Definition

(ICD-O code 8720/3)

Mucosal malignant melanomas (MMMs) are aggressive tumors arising from melanin producing cells (melanocytes), derived from neural crest tissue.

Etiology

While exposure to sun is well established in the development of skin melanoma, no such association could be considered relevant in the clefts and cavities of the nose and sinuses. However, there has been some interest in occupational exposure to formaldehyde.1,2 Further enquiry among British ENT surgeons revealed 181 cases, 26% of which had occurred in the Midlands (Leicestershire, Nottinghamshire, and Birmingham) but unfortunately the rarity of the tumor has precluded any robust epidemiological studies and the concern remains anecdotal. Formaldehyde has been shown to be a respiratory carcinogen in animal models such as the rat3 but nasal MMM has not been reported in embalmers, who might reasonably be expected to be most at risk. However, they do have a higher incidence of skin melanoma and carcinoma of the colon.4

Comparative genomic hybridization has identified chromosomal abnormalities such as 1q and 6p gains not usually seen with cutaneous melanomas.5,6

Incidence

Cutaneous melanomas are the most common type of which 15 to 33% occur in the skin of the head and neck region.1 Melanocytes are also found in mucosa, secretory glands, nasal stroma, and supporting cells of the olfactory epithelium,7 where they can undergo malignant transformation. MMMs are rare, however, accounting for 1.3% of all malignant melanomas, 55% of which occur in the head and neck region.8 Of these, two-thirds originate in the sinonasal region and a quarter in the oral cavity.1,9–13 Overall, less than 1% of malignant melanomas are sinonasal.1,10,14,15 Of the diverse range of sinonasal malignancies, melanoma accounts for ~4%.15

Site

The most common subsite for sinonasal malignant melanomas is the lateral nasal wall followed by, in order of frequency, septum, maxillary sinus, and ethmoids.15 They rarely originate in the sphenoid sinus, nasopharynx, or nasal vestibule.

In our series of 115 cases, 90 cases (78.3%) originated in the nasal cavity, 12 (10.5%) in the ethmoids with/without nasal cavity involvement, and 7 (6.1%) involved the maxilla (in 6 cases disease was too extensive for site of origin to be determined).

Although rarer still, the possibility of the sinonasal lesion representing a metastasis should be considered. This has occurred on two occasions in our series of cases.

Diagnostic Features

Clinical Features

Sinonasal melanomas are equally common in men and women. In previously reported series the mean age (64.3 years) has been older than for those with cutaneous melanomas.1 The incidence is much higher in Japan, where mucosal melanomas make up a one-quarter to one-third of all melanomas, and they may be more common in black populations.10,16 In our own series of 115 cases, which is the largest series of sinonasal mucosal malignant melanomas from a single institution and was collected prospectively for 47 years (1963–2010), 64 patients were female (55.7%) and 51 male (44.3%). The mean age at the time of initial treatment was 65.9 years (range 15 to 91 years).

Presentation, as with other sinonasal malignancies, may mimic inflammatory conditions, leading to a delayed diagnosis. Unilateral nasal obstruction, a visible mass, and in particular frank epistaxis, are the most common features (Fig. 15.27). On endoscopic examination the lesion may appear pigmented but may be amelanotic in at least 10%. It is often vascular and may be necrotic as it outgrows its blood supply. While there may be a polypoid friable mass, the lesion can also be spreading and more sessile. Furthermore, there may be satellite lesions anywhere on the mucosa of one or both sides of the nose and in addition patches of melanosis may be observed that histologically are made up of melanin within macrophages. Thus, determination of the true extent of the tumor can often be difficult. Diplopia, epiphora, and proptosis are late features and only 1 in 10 present with cervical lymphadenopathy, although both this and systemic metastases can develop at any point.

Histological Features and Differential Diagnosis

Melanoma by definition arises from melanocytes, embryological derivatives of neural crest. These are widely distributed in skin and mucosa, including that of the nose, and are found in the glands, in superficial and deep stroma of the septum and turbinates, and in association with the supporting cells of the olfactory epithelium. Early studies7 suggested that they were absent in fetal and neonatal nasal mucosa.

Obvious melanosis is seen quite commonly in the oral cavity and it has been estimated that a proportion proceed to malignancy (though the range is wide, 0.5 to 30%18).

The tumor can pose some difficulties in diagnosis, composed as it is of many polygonal or spindle-shaped cells with many mitoses and must, therefore, be distinguished from sinonasal undifferentiated carcinoma (SNUC), lymphoma, and nasopharyngeal carcinoma (NPC). Immunohistochemistry is required to confirm the diagnosis by demonstrating protein S100 (expressed by virtually all), VIM, HMB-45, melan-A, tyrosinase, and MIFT (microphthalmia transcription factor) in addition to the presence of intracellular melanin itself (Table 15.1).

