Low-Dose Proton Beam Therapy for Circumscribed Choroidal Hemangiomas

Objective  To evaluate the efficacy and safety of proton beam therapy for complicatedcircumscribed choroidal hemangiomas.

Methods  The study was a retrospective nonrandomized investigation. Seventeenconsecutive patients, referred to the Institut Gustave-Roussy, Villejuif,France, for circumscribed choroidal hemangioma associated with serous retinaldetachment were studied. Each eye received a total dose of 20 cobalt grayequivalents (CGEs) delivered in 15-second fractions of 5 CGEs over 4 days.Functional tests included the initial and final best-corrected visual acuity,slitlamp examination, intraocular pressure, fundus examination, fluoresceinangiography, and indocyanine green angiography. Tumor thickness was determinedon B-scan ultrasonography.

Results  The macula was involved in 7 eyes and the lesion was juxtapapillaryin 2 eyes. The mean (SD) tumor thickness was 3.06 (9.0) mm. The mean initialtumor diameter was 6.82 mm (range, 3.2-12.1 mm). The right eye was involvedin 7 cases and the left eye in 10 cases. The mean (SD) follow-up period was52 (58) months (range, 36-90 months). Retinal reattachment was obtained inall cases after a mean period of 2 months (range, 1-12 months; median, 1 month).Tumor regression was obtained in all cases. One recurrence occurred 1 yearafter the initial treatment in an undertreated area. After re-treatment, however,resolution of the retinal detachment occurred, and flattening of the choroidallesion was obtained. Final visual acuity improved to 2 Snellen lines or morein 16 eyes (94%), was stable in 1 eye, and attained 20/40 or more in 12 eyes(70.6%). No radiation therapy complications occurred during follow-up.

Conclusions  Proton beam therapy for choroidal hemangiomas seems to be an effectiveand safe alternative option. A total dose of 20 CGEs delivered in 4 daily15-second fractions of 5 CGEs seems adequate for local control of both thetumor and serous retinal detachment.

Circumscribed choroidal hemangiomas are rare, discrete vascular hamartomasthat develop in isolated form or are a part of Sturge-Weber syndrome. Theyare frequently located at the posterior pole and characterized by a red-orangedome with distinct, regular borders.^1,2 Theselesions are thought to be congenital but only become symptomatic when nodularthickening of a critical size develops causing serous retinal detachment.^2,3 The median age at detection and diagnosisof circumscribed choroidal hemangiomas in most large series is in the fourthto fifth decades of life. Choroidal hemangiomas are benign and do not transforminto malignant tumors: no treatment is required as long as they remain asymptomatic.However, hemangiomas can cause diminished visual acuity via various mechanismssuch as secondary exudative retinal detachment, cystoid macular edema, alterationof the pigment epithelium of the macular region, subretinal fibrosis, areolarmacular atrophy, and neovascular glaucoma, a rare complication.^4-6 Anytreatment for choroidal hemangiomas should ideally achieve complete resolutionof the subretinal fluid and tumor regression with minimal collateral damage.

The advantage of proton beam irradiation is that a homogeneous dosecan be delivered to the target while sparing the healthy tissue surroundingthe tumor.^5,7,8 Whileradiotherapy may destroy the hemangiomas, late complications of proton beamtherapy including radiation maculopathy and optic neuropathy may limit thetherapeutic benefit.

We present the results of proton therapy (20 cobalt gray equivalents[CGEs] delivered in 15-second fractions of 5 CGEs over 4 days) in 17 consecutivepatients with circumscribed choroidal hemangiomas who had a minimum follow-upof 3 years.

Our series is composed of 17 eyes in 17 consecutive patients (8 womenand 9 men) with isolated circumscribed choroidal hemangiomas who were referredto the Institut Gustave-Roussy, Villejuif, France, between January 1, 1995,and June 30, 2000, and treated with proton beam irradiation. The mean agewas 44.7 years (age range, 34-61 years). Inclusion criterion for proton beamirradiation was a circumscribed choroidal hemangioma complicated by retinaldetachment or macular edema causing alteration of visual acuity. Patientswith Sturge-Weber syndrome were excluded from this study.

