A comparison of cause-specific melanoma mortality and all-cause mortality in survival analyses after radiation treatment for uveal melanoma

Abstract

Objective To determine the causes and patterns of mortality after uveal melanoma radiation. Design A cohort study from a single institution was performed. Mortality was modeled using semiparametric survival techniques. All cause and cause-specific mortality analyses were performed. Mortality was compared with expected mortality from the U.S. census data. Participants A total of 731 patients were studied, and 710 (97%) of these had medium or large melanomas. The mean tumor diameter was 11.3 mm, and the mean tumor thickness was 5.8 mm. Ciliary body was involved in 122 (17%) of patients. Complete follow-up was available on 99.6% (728 of 731) of patients. Main outcome measures The authors analyzed the distribution and causes of post-treatment mortality. Results The 5- and 10-year all-cause Kaplan-Meier survival rates were 75.6% and 62.3%, respectively. Both melanoma risk factors (older age, ciliary body involvement, and larger tumor diameter) and nonmelanoma risk factors (older age and medical condition) were significant prognostic factors of all-cause mortality. Deaths from nonmelanoma causes accounted for 91 (42.3%) of 215 deaths. The 5-year and 10-year estimates of nonmelanoma deaths were 8.3% and 15.9%, respectively. Nonmelanoma mortality was similar to that observed in the general U.S. population (91 observed, 98.1 expected). Melanoma metastases accounted for 124 (57.7%) of 215 deaths. The 5- and 10-year estimates for probability of metastatic death were 16.1% and 21.8%, respectively. The largest tumor diameter was the best predictor for melanoma mortality; ciliary body involvement, older age, and distance from the fovea also were significant in multivariate analyses. Conclusion A significant proportion of patients with uveal melanoma die of nonmelanoma causes after radiation. In analyzing prognostic factors, considerable information may be lost if analyses are based on all-cause mortality rather than cause-specific mortality.