BRAF Mutation in Endometrial Carcinoma and Hyperplasia: Correlation with KRAS and p53 Mutations and Mismatch Repair Protein Expression

Endometrial carcinomas are now classified into two histologic subtypes: endometrioid type and nonendometrioid type (1); and the molecular genetic changes involved differ between the two (2–4). In endometrioid-type carcinomas, mutations of PTEN, KRAS, β-catenin, and p53 genes and microsatellite instability have been reported in ∼5% to 50% of cases. By contrast, in nonendometrioid-type carcinomas, as exemplified by serous papillary adenocarcinoma, the p53 gene is reportedly mutated in ∼90% of cases. However, the genetic changes in endometrial carcinoma have not fully been elucidated.

Raf proteins are cytoplasmic serine-threonine kinases and play central roles in the conserved Ras/Raf/mitogen-activated protein/extracellular signal-regulated kinase kinase/extracellular signal-regulated kinase pathway, acting to relay signals from activated Ras proteins via mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1/2 to p42/44 mitogen-activated protein kinase or extracellular signal-regulated kinase 1/2, the key effector of the pathway (5, 6). A recent report has shown the presence of an activating mutation in one of three Raf protein subtypes, B-Raf, in human melanomas and colon cancers, and that the mutation of BRAF transforms NIH 3T3 cells independent of KRAS gene mutation (7). In addition, mutual exclusion between KRAS and BRAF mutations has been reported in colon cancer (8). More recently, BRAF gene mutations have been frequently found in sporadic colorectal carcinomas deficient in mismatch repair (9). However, the occurrence of BRAF gene mutations, as well as their correlation with mutations in other genes and molecules, has not been elucidated in endometrial carcinoma. Therefore, in the present study, we examined mutations of the BRAF gene in DNA samples extracted from normal, hyperplastic, and malignant endometrial tissues using a microdissection technique. We also examined mutations of KRAS and p53 genes in the same tissue DNA to evaluate the correlation with mutations of BRAF. In addition, the occurrence of BRAF gene mutations was compared with the expression of hMSH2 and hMLH1 proteins, which are regarded as key factors in DNA mismatch repair (10, 11).

Case selection and DNA extraction. Ninety-seven patients with endometrial carcinoma visited Shinshu University Hospital between 1993 and 2003, and underwent hysterectomy, bilateral salpingo-oophorectomy, and lymph node dissection or biopsy. The age of the cancer patients ranged from 30 to 76 years. All cases involved Japanese women, and were sporadic without a familial history of endometrial cancer and/or colon cancer. No patients refused participation in the study. According to the Fédération Internationale des Gynaecologistes et Obstetristes (FIGO) classification (1988), 52 patients had stage I, 10 had stage II, 28 had stage III, and 7 had stage IV tumors. The histologic diagnosis was made by two pathologists in the Department of Laboratory Medicine of the hospital. Histologically, 78 had endometrioid adenocarcinoma and 19 had nonendometrioid carcinoma. Of the 78 endometrioid carcinomas, 36 were grade 1, 19 were grade 2, and 23 were grade 3. Of the 19 nonendometrioid carcinomas, 12 were serous papillary adenocarcinomas, 6 were clear cell carcinomas, and 1 was a squamous cell carcinoma. Twenty cases of normal endometrium (proliferative phase, 10 cases; secretory phase, 10 cases) and 9 cases of atypical endometrial hyperplasia were selected from the pathology files of Shinshu University. Formalin-fixed, paraffin-embedded tissue sections were used with the approval of the Ethics Committee of Shinshu University after obtaining written consent from the patients.

DNA was extracted as previously described (12). In brief, four or five 8-μm-thick sections of endometrial tissue were prepared from the paraffin-embedded blocks. In the normal endometrium, endometrial glands were scraped with a razor blade using a microdissection technique. The hyperplastic and malignant tissues were also scraped with a razor from consecutive tissue sections avoiding areas of normal tissue using H&E-stained slides as a template. In neoplastic cases with BRAF mutations, normal tissues were scraped from the same sections adjacent to the neoplastic lesions and examined. For DNA extraction, the tissues were digested for 16 to 24 hours at 55°C in a digestion buffer (2 mg/mL proteinase K and 0.5% Tween 20), then treated with phenol-chloroform.

