BRCA1, TP53, and CHEK2 germline mutations in uterine serous carcinoma




Uterine serous carcinoma (USC) is not recognized as part of any defined hereditary cancer syndrome, and its association with hereditary breast and ovarian carcinoma and Lynch syndrome are uncertain.


Using targeted capture and massively parallel genomic sequencing, 151 subjects with USC were assessed for germline mutations in 30 tumor suppressor genes, including BRCA1 (breast cancer 1, early onset), BRCA2, the DNA mismatch repair genes (MLH1 [mutL homolog 1], MSH2 [mutS homolog 2], MSH6, PMS2 [postmeiotic segregation increased 2]), TP53 (tumor protein p53), and 10 other genes in the Fanconi anemia–BRCA pathway. Ten cases with < 10% serous histology were also assessed.


Seven subjects (4.6%) carried germline loss-of-function mutations: 3 subjects (2.0%) with mutations in BRCA1, 2 subjects (1.3%) with mutations in TP53, and 2 subjects (1.3%) with mutations in CHEK2 (checkpoint kinase 2). One subject with < 10% serous histology had an MSH6 mutation. Subjects with MSH6 and TP53 mutations had neither personal nor family histories suggestive of Lynch or Li-Fraumeni syndromes. Of the 22 women with USC and a personal history of breast carcinoma, the frequency of BRCA1 mutations was 9%, compared to 0.9% in 119 women with no such history.


Approximately 5% of women with USC have germline mutations in 3 different tumor suppressor genes: BRCA1, CHEK2, and TP53. Mutations in DNA mismatch repair genes that cause Lynch syndrome are rare in USC. The germline BRCA1 mutation rate in USC subjects of 2% is higher than expected in a nonfounder population, suggesting that USC is associated with hereditary breast and ovarian carcinoma in a small proportion of cases. Women with USC and breast cancer should be offered genetic testing for BRCA1 and BRCA2 mutations. Cancer 2013. © 2012 American Cancer Society.


Targeted capture and massively parallel sequencing allows for highly reliable mutation analysis of multiple genes of interest simultaneously. We have applied this technique to develop a test we refer to as BROCA, which sequences 30 tumor suppressor genes, including BRCA1 (breast cancer 1, early onset), BRCA2, the DNA mismatch repair genes (MLH1 [mutL homolog 1], MSH2 [mutS homolog 2], MSH6, PMS2 [postmeiotic segregation increased 2]), TP53 (tumor protein p53), and 10 genes in the Fanconi anemia (FA)-BRCA pathway. Using BROCA, we evaluated germline DNA from 360 women with ovarian, fallopian tube, and peritoneal carcinoma: 23% of subjects had germline loss-of-function mutations. Mutations were identified in 12 genes, including every gene in the FA-BRCA pathway tested except ATM (ataxia telangiectasia mutated), suggesting that this pathway is widely implicated in hereditary ovarian carcinoma.1 The potential involvement of the FA-BRCA pathway has not been comprehensively assessed in endometrial carcinoma.

Although ovarian carcinoma has the highest mortality rate, endometrial carcinoma is the most common gynecologic malignancy in the United States. Although uterine serous carcinoma (USC) represents only 10% of all endometrial cancers, it accounts for up to 40% of endometrial cancer–related deaths.2 In contrast to the more common endometrioid histology, USC is more likely to present in advanced stage and carries a worse prognosis. USC mimics the most common histology of ovarian carcinoma and has a similar proclivity for intraperitoneal spread and high frequency of somatic TP53 mutations. Similar to ovarian carcinomas, endometrial carcinomas can be of mixed histologies, and even minor fractions of serous histology seem to confer a more aggressive phenotype in endometrial carcinomas.3

Currently, USC is not recognized as a feature of any hereditary cancer syndrome, but many have speculated a possible association with hereditary breast and ovarian cancer.4-8 Geisler et al found a higher incidence of subsequent breast carcinoma in women with USC compared to endometrioid endometrial carcinoma.9 Studies investigating the association between USC and germline BRCA1 and BRCA2 (BRCA1/2) mutations have revealed conflicting results, included small numbers of subjects, and did not differentiate between USC with purely serous features and those with mixed histologies.5-8 Therefore, it is not certain whether USC is part of the cancer spectrum for BRCA1 and BRCA2 mutation carriers.

