Radiation Retinopathy Following Plaque Radiotherapy for Posterior Uveal Melanoma

Objective  To identify the risk factors that lead to the development of radiation retinopathy following plaque radiotherapy for posterior uveal melanoma. Radiation retinopathy is a slowly progressive, occlusive vasculopathy characterized by radiation-induced endothelial damage.

Methods  Review of the medical records of patients with posterior uveal melanoma treated with plaque radiotherapy.

Results  Of 1300 patients with posterior uveal melanoma treated with plaque radiotherapy from July 1, 1976, through June 30, 1992, radiation retinopathy developed in 560 (43.1%). By using Kaplan-Meier survival estimates, we found that 5% of the patients had nonproliferative radiation retinopathy at 1 year (95% confidence interval [CI], 3%-6%) and 42% at 5 years (95% CI, 38%-45%). The proportion of patients with proliferative retinopathy was 1% at 1 year (95% CI, 0.2%-1.5%) and 8% at 5 years (95% CI, 5%-10%). Multivariate analyses showed that the subset of clinical variables best related to the development of nonproliferative radiation retinopathy were tumor margin of less than 4 mm from foveola (P<.001), tumor limited to the choroid (P=.002), and radiation dose rate of greater than 260 cGy/h to the tumor base (P=.02). The best subset of independent variables related to the development of radiation maculopathy were tumor of less than 4 mm to foveola (P<.001) and the use of radioisotope iridium 192 (¹⁹²Ir) (P=.02) compared with iodine 125 (¹²⁵I). From a multivariate model, the most important factors for the development of proliferative radiation retinopathy included diabetes mellitus (P=.01), radioisotope ¹⁹²Ir (P=.01) compared with ¹²⁵I, and tumor base of greater than 10 mm (P=.02).

Conclusions  Radiation retinopathy is a common finding after plaque radiotherapy for choroidal melanoma, occurring in 42% of patients at 5 years. The main predictors of radiation retinopathy are posterior tumor location with margin near the foveola and high radiation dose rate to the tumor base.

FIRST DESCRIBED in 1933 by Stallard,¹ radiation retinopathy is characterized by a slowly progressive, occlusive vasculopathy with a delayed onset after irradiation. The pathologic hallmark of this condition seems to be radiation-related endothelial changes in the retinal vessels, leading to occlusive vascular disease.² Radiation retinopathy usually leads to irreversible visual impairment.

Radiation retinopathy can occur after external beam irradiation of the eye, orbit, eyelids, face, paranasal sinuses, nasopharynx, and brain.^3-5 The incidence of radiation retinopathy is greater when the treatment site is closer to the eye and the cephalic radiation dose is high.⁵ Although radiation retinopathy has been more commonly observed in patients treated with external beam irradiation, it can also occur after episcleral plaque radiotherapy.⁶

There have been a few reports specifically addressing the role of risk factors for the development of radiation retinopathy after plaque radiotherapy in patients with posterior uveal melanoma.^6,7 Although radiation retinopathy is a potential cause of visual loss in this group of patients, it nevertheless represents a balance between the treatment of a life-threatening ocular tumor and vision-threatening ocular complications. We herein aim to identify the factors that led to the development of radiation retinopathy following plaque radiotherapy in a large group of patients with posterior uveal melanoma.

We reviewed the medical records of patients with posterior uveal melanoma who were treated with plaque radiotherapy from July 1, 1976, through June 30, 1992, at the Ocular Oncology Service of Wills Eye Hospital, Phildelphia, Pa, and retrieved those of patients in whom radiation retinopathy developed. We analyzed the risk factors that led to the development of radiation retinopathy after plaque radiotherapy in this group.

Baseline patient data included age, sex, patient history of diabetes mellitus and systemic hypertension, best-corrected initial visual acuity (>20/40*, 20/40-20/200, >20/200), and baseline intraocular pressure. The variables marked with an asterisk were used as reference for later statistical analyses. Tumor data consisted of tumor quadrant (superior, superotemporal, temporal, inferotemporal, inferior, inferonasal,* nasal, superonasal, and foveal), the distance of the posterior edge of the tumor to the optic disc and foveola, maximum diameter of tumor base (from results of indirect ophthalmoscopy and B-scan ultrasonography), initial thickness (from results of A- and B-scan ultrasonography), shape (dome,* mushroom, diffuse, and plateau as defined by results of indirect ophthalmoscopy and B-scan ultrasonography), the presence of subretinal fluid (from results of indirect ophthalmoscopy and A- and B-scan ultrasonography), and retinal invasion (from results of indirect ophthalmoscopy, B-scan ultrasonography, and fluorescein angiography).

