Indocyanine Green Angiographyand Choroidal N eovascularizationObscured by Hemorrhage

Elias Reichel, MD, Jay S. Duker, MD, Carmen A. Puliafito, MD

Background and Objective: To determine the use of indocyanine green (ICG) an­giography in detecting choroidal neovascularization obscured by hemorrhage.

Study/Design: Indocyanine green angiography was performed on 20 consecutivepat ients who were suspected to have choroidal neovascular membranes (CNVs) thatwere obscured by subretinal or intraretinal hemorrhage on fluorescein angiography. Theetiology of choroidal neovascularization was age-related macular degeneration.

Results: In all patients, ICG angiography was able to detect some choroidal neo­vascularization. The entire extent of the CNV was identified in 12 (60%) of these patientsafter ICG angiography was performed. In the remaining eight eyes (40%) , the CNV stillwas obscured partially by a thick layer of hemorrhage. Difficulty in detection was dueto thick subretinal hemorrhage, lipid exudate, and pigment.

Conclusion: The authors conclude that ICG angiography is superior to fluoresceinangiography in showing the presence and extent of CNVs associated with age-relatedmacular degeneration that are almost entirely obscured by hemorrhage.Ophthalmology 1995;102: 1871-1876

Choroidal ncovascular membranes (CNVs) can be clas­sified into three broad categories based on well-establishedfluorescein angiographic criteria: well-defined or classic,poorly defined or occult, and those obscured by hemor­rhage or exudate. 1-3 In eyes with certain clinical findings,CNVs are presumed to be present even when fluoresceinangiography fails to identify or completely localize a CNVbecause of overlying hemorrhage, pigment, or turbid fluid.In this context, the CNV is considered to be completely,or near completely, obscured, and a CNV can be identifiedin near entirety on fluorescein angiograph y. Therefore,treatment may be performed. The treatment approachfor most CNVs obscured by hemorrhage, in general, isobservation. v " Waiting until the hemorrhage clears even­tually may result in fluorescein angiographic localizationof the CNV. Typically, clearing of the hemorrhage takesseveral months, and during this period of time the CNV

Originally received: May 2, 1995.Revision accepted: June 22, 1995.

From the New England EyeCenter. New England Medical Center, TuftsUniversity, School of Medicine, Boston.

Reprint requeststo Elias Reichel, MD, NewEngland EyeCenter, NewEnglandMedical Center, 750 Washington SI, Box450, Boston, MA 02111.

can increase in size, grow from an extrafoveal to a foveallocation, or cause persistent hemorrhage and exudationunderneath and within the neurosensory retina, resultingin permanent loss of central vision." Improved detectionofCNV obscured by hemorrhage may allow earlier treat­ment and ultimately result in improved preservation ofcentral vision,

Indocyanine green (ICG) dye absorbs (805 nm) andemits (835 nm) light in the near infrared. These absorp­tion and emission characteristics of ICG theoreticallyshould permit better visualization of both normal andabnormal choroidal vessels compared with fluoresceinangiography." Areas of hyperfluorescence observed onICG angiography during the late phase ofthe angiogramare believed to be areas of choroidal neovascularization.?"!'In a situation where there is hemorrhage or serous exudateoverlying a CNV, infrared light should be absorbed readilyby an ICG retaining CNVs, and the emitted fluorescinglight can be detected with a digital video imagining systemthat detects in the infrared. This theoretical advantageover fluorescein angiography should enhance the diagnosisofCNVs that are obscured by hemorrhage. Massive hem­orrhage that is thick may still block tran smission of ICGfluorescence.

1871

Ophthalmology Volume 102, Number 12, December 1995

We performed ICG angiography on 20 patients whohad a large subretinal or intraretinal hemorrhage to com­pare the ability ofICG angiography to detect and localizeCNVs with fluorescein angiography.

Table 1. Classification Scheme of Patterns ofHyperfluorescence on Fluorescein and

Indocyanime Green Angiography of OccultChoroidal Neovascularization Obscurred

by Hemorrhage

by injecting sodium fluorescein 10%(5 ml) into a periph­eral arm vein.

