Radiation Complications and Tumor ControlAfter Plaque Radiotherapy of ChoroidalMelanoma With Macular Involvement

KAAN GUNDUZ, MD, CAROL L. SHIELDS, MD, JERRY A. SHIELDS, MD,JACQUELINE CATER, PHD, JORGE E. FREIRE, MD, AND LUTHER W. BRADY, MD

● PURPOSE: To determine the outcome of plaque radio-therapy in the treatment of macular choroidal melanomaand to identify the risk factors associated with thedevelopment of radiation complications, tumor recur-rence, and metastasis.● METHODS: Chart analysis of 630 consecutive patients(630 eyes) with macular choroidal melanoma managed byplaque radiotherapy between July 1976 and June 1992.● RESULTS: The median largest basal tumor diameter was10 mm, and the median tumor thickness was 4 mm. Bymeans of Kaplan-Meier estimates, visually significantmaculopathy developed at 5 years in 40% of the patients,cataract in 32%, papillopathy in 13%, and tumor recur-rence in 9%. Vision decrease by 3 or more Snellen lineswas found in 40% of the patients at 5 years. Sixty-nineeyes (11%) were enucleated because of radiation com-plications and recurrence. Twelve percent of the patientsdeveloped metastasis by 5 years and 22% by 10 years.

Results of multivariate Cox proportional hazards anal-yses showed that the significant predictors for tumorrecurrence were a distance of tumor margin from theoptic disk of less than 2 mm (P 5 .003) and retinalinvasion (P 5 .009). The significant variables that werepredictive of metastasis included tumor thickness greaterthan 4 mm (P 5 .02) and largest basal tumor diametergreater than 10 mm (P 5 .03).

● CONCLUSIONS: Plaque radiotherapy offers a 91%5-year local tumor control rate for macular choroidalmelanoma. Despite good local tumor control, the risk formetastasis is 12% at 5 years and 22% at 10 years. In11% of the patients, enucleation eventually becamenecessary because of radiation complications and tumorrecurrence. (Am J Ophthalmol 1999;127:579–589.© 1999 by Elsevier Science Inc. All rights reserved.)

T HERE ARE SEVERAL TREATMENT METHODS FOR THE

management of posterior uveal (ciliary body andchoroidal) melanoma.1,2 Posterior uveal melanoma

was treated primarily by enucleation several years ago, butconcern about possible acceleration of metastasis afterenucleation3 brought about a trend toward institutingradiotherapy. A number of studies in which patients werenot randomly assigned have shown similar survival ofpatients treated with numerous methods, including enu-cleation, plaque radiotherapy, and charged particle radio-therapy.4–6 The Collaborative Ocular Melanoma Study iscurrently evaluating in a randomized fashion the effect ofenucleation versus plaque radiotherapy in patients withchoroidal melanoma.7

Plaque radiotherapy has become an established methodof treatment for posterior uveal melanoma. One of theconcerns regarding plaque radiotherapy is that it may beunsuitable for the treatment of choroidal melanoma withmacular involvement, because of the resultant severevisual loss.8 We review our experience with plaque radio-therapy to treat choroidal melanoma involving the maculawith respect to risk factors for the development of radia-tion events, including complications and visual decrementand tumor events of recurrence and metastasis.

PATIENTS AND METHODS

WE ANALYZED THE RECORDS OF ALL PATIENTS WHO HAD

posterior uveal melanoma with macular involvementtreated with plaque radiotherapy on the Ocular Oncology

Accepted for publication Dec 31, 1998.From the Oncology Service, Wills Eye Hospital, Thomas Jefferson

University, Philadelphia, Pennsylvania (Drs Gunduz, C. Shields, J.Shields, and Cater), and Department of Radiation Oncology, AlleghenyUniversity Health System at Hahnemann, Philadelphia, Pennsylvania(Drs Freire and Brady).

Supported by TUBITAK (Dr Gunduz), the Macula Foundation, NewYork, New York (Drs Gunduz and C. Shields), the Paul Kayser Interna-tional Award of Merit in Retina Research, Houston, Texas (Dr J.Shields), the Eye Tumor Research Foundation (Dr C. Shields), and thePennsylvania Lions Sight Conservation and Eye Research Foundation,Philadelphia, Pennsylvania (Drs Gunduz, C. Shields, and J. Shields).

Biostatistical consultation was provided by J. Cater, PhD.Reprint requests to Carol L. Shields, MD, Oncology Service, Wills

Eye Hospital, 900 Walnut Street, Philadelphia, PA 19107; fax: (215)928-1140.

© 1999 BY ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED.0002-9394/99/$20.00 579PII S0002-9394(98)00445-0

Service, Wills Eye Hospital, between June 1976 and June1992. A choroidal melanoma with macular involvementwas defined as a tumor whose margin extended 3 mm orless from the foveola. Although most patients had morethan 5 years of follow-up, those who died, had enucleation,or were lost to follow-up before 5 years were also includedin the study. However, patients who received adjuvantablative laser photocoagulation or transpupillary thermo-therapy were excluded.

The baseline patient data included age, sex, race, sys-temic illness (diabetes mellitus, hypertension), best-cor-rected initial visual acuity (.20/40*, 20/40 to 20/200,,20/200), and intraocular pressure. The asterisked vari-ables were used as a reference for later statistical analysis.Tumor data comprised tumor quadrant (superior, supero-temporal, temporal, inferotemporal,* inferior, inferonasal,nasal, superonasal), the distance (mm) from the posterioredge of the tumor to the optic nerve and foveola, maxi-mum basal tumor diameter, greatest tumor thickness (fromA-scan and B-scan ultrasonography), shape (dome,* mush-room, diffuse, or plateau), the presence of subretinal fluid,and retinal invasion.

