Non-cutaneous melanoma

Overview

• Melanocytes are neural-crest derived and, besides the skin, populate the eye, mucosal surfaces, and meninges, giving rise to non-cutaneous melanomas (non-CMs).
• Non-CMs comprise ~5% of all melanomas: ~3 to 5% uveal and ~1 to 2% mucosal. Most mucosal melanomas are head and neck (~55%: conjunctival, sinonasal, oropharyngeal), then anorectal (~24%) and vulvovaginal (~18%).
• In general, non-cutaneous melanomas are more aggressive and respond less well to immunotherapy than cutaneous melanoma.

Uveal (ocular) melanoma

• Most common primary intraocular malignancy in adults; arises from uveal melanocytes (choroid most common, also ciliary body and iris).
• Biology/genetics: activating GNAQ or GNA11 mutations (G-protein pathway) are present in the majority and are early, initiating events. BAP1 loss is strongly associated with the aggressive class 2 (high metastatic risk), though class is defined by the gene-expression profile, not BAP1 status; EIF1AX associates with low metastatic risk; SF3B1 with intermediate risk and characteristically late-onset metastases. Uveal melanoma has a low incidence of T cells expressing immunoexhaustion markers, which helps explain poor IO responsiveness.
• Risk stratification: the choroid lacks classic lymphatic drainage, so spread is hematogenous, predominantly to the liver (nodal spread is skipped). Gene-expression profiling classifies tumors as low-risk (class 1) or high-risk (class 2); DecisionDx-UM is a commercial 15-gene assay used to tailor surveillance.
• Prognosis: once metastatic, poor (median OS historically ~6 to 13 months). Poor prognostic factors: ECOG ≥1, high LDH, short interval from primary diagnosis to metastasis.
• Local treatment of primary: eye-preserving brachytherapy (iodine-125 plaque) or proton beam, or enucleation. The Collaborative Ocular Melanoma Study (N = 1,317) showed no difference in melanoma-specific OS between iodine-125 plaque radiation and enucleation.
• Metastatic (liver-dominant):
  • Tebentafusp (Kimmtrak), a bispecific gp100 × CD3 (HLA-A*02:01-restricted) engineered T-cell receptor fusion protein:
    • IMCgp100-202 (N = 378, randomized 2:1 vs investigator choice of pembrolizumab, ipilimumab, or dacarbazine): mOS 21.7 vs 16.0 mo (HR 0.51). FDA-approved Jan 2022 for HLA-A*02:01-positive unresectable/metastatic uveal melanoma. First and only therapy to improve OS in metastatic uveal melanoma.
    • 3-yr update (Hassel NEJM 2023; final 5-yr, Piperno-Neumann Ann Oncol 2026): 3-yr OS 27% vs 18% (HR 0.68); 5-yr OS 16% vs 8% (HR 0.67); benefit consistent across subgroups. Effect is independent of tumor shrinkage (many patients benefit without a RECIST response, so do not discontinue for early progression alone if tolerated).
    • Eligibility: HLA-A*02:01 positive (test before starting).
    • Toxicity: cytokine release syndrome (especially first 3 doses, requires inpatient monitoring), rash.
  • Darovasertib (IDE196), a PKC inhibitor active against GNAQ/GNA11-mutant disease. Phase 2 in neoadjuvant primary uveal melanoma showed tumor shrinkage enabling eye-preserving therapy (Shoushtari ASCO 2024); FDA Breakthrough Therapy Designation Mar 2025 (neoadjuvant monotherapy, primary UM requiring enucleation). Metastatic combinations with crizotinib and with tebentafusp in phase 2/3.
  • FOCUS trial (Zager Ann Surg Oncol 2024): melphalan percutaneous hepatic perfusion (Hepzato) in liver-dominant uveal metastases; randomization vs best alternative care was abandoned (single-arm), ORR 36.3%; FDA Aug 14, 2023 for unresectable hepatic metastases affecting <50% of the liver, with no extrahepatic disease or only limited resectable/radiable extrahepatic disease (bone, nodes, subcutaneous, lung). Adds a liver-directed device option alongside selective internal RT and immunoembolization.
  • Standard IO: much less effective than in cutaneous melanoma. Combination nivolumab + ipilimumab ORR ~11.6% (still higher than single-agent PD-1/L1 ~3.6%); two phase 2 studies confirmed a clinical benefit inferior to that in cutaneous and mucosal melanoma. MEK inhibitors have shown little benefit.
  • Surveillance: liver MRI/US every 6 months (higher-risk BAP1/class 2).

Mucosal melanoma

• Sites: head/neck (sinonasal, oral cavity), anorectal, vulvovaginal.
• Genetics: lower somatic mutation burden than CM; oncogenic BRAF (~6.4%) and NRAS (~13.8%) are less frequent than in CM, while KIT mutations are more common. Prognostic factors: depth of primary, number of positive nodes, sites of distant metastasis.
• Behavior: aggressive; often diagnosed at advanced stage; worse prognosis than CM regardless of anatomic site.
• Localized: complete surgical resection with negative margins is prognostically important and can be curative in select cases. SLN mapping utility is controversial; lymph node dissection if clinical/radiologic nodal involvement.
• Metastatic (clinical benefit inferior to CM, partly due to lower PD-L1 positivity and fewer TILs):
  • Nivolumab + ipilimumab preferred IO regimen.
  • Pooled analysis (D'Angelo JCO 2017): nivo + ipi ORR ~37%, mPFS ~5.9 mo, better than single-agent anti-PD-1 (~23%). Confirms IO doublet preference; still less effective than in cutaneous melanoma (mOS ~72 mo in CheckMate 067).
  • c-KIT mutant: imatinib. In a 25-patient phase 2, imatinib 400 mg daily gave ~21% confirmed response (all in mutations, not amplifications); median time to progression was ~3.7 mo for the whole cohort (not response duration); disease control was far higher with KIT mutations (~77%) than amplification alone (~18%), and some KIT-mutant responders had durable responses, median OS ~12.5 mo.
  • BRAFV600+: BRAF-MEK inhibitors.

Acral lentiginous (technically cutaneous, but distinct)

• Sites: palms, soles, subungual.
• A higher proportion of melanomas in Black and Asian persons (because other subtypes are rarer); absolute incidence is low and similar across races (~1.8 per million person-years).
• c-KIT more common than in other cutaneous subtypes.
• Treatment: surgery; advanced disease with IO plus targeted therapy (c-KIT, BRAF if applicable).

High-yield non-cutaneous melanoma pearls

• Uveal: liver-tropic, GNAQ/GNA11 initiating, BAP1 (class 2) high-risk; tebentafusp improves OS in HLA-A*02:01-positive metastatic disease (benefit can be independent of shrinkage).
• Mucosal: KIT-enriched, low mutational burden, aggressive; nivo + ipi preferred.
• Acral: palms/soles/nails; most common subtype in Black and Asian persons (absolute incidence similar across races); KIT-enriched.
• IO is less effective in all non-cutaneous melanomas than in cutaneous melanoma.
• DecisionDx-UM gene-expression profile (class 1 vs 2) tailors uveal surveillance.