Risk prediction models for melanoma: a systematic review

Area of Science:

• Dermatology
• Epidemiology
• Biostatistics

Background:

• Melanoma incidence is rising globally, particularly in fair-skinned populations.
• Early diagnosis significantly improves patient prognosis.
• Risk prediction models can aid in targeted screening and early detection strategies.

Purpose of the Study:

• To systematically review and analyze existing risk prediction models for cutaneous melanoma.
• To identify commonly used risk factors and assess the performance of these models.

Main Methods:

• Systematic literature search of Medline, EMBASE, and Cochrane Library for primary research studies.
• Inclusion of studies reporting or validating melanoma risk prediction models.
• Analysis of 25 distinct risk models, examining 144 potential risk factors.

Main Results:

• Commonly included risk factors were number of nevi, freckles, sunburn history, hair color, and skin color.
• Most models demonstrated similar predictive performance, with a summary ROC area under the curve (AUROC) of 0.755.
• Only two models were validated in independent populations, both showing good discrimination (AUROC 0.79 and 0.70).

Conclusions:

• Existing melanoma risk prediction models show comparable performance.
• Further research should prioritize the validation of established models over the development of new ones.
• Validated models are essential for effective melanoma screening and early detection efforts.