Retinal pigment epithelial cell loss assessed by fundus autofluorescence imaging in neovascular age-related macular degeneration

Retinal Pigment Epithelial Cell LossAssessed by Fundus AutofluorescenceImaging in Neovascular Age-related MacularDegeneration

Nishant Kumar, MD, Sarah Mrejen, MD, Adrian Tien-Chin Fung, MMed, Marcela Marsiglia, MD, PhD,Boon K. Loh, MD, Richard F. Spaide, MD

Purpose: To characterize retinal pigment epithelial (RPE) cell loss as evidenced by autofluorescence imagingin patients with neovascular age-related macular degeneration (AMD).

Design: Retrospective cohort study.Participants: There were 162 eyes of 116 consecutive patients with neovascular AMD examined in a retinal

practice.Methods: Each patient underwent a complete examination including autofluorescence imaging. Areas of

confluent absence of autofluorescence signal of at least 0.5 mm in greatest linear diameter were measured withinthe macular area. Patient demographic and examination data were evaluated in relation to the autofluorescencedata.

Main Outcome Measures: Prevalence and progression of confluent areas of absent autofluorescence andthe relationship these areas had with visual acuity.

Results: The mean age of the patients was 82.9 years, and the mean visual acuity was 20/71 (logarithmminimum angle of resolution [logMAR], 0.55). Confluent loss of autofluorescence was seen in 58.6% of eyes atbaseline, and the median area of absent autofluorescence among those was 1.57 mm2 (interquartile range [IQR],0.62–4.32 mm2). Using generalized estimation equation modeling, the significant predictors for area of confluentabsent autofluorescence at baseline were duration of disease and any previous treatment with photodynamictherapy. The significant predictor of baseline visual acuity was baseline area of confluent absent autofluores-cence. Follow-up was available for 124 (76.5%) eyes, with a mean follow-up of 2.9 years. By then, the meanvisual acuity was 20/90 (logMAR, 0.65), and 79% of eyes had confluent areas of absent autofluorescence, thelarge majority of which affected the central macula. The median area of absent autofluorescence was 3.61 mm2

(IQR, 1.16–7.11 mm2). The best predictor of final visual acuity was the area of absent autofluorescence at the finalfollow-up.

Conclusions: Confluent absence of autofluorescence, a measure signifying RPE loss, was a significantpredictor of visual acuity both at baseline and at final follow-up. This is the first study to document the prevalence,rate of progression, and factors associated with measures of confluent RPE loss in patients with neovascularAMD. Application of strategies to limit RPE cell loss may prove useful in eyes with neovascular AMD.

Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references.Ophthalmology 2013;120:334–341 © 2013 by the American Academy of Ophthalmology.

TafdifaAod

Age-related macular degeneration (AMD) is the most com-mon cause of central vision loss among older individuals indeveloped countries.1 Conventionally, AMD has been di-vided into 2 main types. The nonexudative form has drusenand pigmentary changes as the principle initial manifesta-tions. Over time, some eyes progress to demonstrate a latemanifestation of nonexudative disease known as geographicatrophy (GA). Although the entity is called atrophy, thesalient histopathologic feature is death of affected retinalpigment epithelial (RPE) cells.2 With loss of the RPE cells

334 © 2013 by the American Academy of OphthalmologyPublished by Elsevier Inc.

he first arises from the absence of melanin and results inppearance of GA as seen by ophthalmoscopy or in colorundus photographs: a discrete circular or oval area ofecreased pigmentation in which there is improved visual-zation of the underlying choroidal vessels.3 The secondeature that occurs with the demise of the RPE cells is thessociated loss of lipofuscin these cells ordinarily contain.utofluorescence imaging relies on the stimulated emissionf fluorescence from lipofuscin,4–8 so regions of GA can beetected and measured as areas of absent autofluorescence

ISSN 0161-6420/13/$–see front matterhttp://dx.doi.org/10.1016/j.ophtha.2012.07.076

wacchsaAfltbscomnvdfearttut

cpBadiDorob

alctwisaea0waraaac2amt(ufA

Kumar et al � RPE Loss in Neovascular AMD

is easy to visualize, and so clinical trials often use this typeof imaging as the primary way to monitor eyes over time.Proposed etiologic mechanisms for the development of GAinclude ischemia, senescence, oxidative and photoxidativedamage, inflammation, and improper RNA processing.9–12

Loss of RPE cell function in turn can lead to choriocapillarisloss. There is no proven therapy for GA.

