Prevalence of BRCA1 and BRCA2 mutations in Ashkenazi Jewish families with breast and pancreatic cancer
Germline mutations in the BRCA2 cancer susceptibility gene are associated with an increased risk of pancreatic cancer (PC). Breast-pancreas cancer families with BRCA1 mutations have also been observed. The influence of a family history (FH) of PC on BRCA mutation prevalence in patients with breast cancer (BC) is unknown.
A clinical database review (2000-2009) identified 211 Ashkenazi Jewish (AJ) BC probands who 1) underwent BRCA1/2 mutation analysis by full gene sequencing or directed testing for Ashkenazi founder mutations (BRCA1: 185delAG and 5382insC; BRCA2: 6174delT) and 2) had a FH of PC in a first-, second-, or third-degree relative. For each proband, the pretest probability of identifying a BRCA1/2 mutation was estimated using the Myriad II model. The observed-to-expected (O:E) mutation prevalence was calculated for the entire group.
Of the 211 AJ BC probands with a FH of PC, 30 (14.2%) harbored a BRCA mutation. Fourteen (47%) of the mutations were in BRCA1 and 16 (53%) were in BRCA2. Patients diagnosed with BC at age ≤ 50 years were found to have a higher BRCA1/2 mutation prevalence than probands with BC who were diagnosed at age > 50 years (21.1% vs 6.9%; P = .003). In patients with a first-, second-, or third-degree relative with PC, mutation prevalences were 15.4%, 15.3%, and 8.6%, respectively (P = .58). In the overall group, the observed BRCA1/2 mutation prevalence was 14.2% versus an expected prevalence of 11.8% (O:E ratio, 1.21; P = .15).
BRCA1 and BRCA2 mutations are observed with nearly equal distribution in AJ breast-pancreas cancer families, suggesting that both genes are associated with PC risk. In this population, a FH of PC was found to have a limited effect on mutation prevalence. Cancer 2011;. © 2011 American Cancer Society.
Germline mutations in the BRCA1 and BRCA2 tumor suppressor genes are associated with an increased risk of breast and ovarian cancer.1 Given lifetime risks of 50% to 85% for breast and up to 45% for ovarian cancer, intensive cancer surveillance and/or risk-reducing prophylactic surgeries have become an important component of the management of BRCA1/2 mutation carriers.2-7 Phenotypic features specific to BRCA1 versus BRCA2 carriers have also been recognized. For example, BRCA1 carriers have a higher cumulative risk of ovarian cancer and a propensity toward the development of “triple-negative” breast cancers that lack estrogen and progesterone receptors as well as human epidermal growth factor receptor 2 (HER2)/neu overexpression.8-10 In addition to breast and ovarian cancer, an increased incidence of other malignancies including prostate cancer, male breast cancer, pancreatic cancer, and possibly melanoma have also been observed in families with BRCA mutations, with an apparent stronger association with BRCA2 mutations.11-13 However, the cumulative risks, BRCA gene-specific clinical features, and optimal screening and/or prevention strategies for these cancers are not as well characterized.
Pancreatic cancer is a component of the hereditary breast-ovarian cancer syndrome. The Breast Cancer Linkage Consortium study observed a 3.5-fold increased incidence of pancreatic cancer in families carrying a BRCA2 mutation compared with the general population.14 Inherited mutations in the BRCA2 gene were identified in 17% (5 of 29) of pancreatic cancer kindreds in whom ≥ 3 members were affected with pancreatic cancer.15 In 26 European families with at least 2 first-degree relatives (FDRs) with pancreatic cancer, 12% were found to harbor a BRCA2 mutation.16 A more recent study identified 10 BRCA2 mutation carriers in 180 familial pancreatic cancer kindreds, suggesting a BRCA2 mutation prevalence of 6%.17 To our knowledge, the association between BRCA1 mutations and pancreatic cancer is not as well defined. The relative risk of pancreatic cancer in BRCA1 mutation carriers is estimated to be 2-fold higher than in the general population.12 A recent study noted an absence of BRCA1 mutations in 66 families with a clustering of pancreatic cancer.18
Although relatively rare in the general population, in certain ethnic groups, such as Ashkenazi Jews, mutations in the BRCA1/2 cancer susceptibility genes are more prevalent.19-21 The 3 founder mutations in BRCA1 (185delAG and 5382insC) and BRCA2 (6174delT) are thought to account for the vast majority (approximately 96%) of BRCA1/2 mutations in the Ashkenazim.22 In a small analysis of 23 Ashkenazi Jewish (AJ) families with either a personal or a family history of pancreatic cancer, 8 BRCA mutations were identified, with the BRCA1 185delAG and the BRCA2 6174delT founder mutations being equally represented.23 In patients of Ashkenazi ancestry (unselected for family history) with resected pancreatic cancer, 5.5% were BRCA1/2 mutation positive with 2 mutations in BRCA1 and 6 in the BRCA2 gene.24 These studies suggest a possible role for BRCA1 mutations in a predisposition toward pancreatic cancer, at least in the Ashkenazim. To date, the majority of analyses of the association between BRCA1 mutations and pancreatic cancer predisposition have been limited by sample size and have focused on familial pancreatic cancer kindreds irrespective of breast and ovarian cancer history. In such kindreds, germline mutations in other known (or unknown) pancreatic cancer susceptibility genes, such as p16/CDKN2A, PRSS1, STK11/LKB1, and PALB2, may be responsible for the clustering of pancreatic cancers.25-28
