PREDICTION OF HISTOLOGIC MELANOCYTIC DYSPLASIA FROM CLINICAL OBSERVATION

Background: The dysplastic melanocytic nevus (DMN) is a putative precu rsor for malignant melanoma in persons with a family history of melano ma and multiple large atypical moles. Furthermore, the concept that DM N confers increased risk for melanoma has been extended to those witho ut a family history of melanoma. Some investigators have characterized DMN in terms of size, surface topography, and irregularity of border and color and have provided definitive statements as to major and mino r clinical ''criteria'' for DMN. Objective. In this study we conducted a quantitative test of previously described clinical characteristics in patients diagnosed with melanoma but without a family history of me lanoma in a first-degree relative (N = 153). Methods. Two of us examin ed each patient independently, counting the total number of nevi, deci ding on the most clinically atypical nevus, and recording quantitative ly the size, surface, border, and color characteristics. The most atyp ical nevus was removed from each patient and diagnosed histologically without knowledge of the clinical examiner's description. Results. Of the 153 nevi so examined, 91 (59.5%) were judged to be clinically atyp ical and 23 nevi (15%) were classified as histologically dysplastic. W e then analyzed the ability of each clinical feature to predict histol ogic melanocytic dysplasia. A multivariate analysis was conducted with the blindly scored clinical features as independent variables. Indepe ndent variables included total number of nevi and freckles as well as the features of the particular nevus biopsied (including longest diame ter, macular component, irregular border, ill-defined border, haphazar d coloration). Of the many independent clinical variables, only total number of nevi and macular component were useful in the final multivar iate prediction. Of 21 nevi with a macular component removed from pers ons with more than 24 total body nevi, 7 were histologically dysplasti c, representing a positive predictive value of 33.3% (95% exact confid ence limits 14.6% to 57.0%). Of the 38 nevi without a macular componen t to their most atypical nevus removed from patients with 12 or fewer total body nevi, 37 failed to meet histologic criteria for DMN, repres enting a negative predictive value of 97% (95% exact confidence limits 86.2% to 99.9%). Conclusion: We conclude that in nonfamilial melanoma , clinical criteria suggesting DMN when present will often fail to dis play histologic melanocytic dysplasia. In contrast, absence of a macul ar component in melanocytic nevi in a person with fewer than 13 total body nevi will accurately predict the absence of melanocytic dysplasia on histologic examination.