It is also uncertain whether USC is associated with Lynch syndrome caused by germline mutations in DNA mismatch repair genes (MSH2, MLH1, MSH6, PMS2), in which endometrial carcinoma is a major manifestation. Although the majority of Lynch-associated endometrial carcinoma has endometrioid histology, USC has been described for a small number of Lynch patients.10 Studies of subjects with unselected endometrial carcinoma estimate a germline mutation rate in the DNA mismatch repair genes of approximately 2%, but only included very small numbers of USC cases.11-13 Therefore, the frequency of Lynch syndrome among unselected USC is not known.

A better understanding of inherited mutations associated with USC might allow targeted prevention in at-risk women. We therefore sought to characterize loss-of-function germline mutations in 30 tumor suppressor genes, using BROCA in a large prospectively acquired cohort of patients with USC.


Study Subjects

Individuals eligible for study were those with endometrial carcinoma designated as serous histology (either purely serous or mixed) who underwent surgery at Washington University St. Louis, St. Louis, Missouri, or at the University of Washington, Seattle, Washington, and provided informed consent to participate in the respective institutional gynecologic oncology tissue banks as approved by human subjects divisions of the institutional review boards. All subjects with USC who had lymphocytes collected at or before surgery were included, regardless of age, family history, or other clinical features suggestive of genetic predisposition to cancer. Cases of mixed histology were included, defined by having at least 10% serous histology. Additional cases with any fraction of serous histology < 10% were also analyzed, but were not included in the final mutation rate, because they did not formally meet criteria for mixed USC. We elected to assess these cases, because at least 1 study suggests that the presence of low fractions of serous component (< 10%) worsens prognosis, implying a different biology than purely endometrioid histologies.3 All pathology specimens were reviewed by a gynecologic pathologist to confirm the diagnosis of USC. Information about ethnic backgrounds, clinical information, and personal/family histories was retrieved from medical records. Genomic DNA was extracted from lymphocytes in peripheral blood samples.

Library Construction, Hybridization, and Massively Parallel Sequencing

A total of 3 μg of germline DNA from each subject was sonicated to a peak of 200 base pairs (bp) on a Covaris E series instrument (Covaris, Woburn, Mass). Paired-end libraries were prepared in 96-well plate format using the SureSelectXT enrichment system on a Bravo liquid-handling instrument (Agilent, Santa Clara, Calif). Individual paired-end libraries (500 ng) were hybridized to a 1.1-megabase pair custom design of complementary RNA biotinylated oligonucleotides targeting 30 genomic regions (Table 1). Following capture, each library was amplified by polymerase chain reaction with primers containing a unique 6-bp index and quantified by High Sensitivity chip (Agilent, Santa Clara, Calif). Equimolar concentrations of 96 libraries were pooled to a final concentration of 11 pM, cluster amplified on a single lane of v3 flowcell, and sequenced with 2 × 101-bp paired-end reads and a 7-bp index read using SBS (sequencing by synthesis) v3 chemistry on a HiSeq2000 (Illumina, San Diego, Calif). Sample median depth of coverage was 302× and the percent of targeted bases at > 10× and > 50× depth was 99.4% and 95.7%, respectively.

Table 1. Genomic Regions Targeted for Capture
GeneRegions Targeted for Capturea (UCSC hg19)
  • a

    Repetitive regions removed by RepeatMasker.

MSH2 (+EPCAM)chr2:47595263-47715360
PTEN (+KILLIN)chr10:89618195-89733532

Mutation Analysis

The BROCA panel accurately identifies all classes of mutations, including single base substitutions, small insertions and deletions, and large gene rearrangements.14 Sequence alignment and variant calling were performed against the reference human genome (UCSC hg19), as described.1 Each variant was annotated with respect to gene location and predicted function in Human Genome Variation Society nomenclature. Deletions and duplications of exons were detected by a combination of depth of coverage and split-read analysis, as described.15 All frameshift, nonsense, splice site, and missense mutations predicted to be deleterious to protein function were validated by polymerase chain reaction amplification and Sanger sequencing. Missense mutations and in-frame deletions were classified as deleterious only if a specific functional assessment could be or previously had been carried out. For mutations in Lynch-associated genes, microsatellite instability and MLH1 promoter methylation was assessed, as described.13


Subjects were women with endometrial carcinoma designated as having serous histology (either purely serous or mixed). The series was unselected for family history or any clinical features suggestive of a genetic predisposition to cancer. A total of 151 subjects with pure serous or mixed histology were included in the study. Table 2 provides clinical and pathological characteristics of the study population. An additional 10 subjects with < 10% serous histology were also included in a separate analysis.