Mutational analysis. Because BRAF gene mutations in human cancers have been frequently found in exons 11 and 15 (13), we focused on these two exons. The oligonucleotide primers for sequencing BRAF, designed based on a published sequence, were as follows (14): BRAF exon 11, 5′ TCTGTTTGGCTTGACTTGACTTT 3′ and 5′ CATGCCACTTTCCCTTGTAGAC 3′; BRAF exon 15, 5′ AAACTCTTCATAATGCTTGCTCTG 3′ and 5′ GGCCAAAAATTTAATCAGTGGA3 ′.

The primer sets to detect mutation in exon 1 of KRAS and exons 5 to 8 of p53 were designed according to previous reports (15, 16): KRAS exon 1, 5′ GGTGGAGTATTTGATAGTGTA 3′ and 5′ GGTCCTGCACCAGTAATATGC 3′; p53 exon 5, 5′ GGAATTCCTCTTCCTGCAGTAC 3′ and 5′ GGAATTCGCCCCAGCTGCTCACCATCG 3′; p53 exon 6, 5′ GGAATTCGATGGTGAGCAGCTGG 3′ and 5′ GGAATCAGTTGCAAACCAGACCTCAGG 3′; p53 exon 7, 5′ GGAATTCTCCTAGGTTGGCTCTGAC 3′ and 5′ GGAATTCCAAGTGGCTCCTGACCTGGA 3′; and p53 exon 8, 5′ GGAATTCCCTATCCTGAGTAGTGGTAA 3′ and 5′ GGAATTCCTGCTTGCTTACCTCG 3′.

Each exon was amplified by PCR using 0.5 μg of template DNA, sense and antisense primers, and an AmpliTaq Gold PCR kit (Applied Biosystems, Foster City, CA). A total of 50 μL of reaction mixture was made up according to the instructions of the manufacturer and PCR was commenced at 94°C for 3 minutes, followed by 35 cycles of 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 1 minute, with a final extension for 5 minutes. The PCR products were purified using an UltraClean PCR Clean-up kit (MO BIO Laboratories, Solana Beach, CA) and subjected to direct sequencing. The direct sequencing was done using purified products and the same sets of primers in a capillary automatic sequencer (ABI PRISM 3100 Genetic Analyzer, Applied Biosystems) at Shinshu University. Sequence data were analyzed using the Basic Local Alignment Search Tool (BLAST) software located at the National Center for Biotechnology Information web site (http://www.ncbi.nlm.nih.gov).

Immunohistochemistry. Immunostaining was done using specific monoclonal antibodies against hMSH2 (Oncogene Research Products, San Diego, CA) and hMLH1 (BD Biosciences, San Jose, CA). Indirect immunohistochemical staining was done by the avidin-biotin-peroxidase complex method using a Histofine SAB-PO detector kit (Nichirei, Tokyo, Japan) with microwave pretreatment and proteinase K digestion as previously described (17).

The specific staining of each antibody was identified in the nucleus, and immunoreactivity was evaluated based on the nuclear positive cells. The results were semiquantitatively estimated based on the percentage of positive cells as follows: negative, nuclear-positive cells were absent; weakly positive (+), <10% of cells were positive; moderately positive (++), 10% to 50%; strongly positive (+++), >50%, according to previous studies (17, 18).

Statistical analysis. All statistical analyses to evaluate the difference in the frequency of BRAF gene mutations between positive and negative mismatch repair proteins as well as other factors were made with Fisher's exact probability test. A tied P value of <0.05 was considered significant.

Frequency of BRAF mutation. Of the 97 endometrial carcinomas, 20 (21%) contained a total of 21 BRAF gene mutations at five sites (Fig. 1; Tables 1 and 2). The number of cases mutated in exon 11 and 15 was 5 and 15, respectively (Fig. 1; Table 2). The 21 mutations consisted of 17 missense and 4 nonsense mutations. All five mutations in exon 11 were related to the amino acid substitution Y471S. Amino acid changes caused by 16 mutations in exon 15 consisted of K590R (one case), V599G (seven cases), R602stop (four cases), and L617W (four cases). Five silent mutations were also noted (data not shown) but were not included in the present analysis.

