Analysis of 945 cases of pulmonary metastatic melanoma.

From 1970 to 1990, 7564 patients with melanoma were seen at Duke University Cancer Center. Complete follow-up data were available in all patients. The estimated probability of a pulmonary metastasis developing 5, 10, or 20 years after initial diagnosis was 0.13, 0.19, and 0.30, respectively. Pulmonary metastases were documented in 945 patients (12%), these having 1-, 3-, and 5-year survival rates of 30%, 9%, and 4%, respectively. The methods of diagnosis were chest radiograph (n = 544), computed tomography (n = 157), transthoracic needle biopsy (n = 121), bronchoscopy (n = 14), thoracotomy (n = 112), and autopsy (n = 7). Evidence of advanced pulmonic spread included bilateral disease in 543 and more than two nodules in 595. Univariate predictors for early formation of pulmonary metastases (p less than 0.001) were male sex, black race, increased primary thickness (millimeters), higher Clark's level, nodular or acral lentiginous histology, location on trunk or head and neck, and regional lymph nodes positive for metastasis. Multivariate predictors of improved survival (p less than 0.001) in order of importance were complete resection of pulmonary disease, longer time for formation of metastases, treatment with chemotherapy, one or two pulmonary nodules, lymph nodes negative for metastasis lymph nodes (p less than 0.005), and histologic type (p less than 0.04). Additionally, survival in patients with one nodule and resection (n = 84) was better than in those with similar disease and no resection (n = 142 months, p less than 0.001). These data comprise the largest series to date and emphasize the importance of long-term follow-up, as well as supporting the selective use of resection for isolated pulmonary metastases, increasing the 5-year survival rate from 4% to 20%.