Informed consent was obtained before the patients underwent plaque radiotherapy. Plaque radiotherapy data used in this study include radioactive isotope (iodine 125 [¹²⁵I],* ruthenium 106 [¹⁰⁶Ru], cobalt 60 [⁶⁰Co], and iridium 192 [¹⁹²Ir]); plaque shape (round,* notched, rectangular, or curvilinear); plaque size; hours of radiation; radiation doses to the apex, base, optic disc, and fovea; and radiation rates to the apex, base, optic disc, and fovea.

The follow-up examinations were generally made at 3- to 6-month intervals up to 5 years and at 6- to 12-month intervals thereafter. The development of radiation retinopathy and the time interval to its onset was determined. Results of indirect ophthalmoscopy, fundus photography, and fluorescein angiography were used to determine the development of radiation retinopathy.

Nonproliferative radiation retinopathy was diagnosed if a group of capillary bed changes (microaneurysms, dilation, and nonperfusion), retinal hemorrhage, retinal exudation, retinal edema, nerve fiber layer infarction, and/or vascular sheathing were found. The presence of at least 2 of these findings was required to make the diagnosis. Radiation maculopathy was diagnosed when these changes occurred 3 mm or less from the foveola. A diagnosis of proliferative radiation retinopathy was made when there was retinal or optic disc neovascularization.

The clinical examination records, color photographs, and fluorescein angiograms were compared with those obtained at baseline to exclude retinal vascular changes due to other causes. Only those retinal vascular changes that developed following plaque radiotherapy were considered significant for our purposes.

The major outcome events of our study were the development of nonproliferative radiation retinopathy, radiation maculopathy, and proliferative radiation retinopathy. The effect of individual variables on the development of each outcome event was analyzed using a series of Cox proportional hazards regressions.⁸ The correlation among the variables was determined using Pearson correlations. All variables were analyzed as discrete variables except for age, intraocular pressure, largest basal tumor diameter, tumor thickness, and radiation dose and rates, which were analyzed as continuous variables and later grouped into discrete categories to derive cutoff values. Subsequent multivariate models included variables that were significant on a univariate level (P<.05) and identified the combination of variables best related to the development of events. Kaplan-Meier survival estimates were used to study the development of radiation retinopathy as a function of time.⁹

Five hundred sixty (43.1%) of 1300 patients with posterior uveal melanoma and treated with plaque radiotherapy from July 1, 1976, through June 30, 1992, were found to have radiation retinopathy. Demographics and ocular findings in patients in whom radiation retinopathy developed vs those in whom it did not are listed in Table 1. Of the patients in whom radiation retinopathy developed, 282 were women and 278 were men. The median age was 58 years (range, 20-86 years). Forty-nine patients (8.8%) had systemic hypertension, 29 (5.2%) had diabetes mellitus, and 19 (3.4%) had both.

Choroidal melanoma was found in 532 (95.0%) of the 560 patients; ciliochoroidal melanoma, in 28 (5.0%). The epicenter of the tumor was superior in 27 patients (4.8%), superotemporal in 147 (26.2%), temporal in 67 (12.0%), inferotemporal in 144 (25.7%), inferior in 31 (5.5%), inferonasal in 53 (9.5%), nasal in 26 (4.6%), superonasal in 57 (10.2%), and foveal in 8 (1.4%). The median largest basal tumor diameter was 10 mm (range, 3-18 mm), and the median tumor thickness was 4 mm (range, 2-11 mm). The median distance of the posterior margin of the tumor to the optic disc was 4 mm (range, 0-13 mm) and to the foveola was 3 mm (range, 0-13 mm). Subretinal fluid was present in 401 patients (71.6%). The tumor was dome shaped in 496 patients (88.6%), mushroom shaped in 62 (11.1%), and plateau shaped in 2 (0.4%). Extraocular extension was observed in 7 patients (1.2%). Retinal invasion was present in 5 patients (0.9%).

