Overview

Blurred vision and distortion, especially distorted near vision, are the symptoms most patients with CNV notice first.^3,9 Patients also may complain of decreased vision, micropsia, metamorphopsia, or a scotoma; however, many times they volunteer no symptoms or report only vague visual complaints.⁸ Symptoms generally arise from subretinal fluid, intraretinal fluid, blood, or destruction of photoreceptors and the retinal pigment epithelium (RPE) by fibrous or fibrovascular tissue.^10–13 In some cases, areas of distortion or scotoma can be mapped out on an Amsler grid. Visual acuity, although frequently decreased, may not always be affected. Functional vision generally declines in accordance with Snellen visual acuity. Thus patients with poor Snellen acuity generally report decreased ability to perform functional tasks (e.g., face recognition, telling time) with the affected eye.¹⁴

In some patients with AMD, CNV may appear as a gray–green elevation of tissue deep to the retina with overlying detachment of the neurosensory retina (Fig. 66.1). The gray–green color may arise from hyperplastic RPE in response to the CNV,¹⁵ as has typically been seen in patients, usually younger individuals, with ocular histoplasmosis syndrome (OHS), pathologic myopia, and other conditions complicated by CNV. This gray–green appearance is not always present in older individuals with AMD. Often, the presence of blood or lipid or a sensory retinal detachment in an elderly patient with vision loss indicates the presence of CNV. The CNV capillary network may become more apparent when the overlying RPE has atrophied. Occasionally, a shallow neurosensory detachment may be the only presenting sign of underlying CNV. Elevated RPE, also termed a pigment epithelial detachment (PED), even without overlying subretinal fluid, may also suggest the presence of CNV to be identified subsequently by a fluorescein angiogram. RPE folds beneath a shallow RPE elevation usually indicate the presence of CNV.¹⁶ These subtle clinical findings can easily be missed without careful stereoscopic slit-lamp biomicroscopic examination, facilitated with a contact lens.

Fig. 66.1 Fundus photograph of choroidal neovascularization. Note area of hemorrhage (large arrows), as well as neurosensory retinal detachment (small arrows).

(Reproduced with permission from Elman MJ. Age-related macular degeneration. Int Ophthalmol Clin 1986;26:117–44.)

Retinal pigment epithelial detachments

Retinal PEDs appear clinically as sharply demarcated, dome-shaped elevations of the RPE (Fig. 66.2). They usually transilluminate if they are filled with serous fluid only. Often, there is accompanying RPE atrophy and “pigment figure” formation. Pigment figures, a reticulated pattern of increased pigmentation extending radially over the PED, indicate chronicity of disease and probably have no prognostic significance. Although an overlying sensory retinal detachment may be a clue to the presence of CNV beneath a PED,¹⁷ sometimes a shallow neurosensory detachment may occur as a result of breakdown of the physiologic RPE pump or from disruption of the tight junctions between adjacent RPE cells in the absence of CNV. Unlike a PED, the borders of a neurosensory detachment are not sharply demarcated. The presence of a PED may or may not be a feature of CNV. The fluorescein angiographic pattern (see subsequent discussion) can differentiate a drusenoid PED,¹⁸ which does not have CNV, from a fibrovascular PED, which is a form of occult CNV,¹⁹ as well as from a serous PED, which may or may not overlie an area with CNV.¹⁹ Several clinical signs suggest the presence of CNV underlying an area of PED identified biomicroscopically, including overlying sensory retinal detachment and lipid, blood, and chorioretinal folds radiating from the PED.³ Blood within or surrounding a PED implies the presence of CNV (Fig. 66.3). When confined to the sub-RPE space, the blood may appear as a discretely elevated, green or dark-red mound. The hemorrhage can dissect through the RPE into the subsensory retinal space or into the retina. Rarely, blood may pass through the retina into the vitreous cavity, causing extensive vitreous hemorrhage. The Submacular Surgery Trials (SST) Research Group suggested that this event was more likely in predominantly hemorrhagic lesions that were large (>12 disc areas) or associated with very poor visual acuity (worse than 20/1280 Snellen equivalent).²⁰

Fig. 66.2 Fundus photograph in which a round, sharply demarcated mound indicates the detached retinal pigment epithelium.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

Fig. 66.3 Hemorrhagic retinal pigment epithelial detachment. (A) Sketch of hemorrhagic detachment in which the blood has also dissected underneath the sensory retina. (B) Fundus photograph of hemorrhagic pigment epithelial detachment.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

Breakthrough vitreous hemorrhage

In most cases of neovascular AMD, the peripheral visual field remains unaffected. If bleeding breaks through the retina into the vitreous cavity, however, patients may complain of severe and sudden visual loss involving the peripheral visual field, as well as the central field. This may be accompanied by pain believed to result from stretching of the nerve fibers within the choroid.²¹

