Skip to content

Melanoma risk factors and prevention

TL;DR — Melanoma risk is dominated by two things that cannot be separated in observational data: constitutional phenotype (nevus count, skin and hair colour, freckling, family history) and ultraviolet exposure, particularly intermittent exposure and sunburn. The largest single relative risk in the classical meta-analyses is nevus count — RR 6.89 (95% CI 4.63–10.25) for 101–120 common nevi versus <15, and RR 6.36 (3.80–10.33) for five versus no atypical nevi (Gandini 2005, PMID 15617989) — and about 42% of melanomas are attributable to having ≥25 common nevi (Olsen 2010, PMID 20086181). Mendelian randomisation supports causality for childhood sunburn (OR 6.32, 4.48–8.91) and ease of tanning, and finds no causal role for coffee, alcohol, smoking or socioeconomic status (Liu 2024, PMID 37712456). Only one randomised trial has ever tested a melanoma-prevention intervention with melanoma as an endpoint: daily sunscreen in Nambour, Queensland, which produced HR 0.50 (0.24–1.02, P = .051) for any melanoma and HR 0.27 (0.08–0.97) for invasive melanoma ten years after the trial ended (Green 2011, PMID 21135266). Indoor tanning raises melanoma risk (RR 1.27, 1.16–1.39 overall; 1.75, 1.14–2.69 for early-onset disease) and its prevalence has fallen ~70% in adolescents since the 2009 IARC carcinogenicity classification (An 2021, PMID 34885049; Rodriguez-Acevedo 2020, PMID 31381131). The unresolved gap is that behavioural interventions reliably change sun-protection behaviour but have never been shown to reduce sunburn frequency, let alone melanoma (Henrikson 2018, PMID 29558557).

Constitutional risk factors

Factor Comparison Relative risk (95% CI) Source
Common nevi 101–120 vs <15 6.89 (4.63–10.25) Gandini 2005, PMID 15617989
Common nevi, per lesion each 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
Fair vs dark skin colour — 2.06 (1.68–2.52) PMID 16125929
Actinic damage indicators present vs absent 2.02 (1.24–3.29) PMID 16125929
Family history positive vs negative 1.74 (1.41–2.14) PMID 16125929
Blue vs dark eyes — 1.47 (1.28–1.69) PMID 16125929
Personal history of melanoma vs none 7.28 / 7.24 (sex-specific) Mar 2011, PMID 21605094

An umbrella review of 44 meta-analyses covering 85 associations found that most suffered from large between-study heterogeneity, small-study effects or excess significance bias; only 13 associations reached high credibility, and for melanoma these were freckle density, eye colour, hair colour, personal melanoma history, skin type, sunburns, pre-malignant lesions, and common and atypical nevi (Belbasis 2016, PMID 27663092). This is the honest boundary of the risk-factor literature: the phenotype and nevus associations survive scrutiny, most lifestyle associations do not.

Population attributable fractions convert those relative risks into burden. For ≥1 atypical nevus the PAF is 0.25; for ≥25 common nevi it is 0.42 (0.15 for 25–49 plus 0.27 for ≥50), against 0.04 for 0–10 nevi (PMID 20086181). Under varying nevus-prevalence scenarios the highest melanoma burden always sits in the high-nevus-count group (PAF range 0.31–0.62), which is the quantitative case for risk-targeted rather than universal surveillance.

MC1R and the pigmentation axis

Meta-analysis of the nine most-studied MC1R variants found seven significantly associated with melanoma: p.D84E, p.R142H, p.R151C, p.I155T, p.R160W, p.R163Q and p.D294H, with ORs from 1.42 (1.09–1.85) for p.R163Q to 2.45 (1.32–4.55) for p.I155T. p.R160W and p.D294H associate with both red hair and fair skin; p.D84E, p.R142H and p.R151C associate strongly with red hair only (ORs 2.99, 1.51–5.91 to 8.10, 5.82–11.28); p.V60L and p.V92M showed no association with either. The dissociation between variants that raise melanoma risk and variants that produce the red-hair phenotype implies MC1R acts through non-pigmentary pathways as well (Raimondi 2008, PMID 18366057). Germline architecture beyond MC1R is covered in germline predisposition.