It is clear that mutations in BRCA1 and BRCA2 can be identified in familial breast-pancreas cancer families, and BRCA2 mutation carriers at least are at an increased risk of pancreas cancer. What is less clear is the degree to which a family history of pancreatic cancer influences the likelihood of detecting a BRCA mutation in a woman with breast cancer. Some prediction models do incorporate a family history of pancreas cancer when calculating the chance of detecting a mutation. However, to our knowledge, validation studies of these models have not addressed the specific incremental contribution of a pancreatic cancer history to the risk assessment. In an effort to better elucidate the contribution of pancreatic cancer to hereditary cancer risk assessment, we performed the current study to determine the prevalence of BRCA1 and BRCA2 mutations in AJ breast cancer probands with a family history of pancreatic cancer and to determine whether mutations were more prevalent than expected.
MATERIALS AND METHODS
Through a review of the Memorial Sloan-Kettering Cancer Center (MSKCC) Clinical Genetics Service pedigree and clinical databases, we identified all AJ probands affected with breast cancer who 1) underwent BRCA1 and BRCA2 mutation analysis between 2000 and December 2009 and 2) had a first-, second-, or third-degree family history of pancreatic cancer. All patients received clinical pretest genetic counseling that included construction of a detailed 3-generation pedigree by a genetic counselor and subsequent genetic testing. For each family, clinical data including type, number, and ages at cancer diagnosis was extracted from the pedigree. Breast cancer probands underwent BRCA mutation analysis either by full gene sequencing or by directed testing for the 3 AJ founder mutations (BRCA1: 185delAG and 5382insC; BRCA2: 6174delT). Full gene sequencing analysis of the coding regions and flanking intronic regions was performed by Myriad Genetics Laboratories (Salt Lake City, Utah) as previously described.29 Testing for the AJ BRCA1/2 founder mutations was performed in the Diagnostic Molecular Genetics Laboratory at MSKCC using previously published methods.30, 31 This study was approved by the MSKCC institutional review board.
The pretest probability of identifying a deleterious BRCA mutation in the breast cancer proband was estimated using the Myriad II model, which incorporates family breast and ovarian cancer history, age at cancer diagnosis, and ethnicity. This model is based on mutation prevalence tables originally published by Frank et al.32 Pedigrees in which the family history of pancreatic cancer was segregating through the lineage with the lower Myriad II model score (compared with the opposite lineage) were excluded from analysis. After the exclusion of 4 such families, a total of 211 families were included in the final analysis. Associations between age and BRCA mutation status and between pancreatic cancer family history and BRCA status were assessed using the chi-square test. The observed mutation prevalence was compared with that predicted by the Myriad II model using a single-sample Z test. This test assumes that the predicted values are constant. Receiver operating curves (ROC) and the area under the ROC (AUROC) were calculated to determine how well the Myriad II score predicted actual BRCA1/2 mutation status.
Of the 211 AJ breast cancer probands with a family history of pancreatic cancer, 31% had a FDR, 53% had a second-degree relative (SDR), and 16% had a third-degree relative (TDR) with pancreatic cancer. Twenty-six probands had > 1 relative with a pancreatic cancer diagnosis within the same lineage of the family. In addition to the proband's personal history of breast cancer, 70 (33%) families had ≥ 2 additional breast cancers within the same lineage as the family history of pancreatic cancer, but only 31 (14.7%) probands had a relative with a diagnosis of ovarian cancer within the same lineage. Greater than one-half (52%) of the probands were diagnosed with breast cancer at age ≤ 50 years, whereas the mean age at the time of the pancreatic cancer diagnosis in relatives was 67 years (based on pedigree review). Additional proband and family characteristics are shown in Table 1.