Massive subretinal hemorrhage

Massive subretinal hemorrhage is an unusual complication of neovascular AMD. If – extremely rarely – total hemorrhagic retinal detachment occurs, secondary angle closure glaucoma may develop. These patients may report sudden visual loss followed by pain.²² Anticoagulation therapy may contribute to massive subretinal hemorrhage. In one report,²³ 19% of AMD patients with massive subretinal hemorrhage were taking sodium warfarin or aspirin. Although sodium warfarin therapy may have contributed to the massive subretinal hemorrhage, the antiplatelet therapy was likely a chance association because several Macular Photocoagulation Study (MPS) reports did not observe any increased risk of hemorrhage associated with the use of aspirin.^24–26 Furthermore, comparing baseline characteristics in study participants with predominantly choroidal neovascular lesions in the SST Group N Trial²⁷ with participants with predominantly hemorrhagic lesions,²⁰ no difference in use of aspirin was detected. Recent epidemiology studies found an association of aspirin use with neovascular AMD, but this finding does not necessarily confirm or refute an association of aspirin with the development of predominantly hemorrhagic lesions or massive subretinal hemorrhages.²⁸

We believe the strongest evidence suggests that patients with AMD who need to follow a regimen of aspirin therapy should continue to do so without unnecessary fear they will increase their risk of vitreous hemorrhage.

Retinal pigment epithelial tears

RPE dehiscence or tears have been described as a complication associated with CNV, often in an eye with a serous or fibrovascular PED, and secondary to or unassociated with laser photocoagulation.^29–33 One report³⁴ suggested that CNV underlying a detached RPE can contribute to RPE tear formation. Tears occur at the junction of attached and detached RPE, perhaps when the PED can no longer resist the stretching forces from the fluid in the sub-RPE space emanating from the underlying occult CNV (Fig. 66.4) or from the contractile forces of the underlying fibrovascular tissue that may be intimately associated or entwined with the overlying RPE. When the RPE tears, the free edge of the RPE retracts and rolls toward the mound of fibrovascular tissue. Acutely, a serous detachment of the sensory retina may be caused by the leaking of fluid from the exposed choriocapillaris.¹⁸ This is rarely seen after a few days following the tear.

Fig. 66.4 Sketch of tear or rip of the retinal pigment epithelium (RPE), showing contracted RPE tear. Cc + C, choriocapillaris and choroid.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

Disciform scars

Histologically, CNV usually is accompanied by fibrous tissue, even when no fibrous tissue is readily apparent on initial presentation to an ophthalmologist.^10,35,36 This fibrous tissue may be accompanied by CNV (fibrovascular tissue) or not (fibroglial tissue).¹⁰ The fibrous tissue complex may be beneath the RPE (usually proliferating within the inner aspect of an abnormally thickened Bruch’s membrane), and has been termed type I, or between the RPE and the photoreceptors, termed type II.³⁴ While some people speculate that these types differentiate classic CNV from occult CNV,^37,38 there is little evidence to support that these histologic types are always differentiated by histopathologic correlation.³⁹ Often, over time, the plane of the RPE is destroyed by the fibrovascular or fibroglial tissue, so the location of the CNV with respect to the RPE can no longer be identified readily. When the fibrous tissue becomes apparent clinically, the CNV and fibrous tissue complex may be termed a disciform scar.

Clinically, disciform lesions may vary in color, although typically they appear white to yellow. Hyperpigmented areas may be present depending on the degree of RPE hyperplasia within the scar tissue. Disciform fibrovascular scars may continue to grow, with neovascularization recurring along the edge, invading previously unaffected areas (Fig. 66.5). Varying degrees of subretinal hemorrhage and lipid may overlie or surround the scar. Occasionally, fibrovascular scars may precipitate massive transudation of fluid, mimicking a retinal detachment. The scars may be accompanied by massive lipid, as might be seen in retinal telangiectasis from Coats disease, and hence are sometimes called a “senile Coats response” in AMD. Disciform scars occasionally masquerade as choroidal tumors when much pigment is seen.³⁴ Not infrequently, anastomoses are observed between the retina and the fibrovascular tissue.¹³ As a rule, most fibrovascular scars involve the fovea and cause severe visual loss. However, surviving islands of intact photoreceptor cells noted histologically may explain the better visual performance than would be predicted from the morphologic appearance alone in some scars. Reading vision, rarely better than 20/200, becomes severely compromised in most cases with extensive scars.