Ultraviolet exposure — and the intermittent-exposure paradox

The classical meta-analysis of 57 studies found intermittent sun exposure and sunburn history to be substantial melanoma risk factors while high occupational (chronic) sun exposure was inversely associated with melanoma; latitude and adjustment for phenotype significantly modified the estimates, and higher-latitude studies reported higher sunburn risks (Gandini 2005, PMID 15617990). A 20-year systematic review restricted to Fitzpatrick I–IV confirmed the direction: 20 of 26 studies found a significant UV–melanoma association, with pooled unadjusted OR 1.66 (1.40–1.97) and adjusted OR 1.23 (1.04–1.46) for sunburn history across 3,417 melanomas, and cumulative exposure ORs from 1.1 (1.0–1.2) to 5.2 (2.1–12.5) (Kwa 2025, PMID 39230206).

The occupational literature remains genuinely contradictory. A systematic review of 14 studies found no increased cutaneous melanoma risk in outdoor versus indoor occupational groups (Maduka 2023, PMID 36423476), while a 1,417-patient Florentine cohort found occupational exposure associated with lentigo maligna histotype, head-and-neck site, and greater Breslow thickness at diagnosis (De Giorgi 2025, PMID 40867334). Both findings can be true simultaneously if chronic exposure shifts melanoma subtype and site rather than total incidence — which is the divergent-pathway hypothesis, and it has not been tested prospectively.

Causal inference. Two-sample Mendelian randomisation on the largest melanoma GWAS meta-analysis (30,134 cases) found elevated risk for genetic predisposition to ease of skin tanning (OR 2.842, 2.468–3.274, P < .001) and childhood sunburn (OR 6.317, 4.479–8.909, P < .001), and no significant effect for coffee intake, alcohol, lifetime smoking or socioeconomic status (Liu 2024, PMID 37712456). The authors conclude that curtailing UVR is the single best preventive strategy. A separate MR study using five 25(OH)D-associated SNPs across 12,874 cases and 23,203 controls found no causal association between vitamin D concentration and melanoma risk (OR 1.06, 0.95–1.19 per 20 nmol/L decrease; meta-analysed with UK Biobank, OR 1.02, 0.92–1.13) — which removes the main proposed benefit-side counterweight to sun avoidance in melanoma-specific terms (Liyanage 2020, PMID 31218665).

Indoor tanning

Outcome Summary RR (95% CI) Source
Melanoma, ever indoor tanning (36 studies, 14,583 cases) 1.27 (1.16–1.39) An 2021, PMID 34885049
Early-onset melanoma (<50 years) 1.75 (1.14–2.69) PMID 34885049
Melanoma, first exposure ≤20 years 1.47 (1.16–1.85) PMID 34885049
Melanoma, ≥10 sessions/year 1.52 (1.22–1.89) PMID 34885049
Melanoma (27 studies, earlier meta-analysis) 1.20 (1.08–1.34) Gandini 2019, PMID 30811691
Melanoma, exposure at younger age 1.59 (1.36–1.85) PMID 30811691
Non-melanoma skin cancer 1.40 (1.18–1.65) PMID 34885049

IARC classified artificial tanning devices as carcinogenic to humans in July 2009 (PMID 30811691). Case-control evidence also links sunbed use to ocular melanoma, with greater risk for first exposure at younger ages (PMID 30811691) — relevant to uveal melanoma.

The regulatory experiment. Countries with nationwide legislation restricting indoor tanning for those ≤18 rose from 2 in 2003 to 11 in 2011 (Pawlak 2012, PMID 22801924), and to 26 by the late 2010s (PMID 31381131). Global adolescent indoor-tanning prevalence fell from 22.0% (17.2–26.8) in 2007–2012 to 6.5% (3.3–10.6) in 2013–2018, a 70% reduction; adult prevalence fell 35%, from 18.2% to 10.4% (5.7–16.3) (PMID 31381131). Gandini's review notes that information-only measures and parental-authorisation requirements proved inefficient in several studies, while total bans (Iceland, Brazil, Australian states) remain to be formally evaluated (PMID 30811691). Whether the prevalence decline translates into a melanoma incidence decline is a cohort question that will resolve over the next two decades — and the young-cohort incidence reversals described in epidemiology and global burden are its first candidate evidence.

