Not yet independently audited
This page was written from verified sources but has not yet passed the independent citation audit. Treat its claims as provisional.
Melanoma — master index¶
Last curated: 2026-09-01 · status: audited (authored by claude; independently audited by codex) · 26 curated pages · 636 unique PubMed records · 63 live-verified NCT identifiers · 7 landmark notes · 32-document guideline registry plus 6 web resources · 117-record statistics sheet · four-file patient-voice layer · 21 open questions with 12 dots
The condition in five sentences. Melanoma is a malignancy of melanocytes that behaves as at least four diseases — cutaneous (UV-driven, BRAF/RAS/NF1, checkpoint-responsive), uveal (GNAQ/GNA11, hepatotropic, checkpoint-refractory), acral and mucosal (structural-variant-driven, under-represented in the trials that set the standard of care) — with an estimated 325,000 new cases and 57,000 deaths worldwide in 2020 and incidence varying more than fortyfold between populations (Arnold 2022, PMID 35353115). It is the disease in which checkpoint immunotherapy first proved durable: CheckMate 067's ten-year results give median overall survival 71.9 months with nivolumab plus ipilimumab against 19.9 months with ipilimumab, and 96% ten-year melanoma-specific survival among patients progression-free at three years (Wolchok 2025, PMID 39282897). The same period produced its central controversy: incidence rose several-fold while mortality stayed flat or fell, and an estimated 59–60% of US melanoma diagnoses in White patients in 2014 may have been overdiagnosis, with county-level incidence correlating with melanoma mortality at r = 0.09 against r = 0.96 for lung cancer in the same counties (Adamson 2022, PMID 35293957; PMID 36190719). Therapy is now moving earlier — adjuvant treatment has reached node-negative stage IIB/IIC and neoadjuvant sequencing has reached phase 3 — on recurrence endpoints whose trial-level surrogacy for overall survival is only moderate (Luke 2025, PMID 40198940; Blank 2024, PMID 38828984; Coart 2020, PMID 32777716). Melanoma is simultaneously the best solved and the least well-defined common cancer.
Start here: wiki/overview.md. Research frontier: OPEN-QUESTIONS.md. Growth history: LOG.md.
Reading paths¶
The scoping decision that shapes this condition¶
A PubMed search for melanoma returns 184,619 records (checked 2026-09-01) — the second-largest literature in this repository after hypertension. As with hypertension, the failure mode is sprawl, not thinness, and the condition is therefore scoped by border rather than by breadth target.
What this condition owns. Melanoma of all primary sites: cutaneous as the spine, plus uveal, mucosal and acral as dedicated pages. The systemic-therapy story — checkpoint inhibition, targeted therapy, cellular therapy — is owned here, because melanoma is where most of it was established.
What it does not own. Keratinocyte carcinomas (basal and squamous cell) are a different disease with a different natural history and are not in this repository; melanoma pages mention them only where a shared screening or prevention trial forces it. General immune-related toxicity beyond melanoma cohorts stays out. Paediatric intraocular malignancy belongs to retinoblastoma, not to the uveal melanoma page.
Subtype scoping, and one deliberate deferral. Uveal melanoma has 8,280 records of its own, mucosal 3,476, acral 2,099 (all checked 2026-09-01). Uveal melanoma is biologically distinct from cutaneous melanoma at almost every level and is managed by ocular oncologists rather than dermatologists. It is nonetheless kept inside this condition, as a dedicated page rather than a separate condition, because the immunotherapy contrast — why checkpoint blockade transformed cutaneous melanoma and did not transform uveal melanoma — is one of the most informative things in the field, and it is lost if the two are filed apart. Uveal melanoma is recorded in CONDITIONS-ROADMAP.md as a candidate future standalone condition. If it is ever promoted, this page becomes the overlap page and must cross-link rather than duplicate.
The shared border with retinoblastoma¶
Survivors of heritable retinoblastoma carry significantly increased melanoma risk; survivors of non-heritable retinoblastoma do not. The build located the melanoma-specific figures the scoping session lacked: across 1,851 White long-term survivors diagnosed 1914–2006 and followed to 2016, melanoma occurred in 33 of 1,020 heritable and 7 of 831 non-heritable survivors, with 50-year cumulative incidence 4.5% versus 0.7% and a median age at skin-cancer diagnosis about 20 years younger in heritable survivors; sun sensitivity and phenotype did not vary by skin-cancer status, pointing to a genetic rather than exposure mechanism (Kleinerman 2021, PMID 34153328). Any-cancer figures come from a separate cohort: any-SMN SIR 11.9 (95% CI 10.4–13.5) for heritable versus 0.8 (0.5–1.2) for non-heritable survivors, and the frequently quoted 3.1–17 figure is a range across melanoma, CNS, oral-cavity and breast subsequent neoplasms, not a melanoma-specific SIR (Schonfeld 2021, PMID 33473166).