Fig. 66.5 Disciform scar. (A) Sketch demonstrating that most of the sensory retina, pigment epithelium, and inner choroid has been replaced by a fibrovascular scar. (B) Fundus photograph of a disciform scar after choroidal neovascularization. (C) Fundus photograph of a disciform scar in which continued subretinal fluid and lipid from persistent choroidal neovascularization at the periphery of the fibrous tissue can be seen.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

Overview

Whenever one suspects CNV for which treatment might be indicated, one should consider obtaining stereoscopic fluorescein angiography promptly (Fig. 66.6), even in an era of optical coherence tomography (OCT). The treating ophthalmologist is about to embark on a recommendation for treatment involving drugs which carry risks, potentially large expenses, and requiring many years of follow-up. Although the clinical picture may be “obviously” CNV, other lesions masquerading for CNV can exist (see below) having a fluorescein angiogram at the time of diagnosis reduces the possibility that an error in diagnosis will be made. In addition, fluorescein angiography frequently allows one to determine the pattern (classic or occult), boundaries (well defined or poorly defined), composition (e.g., predominantly CNV, predominantly classic CNV, predominantly CNV with a minimally classic composition, predominantly CNV with an occult with no classic composition, predominantly hemorrhagic), and location of the neovascular lesions with respect to the geometric center of the foveal avascular zone (FAZ). Although many physicians no longer refer to CNV composition, entry criteria for many and even most of the treatment trials cited later in this chapter relied in part on lesion composition. If one chooses to treat only patients who would have been eligible for, e.g., the MARINA trial, baseline angiography is necessary. High-quality stereoscopic fluorescein angiograms, together with meticulous slit-lamp biomicroscopic examination (ideally with a contact lens examination using topical anesthesia and hard contact lens wetting solution to avoid degradation of any subsequent image acquisition that might occur if an ophthalmic demulcent is used), facilitate detecting obvious and subtle features of CNV on angiography.^14,19,40 It should be noted that the descriptive terms below refer to patterns of fluorescence on fluorescein angiography that have been shown to be reliable and reproducible in multicenter clinical trials,^19,41,42 and in practice, and are not related to terms based on other imaging such as OCT, indocyanine green angiography, histopathology, or immunohistochemistry.

Fig. 66.6 Algorithm for management of neovascular age-related macular degeneration suspected or diagnosed on clinical examination. AMD, age-related macular degeneration; CNV, choroidal neovascularization; OCT, optical coherence tomography; MPS, Macular Photocoagulation Study; VEGF, vascular endothelial growth factor.

(© Neil M. Bressler, MD, Johns Hopkins University, 2011.)

Classic choroidal neovascularization

The fluorescein angiographic appearance of classic CNV consists of a discrete, well-demarcated focal area of hyperfluorescence that can be discerned in the early phases of the angiogram, sometimes before dye has completely filled the retinal vessels during choroidal filling.^19,41,42 Although fluorescein can occasionally be observed within the actual capillary network of CNV in the early phase of the angiogram (Fig. 66.7A), the ability to visualize the appearance of actual new vessels is not needed to diagnose classic CNV and is not a specific feature of classic versus occult CNV.^19,41–43 Since both classic and occult patterns of CNV contain new vessels histologically, early-phase angiography may be able to demonstrate these vessels in either pattern. As the angiogram is evaluated within the area of classic CNV, hyperfluorescence increases in intensity and extends beyond the boundaries of the hyperfluorescent area identified in earlier phases of the angiogram through mid- and late-phase frames. Fluorescein may also pool in subsensory retinal fluid overlying the classic CNV (Fig. 66.7B), best seen when visualizing early- and late-phase frames of classic CNV on stereoscopic images. This presentation of classic CNV is in contrast to the appearance of an area of RPE atrophy on fluorescein angiography. RPE atrophy, like classic CNV, is hyperfluorescent during the early phase of the angiogram (Fig. 66.8A). The increased fluorescence through the atrophic patch results from increased transmission of fluorescein through an overlying RPE with a reduced amount of pigment that normally obscures the choroidal blush (sometimes termed a window, or transmission, defect). Unlike the increase in extent and intensity of hyperfluorescence due to leakage from the fluorescence of classic CNV, RPE atrophy does not show leakage of fluorescein at its boundaries through the mid- and late-phase frames. The fluorescence fades after several minutes (Fig. 66.8B), without leakage of fluorescein beyond the boundaries of hyperfluorescence defined in the early stages. Two other lesions in AMD that may show an area of discrete hyperfluorescence in the early phase of the angiogram include a serous PED and a rip or tear of the RPE (angiographic features that differentiate these abnormalities from classic CNV are discussed later). Neither one of these latter abnormalities should show fluorescein leakage in later phases of the angiogram at the boundary of the hyperfluorescence noted in earlier phases.

