Photodynamic Therapy for Occult Choroidal Neovascularization With PigmentEpithelium Detachment in Age-Related Macular Degeneration

Objective  To study the visual and angiographic outcome of eyes with neovascularage-related macular degeneration associated with pigment epithelium detachment(PED) treated by photodynamic therapy.

Methods  Review of the medical charts and the fluorescein and indocyanine greenangiograms of all consecutive patients with age-related macular degenerationassociated with choroidal neovascularization and serous PED of at least 1disc diameter, who received photodynamic therapy from January 1, 2000, toAugust 31, 2002.

Results  Thirty patients (34 eyes) met the study criteria. Each underwent 1 to8 treatments (mean, 4); duration of follow-up was 12 to 36 months (mean, 19months). Nineteen eyes (56%) lost 3 or more Snellen lines of visual acuity,7 eyes (21%) lost 1 or 2 lines, 6 eyes (18%) maintained their initial acuity,and 2 eyes (6%) gained 1 or 2 lines. Subretinal hemorrhage occurred in 5 eyesand retinal pigment epithelium tears in 4 eyes. In 4 eyes, visual acuity decreasedto counting fingers, hand motions, or light perception.

Conclusions  Although 44% of the 34 eyes with age-related macular degeneration andPED lost fewer than 3 Snellen lines in acuity, severe visual loss to countingfingers or less occurred in 4 eyes, 3 of them with choroidal neovascularizationinside the PED. Further studies and treatment modalities are required to improveprognosis of neovascular age-related macular degeneration with serous PED.

Photodynamic therapy (PDT) with verteporfin has been shown to be beneficialfor the treatment of predominantly classic^1,2 andoccult with no classic³ subfoveal choroidalneovascularization (CNV) secondary to age-related macular degeneration (AMD).The Macular Photocoagulation Study (MPS)⁴ recognized2 forms of occult CNV: fibrovascular pigment epithelium detachment (PED) andleakage of undetermined source. Both forms were included as a single groupin the Verteporfin in Photodynamic Therapy study.³ Tothe best of our knowledge, a separate subgroup analysis of the value of PDTin eyes with neovascular AMD with serous PED associated with CNV has not yetbeen published.

We herein report the visual and angiographic outcome of a consecutiveseries of patients with subfoveal occult CNV associated with serous PED ofat least 1 disc diameter, who received PDT in a clinical setting.

The medical charts and the fluorescein (FA) and indocyanine green (ICGA)angiograms of all consecutive patients with AMD associated with CNV and serousPED who received PDT from January 1, 2000, to August 31, 2002, were reviewed.The study was approved by the Legal Department of the Mor Institute for MedicalData, Bnei Brak, Israel, and all the patients signed an informed consent.The PDT was administered according to the protocol of the Treatment of Age-RelatedMacular Degeneration With Photodynamic Therapy study,^1,2 withfollow-up every 3 months for at least 12 months (to September 2003). The needfor repeated treatment was determined by the clinician, usually after 3 months.The following data were collected: age and sex; best-corrected Snellen visualacuity (VA) at initial examination and at the end of the follow-up period;number of PDT sessions; and adverse effects of PDT. The FAs and the ICGAswere reviewed by 2 physicians (R.A.-S. and R.E.). The initial diagnosis ofCNV associated with serous PED of at least 1 disc diameter was made by FA,and further analysis was performed with ICGA. The CNV was divided into 3 categoriesby location with respect to the PED, as suggested by Yannuzzi et al⁵: beneath, CNV located beneaththe PED and entirely within its boundaries (Figure 1); contiguous, CNV that interruptedthe continuity of the margin of the serous PED (Figure 2); and remote, CNV distinctly separatefrom the edge of the serous PED (Figure 3).

Correlations were calculated between the initial and final best-correctedSnellen VA, as well as the change in VA after PDT, and the difference in sizeof the PED, the location of the CNV, and the presence of retinal pigment epithelial(RPE) tear and subretinal hemorrhage.

The χ² test, Fisher exact test, Pearson correlation,and analysis of variance were performed, as appropriate, with the use of SPSSsoftware (version 10, Professional Statistics Release; SPSS Inc, Chicago,Ill).

The cohort included 30 patients (34 eyes) with neovascular AMD associatedwith PED. There were 18 men and 12 women aged 60 to 90 years (mean, 71.3 years).The number of treatments was 1 to 8 (mean, 4 treatments), and the follow-upperiod ranged from 12 to 36 months (mean, 19 months).

The best-corrected Snellen VA results are presented in Table 1 and Table 2.An improved VA of 1 or 2 Snellen lines was noted in 2 eyes (6%), no changein VA in 6 eyes (18%), a decrease of 1 or 2 lines in 7 eyes (21%), and a decreaseof more than 3 lines in 19 eyes (56%).

