During embryogenesis, melanocytes migrate from the neural crest and eventually are widely distributed throughout skin, mucosa, and other sites. Melanomas arising in noncutaneous sites are a heterogenous group that account for less than 1.5% of all melanomas. Sinonasal and vulval melanomas account for the majority of these cases; however, other sites include the uveal tract and the mucosa of the respiratory, genitourinary, and gastrointestinal tracts and glaborous (nonhair bearing) skin. Apart from some orbital melanomas, these sites are not exposed to ultraviolet (UV) radiation and recent genomic analyses indicate that they lack the typical signature of UV exposure seen in many cutaneous melanomas. Recent genomic studies have also identified unique mutations in a proportion of noncutaneous melanomas that distinguish them from cutaneous melanomas. In very rare cases, melanomas may arise in unusual mucosal sites such as the bowel, bronchus, and gall bladder. In these situations, it may not be possible to determine whether these are truly primary lesions or isolated metastases from an unknown primary.

Noncutaneous melanomas have a slight female preponderance due to the excess of female genitourinary melanomas, but otherwise there are no gender differences. Noncutaneous melanoma is a disease of older patients and unlike cutaneous melanoma rarely occurs in younger patients. Comorbidities associated with older age may complicate the management of these patients.

The rarity of noncutaneous melanomas has a number of consequences for their management, not the least of which is a reliance on data derived from patients with cutaneous melanomas, which may not be entirely appropriate. The quality of the evidence available is poor as it is invariably based on small series, frequently from a single institution and often collected over many decades. Furthermore, patients with noncutaneous melanomas are usually excluded from large randomized controlled trials in melanoma. Management of these patients is therefore based on consensus recommendations in the absence of high-level evidence. Although many noncutaneous melanomas present at an advanced stage, the lack of validated staging systems makes prediction of outcomes and comparison of results difficult. In the case of lesions arising in mucosa, the lack of a dermal/epidermal junction means tumor thickness is a less reliable prognostic factor than for cutaneous melanoma. There are no site-specific staging systems for most regions where noncutaneous melanomas arise. The use of staging systems for more common cancers located in the same region, for example, the International Federation of Obstetrics and Gynecology (FIGO) staging system for vulval squamous cell cancer, has generally been unsatisfactory. Most recent reports employ the current AJCC/UICC Staging System for Cutaneous Melanoma, which although it has limitations where tumor thickness cannot be appropriately assessed, provides acceptable results.

At the time of writing this chapter, the management of disseminated cutaneous melanoma, as well as the prospects for effective adjuvant therapy, was undergoing an extraordinary revolution. At the present time, there is little information on the role of targeted therapies or immune checkpoint inhibitors in patients with noncutaneous melanomas.¹ Mucosal melanomas and most other noncutaneous melanomas do not have an activating BRAF mutation but up to a quarter may have a mutation in CKIT. Initial experience with specific inhibitors of CKIT has been disappointing; however, with experience and better selection of patients, for example, those with mutations in exons 11 and 9, improved results have been reported.²

Immune checkpoint inhibition with the anti cytotoxic T lymphocyte antigen -4 (CTLA-4) agent ipilimumab has been reported for mucosal melanoma. Response rates appear to be lower than for cutaneous melanoma but the data is sparse.³

Apart from the technical challenge of major resections of noncutaneous melanoma in many sites, the major morbidity and the high risk of recurrence suggest that patients should be discussed or managed at an experienced referral center. It is likely that with increasing experience with the new generation of melanoma treatments for advanced disease, high-risk subgroups, which include the majority of patients with noncutaneous melanoma, will be eligible for adjuvant studies and along with patients who have metastatic disease, patients should be considered for enrolment in these studies.

Orbital melanomas arise from the conjunctiva, uveal tract, and eyelid. They are uncommon, accounting for less than 4% of all cases of melanoma. In the United States, their incidence is six per million (compared to 243 per million for cutaneous melanoma). Spread of melanoma to the orbit from a primary lesion elsewhere is rare. The great majority of orbital melanomas arise in the uveal tract, with conjunctival and eyelid melanomas accounting for approximately 5% each.⁴

Uveal Melanoma

Melanomas arising in the uveal tract account for 90% of orbital melanomas. The commonest site is the posterior uvea, made up of the choroid (90%) and ciliary body (7%). The iris accounts for only 3% and patients with lesions in this site appear to have better survival. Uveal melanomas arise either de novo or from an existing nevus. Nevi are common but the rate of malignant degeneration is very low. The majority of patients are over the age of 60 years. In comparison to cutaneous melanoma, the incidence of uveal melanoma has not increased in recent years and the association with UV exposure is weak. Most patients present with blurred vision, a field defect or orbital discomfort. Distant disease at the time of presentation is uncommon (2%).⁵

