Recurrent KRAS Mutation and Co-mutation of KIT and SF3B1 in Melanoma of the Female Genital Tract

Background: Malignant melanoma of the female genital tract is relatively uncommon and accounts for 5% of all melanomas in women. This study aimed to identify driver genes in melanoma of the female genital tract with the purpose of enhancing understanding of disease pathogenesis and identifying potential new therapeutic targets to develop effective therapies. Methods: KIT (CD117) and BRAF expression were detected immunohistochemically. Polymerase Chain Reaction (PCR) and Sanger sequencing techniques were performed to identify the mutational status of BRAF, NRAS, KRAS, NF1, KIT, PDGFRA and SF3B1 on 19 melanomas of the female genital tract, paired with 25 cutaneous melanomas, 18 acral melanomas and 11 melanomas of nasal cavity. Results: Somatic variant analysis identied KRAS (6/19; 32%) as the most commonly mutated gene, followed by KIT (4/19; 21%), SF3B1 (3/19; 16%) and NRAS (1/19; 5%). None of the cases were found to harbor BRAF, NF1 and PDGFRA mutations in melanomas of the female genital tract. However, none of the cases were found to harbor SF3B1 and KIT mutations in cutaneous melanomas, acral melanomas and melanomas of nasal cavity. Recurrent KIT mutations, as well as mutations in the less frequently mutated genes NRAS and SF3B1, were exclusively detected in vulvovaginal melanomas, but not in tumors arising in the cervix. However, recurrent KRAS mutations were detected in similar frequencies in tumors of the vulva, vagina, and cervix. Additionally, recurrent KRAS and KIT mutations occurred predominantly in polygonal and epithelioid cell types of melanoma in the female genital tract. Immunohistochemistry revealed moderate or strong cytoplasmic CD117 expression in 6 of the 19 cases (31.6%). Conclusions: We observed that gynecologic melanoma harbored distinct mutation rates in the KIT, BRAF, SF3B1, KRAS, and NRAS genes. Our ndings support the notion that gynecologic melanoma is a distinct entity from nongynecologic melanoma, and these ndings offer insights into future therapeutic options for these patients. spindle cells, in melanoma of the female genital tract. We next examined gene mutations in 25 cutaneous melanomas, 8 acral melanomas and 11 melanomas of nasal cavity. We identied BRAF V600E (11/25; 44%) as the most commonly mutated gene, followed by NRAS (2/25; 8%), and KRAS (1/25; 4%) in cutaneous melanomas. While, BRAF V600E mutation was less frequent in acral melanomas (1/18;6%). In total, 28% (3/18) of acral melanomas showed either a KRAS or NRAS mutation (11% KRAS mutation, 17% NRAS mutation). Interestingly, the BRAF V600E hotspot mutation was not detected in the melanomas of nasal cavity. We identied ve cases with NRAS mutations (5/11;45%) and two cases with KRAS mutations (2/11;18%) in the melanomas of nasal cavity. Notably, none of the cases were found to harbor SF3B1, NF1, KIT and PDGFRA mutations in cutaneous melanomas, acral melanomas and melanomas of nasal cavity. In a word, compared with gynecologic melanoma, non-gynecologic melanoma harbored distinct mutation rates in c-KIT, BRAF, SF3B1, KRAS and NRAS genes.