Table 2. Subject Characteristics
  • Abbreviation: UPSC, uterine papillary serous carcinoma.

  • a

    Based on International Federation of Gynecology and Obstetrics (FIGO) 2009 staging guidelines.

Median age at diagnosis68 y
 In = 61 (40%)
 IIn = 16 (11%)
 IIIn = 34 (23%)
 IVn = 38 (25%)
 Unstagedn = 2 (1%)
 100% UPSCn = 74 (49%)
 Mixedn = 77
  ≥50% serousn = 31 (21%)
  <50% serousn = 46 (30%)
 Caucasiann = 100 (66%)
 African Americann = 39 (26%)
 Asiann = 1 (1%)
 Othern = 3 (2%)
 Unknownn = 8 (5%)

Seven of 151 subjects (4.6%) had germline loss-of-function mutations in 3 genes, and no subject had more than 1 mutation. Deleterious mutations included 3 in BRCA1 (2%), 2 in CHEK2 (checkpoint kinase 2) (1.3%), and 2 in TP53 (1.3%; Table 3). In addition, of the 10 subjects with < 10% serous histology, 1 subject had a deleterious MSH6 mutation. Characteristics of USC cases with a mutation are detailed in Table 3. The mutations were similarly distributed between cases of pure and mixed histology, with 3 mutations in subjects with pure USC and 4 in mixed USC. The median age of cancer diagnosis in mutation carriers was 65 years, which was not significantly different than the median age of 68 years for all USC subjects.

Table 3. Characteristics of Mutation Carriers
MutationMutation TypeAge at Diagnosis (y)EthnicityHistologyStagePersonal History of Other CancersFamily History of Cancer
  • Abbreviation: USC, uterine serous carcinoma.

  • a

    Previously demonstrated to be damaging.16

  • b

    Previously demonstrated to be damaging.18

  • c

    previously demonstrated to be damaging.20

BRCA1 p.E1535XNonsense38African AmericanMixed (endometrioid, serous)IIIABreast cancer (age 37)Two paternal aunts with breast cancer (ages 45, 53), maternal half-brother with leukemia (age 16), maternal grandfather with lung cancer (age 67)
BRCA1 c.2594delCFrameshift67Caucasian100% USCIASynchronous ovarian cancer (age 67), breast cancer (age 50)Sister with breast cancer, another sister with ovarian cancer
BRCA1 c.713-2A>CSplice sitea66Caucasian100% USCIBEsophageal cancer (age 74)Daughter with cervical cancer
TP53 p.R290HMissenseb48CaucasianMixed (endometrioid, serous)IIIC1NoneFather with lymphoma (age 39), grandmother with multiple myeloma
TP53 p.R273HMissenseb77Caucasian100% USCIVAUnknownUnknown
CHEK2 c.1100delCFrameshift79CaucasianMixed (endometrioid, serous)IVAColon cancerNone
CHEK2 p.I157TMissensec64CaucasianMixed (clear cell, serous)IANoneSon with ependymoma, father with prostate cancer