Table 1. Proband and Family Characteristics
|Proband|| || |
| Mean age of proband at BC diagnosis (range), y||50.5||(26-76)|
| No. of probands diagnosed with BC aged ≤50 y||109||52|
| No. of probands diagnosed with BC aged >50 y||102||48|
| No. of probands with >1 primary BC||33||16|
| Male breast cancer||2||0.9|
|Family history|| || |
| No. of probands with FDR with PC||65||31|
| No. of probands with SDR with PC||111||53|
| No. of probands with TDR with PC||35||16|
| Mean age of FDR, SDR, or TDR with PC (range), y||66.6||30-97|
| No. of probands with >1 PC in same side of family||26||12|
| No. of probands with ≥2 additional BCs in same side of family||70||33|
| No. of families with ≥1 OC in same side of family||31||14.7|
BRCA Mutation Prevalence
A total of 30 (14.2%) BRCA mutations (14 BRCA1 mutations and 16 BRCA2 mutations) were identified in the 211 AJ probands with a personal history of breast cancer and a family history of pancreatic cancer (Table 2). BRCA1/2 mutation prevalence was higher in patients diagnosed with breast cancer at age ≤ 50 years compared with probands diagnosed with breast cancer at age > 50 years (21.1% vs 6.9%; P = .003) and was comprised of 9 BRCA1 and 14 BRCA2 mutations. Of all BRCA1 mutation carriers, 13 patients had the 185delAG founder mutation whereas 1 patient harbored the 5382insC mutation. With the exception of 1 nonfounder mutation, all mutations in BRCA2 were because of the 6174delT mutation. The prevalence of BRCA1/2 mutations was similar at 15% in probands with a FDR or SDR with pancreatic cancer and was 8.6% in probands with a TDR with pancreatic cancer (FDR, 15.4% vs SDR, 15.3% vs TDR, 8.6%; P = .58). Mutation prevalence according to age at breast cancer diagnosis and extent of family history of pancreatic cancer is detailed in Table 3.
Table 2. Prevalence of BRCA1 and BRCA2 Mutations in Ashkenazi Jewish BC Probands With a Family History of Pancreatic Cancer
|Overall||211||181 (85.8%)||30 (14.2%)||14 (46.7%)||16 (53.3%)|
|185delAG: 13||6174delT: 15|
|5382insC: 1||3036del4: 1a|
|BC diagnosed age ≤50 y||109||86 (78.9%)||23 (21.1%)||9 (39.1%)||14 (60.9%)|
|185delAG: 8||6174delT: 13|
|5382insC: 1||3036del4: 1a|
|BC diagnosed age >50 y||102||95 (93.1%)||7 (6.9%)||5 (71.4%)||2 (28.6%)|
|185delAG: 5||6174delT: 2|
Table 3. BRCA Mutation Prevalence According to Family History of Pancreatic Cancer
| First-degree relative||65||55 (84.6%)||10 (15.4%)|
|BRCA1: 4 BRCA2: 6|
| Second-degree relative||111||94 (84.7%)||17 (15.3%)|
| Third-degree relative||35||32 (91.4%)||3 (8.6%)|
|BC in proband diagnosed age ≤50 y|
| First-degree relative||27||22 (81.5%)||5 (18.5%)|
| Second-degree relative||66||51 (77.3%)||15 (22.7%)|
| Third-degree relative||16||13 (81.3%)||3 (18.7%)|
|BC in proband diagnosed age >50 y|
| First-degree relative||38||33 (86.8%)||5 (13.2%)|
| Second-degree relative||45||43 (95.6%)||2 (4.4%)|
| Third-degree relative||19||19 (100%)||0 (0.0%)|
In the 30 BRCA1/2 mutation-positive breast cancer probands, the mean age at breast cancer diagnosis was 44 years (range, 22 years-65 years) and 27% had been diagnosed with > 1 primary breast cancer. The mean age at diagnosis of family members with pancreatic cancer was 67.4 years. Of the 30 BRCA1/2 mutation-positive families, 3 families (1 with a BRCA1 mutation and 2 with a BRCA2 mutation) had ≥ 2 pancreatic cancers segregating in the same side of the family; approximately 64% (21 of 33) of pancreatic cancers were diagnosed in women. Of the 21 female relatives with pancreatic cancer from mutation-positive kindreds, 24% also had a prior diagnosis of breast cancer, with 2 relatives being from BRCA1 mutation-positive and 3 from BRCA2 mutation-positive families. Two of the male relatives with pancreatic cancer had double primary tumors: one relative, from a BRCA2 mutation-positive family, had both breast and pancreatic cancer; the other, from a BRCA1 mutation-positive family, had both prostate and pancreatic cancer.
