18–0.69).[134]
One study provided prospective 10-year risks of developing cancer among asymptomatic carriers at various ages.[131] Nonetheless, making precise penetrance estimates in an individual carrier is difficult. The lifetime risks of ovarian cancer are 5.2% in carriers of RAD51C pathogenic variants, 5.8% in carriers of BRIP1 pathogenic variants, and 12% in carriers of RAD51D pathogenic variants. Risk-reducing salpingo-oophorectomy (RRSO) may be considered for these patients upon completion of childbearing.[135,136]
Data from the Consortium of Investigators of Modifiers of BRCA1/BRCA2 (CIMBA), comprising 19,581 carriers of BRCA1 pathogenic variants and 11,900 carriers of BRCA2 pathogenic variants, were analyzed to estimate HRs for breast cancer and ovarian cancer by pathogenic variant type, function, and nucleotide position.[137] Breast cancer cluster regions and ovarian cancer cluster regions were found in both genes. Risks for incidence of breast cancer and ovarian cancer and age at diagnosis differed by variant class. Further evaluation of these findings is needed before they can be translated into clinical practice.
Another study from the CIMBA group looked at the phenotype of women with breast cancer who had inherited pathogenic variants in both BRCA1 and BRCA2.[138] The majority of women carried the common Jewish pathogenic variants. Compared with women who were heterozygous for the same pathogenic variant (heterozygote controls), women who were heterozygous for both BRCA1 and BRCA2 were more likely to be diagnosed with breast cancer than women who were heterozygote controls, and more likely to be diagnosed with ovarian cancer than women who were heterozygote controls with BRCA2, but not those with BRCA1 pathogenic variants. Similarly, age at onset of breast cancer was younger in carriers of both variants compared with women who were heterozygote controls with BRCA2, but not compared with those with BRCA1 pathogenic variants. The percentage of women with both variants and estrogen receptor–positive and progesterone receptor–positive breast cancer was intermediate between the heterozygote controls with BRCA1 pathogenic variants and those with BRCA2 pathogenic variants. The authors concluded that women who inherit pathogenic variants in both BRCA1 and BRCA2 may be managed similarly to carriers of only a BRCA1 variant.
Several studies have suggested that BRCA pathogenic variants may be associated with genetic anticipation. One study evaluated 176 families with BRCA1 or BRCA2 pathogenic variants and at least two consecutive generations of the same cancer. The probands' generations were diagnosed with breast cancer an estimated 6.8 years earlier than the parents' generations and 9.8 years earlier than the grandparents' generations.[139] Similarly, another study showed a difference in age at breast cancer diagnosis between 80 mother-and-daughter paired pathogenic variant carriers but only if the mother was diagnosed with breast cancer after age 50 years.[140] Another cohort study of 106 paired women from two consecutive generations with a known BRCA pathogenic variant in the family estimated a 6- to 8-year earlier age at onset in subsequent generations.[141]
RRSO and/or use of oral contraceptives have been associated with the risk of breast cancer.[66,130,142,143,144,145,146,147] (Refer to the RRSO section and the Oral contraceptives section of this summary for more information.) Other potentially modifiable reproductive and hormonal factors can also affect risk.[148,149,150,151,152] Genetic modifiers of penetrance of breast cancer and ovarian cancer are increasingly under study but are not clinically useful at this time.[153,154,155] (Refer to the Modifiers of risk in carriers of BRCA1 and BRCA2 pathogenic variants section for more information.) While the average breast cancer and ovarian cancer penetrances may not be as high as initially estimated, they are substantial, both in relative and absolute terms, particularly in women born after 1940. A higher risk before age 50 years has been consistently seen in more recent birth cohorts,[64,66,141] and additional studies will be required to further characterize potential modifying factors to arrive at more precise individual risk projections. Precise penetrance estimates for less common cancers, such as pancreatic cancer, are lacking.
Contralateral breast cancer (CBC) in carriers ofBRCApathogenic variants
The increased risk of CBC among carriers of BRCA1 and BRCA2 pathogenic variants has been confirmed in several large studies, with fairly consistent results, as summarized in Table 6.
Study | BRCA1Carriers (%) | BRCA2Carriers (%) |
---|---|---|
Graeser et al. (2009)[156] | 18.5 | 13.2 |
Malone et al. (2010)[157] | 20.5 | 15.9 |
van der Kolk et al. (2010)[158] | 34.2 | 29.2 |
Metcalfe et al. (2011)[159] | 23.8 | 18.7 |
Molina-Montes et al. (2014)[160] | 27 | 19 |
Basu et al. (2015)[161] | 25.7 | 19.5 |
van den Broek et al. (2016)[162] | 21.1 | 10.8 |
Published results include a large study by the German Consortium for Hereditary Breast and Ovarian Cancer, which estimated the risk of CBC in members of families with known BRCA1 and BRCA2 pathogenic variants. At 25 years after the first breast cancer, the risk of CBC was close to 50% in both BRCA1 and BRCA2 families. The risk was also inversely correlated with age in this study, with the highest risks seen in women whose first breast cancer was before age 40 years.[156]
Subsequently, results from the Women's Environmental Cancer and Radiation Epidemiology (WECARE) study, a large, population-based, nested case-control study of CBC, reported a 10-year risk of CBC of 15.9% among carriers of BRCA1/BRCA2 pathogenic variants and a risk of 4.9% among noncarriers. Risks were also inversely related to age at first diagnosis in this study and were 1.8-fold higher in those with a first-degree relative (FDR) with breast cancer.[157]
A larger study of members of BRCA1/BRCA2 families in the Netherlands reported similar 10-year risks of CBC for women from BRCA1 and BRCA2 families (34.2% and 29.2%, respectively).[158]
A comparison of 655 women with BRCA1/BRCA2 pathogenic variants undergoing either breast-conserving therapy or mastectomy noted that both treatment groups experienced high rates of CBC, exceeding 50% by 20 years of follow-up. Rates were significantly higher among women with BRCA1 pathogenic variants than in women with BRCA2 pathogenic variants, and among women whose first breast cancer occurred at or before age 35 years.[163]
In a study of 810 women with stage I or stage II breast cancer who had a BRCA1 or BRCA2 pathogenic variant identified in the family, 149 (18.4%) developed CBC; the 15-year actuarial risk was 36.1% among carriers of BRCA1 pathogenic variants and 28.5% among carriers of BRCA2 pathogenic variants.[159] Risks were higher among women diagnosed before age 50 years than among women diagnosed at age 50 years or older (37.6% vs. 16.8%; P = .003). Furthermore, the risk of CBC varied by family history among women whose initial breast cancer was diagnosed before age 50 years. For these women, the CBC risk among those with 0, 1, or 2 or more FDRs with breast cancer diagnosed before age 50 years was 33.4%, 39.1%, and 49.7%, respectively.
The risk of CBC after a first breast cancer in BRCA1 and BRCA2 carriers has been examined in both retrospective and prospective observational epidemiological studies. A systematic review and quantitative meta-analysis of these epidemiologic studies (18 retrospective and 2 prospective cohort studies) reported 5-year cumulative risks of CBC of 15% (95% CI, 9.50%–20%) in BRCA1 carriers and 9% (95% CI, 5%–14%) in BRCA2 carriers.[160] When the prospective studies were analyzed separately, the 5-year cumulative risk increased to 23.4% (95% CI, 9.1%–39.5%) in BRCA1 carriers and to 17.5% (95% CI, 9.1%–39.5%) in BRCA2 carriers. The discrepancies in the reported frequencies may be inherent due to the potential for biases introduced in retrospective series.
Similarly, in a Dutch cohort of 6,294 patients (including 200 BRCA1 carriers and 71 BRCA2 carriers) with invasive breast cancer diagnosed before age 50 years, and a median follow-up of 12.5 years, the 10-year risks of CBC were 21.1% (95% CI, 15.4%–27.4%) for BRCA1 carriers, 10.8% (95% CI, 4.7%–19.6%) for BRCA2 carriers, and 5.1% (95% CI, 4.5%–5.7%) for noncarriers.[162] Age at first breast cancer diagnosis was predictive of the 10-year cumulative risk of CBC among BRCA1/BRCA2 carriers only. Specifically, the CBC risk among BRCA1/BRCA2 carriers diagnosed before age 41 years was 23.9% (BRCA1, 25.5%; BRCA2, 17.2%); in contrast, CBC among those diagnosed between ages 41 and 49 years was 12.6% (BRCA1, 15.6%; BRCA2, 7.2%).
In an English study of 506 BRCA1 carriers and 505 BRCA2 carriers with a diagnosis of breast cancer at any age and median follow-up of 7.8 years, the 10-year risks for CBC were 25.7% for BRCA1 carriers and 19.5% for BRCA2 carriers.[161] Earlier age at first breast cancer diagnosis for BRCA1 and BRCA2 carriers combined was significantly associated with a higher CBC risk, with a 20-year rate of 55.4% among those younger than 40 years, compared with 36.4% among those older than 50 years. Additionally, differences were more pronounced among BRCA1 carriers when compared with BRCA2 carriers.
An international, multicenter, prospective cohort study followed 1,305 BRCA1 and 908 BRCA2 female carriers with a diagnosis of breast cancer (without any other cancers) for a median follow-up time of 4 years (range, 2–7 y).[132] Participants had a median age of 47 years (range, 40–55 y) at the start of follow-up. The authors reported a cumulative risk of CBC 20 years after the initial breast cancer diagnosis of 40% (95% CI, 35%–45%) for BRCA1 carriers and 26% (95% CI, 20%–33%) for BRCA2 carriers. These 20-year estimates are comparable to 10-year cumulative risk estimates reported in Table 6.
The effect of radiation therapy on CBC risk has also been investigated. A population-based case-control study of 708 women with CBC and 1,399 controls with unilateral breast cancer showed no clear association between the use of radiation therapy and an increased risk of CBC in patients with BRCA1/BRCA2 pathogenic variants.[164] There was no evidence of increased CBC risk following radiation use in carriers of pathogenic/likely pathogenic ATM variants or in carriers of the pathogenic 1100delC CHEK2 variant.
Thus, in summary, despite differences in study design, study sites, and sample sizes, the data on CBC among women with BRCA1/BRCA2 pathogenic variants show several consistent findings:
Risk-reducing strategies
Refer to the Risk-reducing mastectomy section of this summary for information about the use of risk-reducing surgery in carriers of BRCA pathogenic variants. Refer to the Chemoprevention section of this summary for information about the use of tamoxifen as a risk-reduction strategy for CBC in carriers of BRCA pathogenic variants.
Breast cancer as a second malignancy in carriers ofBRCApathogenic variants
Two genetic registry–based studies have recently explored the risk of primary breast cancer after BRCA-related ovarian cancer. In one study, 164 BRCA1/BRCA2 carriers with primary epithelial ovarian, fallopian tube or primary peritoneal cancer were followed for subsequent events.[165] The risk of metachronous breast cancer at 5 years after a diagnosis of ovarian cancer was lower than previously reported for unaffected BRCA1/BRCA2 carriers. In this series, overall survival was dominated by ovarian cancer-related deaths. A similar study compared the risk of primary breast cancer in BRCA-related ovarian cancer patients and unaffected carriers.[166] The 2-year, 5-year, and 10-year risks of primary breast cancer were all statistically significantly lower in patients with ovarian cancer. The risk of CBC among women with a unilateral breast cancer before their ovarian cancer diagnosis was also lower than in women without ovarian cancer, although the difference did not reach statistical significance. These studies suggest that treatment for ovarian cancer, namely oophorectomy and platinum-based chemotherapy, may confer protection against subsequent breast cancer. In a single-institution cohort study of 364 patients with epithelial ovarian cancer who underwent BRCA pathogenic variant testing, 135 (37.1%) were found to carry a germline BRCA1 or BRCA2 pathogenic variant. Of the 135 BRCA1/BRCA2 carriers, 12 (8.9%) developed breast cancer. All breast cancers were stage 0 to stage II and diagnosed as follows: mammogram (7), palpable mass (3), and incidental finding during risk-reducing mastectomy (2). At median follow-up of 6.3 years, of the 12 patients with breast cancer after ovarian cancer, three died of recurrent ovarian cancer and one died of metastatic breast cancer.[167] The majority of these cancers were detected with mammogram or clinical exam, suggesting additional breast surveillance with other modalities or risk-reducing surgery may be of questionable value. Mathematical modeling suggests that for women with BRCA-associated ovarian cancer, breast cancer screening should consist of mammography and clinical breast exam. The consideration of breast magnetic resonance imaging (MRI) and/or risk-reducing mastectomies may be beneficial for women with early-stage ovarian cancer or for long-term ovarian cancer survivors.[168]
Cancers other than female breast/ovarian
Female breast and ovarian cancers are clearly the dominant cancers associated with BRCA1 and BRCA2. BRCA pathogenic variants also confer an increased risk of fallopian tube and primary peritoneal carcinomas. One large study from a familial registry of carriers of BRCA1 pathogenic variants has found a 120-fold RR of fallopian tube cancer among carriers of BRCA1 pathogenic variants compared with the general population.[6] The risk of primary peritoneal cancer among carriers of BRCA pathogenic variants with intact ovaries is increased but remains poorly quantified, despite a residual risk of 3% to 4% in the 20 years after RRSO.[169,170] (Refer to the RRSO section in the Ovarian cancer section of this summary for more information.)
Pancreatic, male breast, and prostate cancers have also been consistently associated with BRCA pathogenic variants, particularly with BRCA2. Other cancers have been associated in some studies. The strength of the association of these cancers with BRCA pathogenic variants has been more difficult to estimate because of the lower numbers of these cancers observed in carriers of pathogenic variants.
Men with BRCA2 pathogenic variants, and to a lesser extent BRCA1 pathogenic variants, are at increased risk of breast cancer with lifetime risks estimated at 5% to 10% and 1% to 2%, respectively.[6,8,9,171] Men carrying BRCA2 pathogenic variants, and to a lesser extent BRCA1 pathogenic variants, have an approximately threefold to sevenfold increased risk of prostate cancer.[7,8,12,109,172,173,174,175]BRCA2-associated prostate cancer also appears to be more aggressive.[176,177,178,179,180,181] (Refer to the BRCA1 and BRCA2 section in the PDQ summary on Genetics of Prostate Cancer for more information.)
Studies of familial pancreatic cancer (FPC) [182,183,184,185,186] and unselected series of pancreatic cancer [187,188,189] have also supported an association with BRCA2, and to a lesser extent, BRCA1.[7] Overall, it appears that between 3% to 15% of families with FPC may have germline BRCA2 pathogenic variants, with risks increasing with more affected relatives.[182,183,184] Similarly, studies of unselected pancreatic cancers have reported BRCA2 pathogenic variant frequencies between 3% to 7%, with these numbers approaching 10% in those of AJ descent.[187,188,190] The lifetime risk of pancreatic cancer in BRCA2 carriers is estimated to be 3% to 5%,[8,12] compared with an estimated lifetime risk of 0.5% by age 70 years in the general population.[191] A large, single-institution study of more than 1,000 carriers of pathogenic variants found a 21-fold increased risk of pancreatic cancer among BRCA2 carriers and a 4.7-fold increased risk among carriers of BRCA1 pathogenic variants, compared with incidence in the general population.[175] Other cancers associated with BRCA2 pathogenic variants in some, but not all, studies include melanoma, biliary cancers, and head and neck cancers, but these risks appear modest (<5% lifetime risk) and are less well studied.[12]
Cancer Sites[6,7,8,12,63,174] | BRCA1 | BRCA2 | ||
---|---|---|---|---|
a Refer to the PDQ summary on Genetics of Prostate Cancerfor more information about the association ofBRCA1andBRCA2with prostate cancer. | ||||
+++ Multiple studies demonstrated association and are relatively consistent. | ||||
++ Multiple studies and the predominance of the evidence are positive. | ||||
+ May be an association, predominantly single studies; smaller limited studies and/or inconsistent but weighted toward positive. | ||||
Strength of Evidence | Magnitude of Absolute Risk | Strength of Evidence | Magnitude of Absolute Risk | |
Breast (female) | +++ | High | +++ | High |
Ovary, fallopian tube, peritoneum | +++ | High | +++ | Moderate |
Breast (male) | + | Undefined | +++ | Low |
Pancreas | ++ | Very Low | +++ | Low |
Prostatea | + | Undefined | +++ | High |
The first Breast Cancer Linkage Consortium study investigating cancer risks reported an excess of colorectal cancer in BRCA1 carriers (RR, 4.1; 95% CI, 2.4–7.2).[192] This finding was supported by some,[6,7,193] but not all,[8,63,71,109,194,195,196] family-based studies. However, unselected series of colorectal cancer that have been exclusively performed in the AJ population have not shown elevated rates of BRCA1 or BRCA2 pathogenic variants.[197,198,199] Taken together, the data suggest little, if any, increased risk of colorectal cancer, and possibly only in specific population groups. Therefore, at this time, carriers of BRCA1 pathogenic variants should adhere to population-screening recommendations for colorectal cancer.
