Sinonasal melanoma is a rare and aggressive malignancy derived from melanocytes in the nasal cavity and paranasal sinuses. Symptoms vary by location, with paranasal sinus tumors often presenting at advanced stage because of their insidious growth. The staging system reflects their aggressive nature, with TNM staging beginning at T3. Optimal treatment is complete surgical resection. Postoperative radiation therapy is recommended for local control despite the absence of any proven benefit on overall survival (OS). Adjuvant chemotherapy or immunotherapy may incur an OS benefit. Recurrent, metastatic, or inoperable disease may be managed using primary radiation, chemotherapy, immunotherapy, and, more recently, checkpoint inhibitors. Despite these new therapies, estimated 10-year survival rate remains low, at 7%.
32 Melanoma of the Nasal Cavity and Paranasal Sinuses
Melanoma is a malignancy of ectodermal origin that involves the skin in the vast majority of cases. The disease is classically divided according to the site of origin of the primary tumor: for example, cutaneous, noncutaneous, or unknown primary melanoma. Noncutaneous melanomas are infrequent and may be found in the retina, genitourinary tract, anus, and upper aerodigestive tract. The most frequent origin of noncutaneous melanoma is the eye (5.3%), followed by melanomas of unknown primary (2.2%) and mucosal melanomas (1.3%).1 Mucosal melanomas of head and neck origin can arise in the oral cavity, oropharynx, nasal cavity, and paranasal sinuses.
Melanoma of the sinonasal cavity is a rare neoplasm that can involve various compartments of the respiratory mucosa. The first description of mucosal melanoma was by Weber in 1856.2 Thirteen years later, Lucke reported the first resection of a nasal mucosal melanoma in a 52-year-old man who had “melanotic sarcoma.”3 One of the first reports of sinonasal melanoma is attributed to Viennois, who described the surgical extirpation of “polype melanique du nez” infiltrating the globe.4 In 1885, Lincoln made the first report in the English literature of a “melanosarcoma” arising in the nasal antrum and treated with galvanocauterization.5 Since then more than 1,950 cases of mucosal melanoma of the head and neck have been reported in the English literature, a third originating in the sinonasal area.6 Significant progress has been made during the past decade in understanding the biology of melanomas, improved surgical technique, and its sensitivity to chemotherapy, radiotherapy, and immunotherapy.7 However, because of the low prevalence of sinonasal melanoma, the pathophysiology of the disease, as well as the roles of radiotherapy and immunotherapy, remains controversial.
In this chapter, we review the current literature on paranasal and nasal mucosal melanomas, as well as the epidemiology, pathology, staging, and treatment of this disease.
32.2 Incidence and Epidemiology
The incidence of melanoma has been increasing 4 to 6% per year since 1973, a greater rate than for any other human cancer in the United States.8 It is estimated that nearly 1 in 75 persons will develop melanoma during his or her lifetime.9 The main factor believed to be involved in the significant rise in incidence of cutaneous melanoma is exposure to sunlight and ultraviolet radiation. Unlike melanoma of the skin, mucosal melanoma did not show any increase in incidence during this period, suggesting that a distinct pathophysiologic mechanism is associated with this tumor. However, a recent study by Marcus et al has shown an increase in sinonasal melanoma since 1989 and particularly in the decade from 1999 to 2009, although the researchers were unable to identify any causative factors.10
An analysis of the National Cancer Database performed over 9 years (1985–1994) showed that among 84,836 cases of malignant melanoma only 1,074 involved the mucosal membrane, half of which arose in the head and neck, with the sinonasal cavity being the most common site.1 A more recent study pooling statistics from cancer registries across 27 states and 1 metropolitan area in the United States from 1996 to 2000 showed similar results, with only 1,806 cases of mucosal melanoma among 133,209 melanoma cases.11 According to a report by the Armed Forces Institute of Pathology, sinonasal mucosal melanomas account for only 1% of all melanomas and for 0.6 to 4% of all tumors of the nasal cavity and paranasal sinuses.12 Conley reported in his series that melanoma involved up to 6.7% of all sinonasal malignancies.13 Ganly et al studied 1,307 patients who had malignant skull base tumors and found that 4% had sinonasal melanoma.14 Lewis and Martin studied the incidence of malignant melanoma of the nasal cavity in Ugandan Africans and found that this disease accounted for 2.6% of all cases of melanoma.15 In Denmark, melanoma of the upper aerodigestive tract accounts for 0.8% of all melanoma cases and for 8% of all head and neck melanomas.16 Japanese have the highest rate of mucosal melanomas compared with Caucasians (4:1 relative rate), most commonly in the oral cavity.17 Interestingly, the incidence of cutaneous melanoma among Japanese is lower than among Caucasians.