Introduction
Primary malignant melanomas arise from melanocytes which present in the genital mucosal epithelium of the female genital tract are extremely rare, accounting for only 1% of all melanomas. [1][2] The most common site of melanomas of the female genital tract is the vulva (75%), followed by the vagina, and 20% of cases are multifocal. 3 Other sites such as cervix, uterus and ovary are very rare. Patients more frequently present with early metastatic spread and poor prognosis, and the 5-year overall survival rate is approximately 10%. 4 In recent years, many of the advances made in the treatment of cutaneous melanoma with the aim of improving overall survival, but not readily applicable to mucosal melanomas. 5,6 Mucosal melanomas are not related to sun exposure or other known environmental factors and are not familial; therefore, mucosal melanomas have been shown to have distinct molecular pro les compared with cutaneous melanoma arising on sun-exposed sites. 7,8 Some common activating driver mutations in genes identi ed in cutaneous melanoma, such as mutated BRAF V600E, 9 have not been identi ed in mucosal melanoma. [10][11][12][13][14][15][16][17][18] Therefore, the oncogenic drivers of mucosal melanoma remain unclear and no consistent molecular targets have been described. 19 To better understand the mutational pro le and identify potential new therapeutic targets of mucosal melanoma, several investigators have performed next-generation sequencing on a small number of mucosal melanomas from different anatomical sites and compared the results with cutaneous melanoma. 7,20−23 Hayward and colleagues 21 found that acral and mucosal melanomas were dominated by structural changes and mutation signatures of unknown etiology that were not previously identi ed in melanoma, such as in SF3B1.
Newell and colleageues 22 also found that mucosal melanomas had a low point mutation burden and high numbers of structural variants; the signi cantly mutated genes in mucosal melanomas included NRAS, BRAF, NF1, KIT, SF3B1, TP53, SPRED1, ATRX, HLA-A and CHD8 genes. These studies have provided the foundation for understanding the molecular pro les of mucosal melanoma and expanded our knowledge of this rare disease. The distinct mutation pro les of cutaneous and mucosal melanoma are striking and support a model of different developmental pathways.
To better understand the mutational pro le and provide important new clues into the molecular changes that occur in melanomas of the female genital tract, our study analyzed the molecular drivers of 19 melanomas of the female genital tract (vulva, vagina and cervix), paired with 25 cutaneous melanomas, 18 acral melanomas and 11 melanomas of nasal cavity. Our ndings expand this knowledge by contributing the large cohort of mucosal melanoma of the female genital tract known to date with validated mutations and may lead to a better understanding of the molecular drivers, and hence improved therapeutics for these rare forms of melanoma.

Material And Methods
Case collection and histological assessment Nineteen melanomas of the female genital tract, 25 cutaneous melanomas, 18 acral melanomas and 11 melanomas of nasal cavity were obtained from the case les of the Department of Pathology of Fujian Cancer Hospital, China, from October 2010 to September 2019. The biopsy material was xed in 10% formalin and embedded in para n after routine histological tissue processing. Formalin-xed para n-embedded tissue (FFPE) sections (3-4 µm thick) were stained with hematoxylin-eosin for microscopic examination. All cases were carefully reviewed independently by two melanoma pathologists to con rm the histological diagnosis according to the criteria of the 2014 WHO Classi cation of Tumors Female Reproductive Organ, and extensive review of medical records was performed. This study was carried out in accordance with the Declaration of Helsinki and written informed consent was obtained from the patients or their legal guardians. The study protocol was approved by the institutional review boards of Fujian Cancer Hospital.

Immunohistochemistry
Immunohistochemistry assays were performed on diagnosed patient tissues available in the form of FFPE tissue blocks using Bond-III Autostainer (Leica Biosystems, Melbourne, Australia) in the Department of Pathology's clinical immunohistochemistry laboratory. Immunostaining on para n sections was done for CK, EMA, S-100, HMB45, Melan-A, SOX-10, Ki67, KIT (CD117) and BRAF (using antibody clones from Maixin Biotech Co., Ltd., Fuzhou China). Conditions for individual immunohistochemistry assays including antigen retrieval and antibody dilutions varied according to the antibody used and were determined by standard optimization and validation procedures. Positive and negative staining controls were included as appropriate.

DNA isolation
Five pieces (5-µm-thick sections) were cut from FFPE tumor tissues. The sections were depara nized and manually microdissected according to standard procedures. Genomic DNA was isolated using the QIAamp FFPE DNA Tissue Kit (Qiagen, Germantown, MD, USA) according to the manufacturer's instructions. Samples were quanti ed using the Qubit DNA high sensitivity assay (Life Technologies, Carlsbad, USA).