The frequency of BRCA1 germline loss-of-function mutations in subjects with USC was 2% (95% confidence interval [CI] = 0.7%-5.7%). All 3 BRCA1 mutations identified have been previously reported,16 but none were Ashkenazi founder mutations (Table 3). Two of the 3 subjects with deleterious BRCA1 mutations had personal histories of breast carcinoma and both had significant family histories of breast and/or ovarian carcinoma. Of 134 USC subjects with personal and family history data available, 22 subjects (16.4%) had a personal history of breast carcinoma, and 40 subjects (29.9%) had family histories of breast and/or ovarian carcinoma. Of the 22 women with USC and a personal history of breast carcinoma, the frequency of BRCA1 mutations was 9% (95% CI = 3%-28%), compared to 0.9% in 112 women with no such history (P = .07; Table 4). Only clearly deleterious germline mutations were included in our total mutation rate. In addition, 36 uncommon nonsynonymous missense variants (< 1% frequency rate reported on the Exome Variant Server [EVS],17 on March 30, 2012) were detected in BRCA1/2 (Table 5), 12 of which are reported as benign on the Breast Cancer Information Core database (BIC),16 on March 30, 2012. Only 1 of these missense variants, BRCA2 p.K912Q, had not been previously reported on either EVS or BIC. BRCA2 p.K912Q was predicted to be possibly damaging by PolyPhen2 (Polymorphism Phenotyping, v2), and benign by the SIFT (Sorting Intolerant From Tolerant) program.

Table 4. Studies of USC and BRCA1/2 Mutations
StudyNo. of USC SubjectsPopulationMutation AscertainmentBRCA1 MutationsBRCA2 MutationsPersonal History of Breast CancerBRCA1/2 in Breast Cancer SubjectsBRCA1/2 in Non– Breast Cancer Subjects
  • Abbreviations: BRCA, early onset breast cancer gene; USC, uterine serous carcinoma.

  • a

    Out of 134 subjects with personal history data available.

  • b

    Out of 112 subject without a history of breast carcinoma.

Biron- Shental et al522Israeli JewishAshkenazi founder mutations3 (13.6%)3 (13.6%)7 (31.8%)3 (42.9%)3 (20%)
Levine et al617Ashkenazi JewishAshkenazi founder mutations00Not availableNot availableNot available
Lavie et al759Ashkenazi JewishAshkenazi founder mutations7 (11.9%)1 (1.7%)15 (25.4%)3 (20%)5 (11.4%)
Goshen et al856CanadianSelected founder mutations and protein truncation test006 (10.7%)00
Current series151AmericanComprehensive analysis3 (2.0%)022 (16.4%)a2 (9.1%)1 (0.9%)b
Table 5. Uncommon Nonsynonymous Missense Mutations Detected in BRCA1 and BRCA2
GeneVariantaReported on BICEVS EA Allele FreqEVS AA Allele FreqPoly Phen2SIFT
  • Abbreviations: BIC, Breast Cancer Information Core database;16 BRCA, early onset breast cancer gene; EVS, Exome Variant Server;17 Freq, frequency.

  • a

    Using RefSeq NM_007294.3 for BRCA1 and NM_000059.3 for BRCA2

BRCA1p.T1561IUncertain significance0/702020/37180.9880
BRCA1p.R1347GUncertain significance42/69784/37340.2550.01
BRCA1p.P1238LUncertain significance1/70190/37360.2640.06
BRCA1p.N723DUncertain significance0/702026/37100.1640.1
BRCA1p.L512FUncertain significanceNot reportedNot reported10
BRCA1p.H476RUncertain significance0/70209/37290.970.07
BRCA1p.V1534MUncertain significance2/701814/37240.070.22
BRCA1p.R1347GUncertain significance42/69784/37340.2550.01
BRCA1p.K912QNot reportedNot reportedNot reported0.4920.45
BRCA1p.H476RUncertain significance0/70209/7290.970.07
BRCA1p.D214GUncertain significance1/70190/37360.0010.34
BRCA2p.R1190QUncertain significanceNot reportedNot reported01
BRCA2p.V1639IUncertain significanceNot reportedNot reported0.090.14
BRCA2p.K1690NUncertain significance2/69940/39640.9890.13
BRCA2p.S1733FUncertain significance1/700512/37180.1590.01
BRCA2p.E1879KUncertain significance2/70180/37380.0680.54
BRCA2p.R2034CUncertain significance35/69839/37290.5560.16
BRCA2p.Q2384KUncertain significance0/702022/371601
BRCA2p.T225AUncertain significanceNot reportedNot reported0.410.12
BRCA2p.H1561NUncertain significance0/702036/37020.6650.28
BRCA2p.R2034CUncertain significance35/69839/37290.5560.16
BRCA2p.V2138FUncertain significance0/700033/36990.2550.08
BRCA2p.K2729NUncertain significanceNot reportedNot reported10.07
BRCA2p.D3272EUncertain significanceNot reportedNot reported0.9810.13

