Statistics — melanoma¶
Last curated: 2026-09-01. Every figure below carries its source, year, population and method. Conflicting estimates are shown side by side and are never averaged. All PMIDs were live-resolved from PubMed E-utilities on 2026-09-01.
Global burden¶
| Measure | Estimate | Population / year | Method | Source |
|---|---|---|---|---|
| New cases worldwide | 325,000 (174,000 M / 151,000 F) | 2020 | GLOBOCAN 2020 | Arnold 2022, PMID 35353115 |
| Deaths worldwide | 57,000 (32,000 M / 25,000 F) | 2020 | GLOBOCAN 2020 | PMID 35353115 |
| New cases worldwide | 331,722 | 2022 | GLOBOCAN 2022 | Wang 2025, PMID 39682020 |
| Deaths worldwide | 58,667 | 2022 | GLOBOCAN 2022 | PMID 39682020 |
| New cases worldwide | ~330,000 | 2022 | GLOBOCAN + CI5plus + WHO mortality | Oh 2026, PMID 42502459 |
| Projected cases | 510,000 (≈+50%) | 2040, if 2020 rates hold | rate-fixed projection | PMID 35353115 |
| Projected deaths | 96,000 (+68%) | 2040, if 2020 rates hold | rate-fixed projection | PMID 35353115 |
| Age-standardised incidence | 3.4 per 100,000 | worldwide, 2020 | CI5 / NORDCAN / SEER / IARC synthesis | Huang 2023, PMID 37296344 |
| Age-standardised mortality | 0.55 per 100,000 | worldwide, 2020 | same | PMID 37296344 |
| Rank among cancers | 17th most common | worldwide, 2022 | GLOBOCAN | PMID 39682020 |
| Share of cases attributable to UV radiation | ~88% (5th-percentile reference population) | worldwide, 2022 | population attributable fraction | PMID 42502459 |
| Non-melanoma skin cancer deaths, for comparison | 69,416 | worldwide, 2022 | GLOBOCAN | PMID 39682020 |
| Total cancer cases worldwide | 20.6 million | 2024 | GLOBOCAN 2024 | Sung 2026, PMID 42417444 |
Conflict to preserve: GLOBOCAN 2020 gives 325,000 cases and GLOBOCAN 2022 gives 331,722; these are different estimation vintages, not a measured 2% increase.
Geographic gradient (age-standardised incidence per 100,000 person-years, 2020)¶
| Region | Male | Female | Source |
|---|---|---|---|
| Australia / New Zealand | 42 | 31 | Arnold 2022, PMID 35353115 |
| Western Europe | 19 | 19 | PMID 35353115 |
| North America | 18 | 14 | PMID 35353115 |
| Northern Europe | 17 | 18 | PMID 35353115 |
| Most of Africa and Asia | commonly <1 | commonly <1 | PMID 35353115 |
| Queensland, Australia (invasive, 2010–2014) | 72 (both sexes, age-standardised) | — | Aitken 2018, PMID 29105744 |
| Peak mortality (New Zealand) | 5 per 100,000 | — | PMID 35353115 |
Incidence trends¶
| Population | Trend | Period | Source |
|---|---|---|---|
| US non-Hispanic White | 20.7 (20.5–20.9) → 28.2 (28.0–28.5) per 100,000; APC 3.90% (2.36–5.48) 2001–2005, then 1.68% (1.37–1.99) | 2001–2015 | Thrift 2020, PMID 31346623 |
| Europe, invasive (18 registries, 117 M people, ~415,000 lesions) | AAPC 4.0% (M) / 3.0% (F) | 1995–2012 | Sacchetto 2018, PMID 29395684 |
| Europe, in situ | AAPC 7.7% (M) / 6.2% (F) | 1995–2012 | PMID 29395684 |
| Europe, thin invasive | AAPC 10% (M) / 8.3% (F) | 1995–2012 | PMID 29395684 |
