Main

Malignant melanoma of the vulva and vagina are uncommon tumors representing <1% of all female genital tract malignancies. Typically, older women are affected; in population-based series, an average patient age of 70 years has been reported.1 Overall prognosis largely depends on tumor stage and is poor in cases with an increased tumor thickness, ulceration, or lymph node metastases.2, 3

Recently, there has been a substantial increase in understanding the biology of malignant melanoma. It has become clear that ultraviolet irradiation-induced melanoma differs in clinical presentation, location, and underlying biological alterations from non-sun-induced tumors. Furthermore, melanomas arising in different mucosal sites have been shown to differ not only from cutaneous tumors but also from site to site with a substantial heterogeneity of alterations in a number of genes, some of which such as BRAF or KIT may be targeted by specific pharmacological inhibitors.

In the present study, we evaluated a large series of melanomas arising in the female genital tract and assessed mutations of the BRAF, NRAS, KIT, and EGFR genes as well as KIT amplifications.

Materials and Methods

A total of 65 primary malignant melanomas of the vulva or vagina were collected from the archives of the Pathology Department at Vancouver General Hospital, Vancouver, British Columbia, Canada (39 cases diagnosed between 1985 and 2010), the Institute of Pathology, A2,2, Mannheim, Germany (16 cases diagnosed between 2000 and 2010) and the Institute of Pathology of the University Hospital, Heidelberg Germany (10 cases diagnosed between 1998 and 2012). Patients with a history of extragenital melanoma or with synchronous extragenital melanomas detected on clinical examination were excluded. Of the 65 tumors, 50 were located on the vulva and 15 originated in the vagina. The slides were reviewed for tumor depth according to Breslow, tumor type (superficial spreading, mucosal lentiginous or nodular), the presence of ulceration or pigmentation and the predominant cell type (epitheloid or spindle cell). Follow-up information was available in 48 cases (median follow-up 15 months, range 1–188 months), within this time, 35 patients had died of disease, 11 were alive, and 2 had died of unrelated causes.

KIT immunostains were performed according to standard procedures. In brief, deparaffinized slides were subjected to heat-induced epitope retrieval (citrate buffer, pH 6.0, Dako, Hamburg, Germany) followed by incubation with a polyclonal KIT antiserum (Dako) at a dilution of 1:50. For visualization, a modified avidin-biotin-complex method was employed using the LSAB+ Kit (Dako) according to the manufacturer’s instructions.

For PCR analysis, tumor tissue was microdissected using glass capillaries and digested as described previously.4 After heat inactivation of the enzyme, the lysate was directly used for PCR under standard conditions using previously published primer combinations for NRAS (exons 2 and 3,5), BRAF (exon 15,5), KIT (exons 11, 13, 17, and 18,6) and EGFR (exons 18, 19, 20, and 21,4). PCR products were directly sequenced on an ABI Prism 377 sequencer (Applied Biosystems, Darmstadt, Germany).

To assess copy number alterations of the KIT gene, a fluorescence in situ hybridization (FISH) probe was generated from BAC clones RP11-586A2 and RP11-273B19 (obtained from Imagenes, Berlin, Germany). In brief, BAC-DNA was isolated (Maxi Prep Kit, Quiagen, Hilden, Germany), fragmented using sonification and fluorescent-labeled using the Platinum Bright 547 nucleic acid labeling kit (Kreatech, Amsterdam, Netherlands). Following co-precipitation of the probe with COT1-DNA (Roche, Mannheim, Germany), the DNA mixture was hybridized onto pre-treated paraffin sections of the tumors as described previously.7

Results

Patient and tumor characteristics are summarized in Tables 1, 2, 3. The average patient age at the time of diagnosis was 72 years (range, 40–89 years). The predominant growth pattern was nodular (63%) followed by mucosal lentiginous (21%), and superficial spreading (16%). Ulceration was present in 72% of cases. In all, 59 of 65 cases (91%) showed an epithelioid morphology, spindle cell differentiation was seen in 6 tumors. Production of melanin was observed in 30 tumors (46%), the majority of cases were deeply infiltrative (36 of 60 informative cases, 60%), a tumor depth of >10 mm was seen in 18 cases (30%). Tumor depth and the presence of lymph node metastases were significantly associated with poorer patient outcome (overall survival) in univariate analysis (Figure 1); multivariate analysis was not performed owing to the low numbers of patients. Vulvar melanomas differed from tumors originating within the vagina with respect to the growth pattern. The superficial spreading and mucosal lentiginous types were significantly associated with vulvar location while most vaginal tumors were nodular melanomas (P=0.016). In addition, spindle cell morphology was only seen in six tumors of the vulva. Vaginal tumors showed a tendency toward greater tumor depth, but these differences failed to reach significance (Table 3). Typical histological features, exemplary immunostains and FISH results are shown in Figure 2.

Table 1 Clinical, pathological and molecular features of 50 vulvar melanomas
Table 2 Clinical, pathological, and molecular features of 15 vaginal melanomas
Table 3 Comparison of clinical and pathological features between vulvar and vaginal melanomas
Figure 1
figure 1

(a) Overall survival for all patients with vulvar or vaginal melanomas. Greater tumor depth (b) and the presence of lymph node metastases (c) were significantly associated with poorer patient outcome (univariate analysis, log rank test), whereas location (vulva vs vagina) was not (d).