The serous PED composed more than 50% of the lesion in 31 of 34 eyes.Seventeen eyes had CNV beneath the PED, 9 had remote CNV, and 8 had contiguousCNV. The eyes with CNV beneath the PED had the worst VA outcome, with a 65%rate (11/17) of visual loss of 3 lines or more, compared with 44% (4/9) forthe group with remote CNV and 50% (4/8) for the group with contiguous CNV.The correlation between the location of the CNV and the change in VA did notreach statistical significance (P = .60, χ² test). Four patients had a final VA of counting fingers, hand motions,or light perception, of whom 3 had CNV beneath the PED and 1 had contiguousCNV. The profound visual loss was due to subretinal hemorrhage in 3 patients(one of whom had an RPE tear as well) and to disciform scar in another patient.

An RPE tear occurred in 4 eyes after 1 or 2 treatments (Table 3). Three of these eyes had CNV beneath the PED, and 1 hadcontiguous CNV. The VA of these eyes ranged from 20/30 to hand motions. Oneadditional patient had an extrafoveal RPE tear before PDT, and his initialVA of 20/80 remained relatively stable at 20/100 after treatment and throughout15 months' follow-up.

Of the remaining 29 eyes, the total lesion size increased in 14 eyes(48%), did not change in 11 (38%), and decreased in 4 (14%) (Table 4). No correlation was found between the change in lesionsize and the change in VA (P = .25, χ² test).

Subretinal hemorrhage occurred in 5 eyes, 3 of them with CNV beneaththe PED, 1 with remote CNV, and 1 with contiguous CNV. Three of these patientshad a final VA of hand motions or light perception, 1 patient had a VA of20/800, and 1 patient had a VA of 20/200. There was a statistically significantcorrelation between the presence of subretinal hemorrhage and profound visualloss (P = .03, Fisher exact test).

Four eyes had retinochoroidal anastomoses on ICGA. The VA decreasedfrom 20/60 to 20/150, from 20/25 to 20/50, and from 20/50 to 20/300 in 1 eyeeach. In the remaining eye, final VA was light perception.

No statistically significant correlation was found between the followingvariables: initial and final VA; initial VA and subretinal hemorrhage; RPEtear and final VA; final VA and location of CNV; and subretinal hemorrhageand location of CNV (analysis of variance).

The findings of the present study indicate that PDT may be of questionablebenefit in eyes with AMD with occult CNV and serous PED. According to thedefinitions proposed by the MPS,⁴ there areessentially 2 forms of occult CNV: fibrovascular PED and leakage of undeterminedsource (without PED). The MPS guidelines (which were used in the Verteporfinin Photodynamic Therapy study³) distinguishbetween the fibrovascular PED and the typical classic serous detachments ofthe RPE, which may or may not have detectable CNV on FA. Further modificationof the definition of subfoveal neovascular lesions with PED was suggestedby Yannuzzi et al,⁵ who introduced the term vascularized PED to describe the combination of CNV andserous PED of at least 1 disc diameter. This description differs from theMPS terminology, since it includes ICGA findings, whereas the MPS guidelines(which were used in the verteporfin studies^1-3)include only FA features. On ICGA, the serous component of the PED is hypofluorescent,whereas the vascularized component is hyperfluorescent, while on FA, bothcomponents demonstrate late hyperfluorescence or leakage. Our series includedeyes with CNV associated with serous PED of at least 1 disc diameter, andmost of the eyes (31/34) included serous PED that was larger than 50% of thelesion (eyes ineligible for verteporfin treatment³).In our study, the diagnosis of exudative AMD was made on the basis of clinicalexamination, and the presence of occult CNV with serous PED was determinedby FA. Pigment epithelium detachment larger than 1 disc diameter associatedwith CNV was then analyzed by confocal ICGA. On FA, all pigment epithelialelevations, both serous and vascular, stain, whereas on confocal ICGA, theserous PED remains totally dark and the area of hyperfluorescence localizesthe neovascularization, thereby segregating the serous and vascularized components.Several reports on the clinicopathological correlation of the hyperfluorescenceon ICGA and histopathological proof of CNV support the localization of theCNV at the hyperfluorescent area.^6-8

Vascularized PED (serous PED with CNV) accounts for an estimated 26%to 31% of all newly diagnosed cases of exudative AMD.^5,9 Itsnatural course is poor because of disciform scar formation or RPE tears.^9-12 Previousstudies on ICGA-guided laser treatment of vascularized PED yielded disappointingvisual results.^13-15 Slakteret al¹³ reported anatomic success in only 43%of eyes with vascularized PED compared with 66% in eyes with occult CNV notassociated with serous PED. Lim et al¹⁴ noteda deterioration in VA in 82% of 20 patients treated by ICGA-guided laser,and Brancato et al¹⁵ found a decrease in VAafter ICGA-guided laser in 82% of eyes with CNV beneath the PED and in 62.5%of eyes with CNV at the margin of the PED.