The radioactive plaque size was adjusted to be 2 mmlarger than the tumor on all sides. Transillumination andindirect ophthalmoscopy localization were used in theprecise localization of the macular choroidal melanoma.Plaque radiotherapy data included radioactive isotope(Iodine-125,* Ruthenium-106, Cobalt-60, or Iridium-192); plaque shape (round,* notched, rectangular, orcurvilinear); plaque size (diameter in millimeters); hours ofradiation; radiation doses to the apex, base, optic disk,fovea, and lens (cGy 5 centigray); and radiation rates tothe apex, base, optic disk, fovea, and lens (cGy/hr).

The follow-up examinations were performed at 3-monthto 6-month intervals up to 5 years and at 6-month to12-month intervals thereafter. In addition to indirectophthalmoscopy, ultrasonography, fundus photography,and fluorescein angiography were used in following thetumor response. Information on Snellen visual acuity andthe development of nonproliferative and proliferative ret-inopathy, cataract, papillopathy, vitreous hemorrhage,neovascular glaucoma, scleral necrosis, recurrence, metas-tasis, and melanoma-related mortality was recorded. Theinterval to the development of each event was determined.For patients with enucleated eyes, the findings at the timeof the last examination before enucleation were consideredin the assessment of radiation events and recurrence.However, these patients were still followed up for thedevelopment of metastasis.

We defined radiation maculopathy as retinal capillarybed changes (nonperfusion, dilation, microaneurysms, ret-inal hemorrhage), retinal exudation, retinal edema, nervefiber layer infarctions, or vascular sheathing that was foundin the foveal area and impaired central vision. In additionto the clinical examination, fundus photography andfluorescein angiography were performed on a regular basis

to determine the development of radiation maculopathy.Radiation papillopathy was diagnosed when peripapillaryexudation, hemorrhage, or optic disk swelling was present.Radiation cataract was defined as the development of alens opacity, especially a posterior subcapsular cataract,which was not present before radiation treatment. Thediagnosis of vitreous hemorrhage was rendered when anyamount of vitreous hemorrhage was detected. Radiationscleral necrosis was diagnosed when thinning of the sclerawas to the point of visualizing underlying uveal tissue. Thediagnosis of neovascular glaucoma was based on thepresence of iris or angle neovascularization and an increasein intraocular pressure.

Tumor recurrence was defined as any degree of enlarge-ment of the residual tumor in base or thickness, as detectedby ophthalmoscopy and ultrasonography. This definitionwas based on the fact that with any malignancy, any de-gree of enlargement is significant and should be con-sidered regrowth. The final tumor thickness (as measuredby A-scan and B-scan ultrasonography), best-correctedSnellen visual acuity, and systemic metastatic status were,determined. Melanoma-related deaths and deaths fromother causes were recorded.

The major outcomes analyzed in this study were thedevelopment of radiation events including maculopathy,cataract, papillopathy, vitreous hemorrhage, neovascularglaucoma, scleral necrosis, visual acuity decrement, andtumor events including recurrence, metastasis, and mela-noma-related mortality. The effect of individual clinicalvariables on the development of each outcome event wasanalyzed by a series of univariate Cox proportional hazardsregressions.9 The correlation among the variables wasdetermined by means of Pearson correlations. All variableswere analyzed as discrete variables, except for patient age,intraocular pressure, basal tumor diameter, tumor thick-ness, radiation dose, and radiation rate, which were ana-lyzed as continuous variables and later grouped intodiscrete categories to derive cut-off values. The variablesthat were significant on a univariate level (P , .05) wereentered into a stepwise regression analysis. For variablesthat showed a high degree of correlation, only one variablefrom the set of associated variables was entered at a time insubsequent multivariate models. A final multivariatemodel fitted variables that were identified as significantpredictors (P , .05) in the stepwise model, as well asvariables deemed clinically important for the specificoutcome.

Kaplan-Meier survival estimate curves were used toanalyze the development of radiation maculopathy, cata-ract, papillopathy, vitreous hemorrhage, visual acuity dec-rement of at least 3 Snellen lines, tumor recurrence,metastasis, and melanoma-related mortality as a functionof time.10 With respect to visual acuity analysis, patientswhose initial vision was 20/200 or worse were not includedin the analysis. The point at which the patient experi-enced a decrement of at least 3 Snellen lines of vision was

AMERICAN JOURNAL OF OPHTHALMOLOGY580 MAY 1999

computed and analyzed by means of a Cox proportionalhazards model, with time to the event as the end point.

RESULTS

BETWEEN JULY 1976 AND JUNE 1992, 630 PATIENTS (630 EYES)

with choroidal macular melanoma received treatment bymeans of plaque radiotherapy. Of these 630 patients, 317were women and 313 were men. The median age was 59years (range, 20 to 91 years). Forty-five patients (7%) hadsystemic hypertension, 39 (6%) had diabetes mellitus, and15 (2%) had both of these conditions. The initial visualacuity was better than 20/200 in 91% of the patients andbetter than 20/40 in 55%. The median intraocular pressurewas 15 mm Hg (range, 8 to 27 mm Hg).