The mean age of presentation of sinonasal melanoma is 65, with a range of 50 to 80 years.18 , 19 , 20 The gender distribution of sinonasal melanomas shows a slight male predominance.21 , 22 , 23 , 24 , 25 Ganly et al have reported that of a total of 53 patients who had anterior skull base melanoma, 70% were males and 30% females.14 Manolidis and Donald reviewed 172 cases of nasal melanomas and reported that 57% of the cases were males and 43% females.26 Female patients who have cutaneous melanoma tend to have a better prognosis than men. However, similar outcomes occurred for males and females who had sinonasal melanoma in regard to long-term survival.27 , 28 , 29
The main factor involved in the development of cutaneous melanoma is exposure to sunlight and ultraviolet radiation, whereas the etiology of the ultraviolet light–protected mucosal melanoma remains unknown. Holmstrom and Lund have suggested that prolonged occupational exposure to formaldehyde may cause significant mucosal irritation, eventually causing paranasal malignant melanoma.30 Similarly, Thompson et al reported formaldehyde exposure in 9 of 115 patients (7.8%) in whom a work history could be elicited.12 The majority of these patients were painters, furniture makers, construction workers, and laundry workers. A joint Danish–Finnish–Swedish case-referent investigation initiated in 1977 studied the connection between nasal and sinonasal cancer and various occupational exposures. The authors found that formaldehyde was evenly distributed among cases involving different tumors of paranasal origin.31
The nasal cavity is the most common origin of head and neck mucosal melanomas (55–79%), followed by the oral cavity.32 , 33 , 34 , 35 The lateral wall of the nasal cavity and the inferior turbinate are the most common origin of melanoma of the nasal mucosa. In the sinuses, the exact site of origin of melanoma is difficult to identify, because most tumors are diagnosed at advanced stage and frequently infiltrate multiple compartments. The most common site of origin for melanoma of the sinuses is the maxillary sinus, followed by the ethmoid and sphenoid sinuses.26 , 29 Melanomas of the frontal sinus, as well as of the middle turbinate, superior turbinate, and cribriform plate, are very rare.36 Table 32.1 shows the sites of origin of sinonasal melanomas among 563 patients reviewed in three different studies.12 , 26 , 37
Manolidis and Donald
Nandapalan et al
Thompson et al a
Floor of nose
Abbreviation: NOS, not otherwise specific.
aIn the other 39 patients, the specific compartment was not specified.
Advanced-stage tumors most frequently involve multiple compartments. For example, in a recent multicenter study, Ganly et al reported that 50% of the patients who had skull base melanomas had tumors infiltrating the cribriform plate, 34% periorbital invasion, 26% orbital invasion, and 17% dural invasion.14
The majority of patients who have sinonasal melanomas present without regional or distant metastases at the time of diagnosis. Positive lymph nodes are found in 4 to 18% of patients at the time of diagnosis.12 , 22 , 23 , 25 The incidence of lymph node involvement in patients who have mucosal melanoma of the oral cavity and oropharynx is 4.7 times higher than in patients who have sinonasal melanoma, owing to the dense lymphatic drainage of the oral cavity and oropharyngeal mucosa.32 Distant metastases of sinonasal melanoma are rare at presentation. For example, Harrison et al reported that none of the 40 patients in their series had distant metastases at presentation.22 In those patients who develop distant metastases, the most common sites are the lungs, bone, liver, brain, and skin.32
The primary cell of origin of melanoma is the melanocyte, which can be found in the nasal and paranasal mucosa. These melanocytes have migrated as neuroectodermal derivatives and embedded in the endodermally derived respiratory mucosa.38 It is estimated that the distribution of melanocytes in the respiratory mucosa is 1,500 cells per square millimeter—less than two-thirds the figure for those found in the skin.39
Mucosal melanosis is defined as a benign pigmented lesion characterized by pigmentation of basal keratinocytes with a normal or slightly increased number of melanocytes.40 Association between mucosal melanosis and increased incidence of nasal or paranasal melanoma has been suggested by several authors.15 , 41 , 42 Preexisting pigmentation of the sinonasal mucosa is seen in less than 10% of patients who have mucosal malignant melanoma.43 Similarly, Thompson et al found preexisting melanosis in only 8% of the patients who had sinonasal melanoma.12 The low rate of melanosis in patients who have mucosal melanoma indicates that most sinonasal melanomas arise de novo.