For each breast cancer proband, we used the Myriad II model to estimate the pretest probability of identifying a BRCA mutation. For the overall study population, the Myriad II model predicted a mutation prevalence of 11.8%; a BRCA1/2 mutation was identified in 14.2% (observed-to-expected [O:E] ratio, 1.21; P = .15) (Table 4). The increased O:E mutation prevalence ratio was upheld regardless of the age at breast cancer diagnosis in the proband, although the differences did not meet statistical significance in any subgroup. As expected, the predicted mutation prevalence in the BRCA-positive group was higher at 18.8% compared with the overall group. The predicted mutation prevalence in the BRCA1 mutation-positive and BRCA2 mutation-positive groups was nearly identical at 18.7% and19.0%, respectively. We constructed ROCs to determine how well the Myriad II model predicted actual BRCA1/2 mutation status. In the overall group, the AUROC was 0.74 (95% confidence interval [95% CI], 0.64-0.82), which is comparable to Myriad II AUROCs of 0.75 and 0.76 reported in prior studies of unselected individuals presenting for BRCA risk assessment in familial cancer clinics.33 In the breast cancer group diagnosed at age > 50 years, the AUROC was similar at 0.73 (95% CI, 0.53-0.97); however, in the group of probands with a diagnosis of breast cancer at age ≤ 50 years, the AUROC decreased to 0.63 (95% CI, 0.50-0.78).
Table 4. Observed Versus Expected BRCA Mutation Prevalence
|BC diagnosed age ≤50 y||109||21.1%||17.3%||1.22||.089|
|BC diagnosed age >50 y||102||6.9%||5.8%||1.18||.30|
Several studies have shown that the third most common cancer associated with BRCA mutations is pancreatic cancer.34 However, the extent to which BRCA1 mutations, in addition to BRCA2 mutations, contribute to this risk is uncertain. The results of the current study suggest that the distribution of BRCA1 versus BRCA2 mutations is nearly equal (47% vs 53%) in AJ breast-pancreas cancer families harboring any mutation. The distribution is similar to unselected AJ populations in whom BRCA1 mutations (185delAG and 5382insC) are reported to account for 45% and BRCA2 6174delT mutations for 54% of all mutations.35 The current study results also support the findings of 2 small studies of BRCA-positive breast/ovarian cancer families with a history of pancreatic cancer. In a study from the United States of 19 BRCA-positive breast/ovarian-pancreas cancer families, 15 mutations were localized to the BRCA1 gene and 4 were localized to the BRCA2 gene.36 BRCA1 and BRCA2 mutations were equally represented in an Israeli study of 23 similar families.23 Conversely, in a recent study of 66 patients with familial pancreatic cancer, BRCA1 mutations were not identified despite the finding that greater than one-half of pancreatic cancer probands reported a family history of breast and/or ovarian cancer.18 Notably, of these 34 breast/ovarian-pancreas cancer families, only 1 family was of AJ descent and this difference in population structure may account for the discordant results. Similar to previous studies of familial pancreatic cancer, of our 26 kindreds who had > 1 family member with pancreatic cancer segregating through the same lineage, 3 (11.5%) were found to harbor a BRCA mutation.15-17 In the other 23 BRCA mutation-negative families, known (or unknown) pancreatic cancer susceptibility genes may account for the clustering of pancreatic cancers. For example, mutations in PALB2 have been identified in a small percentage of families with apparent predispositions toward pancreatic cancer, although the prevalence of PALB2 mutations in breast-pancreas cancer families appears to be low.37, 38
Currently, the Myriad II model and the extended Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation (BOADICEA) model have the highest specificity and sensitivity for BRCA1 and BRCA2 mutation prediction in AJ families.33 However, these models may underestimate mutation prevalence in families with breast and pancreatic cancer. In an assessment of carrier prediction models based on 472 at-risk families, Barcenas et al noted that in a subset of 26 families with a FDR or SDR with pancreatic cancer, 39% tested positive for a BRCA mutation, but the mean probabilities estimated by the BRCAPRO and BOADICEA models were 24.4% and 13%, respectively.33 Applying the Myriad II model to an AJ cohort, we also found that the observed BRCA1/2 mutation prevalence was higher than predicted in breast-pancreas cancer families, although this did not reach statistical significance. Similarly, the constructed ROCs demonstrate a trend toward inferior performance of the Myriad II model in probands with early onset breast cancer with a family history of pancreatic cancer, although