Primary prevention: what the trials actually show

Intervention Design Melanoma result
Daily sunscreen (Nambour, Queensland) 1,621 residents aged 25–75 randomised 1992–1996 to daily vs discretionary sunscreen (± beta-carotene), followed to 2006 11 vs 22 new primary melanomas; HR 0.50 (0.24–1.02), P = .051. Invasive melanoma 3 vs 11, HR 0.27 (0.08–0.97). Preinvasive HR 0.73 (0.29–1.81) (Green 2011, PMID 21135266)
Nicotinamide 500 mg twice daily (ONTRAC) 386 participants with ≥2 prior non-melanoma skin cancers, 12 months 23% (4–38) fewer new non-melanoma skin cancers; no melanoma endpoint; benefit did not persist after discontinuation (Chen 2015, PMID 26488693)
Behavioural counselling 21 trials, N = 20,561, USPSTF evidence review Small-to-moderate increases in sun-protection behaviour; no consistent reduction in sunburn in children (3 trials, n = 2,508) or adults (6 trials, n = 3,959); one adult trial of skin self-examination found 0 vs 1 melanoma at 12 months (Henrikson 2018, PMID 29558557)

Nambour is the load-bearing trial for melanoma prevention worldwide, and it is a single trial in one very-high-ambient-UV township whose primary melanoma result did not reach conventional significance (P = .051); only the invasive-melanoma subgroup did (PMID 21135266). No replication exists. Nicotinamide is often discussed as melanoma chemoprevention but ONTRAC did not test melanoma, and it did not include melanoma among its endpoints (PMID 26488693); its transplant-recipient successor trial addressed keratinocyte cancers (PMID 36856616).

Population campaigns are supported by economic rather than randomised evidence. SunSmart in Victoria, Australia was modelled to have averted 28,000 DALYs (≈22,000 life-years) since 1988 while saving money on skin-cancer management, with an estimated AU$2.30 returned per dollar invested and a further 120,000 DALYs projected from an upgraded 20-year national program (Shih 2009, PMID 19747936). The USPSTF's positive recommendations follow the behavioural evidence rather than a cancer endpoint: B for counselling ages 6 months to 24 years with fair skin, C for selective counselling above 24, I for counselling adults about skin self-examination (Grossman 2018, PMID 29558558).

Translating risk into surveillance

Risk-prediction tools convert the factor table into a targeting instrument.

Tool Inputs Performance
Pooled 16-study melanoma risk score Hair colour, skin type, family history, freckling, nevus count, number of large nevi, sunburn history AUC 0.75 (0.73–0.78) in an independent UK case-control dataset; 29% of cases vs 7% of controls in the top risk group (Davies 2015, PMID 25713022)
Victorian Melanoma Service calculator Common nevi, atypical nevi, hair colour, freckles, family history, personal NMSC history Population-specific 5-year absolute risk; personal melanoma history the strongest single factor (RR 7.28/7.24) (Mar 2011, PMID 21605094)
Disease risk score (23andMe cohort, 210,000 participants) 32 genetic and non-genetic factors Top-percentile DRS associated with up to 13-fold risk increase vs median; age-independent variant identified high-risk asymptomatic individuals and predicted diagnosis up to 14 years earlier (Fontanillas 2021, PMID 33420020)

These tools discriminate but none has been shown to improve an outcome when used to allocate surveillance — the same evidentiary gap that stops population screening being recommended (see screening and overdiagnosis).

Other and disputed associations

  • MC1R adds information beyond visible phenotype. In 960 cases and 396 controls, carriage of two low-risk or any high-risk MC1R variants raised melanoma risk (OR 1.7, 1.0–2.8 and OR 2.2, 1.5–3.0), but the association was stronger in people with protective phenotypes — those who tanned well (OR 2.4, 1.6–3.6), had dark hair (2.4, 1.5–3.6) or dark eyes (3.2, 1.8–5.9) — and in those reporting little sun exposure or no burns (Kanetsky 2010, PMID 20301115). Genotype therefore identifies risk in exactly the people phenotype screening would miss.
  • Childhood sunburn, again by Mendelian randomisation. A separate MR analysis found childhood sunburn causally associated with melanoma (OR 4.74, 1.31–17.19), melanoma in situ overall (4.02, 2.00–8.08), in situ of the face (18.28, 5.28–63.35) and of the trunk (7.05, 2.06–24.13), independent of hair and skin colour, facial ageing, vitamin D, BMI, alcohol and smoking (Li 2023, PMID 37962759). The very wide confidence intervals are a caution about instrument strength, not a refutation.
  • Melanoma and Parkinson's disease share something genetic. Across 157,036 participants followed 14–20 years with 616 incident Parkinson's cases, a first-degree family history of melanoma was associated with Parkinson's risk (multivariate RR 1.85, 1.2–2.8, P = .004) while family history of colorectal, lung, prostate or breast cancer was not (Gao 2009, PMID 19841380). An updated analysis of 131,342 people followed ~30 years found family history of melanoma associated with a 1.4-fold higher Parkinson's risk while hair colour was not strongly associated — implicating non-pigmentary as well as pigmentary pathways (Flores-Torres 2024, PMID 38142631). This is a genuine biological signal, not a melanoma risk factor.
  • Aspirin is a weak and unconfirmed signal. Meta-analysis of 13 observational studies found daily 50–400 mg aspirin associated with reduced skin-cancer risk (OR 0.94, 0.90–0.99, P = .02), driven by non-melanoma skin cancer (OR 0.97, 0.95–0.99); no randomised evidence exists and the authors call for trials (Zhu 2015, PMID 25663859).

