Three-Dimensional Angiography of Classic and Occult Lesion Types in Choroidal Neovascularization

Abstract

purpose. To identify characteristic features of classic and occult choroidal neovascularization (CNV) by using a novel technique of topographic angiography for three-dimensional (3D) visualization.

methods. A confocal scanning laser ophthalmoscope (SLO, Heidelberg Retina Angiograph; Heidelberg Engineering, Dossenheim, Germany) was used to perform fluorescein (FA) and indocyanine green (ICGA) angiography in158 patients. Ninety-four eyes had predominantly classic and 64 eyes had occult lesions. With an image frequency of 20 Hz, a tomographic series of 32 images per set were taken over a depth of 4 mm. Axial analyses for each x/year position were performed, to determine the fluorescence distribution along the z-axis. After the axial location of hyperfluorescence was detected, a depth profile was generated. All results were integrated into a gray-scale-coded depth image and imaged as a 3D relief.

results. Characteristic features of classic and occult lesions were distinguished. Classic CNV appeared as a well-demarcated lesion with steep, prominent borders, often craterlike, and frequently surrounded by a halo, suggesting choroidal perfusion changes. Occult CNV was documented by 3D as a convex lesion with flat, ill-defined borders and without any surrounding halo. Topographic imaging is superior to conventional angiography regarding definition of lesion type, configuration, and extension, because masking phenomena do not interfere.

conclusions. Topographic angiography allows a realistic 3D representation of CNV. Characteristic features based on the neovascular architecture and the differences in leakage behavior of different lesion types are clearly identified.

Age-related macular degeneration (AMD) is the leading cause of visual impairment among elder individuals in developed countries. ^{1 2} AMD occurs in two main patterns: the dry, atrophic form and the exudative neovascular form. Exudative AMD develops when choroidal capillaries start growing through Bruch’s membrane—a phenomenon referred to as choroidal neovascularization (CNV). Because CNV is usually accompanied by exudation, subretinal hemorrhage, and/or RPE detachment, it is often the cause of severe, sudden, and usually irreversible vision loss. ³ Neovascular AMD can be classified according to its appearance during fluorescence angiography into classic or occult lesion types. The more aggressive classic type is often associated with early and substantial vision loss due to direct photoreceptor damage, whereas occult lesions are often present with long-term maintenance of vision until RPE decompensation occurs. ⁴  

Based on lesion size and type, treatment recommendations for CNV differ significantly. Photocoagulation was indicated for classic CNV only, photodynamic therapy (PDT) was recommended for classic CNV of any size, but only for small occult lesions and no official guideline allowed PDT of minimally classic CNV. Antiangiogenic therapy appears to allow an all-lesion-types approach. However, subgroup analysis has demonstrated a superior effect in predominantly classic CNV and smaller lesions in the VISION (VEGF Inhibition Study in Ocular Neovascularization) trial, and best results for minimally classic CNV. ⁵ Promising results concerning the prevention of vision loss were found in the ANCHOR (Anti-VEGF Antibody for the Treatment of Predominantly Classic Choroidal Neovascularization in AMD) trial (Brown DM et al. IOVS 2006;47:E-Abstract 2963) including predominantly classic CNV, where over 94% of patients (94.3% receiving 0.3 mg ranibizumab; 96.4% receiving 0.5 mg ranibizumab) lost fewer than 15 letters (ETDRS [Early Treatment Diabetic Retinopathy Study]chart) from baseline visual acuity, versus 64.3% of patients in the PDT group. The MARINA trial (Minimally classic/occult trial of the Anti-VEGF antibody Ranibizumab In the treatment of Neovascular AMD) focusing on minimally classic and occult lesion types showed that nearly 95% of patients receiving ranibizumab (0.3 or 0.5 mg) lost fewer than 15 letters compared with 62% of patients receiving a sham injection (Heier JS et al. IOVS 2006;47:E-Abstract 2959; Webster MK et al. IOVS 2006;47:E-Abstract 2206; Chang TS et al. IOVS 2006;47:E-Abstract 5252). The proportion of patients showing a significant improvement of more than 3 lines was highest in the ANCHOR trial, with predominantly classic lesions. Current pathway-based therapy results depend much less on lesion morphology, as defined using conventional fluorescein angiography (FA). Because anti-VEGF therapy is basically an antileakage strategy, the exudative morphology of a lesion becomes more relevant. Clearly, the nature of the neovascular process and, with antiangiogenic therapy, the leakage characteristics of a CNV lesion have to be understood. To evaluate the efficacy of any treatment modality, to develop adequate treatment strategies and retreatment regimens, an accurate diagnosis and identification of exudative patterns is fundamental. The main diagnostic tool is FA, where information is often limited by masking phenomena. Because of the short wavelength, fluorescence is nearly completely absorbed by the RPE, blood, or fluid, and underlying processes remain obscure. ^{6 7 8 9} Information about activity and extent of subretinal structures is more easily accessible by indocyanine green angiography (ICGA). The dye is more effective in the near infrared spectrum, enables better transmission through pigment and exudation, ¹⁰ and should therefore improve imaging of occult lesions, leakage, and its origin. ^{11 12} With the introduction of the confocal scanning laser ophthalmoscope (SLO), ¹³ the diagnostic efficacy has been improved by combining optimal contrast, high sensitivity, and resolution up to 300 μm. ^{14 15} The confocal modality allows sequential tomographic imaging with 32 angiographic sections over a 4- to 6-mm thickness. ⁷ Localization of fluorescence to the individual tomographic section offers a depth profile of fluorescence compared with a flat conventional intensity profile. Three-dimensional (3D) reconstruction of the fluorescence distribution results in a 3D relief of vascular structures and dynamic changes such as perfusion and leakage. Hence, topographic angiography offers a realistic imaging of the lesion architecture, perfusion, and extravasation. The purpose of our study was to apply this recent technique of topographic image processing to two main CNV types, classic and occult, and to identify characteristic features. The diagnostic potential of the new method was further evaluated by comparison with conventional 2D angiography. 