The diagnosis of choroidal hemangioma was based on the following criteria:a posteriorly located red-orange choroidal tumor associated with an overlyingserous retinal detachment, a fluorescein angiogram showing early filling ofthe tumor with dye with gradually intense hyperfluorescence and leakage ofdye in the subretinal space during the late phase, early hyperfluorescenceon indocyanine green angiography with a washing-out aspect during the latephase, and B-scan ultrasonography showing a highly reflective hyperechoictumor with no choroidal excavation (Figure1, Figure 2, and Figure 3).

The treatment technique was the same as that used for proton therapyof uveal melanomas.⁹ The tumor was visualizedintraoperatively using indirect ophthalmoloscopy, and 4 tantalum rings weresutured to the sclera using 5/0 Dacron sutures (Alcon Laboratories Inc, ForthWorth, Tex). The 3-dimensional coordinates of the clips were obtained frompairs of x-ray films during a simulation and were then introduced into thetreatment-planning program, which provides a model of the eye and the tumor.A 200-MeV proton beam was degraded to 70 MeV and modulated to administer auniform dose to the target volume. Each eye received a total of 20 CGEs (CGE:physical dose ×1.1 estimated relative biological effectiveness of protonscompared with photons) delivered in four 15-second fractions of 5 CGEs over4 days. During irradiation, the patient's head was immobilized using a stereotactichead-holding device, and voluntary visual fixation was requested. A macularshield was used when the tumor involved the macula. Choroidal hemangiomasdo not represent a malignant condition. Macula-sparing radiotherapy was decidedon and the macula received less than 50% of the total dose. Dosimetry wasestablished to evaluate the individual amount delivered to the macula, theoptic disc, and the lens.

Patients were informed about the study and gave their consent in accordwith the Declaration of Helsinki . The study was not submitted to institutionalreview board approval.

The Pearson-Spearman r test was used for statisticalanalysis.

Follow-up examinations were performed on days 30 and 90, at 6 and 12months after treatment, and yearly thereafter. Ophthalmologic parameters assessedincluded best-corrected visual acuity in the affected eye and fellow eye,tested at a distance of 5 m with standard visual acuity charts, intraocularpressure measurement, slitlamp examination, and indirect ophthalmoscopy. B-scanultrasonography was performed at 3 and 6 months and yearly thereafter. Fluoresceinangiography and indocyanine green angiography were performed yearly. Follow-upexaminations verified that irradiation complications such as cataracts, neovascularglaucoma, and radiation retinopathy had not occurred. The mean (SD) follow-upwas 52 (58) months (range, 36-90 months).

All patients were initially seen with diminished best-corrected visualacuity (Table 1) related to serousretinal detachment (Figure 1). Cystoidmacular edema was present in 1 eye. The time elapsed between the onset ofthe first symptoms and treatment varied between 1 month and 7 years (mean,11 months; median, 4 months); none of the eyes had undergone previous lasertreatment.

The clinical characteristics of the 17 hemangiomas are detailed in the Table 1. Most eyes were initially seenwith a tumor located in the temporal part of the choroid (15 eyes), the maculawas involved in 7 eyes, and the lesion was juxtapapillary in 2 eyes. The mean(SD) tumor thickness was 3.6 (0.9) mm. The mean initial tumor diameter was6.82 mm (range, 3.2-12.1 mm). The right eye was involved in 7 cases and theleft eye in 10 cases.

The retina reattached within 1 month after treatment in 11 eyes and,in 3 patients within 3 months; in 2 patients, a flat retina was seen 6 monthsafter treatment. Retinal detachment disappeared in 1 patient after 12 months(mean, 2 months; median, 1 month). A decrease in tumor thickness was observedin all eyes. Tumors lost at least 50% of their initial thickness 6 monthsafter treatment in 10 eyes (47%) (Figure 2). Twelve months after treatment tumors regressed completely in8 eyes and a flat scar 0.4 to 1.6 mm thick remained in 6 eyes. A recurrenceoccurred in 1 eye, 1 year after treatment. The hemangioma was located in thejuxtapapillary area. This region did not receive irradiation at the initialtreatment. Thickening of the juxtapapillary hemangioma occurred and was complicatedby a cystoid macular edema. Proton beam irradiation was successfully deliveredto the juxtapapillary area. One year after re-treatment the lesion had completelydisappeared and cystoid macular edema had regressed.