No increased prevalence of hereditary BRCA pathogenic variants was found among 200 Jewish women with endometrial carcinoma or 56 unselected women with uterine papillary serous carcinoma.[200,201] (Refer to the Risk-reducing salpingo-oophorectomy section in the Ovarian cancer section of this summary for more information.)
Cancer risk in individuals who test negative for a known familialBRCA1/BRCA2pathogenic variant ("true negative")
There is conflicting evidence as to the residual familial risk among women who test negative for the BRCA1/BRCA2 pathogenic variant segregating in the family. An initial study based on prospective evaluation of 353 women who tested negative for the BRCA1 pathogenic variant segregating in the family found that five incident breast cancers occurred during more than 6,000 person-years of observation, for a lifetime risk of 6.8%, a rate similar to the general population.[145] A report that the risk may be as high as fivefold in women who tested negative for the BRCA1 or BRCA2 pathogenic variant in the family [202] was followed by numerous letters to the editor suggesting that ascertainment biases account for much of this observed excess risk.[203,204,205,206,207,208] Four additional analyses have suggested an approximate 1.5-fold to 2-fold excess risk.[207,209,210,211] In one study, two cases of ovarian cancer were reported.[211] Several studies have involved retrospective analyses; all studies have been based on small observed numbers of cases and have been of uncertain statistical and clinical significance.
Results from numerous other prospective studies have found no increased risk. A study of 375 women who tested negative for a known familial pathogenic variant in BRCA1 or BRCA2 reported two invasive breast cancers, two in situ breast cancers, and no ovarian cancers diagnosed, with a mean follow-up of 4.9 years. Four invasive breast cancers were expected, whereas two were observed.[212] Another study of similar size but longer follow-up (395 women and 7,008 person-years of follow-up) also found no statistically significant overall increase in breast cancer risk among variant-negative women (observed/expected [O/E], 0.82; 95% CI, 0.39–1.51), although women who had at least one FDR with breast cancer had a nonsignificant increased risk (O/E, 1.33; 95% CI, 0.41–2.91).[213] A study of 160 BRCA1 and 132 BRCA2 pathogenic variant–positive families from the Breast Cancer Family Registry found no evidence for increased risk among noncarriers in these families.[214] In a large study of 722 variant-negative women from Australia in whom six invasive breast cancers were observed after a median follow-up of 6.3 years, the standardized incidence ratio (SIR) was not significantly elevated (SIR, 1.14; 95% CI, 0.51–2.53).[215] Based on available data, it appears that women testing negative for known familial BRCA1/BRCA2 pathogenic variants can adhere to general population screening guidelines unless they have sufficient additional risk factors, such as a personal history of atypical hyperplasia of the breast or family history of breast cancer in relatives who do not carry the familial pathogenic variant.
Breast and ovarian cancer risk in breast cancer families without detectableBRCA1/BRCA2 pathogenic variants ("indeterminate")
The majority of families with site-specific breast cancer test negative for BRCA1/BRCA2 and have no features consistent with Cowden syndrome or Li-Fraumeni syndrome.[32] Five studies using population-based and clinic-based approaches have demonstrated no increased risk of ovarian cancer in such families. Although ovarian cancer risk was not increased, breast cancer risk remained elevated.[214,216,216,217,217,218,218,219,220]
Modifiers of risk in carriers ofBRCA1andBRCA2pathogenic variants
Pathogenic variants in BRCA1 and BRCA2 confer high risks of breast and ovarian cancers. The risks, however, are not equal in all pathogenic variant carriers and have been found to vary by several factors, including type of cancer, age at onset, and variant position.[221] This observed variation in penetrance has led to the hypothesis that other genetic and/or environmental factors modify cancer risk in carriers of pathogenic variants. There is a growing body of literature identifying genetic and nongenetic factors that contribute to the observed variation in rates of cancers seen in families with BRCA1/BRCA2 pathogenic variants.
Genetic modifiers of breast and ovarian cancer risk
The largest studies investigating genetic modifiers of breast and ovarian cancer risk to date have come from CIMBA, a large international effort with genotypic and phenotypic data on more than 15,000 BRCA1 and 10,000 BRCA2 carriers.[222] Using candidate gene analysis and GWAS, CIMBA has identified several loci associated both with increased and decreased risk of breast cancer and ovarian cancer. Some of the SNVs are related to subtypes of breast cancer, such as hormone-receptor and HER2/neu status. The risks conferred are all modest but if operating in a multiplicative fashion could significantly impact risk of cancer in carriers of BRCA1/BRCA2 pathogenic variants. Currently, these SNVs are not being tested for or used in clinical decision making.
Genotype-phenotype correlations
Some genotype -phenotype correlations have been identified in both BRCA1 and BRCA2 pathogenic variant families. None of the studies have had sufficient numbers of pathogenic variant–positive individuals to make definitive conclusions, and the findings are probably not sufficiently established to use in individual risk assessment and management. In 25 families with BRCA2 pathogenic variants, an ovarian cancer cluster region was identified in exon 11 bordered by nucleotides 3,035 and 6,629.[11,223] A study of 164 families with BRCA2 pathogenic variants collected by the Breast Cancer Linkage Consortium confirmed the initial finding. Pathogenic variants within the ovarian cancer cluster region were associated with an increased risk of ovarian cancer and a decreased risk of breast cancer in comparison with families with variants on either side of this region.[224] In addition, a study of 356 families with protein-truncating BRCA1 pathogenic variants collected by the Breast Cancer Linkage Consortium reported breast cancer risk to be lower with variants in the central region (nucleotides 2,401–4,190) compared with surrounding regions. Ovarian cancer risk was significantly reduced with variants 3' to nucleotide 4,191.[225] These observations have generally been confirmed in subsequent studies.[130,226,227] Studies in Ashkenazim, in whom substantial numbers of families with the same pathogenic variant can be studied, have also found higher rates of ovarian cancer in carriers of the BRCA1:185delAG variant, in the 5' end of BRCA1, compared with carriers of the BRCA1:5382insC variant in the 3' end of the gene.[228,229] The risk of breast cancer, particularly bilateral breast cancer, and the occurrence of both breast and ovarian cancer in the same individual, however, appear to be higher in carriers of the BRCA1:5382insC pathogenic variant compared with carriers of BRCA1:185delAG and BRCA2:6174delT variants. Ovarian cancer risk is considerably higher in carriers of BRCA1 pathogenic variants, and it is uncommon before age 45 years in carriers of the BRCA2:6174delT pathogenic variant.[228,229]
In an Australian study of 122 families with a pathogenic variant in BRCA1, large genomic rearrangement variants were associated with higher-risk features in breast and ovarian cancers, including younger age at breast cancer diagnosis and higher incidence of bilateral breast cancer.[230]
Pathology of breast cancer
BRCA1pathology
Several studies evaluating pathologic patterns seen in BRCA1-associated breast cancers have suggested an association with adverse pathologic and biologic features. These findings include higher than expected frequencies of medullary histology, high histologic grade, areas of necrosis, trabecular growth pattern, aneuploidy, high S-phase fraction, high mitotic index, and frequent TP53 variants.[231,232,233,234,235,236,237,238] In a large international series of 3,797 carriers of BRCA1 pathogenic variants, the median age at breast cancer diagnosis was 40 years.[238] Of breast tumors arising in BRCA1 carriers, 78% were ER-negative; 79% were PR-negative; 90% were HER2-negative; and 69% were triple-negative. These findings were consistent with multiple smaller series.[92,234,239,240,241] In addition, the proportion of ER-negative tumors significantly decreased as the age at breast cancer diagnosis increased.[238]
There is considerable, but not complete, overlap between the triple-negative and basal-like subtype cancers, both of which are common in BRCA1-associated breast cancer,[242,243] particularly in women diagnosed before age 50 years.[92,93,94] A small proportion of BRCA1-related breast cancers are ER-positive, which are associated with later age of onset.[244,245] These ER-positive cancers have clinical behavior features that are intermediate between ER-negative BRCA1 cancers and ER-positive sporadic breast cancers, raising the possibility that there may be a unique mechanism by which they develop.
The prevalence of germline BRCA1 pathogenic variants in women with TNBC is significant, both in women undergoing clinical genetic testing (and thus selected in large part for family history) and in unselected triple-negative patients, with pathogenic variants reported in 9% to 35%.[94,95,239,246,247,248,249] Notably, studies have demonstrated a high rate of BRCA1 pathogenic variants in unselected women with TNBC, particularly in those diagnosed before age 50 years. A large report of 1,824 patients with TNBC unselected for family history, recruited through 12 studies, identified 14.6% with a pathogenic variant in an inherited cancer susceptibility gene.[249]BRCA1 pathogenic variants accounted for the largest proportion (8.5%), followed by BRCA2 (2.7%); PALB2 (1.2%); and BARD1, RAD51D, RAD51C and BRIP1 (0.3%–0.5% for each gene). In this study, those with pathogenic variants in BRCA1/BRCA2 or other inherited cancer genes were diagnosed at an earlier age and had higher grade tumors than those without pathogenic variants. Specifically, among carriers of BRCA1 pathogenic variants, the average age at diagnosis was 44 years, and 94% had high-grade tumors. One study examined 308 individuals with TNBC; BRCA1 pathogenic variants were present in 45. Pathogenic variants were seen both in women unselected for family history (11 of 58; 19%) and in those with family history (26 of 111; 23%).[250] A meta-analysis based on 2,533 patients from 12 studies was conducted to assess the risk of a BRCA1 pathogenic variant in high-risk women with TNBC.[251] Results indicated that the RR of a BRCA1 pathogenic variant among women with versus without TNBC is 5.65 (95% CI, 4.15–7.69), and approximately two in nine women with triple-negative disease harbor a BRCA1 pathogenic variant. Interestingly, a study of 77 unselected patients with TNBC in which 15 (19.5%) had a germline pathogenic variant or somatic BRCA1/BRCA2mutation demonstrated a lower risk of relapse in those with BRCA1 pathogenic variant–associated TNBC than in those with non-BRCA1-associated TNBC; this study was limited by its size.[247] A second study examining clinical outcomes in BRCA1-associated versus non-BRCA1-associated TNBC showed no difference, although there was a trend toward more brain metastases in those with BRCA1-associated breast cancer. In both of these studies, all but one carrier of BRCA1 pathogenic variants received chemotherapy.[252] In contrast, HER2 positivity and young age alone in the absence of family history or a second primary cancer does not increase the likelihood of a pathogenic variant in BRCA1, BRCA2, or TP53.[253]
It has been hypothesized that many BRCA1 tumors are derived from the basal epithelial layer of cells of the normal mammary gland, which account for 3% to 15% of unselected invasive ductal cancers. If the basal epithelial cells of the breast represent the breast stem cells, the regulatory role suggested for wild-type BRCA1 may partly explain the aggressive phenotype of BRCA1-associated breast cancer when BRCA1 function is damaged.[254] Further studies are needed to fully appreciate the significance of this subtype of breast cancer within the hereditary syndromes.
The most accurate method for identifying basal-like breast cancers is through gene expression studies, which have been used to classify breast cancers into biologically and clinically meaningful groups.[240,255,256] This technology has also been shown to correctly differentiate BRCA1- and BRCA2-associated tumors from sporadic tumors in a high proportion of cases.[257,258,259] Notably, among a set of breast tumors studied by gene expression array to determine molecular phenotype, all tumors with BRCA1 alterations fell within the basal tumor subtype;[240] however, this technology is not in routine use due to its high cost. Instead, immunohistochemical markers of basal epithelium have been proposed to identify basal-like breast cancers, which are typically negative for ER, PR, and HER2, and stain positive for cytokeratin 5/6, or epidermal growth factor receptor.[260,261,262,263] Based on these methods to measure protein expression, a number of studies have shown that the majority of BRCA1-associated breast cancers are positive for basal epithelial markers.[92,234,262]
There is growing evidence that preinvasive lesions are a component of the BRCA phenotype. The Breast Cancer Linkage Consortium initially reported a relative lack of an in situ component in BRCA1-associated breast cancers,[232] also seen in two subsequent studies of BRCA1/BRCA2 carriers.[264,265] However, in a study of 369 ductal carcinoma in situ (DCIS) cases, BRCA1 and BRCA2 pathogenic variants were detected in 0.8% and 2.4%, respectively, which is only slightly lower than previously reported prevalence in studies of invasive breast cancer patients.[266] A retrospective study of breast cancer cases in a high-risk clinic found similar rates of preinvasive lesions, particularly DCIS, among 73 BRCA-associated breast cancers and 146 pathogenic variant–negative cases.[267,268] A study of AJ women, stratified by whether they were referred to a high-risk clinic or were unselected, showed similar prevalence of DCIS and invasive breast cancers in referred patients compared with one-third lower DCIS cases among unselected subjects.[269] Similarly, data about the prevalence of hyperplastic lesions have been inconsistent, with reports of increased [270,271] and decreased prevalence.[265] Similar to invasive breast cancer, DCIS diagnosed at an early age and/or with a family history of breast and/or ovarian cancer is more likely to be associated with a BRCA1/BRCA2 pathogenic variant.[272]
Overall evidence suggests DCIS is part of the BRCA1/BRCA2 spectrum, particularly BRCA2; however, the prevalence of pathogenic variants in DCIS patients, unselected for family history, is less than 5%.[266,269]
BRCA2pathology
The phenotype for BRCA2-related tumors appears to be more heterogeneous and is less well-characterized than that of BRCA1, although they are generally positive for ER and PR.[232,273,274] A large international series of 2,392 carriers of BRCA2 pathogenic variants found that only 23% of tumors arising in carriers of BRCA2 pathogenic variants were ER-negative; 36% were PR-negative; 87% were HER2-negative; and 16% were triple-negative.[238] A large report of 1,824 patients with TNBC unselected for family history, recruited through 12 studies, identified 2.7% with a BRCA2 pathogenic variant.[249] (Refer to the BRCA1 pathology section of this summary for more information about this study.) A report from Iceland found less tubule formation, more nuclear pleomorphism, and higher mitotic rates in BRCA2-related tumors than in sporadic controls; however, a single BRCA2 founder pathogenic variant (999del5) accounts for nearly all hereditary breast cancer in this population, thus limiting the generalizability of this observation.[275] A large case series from North America and Europe described a greater proportion of BRCA2-associated tumors with continuous pushing margins (a histopathologic description of a pattern of invasion), fewer tubules and lower mitotic counts.[276] Other reports suggest that BRCA2-related tumors include an excess of lobular and tubulolobular histology.[233,273] In summary, histologic characteristics associated with BRCA2 pathogenic variants have been inconsistent.