Fig. 66.7 (A) Early transit phase of fluorescein angiogram showing fine net of vessels corresponding to part of choroidal neovascular lesion (black arrows). (B) Late phase of the fluorescein angiogram, demonstrating an increase in the degree and size of fluorescence. In both panels (A) and (B) there is blocked fluorescence resulting from overlying hemorrhage (white arrows).

(Reproduced with permission from Elman MJ. Age-related macular degeneration. Int Ophthalmol Clin 1986;26:117–44.)

Fig. 66.8 (A) Transit phase of fluorescein angiogram, showing hyperfluorescence corresponding to atrophic zones of the retinal pigment epithelium (transmission, or window, defect) and easily visualized choroidal vessels (too large to be vessels of choroidal neovascularization). (B) Hypofluorescence does not increase in size and fades with the later phases of the angiogram. This is in contrast to the pattern seen in choroidal neovascularization (Fig. 66.5).

(Reproduced with permission from Elman MJ. Age-related macular degeneration. Int Ophthalmol Clin 1986;26:117–44.)

Occult choroidal neovascularization

Occult CNV refers to two hyperfluorescent patterns on fluorescein angiography.^19,41,42 The first pattern, termed a fibrovascular pigment epithelial detachment (FVPED), is best appreciated with stereoscopic views, usually at approximately 1–2 min after dye injection. It appears as an irregular elevation of the RPE, often stippled with hyperfluorescent dots (Fig. 66.9). The boundaries may or may not show leakage in the late-phase frames as fluorescein collects within the fibrous tissue or pools in the subretinal space overlying the FVPED. The exact boundaries of a FVPED can usually be determined most accurately only when fluorescence sharply outlines the elevated RPE. The amount of elevation depends on the quality of the stereoscopic photographs and the thickness of the fibrovascular tissue. Stereoscopic pairs of fluorescein angiogram frames can sometimes facilitate identification of the boundaries of the elevated RPE, although not always, as the elevation can slope gradually down to the normal level of the RPE. The second pattern, late leakage of an undetermined source (Fig. 66.10), refers to late choroidal-based leakage in which there is no clearly identifiable classic CNV or FVPED in the early or mid-phase of the angiogram to account for an area of leakage in the late phase. Often this pattern of occult CNV can appear as speckled hyperfluorescence with pooling of dye in the subretinal space overlying the speckles. Usually the boundaries of this type of CNV cannot be determined precisely, and a lesion with this component should not be considered for photocoagulation treatment if it contributes to poorly demarcated boundaries of the lesion.

Fig. 66.9 Classic and occult choroidal neovascularization (CNV) with well-demarcated borders. (A) Color photograph shows scar from prior photocoagulation surrounded by several clues to suggest recurrent CNV, including subretinal hemorrhage, subretinal lipid, irregular elevation of retinal pigment epithelium (RPE) below the area of prior laser treatment, and overlying subretinal fluid. (B) Early phase of fluorescein angiogram shows area of classic CNV, scar from prior laser treatment, and irregular elevation of RPE with stippled hyperfluorescence representing fibrovascular pigment epithelial detachment (PED) inferior and temporal to the scar. (C) Stereoscopic photograph of the same eye 1 min after fluorescein injection. Note fluorescein leakage already apparent from classic CNV and increased intensity of stippled hyperfluorescence corresponding to fibrovascular PED. The boundaries of the fibrovascular PED remain well demarcated. (D) Angiogram taken 10 min after fluorescein injection shows persistence of fluorescein staining and leakage within a sensory retinal detachment overlying the lesion. It is difficult to determine the precise demarcation of fluorescence outlining the elevated RPE from these photographs alone. Although a fairly well-demarcated border can be seen in (C), the intensity of fluorescence at the boundary of elevated RPE is quite irregular in these late-phase photographs, with some areas fading relative to fluorescence of remaining areas of elevated RPE (D). (E and F) Composite drawings using multiple stereoscopic photographs from angiogram show interpretation of the boundaries of the lesion. At each clock-hour, the boundary of the lesion is clearly demarcated; the lesion included classic CNV, which occupies the foveal center.

(Reproduced with permission from Macular Photocoagulation Study Group. Subfoveal neovascular lesions in age-related macular degeneration: guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch Ophthalmol 1991;109:1242–57.)

Fig. 66.10 Occult choroidal neovascularization (CNV) with poorly demarcated boundaries accompanied by classic CNV. (A) Subretinal fluid and hemorrhage in eye with drusen. (B) Early phase of fluorescein angiogram shows both feeder vessels to classic CNV and fibrovascular pigment epithelial detachment (PED). Blocked fluorescence due to thick blood obscures inferior boundary of occult CNV. (C) Mid-phase stereoscopic photographs of angiogram show leakage from classic CNV. (D) Late phase of angiogram shows other areas of late leakage of undetermined source with no discernible, discrete, well-demarcated area of hyperfluorescence from classic CNV or fibrovascular PED detectable in early or mid-phase frames of angiogram that might be considered a source of late leakage.