The second major clinical phenotype, neovascular AMD,is manifested by the invasion of tissue containing newvessels, called choroidal neovascularization (CNV), into thesub-RPE and subretinal space. In exudative AMD there areleakage, hemorrhage, and scarring, with the ultimate ex-pression being a disciform scar in untreated cases. This wasthought to be a stable lesion until Sarks et al13 described asmall series of eyes with disciform scars in which a ring ofatrophy developed and expanded around the scar. The pro-posed etiologic stimuli for the development of CNV aremuch the same as for GA and include ischemia, senescence,oxidative and photoxidative damage, and inflammation.9–11

The treatment of exudative AMD has been revolutionizedby intravitreal injections of agents directed against vascularendothelial growth factor (VEGF). In 2 large phase 3 trialsevaluating intravitreal ranibizumab (Lucentis; Genentech,South San Francisco, CA), ANCHOR (Anti-VEGF Antibodyfor the Treatment of Predominantly Classic Choroidal Neo-vascularization in Age-related Macular Degeneration)and MARINA (Minimally Classic/Occult Trial of the Anti-VEGF Antibody Ranibizumab in the Treatment of Neovascu-lar Age-Related Macular Degeneration), 95% of eyes achievedvisual stabilization and at least 30% attained visual improve-ment at 1 year.14,15 By the end of the second year, theproportion of eyes losing 3 lines of vision increased from5% to approximately 10%.16 Rosenfeld et al17 later evalu-ated color fundus photographs of eyes that lost 3 lines ormore of vision in ANCHOR and MARINA and comparedthem with those that gained 3 lines or more of vision. Anincreased area of RPE abnormality was found to be associ-ated with visual acuity loss, but the area of what they calledGA was not. However, it is not obvious how loss of RPEcells in eyes with potential scarring and other manifestationsof exudative AMD could be evaluated from color photo-graphs using conventional definitions of GA.

The amount of RPE loss could be evaluated by autofluo-rescence imaging without using indirect measures such asthe improved ability to visualize choroidal vessels, whichmay be a problematic end point because of concurrentintervening scarring and neovascular tissue. The prevalenceof loss, rate of enlargement over time, and influence of RPEloss on visual acuity in eyes with CNV is not known, butcould be determined with the help of autofluorescence im-aging. Therefore, this study was performed to investigateRPE loss as evidenced by autofluorescence imaging in eyeswith neovascular AMD.

Patients and Methods

The study had institutional review board approval by the WesternInstitutional Review Board, Olympia, Washington, complied withthe Health Insurance Portability and Accountability Act of 1996,

ith neovascular AMD examined in the month of September 2011t Vitreous Retina and Macula Consultants New York (under theare of RFS) were eligible. All patients had signed an informedonsent for study participation. All patients underwent a compre-ensive ophthalmologic examination including best-corrected vi-ual acuity, color fundus photography, and autofluorescence im-ging using the Topcon TRC-50DX Retinal Camera (Topconmerica, Paramus, NJ) as part of their routine evaluation. Auto-uorescence images were obtained using filters that had an exci-

ation light bandwidth from 535 to 585 nm and a barrier filterandwidth from 605 to 715 nm. Patients with neovascularizationecondary to myopia or infective, inflammatory, or hereditaryauses were excluded. Other exclusion criteria included high my-pia (spherical equivalent more than �6 diopters), previous ther-al laser photocoagulation for neovascular AMD, diabetic reti-

opathy or maculopathy, previous retinal vascular occlusion,itreoretinal or glaucoma surgery, macular hole, history of retinaletachment, presence of an RPE tear at baseline or at the lastollow-up visit, and central serous chorioretinopathy. Additionalxclusionary criteria were media opacities significant enough toffect quality of imaging, photographic artifacts, and retinal, sub-etinal, or sub-RPE hemorrhages that obscured evaluation of au-ofluorescence images at baseline or final follow-up. Only 3 pa-ients with exudative AMD evaluated in this period had notndergone autofluorescence imaging and therefore were not en-ered into the study.

Demographic data including age, gender, smoking status (past,urrent, or never smoked), patient’s lens status (phakic, pseudo-hakic, or aphakic), and history of diabetes mellitus were recorded.aseline in this study was defined as the visit with the firstutofluorescence image that could be either on the visit CNV wasiagnosed or after treatment for it had commenced. The final visitn this study was the date with the last autofluorescence image.ata collected included visual acuity at each visit analyzed; detailsf treatments with photodynamic therapy; and any intravitrealanibizumab, bevacizumab (Avastin; Genentech), triamcinolone,r dexamethasone injections before baseline and between theaseline and final visits.