The radioisotope ¹²⁵I was used in 329 patients (58.7%), ⁶⁰Co in 153 (27.3%), ¹⁹²Ir in 58 (10.4%), and ¹⁰⁶Ru in 20 (3.6%). A regular, round plaque was used in 403 patients (72.0%), a notched plaque in 149 (26.6%), a rectangular plaque in 6 (1.1%), and a custom-designed curvilinear plaque in 2 (0.4%). The plaque size ranged from 10 to 25 mm. The most frequently used plaque size was 15 mm (85.4%), followed by 20 mm (5.4%), 10 mm (3.9%), 18 mm (3.2%), and 25 mm (2.1%). The radiation characteristics for patients in whom radiation retinopathy developed vs those who were free of retinopathy are shown in Table 2.

Follow-up ranged from 20 to 232 months (median, 61 months); 501 patients had regular follow-up examinations. The median follow-up times of patients treated with radioisotope ¹²⁵I was 57 months; ¹⁰⁶Ru, 64 months; ¹⁹²Ir, 72 months; and ⁶⁰Co, 81 months.

All patients with radiation retinopathy had nonproliferative changes. Radiation maculopathy was found in 240 patients (42.8%) and proliferative retinopathy in 51 patients (9.1%).

By using Kaplan-Meier survival curves, we found that the proportion of patients in whom nonproliferative radiation retinopathy developed was 5% at 1 year (95% confidence interval [CI], 3%-6%) and 42% at 5 years (95% CI, 38%-45%) (Figure 1). Proliferative radiation retinopathy was observed in 1% of the patients at 1 year (95% CI, 0.2%-1.5%) and 8% at 5 years (95% CI, 5%-10%) (Figure 2).

Of the 560 patients, 222 (39.6%) also had radiation cataract; 106 (18.9%), papillopathy; 43 (7.7%), vitreous hemorrhage; 40 (7.1%), neovascular glaucoma; and 7 (1.2%), scleral necrosis. Twelve patients (2.1%) were treated for recurrence with a second plaque.

Forty-seven eyes (8.4%) were enucleated due to recurrence (20 eyes), neovascular glaucoma (14 eyes), vitreous hemorrhage (11 eyes), or scleral necrosis (2 eyes). Fifty-five patients (9.8%) died of melanoma-related causes, and 4 (0.7%) of other causes. Of the 55 patients with melanoma-related mortality, 6 previously had undergone enucleation.

Results of the univariate Cox proportional hazards regression analyses showed that the factors predictive of the development of nonproliferative radiation retinopathy were tumor of less than 4 mm to foveola (P<.001), tumor limited to the choroid with no anterior uveal involvement (P<.001), tumor of less than 5 mm to optic disc (P<.001), radiation dose rate of greater than 260 cGy/h to the tumor base (P=.002), subretinal fluid (P=.003), superotemporal tumor meridian (P=.005), radiation dose rate of greater than 75 cGy/h to the tumor apex (P=.007), radiation dose rate of greater than 44 cGy/h to the optic disc (P=.01), radiation dose of greater than 4367 cGy to the optic disc (P=.02), radiation dose of greater than 5742 cGy to the fovea (P=.03), and radiation dose rate of greater than 53 cGy/h to the fovea (P=.05) (Table 3). From a multivariate standpoint, the subset of clinical variables best related to the development of nonproliferative radiation retinopathy included tumor of less than 4 mm to foveola (P<.001), tumor limited to the choroid (P=.002), and radiation dose rate of greater than 260 cGy/h to the tumor base (P=.02) (Table 3).

The factors predictive of radiation maculopathy on a univariate level were tumor margin of less than 4 mm to foveola (P<.001), radiation dose rate of greater than 75 cGy/h to the tumor apex (P<.001), radiation dose rate of greater than 260 cGy/h to the tumor base (P<.001), tumor of less than 5 mm to optic disc (P<.001), radiation dose of greater than 9000 cGy to the tumor apex (P=.003), tumor limited to choroid (P=.003), diabetes mellitus (P=.02), use of radioisotope ¹⁹²Ir (P=.03) compared with ¹²⁵I, and tumor thickness of greater than 5 mm (P=.05) (Table 4). By multivariate analysis, the best subset of independent variables related to the development of radiation maculopathy included tumor of less than 4 mm to foveola (P<.001) and use of radioisotope ¹⁹²Ir (P=.02) compared with ¹²⁵I (Table 4).