Of the 630 patients with choroidal melanoma withmacular involvement, only 15 (2%) had a portion of thetumor located beneath the foveola. Of the remainingmacular choroidal melanomas, the tumor was superior in21 patients (4%), superotemporal in 165 (26%), temporalin 116 (19%), inferotemporal in 193 (31%), inferior in 40(7%), inferonasal in 30 (5%), nasal in 19 (3%), andsuperonasal in 31 (5%). The median largest basal tumordiameter was 10 mm (range, 3 to 20 mm), and the mediantumor thickness was 4 mm (range, 1 to 12 mm). Themedian distance of the posterior margin of the tumor tothe optic nerve was 3 mm (range, 0 to 6 mm), and themedian distance to the foveola was 2 mm (range, 0 to 3mm). Sixty-three of 630 melanomas (10%) contacted theoptic disk. Subretinal fluid was present in 496 patients(79%). The tumor was dome shaped in 565 patients(89%), mushroom shaped in 62 (10%), plateau in two(,1%), and diffuse in one (,1%). Retinal invasion wasclinically present in five patients (1%). Extraocular exten-sion was observed in seven patients (1%).

An evolution occurred in the radioactive isotope usedfor plaque radiotherapy. Cobalt-60 was used in the earlieryears, but recently Iodine-125 has been used almost exclu-sively. The radioisotope Iodine-125 was used in 382 cases(61%), Cobalt-60 in 158 (25%), Iridium-192 in 72 (11%),and Ruthenium-106 in 18 (3%). Depending on locationand shape of the tumor, a round plaque was used in 412cases (65%), a notched plaque in 209 (33%), a rectangularplaque in four (1%), and a custom-designed curvilinearplaque in three (1%). Plaque sizes ranged from 10 to 25mm. The most frequently used plaque size was 15 mm(85%), followed by 10 mm (6%) and 20 mm (4%). Otherplaque sizes, such as 18 mm and 22 mm, were used lessfrequently. The radiation time, dose, and rate of radiationto the tumor apex, base, optic disk, fovea, and lens areshown in Table 1. The median radiation time was 120hours. The median apex dose was 9,120 cGy, and themedian base dose was 32,900 cGy. The median apex doserate was 80 cGy per hour, and the median base dose ratewas 266 cGy per hour.

The follow-up period was 16 to 224 months (median, 63months), and follow-up data were available for 100% ofpatients. The final visual acuity was better than 20/200 in44% of the patients and better than 20/40 in 9%. Themedian final tumor thickness was 2 mm (range, 1 to 6mm).

Table 2 shows the Kaplan-Meier estimates of the prob-ability of developing major radiation complications andtumor events after plaque radiotherapy of macular choroi-dal melanoma. Radiation maculopathy was the most fre-quent complication and was observed in 240 patients(38%). By means of Kaplan-Meier estimates (Figure 1), wefound that the probability of developing radiation macu-lopathy was 40% at 5 years and 59% at 10 years. Radiationpapillopathy was observed in 74 patients (12%). Kaplan-Meier estimates showed that the probability of developingradiation papillopathy was 13% at 5 years and 22% at 10years (Figure 1).

Radiation cataract was noted in 192 patients (30%), andby means of Kaplan-Meier estimates, the probability ofdeveloping radiation cataract was 32% at 5 years and 59%at 10 years (Figure 2). Vitreous hemorrhage was docu-mented in 59 patients (9%), with 9% developing vitreoushemorrhage at 5 years and 18% at 10 years (Figure 2).Twenty of the patients with vitreous hemorrhage hadproliferative radiation retinopathy, whereas in the remain-der vitreous hemorrhage presumably developed secondaryto tumor necrosis. Vision decrement by 3 or more Snellenlines was noted in 252 patients (40%). The probability ofdeveloping a vision decrement of 3 or more Snellen lineswas 40% at 5 years and 82% at 10 years (Figure 3).Neovascular glaucoma was diagnosed in 47 patients (8%),and scleral necrosis was found in two patients (,1%).

Tumor recurrence was noted in 51 patients (8%).Kaplan-Meier estimates showed that 9% developed tumorrecurrence at 5 years and 12% at 10 years (Figure 4). Therewas enlargement of the residual tumor in the base in 36

TABLE 1. Radiation Parameters of Plaque Radiotherapyfor 630 Patients With Macular Choroidal Melanoma

Variable Median Range

Radiation time 120 hours 26–374 hours

Apex dose 9,120 cGy 2,960–16,300 cGy

Apex dose rate 80 cGy/hr 13–229 cGy/hr

Base dose 32,900 cGy 9,600–50,500 cGy

Base dose rate 266 cGy/hr 88–761 cGy/hr

Optic disk dose 5,904 cGy 1,071–27,444 cGy

Optic disk dose rate 54 cGy/hr 9–270 cGy/hr

Fovea dose 12,207 cGy 1,263–31,900 cGy

Fovea dose rate 109 cGy/hr 12–270 cGy/hr

Lens dose 1,168 cGy 554–7,326 cGy

Lens dose rate 12 cGy/hr 6–41 cGy/hr

cGy 5 centigray.