The majority of melanomas of the respiratory mucosa are large and polypoid and have a median thickness of 9 mm, significantly thicker than melanoma of the oral cavity.44 Thompson et al reported a mean thickness of 7.2 mm (range 2–19 mm) and size of 24 mm (range 5–65 mm), and McLean et al had an average tumor depth > 10 mm for sinonasal melanoma, compared with 3.8 mm for oral.12 , 45 Infrequently, mucosal melanoma in situ can be identified after biopsy of a suspected lesion (Fig. 32.1). The authors found no correlation between survival and tumor thickness. Lee et al have found that depth of invasion > 7 mm is an independent factor for poor prognosis in patients who have mucosal melanomas of the head and neck.46
Melanoma has a notorious tendency to mimic other tumors, and in the sinonasal mucosa it can be easily confused with other tumors, which occur more commonly in this region. The challenge in making the diagnosis of sinonasal melanoma is more significant in cases of amelanotic melanoma and in the presence of ulceration. The Armed Forces Institute of Pathology study has previously shown that more than two-thirds of cases of sinonasal melanoma are misclassified as another neoplasm on initial pathologic evaluation.12
Histologically, mucosal melanoma cells may have different characteristics, such as small cells, spindle cells, epithelioid cells, and, rarely, pleomorphic cells.24 , 47 Spindle cell melanoma appears sarcomatous and is composed of cells whose eosinophilic cytoplasm and nuclei may vary in shape and number. Epithelioid-type melanoma is characterized by large cells that have eosinophilic cytoplasm and acentric nucleus. Although melanomas display morphologic diversity, undifferentiated small round cells or polygonal cells are the most prominent cells found in sinonasal melanomas.44 , 48 A pseudopapillary growth pattern is found in up to 25% of patients who have sinonasal melanomas, but not in melanoma of the oral cavity (Fig. 32.2).24 , 44
The most significant factor in establishing the diagnosis of melanoma is melanin production and the appearance of junctional changes (Fig. 32.3). Melanin pigment is found in two-thirds of cases of sinonasal melanoma and should be considered in a sinonasal myxoid tumor with melanin.12 Such an example is melanoma botryoides, a polypoid tumor that contains small amounts of melanin with a botryoid or myxoid pattern.
Amelanotic tumors show similar biological behavior and prognosis as melanotic nasal melanomas but are more difficult to diagnose than conventional melanomas.49 A variant of melanoma, which does not contain melanin, is desmoplastic melanoma.50 Rarely found in sinonasal melanoma, these cells are composed of amelanotic, poorly circumscribed fascicles and bundles of spindle cells with hyperchromatic nuclei.51 Desmoplastic melanoma is a neurotrophic tumor, which frequently expresses aberrant p53 protein.50 This variant of amelanotic malignant melanoma is difficult to differentiate from other soft tissue tumors of the nasal cavity, such as esthesioneuroblastoma, sarcoma, spindle cell carcinoma, and malignant peripheral nerve sheath tumors.52
Melanoma of the sinonasal mucosa often demonstrates deep invasion, necrosis, and vascular invasion. These characteristics are well-established predictors of reduced survival in cutaneous melanomas and melanomas of the head and neck.44 Prasad et al have shown that 60% of paranasal melanomas are detected at an advanced stage, presenting with significant infiltration of skeletal muscles, cartilage, and bone at the time of surgery.44
Rarely nasal mucosal melanomas may show bone and osteoid formation.12 , 53 Osteoid and metaplastic bone formation may be caused by repeated trauma, mesenchymal metaplasia, reparative reaction to bone invasion, and induction of bone formation of the surrounding tissues.12
Owing to the complicated differential diagnosis of this tumor, immunocytochemical staining is frequently required to establish the diagnosis of melanoma of the paranasal sinuses, particularly in cases of amelanotic variants. Melanoma stains positive for vimentin, HMB-45, and S100 protein. In contrast to cutaneous melanomas, paranasal melanoma is negatively stained for synaptophysin and actin leukocytic common antigen.54 , 55 Interestingly, staining for P-97, which is frequently positive in melanoma of the esophageal mucosa, is not found in melanomas originating in the paranasal sinuses or nasal mucosa.56 , 57 Other antigens, which are specifically stained in paranasal melanomas, are KC-2, SK-46, KBA-62 and PNL2.58 , 59 Other melanoma-associated antigens found in the nasal cavity are tyrosinase (T311), D5, A103 (anti-melan-A/MART-1), and TRP-1.42
In a study of 44 sinonasal melanomas, Prasad et al found that all tumors were positive for tyrosinase, 98% for HMB-45, 95% for S100 protein, and 91% for D5.60 They concluded that tyrosinase is the most sensitive marker for melanomas of nasal or paranasal origin. Thompson et al studied 115 cases of sinonasal melanoma and found positive staining for S100 in 91% of the cases and HMB-45 in 76% of the cases.12 Tyrosinase antigens were expressed in 77.7% of the cases. Table 32.2 shows the different antigens expressed in various tumors of the paranasal sinuses, including malignant melanoma.