Areas of confluent absent autofluorescence were defined asbsent autofluorescence measuring at least 0.5 mm in greatestinear diameter. Only areas of abnormal autofluorescence within aircle centered on the macula defined by the superior and inferioremporal arcades were analyzed; areas of peripapillary atrophyere not included in the measurements. The autofluorescence

mages were reviewed and measured using Topcon IMAGEnetoftware. Before analyzing the autofluorescence images, each im-ge was standardized by stretching the pixel histogram using thentire range of available pixel values. For images with multiplereas of confluent absent autofluorescence, all areas of more than.5 mm in greatest diameter were outlined and the measurementsere summed and recorded. If the area created while tracing the

rea of absent autofluorescence had an area of normal autofluo-escence enclosed within it, this was outlined separately and thisrea was subtracted to obtain the total area of confluent absentutofluorescence loss. Two independent observers measured thereas of confluent absent autofluorescence. On completion of dataollection, if there was a difference of more than 15% between the

observers for a particular measurement, the difference wasrbitrated through open adjudication. The mean of the measure-ents of both observers was used for statistical analysis. As part of

he evaluation, the Early Treatment Diabetic Retinopathy StudyETDRS) macular grid in Topcon IMAGEnet (version 2.55, foundnder the utilities tab under stereo analyzer) was placed over theundus image and centered on the geometric center of the fovea.reas of confluent absence of autofluorescence were graded ac-

ording to which of the 3 rings were occupied. The diameters of

335

tacfioTsa

orntciancahEswaaabhots

prirodeoaaoashficaOhOrEfflMbps

Ophthalmology Volume 120, Number 2, February 2013

the circles are 1000 �m, 3000 �m, and 6000 �m by convention.This evaluation was performed by 2 masked readers and anydifference was arbitrated through open adjudication.

Statistical Analysis

The data obtained were analyzed with frequency and descriptivestatistics. The visual acuity measurements were converted to log-arithm of the minimum angle of resolution (logMAR) units beforeanalysis. Generalized estimating equation modeling was used toevaluate predictors of RPE cell loss, enlargement of RPE cell loss,and visual acuity. Variables were entered into the model, and thenvariables were removed in a backward step-wise fashion until allremaining variables met the criteria of P�0.05. For all tests,P�0.05 was considered significant. The statistical analyses wereperformed with IBM-SPSS software version 20.0 (IBM-SPSS, Inc,Chicago, IL).

Results

There were 162 eyes of 116 patients with a mean age�standarddeviation of 82.9�7.9 years at baseline, 72 (62.1%) of whom werefemale. The mean duration of disease to the first autofluorescenceimaging was 2.1�2.4 years. Of the 162 eyes, 38 were evaluatedonce and follow-up was available for 124 (76.5%) of eyes that hada mean follow-up time of 3.0�1.8 years. The mean visual acuityat baseline was 20/71 (logMAR, 0.55). At baseline, 95 (58.6%)eyes had an area of confluent RPE cell loss of 0.5 mm or more ingreatest diameter, and among those, the median area of absentautofluorescence was 1.57 mm2 (interquartile range, 0.62–4.32mm2). At last follow-up, the mean visual acuity was 20/90 (log-MAR, 0.65). By final follow-up, 98 (79%) of 124 eyes hadconfluent loss of autofluorescence, and among those, the medianarea of absent autofluorescence was 3.61 mm2 (interquartile range,1.16–7.11 mm2). The mean change in area of absent autofluores-cence was an enlargement of 0.94 mm2 per year (Table 1; addi-

Table 1. Key Patient Information

Parameter Quantity

No. of eyes 162No. of patients 116Mean age (yrs) 82.9�7.9Female gender 72 (62.1%)Duration of disease to baseline

autofluorescence imaging (yrs)2.1�2.5

Mean visual acuity at baseline 20/71Median baseline area of absent

autofluorescence (mm2)1.57 (IQR, 0.62–4.32)

Mean baseline area of absentautofluorescence (mm2)

3.18�3.93

Duration of follow-up after baselineimaging (yrs)

2.9�1.8

Median area of absent autofluorescenceat last follow-up (mm2)

3.61 (IQR, 1.16–7.11)

Mean area of absent autofluorescenceat last follow-up (mm2)

5.28

Change in mean area of absentautofluorescence per year (mm2)

0.94

Mean visual acuity at final follow-up 20/90

IQR � interquartile range.

tional information is in Table 2, available at http://aaojournal.org). r

336

Figure 1 (available at http://aaojournal.org) shows representa-ive autofluorescence images of a patient with no confluent loss ofutofluorescence at baseline in which areas of absent autofluores-ence measuring more than 0.5 mm in diameter developed duringollow-up. Figures 2 and 3 are representative autofluorescencemages of 2 patients demonstrating enlargement of confluent areasf absent autofluorescence in patients with CNV during follow-up.he areas of absent autofluorescence were both preceded andurrounded by granular areas of abnormal autofluorescence (Figs 2nd 3).

