The search strategy yielded 556 studies between 2005 and 2014, of which 29 were included in this review with the earliest published in 2005 and the latest in 2010. These included 22 studies of plaque therapy and 7 studies of proton beam therapy for choroidal melanoma. For studies of plaque therapy, sample sizes ranged from 24 to 650 patients and duration of follow-up ranged from an average of 23 to 60 months. For studies of proton beam therapy, sample sizes ranged from 18 to 349 patients and duration of follow-up ranged from an average of 28 to 78 months.

Studies examining visual acuity following plaque therapy for choroidal melanoma are outlined in Table 1 and Table 2. Visual acuity can be temporarily or permanently compromised following treatment for choroidal melanoma for a variety of reasons including radiation-induced damage to the anterior segment, retina, and optic nerve. Of the 15 included studies [10] - [20] [29] - [31] that examined visual acuity as an outcome following plaque therapy, sample sizes ranged from 24 to 650 patients and follow-up ranged from a mean of 23 months to five years. In five of six studies that examined the proportion of patients with visual acuity better than 20/40, less than 50% of patients were able to preserve visual acuity equal to or better than 20/40 at last follow-up. Fourteen studies examined the proportion of patients with visual acuity equal to or better than 20/200 with a wide range of proportions reported (23% - 94.4%). Factors shown to be associated with poor long term visual acuity following plaque therapy in multivariate models included radiation dose and tumor location [13] [30] [32] , with two studies [13] [32] demonstrating that tumor location near the optic nerve predicted poor long term visual function.

There were seven studies [9] [23] - [26] [34] included in this review that examined visual acuity as an outcome following proton beam therapy. The sample size in these studies ranged from 18 to 147 patients, and follow-up ranged from a median of 28 to 78 months. Compared to studies of plaque therapy, there were more studies that focused on large tumors and tumors close to the optic nerve. Most studies did not examine preservation of visual acuity equal to or better than 20/40 as an outcome, with only one study [23] reporting that 44.8% of patients had visual acuity equal to or better than 20/40 at five years following the initiation of treatment. Presumably, most studies only focused on preservation of visual acuity equal to or better than 20/200 as a long term outcome because proton beam therapy was associated with more visual loss compared to plaque therapy; the majority of studies [9] [24] [25] [33] [34] included in this review found that less than 50% of patients were able to preserve visual acuity equal or greater than 20/200 at the end of follow-up after proton beam therapy. The only study that examined predictors of poor visual acuity following proton beam therapy [27] found that initial degree of ocular involvement, tumor height, and initial visual acuity were predictive of poor visual acuity following therapy.

Studies examining eye preservation following plaque and proton beam therapy for choroidal melanoma are outlined in Table 3 and Table 4. Secondary enucleation for choroidal melanoma is most often performed for tumor recurrence or treatment toxicity [33] . While enucleation may achieve local tumor control and provide symptomatic relief, it may also be associated with cosmetic side effects and potential detriment to quality of life [34] . There were 13 studies [10] - [14] [16] - [18] [20] [29] - [32] included in this review that examined eye preservation following plaque therapy for choroidal melanoma. Sample sizes ranged from 24 to 650 patients and mean follow-up ranged from 23 months to five years. The majority of studies of plaque therapy reported less than 10% enucleation rates for tumor recurrence [10] [11] [13] [17] [18] [20] [31] , while enucleation rates for treatment toxicity ranged from 1.6% [13] to 14.2% [11] . The only study that examined predictors of enucleation [32] following plaque therapy found that plaque size, tumor recurrence, neovascular glaucoma, and nonproliferative retinopathy were predictive of requiring enucleation.