Electron microscopy can be used as an adjunct for the diagnosis of melanoma in borderline cases and is very specific in identifying premelanosomes, a subcellular organelle present in melanomas. Wright and Heenan identified a subclass of premelanosomes with a high propensity to metastasize, but these premelanosomes are not found in all cases of mucosal melanoma.22 , 61 , 62
Because melanoma of the paranasal sinuses accounts for less than 7% of neoplasms involving this anatomical compartment, a proper differential diagnosis work-up is crucial for management of the disease. The microscopic differential diagnosis of sinonasal melanoma can be divided into three groups: small round blue cell tumors, pleomorphic cell tumors, and spindle cell tumors.12 The first group includes tumors such as olfactory neuroblastoma, primitive neuroectodermal tumor, Ewing sarcoma, melanocytic neuroectodermal tumor of infancy, pituitary adenoma, lymphoma, plasmacytoma, small cell neuroendocrine carcinoma, and mesenchymal chondrosarcoma. The pleomorphic cells group includes sinonasal undifferentiated carcinoma, anaplastic large cell lymphoma, and rhabdomyosarcoma. The group of tumors characterized by spindle cells includes malignant peripheral nerve sheath tumor, fibrosarcoma, malignant fibrous histiocytoma, and synovial sarcoma.12 , 48 , 63 , 64 , 65 , 66 , 67 Olfactory esthesioneuroblastoma can show similar morphology to melanoma; however, this tumor frequently shows neurofibrillary background, Homer-Wright rosettes, and focal sustentacular cell pattern even in high-grade tumors (Hyman grade II–IV). Furthermore, most esthesioneuroblastomas stain positive for neuron-specific enolase and chromogranin and negative for vimentin.
As described earlier, mucosal melanomas frequently lack melanin pigment and thus can be indistinguishable from other high-grade tumors, such as sinonasal undifferentiated carcinoma, undifferentiated nasopharyngeal carcinoma, poorly differentiated nonkeratinizing squamous cell carcinoma, small cell carcinoma, and anaplastic large cell lymphoma.68 Fortunately, most epithelial tumors show strong cytokeratin immunostaining and fail to express S100 protein. Anaplastic large cell lymphoma does not stain for cytokeratin but expresses CD30 and anaplastic large cell lymphoma kinase proteins. Malignant peripheral nerve sheath tumors may also express S100 protein, complicating the pathologic diagnosis of melanoma. Malignant peripheral nerve sheath tumor does not express other antigens, such as HMB-45, which are frequently found in sinonasal melanoma.
32.4 Clinical Presentation and Findings
The clinical symptoms in patients who have paranasal sinus melanoma are frequently nonspecific and include pain, malaise, and weight loss. Other symptoms may be directly associated with the location of the primary tumor. Melanomas originating in the nasal septum may cause irritation and be visible to the patients.69 This may explain why 75% of patients who have nasal mucosal melanoma are diagnosed early with a clinically localized disease, in comparison with those who have paranasal sinus melanoma. Most patients who have paranasal melanoma suffer from nasal discharge, recurrent epistaxis, or nasal obstruction (85–90% of cases). Thompson et al reported that 52 of 115 patients who had melanoma of the sinuses had epistaxis, 42 a visible mass, and 32 obstructive symptoms.12 Pain (20%) and a visible facial mass (9%) occur in more advanced disease stages.23 Signs representing orbital involvement are proptosis, ophthalmoplegia, decreased visual fields, and monocular blindness and are associated with poor prognosis.14 , 46
Most skin or oral cavity melanomas are more likely to be discovered by the patient or by the primary health care physician on routine examination, whereas sinonasal melanomas are inaccessible to self-examination and are routinely diagnosed at an advanced stage. The duration of the symptoms depends on the biological behavior of the disease. In cases of a slowly growing tumor of paranasal origin, airway obstruction may develop slowly, and the disease, which is obscured from the patient and the physician, may develop months or years before the diagnosis is established.22 Most authors report mean symptom duration of 2 to 8 months.16 , 23 , 34 , 46 , 70 Holdcraft and Gallagher have reported that 50% of their patients had suffered from symptoms for 1 to 4 months before diagnosis and that 30% had had symptoms for 6 to 24 months.21 Thompson et al reported a mean duration of 8.2 months ranging from 2 weeks to 8 years.12
Evaluation of patients who have suspected sinonasal melanoma should include complete history and physical evaluation with emphasis on the head and neck. Fiberoptic evaluation of the paranasal sinuses and upper aerodigestive tract is indicated in all patients to evaluate the tumor extent and potential for resection (Fig. 32.4). Radiologic evaluation should always include both CT and MRI of the head, neck, and paranasal sinuses for evaluation of bony and soft tissue involvement, respectively.71 Patients should be evaluated for involvement of cranial nerve, orbit, skull base, dural, or brain infiltration using both CT and MRI.
