INTEROBSERVER VARIABILITY ON THE HISTOPATHOLOGIC DIAGNOSIS OF CUTANEOUS MELANOMA AND OTHER PIGMENTED SKIN-LESIONS

Purpose: To assess the interobserver agreement on the diagnosis and cl assification of cutaneous melanoma. Materials and Methods: A set of 14 0 slides of cutaneous melanoma, including a small subset of benign pig mented skin lesions, were circulated to four experienced histopatholog ists. The kappa statistic for multiple ratings per subject was calcula ted using the method described by Fleiss. Results: The kappa value on the diagnosis of cutaneous melanoma versus benign lesions was 0.61, Th ere was some discordance on the diagnosis in 37 of 140 cases (26%). Fo r the histologic classification of cutaneous melanoma, the highest kap pa values were attained for Breslow thickness (kappa = 0.76) and prese nce of ulceration (kappa = 0.87). The agreement was generally poor for other histologic features, such as level of dermal invasion (kappa = 0.38), presence of regression (kappa = 0.22), and lymphocytic infiltra tion (kappa = 0.27). Conclusion: Our study suggests considerable disag reement among pathologists on the diagnosis of melanoma versus other p igmented lesions. Tumor thickness and presence of ulceration are the m ost reproducible histologic features of cutaneous melanoma. (C) 1996 b y American Society of Clinical Oncology.