Integrating static and dynamic features of melanoma: the DynaMel algorithm

Abstract

Background: Sequential digital dermatoscopy identifies dynamic changes in melanocytic lesions. However, no algorithm exists that systematically weights dynamic changes regarding their association with melanoma.

Objective: We sought to identify relevant dynamic changes and to integrate these into a novel diagnostic algorithm.

Methods: During follow-up (mean 44.28 months) of 688 patients at high risk, 675 pigmented lesions with prospectively documented dynamic changes were excised. The association between specific changes and melanoma was assessed.

Results: We detected 61 melanomas (38 invasive, median thickness 0.42 mm) with dynamic changes. Multivariate logistic regression analyses revealed a significant association between the diagnosis of melanoma and 5 dynamic criteria. According to the observed odds ratios we defined two dynamic major criteria (2 points each: asymmetric-multifocal enlargement and architectural change) and 3 dynamic minor criteria (1 point each: focal increase in pigmentation, focal decrease in pigmentation, and overall decrease in pigmentation when not accompanied by a lighter pigmentation of the adjacent skin). The DynaMel score was generated by addition of dynamic and 7-point checklist scores with a threshold for excision of 3 or more points. Including information about dynamic changes increased the sensitivity of the 7-point checklist from 47.5% (29 of 61 melanomas detected) to 77.1% (47 of 61 melanomas detected). The specificity slightly decreased from 99.0% to 98.1%.

Limitations: Before broad application the DynaMel algorithm needs to be validated using data from a different prospective study.

Conclusions: The DynaMel algorithm integrates a scoring system for dynamic dermatoscopic changes into the 7-point checklist for dermatoscopy and thereby increased the sensitivity of melanoma detection.