One month after treatment visual acuity improved to 2 Snellen linesor more in 9 patients, was stable in 6 eyes, and decreased in 2 eyes. Twoyears after treatment visual acuity improved to 2 Snellen lines or more in16 eyes, was stable in 1 eye (Figure 4),and reached 20/40 or more in 12 eyes (70.6%) (Table 1). We did not observe a decrease of visual acuity.

Two years after treatment visual acuity was significantly correlatedwith initial visual acuity (Pearson-Spearman r test,0.77, P<.001). The time elapsed between the onsetof the first symptoms and treatment was significantly correlated with finalvisual acuity (Pearson-Spearman r test, −0.60, P = .02). The patients treated earlier tended to have abetter final visual acuity.

Initial visual impairment seems to be correlated with the location ofthe hemangioma. When the tumor manifests a subfoveolar extension, initialvisual acuity was less than 20/50. However, tumor thickness was not correlatedwith initial visual acuity. Three patients exhibited some granularity of theretinal pigment epithelium 4 years after treatment, and mild subretinal fibrosiswas present at the periphery of the site occupied by the original tumor in1 eye, 3 years after irradiation. No visual impairment was reported and noradiation treatment complications, such as cataract, irradiation retinopathy,or neovascular glaucoma, were observed during follow-up.

Usually, choroidal hemangiomas are not progressive tumors, but slightenlargements have been reported, albeit rarely.^10,11 Althoughchoroidal hemangiomas are usually asymptomatic and require no treatment, theymay be sight threatening when they involve the macula or when they are associatedwith serous retinal detachment or macular edema and, thus, may constitutea therapeutic dilemma. The spectrum of treatment options is limited when tumorsare centrally located and in close proximity to the optic nerve and the fovea.

Laser therapy has been recommended in eyes with visual impairment dueto an exudative retinal detachment overlying a choroidal hemangioma, the aimbeing to induce chorioretinal adhesion thereby enabling retinal detachment.However, because laser therapy failed to achieve tumor eradication, it isno longer recommended for this disorder.^7,12,13 Thepersistence of abnormal vascular tissue could, in fact, cause the recurrenceof subretinal fluid with progressive impairment of vision. Such therapy achievedretinal reattachment in 82.5% of the eyes, but relapses were reported in 51.5%of the patients within a mean period of 24 months.¹² Moreover,even when modified to a mild scatter technique, laser photocoagulation failedto maintain visual acuity above 20/50 in 75% of treated eyes¹⁴ andirreversible scotoma may occur when the choroidal lesion is close to the opticdisc or the macula.¹² Such a procedure is appliedeasily, except in cases of bullous retinal detachments that occasionally haverequired drainage of subretinal fluid before photocoagulation.¹²

Low doses of fractionated external irradiation or brachytherapy areboth effective and achieve retinal reattachment and tumor regression^15-18 Externalx-ray irradiation delivers a virtually homogeneous dose to the entire choroid.Brachytherapy allows more precise irradiation of the hemangioma, but the doseto the tumor is not homogeneous, with the base receiving a higher dose thanthe apex, and suturing the radioactive plaque posteriorly may be difficult.

Radiation therapy–induced retinopathy following fractionated treatmentis not expected with doses below 30 to 40 Gy. The late carcinogenic effectof irradiation with an increased risk of developing radiation therapy–inducedmalignant tumors is well known.¹⁹ Radiotherapyfor benign lesions should, therefore, consider each individual case carefullyand if unavoidable, irradiation of the smallest area is preferable.