| Italy (21 registries, 15.8 M people) | 3.6%/yr (3.2–4.0) M; 2.5%/yr (2.0–3.1) F | 1994–2013 | Bucchi 2021, PMID 33405292 |
| Olmsted County, Minnesota (n = 2,310) | 11.1-fold increase since the 1970s; 1.21-fold in the last decade | 1970–2020 | Reinhart 2024, PMID 38957842 |
| Acral melanoma, Sweden (n = 1,000) | No significant change in standardised incidence | 1996–2020 | Helkkula 2024, PMID 39140487 |
| Uveal melanoma, US (n = 5,563) | Mean age-adjusted incidence 5.6 per million (5.5–5.7); APC +0.5% in Whites | 1975–2020 | Weinberger 2025, PMID 40225965 |
Birth-cohort reversals¶
| Population | Finding | Source |
|---|---|---|
| Queensland, Australia | Age-specific invasive incidence falling under 40 from cohorts born ~mid-1960s, steepest for those born ~1980+ | Aitken 2018, PMID 29105744 |
| New South Wales, Australia | Invasive incidence stable or falling under 60, rising ≥60 especially in men | Blazek 2022, PMID 36174452 |
| US non-Hispanic White | 1991 cohort IRR 0.85 (0.77–0.94) vs 1956 cohort | Thrift 2020, PMID 31346623 |
| Sweden (34,800 melanomas, <60 years) | Incidence peaked 2013–2015 in ages 20–49 then stable or declining; mortality fell in ages 30–59 but not ≥60 | Helgadottir 2024, PMID 39245436 |
| Canada | Incidence falling in females <30 and males <40; cohorts born 1993–2007 lower than baby-boom reference | O'Sullivan 2026, PMID 41932245 |
| Italy | Cohort IRR rose to the 1973 (F) / 1975 (M) cohorts then declined toward the 1949 reference | Bucchi 2021, PMID 33405292 |
| Iceland | Invasive world-standardised rate peaked 2002–2006 then fell; melanoma mortality falling since 2012 | Thomas 2025, PMID 39444324 |
Mortality trends¶
| Population | Trend | Period | Source |
|---|---|---|---|
| US, all | 2.7 → 2.0 per 100,000 (−1.3%/yr); 184,416 deaths | 1999–2020 | Didier 2024, PMID 38956559 |
| US non-Hispanic White | −6.1%/yr after 2013 | 2013–2020 | PMID 38956559 |
| US White, SEER-9 | +7.5% overall 1986–2013, then −17.9% 2013–2016 (APC −6.2%, −8.7 to −3.7) | 1986–2016 | Berk-Krauss 2020, PMID 32191523 |
| US White men ≥50 | APC −8.3% (−12.2 to −4.1) from 2014 | 2014–2016 | PMID 32191523 |
| Australia | Mortality down 40–50% over recent decades | to 2010s | De Pinto 2024, PMID 38391175 |
| Germany vs 9 neighbouring countries (pooled APC) | −1.8% (−2.3 to −1.4) vs −2.2% (−2.8 to −1.6); P = .42 | 2009–2022 | Hübner 2026, PMID 42268621 |
Survival¶
| Endpoint | Value | Population / era | Source |
|---|---|---|---|
| US 5-year relative survival, metastatic melanoma | 16% → 35% | mid-1990s → 2015–2021 | Siegel 2026, PMID 41528114 |
| US 1-year net survival, distant stage | ~43% (stable 2001–2010) → 58.9% (56.6–61.2) | CONCORD-3, diagnosed 2013 | Di Carlo 2020, PMID 33409455 |
| Share of US melanomas diagnosed at distant stage | 4.4% of 425,915 | 2001–2013 | PMID 33409455 |
| Melanoma-specific mortality, Olmsted County | 26.7% (1970s) → 1.5% (2010s) | 50-year cohort | Reinhart 2024, PMID 38957842 |
| 5-year relative survival, cutaneous | 89% | SEER 1988–2010, n = 219,890 | Bishop 2014, PMID 24272143 |
