Melanoma

Overview and epidemiology

• Epidemiology (US 2025): ~107,000 cases (invasive), ~8,300 deaths. Cutaneous melanoma (CM) accounts for ~5% of all new cancer cases but only ~1.4% of cancer deaths because of screening, early detection, and adequate surgery of early-stage disease. Fifth most common cancer in both men and women.
• Stage at diagnosis: ~84% localized (stages I and II), ~8% regional (stage III), ~4% distant metastatic (stage IV). Most early-stage CMs are highly treatable.
• Trends: melanoma-specific mortality has dropped steeply since 2015 (~4% per year) because of advances in systemic treatment; incidence appears to have stabilized after decades of increase.
• Risk factors: UV exposure, fair skin (Fitzpatrick I/II), multiple atypical/dysplastic nevi (odds ratio ~17.4 for atypical nevi), personal history of melanoma (OR ~12.8), personal history of nonmelanoma skin cancer, family history, immunosuppression.
• Hereditary syndromes: FAMMM (CDKN2A), Cowden (PTEN), retinoblastoma (RB1), Li-Fraumeni (TP53), xeroderma pigmentosum.

Biology and genetics

• Melanocytes are neural-crest derived; they produce melanin (protects against UV-induced DNA damage) but the enzymatic steps are pro-oxidative, generating reactive oxygen species and a high somatic mutation burden. Growth-factor signals run through the Ras-Raf-Mek-Erk (MAPK) pathway.
• Progression: benign nevus → dysplastic nevus → melanoma in situ → invasive melanoma, with accumulating somatic mutations. BRAFV600E is the most frequent mutation in benign nevi; additional events (TERT promoter, then CDKN2A loss) precede invasion; BRAFV600 and NRASQ61 are mutually exclusive initiating drivers, not successive steps.
• TCGA landscape (n = 333): ~52% BRAF-mutant (~46% V600), ~28% RAS, ~14% NF1 mutations, ~31% CDKN2A loss, ~15% TP53. Four genomic subgroups: BRAF-mutant, (N)RAS-mutant, NF1-mutant, triple wild type.
  • BRAF (V600E ~70% of BRAF, plus V600K, others): in ~50% of CM; younger patients, intermittent sun exposure; targetable.
  • NRASQ61 (~15 to 20%): older patients, chronic sun damage; no direct targeted therapy.
  • NF1, c-KIT (acral, mucosal): consider imatinib in KIT-mutant disease.
  • GNAQ/GNA11 in uveal melanoma.
  • TERT promoter: frequent (escape from senescence).
• Familial melanoma (~10% of CM): CDKN2A is the strongest and best-studied factor (60-fold higher risk, penetrance 60 to 90%). BAP1 germline carriers (OR ~17.3) also at high risk (and uveal melanoma, RCC, mesothelioma). MITF E318K variant raises risk of multiple primaries or RCC. MC1R red-hair variants confer moderate risk (OR ~2.4). Offer genetic counseling if ≥3 affected relatives on one side or early-onset multiple primaries.

Histology subtypes

• Superficial spreading (~70%): trunk/extremities, intermediate Breslow.
• Nodular (~15%): early vertical growth phase; worst prognosis at a given thickness.
• Lentigo maligna: sun-damaged skin in the elderly; chronic UV.
• Acral lentiginous: palms, soles, subungual; more common in non-Caucasian patients. Often c-KIT.
• Desmoplastic: sclerotic; lymph-node spread less common; high mutational burden, immunotherapy-responsive.
• Mucosal: genital, anal, sinonasal (separate biology; see non-cutaneous melanoma note).
• Uveal: separate section (very different biology).