Methods

Patient Selection

The study was performed according to the tenets of the Declaration of Helsinki. Informed consent was obtained from all subjects. Data for 158 patients (52.1% men, 47.9% women) were analyzed. Ninety-four eyes had classic or predominantly classic CNV, and 64 eyes had occult CNV. All patients presenting consecutively with exudative AMD in our clinic underwent conventional angiography. Lesion types were analyzed and labeled based on routine angiography as predominantly classic, minimally classic, or occult only or a different diagnosis. All consecutive patients with predominantly classic or occult-only lesions were included in the 3D topographic analyses based exclusively on conventional diagnosis without any specific selection or bias. 

Data Acquisition and Processing

A confocal SLO (Heidelberg Retina Angiograph; Heidelberg Engineering, Dossenheim, Germany) was used to perform all FA (5 mL of 10% fluorescein; Alcon Pharma GmbH, Freiburg, Germany) and ICGA (50 mg solution of ICG; ICG Pulsion, München, Germany) examinations. For complete sectioning of elevated lesions and to create a preretinal initial focus, a +3-D refractive correction was added by using the internal focus adjustment. An early FA/ICGA series of 32 tomographic sections was taken over a depth of 4 mm, each separated by 125 μm. A late series was produced after 10 minutes during FA and after 15 minutes during ICGA. Each data set was recorded within 1.6 seconds and was digitized to a grid of 256 × 256 pixels with an 8-bit intensity resolution. Images from a tomographic series were aligned to correct for artifacts due to eye movements. Based on the maximum fluorescence intensity for each image point of the 32 aligned sections, an intensity image was generated. The resultant intensity image indicated areas with high (bright) or low (dark) amounts of the fluorescent marker and was consistent with a contrast-enhanced version of conventional FA or ICGA. 

Topographic Analysis

Depth distribution of fluorescence within the individual serial planes was determined by extracting an axial intensity profile from the aligned stack of cross-sectional images for each point in the x/year plane. At a defined threshold intensity of 75%, the location of onset of fluorescence was identified, and subsequently an axial depth profile was generated. In an area with physiological chorioretinal vasculature, the topographic intensity distribution in the z-axis normally resembled a bell-shaped curve (Fig. 1A) . The choriocapillaris or the large retinal vessels with the highest dye concentration under physiological conditions are represented by the maximum level of the curve. The depth value on the z-axis indicated the precise location of threshold fluorescence for each lateral position. These sites are regularly situated on the surface of the fluorescent structure whether it be physiological (e.g., retinal vasculature) or pathologic (e.g., prominent choroidal neovascularization or extravasation of fluorescent molecules). As the absolute fluorescence intensity throughout the section was normalized to 1.0, the depth location was independent of the absolute intensity of fluorescence and depended only on the axial position where the fluorescence peak appeared first. Horizontal and vertical cross-sections through the CNV lesion represented the linear topographic profile (i.e., the specific surface configuration), at a given x/year position. 