PLAQUE RADIOTHERAPY OF CHOROIDAL MELANOMAVOL. 127, NO. 5 581

patients and in thickness in 15 patients. Thirty-two (63%)of 51 patients were treated with enucleation and 19 (37%)with a second plaque application. There were three recur-rences after the second plaque treatment, and these casesunderwent enucleation. Of the 630 patients, 77 (12%)developed metastasis. Kaplan-Meier estimates demon-strated that the probability of developing metastasis was12% at 5 years and 22% at 10 years (Figure 4). Sixty-eight(11%) patients died of melanoma-related causes and 10(2%) died of other unrelated causes. In Figure 5, Kaplan-Meier estimates show that the probability of developing

melanoma-related mortality was 10% at 5 years and 20% at10 years.

Results of the univariate Cox proportional hazardsregression analyses (Table 3) showed that the factorspredictive of the development of radiation maculopathywere the use of radioisotope Iridium-192 (P 5 .02)compared with Iodine-125, tumor apex dose rate greaterthan 80 cGy per hour (P 5 .003), presence of preoperativesubretinal fluid (P 5 .003), and initial visual acuity worsethan 20/200 compared with better than 20/40 (P 5 .02).The factors predictive of radiation cataract were basal

FIGURE 1. Kaplan-Meier estimates of patients free of radiation maculopathy and papillopathy after plaque radiotherapy for macularchoroidal melanoma.

TABLE 2. Kaplan-Meier Estimates of Probability of Developing Major Radiation Complications and Tumor Events After PlaqueRadiotherapy of Macular Choroidal Melanoma

Outcome Event

Probability at 5 Years

(95% Confidence Intervals)

Probability at 10 Years

(95% Confidence Intervals)

Radiation event

Maculopathy 0.40 (0.36, 0.45) 0.59 (0.53, 0.66)

Cataract 0.32 (0.27, 0.36) 0.59 (0.52, 0.66)

Papillopathy 0.13 (0.10, 0.16) 0.23 (0.16, 0.29)

Vitreous hemorrhage 0.12 (0.06, 0.17) 0.19 (0.13, 0.26)

Vision decrement . 3 Snellen lines 0.40 (0.34, 0.44) 0.80 (0.77, 0.84)

Tumor event

Recurrence 0.09 (0.05, 0.11) 0.12 (0.08, 0.16)

Metastasis 0.12 (0.08, 0.15) 0.22 (0.18, 0.26)

Melanoma-related mortality 0.10 (0.07, 0.13) 0.20 (0.16, 0.24)

AMERICAN JOURNAL OF OPHTHALMOLOGY582 MAY 1999

tumor diameter greater than 10 mm (P 5 .0001), tumorthickness greater than 4 mm (P 5 .0001), and age morethan 60 years (P 5 .005). The development of radiationpapillopathy was significantly related to the use of radio-isotopes Cobalt-60 (P 5 .0001) and Iridium-192 (P 5

.009) compared with Iodine-125, apex dose rate more than80 cGy per hour (P 5 .005), distance of the tumor marginfrom the optic disk less than 2 mm (P 5 .02), and opticdisk dose more than 5,904 cGy (P 5 .04).

The factors predictive of the development of vitreous

FIGURE 2. Kaplan-Meier estimates of patients free of cataract and vitreous hemorrhage after plaque radiotherapy for macularchoroidal melanoma.

FIGURE 3. Kaplan-Meier estimates of patients free of vision decrement by 3 or more Snellen visual acuity lines after plaqueradiotherapy for macular choroidal melanoma.

PLAQUE RADIOTHERAPY OF CHOROIDAL MELANOMAVOL. 127, NO. 5 583

hemorrhage were a tumor with a mushroom shape com-pared with a dome shape (P 5 .004), age greater than 60years (P 5 .01), and tumor thickness greater than 4 mm(P 5 .01). There was no statistically significant risk factor

predictive of the development of neovascular glaucoma.Because of the small number of patients who developedscleral necrosis, the effect of different variables on thedevelopment of this event could not be analyzed statisti-

FIGURE 4. Kaplan-Meier estimates of patients free of recurrence and metastasis after plaque radiotherapy for macular choroidalmelanoma.

FIGURE 5. Kaplan-Meier estimates of patients free of melanoma-related deaths after plaque radiotherapy for macular choroidalmelanoma.

AMERICAN JOURNAL OF OPHTHALMOLOGY584 MAY 1999

cally. Vision decrement of 3 or more Snellen lines wasfound to be related to the use of Cobalt-60 radioisotope(P 5 .003) relative to Iodine-125 and patient age greaterthan 60 years (P 5 .005). Tumor recurrence was signifi-cantly related to a distance of the tumor margin from theoptic disk of less than 2 mm (P 5 .0003) and the presenceof retinal invasion (P 5 .001). The factors predictive ofdevelopment of metastasis were largest basal tumor diam-eter greater than 10 mm (P 5 .0001), tumor thicknessgreater than 4 mm (P 5 .02), apex dose less than 9,120cGy (P 5 .03), and apex dose rate less than 75 cGy perhour (P 5 .03). The risks for melanoma-related death werepresence of metastasis (P 5 .001), tumor thickness greaterthan 4 mm (P 5 .02), and largest basal tumor diametergreater than 10 mm (P 5 .05).