Staging: AJCC 8th edition (2018)

• Stage at diagnosis is the single most significant predictor of OS. SEER 5-year relative survival: localized ~100%, regional ~74%, distant ~35% (summary stage, approximate mapping to AJCC groups) (SEER 2016 to 2022).
• Key AJCC 8th changes vs 7th: retains Breslow depth and ulceration for T and nodal count (plus satellite/microsatellite/in-transit) for N; reports Breslow to the nearest 0.1 mm; reclassifies thin melanoma (T1a is <0.8 mm and nonulcerated; T1b is 0.8 to 1.0 mm, or <0.8 mm with ulceration); uses clinically occult vs clinically detected nodal terminology; adds stage IIID; and refines M with an M1d (CNS) substage and an LDH modifier ((0) normal, (1) elevated) across each M substage.
• Mitoses: mitotic rate is no longer part of AJCC 8th T staging but remains an important independent prognostic factor (as does TIL density). Mitoses can influence the decision to offer SLNB in thin lesions.

• Younger women, White patients, and immunocompetent patients fare better than older men, Black patients, and immunosuppressed patients.
• Beyond AJCC parameters: mitotic rate and TIL density (in primary, nodes, or metastases) remain important.
• DecisionDx-Melanoma (31-gene expression profile) stratifies stage I to II risk (class 1 vs 2), but Breslow thickness and SLN status remain stronger predictors; not yet routine.

Patient evaluation and workup

• Diagnosis: excisional biopsy (full-thickness, narrow margins). Prefer complete excisional biopsy; avoid superficial shave that would transect the lesion, but broad deep shave (saucerization) is an acceptable technique.
• Pathology elements: Breslow depth, ulceration, mitotic rate, regression, lymphovascular invasion, perineural invasion, growth phase.
• Clinical history: document family history of CM (invasive CM or pancreatic cancer in ≥3 members on one side), or uveal melanoma, RCC, mesothelioma; personal history of ≥3 CMs including one before age 45, regressing lesions, or nonmelanoma skin cancer. Patients diagnosed at ages 15 to 30 have a higher rate of BRAFV600E. New headaches may signal CNS metastases. In stage IV, high WBC, neutrophilia, monocytosis, low lymphocyte count, and elevated LDH portend worse outcomes.
• Sentinel lymph node biopsy (SLNB): recommended to complete regional staging for melanomas >1 mm (≥T2); discuss and consider SLNB for T1b after individualized counseling; offer to all ≥T2 lesions; consider for thinner lesions with aggressive features (high mitotic rate, ulceration, T1b), particularly in young patients. Standardized probes (technetium-99 sulfur colloid plus blue dye) give ~95% SLN identification. SLN status is an independent prognostic factor for OS.
  • MSLT-I (N = 2,001, ≥T2): immediate CLND if SLN positive vs no SLN mapping showed no significant 10-year melanoma-specific survival difference in the overall population but improved disease-free survival; SLN mapping is primarily prognostic and does not prove an overall survival benefit.
• Radiographic staging: PET/CT or contrast CT has lower sensitivity than SLN mapping (cannot detect microscopic, ≤3 mm, nodal disease). Cross-sectional imaging is controversial in low-risk stage I to II but should be considered for stage ≥IIB; add brain MRI for stage III/IV.

Localized treatment (Tis to T4, stages 0 to II)

• Wide local excision is the standard for localized disease; margins are measured clinically from the tumor edge or scar, not from histopathologic margins.

• Historical adjuvant for stage IIB/IIC: high-dose interferon alfa-2b (E1684, E1690, E1694) is now largely abandoned given toxicity, logistics, and unclear OS benefit in the checkpoint-inhibitor era.
• Adjuvant anti-PD-1 for resected stage IIB/IIC:
  • KEYNOTE-716 (Luke Lancet 2022; Long Lancet Oncol 2022 for DMFS; Luke JCO 2024 final DMFS): adjuvant pembrolizumab × 1 yr in resected stage IIB/IIC melanoma, RFS HR 0.61; DMFS HR 0.64. FDA Dec 2021 expanded adjuvant pembrolizumab to IIB/IIC (previously stage III only).
  • Adjuvant nivolumab also approved for IIB/IIC (CheckMate 76K, Kirkwood Nat Med 2023), FDA Oct 13, 2023.

