Objectives: To describe the dermoscopic structures and patterns associated with solitary angiokeratomas and to determine the sensitivity, specificity, positive predictive value, negative predictive value, and reproducibility of these dermoscopic features.
Design: Multicenter retrospective study.
Setting: University hospitals in Spain, Italy, Argentina, New York City, and Austria.
Patients: There were 256 patients total, and 32 specimens each of solitary angiokeratomas, melanocytic nevi, Spitz-Reed nevi, malignant melanomas, pigmented basal cell carcinomas, dermatofibromas, seborrheic keratoses, and other vascular lesions (19 angiomas, 7 pyogenic granulomas, 3 spider nevi, 2 lymphangiomas, and 1 venous lake) were consecutively collected from the laboratories of 8 hospitals. Diagnoses of all patients' lesions were confirmed histopathologically.
Intervention: Dermoscopic examination.
Main outcome measures: The frequency, sensitivity, specificity, positive predictive value, negative predictive value, intraobserver agreement, and interobserver agreement of the different dermoscopic features associated with solitary angiokeratomas were calculated, and the differences were evaluated using the chi(2) or Fisher exact test.
Results: Six dermoscopic structures were evident in at least 50% of the solitary angiokeratomas: dark lacunae (94%), whitish veil (91%), erythema (69%), peripheral erythema (53%), red lacunae (53%), and hemorrhagic crusts (53%). Dark lacunae exhibited a sensitivity of 93.8% and a specificity of 99.1% (P<.001 for both), not being found in malignant melanomas or pigmented basal cell carcinomas. The positive predictive value was 93.8%, and the negative predictive value was 99.1%. The intraobserver agreement was perfect (kappa, 1.00), and the interobserver agreement was excellent (kappa range, 0.83-1.00) (P<.001 for both). Pattern 1, consisting of dark lacunae and whitish veil, exhibited a sensitivity of 84.4% and a specificity of 99.1% and was not found in malignant melanomas or pigmented basal cell carcinomas. The positive predictive value was 93.1%, the negative predictive value was 97.8%, the intraobserver agreement was perfect (kappa, 1.00), and the interobserver agreement was excellent (kappa range, 0.83-1.00) (P<.001 for all). Conclusion Dermoscopy is helpful in improving the diagnostic accuracy of solitary angiokeratomas and allows the observer to differentiate them from other cutaneous tumors such as malignant melanomas and pigmented basal cell carcinomas.