(Reproduced with permission from Macular Photocoagulation Study Group. Subfoveal neovascular lesions in age-related macular degeneration: guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch Ophthalmol 1991;109:1242–57.)

Other terms relevant to interpreting fluorescein angiography of choroidal neovascularization

The terms lesion component versus lesion are important to differentiate in the discussion of fluorescein interpretation and treatment of CNV.^19,41,42 Lesion component is classic or occult CNV or any of four angiographic features that could obscure the boundaries of classic or occult CNV. These four features include: (1) blood that is visible on color fundus photographs and thick enough to obscure the normal choroidal fluorescence; (2) hypofluorescence due to hyperplastic pigment or fibrous tissue, or blood not visible on color fundus photographs; (3) a serous detachment of the RPE (Fig. 66.11); and (4) scar from CNV which either stains or blocks fluorescence (depending on the extent of RPE within the scar). The first two of these four features block the angiographic view of the choroid, making it impossible to determine whether CNV is located in the area of this component. The bright, reasonably uniform, early hyperfluorescence associated with a serous detachment of the RPE (described later) may obscure hyperfluorescence from classic or occult CNV and therefore interfere with the ability to judge whether CNV extends under the area of the serous detachment. The term lesion, in contrast, refers to the entire complex of lesion components.

Fig. 66.11 Fluorescein angiogram of serous retinal pigment epithelial detachment. (A) Early transit phase of a fluorescein angiogram demonstrates uniform fluorescence under the dome of the detachment. Note the deformation of the otherwise round detachment by a notch of the hyperfluorescence (arrow). (B) Later phase of the fluorescein angiogram demonstrates persistent hyperfluorescence that does not extend beyond the margins of the hyperfluorescence seen in the early transit phase.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

The terms well-defined (synonymous with well-demarcated) and poorly defined (synonymous with poorly demarcated or ill-defined) refer to a description of the boundaries of the entire lesion (not of individual lesion components). In a well-defined lesion, the entire boundary for 360 degrees is well demarcated (for example, Figs 66.9, 66.12, 66.13). If the entire boundary is not well demarcated for 360 degrees, then the lesion is poorly defined (for example, Fig. 61.9). Thus the terms well-defined and classic should not be used interchangeably, nor should poorly defined and occult. Well-defined and poorly defined describe lesion boundaries (for a lesion that may be composed of classic CNV, or occult CNV, or both). Classic and occult CNV refer to patterns of fluorescence. In addition, the term poorly defined should not be used to describe situations in which blood blocks the ability to see fluorescence from CNV,^19,41,42 even though earlier publications had alluded to descriptions incorporating this possibility.⁴⁴

Fig. 66.12 Classic and occult choroidal neovascularization (CNV) and elevated blocked fluorescence (EBF) with well-demarcated borders. (A) Recurrent subfoveal CNV. Note small area of hemorrhage temporal to recurrence. (B) Early phase of fluorescein angiogram shows sharp demarcation of hyperfluorescence of classic CNV. (C) Mid-phase photograph of angiogram with fluorescein leakage from classic CNV and sharply demarcated hyperfluorescence of elevated retinal pigment epithelium due to fibrovascular pigment epithelial detachment and indicative of occult CNV. Elevated blocked fluorescence still obscures choroidal fluorescence and possibly the inferior boundary of CNV. (D) Late phase of angiogram demonstrates fluorescein leakage from both classic and occult neovascularization. Note hardly discernible EBF. (E) Composite drawing using multiple stereoscopic photographs of angiogram shows interpretation of boundaries of lesion. Since each lesion component (classic CNV, occult CNV, blood, and EBF) has well-demarcated boundaries, boundaries of entire lesion are considered well demarcated.

(Reproduced with permission from Macular Photocoagulation Study Group. Subfoveal neovascular lesions in age-related macular degeneration: guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch Ophthalmol 1991;109:1242–57.)

Fig. 66.13 (A) Subfoveal choroidal neovascularization (CNV) and contiguous blood. (B) Early phase of fluorescein angiogram shows classic CNV with blocked fluorescence corresponding to contiguous blood that obscures boundaries of CNV along temporal border. Remaining blocked fluorescein surrounding CNV (elevated when viewed stereoscopically) was probably due to the fibrous component of CNV. (C) Late phase of fluorescein angiogram demonstrates that borders of CNV, blood, and elevated blocked fluorescence (green, red, and blue, respectively, in (D) were derived from viewing the entire stereoscopic fluorescein angiogram taken according to study protocol. (D) Drawing demonstrates that combined areas of blood and elevated blocked fluorescence that obscured borders of CNV did not exceed area of visible CNV.