| 5-year relative survival, ocular | 78% | SEER 1988–2010, n = 7,069 | PMID 24272143 |
| 5-year relative survival, mucosal | 34% aggregate (range 3–69% by site) | SEER 1988–2010, n = 2,755 | PMID 24272143 |
| 5-year relative survival, uveal | 82.8%, unchanged | SEER 1975–2016 | Weinberger 2025, PMID 40225965 |
| 5-year disease-specific survival, acral lentiginous | 77.8% (75.9–79.9) | SEER 2000–2016, n = 2,245 | Yan 2022, PMID 34363907 |
| 15-year melanoma-specific survival after melanoma in situ | 98.4% (98.3–98.5) | SEER 2000–2018, n = 137,872 | Patel 2023, PMID 37285145 |
| 15-year relative survival after melanoma in situ | 112.4% (112.0–112.8) | same | PMID 37285145 |
| All-cause SMR after melanoma in situ | 0.68 (0.67–0.70) | same | PMID 37285145 |
| Melanoma-specific SMR after melanoma in situ | 1.89 (1.77–2.02) | same | PMID 37285145 |
Stage-specific survival (AJCC 8th edition)¶
| Stage | 5-year MSS, Ontario 2007–2012 (n = 6,414) | 5-year DSS, Japan (n = 3,097) | 3-year OS, SEER-18 2010–2015 (n = 126,408) |
|---|---|---|---|
| I | 98.4% | IA 97.9% / IB 96.2% | IA 97% / IB 95% |
| II | 82.5% | IIA 94.1% / IIB 84.4% / IIC 72.2% | IIA 87% / IIB 76% / IIC 57% |
| III | 66.4% | IIIA 87.5% / IIIB 72.6% / IIIC 55.3% / IIID 26.0% | IIIA 86% / IIIB 69% / IIIC 56% / IIID 30% |
| IV | 14.4% | — | 24% |
Sources: Hynes 2022, PMID 35778691; Fujisawa 2019, PMID 31023613; Tjokrowidjaja 2022, PMID 34811927. These are three different populations, eras and endpoints and are shown side by side rather than pooled. Note that stage IIC is worse than stage IIIA in all three.
Overdiagnosis estimates¶
| Estimate | Population | Method | Source |
|---|---|---|---|
| 59% (95% CI 45–70) of White women; 60% (32–75) of White men | US, diagnosed 2014 | Black-patient mortality trends as marker for care improvement | Adamson 2022, PMID 35293957 |
| 29–60% across three studies | international scoping review of 35 studies | various | Bjørch 2024, PMID 37793786 |
| >60–80% of melanoma in situ in some settings | narrative review with systematic search elements | various | Greco 2026, PMID 42279200 |
| 6.9 extra melanomas per 1,000 additional biopsies (3.1–10.8), confined to in situ and local disease | SEER, ≥65 years, 1986–2001 | ecological, biopsy-rate linked | Welch 2005, PMID 16081427 |
| Melanoma listed among cancers where population data suggest overdiagnosis | cross-cancer synthesis | narrative | Welch 2010, PMID 20413742 |
Counter-estimates and caveats. Incidence rose in almost every stratum defined by thickness/stage and small-area socioeconomic status in 58,000+ California melanomas, with the largest relative increases in regional, distant and ulcerated disease among low-SES men (Clarke 2017, PMID 28736233). County-level US incidence correlates with median household income (r = 0.43) and not with UV daily dose (r = 0.03), and correlates with melanoma mortality at only r = 0.09 against r = 0.96 for lung cancer in the same counties (Adamson 2022, PMID 36190719).