Rules for the border, held throughout this build: wiki/germline-predisposition.md and wiki/special-populations.md cross-link retinoblastoma's second-cancers-and-survivorship.md, and neither condition restates the other's evidence. The clinically load-bearing distinction is heritable versus non-heritable retinoblastoma, never "retinoblastoma survivors" as one group. Both conditions contain an intraocular malignancy — retinoblastoma in children, uveal melanoma in adults — separated by age and biology; they cross-link and never merge.
Wiki pages¶
Twenty-six canonical pages: 24 topical plus the two standing house pages. This table is the canonical page list for this condition (CONVENTIONS §5); link only to these filenames.
| # | Section | Page | Scope | Lines | Refs | Status |
|---|---|---|---|---|---|---|
| 1 | Foundations | overview.md |
What melanoma is, why subtype matters, map of the condition | 164 | 55 | curated |
| 2 | Foundations | epidemiology-and-global-burden.md |
Incidence and mortality by region, cohort reversals, 2040 projections, the incidence–mortality divergence | 179 | 36 | curated |
| 3 | Foundations | risk-factors-and-prevention.md |
UV exposure, phenotype, nevus count, sunbeds; sun protection and prevention trials | 152 | 32 | curated |
| 4 | Foundations | germline-predisposition.md |
CDKN2A, MC1R, BAP1, POT1, MITF; familial syndromes, testing thresholds, retinoblastoma border | 155 | 30 | curated |
| 5 | Mechanism | molecular-subtypes-and-genomics.md |
TCGA framework; precursor evolution; mutational signatures; resistance genomics | 173 | 29 | curated |
| 6 | Clinical | clinical-diagnosis-and-dermoscopy.md |
Dermoscopy accuracy, algorithms, confocal microscopy, AI, biopsy technique | 152 | 31 | curated |
| 7 | Clinical | histopathology-and-prognostic-factors.md |
Breslow, ulceration, mitotic rate, TILs, regression; diagnostic reproducibility | 155 | 36 | curated |
| 8 | Controversy | screening-and-overdiagnosis.md |
USPSTF I statement, overdiagnosis estimates, the German program, the incidence–mortality gap | 160 | 39 | curated |
| 9 | Clinical | staging.md |
AJCC 8th edition, stage migration, the IIC/IIIA inversion, non-cutaneous systems | 161 | 30 | curated |
| 10 | Clinical | sentinel-node-and-nodal-management.md |
MSLT-I/II, DeCOG-SLT, tumour burden, technique, the inheritance problem | 154 | 32 | curated |
| 11 | Treatment | surgical-management-and-margins.md |
Margin trials, MelMarT-II, Mohs and lentigo maligna, in-transit disease, metastasectomy | 159 | 35 | curated |
| 12 | Treatment | adjuvant-therapy.md |
Checkpoint and BRAF/MEK, stage II expansion, surrogate endpoints, absolute benefit vs toxicity | 158 | 41 | curated |
| 13 | Treatment | neoadjuvant-therapy.md |
NADINA, S1801, PRADO, pathological response, the time-bias critique | 160 | 25 | curated |
| 14 | Treatment | immunotherapy-advanced-disease.md |
CheckMate 067 10-year, relatlimab, treatment-free survival, sequencing, response patterns | 163 | 40 | curated |
| 15 | Treatment | targeted-therapy.md |
BRAF/MEK combinations, COLUMBUS 7-year, DREAMseq, resistance, pyrexia management | 151 | 37 | curated |
| 16 | Clinical | brain-metastases.md |
ABC 7-year, CheckMate 204, COMBI-MB, WBRT-Mel, radionecrosis, leptomeningeal disease | 157 | 41 | curated |
| 17 | Treatment | cellular-therapy-and-resistance.md |
TIL therapy, lifileucel, oncolytic agents, neoantigen vaccines, resistance programmes | 150 | 33 | curated |
| 18 | Safety | immune-related-adverse-events.md |
Frequency by regimen and schedule, fatal toxicity, chronic events, steroids and efficacy | 150 | 33 | curated |
| 19 | Subtype | uveal-melanoma.md |
GNAQ/GNA11/BAP1, COMS, tebentafusp, liver-directed therapy, surveillance | 156 | 41 | curated |
| 20 | Subtype | acral-and-mucosal-melanoma.md |
Distinct genomics, KIT, the efficacy gap, disparities, recognition | 162 | 45 | curated |
| 21 | Clinical | special-populations.md |
Transplant, pregnancy, paediatric, older adults, HIV, autoimmune disease, unknown primary | 169 | 53 | curated |
| 22 | Survivorship | surveillance-and-survivorship.md |
MELFO, ultrasound adherence, ctDNA, second primaries, fear of recurrence | 163 | 40 | curated |
| 23 | Reference | guidelines.md |
EADO, ESMO, NCCN, AAD, USPSTF, Japanese and Australian guidance; where they diverge | 156 | 39 | curated |
| 24 | Frontier | clinical-trials-landscape.md |
58 live-verified selected NCT records across settings; what is being tested and what is not | 178 | 27 | curated |
| 25 | Human | red-flags-and-safety-concerns.md |
Amelanotic, nodular, nail-unit and acral presentations; immunotherapy emergencies; delayed toxicity | 153 | 52 | curated |
| 26 | Human | patient-experience-and-advocacy.md |
Unmet needs, fear of recurrence, prognostic uncertainty, verified organisations | 158 | 35 | curated |
All 26 pages are status: curated. They were authored by claude and independently audited by codex on 2026-09-01.