Stage III: management of regional nodes

• Completion lymph node dissection (CLND) was historically standard after a positive SLN, but has been replaced by observation with nodal ultrasound surveillance:
  • MSLT-II (N = 1,939): CLND did not improve melanoma-specific OS vs surveillance ultrasound, though it improved regional disease control (CLND identifies additional nonsentinel positive nodes in ~11.5% (lymphedema 24.1% vs 6.3% with observation), at the cost of lymphedema/pain).
  • DeCOG-SLT (N = 483): also showed no OS benefit for CLND.
  • CLND is no longer standard for all SLN-positive patients; discuss risks/benefits and offer serial nodal ultrasound. Note that omitting CLND affects staging.

• Therapeutic lymph node dissection (levels 1 to 3 axilla, superficial groin, levels 2 to 5 neck) is standard for clinically detected stage III without distant disease and offers the best locoregional control.
• Adjuvant nodal radiation (ANZMTG 01.02/TROG 02.01): lowered in-field nodal relapse (HR 0.52) but did not improve OS or RFS. Reserve for select high-risk features (multiple/matted nodes, extracapsular extension).
• Isolated limb infusion/perfusion (melphalan) is a regional option for advanced locoregional extremity disease (response ~81% heated perfusion, ~43% infusion); no proven OS benefit, largely fallen out of favor.

• Pembrolizumab (KEYNOTE-054): improved RFS vs placebo (HR 0.56) and DMFS; adjuvant × 1 yr (200 mg IV q3wk × 18 or 400 mg q6wk).
• Nivolumab (CheckMate 238): improved RFS vs ipilimumab (HR 0.65); × 1 yr (240 mg q2wk or 480 mg q4wk). Included resected stage IV.
• CheckMate 238 mature (Ascierto NEJM 2025, final 9-yr analysis): 9-yr RFS 44% (nivo) vs 37% (ipi) (HR 0.76); 9-yr OS 69% vs 65% (HR 0.88, not significant). Confirms durable long-term RFS advantage of adjuvant nivo over ipi in resected stage IIIB to IV. At minimum 4-year follow-up there was no OS benefit of nivo over ipi.
• Dabrafenib + trametinib (COMBI-AD): for BRAFV600-mutant, improved RFS vs placebo (HR 0.51 at 59-month follow-up); greatest benefit in tumors with low mutational burden and/or high IFN pathway activation.
• Ipilimumab adjuvant (EORTC 18071): mostly historical; high toxicity. E1609 showed only low-dose (3 mg/kg) ipilimumab improved OS vs high-dose interferon.
• CheckMate 915: adjuvant nivolumab plus low-dose ipilimumab (1 mg/kg q6wk) did NOT improve RFS over nivolumab monotherapy; combination adjuvant IO is not standard.
• IMMUNED (resected stage IV, no evidence of disease): 1-year RFS 75% (ipi+nivo) vs 52% (nivo) vs 32% (placebo); both nivo arms superior to observation.
• Active surveillance without adjuvant therapy remains acceptable given no clear OS benefit of adjuvant anti-PD-1 monotherapy; shared decision-making.

Neoadjuvant therapy (new standard for macroscopic stage III)

• Pooled International Neoadjuvant Melanoma Consortium data (189 patients, 138 IO and 51 targeted): pathologic complete response (pCR) 37 to 47%; 2-year RFS 89% (pCR) vs 50% (no pCR). About 38% of IO patients with RECIST stable disease still had complete or near-complete pathologic response. OpACIN-neo: high IFN-gamma gene expression and high mutational load predicted pCR.
• NADINA (Blank et al, NEJM 2024; PMID 38828984): neoadjuvant nivolumab + ipilimumab × 2 → surgery → response-adapted: adjuvant omitted for major pathologic response (≤10% viable tumor); non-responders get adjuvant nivolumab, or dabrafenib + trametinib if BRAF V600-mutant. 12-mo EFS 83.7% vs 57.2% (adjuvant nivo); 24-mo EFS 77.3% vs 55.7% (HR 0.32); pCR/near-pCR ~60% skip adjuvant therapy. NCCN category 1 standard for resectable macroscopic stage III (no specific FDA approval).
• SWOG S1801 (Patel NEJM 2023; 3-yr update ASCO 2024): neoadjuvant pembrolizumab × 3 → surgery → adjuvant pembrolizumab × 15 vs upfront surgery plus adjuvant. 3-yr EFS 71% vs 49% (HR 0.51); OS not reached. Cements neoadjuvant IO as standard for clinically detectable (macroscopic) resectable stage IIIB to IV disease; not studied in microscopic (SLN-only) stage III.