Diagnostic performance¶
| Test | Sensitivity | Specificity | Population | Source |
|---|---|---|---|---|
| Dermoscopy vs naked eye (relative diagnostic OR) | 15.6 (2.9–83.7); 9.0 (1.5–54.6) excluding two outliers | — | 9 clinical-setting studies | Vestergaard 2008, PMID 18616769 |
| Dermoscopy, in-person vs image-based (RDOR) | 4.6 (2.4–9.0) favouring in-person | — | 103 cohorts, 42,788 lesions | Dinnes 2018, PMID 30521682 |
| RCM vs dermoscopy, any suspicious lesion (at fixed 90% sensitivity) | — | 82% vs 42% | 9 RCM datasets, 1,452 lesions | Dinnes 2018, PMID 30521681 |
| RCM vs dermoscopy, equivocal lesions | — | 86% vs 49% | 7 RCM datasets, 1,177 lesions | PMID 30521681 |
| Dermoscopy for amelanotic/hypomelanotic melanoma | 61% (37–81) | 90% (74–97) | 7 studies, 1,111 lesions | Lan 2020, PMID 31747045 |
| RCM for amelanotic/hypomelanotic melanoma | 67% (51–81) | 89% (86–92) | same | PMID 31747045 |
| Deep learning, pooled | 82% (77–86) | 87% (84–90); AUC 0.92 (0.89–0.94) | 27 pooled studies | Ye 2024, PMID 39088883 |
| CNN vs 58 dermatologists | dermatologists 86.6% ±9.3 | CNN 82.5% vs dermatologists 71.3% at matched sensitivity | 100-image test set | Haenssle 2018, PMID 29846502 |
| Nodal ultrasound surveillance | 0.879 (0.878–0.879) | 0.969 (0.968–0.970); DOR 224.5 | 36 studies, 18,273 patients | Liu 2024, PMID 38635019 |
| FDG-PET for melanoma metastases | 0.79 (0.66–0.93) | 0.86 (0.78–0.95); DOR 33.1 | 6 poolable studies | Mijnhout 2001, PMID 11301402 |
| 7-point checklist, prospective | 62% | 97% | 688 high-risk patients, 10 years | Haenssle 2010, PMID 20226567 |
| 7-point checklist, retrospective (for contrast) | 78–95% | 65–87% | literature | PMID 20226567 |
Diagnostic reproducibility¶
| Interpretation class | Intraobserver reproducibility | Accuracy vs expert consensus | Source |
|---|---|---|---|
| Nevus / mild atypia | 76.7% | 92% (90–94) | Elmore 2017, PMID 28659278 |
| Moderate atypia | 35.2% | 25% (22–28) | PMID 28659278 |
| Severe atypia / melanoma in situ | 59.5% | 40% (37–44) | PMID 28659278 |
| pT1a invasive melanoma | 63.2% | 43% (39–46) | PMID 28659278 |
| ≥pT1b invasive melanoma | 82.6% | 72% (69–75) | PMID 28659278 |
| Population-level verification by expert panel | 82.8% (81.0–84.5) verified; 8.0% overinterpreted; 9.2% underinterpreted | — | PMID 28659278 |
| Misclassification, general pathologists no second opinion | 52.8% (51.3–54.3) | — | Piepkorn 2019, PMID 31603483 |
| Misclassification, universal dermatopathologist second opinions | 36.7% (33.1–40.7) | — | PMID 31603483 |
Number needed to excise / biopsy¶
| Setting | NNE / NNB | Source |
|---|---|---|
| Single German academic department, 118,668 pigmented lesions, 2004–2013 | 17.2 | Schreieder 2024, PMID 39682200 |
| US academic centre, 2,643 biopsies in one year (PPV 6.4%, 5.5–7.4) | ~16 | Soltani-Arabshahi 2015, PMID 25582536 |
| MoleMap NZ teledermoscopy, 1,571 lesions with pathology | 6 (benign:malignant 5.0:1) | Greenwald 2021, PMID 32114083 |
| Referral centre with adjunctive RCM (randomised) | 3.0 vs 5.3 without (−43.4%) | Pellacani 2022, PMID 35648432 |
Risk factors — relative risks¶
| Factor | Comparison | RR / OR (95% CI) | Source |
|---|---|---|---|
| Common nevi | 101–120 vs <15 | 6.89 (4.63–10.25) | Gandini 2005, PMID 15617989 |
| Common nevi | per additional nevus | 1.017 (1.014–1.020) | Olsen 2010, PMID 20086181 |
| Atypical nevi | 5 vs 0 | 6.36 (3.80–10.33) | PMID 15617989 |
| Atypical nevi | ≥1 vs 0 | 3.63 (2.85–4.62) | PMID 20086181 |
| Pre-malignant / skin-cancer lesions | present vs absent | 4.28 (2.80–6.55) | Gandini 2005, PMID 16125929 |
| Red vs dark hair | — | 3.64 (2.56–5.37) | PMID 16125929 |
| Freckle density | high vs low | 2.10 (1.80–2.45) | PMID 16125929 |
| Skin type I vs IV | — | 2.09 (1.67–2.58) | PMID 16125929 |
| Family history | positive vs negative | 1.74 (1.41–2.14) | PMID 16125929 |
| Personal history of melanoma | vs none | 7.28 / 7.24 (sex-specific) | Mar 2011, PMID 21605094 |
| Indoor tanning | ever vs never | 1.27 (1.16–1.39) | An 2021, PMID 34885049 |
| Indoor tanning, early-onset melanoma | ever vs never | 1.75 (1.14–2.69) | PMID 34885049 |
| Indoor tanning (earlier meta-analysis, 27 studies) | ever vs never | 1.20 (1.08–1.34) | Gandini 2019, PMID 30811691 |
| Sunburn history (Fitzpatrick I–IV) | adjusted | 1.23 (1.04–1.46) | Kwa 2025, PMID 39230206 |
| Childhood sunburn (Mendelian randomisation) | genetic liability | OR 6.317 (4.479–8.909) | Liu 2024, PMID 37712456 |
| Ease of skin tanning (Mendelian randomisation) | genetic liability | OR 2.842 (2.468–3.274) | PMID 37712456 |
| Vitamin D (Mendelian randomisation) | per 20 nmol/L decrease | 1.06 (0.95–1.19); meta 1.02 (0.92–1.13) — null | Liyanage 2020, PMID 31218665 |
Population attributable fractions: ≥1 atypical nevus 0.25; ≥25 common nevi 0.42; 0–10 nevi 0.04 (Olsen 2010, PMID 20086181).