From a univariate standpoint, the factors predictive of proliferative radiation retinopathy were tumor thickness of greater than 5 mm (P=.009), use of radioisotope ¹⁹²Ir (P=.02) compared with ¹²⁵I, diabetes mellitus (P=.04), and largest tumor base of greater than 10 mm (P=.05) (Table 5). By multivariate analysis, the most important factors for the development of proliferative radiation retinopathy included diabetes mellitus (P=.01), use of radioisotope ¹⁹²Ir (P=.01) compared with ¹²⁵I, and largest tumor base of greater than 10 mm (P=.02) (Table 5).

Iodine 125 was the most commonly used radioisotope. The risk factors predictive of the development of radiation retinopathy in 329 patients treated with ¹²⁵I were radiation dose rate of greater than 260 cGy/h to the tumor base (P=.03), tumor limited to choroid (P=.03), largest tumor base of greater than 10 mm (P=.04), and tumor thickness of greater than 5 mm (P=.05) on a univariate level (Table 6). Multivariate analysis showed that the most important factors predictive of the development of radiation retinopathy in this group of patients were tumor thickness of greater than 5 mm (P=.001), tumor limited to choroid (P=.03), and radiation dose rate of greater than 260 cGy/h to the tumor base (P=.05) (Table 6).

Radiation retinopathy is an important complication of plaque radiotherapy of posterior uveal melanoma, often leading to irreversible visual impairment. The diverse manifestations of radiation-induced retinal vascular changes can be divided into nonproliferative and proliferative retinopathy. Nonproliferative radiation retinopathy consists of capillary changes (microaneurysms, dilation, nonperfusion), retinal hemorrhage, retinal exudation, retinal edema, nerve fiber layer infarctions, and/or vascular sheathing. Radiation maculopathy is a subgroup of nonproliferative retinopathy, diagnosed when any of these changes occur within 3 mm of the foveola. Proliferative radiation retinopathy is diagnosed when there is retinal or disc neovascularization.

The early manifestations of radiation retinopathy include microaneurysms and telangiectasia as well as retinal edema and exudation from compromise of the retinal capillary bed.¹⁰ As areas of capillary loss become confluent, nerve fiber layer infarctions are seen. Larger vessels are involved later in the course of the disease, and vascular sheathing becomes apparent. When there are large or multiple areas of capillary nonperfusion, retinal and disc neovascularization can develop, sometimes leading to vitreous hemorrhage.^11,12 Occasional associations with radiation retinopathy include choroidal neovascularization,¹³ choroidal infarction,¹⁴ retinal artery occlusion,¹⁶ and retinal vein occlusion.¹⁵

Results of histopathologic studies of the eyes with radiation retinopathy confirm the vascular nature of damage produced by radiation therapy.^17,18 There is an unequivocal loss of endothelial cells with relative sparing of the pericytes. Although small retinal vessels are commonly involved, larger retinal and choroidal vessels can also be affected.¹⁸ The inner retinal layers are preferentially damaged as a result of this involvement pattern; however, there may also be destruction of the retinal pigment epithelium.¹⁹

The most important predictor of nonproliferative radiation retinopathy in our study was posterior tumor location. From a multivariate model, the factors related to the development of nonproliferative retinopathy included tumor margin of less than 4 mm to the foveola, tumor limited to the choroid without ciliary body and iris involvement, and radiation dose rate of greater than 260 cGy/h to the tumor base. Our results indicate that the peripheral retina may be less susceptible to the damaging effects of irradiation in terms of developing retinopathy. Similarly in diabetes mellitus, retinal vascular changes most often are found posteriorly near the vascular arcades, more so than in the retinal periphery. It has been reported that clinically and angiographically, radiation retinopathy is virtually identical to diabetic retinopathy.^20,21

There has been more emphasis in the literature on the subject of radiation maculopathy because of its profound visual effects. In our study, the best subset of independent variables related to the development of radiation maculopathy included tumor margin of less than 4 mm to foveola and radioisotope ¹⁹²Ir compared with ¹²⁵I on a multivariate level. The higher rate of radiation maculopathy after the use of radioisotope ¹⁹²Ir can be explained on the basis of its high energy levels (0.38 MeV) compared with ¹²⁵I (0.032 MeV).²²

In 2 previous reports, 18%²³ and 23%²⁴ of the patients with posterior uveal melanoma who were treated with plaque radiotherapy were found to have radiation maculopathy. In our study, radiation maculopathy was found in 42.8% of patients developing radiation retinopathy following plaque radiotherapy. Radiation maculopathy can manifest with macular edema if capillary abnormalities and leakage predominate and with macular ischemia if perifoveal capillary loss predominates. In a recent study, focal laser treatment was found to improve visual acuity modestly at 6 months in patients with radiation-induced macular edema, but sustained benefit was not achieved at 2 years.²⁵