Stage IV: first-line metastatic

• Advanced melanoma biology: BRAFV600E and NRASQ61 are retained between primary and metastasis (basis for targeting BRAF); CDKN2A loss is more frequent in distant metastases. Intratumoral heterogeneity is an adverse prognostic factor. Intracranial and extracranial lesions are often biologically similar, explaining concordant IO responses (CheckMate 204).
• Surgery/metastasectomy: reasonable in selected oligometastatic disease and for palliation/disease control.
• Radiation: hypofractionated (≥2.5 Gy/fraction) for symptomatic lesions; SRS for limited brain metastases. Avoid concurrent MAPK inhibitors with RT (worsens toxicity).

• Historical: HDIL2 (approved 1998; ~6% CR, ~10% PR, durable in a fraction) and single-agent ipilimumab (MDX-020-010) have been supplanted by anti-PD-1. Ipilimumab monotherapy is not recommended first line.
• PD-1 monotherapy: pembrolizumab (KEYNOTE-002 after ipilimumab; KEYNOTE-006 frontline vs ipilimumab; dosed 200 mg q3wk or 400 mg q6wk) and nivolumab (CheckMate 037, CheckMate 066 vs dacarbazine; 240 mg q2wk or 480 mg q4wk).
• Nivolumab + ipilimumab (CheckMate 067):
  • Efficacy: mOS 72.1 vs 36.9 mo (nivo) vs 19.9 mo (ipi); 6.5-yr OS 49% vs 42% vs 23%. Response rate 58% vs 45% vs 19%. Powered to compare combination vs ipilimumab alone, not the two nivolumab arms.
  • Toxicity: grade 3 to 4 immune-related AEs ~55 to 60% (endocrine, GI, hepatic, pneumonitis).
• Nivolumab + relatlimab (RELATIVITY-047) (anti-PD-1 + anti-LAG-3, fixed dose nivolumab 480 mg + relatlimab 160 mg q4wk):
  • Efficacy: mPFS 10.1 vs 4.6 mo (nivo alone) (HR 0.75); confirmed ORR 43.1% vs 32.6%.
  • Toxicity: grade 3 to 4 TRAEs ~18.9% (vs ~59% with ipi/nivo).
  • FDA approval: Mar 18, 2022 for unresectable/metastatic melanoma in adults and children ≥12 yr (Tawbi NEJM 2022; PMID 34986285).
  • RELATIVITY-047 mature 4-yr update (Long ASCO 2024): mPFS 10.2 vs 4.6 mo preserved; mOS numerically higher for rela+nivo but not formally significant; safety advantage confirmed. Positioning: rela+nivo preferred over nivo mono when an IO doublet with less toxicity is desired; ipi/nivo (CheckMate 067) retained for high-risk (brain mets, elevated LDH, mucosal, high tumor burden).
• Pembrolizumab or nivolumab monotherapy: option for less aggressive disease or patients who cannot tolerate a combination.