Although perineural spread of disease occurs more commonly with squamous cell carcinoma and adenoid cystic carcinoma of the paranasal sinuses, malignant melanoma must also be included in this differential diagnosis, particularly if the patient’s pathology is known to be desmoplastic melanoma. Imaging in patients who have malignant melanoma of the paranasal sinuses should focus on the likely potential for perineural spread. In a recent study of nine patients who had melanomas of the facial skin and paranasal sinuses (including five desmoplastic melanomas) with symptomatic cranial neuropathy, MR imaging demonstrated postgadolinium enhancement of the trigeminal nerve in all nine cases and of other cranial nerves in five cases.72 Other findings included abnormal contrast enhancement and soft tissue thickening in the cavernous sinus, Meckel’s cave, and/or the cisternal segment of the trigeminal nerve.
Suspicious neck metastases can be evaluated with CT, MRI, or ultrasound. Ultrasound-guided fine-needle biopsy can be used if indicated. Sentinel lymph node biopsy, which is frequently used for lymphatic mapping of cutaneous melanomas, is not a common practice for sinonasal disease.73 Due to the high yield in staging metastatic disease, PET imaging now replaces staging techniques employing multiple imaging modalities (i.e., chest X-ray, neck and liver ultrasound, total body CT, and bone mapping).74 Goerres et al have claimed that all mucosal melanomas of the head and neck can be visualized using FDG PET.71 Large lesions that have a nodular growth are better demonstrated than lesions that have superficial mucosal spread. Similarly, lesions originating in the nasal vestibule are more challenging to identify than those in the posterior sinonasal complex owing to nonspecific uptake in the skin and muscles of the mouth.
Metastatic cutaneous melanoma to the paranasal sinuses is very rare and accounts for 1% of patients who have cutaneous melanoma.48 , 75 Nevertheless, full work-up should be performed to exclude isolated metastasis of melanoma of the skin to the paranasal sinuses.
Prior to the 2009 release of the seventh edition of the American Joint Committee on Cancer’s staging manual, there was no formal staging system for mucosal melanoma. This system (mmTNM) is still in the infancy of its general adoption, and several retrospective studies have assessed its validity in terms of the accuracy of predicted prognoses.
The current AJCC system for staging cutaneous melanoma has been in use since its major revision in 2003.76 In this system, the prognosis of patients who have localized disease (T1: tumors < 1.0 mm or T2: 1.0–2.0 mm in thickness) is good, whereas for patients who have melanomas > 2.0 mm in thickness, a worse survival rate is expected (T3, T4). Patients who have localized disease and no regional or distant metastases are classified as having stage I or II disease. In patients who have regional metastases (stage III), tumor burden is expressed as the number of positive nodes (N1 for a single node, N2 for 2–3 nodes, and N3 for ≥ 4 nodes). In patients who have distant metastatic disease (stage IV), the sites of metastases determine outcome, and M classification is graded from “a” to “c” accordingly (i.e., skin, lung, and visceral, respectively). As with cutaneous melanoma, the outcome of mucosal melanoma initially depends on the stage at presentation.46 Unfortunately, because of the absence of histologic landmarks identifiable as a papillary and reticular dermis in the respiratory mucosa, the AJCC cutaneous classification system cannot be applied to this disease. Furthermore, sinonasal melanomas are frequently polypoid rather than deeply invasive, and tumor thickness cannot accurately predict the prognosis.12
The mmTNM system is similar to the cutaneous system, however, in that it is not location-based but rather emphasizes depth of invasion. In recognition of its aggressive behavior, the mmTNM staging system includes no early stages (T1/2). Any primary tumor involving epithelium or submucosa is automatically a T3, and those involving deep soft tissue, cartilage, or bone are T4a, whereas brain, dura, and skull base involvement is classified as T4b. Nodal disease is assessed simply for the absence (N0) or presence (N1) of regional metastases. Any distant metastasis qualifies as M1. The eighth edition of the AJCC staging manual no longer proposes prognostic stage grouping for mucosal melanoma (Table 32.3).77
An alternative system for classification of sinonasal melanoma is the AJCC staging criteria for nasal and paranasal epithelial tumors (carTNM).77 As described in detail in Chapter 29 of this book, this staging system considers the extent of the primary lesion (stage I–IV) and the presence of regional metastases (stage III–IV) or distant metastases (stage IV) as main prognostic indicators for survival. Notably, the AJCC TNM classification applies only to histologically confirmed carcinomas and was not validated for melanoma. Considering these limitations, few authors even prior to the introduction of the mmTNM had adopted the AJCC staging system for nasal and paranasal epithelial tumors in cases of sinonasal mucosal melanoma.23 , 32 , 78
The system most commonly used for staging mucosal melanoma of the head and neck was first suggested by Ballantyne.79 In this classification, stage I disease represents tumors confined to the primary site, stage II the existence of positive regional lymph nodes, and stage III a distant metastatic disease. This system does not take into consideration the size and extent of the disease process, but it has been used repeatedly for classification of sinonasal melanomas and melanomas of the upper aerodigestive tract.29 , 33 , 80 Because regional and metastatic disease is not common during initial diagnosis, most patients are characterized in the stage I group. Clearly, the main drawback of this staging system is its lack of ability to differentiate between patients who have localized disease and those who have advanced tumors and poor prognosis (i.e., orbital infiltration or intracranial extension).