Results of multivariate analyses (Table 4) showed thatthe significant variables for the development of radiationmaculopathy were the use of the radioisotope Iridium-192(P 5 .03) compared with Iodine-125, and preoperative

subretinal fluid (P 5 .04). The independent predictors ofradiation cataract were tumor thickness more than 4 mm(P 5 .0001), age more than 60 years (P 5 .004), andlargest basal tumor diameter more than 10 mm (P 5 .009).The significant predictive factors for radiation papillopathywere the use of radioisotopes Cobalt-60 (P 5 .0001) andIridium-192 (P 5 .008) compared with Iodine-125, and adistance of the tumor margin from the optic disk of lessthan 2 mm (P 5 .04). The important subset of factorspredictive of the development of vitreous hemorrhage werea mushroom tumor shape versus a dome shape (P 5 .03)and age greater than 60 years (P 5 .05). The significantvariables for vision decrement were the use of radioisotopeCobalt-60 versus Iodine-125 (P 5 .002) and age greaterthan 60 years (P 5 .004). The important subset ofpredictors for recurrence were distance of tumor marginfrom the optic disk less than 2 mm (P 5 .003) and retinalinvasion (P 5 .009). The important predictors of metas-tasis were tumor thickness greater than 4 mm (P 5 .02)

TABLE 3. Plaque Radiotherapy for 630 Patients With Macular Choroidal Melanoma: Significant Variables Leading to an OutcomeEvent Using Univariate Cox Proportional Hazards Analyses

Outcome Event Significant Variable P Value

Relative

Risk 95% Confidence Interval

Radiation event

Maculopathy Apex dose rate . 80 cGy/hr .003 1.9 1.2, 2.8

Preoperative subretinal fluid .003 0.5 0.3, 0.8

Radioisotope Iridium-192* .02 0.4 0.2, 0.9

Initial vision , 20/200† .02 0.7 0.5, 0.9

Cataract Largest base . 10 mm .0001 1.1 1.0, 1.2

Thickness . 4 mm .0001 1.1 1.0, 1.2

Age . 60 years .005 1.5 1.1, 1.9

Papillopathy Radioisotope Cobalt-60* .0001 0.2 0.1, 0.4

Apex dose rate . 80 cGy/hr .005 2.4 1.5, 3.9

Radioisotope Iridium-192* .009 0.3 0.1, 0.7

Distance to optic disk , 2 mm .02 0.6 0.4, 0.9

Optic disk dose . 5,904 cGy .04 0.5 0.3, 1.0

Vitreous hemorrhage Mushroom shape‡ .004 2.6 1.4, 4.9

Age . 60 years .01 1.2 1.0, 1.4

Thickness . 4 mm .01 1.1 1.0, 1.3

Vision decrement Radioisotope Cobalt-60* .003 0.6 0.5, 0.9

Age . 60 years .005 1.5 1.2, 1.9

Tumor event

Recurrence Distance to optic disk , 2 mm .0003 0.4 0.2, 0.6

Retinal invasion .001 10.5 2.5, 43.9

Metastasis Largest base . 10 mm .0001 3.0 1.8, 4.9

Thickness . 4 mm .02 1.7 1.1, 2.7

Apex dose , 9,120 cGy .03 0.6 0.4, 0.9

Apex dose rate , 75 cGy/hr .03 0.6 0.4, 0.9

Melanoma-related deaths Metastasis present .001 4.3 2.1, 6.4

Thickness . 4 mm .02 2.2 1.2, 3.3

Largest base . 10 mm .05 4.2 2.1, 6.3

*Compared with Iodine-125.†Compared with visual acuity . 20/40.‡Compared with dome shape.

PLAQUE RADIOTHERAPY OF CHOROIDAL MELANOMAVOL. 127, NO. 5 585

and largest basal tumor diameter greater than 10 mm (P 5.03). Finally, the significant risks for mortality related tomelanoma were the presence of metastasis (P 5 .01) andtumor thickness greater than 4 mm (P 5 .03).

Enucleation was performed in 69 (11%) of 630 patientswith macular choroidal melanoma because of tumor recur-rence in 35 patients (51%), neovascular glaucoma in 20(29%), vitreous hemorrhage in 13 (19%), and scleralnecrosis in one (1%). Tumor recurrence was found to be asignificant predictor of eventual loss of the eye (P 5 .0001)and subsequent metastasis (P 5 .001). Enucleation, how-ever, did not appear to be a statistically significant predic-tor of metastasis.

DISCUSSION

THE MANAGEMENT OF CHOROIDAL MELANOMA WITH

macular involvement is difficult because of the posteriorlocation of the tumor and the potential visual conse-quences of treatment. Treatment options for macularchoroidal melanoma include enucleation, irradiation, laserphotocoagulation, and transpupillary thermotherapy.Plaque radiotherapy has been found to offer satisfactorylocal tumor control and survival for patients with posterioruveal melanoma comparable to charged particle irradia-tion11 and enucleation.5 The application of plaque radio-therapy for macular tumors is especially challenging

because of the posterior site of the tumor, often requiringlocalization with indirect ophthalmoscopy and tedioussurgical manipulation for accurate plaque placement. De-spite the ultimate impairment of central vision in patientswith irradiated macular choroidal melanoma, the eyegenerally remains comfortable, and peripheral vision cansometimes be preserved.

In our study, the development of radiation maculopathywas related to the use of radioisotope Iridium-192 com-pared with Iodine-125, and the presence of preoperativesubretinal fluid. Pretreatment subretinal fluid may contrib-ute to radiation maculopathy, because the fluid may causerelative outer layer retinal ischemia, subsequently com-pounding radiation-induced effects.12 The association ofIridium-192 radioisotope with a higher frequency of radi-ation maculopathy is not unexpected, because the energydelivered by this radioisotope (0.38 MeV) to the surround-ing tissues is greater than that of Iodine-125 (0.032MeV).13 In addition, plaques made with this isotopecannot be adequately shielded, whereas partial shielding ispossible with Iodine-125.