• Concurrent BRAF + MEK inhibition is superior to single-agent BRAF inhibitor across RR, PFS, and OS (COMBI-d, COMBI-v, coBRIM). Encorafenib is the most potent (longest target dissociation half-life).
  • Dabrafenib + trametinib (COMBI-d/v): mPFS ~11 mo; 5-yr OS ~34%, ~19% progression-free at 5 yr.
  • Encorafenib + binimetinib (COLUMBUS): mPFS 14.9 mo, mOS 33.6 mo.
  • Vemurafenib + cobimetinib (coBRIM): older option.
• Triplet: atezolizumab + vemurafenib + cobimetinib (IMspire150) improved PFS (15.1 vs 10.6 mo) and is FDA-approved, but not widely adopted; COMBI-i and KEYNOTE-022 triplets were negative for PFS.
• When to start BRAF-MEK: symptomatic, rapidly progressive, or large-volume disease (faster response than IO). Otherwise IO first.

• DREAMseq (ECOG-ACRIN EA6134): first-line nivo+ipi → BRAF-MEK at progression vs the reverse. 2-yr OS 72% (IO first) vs 52% (BRAF-MEK first); BRAF-MEK looked better early but IO overtook at ~10 months. IO-first is the preferred sequence for BRAFV600-mutant metastatic disease.
• MAPK-inhibitor rechallenge after a treatment interval yields response rates ~42%. Durable off-treatment responses can occur after stopping PD-1 (KEYNOTE-001).

Refractory / later-line

• Lifileucel (Amtagvi), autologous tumor-infiltrating lymphocyte (TIL) therapy:
  • FDA approval: Feb 16, 2024, the first FDA-approved TIL (tumor-infiltrating lymphocyte) therapy (sipuleucel-T, 2010, preceded it as a cellular immunotherapy for a solid tumor). C-144-01 (Chesney JITC 2022; PMID 36600653): ORR 31.4% in heavily pretreated metastatic melanoma (post-PD-1, and post-BRAF-MEK if applicable).
  • Process: tumor harvest → TIL expansion → preconditioning chemo (cyclophosphamide + fludarabine) → TIL infusion → IL-2 boluses.
  • Toxicity: cytopenias (lymphodepletion), capillary leak from IL-2.
  • TILVANCE-301 (phase 3, NCT05727904): lifileucel + pembrolizumab vs pembrolizumab in first-line unresectable/metastatic melanoma; confirmatory trial for potential first-line TIL approval; readout awaited 2026 to 2027.
• T-VEC (talimogene laherparepvec): intralesional oncolytic HSV-1 (deleted ICP34.5, expresses GM-CSF); FDA-approved for unresectable stage IIIB to IV cutaneous/subcutaneous/nodal disease (durable response ~19.3% vs 1.4% GM-CSF). MASTERKEY-265 (T-VEC + pembrolizumab) did not improve PFS over pembrolizumab.
• Vusolimogene oderparepvec (RP1, Tudriqev): intratumoral oncolytic HSV-1 given with nivolumab; FDA accelerated approval Aug 6, 2026 for unresectable advanced cutaneous melanoma progressing on a PD-1-based regimen (IGNYTE: ORR ~33% (CR ~15%), durable responses (median DOR >21 mo; 14.1 mo was the lower 95% CI bound)).
• Chemotherapy (dacarbazine, temozolomide): salvage only; temozolomide penetrates the CNS better and gives similar response to dacarbazine (~13.5% vs 12.1%) with a small PFS benefit (1.9 vs 1.5 mo, P=.012) but no OS advantage.
• Other targeted:
  • NRAS-mutant: binimetinib (NEMO) prolonged PFS vs dacarbazine (2.8 vs 1.5 mo) but not clinically meaningful; no OS benefit.
  • c-KIT mutant (acral, mucosal): imatinib (responses in mutations, not amplifications).