Image Presentation

A 2D depth profile was obtained indicating the axial position of the onset of fluorescence plotted as a gray-scale image using a code of 256 scales (Fig. 1B) . Superficial localization appeared bright, and fluorescence from deeper layers appeared dark. Each lesion was represented in a set of three images: the conventional 2D intensity profile (Fig. 2A) , the 2D depth profile (Fig. 2B)and a 3D relief representing the realistic vascular morphology (Fig. 2C) . Areas with very weak fluorescence or areas where the highest tomographic section of the series already exceeded the threshold criterion, did not allow a reliable determination of fluorescence onset. Signals had to be above background threshold—that is, the difference of the maximum intensity and the minimal intensity of each axial scan had to be at least a factor of 2, and at least 75% of all data points had to be located within the 32 sections taken. For complete data transfer (5 minutes) and data analyses (15 minutes), a time investment of approximately 20 minutes should be calculated. 

Data Interpretation

All 2D and 3D images were analyzed according to the following characteristics. 

Masking.

Hypofluorescent areas that occur due to blocking phenomena by subneuroepithelial, subretinal, or epiretinal fluid or blood. 

Lesion Extension.

Lesion size was estimated in equivalents of MPS (Macular Photocoagulation Study) disc areas. 

Lesion Elevation.

Lesion prominence was described in three grades: flat, a prominence barely above the choroidal background (grade 1); moderate, a prominence with clear elevation at the level of a retinal vein (grade 2); and intensive, a prominence more than that of a central retinal vein (grade 3). 

Halo.

A halo was defined as an area appearing as a depression within the surrounding choriocapillary background. 

Geometric Characteristics.

The lesion configuration was described as a prominence with steep or flat borders and a central peak or crater compared with a regular convex prominence. 

Results

Topographic Imaging of Classic CNV Using Fluorescein

During the early phase of conventional FA, classic CNV was characterized by a hyperfluorescent area with or without a vascular pattern. Some lesions were surrounded by a dark ring zone mostly corresponding to blood or exudates (Fig. 2A)that reduced the visibility of the lesion borders in conventional FA. The gray-scale-coded intensity profile analyzed the specific location of hyperfluorescence and clearly highlighted prominent (i.e., bright) areas compared with deep (i.e., dark) areas (Figs. 1B 2B) . Because of the high confocal resolution and the normalization of fluorescence intensity independent of the absolute amount of dye, superficial masking was eliminated, and the borders of the lesion were visualized in detail. By 3D topography (Fig. 2C) , the classic membrane was displayed as an irregular, steep elevation encircled by a prominent ringlike structure consistent with the proliferating and actively leaking zone of the net. Already, the early-phase image revealed the true extension of the lesion better than the conventional technique limited by masking. Typically, the classic membrane appeared as a clearly demarcated lesion with prominent margins associated with lesion activity and a central crater corresponding to more inert areas. A dark halo was also noted surrounding the lesion (Fig. 2C) . 

During late-phase conventional FA, lesion size, and fluorescence intensity increased, and the lesion borders were blurred by diffuse leakage and masking phenomena, which prevented an accurate definition of the lesion’s extension (Fig. 3A) . The intensity profile delineated the borders and configuration of the lesion more realistically (Fig. 3B) . In the 3D relief (Fig. 3C) , the lesion was precisely mapped as a steep increasing hyperfluorescent zone in its typical configuration (i.e., prominent borders and central crater), which was found in 94% of the study cases. Leakage, particularly originating from the active capillary sprouts in the lesion periphery accounts for the increased prominence of the lesion borders during late-phase FA compared with the early-phase image. The surrounding halo was seen more impressively in late 3D FA. 

Topographic Imaging of Classic CNV Using ICG

One minute after ICG injection, conventional imaging (Fig. 4A)did not allow the accurate definition of location or extension of the lesion. In the area of the classic membrane, a zone of discretely visible neovascular patterns surrounded by a hypofluorescent margin was embedded within the choriocapillary background. The topographic technique identified the prominence and dimension of the lesion (Fig. 4B) . Plastic 3D imaging presented the CNV in its characteristic, sharply demarcated neovascular configuration (Fig. 4C) . A deep choriocapillary halo surrounding the proliferation became clearly visible. 