There were seven included studies [9] [23] - [26] [34] [35] that examined eye preservation following proton beam therapy for choroidal melanoma. Sample sizes ranged from 21 to 349 patients and follow-up time ranged from a mean of 28 months to a median of 78 months. While most studies reported similar secondary enucleation rates when compared to studies of plaque therapy, three studies [9] [24] [35] reported cumulative enucleation

rates greater than 20%, which was noticeably higher than rates reported in studies of plaque therapy. One study [9] was limited to extra-large and ciliochoroidal tumors and another [24] to stage T3 and T4 tumors, which might explain the higher rates of enucleation reported in these studies. There were no included studies that examined multivariate predictors of secondary enucleation following proton beam therapy.

Studies examining optic neuropathy following plaque and proton beam therapy for choroidal melanoma are outlined in Table 5 and Table 6. Optic neuropathy following radiation is thought to be secondary to demyelination and neuronal degeneration following glial and endothelial cell injury from radiation, and most often leads to irreversible vision loss over time [8] [36] [37] . There were ten studies [10] - [13] [16] [17] [19] [20] [32] [38] included in this review that examined optic neuropathy following plaque therapy, with sample sizes ranging from 24 to 650 patients and follow-up time ranging from a mean of 23 months to 56 months. Rates of radiation optic neuropathy ranged from 1.6% to 62.5%, with the highest rates being in studies of choroidal melanoma within 1.5 mm of the optic disc [24] and juxtapapillary choroidal melanoma [32] . In the two studies that examined multivariate predictors of optic neuropathy [19] [32] , plaque size and subfoveal or juxtapapillary tumor location were found to be predictive of optic neuropathy following plaque therapy.

There were three studies reviewed [9] [33] [34] that examined optic neuropathy following proton beam therapy for choroidal melanoma. Sample sizes ranged from 21 to 384 patients and follow-up time ranged from a median of 28 months to a mean of 78 months. Similar to the studies of neuropathy following plaque therapy, the two studies of optic neuropathy following proton beam therapy, which focused on peripapillary tumors [33] [34] , found high rates of neuropathy following treatment (89.6% and 67.7%), and proximity of the tumor to the optic disc was found to be predictive of radiation optic neuropathy in one of these studies [33] .

Studies examining radiation retinopathy following plaque and proton beam therapy for choroidal melanoma are outlined in Table 7 and Table 8. Radiation retinopathy is characterized by retinal hemorrhage, exudation, edema, ischemia and neovascularization of the retina. These abnormalities may lead to permanent visual loss [39] [40] . There were 13 included studies [10] [11] [13] [15] - [17] [19] [20] [30] [32] [38] [41] [42] that examined radiation retinopathy following plaque therapy, with sample sizes ranging from 24 to 3841 patients and follow-up time ranging from 23 to 56 months. Rates of maculopathy ranged from 8% to 56%, with the highest rates reported in studies of tumors near the optic disc [16] [32] . Similarly, there was a wide range of retinopathy reported (3.6% - 66%), but higher rates were reported in studies of tumors near the optic disc (60% - 66%) [32] [38] and in large or subfoveal tumors (56% - 66%) [15] [20] . Tumor location was the only predictive factor reported in more than one study [30] [42] .

There were two included studies [9] [34] that examined radiation retinopathy following proton beam therapy. While these studies differed greatly in sample size and follow-up time, the study that focused on juxtapapillary tumors [34] also found a high rate of retinopathy (90.3%) following proton beam therapy, similar to the studies of juxtapapillary tumors treated with plaque therapy. The other study [9] , which focused on large tumors, found a low rate of retinopathy compared to the studies of retinopathy following plaque therapy, suggesting that tumor size may not be related to the development of retinopathy for patients receiving proton beam therapy.

Studies examining neovascular glaucoma after plaque and proton beam therapy for choroidal melanoma are outlined in Table 9 and Table 10. Neovascularization of the retina and the iris is frequently a result of retinal ischemia from radiation retinopathy, leading to elevated intraocular pressure and permanent visual field loss from glaucoma [43] . There were seven studies [10] [11] [16] [20] [31] [32] [38] that examined neovascular glaucoma following plaque therapy for choroidal melanoma. Sample sizes ranged from 24 to 650 patients and follow-up time ranged from a mean of 23 months to 56 months. Rates of neovascular glaucoma in small and medium tumors were low (0.0% - 3%) [16] [31] , while studies of tumors near the optic nerve and large tumors reported slightly higher rates (8% - 17%) [10] [11] [20] [32] [38] . One study found that tumor thickness was predictive of neovascular glaucoma [32] .