Natural History

The majority of patients present with local disease alone and ultimately succumb to local recurrence, which may occur on multiple occasions during the course of the disease. However, a number will present with cervical lymphadenopathy (10 and 18%) and occasionally distant metastasis (4%). Distant spread becomes a major feature of advancing disease; it can occur anywhere but has a predilection for lung, liver, brain, and bone.

In our series at the time of diagnosis 101 patients (91%) had no identifiable lymph node involvement (N−), 10 (9%) had involved lymph nodes (N+), and the status was unknown in 4 of the early cases.

Treatment

In the literature the mainstay of treatment for sinonasal malignant melanomas has been surgical resection, sometimes in combination with postoperative radiotherapy.8,10,15,22,24–27 Traditionally, radical surgery via an open approach was utilized but, as results were unpredictable and generally poor, there has been a growing trend toward endoscopic techniques. Ideally, clear margins should be obtained28 but the proximity or spread of sinonasal tumors to the skull base, orbit, and other vital structures makes this a challenge.

Open approaches have included lateral rhinotomy, midfacial degloving, maxillectomy, rhinectomy, or craniofacial resection with or without orbital clearance with varying degrees of associated morbidity.29–33 Endoscopic resection can often facilitate a similar resection which, although piecemeal, should not be regarded as a less extensive operation.34 The endoscope provides a more detailed (magnified) view of the anatomy and assessment of tumor margins than open techniques, which is particularly important when dealing with a disease that may have satellite lesions, amelanotic areas, and submucosal spread that can be better assessed and meticulously removed. Morbidity is reduced with decreased surgical time, decreased hospital stay, less discomfort, and improved cosmetic outcome.8,35,36 As always, endoscopic techniques can also be combined with open approaches.

Melanoma is traditionally considered a radioresistant tumor but may respond to high doses of radiation.37 Protocols differ from unit to unit but radiotherapy is generally given as adjuvant treatment rather than as a single modality. New radiotherapy techniques such as IMRT and neutron beams are being employed, as is the gamma knife for smaller recurrences.38,39 Despite the lack of statistical proof, the use of postoperative radiotherapy may be justified in patients who are well enough and in the presence of positive margins after resection or of unresectable or recurrent disease.

Future developments are likely to include improved staging techniques such as sentinel node assessment and the use of molecular markers such as S100 and tyrosinase to identify high-risk patients.40–43

Outcome

(See Tables 15.20 and 15.21.)

The prognosis for sinonasal melanomas is poor—worse than for its cutaneous counterpart—and the rarity of the tumor will continue to pose the same difficulties in performing randomized prospective trials. Five-year survival is typically less than 25%, with reports varying between 8% and 48% with a median overall survival of 12.5 to 19.3 months.1,10,11,13,15,22,24,25,46–55 Death is usually a consequence of both local recurrence and metastatic disease.

In our series, virtually all cases were treated with a curative intent. Patients underwent surgery, with selected cases receiving postoperative radiotherapy and/or chemotherapy. Follow-up ranged from 2 to 360 months, mean 37.5 months, in the 109 patients whose follow-up was recorded, 6 being lost to follow-up. The primary management in all cases was surgery, the majority undergoing lateral rhinotomy (n = 71) but this has largely been superseded by endoscopic resection (ESS) (n = 31) in the last 20 years. In addition, 4 patients had a craniofacial resection and 4 a midfacial degloving approach, and all procedures were performed by the same surgeon. Sixty-four (55.7%) had surgery alone, 51 (44.3%) had radiotherapy, 10 received chemotherapy in addition to the adjuvant radiotherapy, and 5 were given adjuvant chemotherapy alone. As can be seen from Table 15.20, the overall median survival was 24 months (standard error [SE] = 5.127; 95% confidence interval [CI] = 13.952–34.048) (Fig. 15.29), 5-year overall survival was 28% and 10-year survival 19.4%. Median disease-free survival as expected was less at 21 months (SE 2.943; 95% CI = 15.232–26.768) with a 5-year disease-free survival of 23.7% and 10-year figure of 9.7%.

If lymph nodes were involved at diagnosis (N+) the overall survival, local control rate, and disease-free survival were significantly worse (Mantel-Cox p < 0.001) with no survivors at 5 years in the N+ group (Fig. 15.30). Postoperative radiotherapy did not confer a survival advantage, either overall or disease-free, nor did it improve local control, but this may reflect a selection bias in addition to the small numbers involved (Fig. 15.31). Consequently, the decision to offer patients postoperative radiotherapy must be on an individual basis and after informed discussion.