(Reproduced with permission from Macular Photocoagulation Study Group. Subfoveal neovascular lesions in age-related macular degeneration: guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch Ophthalmol 1991;109:1242–57.)

The term predominantly CNV indicates that at least 50% of the lesion is composed of either classic CNV or occult CNV, or both, while the term predominantly hemorrhagic indicates that at least 50% of the lesion is composed of hemorrhage.^41,42 These terms are critical in the management of AMD (Fig. 66.14), since treatments for CNV with anti-VEGF therapy, or less frequently, with laser photocoagulation, photodynamic therapy (PDT), or surgery, have been tested only in lesions that are predominantly CNV or predominantly hemorrhagic. After determining whether a lesion’s composition is predominantly CNV, it should be determined whether the lesion is predominantly classic, rather than minimally classic or occult with no classic. If predominantly classic, then treatment could be considered with or without evidence of presumed recent disease progression (defined as evidence of blood associated with CNV, or definite visual acuity loss within 3 months, or definite growth of the lesion within 3 months). If minimally classic or occult with no classic, treatment has been shown to be beneficial compared with no treatment only with evidence of presumed recent disease progression, although a therapeutic trial of anti-VEGF therapy might be considered if visual acuity loss already had occurred and one believed that visual acuity improvement might occur with anti-VEGF therapy because of the presence of intraretinal or subretinal fluid judged to be contributing to visual acuity loss and judged likely to resolve with visual acuity improvement following initiation of anti-VEGF therapy.

Fig. 66.14 Algorithm for verteporfin therapy or laser photocoagulation or observation for symptomatic patients with age-related macular degeneration, pathologic myopia, or other causes of choroidal neovascularization (CNV) in which the natural course is likely worse without treatment. MPS, Macular Photocoagulation Study, and now largely replaced by anti-VEGF therapy as outlined in Fig. 66.21.

Retinal pigment epithelium detachments in age-related macular degeneration

Various changes in an eye with AMD may result in elevation or detachment of the RPE, as seen on stereoscopic biomicroscopic or angiographic evaluation. The term RPE detachment or retinal pigment epithelial detachment (retinal PED) secondary to AMD in the ophthalmic literature remains confusing because various RPE detachments may have quite different compositions, fluorescein angiographic appearances, prognoses, and management. Fortunately, these various RPE detachments can usually be differentiated on the basis of fluorescein angiographic patterns of fluorescence. The patterns include the following: (1) fibrovascular PEDs,¹⁹ which are a subset of occult CNV (see Figs 66.9 and 66.10); (2) serous detachments of the RPE⁴⁵ (see Fig. 66.11); (3) hemorrhagic detachments of the RPE, in which blood from a choroidal neovascular lesion is noted beneath or exterior to the RPE (see Fig. 66.3); and (4) drusenoid RPE detachments,⁹ in which large areas of confluent, soft drusen are noted. It is potentially difficult to differentiate between fibrovascular PEDs and serous PEDs. Using descriptions from the MPS Group, fibrovascular PEDs (as a subset of occult CNV) have been distinguished from a typical serous detachment of the RPE, in that the former do not have uniform, bright hyperfluorescence in the early phase. Instead, they show a stippled fluorescence along the surface of the RPE by the middle phase of the angiogram and may show pooling of dye in the overlying subsensory retinal space in the late phase (see Fig. 66.9). Serous PEDs show uniform, bright hyperfluorescence in the early phase, with a smooth contour to the RPE by the middle phase, and little, if any, leakage at the borders of the PED by the late phase (see Fig. 66.11). The fluorescent pattern of a serous PED obscures the ability to determine whether classic or occult CNV exists within or beneath the area of the serous PED. In contrast, a fibrovascular PED is an area of occult CNV.

A hemorrhagic detachment of the RPE will block choroidal fluorescence because of the mound-like collection of blood beneath the RPE (see Fig. 66.3). Occasionally a hemorrhagic detachment of the RPE may be mistaken for a choroidal melanoma, but usually hemorrhagic detachments of the RPE do not demonstrate low internal reflectivity, as is seen characteristically in choroidal melanomas.