Indocyanine Green Digital Angiographic System

A Topcon fundus camera (TRC-50IA; Topcon Instru­ments, Paramus, NJ) was used that contained a 100-Whalogen bulb and xenon flash. Barrier and excitation filterswere designed for ICG peak absorption (805 nm) andfluorescence (835 nm). Near-infrared antireflectivecoatingwas applied to the lenses. A digitized infrared fundus videocamera system designed by Topcon IMAGEnet wasadapted to the camera part of our Topcon system via an

Indocyanine Green

Indocyaninegreen (Cardiogreen) was prepared in solutionwith the manufacturer-supplied aqueous solvent to a con­centration of 10 mg/ml. Fifty milligrams of ICG wereused for diagnostic studies in most patients. In patientswith very dense subretinal hemorrhage or intraretinalhemorrhage, a small pupil, or a visuallysignificantcataractor other media opacity, a higher dose (range, 50-100 mg)of ICG was used. The dye was injected into a peripheralarm vein. The infusion was followed immediately by a10-ml flush of sterile normal saline.

Patients and Methods

Patients were recruited consecutively from the RetinaService at the New England Eye Center. All had the di­agnosis of age-related macular degeneration and wereolder than 55 years of age. Patients included in this studymet the criteria established by the Macular Photocoagu­lation Study for neovascular or exudative age-relatedmacular degeneration. Eligible patients were those withcentral visual symptoms who had results of clinical ex­aminations suspicious for CNVs due to the presence of aretinal pigment epithelial detachment, exudate, subretinalfluid, macular edema, and/or subretinal and intraretinalhemorrhage. Only patients who were suspected to have aCNV obscured by hemorrhage were included in this study.In general, patients with small amounts of intraretinal orsubretinal hemorrhage (no significant thickening on slit­lamp biomicroscopy) were not included in this study be­cause evaluation with fluorescein angiography typicallyshowed the presence of a well-defined or occult CNV.Exclusion criteria included a known allergy to iodine­based dye and previous laser photocoagulation in the studyeye. Best-corrected Snellen visual acuity was obtained inall patients.

Every patient had a general ophthalmologic and retinalexamination, including indirect ophthalmoloscopy andslit-lamp biomicroscopy with either a 78- or 90-diopterlens. Color stereophotography and fluorescein angiogra­phy were first done and then ICG videoangiography wasperformed on all patients. Clinical examination and pho­tographic and angiographic studies all were completed onthe same day.

Fluorescein Angiography

Fluorescein angiography (Kodak Tri-X film and colorfundus photography (Kodak Ektachrome 64 film) or dig­ital fluorescein angiography were done using the sameangiography system that was used for ICG (see below)with appropriate excitation (480) and barrier) (535) filters.Fluorescein angiography was done in the usual manner

FluroesceinAngiography

No hyperfluorescence

Small areas ofhyperfluorescence

leG = indocyanine green.

leG Angiography

No hyperfluorescenceSingle well-demarcated area of

hyperfluorescencePartially obscured hyperfluorescence

No hyperfluorescenceSingle well-demarcated area of

hyperfluorescencePartially obscured hyperfluorescence

Figure 1. Top and second row, A, fundus photograph of an eye with a choroidal neovascularization that is entirely obscured by hemorrhage. B,fluorescein angiography showed no areas of hyperfluorescence suggestive of choroidal neovascularizat ion . C, a late view of the indocyanine greenangiogram showed a single, well-demarcated area of hyperfluorescence suggestive of a choroidal neovascularizat ion (arrows).

Third row and bottom, Figure 2. A, fundus photograph that shows a subretinal hemorrhage. B, a fluorescein angiogram with a small area of poorlydefined choroidal neovascularization superotemporal to the fovea. C, indocyanine green angiography illustrates a single placoid area of hyperfluorescenceobscured by a contiguous area of thickened blood. The area of hyperfluorescence appears to be larger than what is observed on fluorescein angiographyand the nasal border is better defined underneath the hemorrhage. The subretinal hemorrhage seen in the area that continues to obscure the choroidalneovascularization on indocyanine green angiography appears to be darker in color . Stereo color photographs showed a qualitative difference in theelevation in this area of the retina, corresponding to the area of blocked fluorescence, suggesting a greater degree of subretinal hemorrhage.

1872

1873

Ophthalmology Volume 102, Number 12, December 1995

Results

Table 2. Observed Patterns of Hyperfluorescenceon Fluorescein and Indocyanine Green

Angiography of Occult ChoroidalNeovascularization Obscured by Hemorrhage

interfacing adapter. Single-frame analog images (1 frameper second) were sent from the camera to a modified IBMAT computer (CompuAdd, Austin, TX) where they weredigitized and enhanced. The images then were displayedon a high-resolution monitor (1280 X 1024-pixel monitor)and stored on an optical laser disc in a computer for in­stant recall.