Although the radioisotope Cobalt-60 also has a higherenergy level (1.25 MeV) compared with Iodine-125, it wasnot associated with a substantially higher risk of maculopa-thy. Cobalt-60 was mainly used in the earlier part of thestudy, and many eyes with more posterior tumors may havebeen treated with enucleation at that time. Similar tumorshave been treated with Iodine-125 plaque radiation more

TABLE 4. Plaque Radiotherapy for 630 Patients With Macular Choroidal Melanoma: Significant Factors Related to an OutcomeEvent Using Multivariate Cox Proportional Hazards Analyses

Outcome Event Variables P Value Relative Risk 95% Confidence Interval

Radiation event

Maculopathy Radioisotope Iridium-192* .03 0.5 0.2, 0.9

Preoperative subretinal fluid .04 0.6 0.4, 1.0

Cataract Thickness . 4 mm .0001 1.1 1.1, 1.2

Age . 60 years .004 1.1 1.0, 1.2

Largest base . 10 mm .009 1.0 0.9, 1.1

Papillopathy Radioisotope Cobalt-60* .0001 0.2 0.1, 0.4

Radioisotope Iridium-192* .008 0.3 0.1, 0.7

Distance to optic disk , 2 mm .04 0.8 0.5, 1.0

Vitreous hemorrhage Mushroom shape† .03 1.8 1.1, 3.1

Age . 60 years .05 1.1 1.0, 1.4

Vision decrement Radioisotope Cobalt-60 .002 0.6 0.5, 0.8

Age . 60 years .004 1.1 1.0, 1.2

Tumor event

Recurrence Distance to optic disk , 2 mm .003 0.5 0.3, 1.0

Retinal invasion .009 7.6 1.7, 34.2

Metastasis Thickness . 4 mm .02 1.1 1.0, 1.2

Largest base . 10 mm .03 1.2 1.0, 1.5

Melanoma-related deaths Metastasis present .01 2.5 1.2, 3.8

Thickness . 4 mm .03 1.4 1.2, 1.6

*Compared with Iodine-125†Compared with dome shape.

AMERICAN JOURNAL OF OPHTHALMOLOGY586 MAY 1999

recently, altering the relative frequency of radiation macu-lopathy compared with Cobalt-60.

As an illustration of the differences among the variousradioisotopes, consider a choroidal melanoma 5 mm inthickness treated with an apical dose of 8,000 cGy andbase dose of 35,000 cGy. The calculated dose of radiationat 3 mm from the tumor margin would be 85% to 90% ofthe apical dose (7,200 cGy) with Cobalt-60, 75% to 80%(6,400 cGy) with Iridium-192, and 60% to 70% (5,600cGy) with Iodine-125. These doses vary with the tumorbase dimensions in that the larger the base, the higher thedose to the surrounding tissues. Although the energy levelof Ruthenium-106 is high, it is a b emitter, and its energy(3.5 MeV) decreases faster than Iodine-125, which is aphoton emitter. Therefore, it is not generally used fortumors greater than 5 mm in thickness. Consequently, thenumber of patients treated with Ruthenium-106 in ourstudy is limited, contributing to the lack of significance ofthis radioisotope for radiation complications.

The factors predictive of cataract development afterplaque radiotherapy for choroidal melanoma with macularinvolvement were tumor thickness greater than 4 mm,patient age more than 60 years, and basal tumor diametermore than 10 mm. Previous reports on charged particleirradiation and plaque radiotherapy also found that tumorthickness,14–16 age,14 and basal tumor diameter15 wereassociated with a higher frequency of cataract. The lens isone of the most radiosensitive parts of the eye, and cataractdevelopment is anticipated with a total dose over 1,000cGy.17

Radiation cataract developed at a steady rate throughoutthe follow-up period after plaque radiotherapy of macularchoroidal melanoma, as in other reports.18 Anterior seg-ment complications, such as cataract and neovascularglaucoma, may be slightly greater after charged particleirradiation for choroidal melanoma compared with afterplaque radiotherapy because of the transit of the chargedparticle radiation beam through the lens in its path to thetumor.18,19 In plaque radiotherapy, the radiation is deliv-ered through the sclera to the tumor, allowing onlyspillover radiation to reach the anterior segment. Despitethe remoteness of macular choroidal melanoma from thelens and the posterior irradiation route, radiation cataractis still a common problem after plaque radiotherapy.However, we defined cataract as any degree of lens opacity,usually posterior subcapsular cataract, that was not presentbefore plaque radiotherapy. Most of these were minor,visually insignificant lens opacities and did not requirecataract surgery.

Radiation papillopathy was related to the use of radio-isotopes Cobalt-60 and Iridium-192 compared with Iodine-125 and a distance of the tumor margin from the optic diskof less than 2 mm. Again, in terms of energy consider-ations, Cobalt-60 and Iridium-192 have higher energythan Iodine-125 and provide more irradiation to surround-ing tissues.13 In the univariate analysis, an optic disk dose

greater than 5,904 cGy and a tumor apex dose rate greaterthan 80 cGy per hour were also significantly associatedwith radiation papillopathy. Earlier studies have also indi-cated that radiation papillopathy generally develops withoptic disk doses over 5,000 cGy.20,21 Because of the closeproximity of macular choroidal tumors to the optic disk,the rate of radiation to the tumor apex was correlated withthe development of papillopathy.

