Background: Dermoscopy (dermatoscopy, epiluminescence microscopy) is increasingly employed for the preoperative detection of cutaneous melanoma; dermoscopic features of pigmented skin lesions have been previously defined using histopathology as the key to the code. In a preliminary study on 10 cases evaluated by nine dermoscopists and nine histopathologists, the authors experienced that when at least two dermoscopists disagree in evaluating a melanocytic lesion, even histopathologic consultations may give equivocal results.
Methods: One hundred seven melanocytic skin lesions, consecutively excised because of equivocal clinical and/or dermoscopic features, were retrospectively examined by eight dermoscopists and eight histopathologists; the diagnostic interobserver agreement was calculated by means of the Schouten k statistics. After histopathologic consultations, all 107 lesions underwent unblinded dermoscopic re-evaluation in order to find which dermoscopic features had given rise to histopathologic diagnostic difficulties.
Results: The interobserver ageement was good for both dermoscopy (k = 0.53) and histopathology (k = 0.74). Out of 48 cases evaluated by the dermoscopists in complete accordance, only 8 (16.7%) received at least one conflicting histopathologic diagnosis. Instead, among the remaining 59 cases with at least one disagreeing dermoscopic diagnosis, 21 (35.6%) received at least one disagreeing histopathologic diagnosis. The unblinded dermoscopic re-evaluation showed that five out of seven lesions with clear-cut regression structures were histopathologically controversial.
Conclusions: At least for selected and reasonably difficult lesions, a diagnostic discrepancy among formally trained dermoscopists seems to be predictive for a diagnostic disagreement among histopathologists. Lesions showing clear-cut regression structures are prone to give some histopathologic disagreement.
Copyright 2002 American Cancer Society.