Mapping of Choroidal N eovascularizations

The presence of CNVs was determined after comparingslit-lamp biomicroscopic findings and digital fluoresceinangiographic results with ICG angiographic data. Tracingof the CNV onto a digital ICG angiogram from the ICGangiogram was performed via the digital imaging systemusing software that was provided with the Topcon IM­AGEnet system.

Discussion

be seen partially, having been obscured on fluoresceinangiography by a thin layer of hemorrhage. A contiguousor noncontiguous thicker layer of hemorrhage still ob­scured part(s) of the CNV with ICG angiography. Oneother possibility that was not observed in this group ofpatients was that absolutely no hyperfluorescence couldbe seen on ICG angiography. These categories are dis­played in Table 1.

In 4 (20%) of the 20 patients, the CNV was obscuredtotally by hemorrhage on fluorescein angiography. All fourpatients who had no fluorescein angiographic evidence ofa CNV demonstrated CNVs on ICG angiography thatwere well-demarcated unifocal areas ofhyperfluorescenceand were not obscured by hemorrhage. Of the 20 CNVs,16 (80%) were obscured almost entirely by hemorrhage(i.e., fluorescein angiography showed small areas of latehyperfluorescence that were suggestive of a CNV). In 8ofthese eyes, CNVs were detected using ICG angiographythat were characterized as displaying a single, well-de­marcated area of hyperfluorescence in the late views ofthe ICG angiogram (Fig 1). Eight patients (40%) had anarea of hyperfluorescence on ICG angiography partiallyobscured by a thicker layer of hemorrhage (Fig 2). Table2 summarizes the results found in the 20 patients includedin this study.

Clinical suggestion of exudative age-related maculardegeneration occurs when retinal elevation or retinalpigment epithelial detachment, intraretinal fluid, sub­retinal hemorrhage, intraretinal hemorrhage or lipidexudate is seen in patients older than 55 years of age.Lack of fluorescein angiographic evidence of a well-de­fined or "classic" CNV implies that the CNV is occultor poorly defined.l-" Fluorescein angiographic featuresof poorly defined CNVs include the presence of an ir­regular elevation of the retinal pigment epitheliumwhich stains, and often leaks, fluorescein. Choroidalneovascularizations that are obscured by hemorrhageare difficult to treat because it is not possible to applyconfluent laser treatment to the CNV as recommendedby the Macular Photocoagulation Study protocol, inwhich treatment is designed to cover the entire extentof the CNV. 12 This article focuses on a subgroup ofpatients with CNV obscured by relatively large amountsof intraretinal/subretinal hemorrhage who have beenstudied with ICG angiography.

In the current study, ICG angiography identified atleast some part of the CNV underlying the hemorrhage100% of the time. This result is not surprising in that80% ofthese patients displayed some evidence ofa CNVon fluorescein angiography. This finding does not implythat ICG is 100% sensitive in detecting a CNV. It hasbeen observed that well-defined or classic CNVs iden­tified on fluorescein angiography generally do not dis­play late hyperfluorescence on ICG. 13 Therefore, well­defined CNVs associated with large hemorrhages maynot be observed.

o

8(40%)

4(20%)

8(40%)

leG Angiography

o

o

Single, Well PartiallyNone Demarcated Obscured

Between September 1, 1991, and January 1, 1993, 200patients with the new diagnosis of CNVs secondary toage-related macular degeneration were seen on the RetinaService at the New England Eye Center. All patients un­derwent fluorescein angiography. Of these 200 patients,20 eyes (10%) had intraretinal and/or subretinal hemor­rhage that was presumed to be obscuring a CNV. All 20patients underwent ICG angiography to better delineatea CNV. No complications associated with ICG angiog­raphy were observed.

To systematize the characterization of CNV observedon fluorescein and ICG angiography, we divided the char­acteristics of the CNV into six categories. On fluoresceinangiography, CNVs either were obscured completely or,alternatively, small areas of hyperfluorescence also couldbe observed on fluorescein angiography. These small areasofhyperfluorescence were typically contiguous with largerareas of hemorrhage that obscured the CNV. On ICGangiography, two patterns ofhyperfluorescence were seen.In one group of patients, a single, well-demarcated areaof hyperfluorescence could be observed that suggested thefull extent of the CNV. In some instances, the CNV could

Fluorescein

No hyperfluorescenceSmall areas of

hyperfluorescence

leG = indocyanine green.