Germline prevalence¶
| Measure | Value | Population | Source |
|---|---|---|---|
| Combined pathogenic-variant prevalence, 8 familial melanoma genes | 0.5% (Geisinger MyCode) to 0.9% (UK Biobank) | 696,665 genomically ascertained individuals | Goldstein 2026, PMID 42201696 |
| Above the 2.5% testing threshold | only in multiple primaries or first melanoma <40 years | same | PMID 42201696 |
| CDKN2A share of familial melanoma | ~40% | high-density pedigrees | Robles-Espinoza 2014, PMID 24686849 |
| CDKN2A p.Arg112dup carriers, pancreatic cancer RR | 43.8 (13.8–139.0) | Swedish founder cohort | Helgadottir 2014, PMID 24935963 |
| CDKN2A p16-affecting variants, families with pancreatic cancer | 58% (95/163) vs 0% of p14ARF-only families (0/9) | 172 Dutch families, 649 carriers | Overbeek 2021, PMID 32482799 |
| CDKN2A carriers, cumulative PDAC incidence to age 70 | 20.7% | 347 carriers, 20-year prospective surveillance | Klatte 2022, PMID 35658523 |
| Heritable retinoblastoma survivors, 50-year cumulative melanoma incidence | 4.5% | 1,020 heritable survivors | Kleinerman 2021, PMID 34153328 |
| Non-heritable retinoblastoma survivors, same | 0.7% | 831 non-heritable survivors | PMID 34153328 |
| Heritable retinoblastoma survivors, any-SMN SIR | 11.9 (10.4–13.5) | 1,128 heritable survivors | Schonfeld 2021, PMID 33473166 |
| Non-heritable retinoblastoma survivors, any-SMN SIR | 0.8 (0.5–1.2) | 924 non-heritable survivors | PMID 33473166 |
Driver mutation frequencies¶
| Population | BRAF | NRAS | Other | Source |
|---|---|---|---|---|
| German routine care (n = 217) | 40.1% | 24.4% | 2.3% concurrent; 33.2% wild-type for tested exons | Heppt 2017, PMID 28797232 |
| Japanese cutaneous (n = 37) | 76% | 8% | — | Hida 2024, PMID 39564955 |
| Japanese acral (n = 52) | 9% | 17% | KRAS 8%, KIT 19%, NF1 7% | PMID 39564955 |
| KIT, pooled 32 studies (n = 5,224) | — | — | 9.5% (497/5,224) | Gong 2018, PMID 29746316 |
| TERT promoter, sporadic melanoma | — | — | 74% of metastatic cell lines, 85% of metastatic tissues, 33% of primaries | Horn 2013, PMID 23348503 |
| GNAQ/GNA11, uveal melanoma | — | — | 83% carry a mutation in one | Van Raamsdonk 2010, PMID 21083380 |
| TERT promoter, intermediate lesions and melanoma in situ | — | — | 77% | Shain 2015, PMID 26559571 |
Treatment effect sizes — headline randomised results¶
| Trial | Comparison | Result | Source |
|---|---|---|---|
| CheckMate 067 (10-year) | nivo+ipi vs nivo vs ipi | median OS 71.9 / 36.9 / 19.9 months; HR 0.53 (0.44–0.65) and 0.63 (0.52–0.76) vs ipi | Wolchok 2025, PMID 39282897 |
| KEYNOTE-006 | pembrolizumab vs ipilimumab | 24-month OS 55% / 55% / 43%; HR 0.68 both schedules | Schachter 2017, PMID 28822576 |
| Ipilimumab pooled (n = 1,861) | — | median OS 11.4 months; plateau at 22% from ~year 3 | Schadendorf 2015, PMID 25667295 |
| RELATIVITY-047 (4-year) | nivo+relatlimab vs nivo | 4-year PFS 30.6% vs 23.6%; OS 52.0% vs 42.8% | Lipson 2025, PMID 40513285 |
| DREAMseq | immunotherapy first vs targeted first | 2-year OS 71.8% vs 51.5%, P = .010 | Atkins 2023, PMID 36166727 |