One other feature of AMD that appears as an elevated or detached RPE is a drusenoid RPE detachment,¹⁸ which represents extensive areas of large, confluent drusen. Drusenoid RPE detachments can be distinguished from serous detachments of the RPE in that drusenoid RPE detachments fluoresce faintly during the transit and do not progress to bright hyperfluorescence in the late phase of the angiogram. In contrast, serous detachments of the RPE fluoresce brightly in the early-transit phase and remain brightly hyperfluorescent in the late phase. In addition, serous detachments will usually have a smoother, sharper boundary compared with drusenoid RPE detachments. Drusenoid RPE detachments can be distinguished from fibrovascular PEDs in occult CNV by noting that fibrovascular PEDs show areas of stippled hyperfluorescence with persistence of staining or leakage within a sensory retinal detachment overlying the area in the late phase of the angiogram. RPE detachments due to large, soft, confluent drusen are usually smaller, shallower, and more irregular in outline than are fibrovascular PEDs. In addition, the drusenoid RPE detachments often have reticulated pigment clumping overlying the large, soft, confluent drusen, a scalloped border, and have less fluorescence in late-phase frames as compared with earlier-phase frames.

Other angiographic features

Speckled hyperfluorescence

Speckled fluorescence (Fig. 66.15A) in the absence of fluorescein leakage consists of several punctuate spots of hyperfluorescence, usually within 500 µm of each other that are apparent between 2 and 5 minutes after fluorescein injection and cannot be detected in early phase frames in contrast to drusen.⁴⁶ The fluorescence of these spots persisted or increased in intensity by the late phase of the angiogram (Fig. 66.15B), in contrast to drusen or atrophy of the retinal pigment epithelium which do not remain brightly hyperfluorescent by the late phase of the angiogram. Typically, clusters of speckles would appear at the edge of the CNV lesion, rather than the typical distribution of drusen throughout the macular area. This angiographic feature was previously reported to be associated with recurrent CNV.⁴⁶ If speckled hyperfluorescence is noted in the presence of fluorescein leakage in the late phase frames in the absence of elevation of the RPE, then the pattern would be considered to meet the definition of late leakage of an undetermined source, a type of occult CNV.

Fig. 66.15 (A and B) Example of speckled fluorescence noted in an eye with choroidal neovascularization. Multiple punctuate spots of hyperfluorescence within 500 µm of each other are apparent in a late phase frame (B) but cannot be detected in early phase frame (A). Note the typical appearance of clusters of speckles at the edge of the CNV lesion.

(© Neil M. Bressler, MD, Johns Hopkins University, 2011.)

Fading choroidal neovascularization

CNV may occasionally be recognized in the early- or middle-transit phase of the angiogram with fading in the late phase, so little fluorescein staining or leakage can be discerned in the late phase within an area that was presumed to harbor CNV on the basis of its early-phase features¹⁹ (Fig. 66.16). A vascular pattern of the fibrovascular tissue can occasionally be discerned in early-phase frames.⁴³ Most ophthalmologists are reluctant to treat areas of CNV that fade. These areas are usually not associated with overlying subretinal fluid (in conjunction with the lack of fluorescein leakage on angiography), so one can only presume that this region may proceed to disciform scarring. Perhaps these areas represent CNV histologically, but without evidence of subretinal fluid or late leakage, one cannot be certain that this pattern definitively represents CNV, and laser treatment of this area may damage the retina unnecessarily.

Fig. 66.16 Fading fluorescence of occult choroidal neovascularization (CNV) contiguous with classic CNV. (A) Early phase of angiogram shows classic CNV with contiguous areas of slightly elevated hyperfluorescent retinal pigment epithelium (RPE) with a vascular pattern, presumably a fibrovascular pigment epithelial detachment, and other less well-demarcated areas of hyperfluorescence nasal to the fovea. (B) Later phase of angiogram shows fluorescein leakage from classic CNV. However, areas of elevated hyperfluorescent RPE on early phase of angiogram (left) begin to fade, and vascular pattern is no longer apparent. Faded area is not considered a lesion component because hyperfluorescence does not meet minimal leakage/staining standard for occult CNV.

(Reproduced with permission from Macular Photocoagulation Study Group. Subfoveal neovascular lesions in age-related macular degeneration: guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch Ophthalmol 1991;109:1242–57.)

Choroidal neovascularization

Histopathology

CNV appears as a neovascular sprout growing under or through the RPE through breaks in Bruch’s membrane¹³ (Figs 66.17, 66.18). Usually this occurs in association with evidence of fibroblasts, myofibroblasts, lymphocytes, and macrophages.⁵⁰ Various growth factors are suspected to be involved in the development of this CNV, such as vascular endothelial growth factor (VEGF).⁵¹ However, while drugs designed to interfere with VEGF have been shown to reduce the risk of vision loss,^52–57 vision loss still occurred in many treated cases, and it is not yet clear how much these drugs produce an antiangiogenic effect or an antipermeability effect. Following penetration of the inner aspect of Bruch’s membrane, the new vessels proliferate laterally between the RPE and Bruch’s membrane.¹³ As these neovascular twigs mature, they develop a more organized vascular system stemming from a trunk of feeder vessels off the choroid, as well as proliferation of fibrous tissue. The endothelial cells in the arborizing neovascular tufts lack the barrier function of more mature endothelial cells. Hence these new vessels can leak fluid (and fluorescein) in the neurosensory, subsensory, and RPE layers of the retina. Proteins and lipids may accompany this process and precipitate in any layer of the retina. In addition, the fragile vessels are prone to hemorrhage. Occasionally, blood may extend through all the layers of the retina, breaking through into the vitreous cavity. Ultimately, a fibrovascular scar results, usually causing disruption and death of the overlying sensory retinal tissue accompanied by severe visual loss.

