Clinicopathological Demographics of Malignant Melanoma of the Vulva and Vagina in Japan

Objective: Malignant melanomas of the vulva (VuM) and vagina (VaM) represent a unique subgroup of rare malignant melanomas with critical biological properties and treatment differing from that of other cancers. In Japan, adequate surveys on these have not been performed. The objective of this study was to elucidate the clinicopathological demographics and the outcomes of VuM and VaM in Japan. Methods: This retrospective observational study included women with invasive VuM or VaM, identied from older medical records in Japan. Clinical data were collected and the Kaplan-Meier method was used to analyze progression-free survival (PFS) and overall survival (OS). Univariate and multivariate regression models were used to identify factors signicantly related to survival. Results: A total of 217 patients were identied: 109 (50.2%) with VuM and 108 (49.8%) with VaM. The median PFS was 16.8 months in patients with VuM (95% condence interval [CI] 23.1-87.7) and 15.6 months in patients with VaM (95% CI 8.4-12.6). The median OS was 43.9 months (95% CI 60-138) and 31.1 months (95% CI 24.8-45.3) in patients with VuM and VaM, respectively. Multivariate analysis showed that a >III American Joint Committee on Cancer (AJCC) disease stage (hazard ratio [HR] = 2.063; 95% CI = 0.995-4.278) was associated with poorer PFS, and unknown surgical margin was the only independent factor inuencing OS (HR = 2.188; 95% CI = 1.203-3.977). Conclusions: The overall outcomes of VuM and VaM remain poor in Japan. The AJCC stage and the surgical margin are signicant predictors of survival. (initial lymph node dissection status; or postoperative radiotherapy and concurrent (radiation dose,


Introduction
Vulvar melanoma (VuM) and vaginal melanoma (VaM) are rare carcinomas with an annual age-adjusted reported incidence rate of 1.03 cases per million female population in Asians and Paci c Islanders and 1.90 cases per million in non-Hispanic Whites [1]. They represent only 1-3% of all melanoma cases diagnosed in women [2], as a result of which, epidemiological studies are scarce and evidence-based guidelines for disease management are lacking.
Staging and management for VuM and VaM have therefore been extrapolated from cutaneous melanoma [3]; however, signi cant differences exist in terms of their biology, surgical considerations, and treatments. VuM and VaM are characterized as mucosal melanomas with more aggressive behavior and poorer survival rates than skin melanomas. The reported 5-year overall survival (OS) for mucosal melanoma is signi cantly worse than that for cutaneous melanomas [4] (34% versus 89%) [5]. A small number of retrospective studies evaluated the clinicopathological characteristics of VuM and VaM in speci c populations, con rming clinical differences and worse survival prediction in comparison to cutaneous melanomas [6][7][8][9][10][11][12]. However, these retrospective studies were limited to a single center with a limited number of patients, due to the rarity of the disease. According to a nomogram model presented in a recent SEER population-based study, patients diagnosed with VuM can be divided into high-risk and low-risk groups based on age, race, tumor site, depth of tumor invasion, lymph node status, distant metastasis, tumor size, surgery, chemotherapy, and radiotherapy [8]. To date, only one prospective study in 1994 followed the evolution of patients with VuM who underwent radical hemivulvectomy and led to the conclusion that the American Joint Committee on Cancer (AJCC) staging system was the only independent prognostic factor [13].
The essential rst-line therapy for vulvovaginal melanomas is surgery [14][15]; however, it could be challenging to preserve continence and sexual function. Moreover, compared to other malignant melanomas, vulvar and vaginal melanomas have high recurrence rates of 60% and 80%, respectively [16]. Adjuvant radiotherapy does not seem to improve survival outcomes after surgery [17]. For vulvar melanoma with a depth of invasion greater than 1 mm without the presence of metastasis, sentinel lymph node mapping should be considered. In cases of vaginal melanoma with additional risk factors, resection of the lesion itself is di cult, and sentinel lymph node mapping is not performed [18]. According to a retrospective study by Lopez et al., the sentinel lymph node procedure is capable of identifying occult lymph node metastases in patients with vulvar and vaginal melanoma; however, its clinical application needs to be further validated [19].
Currently, postoperative adjuvant therapies and management of recurrences of VuM and VaM follow the guidelines for cutaneous melanoma [20]. However, recent studies have demonstrated that the molecular pro le of urogenital melanomas differs from other melanoma types [21] and seems to show more similarity to mucosal melanomas than cutaneous melanomas [22].
Retrospective studies are showing the potential bene t of immunotherapy in adjuvant setting in improving the overall survival, especially for patients with distant metastatic disease [7]. The most common mutation in cutaneous melanomas is BRAF V600E; however, it is not present at all in VuM and VaM [23] or is less frequent (7-26%) [24][25]21]. In contrast, the NRAS mutation and c-KIT ampli cation are present in 27.6% of cases of these rare melanomas [24], and the NRAS mutation is associated with poorer survival (33.5 vs. 14.0 months) [26]. C-KIT expression [27] and PARP1 expression [28] have therefore been identi ed as valuable predictors of prognosis and survival. Furthermore, somatic variant analysis of 27 vulvovaginal mucosal melanomas identi ed SF3B1 as the most commonly mutated gene (22% cases) [29]. The knowledge of clinicopathological features and the molecular background of these melanomas would pave the way for targeted therapy and improved survival.
There have been no previous Japanese reports on multiple cases of VuM and VaM. Therefore, the objective of the study was to conduct a survey in Japan to obtain data on patient characteristics and current treatment. However, since it is a rare carcinoma, it required multicentered investigations to accumulate data on a substantial number of cases. Moreover, vulvar and vaginal melanoma were treated by skin cancer specialists in many facilities.
Participation was thus requested among member facilities of the Japan Skin Cancer Society (JSCS) along with members of the Japanese Gynecologic Oncology Group (JGOG) in order to collect data on more patients for a coordinated inter-group study. Access to more extensive data on the current status and treatments for vulvar and vaginal melanoma in Japan should provide more references necessary for planning an effective prospective clinical trial in the future.