Up to 80% of patients with uveal melanoma have mutations in GNAQ or GNA11, which lead to constitutive amplification of the MAP kinase pathway. Melanomas of the iris may represent a distinct biological entity, as GNAQ and GNA11 mutations are rare; however, unlike other uveal tract melanoma BRAF mutations may be seen. A further mutation in the BAP1 gene has been found in one third of patients.⁶

The diagnosis of uveal melanoma can be challenging. Tumor growth over several months is of concern but benign nevi also increase in size. Evaluation of suspected uveal melanoma includes ultrasound, MRI, and fluorescein angiography. Biopsy is controversial even when technically possible.

In a landmark trial, the Collaborative Ocular Melanoma Study Group randomized patients with unilateral choroidal melanomas between 2.5 mm and 10 mm in thickness or diameters of 16 mm or less to enucleation or iodine 131 plaque brachytherapy.⁷ This study demonstrated acceptable quality-of-life and vision for patients treated by radiotherapy and no difference in survival at 12 years between the two treatment arms. Prognostic factors identified were size of the melanoma and age. Melanoma-specific survival at 12 years was 78% in the radiotherapy arm and 83% in the enucleation arm.⁷ Prolonged follow-up from anecdotal and retrospective reports indicates that after 25 years up to 50% of patients with uveal melanoma will have died from metastatic melanoma.⁸

Enucleation is now reserved for more advanced or recurrent choroidal melanomas. Ciliary body melanomas are treated by plaque radiotherapy while most melanomas of the iris can be managed successfully by local excision. As the orbit has no lymphatic drainage, regional lymph node recurrence invariably reflects extension into adjacent tissues. Sentinel node biopsy (SNB) is therefore not indicated for uveal melanoma.

For patients who develop metastatic disease, the overwhelming majority present with hepatic metastases (95%) and the time to recurrence may be prolonged. Lung, bone, skin, and brain are the next most common sites of recurrence. The development of metastatic disease is an ominous sign, with few patients surviving more than 12 months.

Management of liver metastases is notoriously unsuccessful. Responses to standard chemotherapy are unusual and usually of short duration. Surgical resection of hepatic metastases is frequently not possible due to extensive and/or multiple tumors and infrequently leads to prolonged control. Other approaches, including infusional chemotherapy, chemo-embolization, immuno-embolization, and isolated hepatic perfusion with cytotoxic drugs, have not proved particularly successful.⁸

Conjunctival Melanoma

Approximately three quarters of conjunctival melanomas arise in an area of primary acquired melanosis (PAM) with atypia. PAM is relatively common in white persons, and presents as small patches of light brown discoloration of the conjunctiva. However, PAM with atypia is uncommon and reported rates of progression to malignancy vary from 13% to 50%. Of the remaining 25% of conjunctival melanomas, half arise de novo and half from acquired nevi. Congenital nevi rarely progress to melanoma.⁵ In contrast to most other types of noncutaneous melanoma, the incidence of conjunctival melanoma is increasing at a rate comparable to that of cutaneous melanoma.⁹

Patients, usually middle aged or older, generally present having noticed an area of pigmentation. An appropriate biopsy focusing on thickened or prominent areas of pigmentation should be performed. The primary lesion should be completely excised. A minimum 5-mm margin is recommended and supplemental cryotherapy is often recommended. The role of supplemental topical chemotherapy (with mitomycin) or radiotherapy is unresolved, but they are often used in more advanced lesions.⁵ The rate of local recurrence is primarily related to the thickness of the lesion and the completeness of excision, and may be as high as 35% at 5 years. Unlike uveal melanoma, conjunctival melanoma may spread to regional lymph nodes. The experience with SNB is limited but the rate of sentinel node involvement appears to be similar to that of cutaneous melanoma.¹⁰ Selection of patients for SNB is controversial, but there is some consensus that the rate of sentinel node involvement is significant for lesions greater than 2 mm in thickness.¹¹

Adverse prognostic factors include site, nonbulbar conjunctiva, tumor thickness, diffuse or multifocal versus limited disease, epithelioid rather than spindle cell type, elevated mitotic rate, and the presence of lymphatic invasion. Information on specific mutations in conjunctival melanoma is sparse. BRAF mutations have been described in approximately one quarter of patients with conjunctival melanoma but not GNAQ or GNA11, which are seen in uveal melanomas exclusively.⁹

Five-year survival is at least 85%. However, in view of the risk of local, regional, and distant recurrence and the development of new lesions in at least one third of patients, six monthly reviews indefinitely are indicated.