Thompson et al studied a group of 115 patients who had sinonasal melanomas in an attempt to develop a validated staging system by incorporating features of size, site, and regional and distant metastases into a single staging system.12 The T classification of this staging system separates tumors localized to a single anatomical compartment (T1) and those involving more than one anatomical level (T2). The N classification accounts for the absence (N0) or presence (N1) of lymph node metastases (whether ipsilateral, bilateral, or contralateral). Patients who have T1 and T2 disease in the absence of regional or distant metastases are grouped into disease stages I and II, respectively. Stage III disease includes patients who have any T, N1, and M0, whereas patients who have distant metastases are classified as having stage IV disease. The TNM classification suggested by Thompson et al predicted patients’ outcome based on the anatomical site of involvement and metastatic disease.
Prasad et al, at Memorial Sloan Kettering Cancer Center, suggested further microscopic classification of stage I (lymph node–negative) sinonasal melanoma.81 Their microstaging system was performed according to disease invasion into three compartments: level I, melanoma in situ; level II, invasion into the lamina propria; and level III, invasion into deep tissue (i.e., skeletal muscles, bone, or cartilage). Kaplan-Meier analysis showed significant difference in 5-year disease-specific survival of patients who had level I (75%), level II (52%), and level III (23%).
Several studies have aimed to address the prognostic value of the various staging systems. Koivunen et al applied the mmTNM classification to 50 patients diagnosed with sinonasal malignant melanoma between 1990 and 2004 and found statistically significant survival differences in overall survival (OS) according to T stage.82 Similarly, Kuauhyama et al directly compared the mmTNM with Ballantyne’s staging system for 66 patients and found the mmTNM system superior in predicting prognosis.83 However, when Michel et al compared all three systems in 32 patients (Ballantyne/Prasad, mmTNM and carTNM), they found that only the carTNM was significantly correlated with both OS and disease-free survival (DFS). The Ballantyne/Prasad and mmTNM systems correlated with survival only in patients who had metastatic disease in which OS was lower.84 A further study at MD Anderson supported Michel et al’s recommendations of using the carTNM system.85
There is general consensus that surgery remains the treatment of choice for mucosal melanoma of sinonasal origin.80 A recent report of the National Cancer Institute and the Centers for Disease Control and Prevention has demonstrated an absolute gain in OS of melanoma patients during the past decade.86 The improvement in survival of cutaneous melanoma patients, despite the increase in incidence, is attributed to early detection and improvements in therapy. The mode of therapy for sinonasal melanoma awaits further evaluation owing to a lack of prospective studies and objective data for the benefit of adjuvant modalities such as postoperative radiotherapy, immunotherapy, and chemotherapy.
There has been some attempt to define the most appropriate treatment regimes using both meta-analysis and pooling from cancer databases, but owing to limitations with these studies, results are far from conclusive. The largest meta-analysis to date, by Gore et al, which pooled 39 case reports and 423 patients, came to the conclusion that there was no OS benefit for surgery with postoperative radiation therapy but that there might be a significant OS benefit with surgery and chemo-/immunotherapy because of its effect on distant metastatic disease. However, this was only level III evidence with a grade C recommendation.87 In a retrospective study of 695 patients using the National Cancer Database, Konuthula et al analyzed treatment modalities and OS rates, but their only significant finding was that adjuvant radiation failed to confer an OS benefit. What’s more, the database does not account for recurrences, and no data were available regarding its benefit for locoregional control.88 These studies highlight the need for prospective, randomized controlled trials to guide future management.
In the absence of distant metastases, complete tumor extirpation is the mainstay of treatment for malignant melanoma of the paranasal sinuses and nasal mucosa. Although negative margins after surgical resection are reported in more than 85% of patients, it is frequently not possible to achieve en bloc tumor resection, and 75% of patients will eventually develop local recurrence.14 Two possible explanations for the high recurrence rate are (1) presence of multifocal disease and (2) submucosal lymphatic spread of melanoma cells in the respiratory mucosa, which is not clinically or radiographically apparent.29 Freedman et al suggested multicentricity as a main factor predicting local recurrence after surgery.23 Accordingly, wide surgical resection, without unnecessary compromise of function and cosmesis, is essential.
