Supplementary Material for: Validation of a dermatoscopy-based algorithm for the diagnosis of acral melanoma

Introduction:
Diagnosis of acral melanocytic lesions can be challenging. The BRAAFF checklist was introduced as a tool to help differentiate between acral nevi and melanoma but has not been validated.

Methods:
We asked raters with varying expertise in dermatoscopy to diagnose dermatoscopic images of 533 acral nevi and 144 melanomas via an online platform with and without use of the BRAAFF checklist. From the ratings we calculated sensitivity, specificity, and interrater agreement. Additionally, a new simplified version of the checklist was also tested.

Results:
We collected 6880 ratings from 175 readers. The BRAAFF checklist achieved a sensitivity of 92.5% and a specificity of 65.0%, which was similar to diagnosis from pattern recognition (sensitivity 90.0%, specificity: 72.1%). Interrater agreement for the BRAAFF criteria ranged from fair to moderate, with lowest agreement for parallel ridge and fibrillar pattern (alpha=0.31) and highest for asymmetry of colors and structures (alpha=0.46). Agreement and diagnostic accuracy were higher for more experienced readers. A simplified version with only two criteria achieved similar sensitivity (95.0%) and lower specificity (60.0%) as the original BRAAFF checklist.

Conclusion:
The BRAAFF checklist is a useful tool for the diagnosis of melanocytic acral lesions with acceptable sensitivity and reasonable specificity but is not superior to pattern recognition. A simplified version of the checklist could be easier to use with equal sensitivity while exhibiting a modest reduction in specificity.