Interestingly, when endoscopic resection was compared with open approaches, there was a significant improved overall survival with endoscopic surgery up to 5 years (Mantel-Cox p = 0.013). There was a similar trend, albeit not reaching statistical significance, favoring the endoscopic technique for local control (p = 0.225) and disease-free survival. This cannot be explained simply in terms of extent of disease as the endoscopic technique has been applied to all comers irrespective of stage, and this group is directly comparable to those who underwent lateral rhinotomy and midfacial degloving, which it has replaced in the hands of the same surgeon.

Thus it appears that endoscopic surgery carries a better prognosis than the lateral rhinotomy, midfacial degloving, or rhinectomy group (median 19 months, SE = 4.297; CI = 10.578–27.422), which in turn had a better prognosis than the craniofacial resection or maxillectomy group (median 7 month overall survival, SE = 7.115; CI = 0–20.946) (Figs. 15.32 and 15.33). Several series have now shown comparable, if not improved, outcomes with endoscopic techniques for MMM when compared with open approaches.13,22,35,36 Even allowing for selection and surgical bias, which may be of less relevance in this disease, it appears that endoscopic resection does not adversely affect the outcome in selected cases with appropriately skilled surgeons and may even improve survival. A possible explanation may relate to the low morbidity associated with endoscopic procedures in a disease that exhibits an exquisite immunological balance with its host, often an elderly patient. Why this benefit should persist for some years and then disappear is unknown and requires further investigation.

In common with some other series, we have shown no correlation between prognosis and tumor site of origin.11 Some have suggested that septal lesions fare better, perhaps because they present earlier, while tumors originating in the maxillary or ethmoid sinuses have a worse prognosis than those in the nasal cavity.13,33,54 This could be a consequence of later presentation or because tumors in these sites are less amenable to resection with clear margins due to orbital or skull base involvement, both likely to be negative predictive factors.13,35

In the literature, the most relevant outcome predictors would seem to be the presence of lymph node involvement (as demonstrated in this series) or distant metastasis.15 Locoregional recurrence frequently precedes the occurrence of metastases and may be an independent predictor of survival in sinonasal melanomas.55,56

In our series, radiotherapy was originally administered to patients in whom complete resection with clear margins had not been achieved or for more advanced disease. Our first analysis,30 although it did not demonstrate a statistical advantage, indicated a trend to benefit that might have been compromised by the numbers of patients available for analysis at that time. Subsequently, from 1996, all patients were given the choice of radiotherapy irrespective of extent of disease and this was accepted by 30 of the 50 patients (60%). Again in this larger cohort, if radiotherapy does have some limited benefit, it fails to show statistical significance. The few series examining the effects of radiation on mucosal melanoma typically used pooled series from all head and neck sites. These are summarized by Krengli et al in a literature analysis.38 They conclude that postoperative radiotherapy improves local control and recommend its use for unresectable disease, but there is no evidence to show that it alters survival.38 The series of Moreno et al from the MD Anderson Center retrospectively looked at 58 cases, 33 of which had received postoperative radiotherapy. Although there was no survival benefit, those receiving 54 Gy or more had a statistically lower rate of locoregional recurrence than those receiving 30 to 50 Gy.15 This further highlights the need for high doses of radiotherapy to achieve any response. This is often limited in sinonasal disease by the close proximity of vital structures such as the optic nerve. Other series that claim benefit with adjuvant radiotherapy are based on small numbers and/or lack statistical proof,13,32,33,38 and conversely there are several large series, including our own, that do confirm this benefit in local control.29–31

The only study to report on radiotherapy as a primary treatment modality resulted in an 18% 5-year survival, lower than in most surgical series.37

Insufficient patients have been treated with chemotherapy or vaccines thus far to comment on the likely success in MMM. Various protocols using chemotherapy such as cisplatin and hyperfractionated radiotherapy or iodine 125 have been reported in individual cases.57,58 In contrast, Eigentler et al were able to perform a meta-analysis on 41 RCTs on palliative treatment in cutaneous melanoma.59 However, Hocker et al have reviewed some novel techniques based on an increased understanding of the genetics of melanoma, including agents that target the Bcl-2 signaling network.60 Lotem et al considered the use of interleukin-2 in combination with DNP-modified autologous vaccine in metastatic melanoma61 and there is also interest in targeting anti-cytotoxic T-lymphocyte antigen (CTLA-4) monoclonal antibodies, either as monotherapy or combined with chemotherapy or vaccines.62,63 Whether any benefits in skin melanoma can be translated into mucosal disease remains to be seen. Finally, although malignant melanoma is regarded as a molecularly heterogeneous disease, it has been shown that the mucosal tumors show a variable frequency of KIT mutations64,65 and this may have therapeutic implications for the future with the introduction of new biological agents such as imatinib, sorafinib, and dasatinib.66,67