| COLUMBUS (7-year) | enco+bini vs vemurafenib | 7-year PFS 21.2% vs 6.4%; OS 27.4% vs 18.2% | Schadendorf 2024, PMID 38723373 |
| COMBI-d/v pooled (5-year) | dabrafenib+trametinib | 5-year PFS 19%, OS 34% | Robert 2019, PMID 31166680 |
| EORTC 18071 | adjuvant ipilimumab vs placebo | OS HR 0.73 (0.60–0.89), P = .002; 8.7% absolute difference at 7 years | Eggermont 2019, PMID 31400634 |
| COMBI-AD (final) | adjuvant dab+tram vs placebo | OS HR 0.80 (0.62–1.01), P = .06; RFS HR 0.52 (0.43–0.63) | Long 2024, PMID 38899716 |
| KEYNOTE-054 | adjuvant pembrolizumab vs placebo | 3.5-year DMFS 65.3% vs 49.4%, HR 0.60 (0.49–0.73) | Eggermont 2021, PMID 33857412 |
| KEYNOTE-716 (52.8 months) | adjuvant pembrolizumab, stage IIB/C | RFS HR 0.62 (0.50–0.78); DMFS HR 0.59 (0.45–0.77); PRFS2 HR 0.75 (0.56–1.01) | Luke 2025, PMID 40198940 |
| CheckMate 76K | adjuvant nivolumab, stage IIB/C | RFS HR 0.42 (0.30–0.59); 12-month RFS 89.0% vs 79.4% | Kirkwood 2023, PMID 37845511 |
| SWOG S1801 | neoadjuvant-adjuvant vs adjuvant pembrolizumab | 2-year EFS 72% vs 49%, P = .004 | Patel 2023, PMID 36856617 |
| NADINA | neoadjuvant ipi+nivo vs adjuvant nivo | 12-month EFS 83.7% vs 57.2%; HR 0.32 (0.15–0.66) | Blank 2024, PMID 38828984 |
| MSLT-II | completion dissection vs observation | 3-year MSS 86% vs 86%, P = .42; lymphoedema 24.1% | Faries 2017, PMID 28591523 |
| DeCOG-SLT | completion dissection vs observation | 5-year DMFS 64.9% vs 67.6% | Leiter 2019, PMID 31557067 |
| Swedish margins trial (19.6 years) | 2 cm vs 4 cm | melanoma-specific HR 0.95 (0.78–1.16), P = .61 | Utjés 2019, PMID 31280965 |
| UK margins trial (8.8 years) | 1 cm vs 3 cm | melanoma-specific HR 1.24 (1.01–1.53), P = .041 | Hayes 2016, PMID 26790922 |
| Margins meta-analysis (7 RCTs, n = 4,579) | narrow vs wide | all endpoints non-significant; melanoma death RR 1.11 (0.96–1.28) | Hanna 2021, PMID 33722422 |
| TIL vs ipilimumab | phase 3, 86% anti-PD-1 refractory | median PFS 7.2 vs 3.1 months, HR 0.50 (0.35–0.72); ORR 49% vs 21% | Rohaan 2022, PMID 36477031 |
| Tebentafusp (5-year) | vs investigator's choice, uveal | median OS 21.6 vs 16.9 months, HR 0.67 (0.54–0.85); 5-year OS 16% vs 8% | Piperno-Neumann 2026, PMID 42162665 |
| ABC (7-year) | ipi+nivo vs nivo, brain metastases | intracranial response 51% vs 20%; 7-year OS 48% vs 26% | Long 2025, PMID 39978375 |
| WBRT-Mel | adjuvant whole-brain RT vs observation | distant intracranial failure at 12 months 42.0% vs 50.5%, P = .22 | Hong 2019, PMID 31553661 |
| Nambour sunscreen trial | daily vs discretionary sunscreen | melanoma HR 0.50 (0.24–1.02), P = .051; invasive HR 0.27 (0.08–0.97) | Green 2011, PMID 21135266 |
| MELFO | reduced vs conventional follow-up | recurrence HR 0.87 (0.54–1.39), P = .57; DFS HR 1.00 | Moncrieff 2022, PMID 35866644 |
| German population screening (ecological) | Germany vs 9 neighbours | pooled APC −1.8% vs −2.2%, P = .42 | Hübner 2026, PMID 42268621 |
Toxicity¶
| Regimen / measure | Value | Source |
|---|---|---|
| Grade 3/4 TRAEs, adjuvant nivolumab (stage III/IV) | 14.4% | Weber 2017, PMID 28891423 |
