Abstract
Head and neck mucosal melanoma (HNMM) is a rare and aggressive malignancy. The objective of this study was to describe the outcomes of patients with HNMM. Clinical and pathological data from 51 patients with primary HNMM were reviewed. All patients were treated at a single cancer centre between 1954 and 2012. Most tumours involved the nasal cavity (35.3%) and upper gingiva (29.4%). The majority of lesions were ulcerated (54.9%) and pigmented (84.3%). Forty-three patients underwent surgical treatment and 21 (41.2%) underwent adjuvant chemotherapy and/or radiotherapy. Eight patients (15.7%) received palliative treatment. The median follow-up period was 21 months. During this period, 30 (58.8%) patients had tumour recurrences. At the last clinical evaluation, only seven (13.7%) patients were alive with no evidence of disease and three (5.9%) were alive with HNMM. There were significant differences in overall survival probability according to the presence of ulceration ( P = 0.004), metastatic lymph nodes ( P = 0.003), and treatment including a radical surgical procedure ( P < 0.001). On multivariate analysis, ulceration was the only variable associated with an increased risk of death. Despite the poor prognosis, there was significant improvement in overall survival in the most recent years in this sample, mainly due to advances in diagnosis and reconstruction techniques.
Mucosal melanoma of the upper aerodigestive tract is a rare and aggressive malignancy. The reported prevalence is around 1% of all head and neck cancers and less than 2% of all melanomas (cutaneous melanomas and other locations of extracutaneous melanoma, such as ocular, gastrointestinal, and urogenital). Different from cutaneous melanoma, the pathogenesis of mucosal melanomas is unknown. The function of mucosal melanocytes remains uncertain, but in pathological conditions, these cells are able to produce substantial amounts of melanin.
Mucosal melanoma remains a challenge for several reasons: (1) the clinical diagnosis is not usually confirmed before the disease is symptomatic; (2) traditional aspects of cutaneous melanoma clinical staging may not apply; and (3) histological diagnosis can be difficult due to its rarity and variable appearance. The standard histological predictors of a poor prognosis that affect cutaneous melanoma staging, such as the Breslow depth, ulceration, and mitoses, have not been demonstrated to influence outcomes in mucosal melanoma patients.
The aim of this study was to describe the early and long-term outcomes of patients with mucosal melanoma of the upper digestive airways considering the clinical and demographic characteristics, types of treatment used, complications, and rates of recurrence and survival.
Materials and methods
The medical records of 67 consecutive patients with head and neck mucosal melanoma (HNMM) admitted for treatment to the department of head and neck surgery and otorhinolaryngology of the study institution in São Paulo, Brazil, between 1954 and 2012, were reviewed retrospectively. Fifty-one patients with primary HNMM were considered for the analysis and 16 patients were excluded. The reasons for exclusion were the following: other diagnosis after pathological review, melanoma metastases from other sites, and one case in which HNMM had been diagnosed and treated recently.
The 51 patients with primary HNMM had a median age of 59 years (range 30–88 years). There were 31 (60.8%) males and 20 (39.2%) females. Of these, 41 patients (80.4%) were Caucasian. Six (11.8%) patients had a history of cancer in the family and one patient had additional relatives with malignant melanoma. The duration of signs and symptoms presented at diagnosis and comorbidities were also recorded. Lesions were evaluated clinically, and the findings related to the site, size, extension, and macroscopic appearance were registered in the medical charts ( Fig. 1 ). In the final three decades of the study period, computed tomography (CT) or magnetic resonance imaging (MRI) was used for staging and treatment planning.
Different from other head and neck cancers, HNMMs are classified as clinical stage III and IV. Stage III comprises mucosal disease without regional or distant metastases. Stage IV is subdivided into ‘A’, ‘B’, and ‘C’, in which stage IVA comprises T3 and T4a tumours (moderately advanced disease with or without regional metastasis and without distant metastasis), IVB comprises T4b tumours (very advanced disease with any N clinical stage and without distant metastasis), and IVC comprises any T and any N clinical stage, but with distant metastasis confirmed (TNM, Union for International Cancer Control (UICC), 2010). Most tumours were diagnosed in the advanced stages.
The treatment performed varied according to the period of diagnosis. Before 1990, only surgeons and radiation oncologists participated on tumour boards, but most of the time the treatment decision was made by the attending surgeon and in most cases a non-radical procedure such as cryotherapy or electrosurgical resection therapy was done. After 1990, a multidisciplinary tumour board made treatment recommendations including radical resection, whenever feasible ( Figs 2 and 3 ), with and without neck dissection, reconstruction, and postoperative radiotherapy. For patients with unresectable tumours or those who did not accept surgical therapy, radiation alone or combined with chemotherapy was indicated. Clinical data, pathology reports, treatment information, and outcomes were retrieved from the medical charts of all subjects.
Stata v.12.0 software (StataCorp LP, College Station, TX, USA) was used for the statistical analysis. The survival probability was estimated by Kaplan–Meier method, and the log rank test was performed to compare survival curves. The Cox multivariate regression model was used to identify independent prognostic factors. The 5% level of significance was considered for all statistical tests.
