1 Introduction
Primary malignant melanoma of the vulva (VuM) and vagina (VaM) represent an important subgroup of malignancies with significant differences in terms of biology and treatment compared with the more common and well-described squamous cell carcinoma [
1]. Important differences also exist in terms of anatomic considerations and surgical approach compared with other melanomas [
2‐
5]. Traditionally, VuM and VaM were categorized as mucosal melanoma, but this has recently been questioned by studies examining molecular characteristics of these lesions, which showed that melanomas of the female genital tract differ in terms of mutational characteristics from mucosal and cutaneous melanomas. It has therefore been suggested that VuM and VaM represent a unique subclass [
6‐
8].
The literature on female genital melanoma is scarce; to date, there is only one prospective study following 71 women with VuM who underwent radical (hemi-)vulvectomy [
9]. Retrospective series suggest that the prognosis and survival are significantly worse compared with cutaneous melanoma [
10‐
14].
Staging for VuM has been extrapolated from cutaneous melanoma and the American Joint Committee on Cancer (AJCC) system is now being used instead of the International Federation of Gynecology and Obstetrics classification, although it remains unclear whether this accurately reflects the behavior of VuM [
15]. While surgery remains the primary treatment modality, the US Food and Drug Administration approval of checkpoint inhibitors and targeted therapy has drastically changed the medical management of advanced and metastatic melanoma and significantly improved overall- and melanoma-specific survival [
16‐
22]. Based on these recent advances in skin melanoma and the poor overall prognosis of genital melanomas reported in smaller series, comprehensive study of VuM and VaM is warranted. The aim of this study is to describe the epidemiologic, clinical, and histopathologic characteristics of VuM and VaM and to analyze their impact on survival in a large representative cohort.
2 Methods
2.1 Study Population
The Surveillance, Epidemiology and End Results (SEER) database, a registry funded by the National Cancer Institute, was used to identify cases of VuM and VaM. The SEER-18 population (including Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, Utah, Los Angeles, San Jose-Monterey, Rural Georgia, the Alaska Native Tumor Registry, Greater California, Greater Georgia, Kentucky, Louisiana and New Jersey) in its November 2018 submission version (1975–2016) was used [
23].
Patients with a diagnosis of invasive VuM and VaM were identified in the SEER*Stat 8.3.5 database and their clinical data were retrieved; cases of in situ melanoma were not included. Only patients with known age and listing in the research database were included.
Cases that did not meet the International Association of Cancer Registries criteria were excluded from further analyses. Patients’ age at diagnosis, year of diagnosis, ethnicity, SEER stage (i.e., localized, regional, distant disease), location of melanoma (labia majora, labia minora, clitoris, overlapping), histopathology, and type of primary surgery and lymphadenectomy were collected. The AJCC stage was collected from all cases where information was available and included both AJCC sixth and seventh editions. Histologic subtypes were identified using the following International Classification of Diseases for Oncology, Third Revision codes: 8720, 8721, 8722, 8723, 8726, 8730, 8740, 8741, 8743, 8744, 8745, 8746, 8761, 8770, 8771, 8772, 8773, and 8774. Data on surgery were grouped as local, radical, debulking, and surgery not otherwise specified. Cases where only a biopsy or local destructive procedure (i.e., cryosurgery, laser) was performed were labeled as “no surgery performed”.
Information on ulceration (present, absent, unknown), mitotic count (mitoses/mm
2), and tumor thickness (mm) was extracted. Data on mitotic count were further grouped and analyzed as follows; “0”, “1”, “2–10”, and “> 10”, according to the proposed categories by Nagarajan et al. [
24]. Vital status, disease-specific survival, and time from diagnosis to last follow-up or death were collected. Patients who were reported to have died from a different malignancy were excluded from disease-specific survival analyses. The use of SEER data is exempt from ethics board approval and all retrieved data excluded personal identifiers.
2.2 Statistics
Descriptive statistics was used to report demographic data. Continuous variables were compared using the Student t test, Mann–Whitney test, or Wilcoxon test, as appropriate. More than two groups were compared using analysis of variance or the Kruskal–Wallis test. Cross-tables and the Chi-square test were used to compare categorical data. The Kaplan–Meier method with log-rank test was used to analyze disease-specific survival for both vulvar and vaginal primary sites. Survival was calculated from the date of diagnosis to the date of death. Univariate analysis was used to identify factors with a significant association with disease-specific survival. These were incorporated into a multivariate, stepwise-forward, Cox proportional hazards regression model to detect covariates that were independently related to disease-specific survival. Statistical analysis was performed using SPSS Version 25 (IBM, Armonk, NY, USA). A p value ≤ 0.05 was considered statistically significant.
4 Discussion
Vulvar and vaginal melanomas represent rare malignancies of the female genital tract [
13]. The current evidence and management strategies are mainly based on smaller retrospective series and extrapolation from cutaneous melanoma [
2,
10‐
13]. In this study, we used the SEER-18 registry representing 27.8% of the US population [
23] to characterize demographic, clinical, and histopathologic features in VuM and VaM and assess their implication on prognosis.