After 15 minutes, diffuse leakage from the choriocapillary layer had homogenized the background pattern, and the CNV appeared as focal hyperfluorescence within a hypofluorescent vicinity of masked fluorescence (Fig. 5A) . Analysis of the 2D image (Fig. 5B)allowed differentiation between the active parts of the CNV with exudation at the peripheral zones and less active parts in the center. 

By 3D topography (Fig. 5C)the profile of the CNV was still sharply demarcated, appeared typically ring-crater configured, and was consistently surrounded by a choroidal halo. 

Topographic Imaging of Occult CNV Using Fluorescein

The appearance of occult CNV in conventional FA was nonspecific, including numerous small, dispensed, and partly confluent leaking spots without providing any sufficient information about the lesion extension or vascular morphology (Fig. 6A) . According to this angiographic pattern, the lesion represented the definition of an occult CNV with fibrovascular pigment epithelial detachment. Depth analysis revealed a round, prominent lesion with a cohesive texture (Fig. 6B) . 3D topography (Fig. 6C)illustrated the occult CNV as a convex plaque with flat, but more distinct borders than seen with conventional FA. The internal structure appeared homogenous and solid, similar to the pattern of the surrounding choriocapillary, where it was embedded. During late-phase angiography, extravasation occurred in conventional as well as in 3D imaging. Late leakage highlighted the entire lesion area (Fig. 7A) . The lesion remained prominent, although less well demarcated by intensity profile (Fig. 7B)and 3D relief (Fig. 7C) . Compared with the configuration of a classic membrane, the occult lesion showed no central crater and its peripheral borders appeared flat. 

Topographic Imaging of Occult CNV Using ICG

By conventional ICGA (Fig. 8A) , the neovascular net offered little contrast compared with the surrounding choriocapillary layer. No precise information regarding extension or activity of the lesion was obtained. Axial analyses identified the lesion location and size (Fig. 8B) . A realistic morphology of the lesion was only displayed by 3D topography (Fig. 8C) . The lesion appeared markedly larger than assumed based on the conventional FA image, and its borders were not well delineated. By late conventional ICGA, the neovascular process appeared as a hypofluorescent plaque (Fig. 9A) . The location analysis revealed a prominent, well-perfused neovascular plaque (Fig. 9B) . Focal perfusion defects were detected at the temporal position of the lesion and were seen most clearly in the late 3D relief. 

Linear Cross Sections

Linear x/year sections may be used to delineate the surface configuration of background and lesion fluorescence in any location, similar to OCT sectional scanning. Figure 10shows vertical and horizontal cross-sectional images of classic and occult lesion types. The linear scan through the lesion surface highlighted the steep, ring-crater-configured elevation (Fig. 10B , thin arrow) as a typical criterion of classic CNV. Cross-section image of occult CNV profiled the lesion as a gently inclining elevation, without a central crater or surrounding halo. 

A halo, as already mentioned, is a typical finding associated with classic CNV. Per definition it is a sub-background area that surrounds the borders of the lesion and should be consistent with a relative defect in the vascular pattern of the choriocapillary layer. As shown in Figure 10B(thick arrows), a halo appeared much more often in classic than in occult CNV. Choroidal halos often appeared during early-phase imaging (FA and ICG dye), but were most easily detected during late-phase IGCA. 

Surface Configuration of Classic and Occult CNV

Vertical and horizontal cross-sections along the fluorescence profile are used to delineate the precise shape of the lesions. The configurations of classic and occult CNV were seen qualitatively in the 3D relief, but the characteristic features were highlighted more precisely and qualitatively in the cross section. Classic CNV appeared as a sudden distinct prominence with steeply inclining borders (Fig. 10A)and a central space, consistent with a crater configuration (Fig. 10B) . The surrounding choroid often appeared thinned compared with the normal background pattern. The surface of an occult neovascular lesion regularly appeared flat, with no distinct borders, consistent with a convex shape and no surrounding alteration (Fig. 10D) . 