Role ofBRCA1andBRCA2in sporadic breast cancer
Given that germline pathogenic variants in BRCA1 or BRCA2 lead to a very high probability of developing breast cancer, it was a natural assumption that these genes would also be involved in the development of the more common nonhereditary forms of the disease. Although somatic mutations in BRCA1 and BRCA2 are not common in sporadic breast cancer tumors,[277,278,279,280] there is increasing evidence that hypermethylation of the gene promoter (BRCA1) and loss of heterozygosity (LOH) (BRCA2) are frequent events. In fact, many breast cancers have low levels of the BRCA1 mRNA, which may result from hypermethylation of the gene promoter.[281,282,283] Approximately 10% to 15% of sporadic breast cancers appear to have BRCA1 promoter hypermethylation, and even more have downregulation of BRCA1 by other mechanisms. Basal-type breast cancers (ER negative, PR negative, HER2 negative, and cytokeratin 5/6 positive) more commonly have BRCA1 dysregulation than other tumor types.[284,285,286]BRCA1-related tumor characteristics have also been associated with constitutional methylation of the BRCA1 promoter. In a study of 255 breast cancers diagnosed before age 40 years in women without germline BRCA1 pathogenic variants, methylation of BRCA1 in peripheral blood was observed in 31% of women whose tumors had multiple BRCA1-associated pathological characteristics (e.g., high mitotic index and growth pattern including multinucleated cells) compared with less than 4% methylation in controls.[287] (Refer to the BRCA1 pathology section for more information.) Although hypermethylation has not been reported for BRCA2 pathogenic variants, the BRCA2 locus on chromosome 13q is the target of frequent LOH in breast cancer.[288,289] Targeted therapies are being developed for tumors with loss of BRCA1 or BRCA2 protein expression.[290]
Pathology of ovarian cancer
Ovarian cancers in women with BRCA1 and BRCA2 pathogenic variants are more likely to be high-grade serous adenocarcinomas and are less likely to be mucinous or borderline tumors.[291,292,293,294,295] Fallopian tube cancer and peritoneal carcinomas are also part of the BRCA-associated disease spectrum.[72,296]
Histopathologic examinations of fallopian tubes removed from women with a hereditary predisposition to ovarian cancer show dysplastic and hyperplastic lesions that suggest a premalignant phenotype.[297,298] Occult carcinomas have been reported in 2% to 11% of adnexa removed from carriers of BRCA pathogenic variants at the time of risk-reducing surgery.[299,300,301] Most of these occult lesions are seen in the fallopian tubes, which has led to the hypothesis that many BRCA-associated ovarian cancers may actually have originated in the fallopian tubes. Specifically, the distal segment of the fallopian tubes (containing the fimbriae) has been implicated as a common origin of the high-grade serous cancers seen in BRCA pathogenic variant carriers, based on the close proximity of the fimbriae to the ovarian surface, exposure of the fimbriae to the peritoneal cavity, and the broad surface area in the fimbriae.[302] Because of the multicentric origin of high-grade serous carcinomas from Müllerian-derived tissue, staging of ovarian, tubal, and peritoneal carcinomas is now considered collectively by the International Federation of Gynecology and Obstetrics. The term high-grade serous ovarian carcinoma may be used to represent high-grade pelvic serous carcinoma for consistency in language.[303]
High-grade serous ovarian carcinomas have a higher incidence of somatic TP53 mutations.[291,304] DNA microarray technology suggests distinct molecular pathways of carcinogenesis between BRCA1, BRCA2, and sporadic ovarian cancer.[305] Furthermore, data suggest that BRCA-related ovarian cancers metastasize more frequently to the viscera, while sporadic ovarian cancers remain confined to the peritoneum.[306]
Unlike high-grade serous carcinomas, low-grade serous ovarian cancers are less likely to be part of the BRCA1/BRCA2 spectrum.[307,308]
Role ofBRCA1andBRCA2in sporadic ovarian cancer
Given that germline variants in BRCA1 or BRCA2 lead to a very high probability of developing ovarian cancer, it was a natural assumption that these genes would also be involved in the development of the more common nonhereditary forms of the disease. Although somatic mutations in BRCA1 and BRCA2 are not common in sporadic ovarian cancer tumors,[277,278,279,280] there is increasing evidence that hypermethylation of the gene promoter (BRCA1) and LOH (BRCA2) are frequent events. Loss of BRCA1 or BRCA2 protein expression is more common in ovarian cancer than in breast cancer,[309] and downregulation of BRCA1 is associated with enhanced sensitivity to cisplatin and improved survival in this disease.[310,311] Targeted therapies are being developed for tumors with loss of BRCA1 or BRCA2 protein expression.[290]
Other High-Penetrance Syndromes Associated With Breast and/or Gynecologic Cancers
Lynch syndrome
Lynch syndrome is characterized by autosomal dominant inheritance of susceptibility to predominantly right-sided colon cancer, endometrial cancer, ovarian cancer, and other extracolonic cancers (including cancer of the renal pelvis, ureter, small bowel, and pancreas), multiple primary cancers, and a young age of onset of cancer.[312] The condition is caused by germline variants in the mismatch repair (MMR) genes, which are involved in repair of DNA mismatch variants.[313] The MLH1 and MSH2 genes are the most common susceptibility genes for Lynch syndrome, accounting for 80% to 90% of observed pathogenic variants,[314,315] followed by MSH6 and PMS2.[316,317,318,319,320,321] (Refer to the Lynch Syndrome section in the PDQ summary on Genetics of Colorectal Cancer for more information about this syndrome.)
After colorectal cancer, endometrial cancer is the second hallmark cancer of a family with Lynch syndrome. Even in the original Family G, described by Dr. Aldred Scott Warthin, numerous family members were noted to have extracolonic cancers including endometrial cancer. Although the first version of the Amsterdam criteria did not include endometrial cancer,[322] in 1999, the Amsterdam criteria were revised to include endometrial cancer as extracolonic tumors associated with Lynch syndrome to identify families at risk.[323] In addition, the Bethesda guidelines in 1997 (revised in 2004) did include endometrial and ovarian cancers as Lynch syndrome–related cancers to prompt tumor testing for Lynch syndrome.[324,325]
The lifetime risk of ovarian carcinoma in females with Lynch syndrome is estimated to be as high as 12%, and the reported RR of ovarian cancer has ranged from 3.6 to 13, based on families ascertained from high-risk clinics with known or suspected Lynch syndrome.[326,327,328,329,330,331] There may be differences in ovarian cancer risk depending on the Lynch syndrome–associated pathogenic variant. In PMS2-associated Lynch syndrome, one study of 284 families was unable to identify an increased risk of ovarian cancer.[332] Another prospective registry of 3,119 Lynch syndrome–pathogenic variant carriers described the cumulative risk of ovarian cancer to range from 10% to 17% in MLH1, MSH2, and MSH6 carriers. In contrast, 0 of 67 women with a pathogenic variant in PMS2 developed ovarian cancer in 303 follow-up years.[333] Overall, there are too few cases of PMS2 pathogenic variant carriers to make definitive recommendations for ovarian cancer management. Characteristics of Lynch syndrome–associated ovarian cancers may include overrepresentation of the International Federation of Gynecology and Obstetrics stages I and II at diagnosis (reported as 81.5%), underrepresentation of serous subtypes (reported as 22.9%), and a better 10-year survival (reported as 80.6%) than reported both in population-based series and in carriers of BRCA pathogenic variants.[334,335]
The issue of breast cancer risk in Lynch syndrome has been controversial. Retrospective studies have been inconsistent, but several have demonstrated microsatellite instability in a proportion of breast cancers from individuals with Lynch syndrome;[336,337,338,339] one of these studies evaluated breast cancer risk in individuals with Lynch syndrome and found that it is not elevated.[339] However, the largest prospective study to date of 446 unaffected carriers of pathogenic variants from the Colon Cancer Family Registry [340] who were followed for up to 10 years reported an elevated SIR of 3.95 for breast cancer (95% CI, 1.59–8.13; P = .001).[340] The same group subsequently analyzed data on 764 carriers of MMR gene pathogenic variants with a prior diagnosis of colorectal cancer. Results showed that the 10-year risk of breast cancer following colorectal cancer was 2% (95% CI, 1%–4%) and that the SIR was 1.76 (95% CI, 1.07–2.59).[341] A series from the United Kingdom composed of clinically referred Lynch syndrome kindreds, with efforts to correct for ascertainment, showed a twofold increased risk of breast cancer in 157 MLH1 carriers but not in carriers of other MMR variants.[342] Results from a meta-analysis of breast cancer risk in Lynch syndrome among 15 studies with molecular tumor testing results revealed that 62 of 122 breast cancers (51%; 95% CI, 42%–60%) in MMR pathogenic variant carriers were MMR-deficient. In addition, breast cancer risk estimates among a total of 21 studies showed an increased risk of twofold to 18-fold in eight studies that compared MMR variant carriers with noncarriers, while 13 studies did not observe statistical evidence for an association of breast cancer risk with Lynch syndrome.[343]
A number of subsequent studies have suggested the presence of higher breast cancer risks than previously published,[344,345,346,347] although this has not been consistently observed.[348] Through a study of 325 Canadian families with Lynch syndrome, primarily encompassing MLH1 and MSH2 carriers, the lifetime cumulative risk for breast cancer among MSH2 carriers was reported to be 22%.[344] Similarly, breast cancer risks were elevated in a study of 423 women with Lynch syndrome, with substantially higher risks among those with MSH6 and PMS2 pathogenic variants, compared with MLH1 and MSH2 pathogenic variants.[345] In fact, breast cancer risk to age 60 years was 37.7% for PMS2, 31.1% for MSH6, 16.1% for MSH2, and 15.5% for MLH1. These findings are consistent with another study of 528 patients with Lynch syndrome–associated pathogenic variants (including MLH1, MSH2, MSH6, PMS2, and EPCAM) in which PMS2 and MSH6 variants were much more frequent among patients with only breast cancer, compared with those with only colorectal cancer (P = 2.3 x 10-5).[346] Additional data to support an association of MSH6 with breast cancer were provided through a study of over 10,000 cancer patients across the United States who had genetic testing.[347] Findings indicated that MSH6 was associated with breast cancer with an odds ratio (OR) of 2.59 (95% CI, 1.35–5.44). Taken together, these studies highlight how the risk profile among patients with Lynch syndrome is continuing to evolve as more individuals are tested through multigene panel testing, with representation of larger numbers of individuals with PMS2 and MSH6 pathogenic variants compared with prior studies. In the absence of definitive risk estimates, individuals with Lynch syndrome are screened for breast cancer on the basis of family history.[349]
Refer to the Lynch Syndrome section of the Clinical Management of Other Hereditary Breast and/or Gynecologic Cancer Syndromes section of this summary for information about clinical management of Lynch syndrome.
Li-Fraumeni syndrome (LFS)
Breast cancer is also a component of the rare LFS, in which germline variants of the TP53 gene on chromosome 17p have been documented. Located on chromosome 17p, TP53 encodes a 53kd nuclear phosphoprotein that binds DNA sequences and functions as a negative regulator of cell growth and proliferation in the setting of DNA damage. It is also an active component of programmed cell death.[350] Inactivation of the TP53 gene or disruption of the protein product is thought to allow the persistence of damaged DNA and the possible development of malignant cells.[351,352] Widely used clinical diagnostic criteria for LFS were originally developed by Chompret et al. in 2001 (called the Chompret Criteria) [353] and revised in 2009 based on additional emerging data.[354]
LFS is characterized by premenopausal breast cancer in combination with childhood sarcoma, brain tumors, leukemia, and adrenocortical carcinoma.[351,355,356]
Germline variants in TP53 are thought to account for fewer than 1% of breast cancer cases.[357]TP53-associated breast cancer is often HER2/neu-positive, in addition to being ER-positive, PR-positive, or both.[358,359,360] Evidence also exists that patients treated for a TP53-related tumor with chemotherapy or radiation therapy may be at risk of a treatment-related second malignancy.
Historical criteria for defining LFS
The term Li-Fraumeni syndrome was used for the first time in 1982,[361] and the following criteria, which subsequently became the classical definition of the syndrome, were proposed by Li and Fraumeni in 1988 [362]:
Subsequently in 2001, Chompret et al. [353] systematically developed clinical criteria for recommending TP53 genetic testing, with the narrow LFS tumor spectrum defined as sarcoma, brain tumors, breast cancer, and adrenocortical carcinoma. The criteria were as follows:
These criteria were revised in 2009 [354] based on additional emerging data [352,363] as follows:
*The 2009 Chompret criteria defined the LFS tumor spectrum as including the following cancers: soft tissue sarcoma, osteosarcoma, brain tumor, premenopausal breast cancer, adrenocortical carcinoma, leukemia, and lung bronchoalveolar cancer.
In 2015, Bougeard et al. [356] revised the criteria based on data from 415 carriers of pathogenic variants, to include the presence of childhood anaplastic rhabdomyosarcoma and breast cancer before age 31 years as an indication for testing, similar to what is recommended for choroid plexus carcinoma and adrenocortical carcinoma. The criteria were revised as follows:
**The 2015 Chompret criteria defined the LFS tumor spectrum as including the following cancers: premenopausal breast cancer, soft tissue sarcoma, osteosarcoma, central nervous system (CNS) tumor, and adrenocortical carcinoma.
Clinical characteristics of LFS
Germline TP53 pathogenic variants were identified in 17% (n = 91) of 525 samples submitted to City of Hope laboratories for clinical TP53 testing.[352] All families with a TP53 pathogenic variant had at least one family member with a sarcoma, breast cancer, brain cancer, or adrenocortical cancer (core cancers). In addition, all eight individuals with a choroid plexus tumor had a TP53 pathogenic variant, as did 14 of the 21 individuals with childhood adrenocortical cancer. In women aged 30 to 49 years who had breast cancer but no family history of other core cancers, no TP53 variants were found.
Subsequently, a large clinical series of patients from France who were tested primarily based on the 2009 version of the Chompret criteria [354] included 415 carriers of pathogenic variants from 214 families.[356] In this study, 43% of carriers had multiple malignancies, and the mean age at first tumor onset was 24.9 years. The childhood tumor spectrum was characterized by osteosarcomas, adrenocortical carcinomas, CNS tumors, and soft tissue sarcomas (present in 23%–30% collectively), whereas the adult tumor spectrum primarily encompassed breast cancer (79% of females) and soft tissue sarcomas (27% of carriers). The TP53 pathogenic variant detection rate was 6% among females younger than 31 years with breast cancer and no additional features suggestive of LFS. Evaluation of genotype-phenotype correlations indicated a gradient of clinical severity, with a significantly lower mean age at onset among those with dominant-negative missense variants (21.3 years), compared with those with all types of loss-of-function variants (28.5 years) or genomic rearrangements (35.8 years). With the exception of adrenocortical carcinoma, affected children mostly harbored dominant-negative missense pathogenic variants. Among 127 female carriers of pathogenic variants with breast cancer, 31% developed CBC. Receptor status information was available for 40 tumors, which indicated 55% were HER2-positive, and 37% were triple-positive (i.e., ER-positive, PR-positive, and HER2-positive). There was an exceptionally high rate of multiple malignancies (43%) among carriers of pathogenic variants, of which 83% were metachronous. Treatment records were available for 64 carriers who received radiation therapy for treatment of their first tumor; of these, 19 (30%) developed 26 secondary tumors within a radiation field, with a latency of 2 to 26 years (mean, 10.7 y).
Similarly, results of 286 TP53 pathogenic variant–positive individuals in the National Cancer Institute's LFS Study indicated a cumulative cancer incidence of almost 100% by age 70 years for both males and females.[364] They reported substantial variations by sex, age, and cancer type. Specifically, cumulative cancer incidence reached 50% by age 31 years in females and age 46 years in males, although male risks were higher in childhood and late adulthood. Cumulative cancer incidence by sex for the top four cancers is included in Table 8. Of those with one cancer, 49% developed at least one additional cancer after a median of 10 years. Age-specific risks for developing first and second cancers were comparable.
| Cumulative Cancer Risk by Age 70 Years | |
---|---|---|
a Adapted from Mai et al.[364] | ||
b Other cancers, such as adrenocortical carcinoma, leukemia, and lung bronchoalveolar cancer, have been considered part of the LFS cancer spectrum.[354,356] | ||
Cancer Type | Females (%) | Males (%) |
Breast cancer | 54 | – |
Soft tissue sarcoma | 15 | 22 |
Brain cancer | 6 | 19 |
Osteosarcoma | 5 | 11 |
With the increasing use of multigene (panel) tests, it is important to recognize that pathogenic variants in TP53 are unexpectedly being identified in individuals without a family history characteristic of LFS.[365] The clinical significance of finding an isolated TP53 pathogenic variant in an individual or family who does not meet the Chompret criteria is uncertain. Consequently, it remains important to interpret cancer risks and determine optimal management strategies for individuals who are unexpectedly found to have a germline TP53 pathogenic variant, while considering their personal and family histories.
One cohort study evaluated 116 individuals with a germline TP53 pathogenic variant yearly at the National Institutes of Health Clinical Center using multimodality screening with and without gadolinium. Baseline screening identified a cancer in eight patients (6.9%) with a false-positive rate of 34.5% for MRI (n = 40).[366] Screening for breast cancer with annual breast MRI is recommended;[99] additional screening for other cancers has been studied and is evolving.[367,368]
PTENhamartoma tumor syndromes (including Cowden syndrome)
Cowden syndrome and Bannayan-Riley-Ruvalcaba syndrome (BRRS) are part of a spectrum of conditions known collectively as PTEN hamartoma tumor syndromes (PHTS). Approximately 85% of patients diagnosed with Cowden syndrome, and approximately 60% of patients with BRRS have an identifiable PTEN pathogenic variant.[369] In addition, PTEN pathogenic variants have been identified in patients with very diverse clinical phenotypes.[370] The term PHTS refers to any patient with a PTEN pathogenic variant, irrespective of clinical presentation.