Data resource
This was an inter-group study of the JGOG and JSCS on vulvar and vaginal melanoma. This retrospective observational study included patients diagnosed with and treated for vulvar and vaginal melanoma between

Study elogilibity
The following patient data were collected: age; site; Breslow thickness; TNM staging; presence or absence of ulcers; mitotic rate; microsatellite; tumor depth; histological subtype; AJCC 7th staging; treatment-related factors (initial treatment, date of initiation of treatment, and date of completion of treatment); surgery; sentinel lymph node dissection status; preoperative, initial or postoperative radiotherapy and concurrent chemotherapy (radiation eld, method of radiation, total dose, duration of treatment, treatment completion rate); and preoperative, initial, and postoperative chemotherapy (regimen, number of cycles).
The patient outcome was recorded in terms of recurrence, date of con rmation of recurrence, treatment for recurrence, survival, and the last con rmed surviving date.

Statistical analysis
The collected data were analyzed for survival by using the Kaplan-Meier method, and the effects of the clinical factors on OS were investigated using a Cox regression model. In the univariate analysis, we included the following factors: physician in charge (dermatologist or gynecologist), organ (vagina/vulva), age, lymph node metastases, microsatellites, histology (nodular melanoma, super cial spreading melanoma, mucosal lentiginous melanoma, and others), AJCC staging (IA/IB, IIA/IIB/IIC, IIIA/IIIB/IIIC, IV), surgical margin, and surgery. Factors with P < 0.10 in univariate analysis were included to the multivariate model. The magnitude of effect was expressed as hazard ratio (HR) and 95% con dence interval (CI). All statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Inc., Cary, NC). A P < 0.05 value was considered statistically signi cant.  Table 1). There were no statistical differences in patient characteristics between VuM and VaM.  (Fig. 1B).

Patient characteristics
We performed independent analyses of the factors associated with survival for VuM and VaM. The univariate analysis (for VaM and VuM combined) showed that the presence of lymph node metastasis, grade > IIIA according to the AJCC classi cation, positive or unknown surgical margins, and the lack of previous surgery were factors related to both OS and PFS, whereas histology type other than nodular melanoma, super cial spreading melanoma, or mucosal lentiginous melanoma, was related only to OS (Tables 2 and 3).

VuM-speci c outcome
With regard to VuM, the univariate analysis identi ed that the treatment by a dermatologist instead of a gynecologist, age, the presence of lymph node metastasis and microsatellites, and AJCC grade > III are factors related to OS. Only the presence of microsatellites and AJCC grade > III are related to PFS (Tables 2 and 3). According to the multivariate analysis in VuM, the presence of microsatellites was identi ed as a prognostic factor for OS (HR

Discussion
Only a handful of studies have evaluated the clinicopathological features of the rare cancers VuM and VaM, and none of them was conducted in the Japanese population. Therefore, in this retrospective observational study, we analyzed 109 cases of VuM and 108 cases of VaM treated over a period of 20 years at various centers in Japan and aimed to characterize those cases and identify predictors for OS and PFS. According to our study, the OS for VuM and VaM in Japan were 43.6 months and 31.  [15], and Huang Q et al. reported that macroscopic tumor growth and the treatment method are independent prognostic factors for OS [31]. Multivariate progression analysis in a retrospective review of 100 cases of VuM identi ed the tumor thickness, dermal mitotic rate, lymphovascular invasion, microscopic satellite, and absence of precursor nevus as predictors of poor survival [32].
In order to understand the risk factors in the Japanese population, we performed independent analyses of the factors associated with the OS and PFS for VuM and VaM. In a multivariate analysis, we identi ed that an AJCC stage > 4 is associated with poor PFS and that an unknown surgical margin is associated with poorer OS. demonstrated the selection of surgery as a rst-line treatment regardless of the disease stage. Moreover, surgery was identi ed as a positive prognostic factor for OS in VaM (HR = 0.539, 95% CI = 0.284-1.022) and negative surgical margins as a prognostic factor for better OS and PFS in all groups. However, the survival bene ts of radical initial surgery for primary VaM have not been demonstrated yet [33]. Adjuvant chemotherapy and radiotherapy were used in 62.7% and 15.7% of cases, respectively. The relationship between the selected therapy and survival rates was not the objective of this study, and may be explored in further studies.
Overall, we have demonstrated that the survival rates for VuM and VaM remain poor in Japan, and that there are signi cant differences between these subtypes.
Our study has several limitations, the primary one being the inherent limitations of its retrospective design. Salient unmeasured biases that could impact outcomes include medical comorbidities, pathological diagnoses, performance status, surgeons' experiences, patient compliance, and the decision-making processes at multiple participating institutions; all of these factors could affect the treatment approach, operational performance, and survival. In addition, while this study included a diverse range of hospitals in Japan, not all hospitals in Japan participated, which may have resulted in a selection bias. Furthermore, due to the relatively long duration of the study (21 years), there may have been changes in treatment methods resulting in non-uniformity that could have affected the overall survival.
In conclusion, we conducted a large-scale survey and a detailed review of the treatment and clinicopathological features of malignant melanomas of the vulva (VuM) and vagina (VaM) in Japan. Our results show that the overall outcomes of VuM and VaM remain poor in Japan, and that the AJCC stage and surgical margin are signi cant predictors of survival. The results of this study could be used as a basis for decision-making in routine clinical practice and as data for future prospective clinical trials. Declarations