Importance: The University of Southern California Eye Physics plaques compare favorably with the Collaborative Ocular Melanoma Study plaques in terms of late adverse effects from radiation, metastasis, and local tumor recurrence.
Objective: To review the University of Southern California experience using Eye Physics plaques and Plaque Simulator software to treat choroidal melanomas and compare the outcomes with published results of the Collaborative Ocular Melanoma Study.
Design, setting, and participants: A retrospective case series of 82 patients treated for medium-sized choroidal melanoma from January 1, 1990, through December 30, 2010, using iodine 125 plaques and treatment simulation software developed at the University of Southern California. The dosimetric goal was 85 Gy in 7 days to a conformal volume enclosing the apex and a 2-mm margin surrounding the tumor base. Plaque localization was guided by the Plaque Simulator computer modeling system using preoperative imaging studies.
Main outcomes and measures: Primary outcome measures were local tumor control, globe preservation, and metastases. Secondary outcome measures were late radiation adverse effects including postoperative vision changes, optic neuropathy, radiation retinopathy, and cataract.
Results: The median follow-up for 82 patients was 46.8 months (range, 1-171 months). Globe preservation was achieved in 80 patients (97.6%); 2 patients underwent enucleation for local recurrence. Metastatic disease developed in 9 patients (11.0%). Retinopathy was seen in 31 patients (37.8%), optic neuropathy in 12 (14.6%), and cataracts in 26 (31.7%). Postoperatively, 21 patients (25.6%) lost more than 6 lines of Snellen visual acuity.
Conclusions and relevance: When considering rates of local recurrence, metastases, and late radiation adverse effects, the University of Southern California results for medium-sized choroidal melanomas using Eye Physics plaques compared favorably with Collaborative Ocular Melanoma Study data. The Plaque Simulator 3-dimensional tumor-modeling program developed at the University of Southern California is a reliable method for determining plaque positioning preoperatively and for treating this cohort of patients.