The parameters known at baseline, including age at the timef symptom onset, duration of symptoms before first autofluo-escence imaging, gender, smoking status, previous photody-amic therapy, intravitreal triamcinolone, intravitreal pegap-anib, intravitreal bevacizumab, intravitreal ranibizumab, priorataract surgery, and glaucoma, were evaluated with general-zed estimating equation modeling. Only duration of diseasend previous treatment with photodynamic therapy were sig-ificant predictors of the dependent variable baseline area ofonfluent absent autofluorescence (Table 3, available at http://aojournal.org). These same variables also were predictive ofaving loss of autofluorescence within the central circle of theTDRS grid (Table 4, available at http://aaojournal.org). Theame variables plus baseline area of absent autofluorescenceere evaluated in creating a model with baseline visual acuity

s a dependent variable. Only the baseline area of absentutofluorescence was a significant predictor of baseline visualcuity (Table 5), with larger areas of absent autofluorescenceeing correlated with poorer visual acuity. Of the 95 eyes thatad confluent absent autofluorescence at baseline, 75 (78.9%)f these involved the central 1 mm of the ETDRS grid, denotinghat these patients had foveal involvement (P�0.001, chi-quare test).

The parameters at final follow-up evaluated as independentredictors of the dependent variable final area of absent autofluo-escence included duration of time since baseline autofluorescencemaging; the baseline area of absent autofluorescence; the squareoot of the area of baseline absence of autofluorescence;18 and usef intravitreal pegaptanib, bevacizumab, or ranibizumab indepen-ently or in aggregate. The significant predictors in generalizedstimating equation modeling for the dependent variable final areaf absent autofluorescence were square root of the baseline area ofbsent autofluorescence and follow-up duration. Larger baselinereas and longer follow-up were correlated with greater areasf absence at final follow-up. Use of the square root of the baselinerea instead of the untransformed value produced a model withomewhat better goodness-of-fit measures (Table 6, available atttp://aaojournal.org). Evaluation of the variables showed that thenal visual acuity was related to the area of absent autofluores-ence and the baseline visual acuity (Table 7). Eyes with largerreas of absent autofluorescence had lower levels of visual acuity.f the 98 eyes with confluent loss of autofluorescence, 89 (90.8%)ad loss in the central 1-mm ring of the ETDRS grid (P�0.001).f the 20 eyes observed to demonstrate confluent absent autofluo-

escence over the follow-up period, 16 (80%) involved the centerTDRS circle (P�0.001). The change in visual acuity over the

ollow-up was correlated with the change in area of absent auto-uorescence and the baseline logMAR visual acuity (Table 8).odeling the growth rate per year revealed the square root of the

aseline area of absent autofluorescence was a highly significantredictor, and once entered into the model, no other factors wereignificant. Thus, the larger the baseline area, the faster the growth

ltasppoeast

Kumar et al � RPE Loss in Neovascular AMD

Discussion

The RPE phagocytizes outer segments of the photoreceptorsand recycles the contained polyunsaturated fatty acids andretinoids.19,20 Damaged and cross-linked molecules are pro-cessed, and the indigestible material is sequestered into lipo-somes as lipofuscin. Because lipofuscin in RPE cells containsretinoids, it can be made to fluoresce with the wavelengthsused in the present study.19 Because RPE cells contain lipo-fuscin as a fundamental part of being alive and functional,autofluorescence imaging is a means of detecting the presenceof RPE cells.21 The absence of this same autofluorescence

Figure 2. Autofluorescence images from a 79-year-old woman showing prochoroidal neovascularization. A, Choroidal neovascularization developedwas treated with intravitreal injections of ranibizumab. Her visual acuity win a larger area of granular loss of autofluorescence. B, Nearly 1 year later,20/80. C, After an additional year, there was marked expansion of the areyear, her acuity remained 20/60, but the area of loss was larger. She was evnormal type. E, One year later, although the area of loss appeared only sligvisual acuity was 20/400.

signifies that functional cells are not present.6,22,23 The RPE i

oss demonstrated in this study is different in character fromhat typically seen with GA in 3 ways. First, the loss ofutofluorescence involved the center of the macula from thetart, in distinction from GA, which usually develops in aerifoveal location. Second, the loss of autofluorescence ex-anded outward from the center. Third, because of the locationf loss, the visual acuity was correlated with size of RPE lossven in early involvement. Because of the differences in char-cteristics between these 2 methods associated with RPE de-truction, the process in the present series was called a rela-ively generic term, RPE loss, instead of GA.

