Predictors of Outcomes in Iodine-125 Brachytherapy-Treated Uveal Melanomas: A Modern Institutional Study.

PURPOSE/OBJECTIVE(S) Plaque brachytherapy is an effective vision- and eye-sparing procedure for uveal melanoma. Few have investigated longitudinal tumor outcomes while using advanced dosimetry and plaque planning software. We hypothesize that patients with large tumors, pre-operative eye complications, and molecular class 2 tumors will have worse outcomes. MATERIALS/METHODS We constructed an IRB-approved retrospective database from a single academic institution with uveal melanoma patients receiving Iodine-125 brachytherapy, 85 Gy over 72 hours prescribed to the tumor apex, between 2011-2019 with a minimum of 3-month follow-up. Data on patient demographics, tumor characteristics, pre-treatment (pre-Tx) retinal complications, post-plaque treatments (PPTx), local recurrences (LR), distant metastases (DM) and overall survival (OS) were collected. Univariate (UVA) and multivariate (MVA) Cox models for OS and cumulative incidence of LR/DM and death without LR/DM as competing risk were conducted using SAS version 9.4. Hazard ratios (HRs) and corresponding P-values were estimated. RESULTS 262 patients were identified, with a median follow-up time of 33.5 months (3.02 - 97.3 months). 19 patients (7.3%) had local recurrence, 17 patients (6.5%) had distant metastasis, and 20 patients (7.6%) died. For LR, on UVA pre-Tx ocular melanocytosis (HR = 5.48, P = 0.01), Eye Physics plaque (HR = 3.70, P = 0.013), and PPTx with transpupillary thermotherapy (HR = 4.65, P = 0.002) were associated with increased LR. No association was found between class 2 tumors and LR (HR = 0.51, P = 0.297). For DM, on UVA pre-Tx retinal detachment (HR = 7.26, P = 0.002), class 2 tumors (HR = 4.35, P = 0.005), PPTx with intravitreal triamcinolone (HR = 4.01, P = 0.006), and large tumor height (HR = 1.23, P < 0.001), tumor area (HR = 1.02, P < 0.001), distance to fovea (HR = 1.13, P = 0.002), and prescription depth (HR = 1.23, P = 0.001) were all associated with increased DM. No MVA was conducted for LR and DM because there were too few events. For any tumor progression (LR or DM), on UVA pre-Tx ocular melanocytosis (HR = 4.26, P = 0.005), Eye Physics plaque (HR = 3.41, P = 0.002), any PPTx (HR = 4.50, P = 0.012), PPTx with intravitreal triamcinolone (HR = 3.30; P < 0.01), and large tumor area (HR = 1.01, P < 0.001) were associated with progression. On MVA pre-Tx ocular melanocytosis (HR = 5.55, P < 0.001), Eye Physics plaque (HR = 2.45, P = 0.049), and any PPTx (HR = 4.47, P = 0.017) were associated with progression. PPTx with aflibercept was associated with less progression (HR = 0.13, P = 0.019). CONCLUSION Ocular melanocytosis and receiving any post-plaque treatment had the greatest impact on tumor progression. Class 2 tumors predicted for distant metastases but not local recurrences. These findings inform physicians of predictors of brachytherapy outcomes and allows for better shared-decision making with pre-operative patients. Future studies must validate these findings using a prospective cohort study.