PTEN functions as a dual-specificity phosphatase that removes phosphate groups from tyrosine, serine, and threonine. PTEN pathogenic variants are diverse and can present as nonsense, missense, frameshift, or splice-site variants. Approximately 40% of variants are found in exon 5, which encodes the phosphatase core motif; several recurrent pathogenic variants have been observed at this location.[371] Pathogenic variants in the 5' end of PTEN or within the phosphatase core of PTEN tend to affect more organ systems.[372]
Operational criteria for the diagnosis of Cowden syndrome have been published and subsequently updated.[373,374] These include major, minor, and pathognomonic criteria that consist of certain mucocutaneous manifestations and adult-onset dysplastic gangliocytoma of the cerebellum (Lhermitte-Duclos disease). An updated set of criteria based on a systematic literature review has been suggested [375] and is currently utilized in the National Comprehensive Cancer Network (NCCN) guidelines.[99] Contrary to previous criteria, the authors concluded that there was insufficient evidence for any features to be classified as pathognomonic. Increased genetic testing (especially multigene panels) has identified individuals with germline PTEN pathogenic variants who do not meet diagnostic criteria for PHTS. Diagnostic criteria will need to be reconciled with these recently discovered phenotypes. Hence, it is unclear whether PHTS diagnoses should be based on clinical features or a positive PTEN genetic test result. The American College of Medical Genetics and Genomics (ACMG) suggests that referral for genetics consultation be considered for individuals with a personal history of or a first-degree relative with the following: 1) adult-onset Lhermitte-Duclos disease or 2) any three of the major or minor criteria that have been established for the diagnosis of Cowden syndrome.[101] Detailed recommendations, including diagnostic criteria for Cowden syndrome, can be found in the NCCN and ACMG guidelines.[99,101] Additionally, a predictive model that uses clinical criteria to estimate the probability of a PTEN pathogenic variant is available; a cost-effectiveness analysis suggests that germline PTEN testing is cost effective if the probability of a variant is greater than 10%.[376]
Over a 10-year period, the International Cowden Consortium (ICC) prospectively recruited a consecutive series of adult and pediatric patients meeting relaxed ICC criteria for PTEN testing in the United States, Europe, and Asia.[377] Most individuals did not meet the clinical criteria for a diagnosis of Cowden syndrome or BRRS. Of the 3,399 individuals recruited and tested, 295 probands (8.8%) and an additional 73 family members carried a germline PTEN pathogenic variant. The authors concluded that melanoma, kidney cancer, and colorectal cancer should be added to the spectrum of cancers associated with PTEN germline pathogenic variants (in addition to breast cancer, thyroid cancer, and endometrial cancer). This conclusion was based on the high melanoma, kidney, and colorectal cancer lifetime risk estimates found in individuals with PTEN pathogenic variants. A second study of approximately 100 patients with a germline PTEN pathogenic variant confirmed these findings and suggested a cumulative cancer risk of 85% by age 70 years.[378]
Although PTEN pathogenic variants, which are estimated to occur in 1 in 200,000 individuals,[373] account for a small fraction of hereditary breast cancer, the characterization of PTEN function will provide valuable insights into the signal pathway and the maintenance of normal cell physiology.[373,379] Lifetime breast cancer risk is estimated to be between 25% and 50% among women with Cowden syndrome.[380] Other studies have reported risks as high as 85%;[377,378,381,382] however, there are concerns regarding selection bias in these studies. As in other forms of hereditary breast cancer, onset is often at a young age and may be bilateral.[383] Lifetime risk of endometrial cancer is estimated to be between 19% and 28%, depending on the cohort studied, with an increased risk of premenopausal onset.[377,378,384] Because of the low prevalence of PTEN pathogenic variants in the population, the proportion of endometrial cancer attributable to Cowden syndrome is small. There are no data that link PTEN pathogenic variants to an increased risk of ovarian cancer. Skin manifestations include multiple trichilemmomas, oral fibromas and papillomas, and acral, palmar, and plantar keratoses. History or observation of the characteristic skin features raises a suspicion of Cowden syndrome. CNS manifestations include macrocephaly, developmental delay, and dysplastic gangliocytomas of the cerebellum.[385,386] (Refer to the PDQ summaries on Genetics of Colorectal Cancer and Genetics of Skin Cancer for more information about PTEN hamartoma tumor syndromes [including Cowden syndrome].)
Diffuse gastric and lobular breast cancer syndrome
The E-cadherin gene CDH1 was first described in 1998 in three Maori families with multiple cases of diffuse gastric cancer (DGC), leading to the designation of hereditary diffuse gastric cancer (HDGC). There have been multiple subsequent reports of an excess of lobular breast cancer in HDGC families.[387]CDH1 is located on chromosome 16q22.1 and encodes the E-cadherin protein, a calcium-dependent homophilic adhesion molecule that plays a key role in cellular adhesion, cell polarity, cell signaling, and maintenance of cellular differentiation and tissue morphology.[388] E-cadherin binds to various catenins to stabilize the cytoplasmic cell adhesion complex and to maintain the E-cadherin interaction with actin filament.[389] Loss of CDH1 can occur as a result of somatic mutations, LOH, or hypermethylation, and can result in dedifferentiation and invasiveness in human cancers.[390,391] Classic histopathologic findings in gastrectomy specimens include in situ signet ring cells and/or pagetoid spread of signet ring cells. Of all gastric cancers, 1% to 3% are attributed to inherited gastric cancer syndromes.[392]
HDGC is an autosomal dominant syndrome associated with poorly differentiated invasive adenocarcinoma of the stomach presenting as linitis plastica. It is a highly penetrant and highly fatal syndrome, with a risk of clinical DGC ranging from 40% to 83%.[387] The risk of lobular breast cancer, which is characterized by small uniform cells that tend to invade in "single files," is also increased in HDGC. Although invasive lobular breast cancer represents only 10% to 15% of all breast cancers, the lifetime risk of lobular breast cancer in carriers of CDH1 pathogenic variants ranges from 30% to 50%.[389,390] Guidelines for screening for CDH1 vary but include multiple cases of DGC in a family, early age of DGC, or lobular breast cancer in a family with DGC. Approximately 25% of families meeting these criteria are found to have a pathogenic variant in CDH1.[392]CDH1 pathogenic variants have been found in some families with lobular breast cancer but no gastric cancer.[393] The management of individuals with CDH1 pathogenic variants without a family history of gastric cancer is unclear.[393]
Peutz-Jeghers syndrome (PJS)
PJS is an early-onset autosomal dominant disorder characterized by melanocytic macules on the lips, the perioral region, and buccal region; and multiple gastrointestinal polyps, both hamartomatous and adenomatous.[394,395,396] Germline pathogenic variants in the STK11 gene at chromosome 19p13.3 have been identified in the vast majority of PJS families.[397,398,399,400,401] The most common cancers in PJS are gastrointestinal. However, other organs are at increased risk of developing malignancies. For example, the cumulative risks have been estimated to be 32% to 54% for breast cancer [402,403,404] and 21% for ovarian cancer (mainly ovarian sex-cord tumors).[402] The risk for pancreatic cancer has been estimated to be more than 100-fold higher than that in the general population.[402] A systematic review found a lifetime cumulative cancer risk, all sites combined, of up to 93% in patients with PJS.[402,405]Table 9 shows the cumulative risk of these tumors.
Females with PJS are also predisposed to the development of cervical adenoma malignum, a rare and very aggressive adenocarcinoma of the cervix.[406] In addition, females with PJS commonly develop benign ovarian sex-cord tumors with annular tubules, whereas males with PJS are predisposed to development of Sertoli-cell testicular tumors;[407] although neither of these two tumor types is malignant, they can cause symptoms related to increased estrogen production.
Although the risk of malignancy appears to be exceedingly high in individuals with PJS based on the published literature, the possibility that selection and referral biases have resulted in overestimates of these risks should be considered.
Site | Age (y) | Cumulative Risk (%)b | Reference(s) |
---|---|---|---|
GI = gastrointestinal. | |||
a Reprinted with permission from Macmillan Publishers Ltd: Gastroenterology[405], copyright 2010. | |||
b All cumulative risks were increased compared with the general population (P< .05), with the exception of cervix and testes. | |||
c GI cancers include colorectal, small intestinal, gastric, esophageal, and pancreatic. | |||
d Westerman et al.: GI cancer does not include pancreatic cancer.[408] | |||
e Did not include adenoma malignum of the cervix or Sertoli cell tumors of the testes. | |||
Any cancer | 60–70 | 37–93 | [401,402,403,404,408,409] |
GI cancerc,d | 60–70 | 38–66 | [403,404,408,409] |
Gynecological cancer | 60–70 | 13–18 | [403,404] |
Per origin | |||
Stomach | 65 | 29 | [402] |
Small bowel | 65 | 13 | [402] |
Colorectum | 65 | 39 | [402,403] |
Pancreas | 65–70 | 11–36 | [402,403] |
Lung | 65–70 | 7–17 | [402,403,404] |
Breast | 60–70 | 32–54 | [402,403,404] |
Uterus | 65 | 9 | [402] |
Ovary | 65 | 21 | [402] |
Cervixe | 65 | 10 | [402] |
Testese | 65 | 9 | [402] |
Peutz-Jeghers gene(s)
PJS is caused by pathogenic variants in the STK11 (also called LKB1) tumor suppressor gene located on chromosome 19p13.[398,399] Unlike the adenomas seen in familial adenomatous polyposis, the polyps arising in PJS are hamartomas. Studies of the hamartomatous polyps and cancers of PJS show allelic imbalance (LOH) consistent with the two-hit hypothesis, demonstrating that STK11 is a tumor suppressor gene.[410,411] However, heterozygous STK11 knockout mice develop hamartomas without inactivation of the remaining wild-type allele, suggesting that haploinsufficiency may be sufficient for initial tumor development in PJS.[412] Subsequently, the cancers that develop in STK11 +/- mice do show LOH;[413] indeed, compound mutant mice heterozygous for pathogenic variants in STK11 +/- and homozygous for pathogenic variants in TP53 -/- have accelerated development of both hamartomas and cancers.[414]
Germline variants of the STK11 gene represent a spectrum of nonsense, frameshift, and missense variants, and splice-site variants and large deletions.[397,403]
Approximately 85% of variants are localized to regions of the kinase domain of the expressed protein. No strong genotype-phenotype correlations have been identified.[403] Up to 30% of variants are large deletions involving one or more exons of STK11, underscoring the importance of deletion analysis in suspected cases of PJS.[397]
STK11 has been unequivocally demonstrated to cause PJS. Although earlier estimates using direct DNA sequencing showed a 50% pathogenic variant detection rate in STK11, studies adding techniques to detect large deletions have found pathogenic variants in up to 94% of individuals meeting clinical criteria for PJS.[397,405,415] Given the results of these studies, it is unlikely that other major genes cause PJS.
Clinical management
The high cumulative risk of cancers in PJS has led to the various screening recommendations summarized in the table of Published Recommendations for Diagnosis and Surveillance of Peutz-Jeghers Syndrome (PJS) in the PDQ summary on Genetics of Colorectal Cancer.
PALB2
PALB2 (partner and localizer of BRCA2) interacts with the BRCA2 protein and plays a role in homologous recombination and double-stranded DNA repair. Similar to BRIP1 and BRCA2, biallelic pathogenic variants in PALB2 have also been shown to cause Fanconi anemia.[416]
PALB2 pathogenic variants have been screened for in multiple small studies of familial and early-onset breast cancer in multiple populations.[14,417,418,419,420,421,422,423,424,425,426,427,428,429,430,431,432,433,434] Pathogenic variant prevalence has ranged from 0.4% to 3.9%. Similar to BRIP1 and CHEK2, there was incomplete segregation of PALB2 pathogenic variants in families with hereditary breast cancer.[417] Among 559 cases with CBC and 565 matched controls with unilateral breast cancer, pathogenic (truncating) PALB2 pathogenic variants were identified in 0.9% of cases and in none of the controls (RR, 5.3; 95% CI, 1.8–13.2).[428]
Data based on 154 families with loss-of-function PALB2 variants suggest that this gene may be an important cause of hereditary breast cancer, with risks that overlap with BRCA2.[435] In this study, analysis of 362 family members from 154 families with PALB2 pathogenic variants indicated that the absolute risk of female breast cancer by age 70 years ranged from 33% (95% CI, 24%–44%) for those with no family history of breast cancer to 58% (95% CI, 50%–66%) for those with two or more FDRs with early-onset breast cancer. Furthermore, among 63 breast cancer cases in which HER2 status was known, 30% had triple-negative disease. An earlier Finnish study reported on a PALB2 founder pathogenic variant (c.1592delT) that confers a 40% risk of breast cancer to age 70 years [418] and is associated with a high incidence (54%) of triple-negative disease and lower survival.[419] Pathogenic variants have been observed in early-onset and familial breast cancer in many populations.[420,421] A large report of 1,824 patients with TNBC unselected for family history, recruited through 12 studies, identified 1.2% with a PALB2 pathogenic variant.[249] (Refer to the BRCA1 pathology section of this summary for more information about this study.)
In a later Polish study of more than 12,529 unselected women with breast cancer and 4,702 controls, PALB2 pathogenic variants were detected in 116 cases (0.93%; 95% CI, 0.76%–1.09%) and 10 controls (0.21%; 95% CI, 0.08%–0.34%), with an OR for breast cancer of 4.39 (95% CI, 2.30–8.37).[436] The study findings confirm a substantially elevated risk of breast cancer (24%–40%) among women with a PALB2 pathogenic variant up to age 75 years. The 5-year cumulative incidence of CBC was 10% among those with a PALB2 pathogenic variant, compared with 17% among those with a BRCA1 pathogenic variant and 3% among those without a variant in either gene. Furthermore, the 10-year survival for women with a PALB2 pathogenic variant and breast cancer was 48% (95% CI, 36.5%–63.2%), compared with 72.0% among those with a BRCA1 pathogenic variant and 74.7% among those without a variant in either gene. Among PALB2 carriers, breast tumors 2 cm or larger had substantially worse outcomes (32.4% 10-year survival), compared with tumors smaller than 2 cm (82.4% 10-year survival). Approximately one-third of those with a PALB2 pathogenic variant had TNBC, and the average age at breast cancer diagnosis was 53.3 years.
A similar case-control study from China enrolled 16,501 unselected patients with breast cancer and 5,890 controls. These patients were screened for PALB2, BRCA1, and BRCA2 pathogenic variants. The prevalence of PALB2 pathogenic variants was 0.97% in the cases and 0.19% in the controls. The OR for breast cancer for carriers was 5.23 (95% CI, 2.84–9.65; P > .0001). PALB2 carriers were more likely to be age 30 years or younger (6.88% vs. 3.56%; P = .04). PALB2 carriers were also more likely to have TNBCs (22.83% vs. 13.56%; P = .004), larger tumors (tumor size ≥2 cm, 55.93% vs. 45.93%; P = .04), node-positive tumors (49.60% vs. 38.80%; P = .018), and contralateral breast cancers (6.29% vs. 2.01%; P = .003). Additionally, PALB2 carriers were more likely to have a family history of breast and/or ovarian cancers (20.63% vs. 7.96%; P < .0001).[437]
In the largest study to date, 524 families with pathogenic variants in PALB2 recruited through an international effort, the RR of breast cancer in women was reported to be 7.18 (95% CI, 5.82–8.85), with a 53% risk (95% CI, 44%–63%) of breast cancer to age 80 years.[438] Additional elevated risks reported included ovarian cancer (RR, 2.91; 95% CI, 1.40–6.04) with a lifetime risk of 5% (95% CI, 2%–10%), pancreatic cancer (RR, 2.37; 95% CI, 1.24–4.50) with a lifetime risk of 2% to 3% (95% CI for women, 1%–4%; 95% CI for men, 2%–5%), and male breast cancer (RR, 7.34; 95% CI, 1.28–42.18) with a lifetime risk of 1% (95% CI, 0.2%–5%). These findings confirm the role of PALB2 as an inherited breast cancer gene in women, while firmly establishing an association with ovarian, pancreatic, and male breast cancers.