We have shown that the overall survival in VuM and VaM is worse compared with the survival rate in cutaneous melanoma across all stages (Fig.
2). This is in agreement with smaller retrospective series from the MD Anderson Cancer Center, Memorial Sloan Kettering Cancer Center, and the AGO Germany and may be explained by later diagnosis and different biology [
10‐
12]. The survival of women with VaM was especially poor. While more women diagnosed with VaM already had metastatic disease (25.1% vs. 6.7%), survival was consistently worse across all disease stages compared with VuM and therefore the poor prognosis cannot solely be attributed to a later diagnosis in more advanced stages of the disease (Figs.
3b,
5a).
The histopathologic subtypes differed significantly between VuM and VaM with the superficial spreading type being the most common form in VuM. This subtype is generally associated with a better prognosis [
25]. In contrast, the nodular subtype was found in more than half of the VaM and has previously been associated with worse survival [
25,
26].
Consistent with previous findings from the GOG-73 study, where 71 women with VuM were prospectively observed [
9], the AJCC staging system was prognostic of disease-specific survival (Fig.
3b) and can be used in women with VuM. However, in our multivariate Cox model, lymph node status was the most important independent predictor of survival (Table
4; Fig.
3d).
The staging system is currently in its eighth edition and includes Breslow thickness, ulceration, lymph node involvement, and distant metastases. A recent retrospective study from the MD Anderson Cancer Center suggested that dermal mitotic rate is an important independent predictor of overall and disease-specific survival in VuM [
24]. While the previous AJCC edition included mitotic rate in its T-stage, this has been omitted in the current version [
15]. We have validated the findings from this study [
24] in a larger cohort and have shown that mitotic rate is an independent predictor of survival that remained significant in the multivariate analysis; we therefore recommend routine assessment during a pathologic work-up.
Surgery remains the primary treatment modality for all locally resectable melanomas [
2,
4,
5]. A surgical margin of 0.5–1.0 cm for melanoma in situ, 1 cm for invasive melanoma with a Breslow thickness ≤ 1 mm, 1–2 cm for a Breslow thickness of 1.01–2 mm, and 2 cm for a Breslow thickness of ≥ 2.01 mm is generally recommended [
2,
4,
5]. While this may be easily achievable without major functional disturbances in most parts of the body, it can be challenging for VuM and VaM in terms of preservation of continence and sexual function. However, strategies of more radical surgery have been attempted in the past in view of the poor prognosis of genital melanoma [
9]. While prospective data are lacking, retrospective data indicate that there is no benefit to more radical surgical approaches compared to local procedures with the above-mentioned surgical margins [
10‐
12]. This is consistent with the findings in the present study where radical surgery did not have a better outcome (Fig.
3c). In fact, the disease-specific survival was worse, but this is likely attributable to the fact that more radical procedures were performed in cases with advanced disease.
Because of the lack of prospective data, the role of lymph node assessment has been controversial in the past and only 52.9% of women with non-metastatic VuM and 42.9% of women with non-metastatic VaM in the SEER-18 population did undergo a lymphadenectomy or a sentinel-node biopsy. In this study, however, the lymph node status was the most important independent predictor of survival and lymph node involvement was consistently associated with prognosis in previous prospective and retrospective analyses [
9‐
12]. The EORTC 18071, Checkmate-238, and Keynote-054 studies have shown improved survival in surgically resected stage III melanoma treated with adjuvant checkpoint inhibitors. Therefore, ipilimumab, nivolumab, and pembrolizumab have recently been approved for adjuvant treatment [
27‐
29]. This underlines the need for all women with malignant melanoma of the female genital tract > 1 mm depth of invasion to undergo sentinel-node biopsy [
3]. In thin melanomas ≤ 1 mm, other risk features including mitotic rate should be evaluated and a sentinel-node biopsy may be offered in those with higher risk features [
3]. In the MSLT-II trial, immediate completion lymph-node dissection did not increase melanoma-specific survival among patients with sentinel-node metastases and can therefore be omitted [
30]. Although the above-mentioned study protocols allowed inclusion of mucosal and vulvovaginal melanoma, the results have not been separately analyzed or reported. However, preliminary data from pooled subgroup analyses from several randomized clinical trials on cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) and programmed death-ligand 1 (PD-L1) inhibitors indicate similar survival improvements compared to cutaneous melanoma albeit to a lesser extent [
21].
5 Strengths and Limitations
This study investigates a large series of well-described cases of VuM and VaM and is representative of the North American population. Because of the rarity of female genital melanomas, prospective data are scarce. Similarly, previously published retrospective single-center experiences have been limited by the small number precluding firm conclusions. The study is, however, limited by its retrospective design and the use of registry data, which do not allow confirmation and preclude central pathology review. In addition, margin status of the surgical specimens was not available, limiting firm conclusions regarding the extent of surgery. Information on chemotherapy and radiation is limited in the SEER database and has therefore not been included in this study.