Pattern of Failure Study in Ocular Patients with Macular Melanoma using Fusion of Fundus Images in Eyeplan Software

Abstract

To study the pattern of local failure and determine if tumor localization can be improved in patients treated with proton beam radiotherapy (PBRT) for ocular melanoma using the new version of EYEPLAN software, which allows for registration of high-resolution fundus photos. The new version of EYEPLAN (V3.05a) software allows for the registration of a fundus photo so that it can be displayed as a background image for currently displayed graphics. The image is registered by clicking on the center of the fovea and optic disc in the planning fundus view. This is used in conjunction with tantalum marker rings, surgeon's mapping, ultrasound, and 3DMRI to draw the tumor contour accurately. In order to determine if the fundus image helps in tumor delineation and treatment planning, we identified 303 patients with choroidal melanoma who were treated with PBRT between 1995 and 2008. All patients were treated to a dose of 56 GyE/4 fx. The seventy-nine pts with macular lesions, which can be more difficult to outline with rings due to their posterior location, were then chosen and the treatment plans were reviewed and re-planned using EYEPLAN. For patients with local failure (LF), the pre-treatment fundus images were fused to the original treatment plan to check if the original fundus image tumor volume was covered. The process was repeated with post-treatment fundus images to record the exact location of failure with respect to the target volume and dose received. The mean follow-up of macular melanoma patients was 28 ± 21 (range 6 to 91) months. The 3-yr local control rate and overall survival was 84.5 ± 6.1% and 79.9 ± 6% respectively. Tumor growth was seen in 6 eyes. Among the 6 recurrences, 3 were managed by enucleation and 3 had distant metastasis of which 2 patients died. Re-planning these 6 patients with their original fundus photo superimposed showed that in 4 cases the treatment field adequately covered the tumor volume. The remaining 2 patients, had 3 clips placed around the tumor due to difficulty defining these posterior tumors. Overlaid fundus photos clearly indicated the area of marginal miss. Re-planning with the fundus photo showed improved tumor coverage in these macular lesions. Local control was excellent in patients receiving 56 GyE of PBRT for macular melanoma. The superposition of the fundus photo would have clearly allowed improved localization of tumor in one-third of those patients who failed. Posterior lesions are better defined with the additional use of fundus image since they can be difficult to mark at the time of clip surgery. Our current practice standard is to use the direct superimposition of the high-resolution fundus photo in addition to the surgeon's clinical and clip mapping of the tumor and ultrasound to draw the tumor volume.