The route of spread for tumors originating in the anterior skull base and paranasal sinuses is determined by the complex anatomy of the craniofacial compartments. A tumor arising in the ethmoid sinus or paranasal cavity may invade laterally to the orbit, inferiorly to the maxillary antrum and palate, posteriorly to the nasopharynx and pterygopalatine fossa, and superiorly to the dura, brain, or cavernous sinus. The recent improvement of endoscopic technology now allows for the resection of benign neoplasms or early malignant neoplasms, with several studies, including a meta-analysis, having suggested comparable results.89 , 90 , 91 However, for the majority of sinonasal melanoma, an open approach is more suitable for allowing extirpation of tumors in an en bloc fashion and with wide margins, in our opinion.
The type of surgery is planned according to the extent of the tumor. For small tumors involving the nasal septum, resection of the tumor along with the perichondrium and septal cartilage may be performed via lateral rhinotomy incision or endoscopically (Fig. 32.5). However, these tumors also frequently infiltrate adjacent structures such as the hard palate, ethmoid sinuses, and medial maxillary walls. In such cases a unilateral or bilateral medial maxillectomy is performed via a lateral rhinotomy incision.
Conventional exposure of the infra- and suprastructure of the maxilla is achieved via a lateral rhinotomy with lip split or subciliary extension as indicated (Fig. 32.6). Tumors infiltrating the cribriform plate and fovea ethmoidalis are safely accessed via the craniofacial or subcranial approach.14 These approaches offer wide exposure of the tumor, allowing resection of the intracranial and extracranial extensions of the tumor.
Massive orbital involvement or orbital apex infiltration requires orbital exenteration. In this case, orbital exenteration is performed with radical maxillectomy or craniofacial resection as determined by the tumor’s extension. Combinations of the craniofacial approach with transorbital and middle fossa approaches were described by Shah et al for malignant tumors of the anterior skull base and paranasal sinuses.92 A combined facial translocation approach was also described and safely used by Hao et al for malignant tumors of the paranasal cavity.93 , 94
After resection of the primary tumor, the surgical margins should be evaluated for residual disease using multiple permanent sections at the periphery of resection, sampling bone, mucosa, soft tissue, and other tissue as indicated. Although complete tumor resection should be the goal of care for sinonasal melanoma, a recent analysis of 53 patients who had skull base melanoma showed no survival benefit of negative margins in this anatomical area.14
Dural and anterior skull base reconstruction is required after craniofacial resection. Dural reconstruction is performed principally using pericranial, galeal, temporalis fascia, or fascia lata grafts. Bovine pericardium can also be used for reconstruction. Fibrin glue is used to provide additional protection against cerebrospinal fluid leak. Reconstruction of the medial orbital walls is not typically performed. If indicated, a split calvarial bone graft, a fascia lata sling, or three-dimensional titanium mesh covered by pericranium are used for reconstruction of the orbital support. A temporalis muscle flap and a split-thickness skin graft to cover the orbital socket can be used after orbital exenteration. In cases of a radical maxillectomy with or without orbital exenteration, a lateral thigh free flap or a rectus abdominis musculocutaneous free flap may be used to obliterate this large defect and to support the obturator.
Because neck lymph nodes are rarely encountered in cases of sinonasal melanoma, neck dissection should be performed only if regional metastases are identified, based on clinical or radiologic evaluation.
The postoperative complication rate after surgical resection of malignant skull base tumors is 36%.14 Among patients treated with craniofacial resection for excision of sinonasal melanomas, postoperative mortality is 6% and major postoperative complications occur in 26% of the patients.
The use of radiation therapy for treatment of melanoma is controversial. Despite the long-standing debate regarding the radio sensitivity of melanoma, there has been a significant increase in the use of adjuvant radiation therapy for the treatment of this disease.1 Both clinical and basic science data indicate that melanoma cells have the ability to repair cellular damage, providing resistance to conventional fractionated radiation therapy.95 It was therefore speculated that hypofractionation or high-dose-per-fractionation (HDPF) therapy would give more effective radiation treatment to these patients. Several nonrandomized, retrospective studies have reported improved locoregional control rates of patients who had high-risk cutaneous melanoma of the head and neck using conventional or hypofractionation adjuvant therapy compared with surgery alone.96 , 97 Moreover, Raben et al reported 70% local control rate in 10 patients treated with the HDPF regimen after surgical resection of head and neck malignant melanoma, with minimal morbidity.98 However, no change in OS was found in this study after hypofractionation adjuvant radiotherapy, owing to the development of disseminated disease. In patients who had mucosal melanoma, Patel et al reported no advantage of postoperative radiotherapy compared with surgery alone.29
In contrast, Ganly et al reported that postoperative radiation therapy was an independent positive predictor of overall, disease-specific, and recurrence-free survival on a multivariate analysis of patients who had skull base melanoma.14 Patients treated with surgery and postoperative radiotherapy had 39% 3-year OS, compared with 18% in patients treated using surgery alone. Freedman et al reported no survival benefit for patients who had paranasal and nasal cavity melanoma treated using surgery and adjuvant radiotherapy, compared with those treated using surgery alone.23 In the same study, it was reported that none of the 18 patients treated using radiation alone survived after 5 years.