Male breast cancer has been observed in PALB2 pathogenic variant–positive breast cancer families.[14,422,435] In a study of 115 male breast cancer cases in which 18 men had BRCA2 pathogenic variants, an additional two men had either a pathogenic or predicted pathogenic PALB2 variant (accounting for about 10% of germline variants in the study and 1%–2% of the total sample).[14] The RR of breast cancer for male carriers of PALB2 pathogenic variants compared with that seen in the general population was estimated to be 8.30 (95% CI, 0.77–88.56; P = .08) in the study of 154 families.[435]
After the identification of PALB2 pathogenic variants in pancreatic tumors and the detection of germline pathogenic variants in 3% of 96 familial pancreatic patients,[439] numerous studies have pointed to a role for PALB2 in pancreatic cancer. PALB2 pathogenic variants were detected in 3.7% of 81 familial pancreatic cancer families [440] and in 2.1% of 94 BRCA1/BRCA2 pathogenic variant–negative breast cancer patients who had either a personal or family history of pancreatic cancer.[441] Two relatively small studies—one of 77 BRCA1/BRCA2 pathogenic variant–negative probands with a personal or family history of pancreatic cancer, one-half of whom were of AJ descent, and another study of 29 Italian pancreatic cancer patients with a personal or family history of breast or ovarian cancer—failed to detect any PALB2 pathogenic variants.[442,443] A sixfold increase in pancreatic cancer was observed in the relatives of 33 BRCA1/BRCA2-negative, PALB2 pathogenic variant–positive breast cancer probands.[422]
Overall, the observed prevalence of PALB2 pathogenic variants in familial breast cancer varied depending on ascertainment relative to personal and family history of pancreatic and ovarian cancers, but in all studies, the observed pathogenic variant rate was lower than 4%. Data suggest that the RR of breast cancer may overlap with that of BRCA2, particularly in those with a strong family history; thus, it remains important to refine cancer risk estimates in larger studies. Furthermore, the risk of other cancers (e.g., pancreatic) is poorly defined. Given the low PALB2 pathogenic variant prevalence in the population, additional data are needed to define best candidates for testing and appropriate management.
De Novo Pathogenic Variant Rate
Until the 1990s, the diagnosis of genetically inherited breast and ovarian cancer syndromes was based on clinical manifestations and family history. Now that some of the genes involved in these syndromes have been identified, a few studies have attempted to estimate the spontaneous pathogenic variant rate (de novo pathogenic variant rate) in these populations. Interestingly, PJS, PTEN hamartoma syndromes, and LFS are all thought to have high rates of spontaneous pathogenic variants, in the 10% to 30% range,[444,445,446,447] while estimates of de novo pathogenic variants in the BRCA genes are thought to be low, primarily on the basis of the few case reports published.[448,449,450,451,452,453,454,455,456] Additionally, there has been only one case series of breast cancer patients who were tested for BRCA pathogenic variants in which a de novo variant was identified. Specifically, in this study of 193 patients with sporadic breast cancer, 17 pathogenic variants were detected, one of which was confirmed to be a de novo pathogenic variant.[448] As such, the de novo pathogenic variant rate appears to be low and fall into the 5% or less range, based on the limited studies performed.[448,449,450,451,452,453,454,455,456] Similarly, estimates of de novo pathogenic variants in the MMR genes associated with Lynch syndrome are thought to be low, in the 0.9% to 5% range.[457,458,459] However, these estimates of spontaneous pathogenic variant rates in the BRCA genes and Lynch syndrome genes seem to overlap with the estimates of nonpaternity rates in various populations (0.6%–3.3%),[460,461,462] making the de novo pathogenic variant rate for these genes relatively low.
References:
Background
Pathogenic variants in BRCA1, BRCA2, PALB2, and the genes involved in other rare syndromes discussed in the High-Penetrance Breast and/or Gynecologic Cancer Susceptibility Genes section of this summary account for less than 25% of the familial risk of breast cancer.[1] Despite intensive genetic linkage studies, there do not appear to be other high-penetrance genes that account for a significant fraction of the remaining multiple-case familial clusters.[2] However, several moderate-penetrance genes associated with breast and/or gynecologic cancers have been identified. Genes such as CHEK2 and ATM are associated with a 20% or higher lifetime risk of breast cancer;[3,4] similarly, genes such as RAD51C, RAD51D, and BRIP1 are associated with a 5% to 10% risk of ovarian cancer.[5,6] Many of these genes are now included on multigene panels, although the clinical actionability of these findings remains uncertain and under investigation.
Breast and Gynecologic Cancer Susceptibility Genes Identified Through Candidate Gene Approaches
There is a very large literature of genetic epidemiology studies describing associations between various loci and breast cancer risk. Many of these studies suffer from significant design limitations. Perhaps as a consequence, most reported associations do not replicate in follow-up studies. This section is not a comprehensive review of all reported associations. This section describes associations that are believed by the editors to be clinically valid, in that they have been described in several studies or are supported by robust meta-analyses. The clinical utility of these observations remains unclear, however, as the risks associated with these variations usually fall below a threshold that would justify a clinical response.
Fanconi anemia genes
Fanconi anemia (FA) is a rare, inherited condition characterized by bone marrow failure, increased risk of malignancy, and physical abnormalities. To date, 16 FA-related genes, including BRCA1 and BRCA2, have been identified (as outlined in Table 10). FA is mainly an autosomal recessive condition, except when caused by pathogenic variants in FANCB, which is X-linked recessive. FANCA accounts for 60% to 70% of pathogenic variants, FANCC accounts for approximately 14%, and the remaining genes each account for 3% or fewer.[7]
a Refer to theBRCA1 and BRCA2section of this summary for information about the cumulative risk of breast cancer incarriersofBRCA1andBRCA2pathogenic variants. |
b Refer to thePALB2section of this summary for information about the cumulative risk of breast cancer in carriers ofPALB2pathogenic variants. |
c Moderate risk is defined as a statistically significant, twofold or lower increased risk estimate. |
High-Risk Genes |
–BRCA1(FANCS)a |
–BRCA2(FANCD1)a |
–PALB2(FANCN)b |
Moderate-Risk Genes c |
–BRIP1(FANCJ/BACH1) |
–FANCD2 |
–RAD51C(FANCO) |
Genes With Uncertain or No Significantly Increased Risk |
–FANCA |
–FANCB |
–FANCC |
–FANCE |
–FANCF |
–FANCG(XRCC9) |
–FANCI(KIAA1794) |
–FANCL |
–SLX4(FANCP) |
–ERCC4(FANCQ/XPF) |
Progressive bone marrow failure typically occurs in the first decade, with patients often presenting with thrombocytopenia or leucopenia. The incidence of bone marrow failure is 90% by age 40 to 50 years. The incidence is 10% to 30% for hematologic malignancies (primarily acute myeloid leukemia) and 25% to 30% for nonhematologic malignancies (solid tumors, particularly of the head and neck, skin, gastrointestinal [GI] tract, and genital tract). Physical abnormalities, including short stature, abnormal skin pigmentation, radial ray defects (including malformation of the thumbs), abnormalities of the urinary tract, eyes, ears, heart, GI system, and central nervous system, hypogonadism, and developmental delay are present in 60% to 75% of affected individuals.[7]
Variants in some of the FA genes, most notably BRCA1 and BRCA2, but also PALB2, RAD51C (in the RAD51 family of genes), and BRIP1, among others, may predispose to breast cancer in heterozygotes. Given the widespread availability of multigene (panel) tests, genetic testing of many of the FA genes is frequently performed despite uncertain cancer risks and the lack of available evidence-based medical management recommendations for many of these genes.
FA gene pathogenic variant carrier status can have implications for reproductive decision making because pathogenic variants in these genes can lead to serious childhood onset of disease if both parents are carriers of pathogenic variants in the same gene. Partner testing may be considered.
BRIP1
BRIP1 (also known as BACH1) encodes a helicase that interacts with the BRCA1 C-terminal (BRCT) domain. This gene also has a role in BRCA1-dependent DNA repair and cell cycle checkpoint function. Biallelic pathogenic variants in BRIP1 are a cause of FA,[8,9,10] much like such pathogenic variants in BRCA2. Inactivating variants of BRIP1 are associated with an increased risk of breast cancer. In one study, more than 3,000 individuals from BRCA1/BRCA2 pathogenic variant–negative families were examined for BRIP1 variants. Pathogenic variants were identified in 9 of 1,212 individuals with breast cancer but in only 2 of 2,081 controls (P = .003). The relative risk (RR) of breast cancer was estimated to be 2.0 (95% confidence interval [CI], 1.2–3.2; P = .012). Of note, in families with BRIP1 pathogenic variants and multiple cases of breast cancer, there was incomplete segregation of the pathogenic variant with breast cancer, consistent with a low-penetrance allele and similar to that seen with CHEK2.[11] In a case-control study of 3,236 women with ovarian cancer, BRIP1 pathogenic variants were more frequently associated with ovarian cancer risk (RR, 11.2; 95% CI, 3.2–34.1).[12]
CHEK2
CHEK2 is a gene involved in the DNA damage repair response pathway. Based on numerous studies, a polymorphism, 1100delC, appears to be a rare, moderate-penetrance cancer susceptibility allele.[13,14,15,16,17,18] One study identified the pathogenic variant in 1.2% of the European controls, 4.2% of the European BRCA1/BRCA2-negative familial breast cancer cases, and 1.4% of unselected female breast cancer cases.[13] In a group of 1,479 Dutch women younger than 50 years with invasive breast cancer, 3.7% were found to have the CHEK2 1100delC pathogenic variant.[19] In additional European and U.S. (where the pathogenic variant appears to be slightly less common) studies, including a large prospective study,[20] the frequency of CHEK2 pathogenic variants detected in familial breast or ovarian cancer cases has ranged from 0% [21] to 11%; overall, these studies have found an approximately 1.5-fold to 3-fold increased risk of female breast cancer.[20,22,23,24,25] A multicenter combined analysis and reanalysis of nearly 20,000 subjects from ten case-control studies, however, has verified a significant 2.3-fold excess of breast cancer among carriers of pathogenic variants.[26] A subsequent meta-analysis based on 29,154 cases and 37,064 controls from 25 case-control studies found a significant association between CHEK2 1100delC heterozygotes and breast cancer risk (odds ratio [OR], 2.75; 95% CI, 2.25–3.36). The ORs and CIs in unselected, familial, and early-onset breast cancer subgroups were 2.33 (1.79–3.05), 3.72 (2.61–5.31), and 2.78 (2.28–3.39), respectively. However, study limitations included pooling of populations without subgroup analysis, using a mix of population-based and hospital-based controls, and basing results on unadjusted estimates (as cases and controls were matched on only a few common factors); therefore, results should be interpreted in the context of these limitations.[27] In a series of male breast cancer patients, the CHEK2 1100delC variant was significantly more frequently identified than in controls, suggesting that this variant is also associated with an increased risk of male breast cancer.[28]
Two studies have suggested that the risk associated with a CHEK2 1100delC pathogenic variant was stronger in the families of probands ascertained because of bilateral breast cancer.[29,30] Furthermore, a meta-analysis of carriers of 1100delC pathogenic variants estimated the risk of breast cancer to be 42% by age 70 years in women with a family history of breast cancer.[31] Similarly, a Polish study reported that CHEK2 truncating pathogenic variants confer breast cancer risks based on a family history of breast cancer as follows: no family history: 20%; one second-degree relative: 28%; one first-degree relative: 34%; and both first- and second-degree relatives: 44%.[3] Moreover, a Dutch study suggested that female homozygotes for the CHEK2 1100delC variant have a greater-than-twofold increased breast cancer risk compared with heterozygotes.[32] Although there have been conflicting reports regarding cancers other than breast cancer associated with CHEK2 pathogenic variants, this may be dependent on variant type (i.e., missense vs. truncating) or population studied and is not currently of clinical utility.[18,23,33,34,35,36,37,38] The contribution of CHEK2 variants to breast cancer may depend on the population studied, with a potentially higher variant prevalence in Poland.[39] Carriers of CHEK2 variants in Poland may be more susceptible to estrogen receptor (ER)–positive breast cancer.[40]
Currently, the clinical applicability of CHEK variants remains uncertain because of low variant prevalence and lack of guidelines for clinical management.[41]
A large Dutch study of 86,975 individuals reported an increased risk of cancers other than breast and colon for carriers of the CHEK2 1100delC pathogenic variant,[42] although additional studies are needed to further refine these risks.
(Refer to the CHEK2 section in the PDQ summary on Genetics of Colorectal Cancer for more information.)
ATM
Ataxia telangiectasia (AT) is an autosomal recessive disorder characterized by neurologic deterioration, telangiectasias, immunodeficiency states, and hypersensitivity to ionizing radiation. It is estimated that 1% of the general population may be heterozygote carriers of ATM variants.[43] More than 300 variants in the gene have been identified, most of which are truncating variants.[44] ATM proteins have been shown to play a role in cell cycle control.[45,46,47]In vitro, AT-deficient cells are sensitive to ionizing radiation and radiomimetic drugs, and lack cell cycle regulatory properties after exposure to radiation.[48] There is insufficient evidence to recommend against radiation therapy in carriers of a single ATM pathogenic variant (heterozygotes).
Initial studies searching for an excess of ATM pathogenic variants among breast cancer patients provided conflicting results, perhaps due to study design and variant testing strategies.[49,50,51,52,53,54,55,56,57,58,59] However, two large epidemiologic studies have demonstrated a statistically increased risk of breast cancer among female heterozygote carriers, with an estimated RR of approximately 2.0.[4,59] A meta-analysis modeled the risk of breast cancer to be 6.02% by age 50 years and 32.83% by age 80 years.[60] Given these risks, increased screening and other recommendations based on family history and age may be considered.
Some studies have shown an association between ATM and ovarian cancer,[61,62] although, at this time, there is no evidence to suggest an impact on risk management or disease characteristics.
CASP8andTGFB1
The Breast Cancer Association Consortium (BCAC), an international group of investigators, investigated single nucleotide variants (SNVs) identified in previous studies as possibly associated with excess breast cancer risk in 15,000 to 20,000 cases and 15,000 to 20,000 controls. Two SNVs, CASP8 D302H and TGFB1 L10P, were associated with invasive breast cancer with RRs of 0.88 (95% CI, 0.84–0.92) and 1.08 (95% CI, 1.04–1.11), respectively.[63]
RAD51
RAD51 and the family of RAD51-related genes, also known as RAD51 paralogs, are thought to encode proteins that are involved in DNA damage repair through homologous recombination and interaction with numerous other DNA repair proteins, including BRCA1 and BRCA2. RAD51 protein plays a central role in single-strand annealing in the DNA damage response. RAD51 recruitment to break sites and recombinational DNA repair depend on the RAD51 paralogs, although their precise cellular functions are poorly characterized.[64] Variants in these genes are thought to result in loss of RAD51 focus formation in response to DNA damage.[65]
One of five RAD51-related genes, RAD51C has been reported to be linked to both FA-like disorders and familial breast and ovarian cancers. The literature, however, has produced contradictory findings. In a study of 480 German families characterized by breast and ovarian cancers who were negative for BRCA1 and BRCA2 pathogenic variants, six monoallelic variants in RAD51C were found (frequency of 1.3%).[66] Another study screened 286 BRCA1/BRCA2-negative patients with breast cancer and/or ovarian cancer and found one likely pathogenic variant in RAD51C-G153D.[67]RAD51C pathogenic variants have also been reported in Australian, British, Finnish, and Spanish non-BRCA1/BRCA2 ovarian cancer–only and breast/ovarian cancer families, and in unselected ovarian cancer cases, with frequencies ranging from 0% to 3% in these populations.[5,12,68,69,70,71,72,73] In a sample of 206 high-risk Jewish women (including 79 of Ashkenazi origin) previously tested for the common Jewish pathogenic variants, two previously described and possibly pathogenic missense variants were detected.[74] Four additional studies were unable to confirm an association between the RAD51C gene and hereditary breast cancer or ovarian cancer.[75,76,77,78]
In addition to carriers of RAD51C pathogenic variants, there are other RAD51 paralogs, including RAD51B, RAD51D, RAD51L1, XRCC2, and XRCC3, that may be associated with breast and/or ovarian cancer risk,[6,12,79,80,81,82,83] although the clinical significance of these findings is unknown. In a case-control study of 3,429 ovarian cancer patients, RAD51C and RAD51D pathogenic variants were more commonly found in ovarian cancer cases (0.82%) than in controls (0.11%, P < .001).[84]
In addition to germline variants, different polymorphisms of RAD51 have been hypothesized to have reduced capacity to repair DNA defects, resulting in increased susceptibility to familial breast cancer. The Consortium of Investigators of Modifiers of BRCA1/BRCA2 (CIMBA) pooled data from 8,512 carriers of BRCA1 and BRCA2 pathogenic variants and found evidence of an increased risk of breast cancer among women who were BRCA2 carriers and who were homozygous for CC at the RAD51 135G→C SNV (hazard ratio, 1.17; 95% CI, 0.91–1.51).[85]
Several meta-analyses have investigated the association between the RAD51 135G→C polymorphism and breast cancer risk. There is significant overlap in the studies reported in these meta-analyses, significant variability in the characteristics of the populations included, and significant methodologic limitations to their findings.[86,87,88,89] A meta-analysis of nine epidemiologic studies involving 13,241 cases and 13,203 controls of unknown BRCA1/BRCA2 status found that women carrying the CC genotype had an increased risk of breast cancer compared with women with the GG or GC genotype (OR, 1.35; 95% CI, 1.04–1.74). A meta-analysis of 14 case-control studies involving 12,183 cases and 10,183 controls confirmed an increased risk only for women who were known BRCA2 carriers (OR, 4.92; 95% CI, 1.10–21.83).[90] Another meta-analysis of 12 studies included only studies of known BRCA-negative cases and found no association between RAD51 135G→C and breast cancer.[91]
In summary, among this conflicting data is substantial evidence for a modest association between germline variants in RAD51C and breast cancer and ovarian cancer. There is also evidence of an association between polymorphisms in RAD51 135G→C among women with homozygous CC genotypes and breast cancer, particularly among BRCA2 carriers. These associations are plausible given the known role of RAD51 in the maintenance of genomic stability.