Distribution of Lesion Characteristics by FA and ICGA

Table 1shows that classic CNV types appeared more often as prominent lesions by topographic FA, independent of early or late-phase imaging. This observation suggests that extravasation and accumulation of fluorescent molecules within the overlying retina occurred immediately after dye injection. Prominence by topographic FA was clearly related to leakage, as the highest classic lesions were seen during late-phase FA. By topographic ICGA, extravasation was limited, and most lesions appeared relatively flat (i.e. mostly intraluminal fluorescence was imaged and leakage towards overlying layers was limited). 3D ICGA is therefore likely to provide a realistic histology-like morphology of classic CNV. The characteristic morphology for classic CNV types is a well-demarcated lesion with steep borders and deep center, labeled as a “crater” configuration in our study. This configuration was found in 92% to 94% by topographic FA and in 83% by topographic ICGA. Occult CNV consistently appeared as a flat plaque by all topographic modalities during all phases in most cases, about one third of occult lesions were primarily more prominent. Intense leakage did not appear to be a relevant feature in occult lesion types, as no differences in lesion prominence, including the neovascular channels and extravasated dye molecules, were seen between early- and late-phase topographic imaging. A flat, convex configuration was seen in ∼90% of occult lesions in either modality and during either phase. The extensive leakage from the borders of the CNV, present so invariably in classic CNV, was absent in occult lesion types. Choroidal halos were identified in one third of all classic lesions, but were in general a rare finding in occult lesion types. 

Comparison of Lesion Features by Topographic and Conventional Imaging

As seen in Table 2 , by topographic FA, classic CNV often appeared smaller, whereas it usually appeared larger in topographic ICGA, when compared with conventional imaging. Diffuse leakage of fluorescein into the collateral tissue may account for an overestimation of the classic lesion size by conventional FA. However, the true size of classic CNV, as imaged by topographic ICGA may often be larger than anticipated by conventional ICGA, because it is masked by the overlying hemorrhage or fluid. Masking was rarely a hindrance in topographic imaging, either by FA or by ICGA. 

Similar conclusions apply to the topography of occult lesions: Some leakage in late-phase conventional FA may account for overestimation of the size of the neovascular net, whereas topographic ICGA, eliminating masking phenomena, offers a reliable method for identifying the entire occult net. Note that conventional ICGA, particularly designed to image occult lesions, strongly underestimated the real size of occult lesions, and ∼90% of occult CNV was larger by topographic ICGA imaging. Again, masking was not a problem in confocal fluorescence analysis, even with occult lesions. An irregular perfusion pattern with vascular “defects” was appreciated mostly during early FA topography and reflected dynamic perfusion irregularities within the neovascular net of classic lesions. 

Discussion

Impaired perfusion, exudation, and proliferation as a consequence of vascular diseases are common reasons for severe irreversible vision loss in patients of every age. Since a successful therapy strongly depends on the exact definition of the underlying disease and its activity, the need for effective diagnostic tools is obvious. The most common diagnostic procedure in all-day clinical routine is conventional angiography with fluorescein, FA, and occasionally ICGA. Although particularly FA is considered as the gold standard, the diagnostic value is limited by masking phenomena, low contrast and resolution, and insufficient depth resolution. The major drawback of conventional angiography is the capacity for accurate differentiation of leakage from staining due to the flat 2D representation based purely on fluorescence intensity. In contrast, topographic angiography enables analysis of fluorescence location independent of absolute intensity and to identify even weak fluorescent signals. Lesions hidden underneath exudates or hemorrhage or located underneath the RPE can be easily identified. Topographic fluorescence analysis, therefore, offers a realistic 3D anatomy of the neovascular proliferation and clearly identifies amount, location, and dynamics of leakage, independent of staining. 

This study uses confocal topographic angiography to identify characteristic features of classic and occult CNV types with a specific focus on neovascular morphology and leakage characteristics. Both modalities—FA and ICGA—were used for topographic analysis of imaging characteristics. Different timings, early and late phase images, were included. In addition, a comparison with conventional angiographic features was performed. 

Most important, consistent differences in lesion morphology were found. Classic CNV appeared as a characteristic, crater-configured, well-demarcated prominence. As the sensitive confocal topography detected even weak fluorescence signals, the lesion’s prominence clearly included intra- as well as extravascular fluorescence originating from leakage. Observation of lesions with leakier fluorescein and less leaky ICG molecules during different phases allows a precise documentation of leakage patterns. In classic lesion types, the most intense extravasation was found along the borders of the lesion, mainly where neovascular proliferation was most active. In many cases, the classic lesion was surrounded by a halo, most likely consistent with an alteration in the choroidal vascular perfusion. Although this halo may be primarily due to fluid’s pooling in the subretinal space leading to a compression of the underlying choriocapillaris, or secondarily, choroidal perfusion changes may be a reason for this characteristic feature. Occult CNV types presented as flat convex lesions without a central depression or surrounding halo. Extravasation does not change the morphology over time and with different dyes; hence, most likely intraretinal leakage is less pronounced than with classic lesion types. The flat and smooth configuration also supports the hypothesis that occult lesions are histologically buried underneath an intact RPE layer. ^{16 17} Many attempts have been made to obtain a realistic histology of CNV lesions. ¹⁷ Topographic angiography appears to offer such a realistic morphologic identification, and since it is performed over time, it also includes significant information about growth and leakage patterns. 