The significant predictive factors of vision decrease inthe multivariate analysis were the use of radioisotopeCobalt-60 versus Iodine-125 and patient age more than 60years. Cobalt-60 radioisotope was associated with a higherfrequency of papillopathy, explaining the tendency forvision loss. In a previous report from our department onjuxtapapillary melanoma treated with plaque radiotherapy,a loss of 3 Snellen lines of visual acuity was found in 72%of the patients at 5 years versus only 40% in this study.22

However, all the tumors in the juxtapapillary melanomastudy contacted the optic disk, and many involved thefoveola, explaining the higher frequency of vision loss.

The 5-year recurrence rate of posterior uveal melanomaafter plaque radiotherapy has been reported to be 10% to15%11,20,22–25; our 5-year recurrence rate of 9% for macularchoroidal melanoma is consistent with previous data. Thisis reassuring, because the application of a plaque in themacular region is a tedious, difficult procedure, and accu-racy of placement is important.

The development of recurrence was significantly relatedto a distance of the tumor margin from the optic disk of lessthan 2 mm and retinal invasion on a multivariate level.The association of the increased risk of recurrence withclose location to the optic disk may be related to thechallenges of plaque design and placement in the opticnerve region.24 Retinal invasion can be associated with theseeding of melanoma cells on and under the retina and inthe vitreous cavity, leading to recurrence. However, thestatistical analysis was based on the observation of retinalinvasion in only five patients, which is too few to deriveany definitive conclusions. Finally, similar to previousreports,26 the development of recurrence was found in ourstudy to be an important predictor of the eventual loss ofthe eye and metastasis.

The overall 5-year metastasis rate of posterior uvealmelanoma after plaque radiotherapy and charged particleirradiation is approximately 20%.27,28 Specifically for thechoroidal melanoma with macular involvement, we foundthat the 5-year metastasis rate was 12% and the 10-yearmetastasis rate was 22%. The 5-year melanoma-relatedmortality rate was 10%, and the 10-year mortality rate was20%. Similarly, the 10-year mortality rate from metastaticmelanoma after helium ion irradiation was 23%.18

In the multivariate model, the factors predictive ofmetastasis were tumor thickness greater than 4 mm andbasal tumor diameter greater than 10 mm. The associ-ation of basal tumor diameter and thickness with in-creased risk for metastasis after plaque radiotherapy2,26

PLAQUE RADIOTHERAPY OF CHOROIDAL MELANOMAVOL. 127, NO. 5 587

and charged particle irradiation28 –30 has been recog-nized. As with other cancers, size is an importantpredictor of ultimate metastasis.

The radioisotopes Cobalt-60 and Iridium-192 were sig-nificantly associated with a number of radiation events inthis study, such as maculopathy, papillopathy, and visiondecrease. We have been using Iodine-125 as the radioiso-tope of choice because of its accessibility, ease of shieldingcompared with Cobalt-60 and Iridium-192, good tissuepenetration compared with Ruthenium-106, and ability tobe custom designed into different plaque shapes.13 Evi-dently, radiation complications are fewer with Iodine-125compared with Cobalt-60 and Iridium-192.

We recognize that there are limitations in our study.Most importantly, our study is retrospective and encom-passes trends in tumor management and radiation isotopepreference at a single ocular oncology center. In addition,because of our interest and specialization in ocular cancers,more difficult cases may have been included in the sample.Although most of our patients had seemingly adequatefollow-up, the risk of metastasis may increase with longerfollow-up. Despite these shortcomings, our study highlightsimportant observations regarding plaque radiotherapy ofchoroidal melanoma with macular involvement. Wesought to identify the role of clinical parameters inpredicting radiation complications, tumor recurrence, andmetastasis in choroidal melanoma with macular involve-ment. A comprehensive understanding of the expectedoutcome with plaque radiotherapy is important, becausethis treatment modality is often employed for posterioruveal melanoma. More importantly, a recent study hasshown that the quality of life for patients with posterioruveal melanoma appears to be better with plaque radio-therapy than with enucleation and charged particle irra-diation (Tunc M, Char D, Kroll S, unpublished data,presented as a poster at the American Academy of Oph-thalmology Annual Meeting, October 1997).

In conclusion, plaque radiotherapy offers a 5-year 91%local control rate in macular choroidal melanoma. The riskof metastasis is 12% at 5 years and 22% at 10 years.Radiation complications, including retinopathy, papillopa-thy, and cataract, can occur, leading to a decrease of atleast 3 Snellen lines of visual acuity in 40% of the patients5 years after treatment. Enucleation becomes necessary in11% of the patients because of radiation complications andtumor recurrence.

REFERENCES

1. Shields JA, Shields CL. Intraocular tumors: a text and atlas.Philadelphia: Saunders, 1992:171–205.

2. Shields JA, Shields CL, Donoso LA. Management of poste-rior uveal melanoma. Surv Ophthalmol 1991;36:161–195.

3. Zimmerman LE, McLean IW, Foster WD. Does enucleationof the eye containing a malignant melanoma prevent or

accelerate the dissemination of tumor cells? Br J Ophthalmol1978;62:420–425.

4. Adams KS, Abramson DH, Ellsworth RM, et al. Cobaltplaque versus enucleation for uveal melanoma: comparisonof survival rates. Br J Ophthalmol 1988;72:494–497.

