Primary cutaneous melanoma. Identification of prognostic groups and estimation of individual prognosis for 5093 patients

Background. Numerous investigations have examined prognostic factors for patients with primary cutaneous melanoma. However, only a few studies have been published on the definition of prognostic groups. The first aim of the present study was to determine the relative importance of different prognostic factors in a large collective study. The second aim was to define prognostic groups of patients based on combinations of prognostic factors and to define a model that allows the estimation of individual survival probability. Methods. Long term follow‐up of 5264 patients with invasive primary cutaneous melanoma was performed from 1970 to 1988 at four German University Departments of Dermatology (Berlin‐Steglitz, Münster‐Hornheide, Tübingen, and Würzburg). The multivariate Cox model was used to analyze 5093 patients, and 4371 patients with complete information were included in a classification and regression tree analysis (CART). Results. Tumor thickness, sex, anatomic location, and level of invasion were highly significant prognostic factors according to the multivariate analysis (P < 0.0001). However, histologic subtype and age influenced prognosis less significantly (P < 0.05). The CART analysis resulted in 12 groups defined mainly by tumor thickness, sex, and anatomic location, which were combined i five prognostic groups. The prognostic stratification defined by the five groups was superior compared with standard TNM model. Ten‐year survival rates of the five groups ranged from 97% to 14% (P < 0.0001), and an equation was used to calculate individual survival probabilities based on the significant factors of the Cox model. Conclusions. Consideration of all significant prognostic factors of patients with primary cutaneous melanoma investigated in the present study allows for definition of prognostic groups with a more reliable estimation of prognosis than by previous staging systems and also enables calculation of individual survival probabilities.