In the present series, the loss of autofluorescence (and by

on of loss of the retinal pigment epithelial autofluorescence in an eye withpreviously (previous to the date the picture was taken), and the patient

30. Note the smaller areas of confluent loss of autofluorescence embeddedmaller areas of loss were becoming more confluent. Her visual acuity wasbsent autofluorescence, but her visual acuity was 20/60. D, The followinged by a low-vision service because it was nearly impossible for her to readarger, her visual acuity declined to 20/200. F, After 1 additional year, the

gressi1 yearas 20/the s

a of aaluathtly l

nference, loss of the RPE) was highly predictive of visual

337

ctcaua

oidfts

CA

T

Ophthalmology Volume 120, Number 2, February 2013

acuity at baseline and at final follow-up. The change in areaof confluent absent autofluorescence was correlated with thechange in visual acuity observed over the follow-up period.This is not to say the visual acuity loss was solely the resultof RPE loss, because the same factors that cause loss of RPEalso may cause loss of the overlying photoreceptors. Thecorrespondence between decreased central autofluorescenceand decreased visual acuity has been demonstrated in a widevariety of diseases in addition to AMD, including centralserous chorioretinopathy, vitelliform lesions, trauma, retinaldystrophies, myopic degeneration, and drug toxicities.24–30

In some of these conditions, the retinal abnormalities seemto precede the autofluorescence abnormalities, but con-versely, loss of RPE cells in the central macula is not

Figure 3. Autofluorescence images from an 86-year-old man with choroidathe preceding 5 months since his diagnosis. His visual acuity was 20/40. Bof the areas of confluent absence of autofluorescence. C, The loss was seenacuity was 20/60. D, When last examined 43 months after having been diperiod with ranibizumab, his visual acuity was 20/80.

Table 5. Predictors of Baseline Visual Acuity

Parameter �Standard

ErrorWald

Chi-Square Significance

Constant 0.517 0.046 11.123 0.000Baseline area of absent

autofluorescence0.446 0.1437 9.626 0.002

The dependent variable was the logarithm of the minimum angle of

338

onsistent with preservation of central vision. It is likely thathe regression parameter estimates for absent autofluores-ence reflect the direct contribution RPE loss has on visualcuity, as well as the proxy effects for yet unknown ornquantified tissue damage in exudative AMD that may bessociated with loss of the RPE.

The etiologic and genetic risk factors for GA seem toverlap those for CNV.31,32 The progression of early andntermediate forms of AMD with the development of largerrusen and pigmentary abnormalities serves as risk factorsor both GA and for CNV.33 It is not reasonable to think theendency to develop RPE loss stops if CNV develops. Whentressed by oxidative damage and other stimuli, RPE cells

vascularization. A, The patient was receiving ranibizumab injections overmonths later, his visual acuity was 20/50 and he had a slight expansion

pand in this autofluorescence image obtained 10 months later. The visualed with choroidal neovascularization and being treated during the entire

Table 7. Predictors of Final Follow-up Visual Acuity

Parameter �Standard

ErrorWald

Chi-Square Significance

onstant 0.563 0.054 109.9 �0.001rea of absentautofluorescence areaat final follow-up

0.022 0.0096 5.0 0.025

he dependent variable was the logarithm of the minimum angle of

l neo, Tento exagnos

esolution visual acuity at final follow-up.

ttmioaaiyAdowoAifflmra

twifsmtmsedbfimflotsnoacancbA

R

Kumar et al � RPE Loss in Neovascular AMD

secrete increased amounts of VEGF or may undergo celldeath.9 It is possible that the stress placed on RPE cells mayincrease because of the many perturbations induced by theneovascularization. It is also possible that treatment mayinfluence the potential for RPE loss. An early form oftreatment for subfoveal CNV, photodynamic therapy usingverteporfin, is known to damage RPE cells in animal mod-els.34,35 In this study, some eyes had received photodynamictherapy before anti-VEGF agents were available. Photody-namic therapy was found to be associated with RPE loss inthe present study.

Because VEGF antagonists are the best form of availabletreatment and consequently all treatable patients are treatedwith anti-VEGF drugs, the potential that these agents con-tribute to the development of RPE loss could not be eval-uated. In rabbit eyes, intravitreal bevacizumab did not causeany observable damage.36,37 In cell culture, RPE cells ex-posed to hydrogen peroxide were more likely to die if eitherVEGF or the VEGF receptor was blocked by antibodies.38

This work suggests VEGF may be a survival factor for RPEcells under oxidative stress. In the Comparison of Age-Related Macular Degeneration Treatments Trials at the2-year follow-up, geographic atrophy, defined in that studyas round areas of depigmentation in which underlying cho-roidal vessels were more visible, was more likely to bepresent in eyes treated with ranibizumab as compared witheyes treated with bevacizumab.39 In this study, RPE losswas detected by a test that relied on physiologic character-istics of RPE cells themselves and not an indirect measuresuch as improved ability to see underlying choroidal bloodvessels in eyes with intervening CNV. In this study, therewas no significant interaction between the drug used and thedevelopment of RPE loss.