Abraxas
Pathogenic variants in the BRCA1-interacting gene Abraxas were found in three Finnish breast cancer families and no controls.[92] The significance of this finding outside of this population is not yet known.
RECQL
Through full exome sequencing among high-risk Polish and Quebec-based French Canadian families, the RECQL gene was discovered to harbor multiple rare truncating variants in both populations.[93] (Refer to the Clinical Sequencing section in the Cancer Genetics Overview PDQ summary for more information about whole-exome sequencing.) In the same populations, truncating variants in this gene were also identified in two subsequent validation phases among additional breast cancer patients from high-risk families, and among additional breast cancer cases in which the variant frequency was higher than that observed among controls. A case-control study from Belarus and Germany looked at the most common pathogenic variant, c.1667_1667+3delA GTA, and found it to be linked to ER-positive breast cancer. The OR in this study alone was 1.23 (95% CI, 0.44–3.47; P = .69), but in a meta-analysis with a Polish study, the OR was 2.51 (95% CI, 1.13–5.57, P = .02).[94] Although study results suggest that truncating germline RECQL pathogenic variants are associated with an increased risk of breast cancer, the exact magnitude of risk remains uncertain, and future studies are needed to determine clinical usefulness. Furthermore, the significance of this finding outside of these two populations is not yet known.
SMARCA4
SMARCA4 encodes BRG1 and is a catalytic subunit of the SWI/SNF chromatin remodeling complex, which plays a major role in rendering chromatin accessible to regulation of gene expression.
Small cell carcinoma of the ovary, hypercalcemic type (SCCOHT) is a rare, aggressive tumor that has an early age at onset (before age 40 y) and a poor prognosis.[95,96,97] Familial clustering is sometimes present. SCCOHT tumors may be unilateral or bilateral and have been characterized histologically by the presence of small hyperchromatic cells with brisk mitotic activity.[96] A multimodality approach including surgery, chemotherapy, and radiation therapy has been suggested for the treatment of SCCOHT.[96,97] Given the paraneoplastic phenomenon of hypercalcemia in 60% of cases, tracking calcium levels is useful in monitoring the course of disease. With a wide range of differential diagnoses including germ cell tumors, sex cord–stromal tumors, and undifferentiated carcinomas, SCCOHT remains classified by the World Health Organization as a "miscellaneous tumor" but more recently has been sequenced to be a malignant rhabdoid tumor.[98] Through exome sequencing, most cases of SCCOHT have been found to lack functional SMARCA4/BRG1; in fact, pathogenic variants in SMARCA4 may be the sole variants responsible for SCCOHT.
Despite only approximately 300 cases in the literature, three separate research groups showed SCCOHT to be associated with germline pathogenic variants and somatic mutations in the SMARCA4 gene. In one study of 12 young women with SCCOHT, sequencing of paired tumor and normal samples identified inactivating biallelic SMARCA4 pathogenic variants in each case.[99] Only four additional nonrecurrent somatic genes were identified in any of the other 278 genes sequenced. Immunohistochemistry demonstrated loss of SMARCA4 protein expression in seven of nine tested cases, consistent with a tumor-suppressor gene function. In a second study of another 12 patients, next-generation sequencing also identified SMARCA4 as the only recurrently variant gene, with the majority of variants predicted to result in a truncated protein.[100] A third study included three families in whom whole-exome sequencing with Sanger sequencing confirmation identified at least one germline pathogenic variant or somatic mutation in 24 of 26 cases.[101] Overall, 38 of 43 (88%) of SCCOHT tumors showed loss of SMARCA4 expression, in comparison to only 1 of 139 (0.7%) other ovarian tumor types.
Because of the rarity of this tumor, the penetrance of SMARCA4 is unknown. There is currently no consensus for management, yet SMARCA4 is on the larger multigene panels currently available for genetic testing, and risk-reducing surgery has been offered to pathogenic variant carriers.[102]
References:
Polymorphisms underlying polygenic susceptibility to breast and gynecologic cancers are considered low penetrance, a term often applied to sequence variants associated with a minimal to moderate risk. This is in contrast to high-penetrance variants or alleles that are typically associated with more severe phenotypes, for example BRCA1/BRCA2 pathogenic variants leading to an autosomal dominant inheritance pattern in a family, and moderate-penetrance variants such as BRIP1, CHEK2, and RAD51C. (Refer to the High-Penetrance Breast and/or Gynecologic Cancer Susceptibility Genes and the Moderate-Penetrance Genes Associated With Breast and/or Gynecologic Cancer sections of this summary for more information.) Because these types of sequence variants (also called low-penetrance genes, alleles, variants, and polymorphisms) are relatively common in the general population, their overall contribution to cancer risk is estimated to be much greater than the attributable risk in the population from pathogenic variants in BRCA1 and BRCA2. For example, it is estimated by segregation analysis that half of all breast cancer occurs in 12% of the population that is deemed most susceptible.[1] There are no known low-penetrance variants in BRCA1/BRCA2. The N372H variation in BRCA2, initially thought to be a low-penetrance allele, was not verified in a large combined analysis.[2]
Two strategies have attempted to identify low-penetrance polymorphisms leading to breast cancer susceptibility: candidate gene and genome-wide searches. Both involve the epidemiologic case-control study design. The candidate gene approach involves selecting genes based on their known or presumed biological function, relevance to carcinogenesis or organ physiology, and then searching for or testing known genetic variants for an association with cancer risk. This strategy relies on imperfect and incomplete biological knowledge, and, despite some confirmed associations (described below), has been relatively disappointing.[2,3] The candidate gene approach has largely been replaced by genome-wide association studies (GWAS) in which a very large number of single nucleotide variants (SNVs) (approximately 1 million to 5 million) are chosen within the genome and tested, mostly without regard to their possible biological function, but instead to more uniformly capture all genetic variation throughout the genome.
Genome-Wide Searches
In contrast to assessing candidate genes and/or alleles, GWAS involve comparing a very large set of genetic variants spread throughout the genome. The current paradigm uses sets of as many as 5 million SNVs that are chosen to capture a large portion of common variation within the genome based on the HapMap and the 1000 Genomes Project.[4,5] By comparing allele frequencies between a large number of cases and controls, typically 1,000 or more of each, and validating promising signals in replication sets of subjects, very robust statistical signals of association have been obtained.[6,7,8] The strong correlation between many SNVs that are physically close to each other on the chromosome (linkage disequilibrium) allows one to "scan" the genome for susceptibility alleles even if the biologically relevant variant is not within the tested set of SNVs. Although this between-SNV correlation allows one to interrogate the majority of the genome without having to assay every SNV, when a validated association is obtained, it is not usually obvious which of the many correlated variants is causal.
Genome-wide searches are showing great promise in identifying common, low-penetrance susceptibility alleles for many complex diseases,[9] including breast cancer.[10,11,12,13] The first study involved an initial scan in familial breast cancer cases followed by replication in two large sample sets of sporadic breast cancer, the final being a collection of over 20,000 cases and 20,000 controls from the Breast Cancer Association Consortium.[10] Five distinct genomic regions were identified that were within or near the FGFR2, TNRC9, MAP3K1, and LSP1 genes or at the chromosome 8q region. The 8q region and others may harbor multiple independent loci associated with risk. Subsequent genome-wide studies have replicated these loci and identified additional ones.[11,12,14,14,15,16,17,18,19] Numerous SNVs identified through large studies of sporadic breast cancer appear to be associated more strongly with estrogen receptor (ER)–positive disease;[20] however, some are associated primarily or exclusively with other subtypes, including triple-negative disease.[21,22] An online catalog is available of SNV-trait associations from published GWAS for use in investigating genomic characteristics of trait/disease-associated SNVs.
Although the statistical evidence for an association between genetic variation at these loci and breast and ovarian cancer risk is overwhelming, the biologically relevant variants and the mechanism by which they lead to increased risk are unknown and will require further genetic and functional characterization. Additionally, these loci are associated with very modest risk (typically, an odds ratio [OR] <1.5), with more risk variants likely to be identified. No interaction between the SNVs and epidemiologic risk factors for breast cancer have been identified.[23,24] Furthermore, theoretical models have suggested that common moderate-risk SNVs have limited potential to improve models for individualized risk assessment.[25,26,27] These models used receiver operating characteristic (ROC) curve analysis to calculate the area under the curve (AUC) as a measure of discriminatory accuracy. A subsequent study used ROC curve analysis to examine the utility of SNVs in a clinical dataset of more than 5,500 breast cancer cases and nearly 6,000 controls, using a model with traditional risk factors compared with a model using both standard risk factors and ten previously identified SNVs. The addition of genetic information modestly changed the AUC from 58% to 61.8%, a result that was not felt to be clinically significant. Despite this, 32.5% of patients were in a higher quintile of breast cancer risk when genetic information was included, and 20.4% were in a lower quintile of risk. Whether such information has clinical utility is unclear.[25,28]
More limited data are available regarding ovarian cancer risk. Three GWAS involving staged analysis of more than 10,000 cases and 13,000 controls have been carried out for ovarian cancer.[29,30,31] As in other GWAS, the ORs are modest, generally about 1.2 or weaker but implicate a number of genes with plausible biological ties to ovarian cancer, such as BABAM1, whose protein complexes with and may regulate BRCA1, and TIRAPR, which codes for a poly (ADP-ribose) polymerase, molecules that may be important in BRCA1/BRCA2-deficient cells.
Polygenic risk scores for breast and ovarian cancer
The collective influence of many genetic variants has more recently been evaluated using an aggregate score. In 2015, a polygenic risk score (PRS) comprising all of the known breast cancer risk genetic variants or SNVs was estimated in women of European ancestry using 41 studies in the Breast Cancer Association Consortium (BCAC), including more than 33,000 breast cancer cases and 33,000 controls.[32] This early attempt at estimating a PRS for breast cancer included 77 SNVs, which collectively conferred lifetime risks of developing breast cancer by age 80 years of 3.5% and 29% for women in the lowest and highest 1% of the PRS, respectively.[32] Since then, PRSs incorporating additional genetic variants and examining other breast cancer–related outcomes including tumor and pathological characteristics, mode of detection, and contralateral breast cancer (CBC) have been estimated.[33,34,35,36,37,38,39,40] In 2019, the PRS with the highest discriminatory ability to date was developed and prospectively validated in the largest GWAS datasets available (79 studies in BCAC and more than 190,000 women in the U.K. Biobank), which incorporates information on 313 genetic variants and is optimized for ER-positive and ER-negative breast cancer.[39] Compared with women in the middle quintile, those in the highest 1% of PRS313 had 4.04-, 4.37-, and 2.78-fold risks of developing breast cancer overall, ER-positive disease, and ER-negative disease, respectively.[39] Lifetime absolute risk of breast cancer by age 80 years for women in the lowest and highest 1% of PRS313 ranged from 2% to 31% for ER-positive breast cancer, while for ER-negative disease, the absolute risks ranged from 0.55% to 4%.[39]
Common genomic variants associated with the development of a first primary breast cancer are also associated with the development of CBC.[40] Women in the highest quartile of the PRS had a 1.6-fold increased risk of developing CBC compared with the lowest quartile.[40] Moreover, PRSs of breast and ovarian cancers have been assessed in women who are carriers of BRCA1 and BRCA2 pathogenic variants, and have been found to be predictive of cancer risk in these women, supporting the hypothesis of a shared polygenic component of cancer risk between the general population and variant carriers.[36] The PRS for ER-negative disease had the strongest association with breast cancer risk in BRCA1 variant carriers, while the strongest association in BRCA2 variant carriers was seen for the overall breast cancer PRS. BRCA1 variant carriers had cumulative lifetime risks of 56% and 75% of developing breast cancer at the 10th and 90th percentile of the PRS, respectively. The ovarian cancer PRS was strongly associated with risk for both BRCA1 and BRCA2 variant carriers. For BRCA2 variant carriers, the ovarian cancer risk was 6% and 19% by age 80 years for those at the 10th and 90th percentile of PRS, respectively. The authors noted that the incorporation of the PRS into risk prediction models may better inform decisions on cancer risk management for this population.[36]
Several studies have examined the extent to which clinical breast cancer risk prediction models can be improved by including information on known susceptibility SNVs, and reporting improved discriminatory accuracy after inclusion of the PRS.[41,42,43,44,45,46] For example, in a study combining PRS77 with clinical models, the AUC for predicting breast cancer before age 50 years improved by more than 20%.[42] Clinical trials, including WISDOM and MyPeBs, are in progress to study the potential clinical utility of the PRS for making screening decisions and understanding outcomes.[47] Because PRSs have been largely developed and validated in populations of European ancestry, the utility and prediction accuracy of these PRSs in non-European populations is unknown.
A large study examined whether known reproductive and lifestyle risk factors interact with PRSs to increase breast cancer risk and did not find a multiplicative interaction with established risk factors.[48]
Whole-Genome and Whole-Exome Sequencing
In addition to GWAS interrogating common genetic variants, sequencing-based studies involving whole-genome or whole-exome sequencing [49] are also identifying genes associated with breast cancer, such as XRCC2, a rare, moderate-penetrance breast cancer susceptibility gene.[50] (Refer to the Clinical Sequencing section in the PDQ summary on Cancer Genetics Overview for more information about whole-exome sequencing.)
References:
Increasing data are available on the outcomes of interventions to reduce risk in people with a genetic susceptibility to breast cancer or ovarian cancer.[1,2,3,4,5,6,7] As outlined in other sections of this summary, uncertainty is often considerable regarding the level of cancer risk associated with a positive family history or genetic test. In this setting, personal preferences are likely to be an important factor in patients' decisions about risk reduction strategies.
Screening and Prevention Strategies
Breast cancer
Screening/surveillance
Refer to the PDQ summary on Breast Cancer Screening for information on screening in the general population, and to the PDQ summary Levels of Evidence for Cancer Genetics Studies for information on levels of evidence related to screening and prevention.
Breast self-examination
In the general population, evidence for the value of breast self-examination (BSE) is limited. Preliminary results have been reported from a randomized study of BSE being conducted in Shanghai, China.[8] At 5 years, no reduction in breast cancer mortality was seen in the BSE group compared with the control group of women, nor was a substantive stage shift seen in breast cancers that were diagnosed. (Refer to the PDQ summary on Breast Cancer Screening for more information.)