| Grade 3/4 TRAEs, adjuvant ipilimumab 10 mg/kg | 45.9% | PMID 28891423 |
| Grade 3/4 TRAEs, adjuvant nivolumab (stage IIB/C) | 10.3% (vs 2.3% placebo) | Kirkwood 2023, PMID 37845511 |
| Grade 3/4 TRAEs, nivolumab + relatlimab | 18.9% (vs 9.7% nivolumab) | Tawbi 2022, PMID 34986285 |
| Grade 3–4 immune-related events, neoadjuvant ipi 3 + nivo 1 | 40% | Rozeman 2019, PMID 31160251 |
| Grade 3–4 immune-related events, neoadjuvant ipi 1 + nivo 3 | 20% | PMID 31160251 |
| Grade 3–4 immune-related events, neoadjuvant sequential ipi→nivo | 50% (accrual closed early) | PMID 31160251 |
| Any-grade colitis (pooled RCTs) | 14.5% | Da 2019, PMID 32082164 |
| Any-grade hypothyroidism | 13.8% | PMID 32082164 |
| Any-grade hepatitis | 10.4% | PMID 32082164 |
| Any-grade hypophysitis | 10.0% | PMID 32082164 |
| Any-grade pneumonitis | 4.6% | PMID 32082164 |
| Colitis risk, combination vs monotherapy | RR 3.56 (1.56–8.12) | PMID 32082164 |
| Fatal checkpoint toxicity reports (global) | 613 events, 2009–Jan 2018; anti-CTLA-4 deaths 70% colitis | Wang 2018, PMID 30242316 |
| Checkpoint myotoxicity at 6 months | 0.7–0.9% of 172,363 treated adults | Salem 2026, PMID 40884033 |
| Chronic events persisting >12 weeks after adjuvant anti-PD-1 | 43.2% of 387 patients; 96.4% grade 1–2, most unresolved | Patrinely 2021, PMID 33764387 |
| Endocrine toxicity overall | 25–50% of recipients depending on regimen | Wright 2023, PMID 36481794 |
| Lymphoedema after completion dissection | 24.1% | Faries 2017, PMID 28591523 |
| Transplant recipients: acute rejection at 1 year on checkpoint inhibitors | 36.2% (30.7–41.7) | Saleem 2025, PMID 40545616 |
| Transplant recipients: graft loss at 1 year | 18.4% (13.7–23.1) | PMID 40545616 |
| Transplant recipients: objective response at 1 year | 31.6% (25.0–37.7) | PMID 40545616 |
Economic and life-years burden¶
| Measure | Value | Population / era | Source |
|---|---|---|---|
| Adults treated for any skin cancer, annual | 6.1 million (5.6–6.6) | US, 2016–2018 | Kao 2023, PMID 36449145 |
| Annual US skin-cancer treatment cost | $8.0 bn (2012–15) → $8.9 bn (2016–18) | US, MEPS | PMID 36449145 |
| Adults with melanoma, annual | 0.70 million (0.61–0.78) | US, MEPS 2011–2020 | Olateju 2024, PMID 39612258 |
| Total annual expenditure per person with melanoma | $19,427 (vs $13,744 NMSC, $23,741 other cancers) | same | PMID 39612258 |
| Adjusted marginal expenditure, melanoma vs other cancers | −$3,369 (−$5,934 to −$804) | same | PMID 39612258 |
| Years of life lost per metastatic patient, Australia | 19.9 (M) / 22.7 (F) | 12-country model, 2014 | Thiam 2016, PMID 26531249 |
| Years of life lost per metastatic patient, US | 17.9 (M) / 20.6 (F) | same | PMID 26531249 |
| SunSmart Victoria: DALYs averted since 1988 | 28,000 (≈22,000 life-years); AU$2.30 returned per dollar | Australia, modelled | Shih 2009, PMID 19747936 |
| 3D total-body photography, incremental cost | US$945 per person (738–1,157) over 24 months, no QALY difference | randomised, Australia | Lindsay 2025, PMID 40136266 |