Recently stereotactic radiotherapy with a linear accelerator was proposedas an alternative treatment for circumscribed choroidal hemangiomas.²⁰ Kivela et al²⁰ demonstratedresolution of retinal detachment within 6 to 20 months (median, 5 months)and a median decreased tumor height of 24% after 6 months. In our study disappearanceof subretinal fluid usually occurred within 1 to 12 months following treatment(median, 1 month) and tumor height decreased 50% after 6 months, which comparedfavorably with the study of Kivela et al.²⁰

Proponents of transpupillary thermotherapy for primary treatment ofcircumscribed hemangiomas seemed to have gained ground.^8,9,21 Althoughsuccessful tumor regressions have been reported, they were often incompletewith conventional transpupillary thermotherapy. The mild thermal effect oftranspupillary thermotherapy was found to be the cause of complete chorioretinalatrophy with a lack of photoreceptor recovery and an optimal visual outcomeof only 20/50.⁸

Photodynamic therapy is not dependent on a thermal effect but ratheron a photochemical effect. Barbazetta and Schmidt-Erfurth²² andRobertson²³ proposed photodynamic therapy asan attractive treatment option for choroidal hemangiomas. However, only 5cases were reported with a limited follow-up. More recently Schmidt-Erfurthet al²⁴ and Porrini et al²⁵ corroboratedtheir results in a prospective study. In age-related macular degeneration,the photosensitizer used is verteporfin, which accumulates preferentiallyin choroidal new vessels. However, it is thought that choroidal hemangiomasdo not grow. Photodynamic therapy does not specifically target this type ofnonproliferative vessel of choroidal origin. The lack of neovascular selectivitymay lead to focal overtreatment of large vascular lesions. Occlusion of thephysiologic choroid may result from too many re-treatments, and an intervalthat is too short between the subsequent re-treatments. The advantage of protonbeam irradiation is that a homogeneous dose can be delivered to the targetwhile sparing the healthy tissue surrounding the tumor.^5,16,26

In 1997, we reported on 13 cases with choroidal hemangiomas in whichserous retinal detachment was treated with proton beam irradiation. The totaldose delivered was 30 CGEs. Retinal reattachment was obtained in all casesafter a mean period of 52 days, tumor height regressed in all cases, and visualacuity improved to 2 Snellen lines or more in 8 (62%) of 13 eyes and attained20/200 in 9 (69%) of 13 eyes.²⁷ However, themean duration of symptoms was 3 years. Four of the patients had previouslyundergone laser photocoagulation and had experienced retinal detachment recurrences.In this study, the total dose delivered was 20 CGEs in 15-second fractionsof 5 CGEs over 4 days, regression of retinal detachment was obtained in allcases after a mean period of 2 months, tumor height regressed in all cases,and tumor flattening was obtained in 14 (82.0%) of 17 eyes. Visual acuityimproved to 2 Snellen lines or more in 16 (94.0%) of 17 eyes and attained20/40 in 12 (70.6%) of 17 eyes. However, the mean time elapsed before treatmentwas 11 months and no eyes had received any previous treatment. Better visualimprovement should not be the only reason for reducing the total dose delivered;more adverse effects can also be avoided. Zografos et al²⁸ claimedthat a total dose of 27.3 Gy induced radiation optic neuropathy, which reducedvisual acuity and that a dose of 22.7 Gy can induce some retinal telangiectasis5 to 7 years after irradiation. Doses lower than 18.2 Gy appear to be saferand tumor regression can be achieved within 2 to 4 years. Our results do notcorroborate those of Lee and Hungerford²⁹ whoreported optic neuropathy and maculopathy in 2 patients treated with 18 GyCGE within 32 months of proton beam irradiation. The risk for radiation-relatedcomplications depends on factors such as total dose, fractionation, and treatedvolume. Proton beam therapy allows reducing the irradiated volume.

The respective efficacy and safety and safety of photodynamic therapyand proton beam therapy should be compared through large randomized, prospectivestudies. This approach seems difficult because of the reduced prevalence ofchoroidal hemangiomas.

Although it is associated with inconvenience such as the need for surgeryand its cost, proton beam irradiation (20 CGE) seems to be an attractive andeffective option for treating circumscribed choroidal hemangiomas with exudativeretinal detachment. Our study results indicate that the dose we chose is beneficial,especially in eyes with subfoveal involvement, and it is also tolerated atthe posterior pole.

Correspondence: Eric Frau, MD, Department of Ophthalmology, CHU deBicêtre, 78 avenue du Général Leclerc, Le Kremilin BicetreCedex 94270, France (mlfrau@noos.fr).

Submitted for publication August 26, 2003; final revision received March29, 2004; accepted May 7, 2004.