What validated models integrate genetic and clinical risk factors into an overall measurement of high risk from new primary melanoma?
Melanoma risk factors such as skin pigmentation, naevus number and genetic loci are not independent of each other. Optimal clinical risk assessment needs a combination of these measurements that most reliably discriminates people with a high likelihood of future melanoma from those at lower risk. Such measures could inform and motivate preventive behaviours and provide a basis for targeted interventions to improve early detection in the population.
Systematic review evidence
Vuong et al (2014) and Usher-Smith et al (2014) conducted systematic reviews of 28 and 25, respectively, multivariable risk prediction models for incident primary melanoma reported to 2013, and concluded they achieved fair to very good discrimination (AUROC). For example, Vuong et al (2014) assessed 19 eligible studies, which yielded two to 13 predictors; the most common were the presence of nevi, skin type, freckle density, age, hair colour and sunburn history. Only four studies in the two reviews had included genetic factors. Very few studies validated performance in an external dataset and calibration performance was only reported in two studies. Most base studies had used case-control design and therefore have a moderately high risk of bias. Three studies identified high risk individuals using absolute risk cutoffs, which are likely to have greater intelligibility for patients and clinical utility than relative risks. However, relative risks can also be important for targeting sun protection interventions towards younger people at high relative risk, but low absolute risk.
The systematic review conducted for this guidelines process identified a further nine eligible studies that reported discrimination, six of them reporting calibration. Three of these studies included genetic factors. Three studies conducted substantial external validation, including in cohort studies, however genetic factors were only assessed via family history. Discrimination was, in general, high; the models validated externally and were well calibrated. Australian data have been extensively used to generate and validate the models and these outcomes are therefore highly suitable to inform Australian clinical practice.
One limitation in the evidence is that very few studies have externally validated the effect of introduction of measured genetic factors on risk discrimination. Two Australian studies, one measuring genotypes at MC1R and other melanoma susceptibility SNPs, and these modestly improved the discrimination and calibration of a base clinical model. A second limitation is that the lists of clinical risk factors studied and validated may not yet be complete, and further factors may improve future models. Finally, there is a need for suitable on-line tools to support melanoma risk assessment using these better-performing, systematic techniques (see Melanoma risk calculator).
In summary, there is high level evidence that integrated assessment of personal risk factors for cutaneous melanoma, whether self-measured or clinically assessed, stratifies the population by future likelihood of melanoma more reliably than less systematic methods. Data are emerging that measured genetic risk can improve the performance of these models, but this requires further validation.
Evidence summary and recommendations
|Integrated assessment of personal risk factors for cutaneous melanoma, whether self-measured or clinically assessed, effectively stratifies the population by future likelihood of melanoma.||III-3||, , , , |
|Assess all patients for future risk of melanoma, using validated risk factors and a model that integrates personal risk factors into an overall index of risk.||B|
- Vuong K, McGeechan K, Armstrong BK, Cust AE. Risk prediction models for incident primary cutaneous melanoma: a systematic review. JAMA Dermatol 2014 Apr;150(4):434-44 Available from: http://www.ncbi.nlm.nih.gov/pubmed/24522401.
- Usher-Smith JA, Emery J, Kassianos AP, Walter FM. Risk prediction models for melanoma: a systematic review. Cancer Epidemiol Biomarkers Prev 2014 Aug;23(8):1450-63 Available from: http://www.ncbi.nlm.nih.gov/pubmed/24895414.
- Whiteman DC, Green AC. A risk prediction tool for melanoma? Cancer Epidemiol Biomarkers Prev 2005 Apr;14(4):761-3 Available from: http://www.ncbi.nlm.nih.gov/pubmed/15824139.
- Fears TR, Guerry D 4th, Pfeiffer RM, Sagebiel RW, Elder DE, Halpern A, et al. Identifying individuals at high risk of melanoma: a practical predictor of absolute risk. J Clin Oncol 2006 Aug 1;24(22):3590-6 Available from: http://www.ncbi.nlm.nih.gov/pubmed/16728488.
- Mar V, Wolfe R, Kelly JW. Predicting melanoma risk for the Australian population. Australas J Dermatol 2011 May;52(2):109-16 Available from: http://www.ncbi.nlm.nih.gov/pubmed/21605094.
- Bakos L, Mastroeni S, Bonamigo RR, Melchi F, Pasquini P, Fortes C. A melanoma risk score in a Brazilian population. An Bras Dermatol 2013 Mar;88(2):226-32 Available from: http://www.ncbi.nlm.nih.gov/pubmed/23739694.
- Cust AE, Goumas C, Vuong K, Davies JR, Barrett JH, Holland EA, et al. MC1R genotype as a predictor of early-onset melanoma, compared with self-reported and physician-measured traditional risk factors: an Australian case-control-family study. BMC Cancer 2013 Sep 4;13:406 Available from: http://www.ncbi.nlm.nih.gov/pubmed/24134749.
- Davies JR, Chang YM, Bishop DT, Armstrong BK, Bataille V, Bergman W, et al. Development and validation of a melanoma risk score based on pooled data from 16 case-control studies. Cancer Epidemiol Biomarkers Prev 2015 May;24(5):817-24 Available from: http://www.ncbi.nlm.nih.gov/pubmed/25713022.
- Kypreou KP, Stefanaki I, Antonopoulou K, Karagianni F, Ntritsos G, Zaras A, et al. Prediction of Melanoma Risk in a Southern European Population Based on a Weighted Genetic Risk Score. J Invest Dermatol 2016 Mar;136(3):690-5 Available from: http://www.ncbi.nlm.nih.gov/pubmed/27015455.
- Nikolić J, Loncar-Turukalo T, Sladojević S, Marinković M, Janjić Z. Melanoma risk prediction models. Vojnosanit Pregl 2014 Aug;71(8):757-66 Available from: http://www.ncbi.nlm.nih.gov/pubmed/25181836.
- Olsen CM, Neale RE, Green AC, Webb PM, Whiteman DC, QSkin Study., et al. Independent validation of six melanoma risk prediction models. J Invest Dermatol 2015 May;135(5):1377-84 Available from: http://www.ncbi.nlm.nih.gov/pubmed/25548858.
- Penn LA, Qian M, Zhang E, Ng E, Shao Y, Berwick M, et al. Development of a melanoma risk prediction model incorporating MC1R genotype and indoor tanning exposure: impact of mole phenotype on model performance. PLoS One 2014;9(7):e101507 Available from: http://www.ncbi.nlm.nih.gov/pubmed/25003831.
- Sneyd MJ, Cameron C, Cox B. Individual risk of cutaneous melanoma in New Zealand: developing a clinical prediction aid. BMC Cancer 2014 May 22;14:359 Available from: http://www.ncbi.nlm.nih.gov/pubmed/24884419.
- Vuong K, Armstrong BK, Weiderpass E, Lund E, Adami HO, Veierod MB, et al. Development and External Validation of a Melanoma Risk Prediction Model Based on Self-assessed Risk Factors. JAMA Dermatol 2016 Aug 1;152(8):889-96 Available from: http://www.ncbi.nlm.nih.gov/pubmed/27276088.
- Cust AE, Bui M, Goumas C, et al. Contribution of MC1R Genotype and Novel Common Genomic Variants to Melanoma Risk Prediction. 23:566-7 2014.