Diagnostic evaluation of pigmented cutaneous lesions and factors used in clinical diagnosis assessment

Introduction. The necessity of excising melanomas characterized by a slight thickness at an early stage, leads dermatol to remove pigmented lesions wh ich do not correspond to melanomas. The aims of this study were: a) to pros pectively assess the accuracy of melanoma diagnosis, bf to quantify the num ber of excisions performed according to the degree of melanoma suspicion, c ) to determine the specific clinical sign or signs of relevant diagnostic v alue. Patients and methods. This study was conducted prospectively from January 1 996 to August 1997 by dermatologists in private practice and dermatologists from a University Hospital staff. When it was decided to excise a pigmente d lesion, a form was filled out choosing the most appropriate clinical diag nosis, the degree of melanoma suspicion, and clinical signs which lead to s urgery. Based on histological findings as the reference, the sensitivity, s pecificity, accuracy of melanoma diagnosis and the kappa test that evaluate s the concordance between clinical and histological diagnosis, were perform ed. The diagnostic Value of clinical signs was assessed by variance analysi s. Results. Of the 353 excised lesions, 38 (10,7 p. 100) were identified as me lanoma on histologic examination. The sensitivity, the specificity and diag nostic accuracy were: 79 p. too, 94 p, too and 53p too respectively. The ka ppa test concordance between clinical and histological diagnosis was 0.66. Two hundred and two lesions (57 p. too) were excised even though the clinic al suspicion of melanoma was poorly considered. Only one of these 202 legio ns was identified histologically as a true melanoma. Thirty seven (21 ,5 p 100) of the 151 remaining excised lesions with an "average" or "strong" sus picion were true melanomas. The clinical signs considered, alone or associa ted, had a poor predictive positive value < 38 p. loo). An analytical appro ach performed with a logistic model permitted the identification of two ass ociated signs suggesting a best diagnostic value. Discussion. This is the only study, to our knowledge, reported in the liter ature which prospectively assesses the sensitivity, specificity and concord ance between clinical and histological diagnosis of melanoma. Results were considered from average to good. The originality of this study was to asses s the number of pigmented lesions excised according to the degree of melano ma suspicion, suggesting the possibility of reducing the number of nevi rem oved when the melanoma risk was considered clinically poor. Finally,this st udy emphasizes the limits of clinical semiology and the need for future dia gnostic methods in the assessment of melanoma.