Contradictions to keep visible

  • Chronic occupational UV is inversely associated with melanoma in meta-analysis (PMID 15617990) and shows no excess risk in systematic review (PMID 36423476), yet is associated with lentigo maligna, head/neck site and thicker tumours in cohort data (PMID 40867334). Any prevention message that treats "sun exposure" as one exposure is contradicted by its own evidence base.
  • Sunscreen prevented invasive melanoma in Nambour but the overall melanoma result was borderline (PMID 21135266), and the same trial's participants were in the world's highest-incidence setting; transportability to lower-UV populations is untested.
  • Behavioural interventions change behaviour without changing sunburn (PMID 29558557), which severs the assumed causal chain from counselling to melanoma.
  • The UV-attributable fraction of global melanoma is estimated at ~88% (Oh 2026, PMID 42502459) while the overdiagnosis literature holds that much of the incidence being attributed is not consequential disease (Adamson 2022, PMID 35293957) — the attributable-fraction denominator is itself contested.
  • Vitamin D provides no causal counterweight to sun avoidance for melanoma specifically (PMID 31218665), but MR addresses melanoma risk only, not the wider health consequences of reduced UV exposure.

Open questions

  • Does the divergent-pathway model (chronic exposure → lentigo maligna/head-and-neck; intermittent exposure → superficial spreading/trunk) hold prospectively, and should prevention messaging be site- and phenotype-specific (PMID 15617990; PMID 40867334)?
  • Would a sunscreen trial replicate Nambour outside a very-high-UV setting, and is such a trial still ethical or feasible (PMID 21135266)?
  • Will the 70% fall in adolescent indoor tanning since 2009 produce a measurable melanoma incidence decline, and on what lag (PMID 31381131; PMID 29105744)?
  • Can a risk score allocating surveillance improve mortality rather than only discriminate risk (PMID 25713022; PMID 33420020)?
  • Why do behavioural interventions raise sun-protection behaviour without reducing sunburn — measurement error, compensating exposure, or insufficient dose (PMID 29558557)?
  • Does nicotinamide have any melanoma-specific effect? No trial has tested it with a melanoma endpoint (PMID 26488693).