Figure 2
figure 2

Representative histology of a nodular ((a) hematoxylin and eosin, original magnification × 40) and mucosal lentiginous melanoma ((b) hematoxylin and eosin, original magnification × 100). (c,d) KIT overexpression ((c) KIT-Immunostain, original magnification × 100) and amplification ((d) FISH, original magnification × 1000) in a case of vulvar melanoma (case 28). (e,f) Heterogeneous KIT overexpression (KIT-IHC, original magnification × 100) and amplification (FISH, original magnification × 1000) in a malignant melanoma of the vagina (case 9).

The immunohistochemical and molecular findings are summarized in Table 4. Using immunohistochemistry, moderate or strong cytoplasmic KIT expression was detected in 30 of the 65 cases (46%). In 54 cases, KIT sequence analysis was successfully performed revealing four exon 11 point mutations (W557R, V559D, V560D, and R586I), two exon 11 insertions (Y578-H580dup and P585 ins REF), and one exon 17 point mutation (D820V). All of these seven tumors showed strong KIT immunostaining (P=0.0014). Increased KIT gene copy numbers defined as more than four FISH signals per nucleus on average were seen in 7 of 57 successfully hybridized tumors (12%), in 4 of the 7 cases more than 10 signals arranged in clusters were observed, whereas in the remaining 3 tumors an average of between 4 and 8 signals was seen. In one of these cases (a deeply infiltrating vaginal melanoma), a high-level KIT amplification resulting in KIT overexpression was observed in approximately half of the tumor cells showing a sharp demarcation from the (superficial) rest of the tumor (Figure 2). No intratumoral heterogeneity was observed in any of the other amplified cases. Six of the seven tumors with increased KIT copy numbers showed moderate or strong KIT staining (P=0.045), whereas one case with an average of 5.5 signals per nucleus was only weakly positive. Seven melanomas harbored NRAS mutations affecting codons 12, 13, or 61 (G12A, 2 × G12V, G13D, 2 × Q61L, and Q61H), no mutations in the sequenced BRAF or EGFR exons were detected. None of the molecular features was associated with patient survival (Figure 3). Although KIT mutations were exclusively observed in vulvar melanomas (P=0.171), KIT amplifications and increased KIT protein levels were seen in both locations. No difference was observed between vulvar and vaginal tumors regarding NRAS mutations.

Table 4 Comparison of immunohistochemical and molecular findings between vulvar and vaginal melanomas
Figure 3
figure 3

Neither KIT alterations (overexpression (a), mutation (b), or amplification (c)) nor NRAS mutations (d) are associated with a significantly better or worse overall survival (univariate analysis, log-rank test).

Discussion

Over the past years, a surprising molecular heterogeneity of malignant melanoma has emerged. Activating V600E or V600K mutations in the BRAF kinase have been observed in up to 62% of melanomas arising in sun-exposed skin.8 Targeting BRAF using specific inhibitors such as dabrafenib or vemurafenib has led to substantially increased survival rates in BRAF mutated, but not in BRAF wild-type melanoma.9, 10 However, in melanomas arising on mucosal surfaces or non-sun-exposed skin, BRAF mutations have only infrequently been reported.8 Accordingly, none of the gynecological melanomas in our series harbored a BRAF mutation.

As somatic BRAF and NRAS mutations are mutually exclusive,11 we next screened our series for mutations in exons 2 and 3 (including codons 12, 13, and 61) of NRAS. NRAS mutations were present in four (three vulvar and one vaginal) tumors indicating a mutation frequency of 12% in gynecological melanomas which is notably lower than in melanomas arising in chronic sun-damaged skin where mutation rates of up to 24% have been reported.12 Interestingly, in contrast to some TP53, CCKN2A, or PTEN mutations that also may be present in melanomas, NRAS alterations typically are not classical ultraviolet irradiation-induced G:C>A:T exchanges or GG:CC>AA:TT exchanges at dipyrimidine sites which points to a more complex mechanism leading to these mutations.13 In fact, in contrast to melanomas in other mucosal sites, esophageal melanomas were reported to harbor NRAS mutations in >30% of cases14 further underlining the lack of direct association with ultraviolet irradiation. Recently, MEK-inhibition was shown to demonstrate therapeutic activity in NRAS-mutated melanoma opening a novel therapeutic option for these tumors.15

KIT mutations and amplifications have been observed in varying frequencies in melanomas arising from different primary sites.16, 17, 18 In addition, KIT protein expression or overexpression as detected by immunohistochemistry has been reported to show some correlation with KIT mutations or amplifications16 but has been insufficient to predict response to KIT-targeted therapy with imatinib.19 In a recent phase II study, response rates for metastatic melanomas treated with imatinib mesylate were 64.7% in patients with KIT exon 11 mutations, 40% for exon 17 mutations, and 33% for KIT amplifications.20 The frequency of KIT mutations in mucosal melanomas has been reported to be as high as 39%.16 We observed five KIT point mutations (exons 11 and 17) and two in-frame insertions (exon 11) in vulvar melanomas, but none in vaginal tumors indicating an important difference in the underlying biology. Although some authors interpret vulvar tumors as melanomas of the non-sun-exposed skin,8 vaginal melanomas show molecular and morphological similarities to esophageal primaries14 that typically lack KIT mutations but may harbor NRAS alterations.

In conclusion, malignant melanomas of the vulva and vagina typically are aggressive tumors associated with a poor overall survival. Tumors in both locations frequently are deeply infiltrating at the time of diagnosis, highlighting the need for novel adjuvant treatment approaches. As overall survival in patients with gynecological melanomas is very poor, our data provide a rationale for KIT mutation testing and targeted treatment approaches in melanomas of the vulva. Targeting of NRAS-mutated tumors with MEK inhibitors may be beneficial in melanomas of the vulva and vagina.