According to the FIGO classification, the 20 cases with BRAF gene mutations consisted of 12 stage I, 6 stage III, and 2 stage IV tumors. With regard to histologic subtype, 18 endometrioid (18 of 78, 23%) and 2 nonendometrioid (2 of 19, 11%) carcinomas had BRAF gene mutations. Of the 18 endometrioid carcinomas with a mutation, 8 were grade 1, 5 were grade 2, and 5 were grade 3. The percentage of BRAF gene mutations in grade 1, 2, and 3 tumors was 22% (8 of 36), 26% (5 of 19), and 22% (5 of 23), respectively. There was no significant correlation of BRAF gene mutation with FIGO stage, histologic subtype, or histologic grade. Of nine atypical endometrial hyperplasias, one (11%) had a mutation in exon 15. None of the 20 cases of normal endometrium showed a BRAF gene mutation. In addition, the DNA obtained from a total of 21 normal tissue samples taken adjacent to the neoplastic areas of 20 endometrial carcinomas and one endometrial hyperplasia, all of which harbored a BRAF gene mutation, showed no BRAF mutation. These results indicate that all mutations are somatic.

Correlation of BRAF mutation with KRAS and p53 mutations and mismatch repair proteins. Of 97 endometrial carcinomas, 18 (19%) had KRAS gene mutations (Table 1) in codons 12, 13, 14, and 23 of exon 1. Of 18 KRAS gene mutations, 17 (22%) were observed in endometrioid carcinomas and 1 (5%) in nonendometrioid carcinoma. Of the 18 cases, no case had a mutation in the BRAF gene (Table 3), suggesting mutual exclusion between KRAS and BRAF gene mutations (P = 0.009). An analysis of the p53 gene showed that 22 (23%) cases contained a total of 22 sites where mutations occurred (exon 6, 7; exon 7, 9; exon 8, 6). Of the 22 p53 gene mutations, 9 were found in endometrioid carcinomas (9 of 78, 12%) and 13 in nonendometrioid carcinomas (13 of 19, 68%). Of the 22 cases, 6 were also positive for BRAF gene mutations (Table 3). There was no significant correlation between the p53 and BRAF gene mutations.

Results of immunostaining showed that in the normal endometrium, positive staining for hMLH1 and hMSH2 was observed in the glandular cells of the proliferative phase (Fig. 2A and D) and reduced in the secretory phase. In 97 endometrial carcinoma, 65 (67%) were positive for hMLH1 (Fig. 2B and C) with a mean positive percentage of 14.4 (Tables 3 and 4). Of the 65 hMLH1-positive cases, 51 involved endometrioid carcinomas (51 of 78, 65%; grade 1, 24; grade 2, 11; grade 3, 16) and 13 (13 of 19, 68%) involved nonendometrioid carcinomas. BRAF mutation was more frequently found in hMLH1-negative cases (13 of 32, 41%) than in hMLH1-positive cases (7 of 65, 11%) with a significant difference (P = 0.008; Tables 3 and 4). The hMSH2-positive staining (Fig. 2E and F) was observed in 92 (95%) cases with a mean positive percentage of 37.4 (Tables 3 and 4). Of the 92 hMSH2-positive cases, 75 involved endometrioid carcinomas (75 of 78, 96%; grade 1, 35; grade 2, 18; grade 3, 22) and 17 (17 of 19, 89%) involved nonendometrioid carcinomas. There was no significant correlation between BRAF gene mutation and hMSH2 expression.