The Verteporfin in Photodynamic Therapy study reported that PDT is beneficialin eyes with subfoveal occult CNV with no classic CNV.³ Thelesions included in the Verteporfin in Photodynamic Therapy study could containfeatures that obscured the distinction of classic and occult CNV on FA, namely,blood, hypofluorescence due to sources other than visible blood, and serousPED. Both type 1 (fibrovascular PED) and type 2 (leakage of undetermined source)occult CNV were treated, and no separate analysis of these 2 subgroups wasdone. Eyes with serous PED occupying more than 50% of the lesion were ineligiblefor verteporfin. Consequently, there are currently no explicit recommendationsfor PDT with verteporfin for eyes with subfoveal CNV and serous PED, especiallyin which the main component of the lesion is serous PED.

To the best of our knowledge, there are as yet no peer-reviewed publishedstudies on series of patients with CNV and serous PED treated with PDT. However,unpublished data and abstracts are available: Pece and Brancato¹⁶ reportedon 32 PDT-treated eyes of 28 patients with CNV associated with PED followedup for a mean duration of 5.7 months, and Moisseiev et al¹⁷ reportedon 15 patients with vascularized PED in AMD, in lesions where PED is the predominantcomponent, who received PDT, with a follow-up of 3 to 19 months. Lommatzschet al¹⁸ reported decreased VA of more than3 lines in 10 (83%) of 12 eyes with vascularized PED treated with PDT, within3 to 9 months after treatment. Copt and Zografos¹⁹ showedthat 18 (90%) of 20 eyes with vascularized PED who received PDT lost morethan 15 letters of VA from baseline. The results of these series are presentedin Table 5. The visual outcomereported in our cohort was decreased VA of 3 or more lines in 56% of the patients,during a follow-up period of 12 to 36 months (mean, 19 months). Our resultsare similar to those of the smaller series by Moisseiev et al,¹⁷ worsethan those of the larger series by Pece and Brancato,¹⁶ andbetter than those of the series by Lommatzsch et al¹⁸ andCopt and Zografos.¹⁹ Although caution mustbe exercised when different samples are compared, some of the differencesmay be attributed to the variability in the follow-up periods. It is importantto note as well that best-corrected VA from Early Treatment Diabetic RetinopathyStudy charts was used in clinical trials for PDT,^1-3 whereasbest-corrected Snellen VA was used in our retrospective series. Since changesin Snellen lines do not equate with changes in lines on Early Treatment DiabeticRetinopathy Study charts, further caution must be exercised when differentseries are compared.

The anatomic response to PDT in our patients was disappointing. Thelesion size decreased in only 4 eyes, remained stable in 11, and increasedin 14 (Table 4).

Tears of RPE occurred in 4 eyes (12%) (Table 3), compared with the rate of 12% in the study by Pece andBrancato,¹⁶ 33% in the series by Lommatzschet al,¹⁸ and 15% in the study by Copt and Zografos.¹⁹ One of our patients with an RPE tear sustained profoundvisual loss to hand motions because of subretinal hemorrhage, and the other3 patients had a final VA of 20/30 to 20/200. It is not clear whether theseRPE tears are complications of the treatment or represent the natural courseof the disease.

Acute RPE tears may occur as a natural complication of CNV associatedwith PED, or after laser treatment, because of contraction of the underlyingCNV or the pressure caused by the sub-RPE fluid on the taut serous PED.^20-22 The visual impactof the tear depends on the foveal involvement of the rupture or on the presenceof hemorrhages. Several studies reported increased rates of RPE tear afterPDT with verteporfin for eyes with CNV associated with PEDs,^18,19,23,24 butthe mechanism of RPE tear after PDT remains unknown. Immediate optical coherencetomography performed after PDT for subfoveal CNV showed subretinal and intraretinalfluid accumulation caused by leakage from the neovascular complex, which wasconfirmed by ICGA.²⁵ The fluid accumulationstarted 120 minutes after PDT and was maintained for up to 5 days. It is thereforepossible that this fluid beneath the PED exerts additional tension on thetaut RPE, leading to RPE tear.

Subretinal hemorrhage occurred in 5 eyes (one of them with an RPE tear),causing profound visual loss in 4 of them. Three of the eyes had a final VAof hand motions or light perception, one had a VA of 20/800, and one had aVA of 20/200. There was a statistically significant correlation between thepresence of subretinal hemorrhage and profound visual loss (P = .03, Fisher exact test).

In conclusion, the visual and anatomic results of our series show that,while 15 (44%) of the 34 eyes lost fewer than 3 lines of Snellen VA afterPDT, 4 eyes (12%) sustained profound visual loss, 3 of them because of massivesubretinal hemorrhage and 1 because of RPE tear. In addition, 4 eyes (12%)developed RPE tear. No correlation was found between the location of the CNVor the lesion size and the final VA. Although the small sample size precludesdefinitive recommendations, our study shows a high rate of visual and anatomiccomplications after PDT in eyes with CNV associated with serous PED.

Corresponding author and reprints: Ruth Axer-Siegel, MD, Departmentof Ophthalmology, Rabin Medical Center, Beilinson Campus, Petah Tikva 49 100,Israel (e-mail: seegs@netvision.net.il).

Submitted for publication July 23, 2003; final revision received October31, 2003; accepted December 3, 2003.

This study was presented in part at the meeting of the American Societyof Retinal Specialists; August 18, 2003; New York, NY.