There were three studies of neovascular glaucoma following proton beam therapy for choroidal melanoma [9] [25] [26] , with sample sizes ranging from 18 to 59 patients and follow-up time ranging from a median of 28 to 77.2 months. While reported rates of neovascular glaucoma were higher in all of these studies (23% - 38.1%) compared to those of neovascular glaucoma following plaque therapy, two of these studies focused on large tumors and those near the optic nerve. The one study [9] that examined predictors of neovascular glaucoma after proton beam therapy also found that tumor thickness and proximity to the optic nerve were predictive of neovascular glaucoma.

Studies examining cataract formation following plaque and proton beam therapy for choroidal melanoma are

outlined in Table 11 and Table 12. Cataract formation is a well-known complication resulting from radiation- induced opacification of the lens [44] . There were 12 studies [10] - [12] [15] - [17] [20] [31] [38] [41] [44] [45] included that examined cataract development following plaque therapy for choroidal melanoma. Sample sizes ranged from 24 to 532 patients and follow-up time ranged from a mean of 23 months to five years. There was a range of reported cataract development rates from 5.5% to 68%. The highest rate was reported in patients with

choroidal melanoma randomized to iodine-125 brachytherapy in the COMS [44] ; the high rate may be explained by the larger sample size, or more accurate documentation of outcomes due to the study being a randomized trial. Another study that reported similar rates by Krema et al. [38] focused on posterior tumors near the optic disc. Of the two studies that examined predictors of cataract formation following plaque therapy [44] [45] , both found that radiation dose was predictive of cataract formation following treatment.

There were two studies [9] [34] that examined cataract formation following proton beam therapy for choroidal melanoma. While the two studies had very different sample sizes and duration of follow-up and focused on different tumor types, both found similar rates of cataract development following proton beam therapy (28.6% and 31%). One of the studies [9] found that tumor distance from the fovea was predictive of cataract development following proton beam therapy, and the other study did not examine predictors of cataract development.

Studies examining distant metastasis from choroidal melanoma following plaque and proton therapy are outlined in Table 13 and Table 14. Distant metastasis from choroidal melanoma is associated with a poor prognosis with a median survival time of less than one year following metastasis [48] . The most common site of metastasis is the liver, with other sites being lung, bone, and skin [49] . There were 13 included studies [10] - [22] that examined the development of distant metastasis following treatment of choroidal melanoma with plaque therapy. Sample sizes ranged from 24 to 650 patients, and follow-up time ranged from a median of 17.3 months to a mean of 59.47 months. Rates of metastasis ranged from 0.0% to 31.9%, with the highest rates reported in a study of large tumors [20] and in the study with the longest duration of follow-up [18] . Of the studies that reported the time to development of metastasis [13] [14] [17] , the average time ranged from 30.4 months [13] to 58 months [14] . Three studies [13] [17] [18] reported rates of metastasis-free survival, with all reported rates greater than 75%. The only study [22] to examine predictors of metastasis found that higher intraocular pressure and greater tumor base were predictive of distant metastasis following plaque therapy.

There were five studies [9] [23] - [26] that examined the development of metastasis following proton therapy for choroidal melanoma. Sample sizes ranged from 21 - 349 patients, and duration of follow-up ranged from a median of 28 months to 77.2 months. Rates of metastasis were similar to those reported after plaque therapy, ranging from 5.6% to 25.4%. The only study [25] that examined time to metastasis reported a median of 38 months to metastasis, and the studies [24] [25] that examined metastasis-free survival reported survival rates of 72% to 82% at five years. Tumor dimension was the only factor reported to be related to metastasis development in more than one study [9] [23] .