Sanger sequencing
Nested PCR was performed to amplify BRAF exon 15; NRAS exons 2, 3 and 4; KRAS exons 2, 3 and 4; KIT exons 9, 11, 13, 17 and 18; PDGFRA exons 12 and 18; NF1 exon 22; and SF3B1 exon 14 on 19 melanomas of the female genital tract, as previously described. [24][25][26] PCR was carried out using AmpliTaq Gold polymerase (Applied Biosystems, Weiterstadt, Germany) according to the manufacturer's instructions under the following conditions: 95°C for 5 min followed by 38 cycles of denaturation for 30 s at 94°C, annealing for 30 s at 58°C and extension for 60 s at 72°C. The primer pairs for BRAF, NRAS, KRAS, PDGFRA, KIT, NF1 and SF3B1 were designed using Primer Premier 5. The PCR products were routinely puri ed and sequenced in both directions using the BigDye Terminator version 3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA). At least two independent PCR and sequencing experiments were conducted to con rm mutations.

Next-Generation Sequencing Assay
The panel of targeted genes was designed on the basis of large-scale mutation-pro ling studies on melanomas covering BRAF, NRAS, KRAS, NF1, KIT, PDGFRA and SF3B1. The genomes of 25 cutaneous melanomas, 18 acral melanomas and 11 melanomas of nasal cavity with eligible formalin-xed and para n-embedded samples were sequenced using NextsEq. 500 (Illumina, Inc., USA) instrument. The raw data were aligned and analyzed for the detection of insertions/deletions and single-nucleotide variants. The detail process of NGS library preparation, capture-based targeted DNA sequencing and data analysis were described in supplementary material.

Clinicopathologic Findings of patients with melanomas of the female genital tract
This study included 19 patients with primary melanomas of the female genital tract, with a median patient age of 53.0 years and an average age of 55.4 years at diagnosis (range from 34 to 84 years). The primary melanomas of the female genital tract included ve vulvar (26.3%), eight vaginal (42.1%) and six cervical melanomas (31.6%). Six (31.6%) patients had T4 disease, two (10.5%) had T3 disease, three (15.8%) patients had T1 disease and the remaining eight (42.1%) patients had T2 disease. The clinicopathological features of the patients are summarized in Table 1.
Microscopic examination revealed that the tumor cells were distributed in solid islets, nests or band-like formations. Tumor cells were polygonal, epithelioid or spindle in shape and rarely of small cell type, with oval, pleomorphic, hyperchromic nuclei ( Fig. 1A-C). Melanin pigment was diffuse or focal and even absent. Mitoses were easy to see. Immunohistochemical analysis revealed that CK was negative in most cases and weakly focally positive in three cases (3/19, 15.8%) (Fig. 1D and Table 1). The tumor cells were strongly positive for S-100 and SOX-10 protein (Fig. 1E-F). Positive Melan-A and HMB-45 protein expressions were also detected, with somewhat lower intensity of staining (Fig. 1G-H). Tumor cells were negative for EMA (data not shown). These ndings supported the diagnosis of melanoma. Immunohistochemistry revealed moderate or strong cytoplasmic CD117 expression in 6 of the 19 cases (31.6%) ( Fig. 2A-B). None of the cases were positive for BRAF (Fig. 2C).
BRAF, NRAS, KRAS, NF1, KIT, PDGFRA and SF3B1 gene mutation analysis We examined gene mutations in the 19 melanomas of the female genital tract (listed in Table 1). We identi ed six cases with KRAS mutations (four p.G13D, two p.G12D) and one cases with NRAS mutation (p.Q61*) (Fig. 3A-C). In total, 37% of tumors showed either a KRAS or NRAS mutation (32% KRAS mutation, 5% NRAS mutation). The SF3B1 p.R625H hotspot mutation was detected in 16% (3/19) of the mucosal melanomas of the female genital tract (Fig. 3D). Notably, recurrent KIT mutations (p.S476N, p.G498V, p.L640P, p.D810H, p.V852A) were found in 21% (4/19) of the melanomas of the female genital tract; one sample was found to harbor both p.S476N and p.G498V mutations (Fig. 4). In addition, one case had both KRAS and C-KIT mutations. All four tumors with recurrent KIT mutations showed strong CD117 immunostaining. In contrast, oncogenic driver mutations in BRAF, which are commonly identi ed in cutaneous melanoma, were not detected in any sample (data not shown). None of the cases were found to harbor NF1 and PDGFRA mutations (data not shown).
Notably, recurrent KIT mutations, as well as less frequent gene mutations in NRAS and SF3B1, were exclusively detected in vulvovaginal melanomas, but not in tumors arising in the cervix (Table 1). However, recurrent KRAS mutations were detected at a similar frequency in tumors of the vulva, vagina and cervix. Additionally, high numbers of KRAS and KIT mutations were identi ed with frequencies varying according to histological subtype. Interestingly, the recurrent KRAS and KIT mutations occurred predominantly in polygonal and epithelioid cell subtypes, but rarely in spindle cells, in melanoma of the female genital tract.