Known conflicts and caveats¶
- Two GLOBOCAN vintages give different global case counts (325,000 for 2020; 331,722 for 2022). They are separate estimation exercises, not a measured trend (PMID 35353115; PMID 39682020).
- Incidence is a poor proxy for occurrence in this disease. US county-level incidence correlates with melanoma mortality at r = 0.09 versus r = 0.96 for lung cancer in the same counties (PMID 36190719).
- Overdiagnosis estimates range 29–60% and rest on different comparison series with different assumptions; they are not interchangeable (PMID 35293957; PMID 37793786).
- Stage-specific survival tables from Ontario, Japan and SEER are not poolable — different eras, endpoints (MSS / DSS / OS) and follow-up windows (PMID 35778691; PMID 31023613; PMID 34811927).
- Stage IIC has worse survival than stage IIIA in every validation cohort here. This is a real finding, not a data error (PMID 31401470; PMID 31023613; PMID 31977051).
- AJCC 7 and AJCC 8 stage III figures are not comparable. Reclassification moved 44–57% of stage III patients between subgroups, improving every subgroup's apparent survival by migration alone (PMID 35778691; PMID 31111349).
- Relative survival above 100% after melanoma in situ (112.4%) indicates selection: these patients are healthier and better-surveilled than the general population (PMID 37285145).
- Prospective and retrospective diagnostic-algorithm performance differ substantially — the 7-point checklist loses ~20 sensitivity points prospectively (PMID 20226567).
- Number needed to excise is a property of the setting, ranging 3.0 to 17.2 across the four settings tabulated above.
- Registrational trial populations were overwhelmingly White and cutaneous. Anti-PD-1 objective response was 54% in White versus 20% in East Asian, Hispanic and African patients in one 1,135-patient multicentre series (PMID 35293617); acral first-line response was 15.0% versus 39.1% for cutaneous in a predominantly White Spanish registry (PMID 40841506).
- Adjuvant trial results are recurrence endpoints. Only EORTC 18071 shows a significant overall-survival benefit; COMBI-AD missed at P = .06 with 8.33 years of follow-up, and trial-level surrogacy of RFS for OS is R² = 0.59 (95% CI 0.08–1.00) (PMID 31400634; PMID 38899716; PMID 32777716).
- Neoadjuvant results rest on short follow-up — NADINA's headline figure comes from a median 9.9 months (PMID 38828984).
- Melanoma-specific FDG-PET meta-analysis remains technologically dated. The 2001 analysis searched through 1999; a second meta-analysis published in 2010 included studies only through 2006 and found scan-level specificity 0.86 (95% CI 0.77–0.92), positive likelihood ratio 5.86 (3.64–9.43) and diagnostic odds ratio 37.89 (15.80–90.86), with heterogeneity in other estimates (PMID 11301402; PMID 19727717).
- The Schonfeld 3.1–17 figure is a range across melanoma, CNS, oral-cavity and breast subsequent neoplasms, not a melanoma-specific SIR. Melanoma-specific figures come from a different cohort (PMID 33473166; PMID 34153328).