With conventional imaging, the true dimension of CNV lesions can only be estimated, whereas in topographic imaging the membrane shows defined margins. Obviously, the truest evaluation is histology and clinical-pathologic correlation. However, 3D topography offers a much more realistic identification of the lesion borders, because the confocal technology depicts the most subtle neovascular sprout in its respective layer and, since even the lowest fluorescence intensity levels are documented, any leakage along the margins of the CNV lesion is clearly identified. Masking is completely eliminated as a limiting factor. Studies on the reproducibility and agreement of angiographic interpretation of conventional angiography have shown a significant diversity among individual readers and centers. ^{18 19} Computer-assisted demarcation of lesion characteristics reduced the variability in physicians’ estimates, but the reliability largely depends on the algorithms selected. ²⁰ Topographic analysis is more sensitive and includes more factors than just fluorescence intensity and brightness for diagnostic conclusions. The differentiation of leakage and staining is a major issue in angiographic evaluation. Whereas it is hard to make this decision based on conventional angiography, topography identifies characteristic leakage patterns and a quantitative documentation of extravasation compared with staining. 

Optical coherence tomography (OCT) is used to supplement the information missing by angiography and will play an important role in diagnosis and treatment follow-up of macular disease. ^{21 22} However, OCT has the disadvantage that it does not allow the precise appreciation of the morphology and extension of the neovascular complex, ²³ but rather identifies the consequences of neovascular growth such as extravasation and fluid pooling. Furthermore, the radial-scanning principle of OCT may overlook pathologic areas, even if recent appendages using grid scanning protocols seem to minimize this problem. ²⁴ On the one hand, the novel raster scanning mode introduced by the ultra-ultra-high fast-resolution OCT systems allow a realistic topographic OCT imaging. ²⁵ The conventional FA mode, on the other hand, is unable to detect and quantify extravascular exudates and is minor in the detection of cystoid spaces. ²² More sophisticated imaging modalities or combination of various methods, ²⁶ may resolve this dilemma. If 3D topography (e.g., using ICG) offers a better differentiation capability, such analyses may also reflect vision. Nazemi et al. ²⁷ reported differences in results obtained with the 3D Amsler grid. Central visual acuity using ETDRS standards only highlights visual function in the foveal center. Other functional tests such as multifocal ERG or microperimetry should be correlated with topographic angiography to enable analyses of angiographic anatomy and neurosensory function. 

Because novel therapies for exudative AMD have been developed during the past years, of which some are approved, others are emerging, the analyses of lesion response in terms of neovascular growth/regression and leakage activity are essential for therapeutic decisions. With conventional angiography the dynamic of leakage is often unclear. Usually, the fluorescence homogenizes over time, and the true origin of the dye accumulation cannot be defined particularly as staining by fibrotic components, which increases during the course of the disease. 

Until recently, angiographic lesion composition was suspected to be an important prognostic factor. However, subgroup analyses of the TAP and VIP data from the trials using photodynamic therapy raised doubts about this hypothesis. ²⁸ Rather, lesion size was identified as the only relevant factor for prognosis. Some investigators, using pharmacologic intervention such as inhibition of vascular endothelial growth factor (VEGF), state that neither lesion composition nor lesion size is relevant in treatment outcomes, as reported in the trial of the safety and efficacy of pegaptanib (Macugen trial). ⁵ Another paper reported a difference in outcome dependent on baseline lesion morphology. ⁴ Another clinical trial using a novel antiproliferative substance (anecortave acetate) and the standard PDT regimen, treatment outcomes were poor in both arms due to the specific anatomy of the lesions included with small lesions and a large proportion of retinal angiomatous proliferation (RAP). ²⁹ Other lesion entities such as RAP lesions and polypoidal choroidal CNV, among others, should be better classified and be studied with topographic angiography to understand their biology and possible response to treatment. 