Two subjects had deleterious TP53 mutations (1.3%). Both of the TP53 mutations identified in our study subjects have been previously reported in Li-Fraumeni syndrome (LFS) families.18 One mutation carrier had mixed USC and the other had pure USC. The R290H mutation carrier did not have a family history suggestive of LFS, although her family history was notable for a father with lymphoma at age 39 and a grandmother with multiple myeloma. The R290H mutation has been reported in multiple LFS families, although it has a weak mutant phenotype when assessed in functional assays.19 No family history data was available for the subject with the R273H mutation.

Two subjects had mutations in CHEK2 (1.3%), both cases with mixed USC. One mutation was the CHEK2 c.1100delC founder mutation, and the other mutation, p.I157T, is a missense mutation that has been demonstrated through functional assays to be damaging.20

A single DNA mismatch repair gene mutation was identified in a subject with an endometrioid tumor with 1% serous histology. The in-frame MSH6 deletion was classified as deleterious given the tumor microsatellite instability-high (MSI-H) phenotype (defined by having instability in 2 or more of the following markers: BAT25, BAT26, D5S346, D2S123, and D17S250) and lack of MLH1 promoter methylation. The mutation carrier did not have a family history of cancer and did not meet Amsterdam criteria for Lynch syndrome.


To our knowledge, this is the first study to comprehensively evaluate an unselected USC population for germline mutations in 30 tumor suppressor genes, including BRCA1, BRCA2, the DNA mismatch repair genes, TP53, and genes in the FA-BRCA pathway. USC represents only 10% of all endometrial carcinoma cases2; thus, 151 USC cases represent a large series of this relatively rare cancer subtype. Subjects were not selected for age or family history, in an effort to obtain an unbiased mutation rate in the general population. However, both institutions represented in this study are tertiary referral centers, and it is possible that the mutation rate reported here may be influenced by a referral bias for patients with personal and family histories of cancer.

We and others have previously identified germline mutations in 11 different genes in the FA-BRCA pathway in subjects with ovarian carcinoma.1, 21-23 In contrast, in USC, only BRCA1 and CHEK2 have significant rates of germline mutations from the FA-BRCA pathway, suggesting that the FA-BRCA pathway is less widely implicated in USC compared to ovarian carcinoma. Thus, although USC and serous ovarian carcinoma share many pathological and clinical features, the molecular epidemiology of USC does differ from that of ovarian carcinoma.

Our data reveal that approximately 2% of unselected women with USC have a germline BRCA1 mutation. Although the overall mutation rate is low, it is higher than would be expected in the general population, in which BRCA1 mutation frequency has been estimated at approximately 0.06%.24 Two of the 3 mutation carriers had personal histories of breast carcinoma and family histories of breast/ovarian carcinoma. The rate of BRCA1/2 mutations in women with USC and prior breast carcinoma in this series was 9%, suggesting that women with USC and prior breast carcinoma should be offered genetic testing for BRCA1 and BRCA2 mutations. These data support the conclusion that USC is associated with hereditary breast and ovarian cancer, albeit in a small proportion of cases. Although our findings suggest an elevated relative risk of USC in BRCA1 mutation carriers, the absolute risk is likely to be small, given the rarity of USC in the general population. Whether these data should influence the role of hysterectomy in combination with risk-reducing salpingo-oophorectomy for cancer prevention in BRCA1 mutation carriers requires further study.

