The Lethal Thin Melanoma — Wolf in Sheep’s Clothing

Although the survival rate associated with thin melanomas exceeds 90%, about 15% of annual reported deaths in the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database involve thin melanomas. This mortality rate is largely explained by the high number of thin melanomas currently diagnosed in the U.S. and highlights the need for effective identification of high-risk thin melanomas, especially if adjuvant therapy is to be considered.

To create a classification system that predicts thin melanoma risk better than the current American Joint Commission on Cancer (AJCC) staging system, researchers at the University of Pennsylvania examined records on 26,291 melanomas in the SEER database and proposed new parameters that they validated in 2389 melanomas seen by their Pigmented Lesion Group. The AJCC parameters — age, sex, anatomic site, and tumor characteristics (level, thickness, and ulceration) — were available for all patients. The researchers looked at four additional candidate prognostic factors: radical (RGP) or vertical (VGP) growth phase; presence of RGP regression; presence of VGP mitogenicity; and presence of tumor-infiltrating lymphocytes (TILs). They developed a statistical model called a prognostic tree to sequentially divide patients into classes, with each tree node representing an “if-then” stratification decision (e.g., presence or absence of RGP).

Receiver-operating characteristic (ROC) curve analysis showed that the new model performed significantly better than AJCC staging (mean difference in area under the curve, 0.06). For example, in the new system, a man presenting with a 0.81-mm nonulcerated mitogenic melanoma would be assigned to a class with an estimated 10-year survival rate of 90.6% (comparable to stage 1B in AJCC). Under the AJCC system, the same patient would be classified as stage IA or T1a, with an expected 10-year survival rate of 97.4%.