Grading systems for AMD exist at present in attempts toestimate risk of progression to late disease.40–43 This isimportant in quantifying disease progression and in preven-tion strategies. In these classification strategies, earlierforms of disease were subdivided into many categories, andlate disease was lumped together as the dreaded terminationof the process. These classification systems were developedbefore there was a widespread, effective means of treatingexudative disease. Although there is no effective treatmentfor the atrophic forms of AMD, there are apparent differ-ences in phenotypes of treated exudative disease. As such,current classification systems will need to be enlarged to

Table 8. Predictors of Change in Visual Acuity

Parameter �Standard

ErrorWald

Chi-Square Significance

Constant 0.197 0.049 16.1 �0.001Change in area of absent

autofluorescence0.035 0.015 5.6 0.018

Baseline logMAR visualacuity

�0.346 0.068 26 �0.001

The dependent variable was the change in logarithm of the minimumangle of resolution (logMAR) visual acuity.

include various aspects of treated neovascular disease. In

hat regard, the amount of RPE loss seems to be an impor-ant, measurable characteristic. Future studies of the treat-ent of CNV in AMD should evaluate autofluorescence

maging data in addition to other imaging data currentlybtained in clinical practice. In this study, the baselineutofluorescence characteristics were predictive of eventualnatomic and functional measures. Comparative studies us-ng optical coherence tomography have not been publishedet. For example, there are no published studies from theNCHOR, MARINA, PIER (Phase IIIb, Multicenter Ran-omized Double-Masked Sham-injection Controlled Studyf the Efficiency and Safety of Ranibizumab in Subjectsith Subfoveal Choroidal Neovascularization with or with-ut Classic CNV Secondary to AMD), or Comparison ofge-Related Macular Degeneration Treatments Trials stud-

es examining the interrelationships among visible anatomiceatures of the retina, including intraretinal or subretinaluid and visual acuity.14–16,44,45 Integrating anatomic infor-ation from optical coherence tomography with autofluo-

escence data may help to delineate factors related to visualcuity loss among patients with exudative AMD.

To the best of our awareness, this is the first study to quantifyhe prevalence and progression of confluent RPE loss in patentsith CNV and to demonstrate that the area of confluent RPE loss

s a significant predictor of visual acuity both at baseline and finalollow-up. This study has weaknesses conferred by its retro-pective nature and the developmental aspect of the treat-ent for CNV. Some patients were followed up for longer

han others, and consequently they were treated with olderethods such as photodynamic therapy. A prospective

tudy could concentrate on current treatment methods, butven these likely will change over time. The present studyid not examine patterns of autofluorescence loss that maye predictive of future loss and did not evaluate anatomicndings obtainable from analysis of optical coherence to-ographic images and their interrelationships with auto-uorescence imaging. The visual acuities reported were notbtained at prespecified standardized intervals. In addition,he impact on visual function as measured by instrumentsuch as the 25-item Visual Function Questionnaire46 wasot performed. These additional studies are either in processr are in the planning stages. Our study demonstrated thatutofluorescence is a valuable tool in demonstrating con-urrent RPE loss in patients with CNV. Numerous studiesre underway to develop mechanisms to prevent RPE loss inonexudative AMD. The present study suggests that appli-ation of strategies to limit expansion of RPE cell loss maye applicable in eyes with CNV resulting from neovascularMD as well.

eferences

1. Eye Diseases Prevalence Research Group. Prevalence of age-related macular degeneration in the United States. Arch Oph-thalmol 2004;122:564–72.

2. Sarks JP, Sarks SH, Killingsworth MC. Evolution of geo-graphic atrophy of the retinal pigment epithelium. Eye (Lond)

339

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

Ophthalmology Volume 120, Number 2, February 2013

3. Sunness JS, Bressler NM, Tian Y, et al. Measuring geographicatrophy in advanced age-related macular degeneration. InvestOphthalmol Vis Sci 1999;40:1761–9.

4. Göbel AP, Fleckenstein M, Schmitz-Valckenberg S, et al.Imaging geographic atrophy in age-related macular degener-ation. Ophthalmologica 2011;226:182–90.

5. Spaide RF. Fundus autofluorescence and age-related maculardegeneration. Ophthalmology 2003;110:392–9.

6. Sunness JS, Ziegler MD, Applegate CA. Issues in quantifyingatrophic macular disease using retinal autofluorescence. Ret-ina 2006;26:666–72.