Four smaller studies have explored the association of USC with germline BRCA1 or BRCA2 mutations, with conflicting results (summarized in Table 4).5-8 In the only other non-Ashkenazi series, Goshen et al studied 56 Canadian women with USC and failed to detect any BRCA1 or BRCA2 mutations, but used noncomprehensive testing, which would have missed 2 of the 3 BRCA1 mutations identified in our study.8 Three previous studies were in Jewish populations and tested only for the 3 Ashkenazi founder mutations. Two of the 3 Jewish studies found an increased mutation rate in BRCA1 and BRCA2 (14%-27%, compared to an expected 2.3% founder mutation rate),5, 7 whereas the smallest series of 17 subjects found no mutations.6 Interestingly, the largest Ashkenazi series by Lavie et al found a significantly higher mutation rate in BRCA1 (11.9%) than in BRCA2 mutations (1.7%; Table 4).7 Although the number of mutations in our series is small, all occurred in BRCA1. In combining data from all the series on USC, there were 13 BRCA1 mutations in 305 cases compared with 4 BRCA2 mutations (P = .046, Fisher's exact test). Therefore, it may be that USC is more commonly associated with mutations in BRCA1 than BRCA2. The association of BRCA1/2 mutations in endometrial carcinomas appears limited to USC, because the 1 large series of 199 mostly endometrioid endometrial carcinoma in Ashkenazi women did not find a mutation rate different than the expected population frequency.6

The 1.3% rate of TP53 mutations seen in this study was unexpected, because uterine carcinoma is not a known manifestation of LFS. However, the neoplastic spectrum in TP53 mutation carriers is much wider than the specific cancers used to define LFS and/or Li-Fraumeni–like syndrome (LFL), with roughly 20% to 30% of cancers in TP53 mutation–positive families falling outside the classic spectrum.26-28 Both of the specific TP53 mutations in our subjects have been previously reported in LFS families.18 Only 1 carrier of the TP53 mutation had available family history information, and she did not fit criteria for either LFS or LFL. This finding is similar to our previous observations of deleterious TP53 mutations in women with ovarian carcinoma.1 Of several studies exploring the wide spectrum of cancers in LFS/LFL families, 1 reported endometrial carcinoma as a rare finding.28 In some families with TP53 mutations, there may be a wider range of cancer types, lower cancer risk, and later onset of tumors, suggesting that deleterious TP53 mutations may be more frequent in the population than previously assumed. Such families with reduced penetrance and a later average age of onset might go unnoticed due to the generally restrictive guidelines used for TP53 mutation testing. As comprehensive genetic testing is increasingly offered to individuals not selected for established syndromic phenotypes or family history, a wider range of expressivity associated with germline mutations of cancer susceptibility genes may become increasingly apparent.

Our data suggest that the rate of Lynch syndrome in subjects with USC is lower than that of endometrioid carcinomas (1.8%-2.1%).11, 12 No USC subjects included in our primary analysis were found to have germline mutations in any of the mismatch repair genes. Of the additional 10 subjects with < 10% serous histology who were analyzed, 1 had a germline mutation in MSH6. Interestingly, this case with the MSH6 mutation contained only 1% serous and 99% endometrioid components. Although USC has been previously reported in subjects with Lynch syndrome, no prior studies specifically evaluated for germline mutations in the mismatch repair genes in a series of USC cases. Our findings are concordant with a recent study demonstrating no defects in DNA mismatch repair in the neoplastic DNA from 29 pure or mixed USC cases.25 In that study, pancreatic carcinomas were overrepresented in relatives of individuals with USC, but this excess was not explained by DNA mismatch repair defects. We speculate that BRCA1 or BRCA2 mutations could serve as the common link between USC and pancreatic cancer in these families.

Overall, 5% of USC cases had germline loss-of-function mutations in 3 of 30 tumor suppressor genes evaluated. Mutations in DNA mismatch repair genes that cause Lynch syndrome are rare in USC, similar to the rate for ovarian carcinoma. The germline BRCA1 mutation rate in USC subjects of 2% is higher than expected in a nonfounder population, suggesting that USC is associated with hereditary breast and ovarian cancer in a small proportion of cases. In contrast, USC is less likely to be associated with mutations in other FA-BRCA genes compared with ovarian carcinoma.


Supported by National Institutes of Health (NIH) National Cancer Institute (NCI) 1R01CA131965 (to E.M.S.), Department of Defense Ovarian Cancer Research Program OC093285 (to T.W.), NIH R01CA157744 (to T.W., M.-C.K.), and NCI P50 Specialized Program of Research Excellence (SPORE) PJG award CA134254 (to P.G.).


The authors made no disclosure.