1874

Reichel et al . ICG Choroidal Neovascularization

Figure 3. A, preoperative fundus photograph shows a subretinal hemorrhage in the left eye of a patient with age-related macular degeneration. Visualacuity was 20/50. B, fluorescein angiography shows obscured hyperfluorescence. C, indocyanine green angiography shows a small area of hyperflu­orescence (arrows) that was treated with diode laser photocoagulation. D, 3-month follow -up fluorescein angiography shows no evidence of thechoroidal neovascularization. Visual acuity = 20/40.

In 60% of membranes, a single, well-demarcated areaof hyperfluorescence appeared on ICG angiography. In40% of membranes, single , well-demarcated CNVscould be detected but were obscured partially by a thicklayer of hemorrhage.

Indocyanine green angiography appears to be su­perior to fluorescein angiography in showing the pres­ence and extent of CNVs associated with age-relatedmacular degeneration that are obscured by dense sub­retinal/intraretinal hemorrhage. Areas of hyperflu­orescence that appear to be well demarcated and areobserved in entirety may be amenable to laser pho­tocoagulation by tracing the area of hyperfluorescenceto a red-free photograph. Early photocoagulation of aCNV obscured by hemorrhage using a diode laser may ..result in stabilization or improvement of central visualacuity (Fig 3).14 Diode laser photocoagulation was usedbecause experimental studies indicate maximal pen­etration through blood using infrared wavelengthswithout severely damaging the retina.'? Prospectivestudies with long-term follow-up may show a treatment

benefit when using lCG angiography as a guide fortreating eyes with CNV obscured by hemorrhage.

References

1. Bressler, NM, Bressler SB, Gragoudas ES. Clinical charac­teristics of choroidal neovascular membranes ArchOphthalmol 1987;105:209-13.

2. Folk Jc. Aging macular degeneration: Clinical features oftreatable disease. Ophthalmology 1985;92:594-602.

3. Soubrane G, Coscas G, Franscais C, Koenig F. Occult sub­retinal new vessels in age-related macular degneration. Nat­ural history and early laser treatment. Ophthalmology1990;97:649-57.

4. Gass JDM. Pathogenesis of disciform detachment of theneuroepithelium. III. Senile disciform macular degeneration.Am J Ophthalmol 1967;63:617-44.

5. Gass JDM. Pathogenesis of disciform detachment of theneuroepithelium. I. General concepts and classification.AmJ Ophthalmol 1967;63:573-85.

6. Gass JDM. Stereoscopic atlas ofmacular diseases: diagnosisand treatment, 3rd ed. St. Louis: Mosby, 1987;80.

1875

Ophthalmology Volume 102, Number 12, December 1995

7. Bressler SB, Bressler NM, Fine SL, et al. Natural course ofchoroidal neovascular membranes within the foveal avas­cular zone in senile macular degeneration. Am J Opthalmol1982;93:157-63.

8. Cohen SM, Shen JH, Smiddy WE. Laser energy and dyefluorescence transmission through blood in vitro. Am JOphthalmol 1995;119:452-7.

9. Destro M, Puliafito CA. Indocyanine green videoangiog­raphy of choroidal neovascularization. Ophthalmology1989;96:846-53.

10. Guyer DR, Puliafito CA, Mones JM, et al. Digital indocy­anine green angiography in chorioretinal disorders. Oph­thalmology 1992;99:287-91.

II. Yannuzzi LA, Siakter JS, Sorenson JA, et al. Digital indo­cyanine green videoangiography and choroidal neovascu­larization. Retina 1992;12: J91-223.

1876

12. Macular Photocoagulation Study Group. Argon laserphotocoagulation for senile macular degeneration. Re­sults of a randomized clinical trial. Arch Ophthalmol1982; I00:912-8.

13. Avvad FK, Duker JS, Reichel E, et al. The digital indocy­anine green videoangiography characteristics of well-definedchoroidal neovascularization. Ophthalmology 1995;102:401-5.

14. Siakter JS , Yannuzzi LA, Sorenson JA, et al. A pilotstudy of indocyanine green laser photocoagulation ofoccult choroidal neo vascularization in age-relatedmacular degenerat ion . Arch Ophthalmol 1994; 112:465-72.

15. Johnson MW, Hassan TS, Elner VM. Laser photocoagu­lation of the choroid through experimental subretinal hem­orrhage. Arch Ophthalmol 1995;113:364-70.