Discussion
Primary malignant melanomas of the female genital tract are extremely rare. The clinical behavior and molecular characteristics of these melanomas have not been well explored. Although melanoma of the female genital tract is an aggressive disease with histological resemblance to melanomas of other sites, recent studies found the heterogeneity of molecular biology of melanoma of different sites. Up to date, relatively little information is known about the molecular alterations that drive melanoma of the female genital tract. [10][11][12][13][14][15][16][17][18] To better understand the mutational pro le and offer insights into future therapeutic options for patients with melanomas of the female genital tract, our study analyzed the histological and genetic characteristics of 19 melanomas of the female genital tract (vulva, vagina and cervix), paired with 25 cutaneous melanomas, 18 acral melanomas and 11 melanomas of nasal cavity.
Activating V600E or V600K mutations in BRAF kinase have been observed in up to 62% of melanomas arising in sun-exposed skin. However, in melanomas arising on mucosal surfaces or non-sun-exposed skin, BRAF mutations are infrequently reported. 11 Previous studies showed that BRAF was mutated in 0-33% of patients with vulvar and vaginal melanomas with sample sizes ranging from 1 to 51 cases. [27][28][29] In our study, oncogenic driver mutations in BRAF V600E, which were commonly identi ed in 44% cutaneous melanoma, were not detected in the melanomas of female genital tract. Our nding is similar to most published data on vulvovaginal melanomas. [10][11][12][13][14][15][16][17][18] The differences between our ndings and some published studies reporting on BRAF mutations in urogenital melanomas or vulvovaginal melanomas are unclear. We doubt the small number of samples in our series (19 patients) could account for this discrepancy. One explanation may lie in the use of different mutation screening methods, which vary in sensitivity. In our study, Sanger sequencing (covering exon 15) was used to detect BRAF mutation. In contrast, Hou and colleagues 27 used a combination of next-generation sequencing (covering exons 1-18) and Sanger sequencing (covering exons 11 and 15). In addition, many of their samples were metastatic and may have harbored mutations that differed from the molecular makeup of the primary tumor. Notably, only some of the BRAF-mutant vulvar and vaginal melanomas in the literature harbored BRAF V600E mutations. [27][28][29] A literature search of the remaining BRAF variant-mutant in vulvovaginal melanomas revealed possible inactivating mutations that were less likely to respond to vemurafenib, which is the FDA-approved selective inhibitor of the V600E mutant BRAF kinase used to treat patients who have metastatic or unresectable melanoma with BRAF mutations. Our results indicate that none of the patients with melanomas of the female genital tract can be treated with vemurafenib.
According to the literatures, KRAS mutations are common in pancreas, colon and lung cancers 30 , whereas NRAS mutations are common in myeloid leukemias and cutaneous melanomas. [30][31][32] However, we identi ed six KRAS mutations and one NRAS mutation in 19 melanomas of the female genital tract. In total, 37% of tumors showed either a KRAS or NRAS mutation (32% KRAS, 5% NRAS). As reported previously, the mutations were found to be mutually exclusive. In our study, the prevalence of KRAS mutation in melanomas of the female genital tract was notably higher than melanomas of other sites, whereas the prevalence of NRAS mutation in melanomas of the female genital tract was notably lower compared with the prevalence in melanomas arising in nasal cavity, where mutation rates of up to 45%. Our nding is similar to the published data on esophageal melanomas, which harbored NRAS mutations in 30% of cases. 33 Recurrent KRAS or NRAS mutation contribute to poor prognosis. However, in recent years, MEK inhibition was shown to demonstrate therapeutic activity in NRAS-mutated melanoma in clinical trials, opening a novel therapeutic era for these tumors. 