Melanoma Risk Assessment
Patients with a personal history of melanoma or dysplastic nevi should be asked to provide information regarding a family history of melanoma and other cancers to detect the presence of familial melanoma. Age at diagnosis in family members and pathologic confirmation, if available, should also be sought. The presence of multiple primary melanomas in the same individual may also provide a clue to an underlying genetic susceptibility. Approximately 30% of affected individuals in hereditary melanoma kindreds have more than one primary melanoma, versus 4% of sporadic melanoma patients. Family histories should be updated regularly; an annual review is often recommended.
For individuals without a personal history of melanoma, several models have been suggested for prediction of melanoma risk. Data from the Nurses' Health Study were used to create a model that included gender, age, family history of melanoma, number of severe sunburns, number of moles larger than 3 mm on the limbs, and hair color. The concordance statistic for this model was 0.62 (95% CI, 0.58-0.65). Another measure of baseline melanoma risk was derived from a case-control study of individuals with and without melanoma in the Philadelphia and San Francisco areas. This model focused on gender, history of blistering sunburn, color of the complexion, number and size of moles, presence of freckling, presence of solar damage to the skin, absence of a tan, age, and geographic area of the United States. Attributable risk with this model was 86% for men and 89% for women. This predictive tool, the Melanoma Risk Assessment Tool, is available online. However, this tool was developed using a cohort of primarily white individuals without a personal or family history of melanoma or nonmelanoma skin cancer. It is designed for use by health professionals, and patients are encouraged to discuss results with their physicians. Additional external validation is appropriate before this tool can be adopted for widespread clinical use.
Two models have been developed to predict the probability of identifying germline CDKN2A mutations in individuals or families for research purposes (Table 7). MelPREDICT  uses logistic regression and MelaPRO  uses a Mendelian modeling algorithm to estimate the chance of an individual carrying a mutation in CDKN2A.
Table 7. Characteristics of Common Models for Estimating the Likelihood of a CDKN2A Mutation
|Features of Model||Incorporates three different penetrance models ||Uses logistic regression |
|Can input information for large families||Accounts for a number of primary melanomas in family and age of onset |
|Includes information for unaffected individuals on risk of developing melanoma ||�|
|Limitations||The model has not been validated on unaffected probands.||Cannot incorporate complex pedigree structure information into the model |
|�||Does not take into account domain-specific penetrances or geographical differences in penetrance |