Little direct prospective evidence exists regarding BSE in individuals with an increased risk of breast cancer. In the Canadian National Breast Screening Study, women with first-degree relatives (FDRs) with breast cancer had statistically significantly higher BSE competency scores than those without a family history. In a study of 251 high-risk women at a referral center, five breast cancers were detected by self-examination less than a year after a previous screen (as compared with one cancer detected by clinician exam and 11 cancers detected as a result of mammography). Women in the cohort were instructed in self-examination, but it is not stated whether the interval cancers were detected as a result of planned self-examination or incidental discovery of breast masses.[9] In another series of carriers of BRCA1/BRCA2pathogenic variants, four of nine incident cancers were diagnosed as palpable masses after a reportedly normal mammogram, further suggesting the potential value of self-examination.[10] A task force convened by the Cancer Genetics Studies Consortium has recommended "monthly self-examination beginning early in adult life (e.g., by age 18–21 y) to establish a regular habit and allow familiarity with the normal characteristics of breast tissue. Education and instruction in self-examination are recommended."[11]
Level of evidence: 5
Clinical breast examination
Few prospective data exist regarding clinical breast examination (CBE).
The Cancer Genetics Studies Consortium task force concluded, "As with self-examination, the contribution of clinical examination may be particularly important for women at inherited risk of early breast cancer." They recommended that female carriers of a BRCA1 or BRCA2 high-risk pathogenic variant undergo annual or semiannual clinical examinations beginning at age 25 to 35 years.[11]
Level of evidence: 5
Mammography
In the general population, strong evidence suggests that regular mammography screening of women aged 50 to 59 years leads to a 25% to 30% reduction in breast cancer mortality. (Refer to the PDQ summary on Breast Cancer Screening for more information.) For women who begin mammographic screening at age 40 to 49 years, a 17% reduction in breast cancer mortality is seen, which occurs 15 years after the start of screening.[12] Observational data from a cohort study of more than 28,000 women suggest that the sensitivity of mammography is lower for young women. In this study, the sensitivity was lowest for younger women (aged 30–49 y) who had an FDR with breast cancer. For these women, mammography detected 69% of breast cancers diagnosed within 13 months of the first screening mammography. By contrast, sensitivity for women younger than 50 years without a family history was 88% (P = .08). For women aged 50 years and older, sensitivity was 93% at 13 months and did not vary by family history.[13] Preliminary data suggest that mammography sensitivity is lower in BRCA1 and BRCA2 carriers than in noncarriers.[10] Subsequent observational studies have found that the positive predictive value (PPV) of mammography increases with age and is highest among older women and among women with a family history of breast cancer.[14] Higher PPVs may be due to increased breast cancer incidence, higher sensitivity, and/or higher specificity.[15] One study found an association between the presence of pushing margins and false-negative mammograms in 28 women, 26 of whom had a BRCA1 pathogenic variant and two of whom had a BRCA2 pathogenic variant. Pushing margins, a mammographic characteristic of medullary histology, are associated with an absence of fibrotic reaction.[16] In addition, rapid tumor doubling times may lead to tumors presenting shortly after an apparently normal study. In one study, mean tumor doubling time in BRCA1/BRCA2 carriers was 45 days, compared with 84 days in noncarriers.[17] Another study that evaluated mammographic breast density in women with BRCA pathogenic variants found no association between pathogenic variant status and mammographic density; however, in both carriers and noncarriers, increased breast density was associated with increased breast cancer risk.[18]
The randomized Canadian National Breast Screening Study-2 compared annual CBE plus mammography to CBE alone in women aged 50 to 59 years from the general population. Both groups were given instruction in BSE.[19] Although mammography detected smaller primary invasive tumors, more invasive cancers, and more ductal carcinoma in situ (DCIS) than CBE, the breast cancer mortality rates in the CBE-plus-mammography group and the CBE-alone group were nearly identical, and compared favorably with other breast cancer screening trials. After a mean follow-up of 13 years (range, 11.3–16.0 y), the cumulative breast cancer mortality ratio was 1.02 (95% confidence interval [CI], 0.78–1.33). One possible explanation of this finding was the careful training and supervision of the health professionals performing CBE.
Digital mammography refers to the use of a digital detector to find and record x-ray images. This technology improves contrast resolution [20] and has been proposed as a potential strategy for improving the sensitivity of mammography. A screening study comparing digital with routine mammography in 6,736 examinations of women aged 40 years and older found no difference in cancer detection rates;[21] however, digital mammography resulted in fewer recalls. In another study (ACRIN-6652) comparing digital mammography to plain-film mammography in 42,760 women, the overall diagnostic accuracy of the two techniques was similar.[22] When receiver operating characteristic curves were compared, digital mammography was more accurate in women younger than 50 years, in women with radiographically dense breasts, and in premenopausal or perimenopausal women.
In a prospective study of 251 individuals with BRCA pathogenic variants who received uniform recommendations regarding screening and risk-reducing surgery, annual mammography detected breast cancer in six women at a mean of 20.2 months after receipt of BRCA results.[9] The Cancer Genetics Studies Consortium task force has recommended that female carriers of a BRCA1 or BRCA2 high-risk pathogenic variant have, "annual mammography, beginning at age 25 to 35 years. Mammograms should be done at a consistent location when possible, with prior films available for comparison."[11] Data from prospective studies on the relative benefits and risks of screening with an ionizing radiation tool versus CBE or other nonionizing radiation tools would be useful.[23,24,25]
Certain observations raise concerns that carriers of BRCA pathogenic variants may be more prone to radiation-induced breast cancer than women without pathogenic variants. The BRCA1 and BRCA2 proteins are known to be important in cellular mechanisms of DNA damage repair, including those involved in repairing radiation-induced damage. Some studies have suggested intermediate radiation sensitivity in cells that are heterozygous for a BRCA variant, but this is not consistent and varies by experimental system and endpoint.
Three studies have failed to find convincing evidence of an association between ionizing radiation exposure and increased breast cancer risk in carriers of BRCA1 and BRCA2 pathogenic variants.[26,27,28] In contrast, two large international studies found evidence of an increased breast cancer risk due to chest x-rays [29] or estimates of total exposure to diagnostic radiation.[30] A large, international, case-control study of 1,601 carriers of pathogenic variants described an increased risk of breast cancer (hazard ratio [HR], 1.54) among women who were ever exposed to chest x-rays, with risk being highest in women aged 40 years and younger, born after 1949, and exposed to x-rays only before age 20 years.[29] Some of the subjects in this study were also included in a larger, more comprehensive analysis of carriers of pathogenic variants from three European centers.[30] In this study of 1,993 carriers of BRCA1 and BRCA2 pathogenic variants from the United Kingdom, France, and the Netherlands, age-specific total diagnostic radiation exposure (e.g., chest x-rays, mammography, fluoroscopy, and computed tomography) estimates were derived from self-reported questionnaires. Women who were exposed to radiation before age 30 years had an increased risk (HR, 1.90; 95% CI, 1.20–3.00) when compared with those who were never exposed. This risk was primarily driven by nonmammographic radiation exposure in women younger than 20 years (HR, 1.62; 95% CI, 1.02–2.58). Subsequently, a prospective study of 1,844 BRCA1 carriers and 502 BRCA2 carriers without a breast cancer diagnosis at study entry, with an average follow-up time of 5.3 years, observed no significant association between prior mammography exposure and breast cancer risk.[28] Additional subgroup analyses in women younger than 30 years demonstrated no association with breast cancer risk.
With the routine use of magnetic resonance imaging (MRI) in carriers of BRCA1 and BRCA2 pathogenic variants, any potential benefit of mammographic screening must be carefully weighed against potential risks, particularly in young women.[31] One study has suggested that the most cost-effective screening strategy in carriers of BRCA1 and BRCA2 pathogenic variants may be annual MRI beginning at age 25 years, with alternating MRI and digital mammography (so that each test is done annually but screening occurs every 6 months) beginning at age 30 years.[32] The National Comprehensive Cancer Network (NCCN) currently recommends annual breast MRI screening with contrast (or mammogram with consideration of tomosynthesis, only if MRI is unavailable) between ages 25 and 29 years and annual mammogram (with consideration of tomosynthesis and breast MRI screening with contrast) between ages 30 and 75 years.[33]
Magnetic resonance imaging
Because of the relative insensitivity of mammography in women with an inherited risk of breast cancer, a number of screening modalities have been proposed and investigated in high-risk women, including carriers of BRCA pathogenic variants. Many studies have described the experience with breast MRI screening in women at risk of breast cancer, including descriptions of relatively large multi-institutional trials.[34,35,36,37,38,39,40,41,42]
Despite some limitations, these studies consistently demonstrate that breast MRI is more sensitive than either mammography or ultrasound for the detection of hereditary breast cancer. The results of six large studies are presented in Table 11, Summary of MRI Screening Studies in Women at Hereditary Risk of Breast Cancer.[34,36,37,40,43,44] Most cancers in these programs were screen detected, with only 6% of cancers presenting in the interval between screenings. The sensitivity of MRI (as defined by the study methodology) ranged from 71% to 100%. Of the combined studies, 77% of cancers were identified by MRI, and 42% were identified by mammography.
Concerns have been raised about the reduced specificity of MRI compared with other screening modalities. In one study, after the initial MRI screen, 16.5% of patients were recalled for further evaluation and an additional 7.6% of patients were recommended to undergo a short-interval follow-up examination at 6 months.[37] These rates declined significantly during later screening rounds, with fewer than 10% of the subjects recalled for more detailed MRI and fewer than 3% recommended to have short interval follow-up. In a second study, Magnetic Resonance Imaging for Breast Screening (MARIBS), the recall rate for additional evaluation was 10.7% per year.[36] The benign biopsy rates in the first study were 11% at first round, 6.6% at second round, and 4.7% at third round.[37] In the MARIBS study, the aggregate surgical biopsy rate was 9 per 1,000 screening episodes, though this may underestimate the burden because follow-up ultrasounds, core-needle biopsies, and fine-needle aspirations have not been included in the numerator of the MARIBS calculation.[36] The PPV of MRI has been calculated differently in the various series and fluctuates somewhat, depending on whether all abnormal examinations or only the examinations that result in a biopsy are counted in the denominator. Generally, the PPV of a recommendation for tissue sampling (as opposed to further investigation) is in the range of 50% in most series.
These trials appear to establish that MRI is superior to mammography in the detection of hereditary breast cancer; women who participated in these trials, which included annual MRI, were less likely to have a cancer missed by screening.[45] However, mammography may identify some cancers, particularly DCIS, that are not identified by MRI.[46]
Regarding downstaging, one screening study demonstrated that patients at risk of hereditary breast cancer were more likely to be diagnosed with small tumors and node-negative disease than were women in two nonrandomized control groups.[34] A multicenter randomized trial compared annual breast MRI, annual CBE, and biannual mammogram with annual mammogram and CBE among women who tested negative for BRCA1/BRCA2 pathogenic variants but with a lifetime familial risk of breast cancer of 20% or higher. Invasive cancers in the MRI arm were smaller, had fewer positive nodes, and were diagnosed at lower stages than those in the mammogram-alone arm.[47]
Despite the apparent sensitivity of MRI screening, some women in MRI-based programs will develop life-threatening breast cancer. In a prospective study of 51 carriers of BRCA1 pathogenic variants and 41 carriers of BRCA2 pathogenic variants screened with yearly mammograms and MRIs (of whom 80 had risk-reducing oophorectomy), 11 breast cancers (9 invasive and 2 DCIS) were detected. Six cancers were first detected on MRI; three were first detected by mammogram; and two were interval cancers. All breast cancers occurred in carriers of BRCA1 pathogenic variants, suggesting a continued high risk of BRCA1-related breast cancer after oophorectomy in the short term. These results suggest that surveillance and prevention strategies may have differing outcomes in carriers of BRCA1 and BRCA2 pathogenic variants.[41]
A publication combining results from three large studies (MARIBS, a Canadian study, and a Dutch MRI screening study) demonstrated that when MRI was added to mammography, 80% of cancers detected in carriers of BRCA2 pathogenic variants were either DCIS or invasive cancers smaller than 1 cm. In carriers of BRCA1 pathogenic variants, 49% of cancers were DCIS or small invasive cancers. In addition, the authors predicted mortality benefits with the addition of MRI for both carriers of BRCA1 and BRCA2 pathogenic variants. The model predicted breast cancer mortality reductions of 42% to 47% for mammography, 48% to 61% for MRI, and 50% to 62% for combined screening.[48] An additional study examining carriers of BRCA1/BRCA2 pathogenic variants undergoing MRI between 1997 and 2006 has demonstrated that 97% of incident cancers were stage 0 or stage I.[49] A 2015 Dutch case-control study further evaluated 2,308 high-risk patients, including 706 women with known BRCA pathogenic variants, who were screened with mammogram and compared them with those who also had MRI.[50] Of the patients screened, 93 patients were detected to have 97 cancers, 33 patients had a BRCA1 pathogenic variant, and 18 patients had a BRCA2 pathogenic variant. With a median follow-up of 9 years, metastases-free survival was improved in the MRI-screened cohort (90% vs. 77%), but it did not reach statistical significance in the BRCA1 and BRCA2 subset because of a very small sample size. All MRI-screened patients were more likely to be node-negative and receive less chemotherapy. The American Cancer Society and NCCN have recommended the use of annual MRI screening for women at hereditary risk of breast cancer.[33,51]
An additional question regarding the timing of mammography and MRI is whether they should be done simultaneously or in an alternating fashion (so that while each test is done annually, screening occurs every 6 months). One study has suggested that the most cost-effective screening strategy in carriers of BRCA1 and BRCA2 pathogenic variants may be annual MRI beginning at age 25 years, with alternating MRI and digital mammography beginning at age 30 years.[32]
A study of women who had undergone risk-reducing mastectomy reported a low yield from annual MRI screening based on a retrospective review of 159 women, 58% of whom had pathogenic BRCA variants, over an 8-year period. MRI was negative in 98% of patients and was associated with a false-positive rate of 90%. The yield of surveillance MRI was very low in this cohort, bringing into question the added value of routine surveillance MRI in this population.[52]
In summary, evidence strongly supports the integral role of breast MRI in breast cancer surveillance for carriers of BRCA1/BRCA2 pathogenic variants.
Series | Rijnsburger[42] | Warner[37] | MARIBS[36] | Kuhl[40] | Weinstein[43] | Sardanelli[44] | Totals | |
---|---|---|---|---|---|---|---|---|
a Based on the first 1,909 women screened.[34] | ||||||||
b Includes patients with invasive cancer only and patients with both invasive andin situ cancers. | ||||||||
c Includes only 75 cancers detected in women who underwent both mammographic and MRI screening. | ||||||||
d Restricted to studies in which ultrasound was performed. | ||||||||
N Patients | Overall | 2,157 | 236 | 649 | 687 | 609 | 501 | 4,839 |
BRCA1/BRCA2Carriers | 594 | 236 | 120 | 65 | 44 | 330 | 1,389 | |
N Screening Episodes | 6,253 | 457 | 1,881 | 1,679 | 1,592 | 11,862 | ||
N Cancers | Baseline | 22a | 13 | 20 | 10 | 0 | 0 | 65 |
Subsequent | 97 | 9 | 15 | 17 | 18 | 52 | 208 | |
Invasiveb | 78 | 16 | 29 | 8 | 11 | 44 | 186 | |
In situ | 19 | 9 | 6 | 9 | 7 | 8 | 58 | |
Annual Incidence | 10.4/1,000 | 19/1,000 | ||||||
Detected at Planned Screening | 78 | 21 | 33 | 27 | 18 | 49 | 226 (83%) | |
N Detected by Each Modality | Mammography | 31c | 8 | 14 | 9 | 7 | 25 | 94 (42%) |
MRI | 51c | 17 | 27 | 25 | 12 | 42 | 174 (77%) | |
Ultrasoundd | 7 | 10 | 3 | 26 | 46 (41%) | |||
Follow-up | Median of 4.9 y | Minimum of 1 y | 2–7 y | Median of 29.09 mo | 2 y | 3 y |
Level of evidence: 3
Ultrasound
Several studies have reported instances of breast cancer detected by ultrasound that were missed by mammography, as discussed in one review.[53] In a pilot study of ultrasound as an adjunct to mammography in 149 women with moderately increased risk based on family history, one cancer was detected, based on ultrasound findings. Nine other biopsies of benign lesions were performed. One was based on abnormalities on both mammography and ultrasound, and the remaining eight were based on abnormalities on ultrasound alone.[53] A large study of 2,809 women with dense breast tissue (ACRIN-6666) demonstrated that ultrasound increased the detection rate due to breast cancer screening from 7.6 per 1,000 with mammography alone to 11.8 per 1,000 for combined mammography and ultrasound.[54] However, ultrasound screening increases false-positive rates and appears to have a limited benefit in combination with MRI. In a multicenter study of 171 women (92% of whom were carriers of BRCA1/BRCA2 pathogenic variants) undergoing simultaneous mammography, MRI, and ultrasound, no cancers were detected by ultrasound alone.[38] Uncertainties about ultrasound include the effect of screening on mortality, the rate and outcome of false-positive results, and access to experienced breast ultrasonographers.