Documentation of leakage activity is most important, as most anti-VEGF drugs work on antipermeability. Conventional FA is not a modality to allow quantification of leakage activity compared with staining, topographic modalities can easily offer such qualities. Furthermore, the different lesion types, classic, and occult, were shown to have different morphologic and leakage characteristics and may therefore need custom-made treatment regimen or dosages. When ranibizumab was used in predominantly classic CNV (ANCHOR trial), the higher dose of 0.5 mg appeared to be more effective, whereas there was no difference seen between high- and low-dose regimens in the trial including only occult and minimal classic lesions (Webster MK et al. IOVS 2006;47:ARVO E-Abstract 2206). Based on conventional angiography, some investigators ⁵ have found continuous growth of CNV despite continuous therapy, whereas others claim a complete inhibition of further neovascular growth (Brown DM et al. IOVS 2006;47:ARVO E-Abstract 2963). A three-dimensional modality would clearly offer a more reliable measurement of CNV size. 

Regarding the rapid development and the multitude of novel strategies in the treatment of CNV during the past few years and since most mechanisms of therapeutic efficacy are still unclear, an exact, detailed imaging of CNV morphology and activity is warranted for diagnostic identification as well as therapeutic monitoring. CNV is not a homogenous entity, as the lesion types differ anatomically and in leakage behavior and an “allcomers” approach may not be an adequate strategy to optimize treatment. The results of the recent antiangiogenic therapy trials clearly indicate, that the conventional morphology concepts and lesion type ideologies do not apply any more. Other characteristics such as the exudative morphology of a lesion will be considered for prognosis evaluation. Novel imaging modalities will help to improve the understanding of the biology of neovascular disease and to develop adequate guidelines for treatment evaluation. 

 Submitted for publication June 21, 2006; revised September 17, 2006; accepted January 26, 2007.

 Disclosure: U. Schmidt-Erfurth, None; K. Kriechbaum, None; A. Oldag, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) Intensity profile curve showing the intensity and onset of hyperfluorescence at a specific x/year location, following the z-scan through all 32 tomograms. The peak of the curve represents the level of the choriocapillaris or large vessels with high concentration of dye (i.e., high fluorescence intensity). (B) Gray-scale-coded 2D profile demonstrates the distribution of the axial fluorescence within the entire angiographic field: Brightness indicates prominent localization, darkness indicates deep localization. In this example, the typical ring-crater configuration of a classic membrane is shown.

Figure 2.

 

(A) Early-phase conventional FA intensity image of a classic CNV demonstrating the distribution of fluorescence intensity, with central leakage of the CNV surrounded by subretinal hemorrhage. (B) Gray-scale-coded 2D depth image identifying the axial location of fluorescence, the CNV lesion’s prominence, and the ring-shaped configuration of the lesion’s borders. (C) 3D topographic relief of the same lesion. The 3D perspective allows visualization of the lesion with sharp demarcation and a typical ring-crater configuration.

Figure 3.

 

(A) Late-phase conventional FA intensity image 10 minutes after dye injection showing the same classic lesion as in Figure 2 , with intensive central leakage, masking by blood and surrounding subsensory extravasate. (B) Gray-scale-coded 2D depth image detecting the prominence of fluid pooling over the lesion site. (C) The topographic relief delineates the 3D distribution of fluorescent leakage with enhanced prominence of the hyperpermeable borders of the CNV.

Figure 4.

 

(A) Conventional ICGA intensity image of the classic lesion seen by FA in Figures 2 and 3 . (B) In the gray-scale-coded 2D depth image the prominence of the neovascular net is highlighted by its brightness. (C) The 3D topographic image of the same lesion demonstrates the vascular configuration of the lesion as well as perfusion changes within the surrounding choriocapillary bed. A halo, a dark ring zone that surrounds the lesion, is identified.

Figure 5.

 

(A) Conventional ICGA intensity image of the identical classic lesion taken 15 minutes after ICG administration. The choriocapillary fluorescence appears homogenous without distinction of a vascular pattern due to extravasation of the dye molecules. The CNV is stained more intensively than during the early phase, suggesting pooling or leakage of ICG. (B) The 2D depth image identifies the prominence of the entire lesion as well as of individual areas within the lesion. (C) A prominent ring of fluorescent leakage is seen along the borders of the lesion in the topographic relief. As with FA the shape of the CNV is craterlike with elevated margins indicative of peripheral leakage also with ICGA.

Figure 6.

 

(A) Early-phase conventional FA intensity image of an occult lesion demonstrates mild hyperfluorescence without precise demarcation or pattern. (B) The gray-scale-coded 2D depth image identifies an area of fluorescence located within superficial layers and therefore appearing bright, regardless of the low overall intensity of fluorescence. (C) The 3D topographic relief of the same occult lesion identifies a prominent fluorescent plaque with flat borders.