34 KIT mutations have been observed in varying frequencies in melanomas arising at different primary sites. KIT protein expression or overexpression as detected by immunohistochemistry has been reported to show some correlation with KIT gene mutations but has been insu cient to predict response to KIT-targeted therapy with imatinib. In our study, moderate or strong cytoplasmic KIT expression was detected in 6 of the 19 cases (31.6%), and KIT mutations were observed in 21% (4/19) of the mucosal melanomas of female genital tract. All four tumors with KIT mutations showed strong KIT immunostaining. This nding shows that KIT protein expression correlated with both KIT mutations and ampli cation. The frequency of KIT mutation in our series was much higher than rates reported in studies on non-gynecologic melanoma. 27 Interestingly, KIT mutations were associated with histological subtype and tumor site. Notably, recurrent KIT mutations were exclusively detected in vulvovaginal melanomas, but not in tumors arising in the cervix, and KIT mutation varied immensely between vulvar and vaginal sites, with 20% (1/5) of vulvar samples harboring the mutation compared with only 37.5% (3/8) of vaginal samples. This further highlights our conclusion that mucosal melanomas of the female genital tract have a genetic pro le that is distinct from that of mucosal melanomas from different anatomical sites. In addition, we found that KIT mutations occurred predominantly in polygonal and epithelioid cell subtypes, but rarely in spindle cells. However, our ndings are different from those of Hou and colleagues 27 that showed that vulvar melanoma may be associated with a much higher KIT mutation rate than vaginal melanoma. The differences between our ndings and published studies on KIT mutations in vulvar and vaginal melanomas could be due to the small numbers of samples in our series, different methodology or ethnic difference. It is also possible that vulvar tumors were regarded as melanomas of non-sun-exposed areas. In addition, co-mutations of KIT and NF1 have been reported in mucosal melanoma, although they are rare. 35 However, in our study, none of the cases were found to harbor NF1 and PDGFRA mutations in melanomas of the female genital tract, as well as in cutaneous melanomas, acral melanomas and melanomas of nasal cavity. I think the small number of samples in our series could account for this discrepancy. SF3B1 mutations have been identi ed in subsets of solid tumors, as well as in myelodysplastic syndrome and chronic lymphocytic leukemia. [36][37][38] Recently, SF3B1 was identi ed as a signi cantly mutated gene in mucosal melanoma, especially in uveal, female genital and anorectal melanomas. [20][21][22][23] Our study also found that the SF3B1 R625 hotspot mutation occurred in 16% of the mucosal melanomas of the female genital tract and not detected in cutaneous melanomas, acral melanomas and melanomas of nasal cavity. SF3B1 mutations have different prognostic associations in different types of cancers. In uveal melanoma, SF3B1 mutations are associated with a better prognosis, whereas in other mucosal melanomas, 36 SF3B1 mutations are correlated with a worse prognosis. 25 However, the study of these rare tumors was underpowered to detect statistically signi cant differences, and larger studies are required to address this issue. This nding suggests that SF3B1 might be exploited as a novel prognostic and/or therapeutic target in melanomas of the female genital tract.

Conclusion
We observed that gynecologic melanoma harbored distinct mutation rates in c-KIT, BRAF, SF3B1, KRAS and NRAS genes compared with non-gynecologic melanoma. Our ndings support the notion that gynecologic melanoma is a distinct entity from non-gynecologic melanoma, especially cutaneous melanomas. Although our results are preliminary, they highlight the unique molecular landscape of gynecologic melanoma within the spectrum of melanoma malignancies, and these ndings offer insights into future therapeutic options for these patients.

Declarations
Funding This study was funded by the Natural Science Foundation of Fujian Province (grant number 2018J01277), Joint Funds for the Innovation of Science and Technology, Fujian province (grant number 2017Y9081) and Science and Technology Program of Fujian Province, China (grant number 2018Y2003).

Con ict of Interest statement
All authors declare that they have no con ict of interest.

Ethical approval
This study was carried out in accordance with the Declaration of Helsinki and written informed consent was obtained from the patients or their legal guardians. The study protocol was approved by the institutional review boards of Fujian Cancer Hospital.