Brown et al⁶ reported that patients in whom radiation maculopathy developed after plaque radiotherapy received a mean radiation dose of 15,000 cGy to the fovea vs 5000 cGy for those who had no maculopathy. They suggested that plaque radiotherapy doses of less than 5000 cGy to the fovea were probably safe. In our study, a fovea radiation dose of greater than 5742 cGy was associated with a significantly higher risk for nonproliferative radiation retinopathy on a univariate level. We agree that a cutoff dose of approximately 5000 cGy or less may be tolerated safely by the fovea following plaque radiotherapy.

After plaque radiotherapy for posterior uveal melanoma, the most important factors for the development of proliferative radiation retinopathy were diabetes mellitus, radioisotope ¹⁹²Ir compared with ¹²⁵I, and tumor base of greater than 10 mm. The presence of diabetes mellitus exacerbates the effects of radiation retinopathy.²⁶ Radiation retinopathy causes selective loss of endothelial cells, whereas diabetic retinopathy affects the pericytes. Therefore, the 2 components of the capillary wall are destroyed by the combined effect of diabetes and radiation, leaving little cellular support behind.²¹ However, Kinyoun et al²⁷ found that diabetic retinopathy was not a risk factor for proliferative radiation retinopathy. The radioisotope ¹⁹²Ir is associated with a higher risk for proliferative radiation retinopathy compared with ¹²⁵I, due to its higher energy levels. Larger tumor base is associated with a higher incidence of retinopathy because spillover radiation energy delivered to the surrounding retina is greater when the tumor base is larger.

During the past decade, we have been using ¹²⁵I as the radioisotope of choice because of its availability, ease of shielding, good tissue penetration, and ability to be custom-designed.²² Our results did not show an increase in the frequency of radiation retinopathy with the radioisotope ⁶⁰Co compared with ¹²⁵I, despite its considerably higher energy level (1.25 MeV vs 0.032 MeV for ¹²⁵I). The radioisotope ⁶⁰Co was used in the earlier part of the study, and many eyes with more posterior tumors may have been enucleated rather than treated with plaque radiotherapy during that period. Similar cases recently have been treated with ¹²⁵I plaque radiotherapy, altering the relative frequency of radiation retinopathy compared with ⁶⁰Co. The multivariate analysis of the patients treated with ¹²⁵I plaque radiotherapy showed that the factors related to the development of radiation retinopathy were tumor thickness of greater than 5 mm, tumor limited to choroid, and radiation dose rate of greater than 260 cGy/h to the tumor base. In this subset of patients, tumor proximity to the foveola was not a significant factor.

There are certain limitations of our study that should be considered. First, it is a retrospective study, and more difficult cases may have been sent to us for management because of our special interest in ocular tumors. Second, follow-up may be considered short for a study dealing with posterior uveal melanoma. However, we realize from previous reports that radiation retinopathy generally occurs at 1 to 3 years after plaque treatment.⁶ Most of our patients had long enough follow-up, adequate to assess the complication. Third, mild degrees of radiation retinopathy, especially in the peripheral retina, not resulting in visual loss may have been overlooked in our study. We suspect that many patients had some radiation-related change on or near the tumor. However, since we did not routinely use 7-field photographs, we may have underestimated the abnormality.

In conclusion, radiation retinopathy is a common finding after plaque radiotherapy for posterior uveal melanoma, occurring in 42% of patients at 5 years. The main predictors of radiation retinopathy are tumor margin near the optic disc and foveola as well as high radiation dose rate to the tumor base. The presence of diabetes mellitus is another major factor that predisposes to the development of radiation retinopathy.

Accepted for publication December 15, 1998.

This study was supported by the Paul Kayser International Award of Merit in Retina Research, Houston, Tex (Dr J. A. Shields); the Eye Tumor Research Foundation and the Pennsylvania Lions Sight Conservation and Eye Research Foundation (Drs Gündüz, C. L. Shields, and J. A. Shields), Philadelphia; and the Macula Foundation, New York, NY (Drs Gündüz and C. L. Shields).

Reprints: Carol L. Shields, MD, Oncology Service, Wills Eye Hospital, 900 Walnut St, Philadelphia, PA 19107.