Results
The population studied consisted of 51 patients with primary HNMM ( Table 1 ). The median period of complaints was 6 months (range 0–96 months) and the most common symptom was bleeding (28 cases, 54.9%). Most tumours involved the nasal cavity (18 cases, 35.3%) and upper gingiva (15 cases, 29.4%), followed by the hard palate (9 cases, 17.6%). The majority of lesions were ulcerated (28 cases, 54.9%) and pigmented (43 cases, 84.3%). Thirty-three patients (64.7%) had ulcerated lesions at the time of diagnosis.
| Variable | Category | Frequency (%) |
|---|---|---|
| Gender | Male | 31 (60.8) |
| Female | 20 (39.2) | |
| Age, years | Range | 30–88 |
| Median | 59 | |
| Mean (SD) | 59.4 (13.1) | |
| Race | Caucasian | 41 (80.4) |
| Non-Caucasian | 10 (19.6) | |
| Site | Sinonasal cavity | 21 (41.2) |
| Oral cavity | 30 (58.8) | |
| Lesion surface | Flat | 13 (25.5) |
| Ulcerated | 28 (54.9) | |
| Not reported | 10 (19.6) | |
| Pigmentation | No | 2 (3.9) |
| Yes | 43 (84.3) | |
| Not reported | 6 (11.8) | |
| Clinical stage | III | 17 (33.3) |
| IVA | 22 (43.1) | |
| IVB | 6 (11.8) | |
| IVC | 2 (3.9) | |
| X | 4 (7.8) | |
| Treatment period (years) | 1954–1980 | 14 (27.5) |
| 1981–1990 | 14 (27.5) | |
| 1991–2000 | 8 (15.7) | |
| 2001–2012 | 15 (29.4) | |
| Treatment | Surgery (S) | 19 (37.3) |
| Chemotherapy (C) | 8 (15.7) | |
| Radiotherapy (R) | 0 (0.0) | |
| S + R | 17 (33.3) | |
| S + C | 2 (3.9) | |
| S + C + R | 5 (9.8) | |
| Follow-up, months | Range | 0–133 |
| Median | 20.8 | |
| Mean (SD) | 33.4 (32.5) | |
| Recurrence | No | 21 (41.2) |
| Yes | 30 (58.8) | |
| Status | Alive no disease | 7 (13.7) |
| Alive with disease | 3 (5.9) | |
| Death by HNMM | 30 (58.8) | |
| Death by other disease | 5 (9.8) | |
| Lost | 6 (11.8) |
The tumour clinical stage was reviewed on the basis of data registered in the medical charts and was then classified according to the American Joint Committee on Cancer (AJCC) 7th edition (2010): 17 cases (33.3%) were classified as clinical stage III, 22 (43.1%) as IVA, six (11.8%) as IVB, and two (3.9%) as IVC; the clinical stage could not be classified for four patients (7.8%) because of missing data. Considering the differences in staging methods and treatment approach, the cases were separated according to the decade of admission ( Table 1 ).
A total of 44 patients underwent surgical treatment; six (11.8%) of them had only palliative surgery (electrotherapy or cryotherapy), 17 (31.3%) had radical surgery, and 21 (41.2%) had radical surgery and postoperative adjuvant therapy (chemotherapy ± radiotherapy). The radical surgical procedures included wide resection of the primary tumour, and 13 underwent neck dissection (all with clinically positive neck lymph nodes). Of the neck dissections performed, four were radical neck dissections, three were supraomohyoid extended to level 4, and six were supraomohyoid dissections. A group of eight patients (15.7%) had no surgical treatment. The period of hospitalization associated with the initial surgical treatment ranged from 1 to 15 days.
Regarding postoperative complications, there were 37 cases without complications and six cases with complications, such as operative wound infection, dehiscence, fistula, pulmonary infection, and others.
The median follow-up period was 21 months (range from less than 1 month to 133 months). During this period, 30 patients (58.8%) had disease persistence or recurrence (median time of 9.4 months, range 2.3–63.5 months). Most of the persistent or recurrent disease was at the primary site (26 cases, 86.7%); 12 cases (40%) involved the neck lymph nodes. A total of 14 patients (27.5%) had distant metastasis (four isolated and 10 combined with local and/or regional recurrences). At the end of data collection, seven patients (13.7%) were alive without evidence of disease, three patients (5.9%) were alive with recurrent disease, 30 patients (58.8%) had died of persistent, recurrent, or metastatic disease, and five (9.8%) had died of other causes; no follow-up information (from 2 to 96 months) was available for six patients (11.8%).
On pathological analysis, significantly positive results were not obtained, making the Prasad classification impossible ( Figs 4 and 5 ).
Univariate overall survival and disease-free survival varied according to the period of admission, with better results obtained within the more recent decades ( Table 2 ). The presence of ulceration ( P = 0.004), presence of positive nodes ( P = 0.003), and the treatment performed ( P < 0.001) showed associations with overall survival and disease-free survival rates. Multivariate analysis using the Cox regression model identified ulcerated lesion as the only independent predictor associated with an increased risk of death (relative risk 2.85, 95% confidence interval 1.21–6.73) ( Figs 6 and 7 ).