Level of evidence: None assigned
Other screening modalities
A number of other techniques are under active investigation, including tomosynthesis, contrast-enhanced mammography, thermography, and radionuclide scanning. Additional evidence is needed before these techniques can be incorporated into clinical practice.
Level of evidence: None assigned
Risk-reducing surgeries
Risk-reducing mastectomy
Risk-reducing mastectomy (RRM) is a management option for patients who are considered to be at high risk of developing breast cancer. The Society of Surgical Oncology has endorsed RRM as an option for women with BRCA1/BRCA2 pathogenic variants or strong family histories of breast cancer.[55] Historically, a total or simple mastectomy has been performed, which includes removal of all of the breast tissue, including the nipple and areolar complex (NAC). If the patient is interested, reconstruction can be performed simultaneously with the ablative portion of the procedure. Options for reconstruction include tissue expander and implant-based reconstructions or autologous reconstructions, in which the patient's own tissue is used to reconstruct the breast. A number of different tissues can be used to reconstruct the breast, including flaps based on the latissimus dorsi muscle, the transverse rectus abdominis muscle, or the gluteus muscle. Muscle-sparing techniques such as the deep inferior epigastric perforator flap can also be used, but these require advanced microvascular techniques. In the interest of improved cosmetic outcomes, skin-sparing techniques have been utilized in which the entire breast is removed with the NAC, but the entire skin envelope of the breast is preserved. In a further refinement, nipple-sparing techniques have been developed in which all of the breast skin and the nipple are preserved while the underlying glandular tissue is removed.
Risk-reducing mastectomy in unaffected women
Because there are no randomized, prospective trials of RRM versus observation, data are limited to cohort and case-control studies. The available data demonstrate that RRM does decrease breast cancer incidence in high-risk patients,[56,57,58] but overall survival (OS) correlates more closely with the overall risk from the primary incidence of breast cancer. Several studies have analyzed the impact of RRM on breast cancer risk and mortality. In one retrospective cohort study of 214 women considered to be at hereditary risk by virtue of a family history suggesting an autosomal dominant predisposition, three women were diagnosed with breast cancer after bilateral RRM, with a median follow-up of 14 years.[58] Because 37.4 cancers were expected, the calculated risk reduction was 92.0% (95% CI, 76.6%–98.3%). In a follow-up subset analysis, 176 of the 214 high-risk women in this cohort study underwent genetic testing for pathogenic variants in BRCA1 and BRCA2. Pathogenic variants were identified in 18 women, none of whom developed breast cancer after a median follow-up of 13.4 years.[56] Two of the three women diagnosed with breast cancer after RRM were tested, and neither carried a pathogenic variant. The calculated risk reduction among carriers of pathogenic variants was 89.5% to 100.0% (95% CI, 41.4%–100.0%), depending on the assumptions made about the expected numbers of cancers among carriers of pathogenic variants and the status of the untested woman who developed cancer despite mastectomy. The result of this retrospective cohort study has been supported by a prospective analysis of 76 carriers of pathogenic variants who underwent RRM and were monitored prospectively for a mean of 2.9 years. No breast cancers were observed in these women, whereas eight were identified in women who underwent regular surveillance (HR for breast cancer after RRM, 0.00 [95% CI, 0.00–0.36]).[57]
The Prevention and Observation of Surgical Endpoints study group also estimated the degree of breast cancer risk reduction after RRM in carriers of BRCA1/BRCA2 pathogenic variants. The rate of breast cancer in 105 carriers of pathogenic variants who underwent bilateral RRM was compared with that of 378 carriers who did not choose surgery. Bilateral mastectomy reduced the risk of breast cancer by approximately 90% after a mean follow-up of 6.4 years.[3]
Theoretical models have also been utilized to assess the role of RRM in women with pathogenic variants in BRCA1 and BRCA2. Assuming risk reduction in the range of 90%, one model suggests that for a group of women, aged 30 years, with BRCA1 or BRCA2 pathogenic variants, RRM would result in an average increased life expectancy of 2.9 to 5.3 years.[59] A computer-simulated survival analysis using a Monte Carlo model included breast MRI, mammography, RRM, and risk-reducing salpingo-oophorectomy (RRSO) and examined the impact of each intervention separately on carriers of BRCA1 and BRCA2 pathogenic variants.[5] The most effective strategy was found to be RRSO at age 40 years and RRM at age 25 years, with survival at age 70 years approaching that of the general population. However, delaying mastectomy until age 40 years, or substituting RRM with screening with breast MRI and mammography, had little impact on survival estimates. For example, replacing RRM with MRI-based screening in women with RRSO at age 40 years led to a 3% to 5% decrement in survival compared with RRM at age 25 years.[60] As with any models, numerous assumptions cause uncertainty; however, these studies provide additional information for women and their providers who are making these difficult decisions.
Another study of at-risk women showed a 70% time–trade-off value for RRM, indicating that participants were willing to sacrifice 30% of life expectancy to avoid RRM.[61] A cost-effectiveness analysis study of RRM has also been performed. The investigators concluded that, compared with surveillance, risk-reducing surgery (mastectomy and oophorectomy) is cost-effective with regard to years of life saved, but not for improved quality of life.[62] While these data are interesting and may be useful for public policy decisions, they cannot be individualized for clinical care because they include assumptions that cannot be fully tested.
Level of evidence: 3ai
Contralateral risk-reducing mastectomy in affected women
If RRM is effective in lowering breast cancer risk in unaffected women, what is its role in women with unilateral breast cancer? This question often arises in discussions about surgical options with women who have unilateral breast cancer and hereditary risks. This section addresses the role of contralateral risk-reducing mastectomy (CRRM) in women being treated with mastectomy and will not discuss breast conservation therapy. Multiple studies have shown an increase in the rate of CRRM in women with unilateral breast cancer.[63,64] When the appropriateness of CRRM is being assessed for women with unilateral breast cancer, the first task is to determine the risk of contralateral breast cancer (CBC).
In the general population, current estimates of CBC risk after breast cancer treatment are approximately 0.3% per year and are declining.[65] In carriers of BRCA pathogenic variants with a diagnosis of breast cancer, the risk of a second, unrelated breast cancer is related to age at initial diagnosis, biology, and systemic therapies used, but is clearly higher than that in the general population.[66] (Refer to the Contralateral breast cancer in carriers of BRCA pathogenic variants section in the High-Penetrance Breast and/or Gynecologic Cancer Susceptibility Genes section of this summary for more information about the risk of CBC in this population.) In carriers of BRCA pathogenic variants whose first cancer has an excellent prognosis, estimating the risk of a second, unrelated breast cancer event is important for informing their decision to undergo risk-reducing surgery and has been described in this setting to improve survival.[67] The timing of genetic testing and knowledge of BRCA pathogenic variant status may influence surgical decision making, may prevent subsequent surgeries, and may influence follow-up care. Therefore, for individuals at increased risk of carrying a BRCA pathogenic variant, it is important that genetic testing be considered in advance of surgery, when possible.[68]
In a group of 148 carriers of BRCA1 and BRCA2 pathogenic variants with unilateral breast cancer, 79 of whom underwent CRRM, the risk of CBC was reduced by 91% and was independent of the effect of RRSO. Survival increased among women who underwent CRRM, but this result was likely associated with higher mortality caused by the index cancer or metachronous ovarian cancer in the group that did not have surgery.[69] Data from ten European centers on 550 women (including 202 BRCA carriers) with 3,334 woman-years of follow-up indicated that RRM was highly effective. Bilateral RRMs were carried out on women with a lifetime risk of 25% to 80%, with an average expected incidence rate of 1% per year. No breast cancers occurred in this cohort over the follow-up period, though more than 34 breast cancers would have been expected.[70] A retrospective study of 593 carriers of BRCA1 and BRCA2 pathogenic variants included 105 women with unilateral breast cancer who underwent CRRM and had a 10-year survival rate of 89%, compared with 71% in the group that did not undergo contralateral risk-reducing surgery (P < .001).[4] This study was limited by several factors, such as the lack of information regarding breast cancer screening, grade, and estrogen receptor (ER) status in a large portion of this sample.
A Dutch cohort of 583 patients identified between 1980 and 2011, who had both a BRCA pathogenic variant and a diagnosis of unilateral breast cancer, were evaluated for the effect of CRRM.[71] With a median follow-up of 11.4 years, 242 (42%) of the patients underwent RRM (193 carriers of BRCA1 pathogenic variants and 49 carriers of BRCA2 pathogenic variants) at differing times after their diagnoses. Improved OS was observed in the RRM group when compared with the surveillance group (HR, 0.49; 95% CI, 0.29–0.82), with improvements most pronounced in those diagnosed before age 40 years, with low tumor grade, and non–triple-negative subtype. In an attempt to control for the bias of time to surgery, the authors included a separate evaluation of women who were known to be disease free 2 years after the primary cancer diagnosis (HR, 0.55; 95% CI, 0.32–0.95). Additionally, the group who underwent RRM was more likely to undergo bilateral salpingo-oophorectomy and systemic chemotherapy, which may influence the significance of these survival findings.
A retrospective study of 390 women with early-stage breast cancer who were from families with a known BRCA1/BRCA2 pathogenic variant found a significant improvement in survival for women who underwent RRM when compared with those who chose unilateral mastectomy.[67] Patients were followed for a median of 14.3 years (range, 0.1–20.0 y). A multivariate analysis controlling for age at diagnosis, year of diagnosis, treatment, and other prognostic factors found that CRRM was associated with a 48% reduction in death from breast cancer. This was a relatively small study, and although the authors adjusted for multiple factors, residual confounding factors may have influenced the results.
All of these studies are limited by the biases introduced in relatively small, retrospective studies among very select populations. There is often limited data on potential confounding variables such as socioeconomic status, comorbidities, and access to care. It has been suggested that women who elect to undergo RRM are healthier by virtue of being able to tolerate more extensive surgery. This theory is supported by one study that used Surveillance, Epidemiology, and End Results (SEER) Program data to examine the association between CRRM and outcomes among women with unilateral breast cancer stages I through III. Results showed a reduction in all-cause mortality and breast cancer–specific mortality, and also in noncancer event mortality, a finding that would not be expected to be related to CRRM.[72]
Level of evidence: 3ai
Nipple-sparing mastectomy
The option of nipple-sparing mastectomy (NSM) in carriers of BRCA pathogenic variants undergoing risk-reducing procedures has been controversial because of concerns about increased breast tissue left behind at surgery to keep the NAC viable. The ability to leave behind minimal residual tissue, however, may be related to experience and technique. In a retrospective review of NSM performed in carriers of BRCA pathogenic variants at two hospitals between 2007 and 2014, NSM was performed on 397 breasts in 201 carriers of BRCA pathogenic variants.[73] This study included both unaffected and affected women. Incidental cancers were found in 4 of 150 RRM patients (2.7%) and 2 of 51 cancer patients (3.9%). With a mean follow-up of 32.6 months (range, 1.0–76.0 months), there were four subsequent cancer events that included two patients with axillary recurrences, one patient with a local and distant recurrence 11 months after her original NSM, and one patient who developed a new cancer in the inferior portion of her breast, with no recurrences at the NAC. A study of 177 NSMs performed in 89 carriers of BRCA pathogenic variants between 2005 and 2013 reported similar, excellent local control rates. Sixty-three patients had risk-reducing NSM (median follow-up, 26 months; range, 11–42 months), and 26 patients had NSM and a diagnosis of breast cancer (median follow-up, 28 months; range, 15–43 months). Five patients required further nipple excision. There were no local recurrences or newly diagnosed breast cancers.[74] Although the median follow-up time was short (34 months), NSM appeared to be a safe alternative to traditional mastectomy in appropriately selected patients with BRCA pathogenic variants.[75]
Level of evidence: 3aii
Histopathology of RRM specimens
Studies describing histopathologic findings in RRM specimens from women with BRCA1 or BRCA2 pathogenic variants have been somewhat inconsistent. In two series, proliferative lesions associated with an increased risk of breast cancer (lobular carcinoma in situ, atypical lobular hyperplasia, atypical ductal hyperplasia, and DCIS) were noted in 37% to 46% of women with pathogenic variants who underwent either unilateral or bilateral RRM.[76,77,78] In these series, 13% to 15% of patients were found to have previously unsuspected DCIS in the prophylactically removed breast. Among 47 cases of RRM or contralateral mastectomies performed in known carriers of BRCA1 or BRCA2 pathogenic variants from Australia, three (6%) cancers were detected at surgery.[79] In general, histopathologic findings in RRM specimens do not impact management.
Utilization
Individual psychological factors play an important role in decision-making about RRM by unaffected women and CRRM in women with unilateral breast cancer. (Refer to the Psychosocial Aspects of Cancer Risk Management for Hereditary Breast and Ovarian Cancer section in the Psychosocial Issues in BRCA1/BRCA2 section of this summary for information about uptake of RRM in BRCA carriers and the Psychosocial Outcome Studies section for information about psychosocial outcomes of RRM.)
Conclusion
In summary:
Risk-reducing salpingo-oophorectomy (RRSO)
In the general population, removal of both ovaries has been associated with a reduction in breast cancer risk of up to 75%, depending on parity, weight, and age at time of artificial menopause. (Refer to the PDQ summary on Breast Cancer Prevention for more information.) A Mayo Clinic study of 680 women at various levels of familial risk found that in women younger than 60 years who had bilateral oophorectomy, the likelihood of breast cancers developing was reduced for all risk groups.[80] Ovarian ablation, however, is associated with side effects such as hot flashes, impaired sleep habits, vaginal dryness, dyspareunia, and increased risk of osteoporosis and heart disease. A variety of strategies may be necessary to counteract the adverse effects of ovarian ablation.
The evidence for the effect of RRSO on breast cancer has evolved. Early small studies suggested a protective benefit. Initial retrospective studies supported breast cancer and ovarian cancer risk reduction after RRSO in BRCA pathogenic variant–positive women.[81] In support of early small studies,[82,83] a retrospective study of 551 women with disease-associated BRCA1 or BRCA2 variants found a significant reduction in risk of breast cancer (HR, 0.47; 95% CI, 0.29–0.77) and ovarian cancer (HR, 0.04; 95% CI, 0.01–0.16) after RRSO.[81] A prospective, single-institution study of 170 women with BRCA1 or BRCA2 pathogenic variants showed a similar trend. With RRSO, the HR was 0.15 (95% CI, 0.02–1.31) for ovarian, fallopian tube, or primary peritoneal cancer, and 0.32 (95% CI, 0.08–1.2) for breast cancer; the HR for either cancer was 0.25 (95% CI, 0.08–0.74).[84] A prospective, multicenter study of 1,079 women followed up for a median of 30 to 35 months found that while RRSO was associated with reductions in breast cancer risk in both carriers of BRCA1 and BRCA2 pathogenic variants, the risk reduction was more pronounced in BRCA2 carriers (HR, 0.28; 95% CI, 0.08–0.92).[6] A meta-analysis of all reports of RRSO and breast and ovarian/fallopian tube cancers in carriers of BRCA1/BRCA2 pathogenic variants confirmed that RRSO was associated with a significant reduction in breast cancer risk (overall: HR, 0.49; 95% CI, 0.37–0.65; BRCA1: HR, 0.47; 95% CI, 0.35–0.64; BRCA2: HR, 0.47; 95% CI, 0.26–0.84).[85] However, a cohort study of 822 carriers of BRCA1/BRCA2 pathogenic variants conducted in the Netherlands, where carrier screening is performed nationwide, did not observe a reduced risk of breast cancer after RRSO (HR, 1.09; 95% CI, 0.67–1.77).[86] The authors argued that the previous findings were driven by methodological issues including cancer-induced testing bias and immortal person time; the researchers empirically evaluated this by using their own cohort and applying the same assumptions about counting person time from previous studies.[