model performance was adequate in the overall population. These findings are consistent with the conclusion that the presence of pancreatic cancer in the family history should be taken into account when considering genetic testing of AJ women with breast cancer. Of all BRCA carrier prediction models, to our knowledge only the Manchester scoring system and the PENN II model (available at: http://www.afcri.upenn.edu/itacc/penn2/) consider a family history of pancreatic cancer as a variable in predicting the probability of a BRCA mutation.39-42 It is important to note that in both of these models, a family history of pancreatic cancer is considered to increase only the probability of a BRCA2, but not of a BRCA1, mutation. In addition, the Manchester scoring system does not include an algorithm for the inclusion of individuals of AJ ancestry and therefore is not applicable to our study population.39, 40 As such, we did not evaluate the calibration of these models in our ascertainment.
In the current study, the mean age at the time of pancreatic cancer diagnosis in BRCA mutation–positive families was 67 years, which is significantly older then the ages at which BRCA-associated breast and ovarian cancers typically occur. In previous generations, many BRCA mutation-positive women developed fatal early onset breast or ovarian cancer before the age at which pancreatic cancer typically presents. With improvements in the diagnosis, treatment, and prevention of BRCA-associated breast and ovarian cancers, pancreatic cancer may be unveiled in some families and thus the prevalence of pancreatic cancer in BRCA mutation-positive families may be shifting. In the current study, 24% of female relatives with pancreatic cancer (from mutation-positive kindreds) also had a prior diagnosis of breast cancer. Many BRCA mutation carriers, especially those with a family history of pancreatic cancer, are increasingly aware and attentive to their risk of pancreatic cancer. Given the continued poor prognosis of pancreatic cancer, early detection measures for high-risk individuals are needed and studies using different screening modalities are currently underway. In addition, knowledge of the BRCA mutation status of patients with pancreatic cancer has become increasingly important as new therapeutic interventions such as poly(ADP-ribose) polymerase (PARP) inhibitors have shown significant activity in patients with advanced BRCA-associated breast and ovarian cancers and hold potential for the treatment of BRCA-associated pancreatic cancer.
The current study is limited to patients of AJ descent from a single institution and may not be applicable to other populations. For example, it is possible that in ethnicities with a lower baseline probability of harboring a BRCA mutation, the presence of a family history of pancreatic cancer could more significantly impact the pretest probability of finding a BRCA mutation. In addition, because our study focused on breast cancer probands, for the individuals in the family with pancreatic cancer, we did not obtain routine histological confirmation of the pancreatic cancer and these family members were not directly genetically tested for BRCA mutations and may have been phenocopies, leaving the precise burden of pancreatic cancer due to BRCA mutations undefined. To offset this limitation, families were excluded if the pancreatic cancer was segregating in the opposite lineage from the breast and ovarian cancers (ie, the lineage with the higher mutation probability score). Furthermore, of the families transmitting mutations, 23% of the family members with pancreatic cancer had been diagnosed with an additional BRCA-associated malignancy, increasing the likelihood that these individuals were in fact carriers of the BRCA mutation segregating in their close relatives.
Although BRCA1 and BRCA2 mutations appear to be distributed nearly equally in AJ breast-pancreas cancer families with mutations, evaluation of breast-pancreas cancer families of various ancestry is needed before extrapolation of these results. Clinicians should recognize the importance of pancreatic cancer as a BRCA-associated cancer, and a family history of pancreatic cancer should be taken into account when considering referral for hereditary breast-ovarian cancer syndrome genetic risk assessment, especially in families of limited size or structure.
Supported in part by the Breast Cancer Research Foundation, the Robert and Kate Niehaus Clinical Cancer Genetics Research Initiative, the Lymphoma Foundation, and the Truettner Family Research Fund. Dr. Stadler is supported in part by an American Society of Clinical Oncology Cancer Foundation Career Development Award.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.