Overall, the prognosis for thin lesions of melanoma (less than 0.76 mm
) is excellent. However, a number of melanoma patients with seemingly
innocuous lesions have been reported to develop recurrences. For this
reason, we examined histologic sections taken from eight cases of Clar
k level II melanoma that unexpectedly recurred, and compared their his
topathologic features with an equal number of nonrecurrent lesions in
whom reliable clinical follow-up data were available. Prognostic varia
bles including Breslow thickness, mitotic rate, ulceration, the presen
ce or absence of regression, a vertical growth phase component, and an
associated banal nevus were evaluated in a double blind manner. When
attempts were made to predict outcome based on one or more prognostic
variables, the only correlation of statistical significance was the Br
eslow thickness (P = 0.04). A Breslow thickness greater than 0.4 mm wa
s associated with a significantly shorter disease free interval than a
thickness below 0.4 mm. There was no significant correlation between
predicted outcome based on the histologic features examined and the ev
entual outcome based on history of recurrence (P = 0.36). These data i
ndicate that although prognostic models that predict outcome in melano
ma are generally reliable, there is a sizable population of patients w
ith thin melanomas that do worse than would be expected.