References

  1. Gandini S, et al. Meta-analysis of risk factors for cutaneous melanoma: I. Common and atypical naevi. European journal of cancer (Oxford, England : 1990). 2005;41:28-44. PMID 15617989
  2. Olsen CM, et al. Estimating the attributable fraction for cancer: A meta-analysis of nevi and melanoma. Cancer prevention research (Philadelphia, Pa.). 2010;3:233-45. PMID 20086181
  3. Liu M, et al. Analysing the causal relationship between potentially protective and risk factors and cutaneous melanoma: A Mendelian randomization study. Journal of the European Academy of Dermatology and Venereology : JEADV. 2024;38:102-111. PMID 37712456
  4. Green AC, et al. Reduced melanoma after regular sunscreen use: randomized trial follow-up. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2011;29:257-63. PMID 21135266
  5. An S, et al. Indoor Tanning and the Risk of Overall and Early-Onset Melanoma and Non-Melanoma Skin Cancer: Systematic Review and Meta-Analysis. Cancers. 2021;13. PMID 34885049
  6. Rodriguez-Acevedo AJ, et al. Indoor tanning prevalence after the International Agency for Research on Cancer statement on carcinogenicity of artificial tanning devices: systematic review and meta-analysis. The British journal of dermatology. 2020;182:849-859. PMID 31381131
  7. Henrikson NB, et al. Behavioral Counseling for Skin Cancer Prevention: Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA. 2018;319:1143-1157. PMID 29558557
  8. Gandini S, et al. Meta-analysis of risk factors for cutaneous melanoma: III. Family history, actinic damage and phenotypic factors. European journal of cancer (Oxford, England : 1990). 2005;41:2040-59. PMID 16125929
  9. Mar V, et al. Predicting melanoma risk for the Australian population. The Australasian journal of dermatology. 2011;52:109-16. PMID 21605094
  10. Belbasis L, et al. Non-genetic risk factors for cutaneous melanoma and keratinocyte skin cancers: An umbrella review of meta-analyses. Journal of dermatological science. 2016;84:330-339. PMID 27663092
  11. Raimondi S, et al. MC1R variants, melanoma and red hair color phenotype: a meta-analysis. International journal of cancer. 2008;122:2753-60. PMID 18366057
  12. Gandini S, et al. Meta-analysis of risk factors for cutaneous melanoma: II. Sun exposure. European journal of cancer (Oxford, England : 1990). 2005;41:45-60. PMID 15617990
  13. Kwa M, et al. The risk of ultraviolet exposure for melanoma in Fitzpatrick skin types I-IV: A 20-year systematic review with meta-analysis for sunburns. Journal of the European Academy of Dermatology and Venereology : JEADV. 2025;39:1239-1253. PMID 39230206
  14. Maduka RC, et al. Indoor Versus Outdoor: Does Occupational Sunlight Exposure Increase Melanoma Risk? A Systematic Review. The Journal of surgical research. 2023;283:274-281. PMID 36423476
  15. De Giorgi V, et al. Cutaneous Melanoma and Occupational UV Exposure: Associations with Anatomical Site, Histological Subtype, and Breslow Thickness. Cancers. 2025;17. PMID 40867334
  16. Liyanage UE, et al. Is there a causal relationship between vitamin D and melanoma risk? A Mendelian randomization study. The British journal of dermatology. 2020;182:97-103. PMID 31218665
  17. Gandini S, et al. Epidemiological evidence of carcinogenicity of sunbed use and of efficacy of preventive measures. Journal of the European Academy of Dermatology and Venereology : JEADV. 2019;33 Suppl 2:57-62. PMID 30811691
  18. Pawlak MT, et al. Legislation restricting access to indoor tanning throughout the world. Archives of dermatology. 2012;148:1006-12. PMID 22801924
  19. Chen AC, et al. A Phase 3 Randomized Trial of Nicotinamide for Skin-Cancer Chemoprevention. The New England journal of medicine. 2015;373:1618-26. PMID 26488693
  20. Allen NC, et al. Nicotinamide for Skin-Cancer Chemoprevention in Transplant Recipients. The New England journal of medicine. 2023;388:804-812. PMID 36856616
  21. Shih ST, et al. Economic evaluation of skin cancer prevention in Australia. Preventive medicine. 2009;49:449-53. PMID 19747936
  22. Grossman DC, et al. Behavioral Counseling to Prevent Skin Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2018;319:1134-1142. PMID 29558558
  23. Davies JR, et al. Development and validation of a melanoma risk score based on pooled data from 16 case-control studies. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2015;24:817-24. PMID 25713022
  24. Fontanillas P, et al. Disease risk scores for skin cancers. Nature communications. 2021;12:160. PMID 33420020
  25. Kanetsky PA, et al. Does MC1R genotype convey information about melanoma risk beyond risk phenotypes?. Cancer. 2010;116:2416-28. PMID 20301115
  26. Li Y, et al. Childhood sunburn and risk of melanoma and non-melanoma skin cancer: a Mendelian randomization study. Environmental science and pollution research international. 2023;30:122011-122023. PMID 37962759
  27. Gao X, et al. Family history of melanoma and Parkinson disease risk. Neurology. 2009;73:1286-91. PMID 19841380
  28. Flores-Torres MH, et al. Hair color, family history of melanoma, and the risk of Parkinson's disease: An analysis update. Parkinsonism & related disorders. 2024;119:105965. PMID 38142631
  29. Zhu Y, et al. Aspirin for the primary prevention of skin cancer: A meta-analysis. Oncology letters. 2015;9:1073-1080. PMID 25663859
  30. Oh J, et al. Global patterns of melanoma, burden attributable to ultraviolet radiation exposure, and projections to 2050 across 185 countries: A population-based study with global modeling. JAAD international. 2026;27:149-159. PMID 42502459
  31. Adamson AS, et al. Estimating Overdiagnosis of Melanoma Using Trends Among Black and White Patients in the US. JAMA dermatology. 2022;158:426-431. PMID 35293957
  32. Aitken JF, et al. Generational shift in melanoma incidence and mortality in Queensland, Australia, 1995-2014. International journal of cancer. 2018;142:1528-1535. PMID 29105744