Figure 7.

 

(A) Late-phase conventional FA image of the identical occult lesion. Fluorescein was used as the dye. (B, C) Gray-scale-coded 2D depth image and 3D topographic images of the same lesion 10 minutes after dye injection. The occult lesion appears as a flat, well-demarcated area without a central crater or halo. No prominent leakage with extravasation from the borders of the lesion is noted compared with the classic lesion seen in Figures 3 4 5 6 .

Figure 8.

 

(A) Conventional intensity image of the occult lesion seen in Figures 6 and 7 . The CNV was embedded within the vascular pattern of the choriocapillary layer delineated with ICG. (B) The gray-scale-coded 2D depth image demonstrates the site of the lesion and its superficial location compared with the surrounding physiological choriocapillaris. (C) In the 3D topographic image, the prominence and flat configuration of the occult lesion is documented.

Figure 9.

 

(A) Fifteen minutes after ICG injection the choriocapillary pattern has faded, the lesion appears as late plaque by conventional angiographic intensity imaging. (B) The 2D depth image detects the ICGA pooling within the superficial layers of the tomogram. (C) The topographic relief delineates the irregular pattern of the choriocapillary layer as well as the overlying neovascular CNV net. Distinct perfusion defects are recognized at the temporal aspect of the occult lesion.

Figure 10.

 

Vertical and horizontal cross-section images of classic CNV (A, B) and occult CNV (C, D). (A, B) The characteristic shape of classic CNV: steep borders, ring crater configuration (thin arrow) and a halo (thick arrows), defined as a sub-background area that surrounds the borders of the lesion and should be consistent with a relative defect in the vascular pattern of the choriocapillary layer. (C, D) The configuration of occult CNV as a slowly inclining elevation without steep borders, central crater or halo, which appears much more often in classic CNV than in occult CNV.

Table 1.

 

Lesion Characteristics (Topographic Imaging)

	Classic CNV							Occult CNV
		FA			ICGA			FA			ICGA
		Early	Late	Early	Late	Early	Late	Early	Late
Prominence
Flat	46.8	35.1	74.5	72.3	70.3	79.7	70.6	65.6
Moderate	45.7	48.9	25.5	26.6	29.7	20.3	28.1	32.8
High	7.5	16	0	1.1	0	0	1.3	1.6
Configuration
Crater	91.5	93.6	83	83	10.3	12.5	10.3	12.5
Convex	8.5	6.4	17	17	89.7	87.5	89.7	87.5
Halo
Present	34	31.9	36.2	48.9	12.5	12.5	9.4	17.2
Absent	66	68.1	63.8	51.1	87.5	87.5	90.6	82.8

 Lesion prominence was graded as flat (barely detectable), moderate (prominence reaching the elevation of a retinal vessel), or high (exceeding a vein prominence). The configuration was described as a crater (concave with a flat center surrounded by prominent borders) or convex (flat borders and a hill-like regular prominence in the center of the lesion). The data represent the relative proportion (%) of patients demonstrating the specific feature.

Table 2.

 

Imaging Characteristics

	Classic CNV							Occult CNV
		FA			ICGA			FA			ICGA
		Early	Late	Early	Late	Early	Late	Early	Late
Size*
Larger	13.8	3.2	94.7	90.4	67.2	29.7	93.8	87.5
Smaller	86.2	96.8	5.3	9.6	32.8	70.3	6.2	12.5
Masking, †
No	93.6	72.3	90.4	79.8	95.3	95.3	89.1	93.8
Yes	6.4	27.7	9.6	20.2	4.7	4.7	10.9	6.2
Perfusion defects, †
Present	78.7	17	16	9.6	6.3	4.7	6.3	12.5
Absent	21.3	83	84	90.4	93.7	95.3	93.7	87.5

 The data show that classic CNV is imaged most realistically by topographic ICGA, as compared with FA, with respect to demarcation and extension. Masking phenomena (hemorrhage, pigment, and exudate), which always influence the evaluation of conventional angiography, had no impact on topographic imaging, because even low fluorescence intensities were detected. Choroidal perfusion defects surrounding the CNV lesion were seen more often in classic CNV (78.7%, early FLA) than in occult CNV (12.5% late ICG). The data represent the relative proportion (%) of patients demonstrating the specific feature.

*  By topographic versus conventional angiography.

†  Present in topographic angiography.