The present study showed that the BRAF gene was mutated in 20 cases (21%) of endometrial carcinoma and 1 case (11%) of endometrial hyperplasia. BRAF gene mutations have been identified in a wide range of human tumors with various frequencies. The incidence of BRAF gene mutations is highest in melanoma (63-66%) and thyroid carcinoma (36-53%; refs. 7, 14, 19, 20), followed by colorectal (5-10%) and pancreatic carcinomas (4-7%; refs. 7–9, 21, 22), and is rare in gastric (0-2.2%) and lung carcinomas (1.6-3%; refs. 19, 23–25). Interestingly, a study reported only one BRAF gene mutation, of unknown functional significance, among 146 endometrial carcinoma cases (26), but the reason for this discrepancy is not clear. An intriguing finding of the present study is the frequent occurrence of BRAF gene mutations in hMLH1 protein–negative patients. Several studies have reported that the decreased expression of the hMLH1 protein is closely related to the silencing methylation of the hMLH1 gene promoter in colorectal and endometrial carcinomas (10, 11, 27), and the frequency of mutation in mismatch repair genes is low (28). Therefore, the reduced expression of hMLH1 protein in the present study is likely to be caused by promoter methylation. Similarly, BRAF gene mutation was associated with extensive hMLH1 promoter methylation and with microsatellite instability in colorectal carcinoma (9, 29, 30). In endometrial carcinoma, loss of mismatch function is reportedly associated with PTEN gene mutation (3, 4), suggesting that impaired mismatch repair function is an important factor in the early stage of endometrial carcinogenesis. To our knowledge, this is the first report of BRAF gene mutations associated with decreased expression of hMLH1 protein in endometrial carcinomas.

KRAS gene mutations in endometrioid carcinomas and nonendometrioid carcinomas were observed in 22% and 5% of the cases examined in the present study. p53 gene mutations in the two histologic subtypes were observed in 12% and 68% of the cases. These findings are consistent with previous reports (3, 4). In addition, the frequency of BRAF gene mutations tended to be higher in endometrioid carcinomas (23%) than in nonendometrioid carcinomas (11%), supporting a possible difference in pathogenesis between the two subtypes. Interestingly, all 18 KRAS gene mutation–positive cases were BRAF gene mutation negative, indicating mutual exclusion. Mutual exclusion between KRAS and BRAF gene mutations was observed in colorectal tumors as well (8, 9). These findings suggest that either BRAF or KRAS is needed in the pathogenesis of tumors, in which the Ras-Raf pathway is crucial to the tumorigenesis. However, no such inverse relationship was found between BRAF and p53 in the present study. It was reported that there was no association between BRAF and p53 nor APC gene abnormalities in colorectal tumors (29), suggesting that the p53-related pathway is independent of the Ras-related pathway.

The present study revealed the occurrence of a total of 21 BRAF gene mutations at five sites in 97 cases of endometrial carcinoma. Seven mutations involved Val599, a previously identified hotspot in the kinase domain. However, the amino acid switch observed in the present study (V599G) was different from the most common change (V599E). Therefore, the functional involvement of the V599G mutation needs to be clarified. We found three novel missense mutations (i.e., Y471, K590R, and L617W). Y471S locates close to the G-Loop domain, whereas K590R and L617W are close to the activating kinase domain (13). One study suggested that non-V559E mutations in the kinase domain modulate the Ras-Raf-Mek-Erk pathway, and thus are involved in the pathogenesis of the tumor (9). However, another study reported that mutations of the G-Loop site were not related to the increase in kinase activity and transformation activity (31). In addition, the possibility that the R602stop mutation may suppress tumorigenesis cannot be fully excluded.

The present study revealed that only one (11%) atypical endometrial hyperplasia contained a BRAF gene mutation. On the other hand, 23% of endometrioid-type carcinomas in the present study had BRAF mutations with no apparent correlation with tumor stage or histologic grade. These findings suggest that the BRAF gene mutation is important for malignant transformation, rather than the premalignant stages in the tumorigenesis of endometrioid-type carcinomas. This pattern resembles that observed in colorectal carcinoma because the prevalence of BRAF mutations in colorectal adenomas ranged from 1% to 3%, and in carcinomas from 5% to 10% (8, 9). Conversely, in serous borderline tumors of the ovary and nevi, a precursor of malignant melanoma, the prevalence of BRAF gene mutations in premalignant lesions was the same as that in malignant lesions (32, 33), suggesting that the BRAF activating mutation is an early tumorigenic event in these tumors.

In conclusion, the present study indicated that 21% of endometrial carcinomas showed BRAF mutations including several novel sites, and mutation in endometrial carcinoma was associated with a loss of hMLH1 expression. Although the exact molecular mechanisms of these alterations are not yet elucidated, these mutations seem to be involved in the pathogenesis of a subset of endometrial carcinomas.