Literature layer¶
| File | Contents | Size |
|---|---|---|
literature/BIBLIOGRAPHY.md |
Every cited paper, grouped by topic cluster, with tags and citing pages | 636 records |
literature/notes/ |
Deep notes on landmark papers | 7 notes |
literature/guidelines/REGISTRY.md |
Society guidelines worldwide, with supersession chains, disagreements, gaps and a watch list | 32 document rows + 6 verified web resources |
literature/statistics/STATISTICS.md |
Quick-reference tables with source, year, population and method; conflicts shown side by side | 117 records across 14 tables |
literature/patient-voice/ |
README (method and ethics), organizations, themes, sources | 18 verified organisations, 8 themes |
| OPEN-QUESTIONS.md | Tiered questions with stable IDs plus a 12-row "dots not yet connected" table | 21 questions, 12 dots |
Landmark notes¶
| Note | Paper | Why it is landmark |
|---|---|---|
wolchok-2025-checkmate067-10year.md |
Wolchok 2025, PMID 39282897 | The longest randomised follow-up of checkpoint blockade in any cancer |
blank-2024-nadina.md |
Blank 2024, PMID 38828984 | First phase 3 test of neoadjuvant against adjuvant therapy in melanoma |
adamson-2022-overdiagnosis.md |
Adamson 2022, PMID 35293957 | The quantitative overdiagnosis estimate that reframed the field's central controversy |
faries-2017-mslt2.md |
Faries 2017, PMID 28591523 | Removed a routine operation and orphaned the variable that justified it |
nathan-2021-tebentafusp.md |
Nathan 2021, PMID 34551229 | First therapy of any kind to improve survival in metastatic uveal melanoma |
shain-2015-precursor-evolution.md |
Shain 2015, PMID 26559571 | Ordered the genetic events of melanoma genesis and dissolved the benign–malignant boundary |
luke-2025-keynote716.md |
Luke 2025, PMID 40198940 | Extended adjuvant therapy to node-negative disease and exposed the surrogate-endpoint problem |
The five things a reader should carry away¶
- Melanoma is at least four diseases. Cutaneous, uveal, acral and mucosal differ in driver, aetiology, tropism, treatment and survival trend. Effect sizes do not transport between them.
- Incidence is a poor proxy for occurrence. County-level melanoma incidence correlates with melanoma mortality at r = 0.09; the same counties give r = 0.96 for lung cancer (PMID 36190719).
- The field de-escalates surgery on negative trials and escalates systemic therapy on surrogate endpoints. Both are defensible; they are not symmetric in evidentiary strength.
- The durable-response plateau is real and is the reference case for immuno-oncology — 96% ten-year melanoma-specific survival among patients progression-free at three years (PMID 39282897) — and it created a prognostic-uncertainty burden that patients and carers report as harm.
- The evidence base was built in White, cutaneous populations. Anti-PD-1 objective response was 54% versus 20% by ethnicity in one 1,135-patient series (PMID 35293617); first-line response was 15.0% in acral versus 39.1% in cutaneous melanoma in a predominantly White registry (PMID 40841506).
Curation state¶
| Layer | State |
|---|---|
| Wiki pages | 26 of 26 independently audited, all status: curated, 150–179 lines each, 23–55 references each |
| Bibliography | 636 unique PubMed records, all live-resolved and re-verified 2026-09-01 |
| Landmark notes | 7 |
| Guidelines registry | 32 document rows, 6 verified web resources, disagreements section, watch list |
| Statistics sheet | 117 records across 14 tables plus 14 named conflicts and caveats |
| Patient-voice layer | 4 files; 18 organisations verified live; 8 themes each with ≥2 sources |
| Open questions | 21 questions across three tiers, 12 dots, 9 seed items explicitly closed |
| Trials | 63 NCT identifiers, every one resolved live against the ClinicalTrials.gov v2 API on 2026-09-01 |
| Audit | Passed. codex independently audited material authored by claude; no substantive flags remain |
Built 2026-09-01 by claude; independently audited 2026-09-01 by codex. The audit re-fetched every identifier, checked claim–citation pairs and quoted numbers, re-ran asserted-absence searches, corrected the corpus, and promoted all pages only after the final verification pass. See LOG.md for the full audit record.