7. Schmitz-Valckenberg S, Holz FG, Bird AC, Spaide RF. Fun-dus autofluorescence imaging: review and perspectives. Ret-ina 2008;28:385–409.

8. Bearelly S, Cousins SW. Fundus autofluorescence imaging inage-related macular degeneration and geographic atrophy.Adv Exp Med Biol 2010;664:395–402.

9. Spaide RF, Armstrong D, Browne R. Continuing medical edu-cation review: choroidal neovascularization in age-related macu-lar degeneration—what is the cause? Retina 2003;23:595–614.

10. Zarbin MA. Current concepts in the pathogenesis of age-related macular degeneration. Arch Ophthalmol 2004;122:598–614.

11. Scholl HP, Fleckenstein M, Charbel Issa P, et al. An update onthe genetics of age-related macular degeneration. Mol Vis[serial online] 2007;13:196–205. Available at: http://www.molvis.org/molvis/v13/a23/. Accessed June 24, 2012.

12. Kaneko H, Dridi S, Tarallo V, et al. DICER1 deficit inducesAlu RNA toxicity in age-related macular degeneration. Nature2011;471:325–30.

13. Sarks J, Tang K, Killingsworth M, et al. Development ofatrophy of the retinal pigment epithelium around disciformscars. Br J Ophthalmol 2006;90:442–6.

14. Brown DM, Kaiser PK, Michels M, et al, ANCHOR StudyGroup. Ranibizumab versus verteporfin for neovascular age-related macular degeneration. N Engl J Med 2006;355:1432–44.

15. Rosenfeld PJ, Brown DM, Heier JS, et al, MARINA StudyGroup. Ranibizumab for neovascular age-related macular degen-eration. N Engl J Med 2006;355:1419–31.

16. Brown DM, Michels M, Kaiser PK, et al, ANCHOR StudyGroup. Ranibizumab versus verteporfin photodynamic therapyfor neovascular age-related macular degeneration: two-year re-sults of the ANCHOR Study. Ophthalmology 2009;116:57–65.

17. Rosenfeld PJ, Shapiro H, Tuomi L, et al, MARINA andANCHOR Study Groups. Characteristics of patients losingvision after 2 years of monthly dosing in the phase III ranibi-zumab clinical trials. Ophthalmology 2011;118:523–30.

18. Yehoshua Z, Rosenfeld PJ, Gregori G, et al. Progression ofgeographic atrophy in age-related macular degeneration im-aged with spectral domain optical coherence tomography.Ophthalmology 2011;118:679–86.

19. Schmitz-Valckenberg S, Holz FG, Bird AC, Spaide RF. Fun-dus autofluorescence imaging: review and perspectives. Ret-ina 2008;28:385–409.

20. Sparrow JR, Hicks D, Hamel CP. The retinal pigment epithe-lium in health and disease. Curr Mol Med 2010;10:802–23.

21. Delori FC, Dorey CK, Staurenghi G, et al. In vivo fluores-cence of the ocular fundus exhibits retinal pigment epitheliumlipofuscin characteristics. Invest Ophthalmol Vis Sci 1995;36:718–29.

22. Schmitz-Valckenberg S, Fleckenstein M, Scholl HP, Holz FG.Fundus autofluorescence and progression of age-related mac-ular degeneration. Surv Ophthalmol 2009;54:96–117.

23. Choudhry N, Giani A, Miller JW. Fundus autofluorescence ingeographic atrophy: a review. Semin Ophthalmol 2010;25:

340

4. von Rückmann A, Fitzke FW, Bird AC. Distribution of pig-ment epithelium autofluorescence in retinal disease state re-corded in vivo and its change over time. Graefes Arch ClinExp Ophthalmol 1999;237:1–9.

5. Lorenz B, Wabbels B, Wegscheider E, et al. Lack of fundusautofluorescence to 488 nanometers from childhood on inpatients with early-onset severe retinal dystrophy associatedwith mutations in RPE65. Ophthalmology 2004;111:1585–94.

6. Kellner U, Renner AB, Tillack H. Fundus autofluorescenceand mfERG for early detection of retinal alterations in patientsusing chloroquine/hydroxychloroquine. Invest OphthalmolVis Sci 2006;47:3531–8.

7. Mustafa MS, McBain VA, Scott CM. Autofluorescence im-aging—a useful adjunct in imaging macular trauma. ClinOphthalmol 2010;4:1497–8.

8. Imamura Y, Fujiwara T, Spaide RF. Fundus autofluorescenceand visual acuity in central serous chorioretinopathy. Ophthal-mology 2011;118:700–5.