Eyelid Melanoma

Melanoma of the eyelid is also rare but there is even less information to guide management. Histologically, the skin of the eyelids is similar to skin elsewhere in the body and overall, eyelid melanomas behave similarly to cutaneous melanomas elsewhere and have similar outcomes. Lentigo maligna melanoma of the eyelid is more common than elsewhere on the body and lesions on the upper eyelid have a poorer prognosis than lesions on the lower eyelid.

The proximity of major structures limits excision margins and may pose difficulties for restitution of function of the eyelid and adjacent structures and an acceptable cosmetic appearance. Generally a 5-mm margin of excision is recommended; however, a recent study suggested that a smaller margin (2 mm) may be adequate for melanomas less than 1 mm in thickness.¹² Data from the small number of cases of SNB that have been reported indicate that the procedure is technically feasible, but do not allow its efficacy for eyelid melanomas to be assessed.

Anorectal melanomas are rare, with an incidence of approximately 0.4 per million, and they account for less than 1% of all anorectal tumors. A small increase in their incidence rate has been noted.¹³ Anorectal melanomas may arise anywhere from the mucosa above the dentate line to the modified epithelium of the anal canal or to the perianal skin, and although potentially biologically distinct, these melanomas are traditionally grouped as one condition. In over 60% of cases, the melanomas are located in the anal canal, one quarter in the anorectal region, above the dentate line, and the remainder in perianal skin.¹⁴ Anal melanomas frequently spread to inguinal lymph nodes, while rectal melanomas invade locally and initially spread to pelvic lymph nodes. Although rectal melanomas usually present at a more advanced stage than anal or anorectal melanomas, there is little difference in survival by site.¹⁰

Most patients are elderly, with the median age at presentation in the seventh or eighth decade. Patients most commonly present with rectal bleeding and/or a perianal mass. In many cases, the diagnosis is confirmed after excision of a presumed hemorrhoid. Most but not all lesions, particularly those arising in the anal canal and perianal skin, are pigmented. Rectal melanomas are much less likely to be pigmented. Most lesions are raised and may be polypoid in appearance. At presentation 30% of patients have evidence of regional or distant spread.^13,15

Preoperative evaluation should include histologic confirmation as well as staging to exclude distant metastatic disease. Computed tomography (CT) scanning of the chest, abdomen, and pelvis as well as evaluation of the brain either by CT or magnetic resonance imaging (MRI) is indicated. Where available, positron emission tomography (PET) scanning is recommended because of its increased sensitivity for the detection of metastatic disease. Evaluation of the thickness and depth of penetration of the tumor by ultrasound or MRI may assist in deciding on the surgical procedure.

Traditionally, abdomino-perineal resection (APR) has been the surgical procedure of choice because of the depressingly high rate of local recurrence if it was not performed. However, in recent decades wide excision of the melanoma with sphincter preservation has been shown to offer similar survival in appropriately selected patients. A clear margin of 10 mm has been recommended.¹⁴ There has been no prospective randomized evaluation of APR versus wide local excision but a consistent picture emerges from a wide variety of single institution reports, collected reviews, and population-based studies indicating that after wide local excision survival is equivalent to that of APR.^13–18 Approximately 60% of cases are currently managed by wide local excision. Unselected series suggest a local recurrence rate of at least 60% but more recent series have rates of local control approaching those achieved by APR. More recently, the use of adjuvant radiotherapy has provided superior rates of local control, 82% at 5 years in a small series from the MD Anderson Cancer Center but without any impact on overall survival (30% at 5 years).¹⁷ Despite the failure of radical extirpation with APR to provide a survival difference, close or involved margins after either procedure are associated with poorer survival.¹⁴

There is only limited information on which to make recommendations about management of the regional lymph nodes. Lymph node status is a very powerful predictor of outcome but prophylactic lymphadenectomy has not been shown to influence survival.^18,19 In the absence of palpable or imaging-detected lymph node involvement, lymphadenectomy is not indicated. Given the limited information to hand, the role of SNB is at present undefined. Based on experience with cutaneous melanoma, undertaking SNB would seem reasonable in patients with more favorable disease, that is, smaller and/or thinner tumors or where knowledge of the lymph node status may affect treatment decision making. In a small, selected group of patients, unsuspected inguinal lymph node deposits were found in 21%.¹⁷ Although palpable lymph node metastases at presentation are associated with poorer survival, there is currently no evidence that removing small volume disease identified by SNB impacts on survival, particularly given the very high rate of distant recurrence. Periodic high-resolution ultrasound assessment of the regional nodes in the course of follow-up may be a reasonable alternative strategy. Adjuvant radiation therapy to the draining lymph node fields is not recommended as it has no impact on survival and can cause significant morbidity.¹⁷