Owens et al recently reported a retrospective evaluation of patients who had mucosal melanoma (23% with sinonasal disease) treated using surgery alone, surgery and adjuvant radiotherapy, or surgery and biochemotherapy, with or without adjuvant radiotherapy.99 Radiation therapy was generally used as an adjuvant therapy for patients who had extensive disease. Patients who had sinonasal tumors received 6,000 cGy in 30 fractions, whereas those who had oral lesions received 3,000 cGy in 5 fractions. Biochemotherapy (cisplatin, vinblastine, and dacarbazine, with or without the addition of interferon alfa-2b and interleukin 2) was used almost exclusively in patients who had recurrent disease or distant metastases. The addition of radiotherapy tended to decrease the rate of local failure but did not prevent distant metastases or improve OS. Biochemotherapy regimens used for metastatic or recurrent disease had no significant impact on survival.
An evaluation of the impact of postoperative radiotherapy on local control and survival of patients who had head and neck mucosal melanoma was reported by Temam et al at the Institut Gustave-Roussy (Villejuif).32 Sixty-nine patients who had local disease were managed by surgery without postoperative radiotherapy, two-thirds of whom had sinonasal disease. The study suggested that postoperative radiotherapy increased local control in patients who had small tumors but did not impact survival.
Most reports of definitive radiation therapy for mucosal melanomas involve small series of patients. Gilligan and Slevin reported one of the largest series of melanomas of the paranasal sinuses and nasal cavity treated using radiation alone.25 Complete response was achieved in 79% of the 28 patients included in the study, with relatively low treatment morbidity. In this study, the overall 3- and 5-year survival rates were 49% and 18%, respectively (Table 32.4).
Stern and Guillamondegui at MD Anderson Cancer Center reported two of five patients alive and disease-free 5 years after radiotherapy alone.34 At Princess Margaret Hospital in Toronto, Canada, Harwood and Cummings reported 50% local control rate at 6 months to 4.2 years after primary radiotherapy (n = 10 patients).100 Trotti and Peters reviewed a series of reports using radiotherapy alone for mucosal head and neck melanoma and concluded that 50 to 75% of the patients had documented complete response, with long-term control in 50 to 66%.101 They concluded that in view of the poor results of radical surgery, radiation should be seriously considered as the initial treatment of choice for primary mucosal melanomas of the head and neck. Albertsson et al reported the result of hyperfractionation radiation in combination with cisplatinum.102 Three of four patients treated for local recurrence achieved local control. Radiation may also be appropriate as a primary treatment for patients who have unresectable disease, elderly patients, and patients who have high surgical risk or palliative intent.103 External beam radiation may achieve local control, relieve pain, and decrease tumor compression on vital structures, including cranial nerves, the orbit, airway, and brain.
Radiation therapy has the potential for complications, especially if applied to the cranial base. Severe morbidity associated with radiation of the anterior skull base includes osteoradionecrosis, encephalomalacia, optic neuropathy, and retinopathy.104 Radiation therapy has also been associated with a decreasing quality of life in patients with skull base malignancies.105 Use of heavy particle radiation sources (i.e., proton or carbon ions), as well as accurate delivery using intensity-modulated radiation therapy, may be beneficial in enhancing therapeutic outcomes and reducing complication rates.
Fast neutron therapy was used to treat primary, recurrent, or metastatic cutaneous and mucosal melanoma in 48 patients, showing complete regression in 71%, with a 9% incidence of local recurrence.106 The median survival was 14.5 months, and complications, including fibrosis and necrosis, occurred in 22% of patients. In another series, Linstadt et al reported local control in two of six patients treated with neon ions for melanoma located in a variety of sites, including paranasal sinus.107 Promising results were found in a dose escalation study using carbon ions, reporting 100% 5-year local control rate in five patients who had mucosal melanoma.108
Currently, most centers are using radiation therapy in an adjuvant setting following surgical resection or in cases in which surgery in not possible or is refused by the patient both as an attempt at cure or for palliation.