9. Robson AG, Tufail A, Fitzke F, et al. Serial imaging andstructure-function correlates of high-density rings of fundusautofluorescence in retinitis pigmentosa. Retina 2011;31:1670–9.

0. Roe RH, Jumper JM, Gualino V, et al. Retinal pigment epi-theliopathy, macular telangiectasis, and intraretinal crystaldeposits in HIV-positive patients receiving ritonavir. Retina2011;31:559–65.

1. Sepp T, Khan JC, Thurlby DA, et al. Complement factor Hvariant Y402H is a major risk determinant for geographicatrophy and choroidal neovascularization in smokers and non-smokers. Invest Ophthalmol Vis Sci 2006;47:536–40.

2. Cameron DJ, Yang Z, Gibbs D, et al. HTRA1 variant conferssimilar risks to geographic atrophy and neovascular age-related macular degeneration. Cell Cycle 2007;6:1122–5.

3. Smith W, Assink J, Klein R, et al. Risk factors for age-relatedmacular degeneration: pooled findings from three continents.Ophthalmology 2001;108:697–704.

4. Husain D, Kramer M, Kenny AG, et al. Effects of photody-namic therapy using verteporfin on experimental choroidalneovascularization and normal retina and choroid up to 7weeks after treatment. Invest Ophthalmol Vis Sci 1999;40:2322–31.

5. Miller JW, Walsh AW, Kramer M, et al. Photodynamic ther-apy of experimental choroidal neovascularization using lipo-protein-delivered benzoporphyrin. Arch Ophthalmol 1995;113:810–8.

6. Bakri SJ, Cameron JD, McCannel CA, et al. Absence ofhistologic retinal toxicity of intravitreal bevacizumab in arabbit model. Am J Ophthalmol 2006;142:162–4.

7. Feiner L, Barr EE, Shui YB, et al. Safety of intravitreal injectionof bevacizumab in rabbit eyes. Retina 2006;26:882–8.

8. Byeon SH, Lee SC, Choi SH, et al. Vascular endothelialgrowth factor as an autocrine survival factor for retinal pig-ment epithelial cells under oxidative stress via the VEGF-R2/PI3K/Akt. Invest Ophthalmol Vis Sci 2010;51:1190–7.

9. Comparison of Age-related Macular Degeneration TreatmentsTrials (CATT) Research Group Writing Committee, MartinDF, Maguire MG, Fine SL, et al. Ranibizumab and bevaci-zumab for treatment of neovascular age-related maculardegeneration: two-year results. Ophthalmology 2012;119:1389–98.

0. Bird AC, Bressler NM, Bressler SB, et al, International ARMEpidemiological Study Group. An international classification andgrading system for age-related maculopathy and age-related mac-

4

4

4

Kumar et al � RPE Loss in Neovascular AMD

41. Klaver CC, Assink JJ, van Leeuwen R, et al. Incidence andprogression rates of age-related maculopathy: the Rotter-dam Study. Invest Ophthalmol Vis Sci 2001;42:2237–41.

42. Age-Related Eye Disease Study Research Group. The Age-Related Eye Disease Study severity scale for age-related mac-ular degeneration: AREDS report no. 17. Arch Ophthalmol2005;123:1484–98.

43. Seddon JM, Sharma S, Adelman RA. Evaluation of the clin-ical age-related maculopathy staging system. Ophthalmology

Footnotes and Financial Disclosures

The author(s) have made the following disclosure(s):

R

T

S

CRYr

4. Kaiser PK, Brown DM, Zhang K, et al. Ranibizumab for pre-dominantly classic neovascular age-related macular degeneration:subgroup analysis of first-year ANCHOR results. Am J Ophthal-mol 2007;144:850–7.

5. CATT Research Group, Martin DF, Maguire MG, Ying GS, etal. Ranibizumab and bevacizumab for neovascular age-relatedmacular degeneration. N Engl J Med 2011;364:1897–908.

6. Mangione CM, Lee PP, Gutierrez PR, et al, National Eye Insti-tute Visual Function Questionnaire Field Test Investigators. De-velopment of the 25-item National Eye Institute Visual Function

Originally received: March 23, 2012.Final revision: July 24, 2012.Accepted: July 24, 2012.Available online: November 6, 2012. Manuscript no. 2012-432.

From the Vitreous Retina Macula Consultants of New York, New York,New York.

Financial Disclosure(s):

ichard F. Spaide: Royalty payments—Topcon Medical Systems, Inc.

he remaining authors have no potential conflicting relationships to report.

upported by the Macula Foundation, New York, New York.

orrespondence:ichard Spaide, MD, Vitreous Retina Macula Consultants of Nework, 460 Park Avenue, Fifth Floor, New York, NY 10022. E-mail:

[email protected].

341