5. Augsburger JJ, Gamel JW, Lauritzen K, Brady LW. Cobalt 60plaque radiotherapy vs enucleation for posterior uveal mel-anoma. Am J Ophthalmol 1990;109:585–592.

6. Seddon JM, Gragoudas ES, Egan KM, et al. Relative survivalrates after alternative therapies for uveal melanoma. Oph-thalmology 1990;97:769–777.

7. Straatsma BR, Fine SL, Earle JD, et al. Enucleation versusplaque irradiation for choroidal melanoma. Ophthalmology1988;95:1000–1004.

8. Shields JA, Shields CL, De Potter P, et al. Plaque radiother-apy of uveal melanoma. Int Ophthalmol Clin 1993;33:129–135.

9. Lee ET. Statistical methods for survival data analysis, 2nd ed.New York: John Wiley and Sons, 1992.

10. Kaplan E, Meier P. Nonparametric estimation from incom-plete observations. J Am Stat Assoc 1958;53:457–481.

11. Char DH, Quivey JM, Castro JR, et al. Helium ion versusiodine 125 brachytherapy in the management of uvealmelanoma. A prospective, randomized, dynamically bal-anced trial. Ophthalmology 1993;100:1547–1554.

12. Foss AJE, Whelehan I, Hungerford JL, et al. Predictivefactors for the development of rubeosis following protonbeam radiotherapy for uveal melanoma. Br J Ophthalmol1997;81:748–754.

13. Shields CL, Shields JA, Gunduz K, Freire JE, Mercado GM.Radiation therapy of uveal malignant melanoma. Ophthal-mic Surg Lasers 1998;29:397–409.

14. Meecham WJ, Char DH, Kroll S, et al. Anterior segmentcomplications after helium ion radiation therapy for uvealmelanoma: radiation cataract. Arch Ophthalmol 1994;112:197–203.

15. Kleineidam M, Augsburger JJ, Hernandez C, et al. Catarac-togenesis after cobalt 60 eye plaque radiotherapy. Int J RadiatOncol Biol Phys 1993;26:625–630.

16. Gragoudas ES, Egan KM, Walsh SM, et al. Lens changesafter proton beam irradiation for uveal melanoma. Am JOphthalmol 1995;119:157–164.

17. Merriam GR, Focht EF. A clinical study of radiation cata-racts and the relationship to dose. Am J Roentgenol 1957;77:759–785.

18. Char DH, Kroll SM, Castro J. Ten-year follow-up of heliumion therapy for uveal melanoma. Am J Ophthalmol 1998;125:81–89.

19. Gragoudas ES. Long-term results after proton irradiation ofuveal melanomas. Graefes Arch Clin Exp Ophthalmol 1997;235:265–267.

20. Lommatzsch PK, Alberti W, Lommatzsch R, Rohrwacher F.Radiation effects on the optic nerve observed after brachy-therapy of choroidal melanomas with 106 Ru/106 Rhplaques. Graefes Arch Clin Exp Ophthalmol 1994;232:482–487.

21. Brown GC, Shields JA, Sanborn G, et al. Radiation opticneuropathy. Ophthalmology 1982;89:1489–1493.

22. De Potter P, Shields CL, Shields JA, et al. Plaque radiother-apy for juxtapapillary choroidal melanoma. Arch Ophthal-mol 1996;114:1357–1365.

23. Karlsson UL, Augsburger JJ, Shields JA, et al. Recurrence ofposterior uveal melanoma after cobalt 60 episcleral plaquetherapy. Ophthalmology 1989;96:382–388.

AMERICAN JOURNAL OF OPHTHALMOLOGY588 MAY 1999

24. Quivey JM, Augsburger JJ, Snelling L, Brady LW. Iodine-125plaque therapy for uveal melanoma. Analysis of the impact oftime and dose factors on local control. Cancer 1996;77:2356–2362.

25. Gunduz K, Shields CL, Shields JA, Cater J, Freire JE,Brady LW. Plaque radiotherapy of uveal melanoma withpredominant ciliary body involvement. Arch Ophthalmol1999;117:170 –177.

26. Vrabec T, Augsburger JJ, Gamel JW, et al. Impact of localtumor relapse on patient survival after cobalt 60 plaqueradiotherapy. Ophthalmology 1991;98:984–988.

27. Augsburger JJ, Gamel JW, Shields JA, et al. Post-irradiation

regression of choroidal melanomas as a risk factor for deathfrom metastatic disease. Ophthalmology 1987;94:1173–1177.

28. Gragoudas ES, Seddon JM, Egan KM, et al. Metastasis fromuveal melanoma after proton beam irradiation. Ophthalmol-ogy 1988;95:992–999.

29. Gragoudas ES, Seddon JM, Egan KM, et al. Prognosticfactors for metastasis following proton beam irradiation ofuveal melanomas. Ophthalmology 1986;93:675–680.

30. Glynn RJ, Seddon JM, Gragoudas ES, et al. Evaluation oftumor regression and other prognostic factors for early andlate metastasis after proton irradiation of uveal melanoma.Ophthalmology 1989;96:1566–1573.

Authors InteractivetWe encourage questions and comments regarding this article via the Interneton Authors Interactivet at http://www.ajo.com/ Questions, comments, andauthor responses are posted.

PLAQUE RADIOTHERAPY OF CHOROIDAL MELANOMAVOL. 127, NO. 5 589