Toxicity of systemic therapy

• Serious TRAEs are far higher with ipi+nivo than nivo alone (~55.4% vs 9.9%). Immune toxicities are generally reversible with corticosteroids; grade 3 warrants suspension and grade 4 (except controlled endocrinopathy) permanent discontinuation. Steroid-refractory colitis: infliximab or vedolizumab; steroid-refractory hepatitis: mycophenolate. Endocrinopathies need hormone replacement. Fatal toxicities: neurologic (Guillain-Barre, myasthenia, encephalitis, transverse myelitis), pneumonitis, myocarditis.
• Discontinuation of ipi/nivo for toxicity does not clearly reduce efficacy. Anti-PD-1 rechallenge after high-grade toxicity carries ~30 to 50% recurrent-toxicity risk.
• BRAF/MEK class effects: pyrexia (dabrafenib/trametinib, >50%, median onset ~19 days), diarrhea/transaminitis/photosensitivity (vemurafenib/cobimetinib), vomiting/retinal detachment (encorafenib/binimetinib). Watch decreased LVEF, hypertension, and QTc prolongation. Avoid concurrent MAPK inhibitor with radiation.

Brain metastases

• Common (~50% of stage IV develop CNS disease); 80% of brain metastases are hemispheric.
• Nivolumab + ipilimumab (CheckMate 204): intracranial clinical benefit ~57% in asymptomatic patients with lesions <3 cm, higher than ipilimumab (~24%) or pembrolizumab (~22%) monotherapy. Preferred upfront for small, asymptomatic parenchymal metastases not requiring urgent local therapy, with close monitoring.
• BRAF-MEK has CNS activity (COMBI-MB) but less durable intracranial responses than ipi/nivo; brain relapses occur earlier.
• SRS for limited brain metastases; WBRT reserved for extensive disease/palliation (adjuvant WBRT after local therapy does not improve OS and worsens neurocognition). Clinical benefit is poor for leptomeningeal or symptomatic parenchymal disease regardless of systemic agent. Multidisciplinary care with surgery/SRS for symptomatic lesions.

Surveillance and survivorship

• History and skin/lymph-node exam every 3 to 6 months for the first 3 to 5 years, then annually for life; lifelong annual skin screening. About half of recurrences are in the regional nodal bed, ~20% locoregional, ~30% distant, most within the first 2 years.
• Imaging surveillance for stage IIB and higher (PET/CT or CT/MRI).
• Survivors carry a 1 to 8% lifetime risk of a second invasive melanoma and higher rates of nonmelanoma skin cancer; lymphedema and financial/psychosocial burden are common.
• Older adults: melanoma tends to be thicker with fewer positive SLNs and higher UV-associated mutation burden, with enrichment of BRAFV600K and NRAS mutations (the UV signature itself is C>T at dipyrimidines, including CC>TT); responses are at least as good as in younger patients, but severe AEs are more frequent. Use geriatric assessment to guide combination vs single-agent IO.

High-yield melanoma pearls

• Excisional biopsy for suspected melanoma (avoid superficial shave; deep saucerization is acceptable).
• Breslow depth is the most important prognostic factor in localized melanoma (overall prognosis is driven by stage, incorporating nodal and distant disease); T1a is <0.8 mm and nonulcerated.
• SLNB offered for ≥T2 (>1 mm); discuss and consider for T1b (0.8 to 1.0 mm, or ulcerated) and for thin lesions with high mitotic rate.
• MSLT-II / DeCOG-SLT: observation with nodal ultrasound acceptable after a positive SLN.
• NADINA: neoadjuvant nivo + ipi for macroscopic stage III (new standard; omit adjuvant if major pathologic response, ≤10% viable tumor).
• SWOG S1801: neoadjuvant then adjuvant pembrolizumab beats upfront surgery.
• CheckMate 067: nivo + ipi first-line metastatic, mOS 72.1 mo.
• RELATIVITY-047: nivo + relatlimab, less toxic IO combination.
• DREAMseq: IO first, then BRAF-MEK at progression.
• BRAF-MEK doublets (dabrafenib + trametinib, encorafenib + binimetinib, vemurafenib + cobimetinib): no head-to-head OS comparison; choose by toxicity profile.
• Lifileucel (Amtagvi): first FDA-approved TIL therapy (Feb 2024) for refractory disease.
• Brain metastases: nivo + ipi (CheckMate 204) for small asymptomatic lesions, or SRS; BRAF-MEK less durable.