Fig. 66.17 Photomicrograph of choroidal neovascularization (outlined by small arrows) beneath the retinal pigment epithelium growing through a break in Bruch’s membrane (large arrows).

(Reproduced with permission from Elman MJ. Age-related macular degeneration. Int Ophthalmol Clin 1986;26:117–44.)

Fig. 66.18 Sketch of choroidal neovascularization showing ingrowth of vessels from the choriocapillaris, through a break in Bruch’s membrane, into the subretinal pigment epithelial space.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

Choroidal neovascularization

CNV may arise in association with a number of conditions other than AMD, such as OHS, pathologic myopia, angioid streaks (especially when associated with pseudoxanthoma elasticum), choroidal ruptures, and idiopathic causes. Whether AMD is present when CNV arises in patients older than 50 years without drusen is controversial.

CNV may masquerade as central serous chorioretinopathy. Although the classic case of central serous chorioretinopathy shows a “smokestack” configuration on fluorescein angiography, the more common presentation is a dot of hyperfluorescence that merely increases in size and intensity of fluorescence, much like a small area of CNV. One must strongly consider CNV in patients aged 50 and older who have typical central serous chorioretinopathy. In these cases the increased age contrasts to the third to fifth decades, when central serous chorioretinopathy is most commonly seen. Presentation of central serous chorioretinopathy in an elderly patient with drusen can be difficult to distinguish and should alert one to the possibility of CNV, necessitating careful follow-up. If CNV is indeed present, growth of the area of hyperfluorescence might be detected while treatment may still be beneficial.

Basal laminar, or cuticular, drusen may be complicated by foveal detachments of vitelliform material in one or both eyes that may mimic neovascular AMD. Contact lens examination with transillumination of the fundus in these cases reveals an appearance similar to that of pigskin, with a myriad of small drusen. Angiographically, literally hundreds of bright spots appear very early during the angiogram, an appearance that has been described as a “starry sky.” These patients are usually asymptomatic until they accumulate vitelliform lesions in the fovea. The ensuing foveal detachment by this material, which can be unilateral or bilateral, simulates the foveal detachment that occurs with subfoveal CNV. The hyperfluorescence usually progressively fills the area of vitelliform material, rather than showing one area of bright fluorescence early that leaks late in classic CNV. Unlike subfoveal CNV, which most often progresses to a disciform scar with visual acuity of 20/200 or worse, eyes with a foveal detachment secondary to cuticular drusen resolve without scarring. The retina then may or may not have an area of RPE atrophy about one to two disc areas in size. Foveal acuity with atrophy is more often in the range of 20/80 to 20/125, in contrast to the more severe visual loss that accompanies true subfoveal CNV. Patients with cuticular drusen can still develop typical drusen associated with AMD or CNV.

Pattern dystrophies of the RPE may also have vitelliform detachments in which the angiographic pattern can mimic CNV. When one identifies any condition with a vitelliform detachment, it is critical to determine if the late-phase bright fluorescence is progressive fluorescein staining of the vitelliform material or leakage of fluorescein from CNV. The former would not benefit from laser photocoagulation, whereas the latter might. As discussed earlier, usually the fluorescence from a vitelliform detachment shows one or more areas of hyperfluorescence in the early phase of the angiogram with progressive hyperfluorescence of the entire area of yellow vitelliform material by the late phase. In contrast, a diffuse area of hyperfluorescence in the late phase that corresponds to classic CNV (rather than the stippled fluorescence of occult CNV) should show an area of bright hyperfluorescence in the early phase that is only slightly smaller than the area of hyperfluorescent leakage in the late phase.

Vitreous hemorrhage

If a new patient has vitreous hemorrhage in one eye and signs of AMD in the other eye, other causes of vitreous hemorrhage must first be ruled out. Common causes of vitreous hemorrhage include retinal tear formation and retinal vascular diseases such as diabetic retinopathy or branch vein occlusion. Ultrasonography usually differentiates breakthrough vitreous hemorrhage secondary to neovascular AMD or retinal vascular causes from that caused by a tumor (Fig. 66.19).

Fig. 66.19 Ultrasound image of a vitreous hemorrhage associated with neovascular age-related macular degeneration in which a relatively flat and broad-based posterior pole lesion (arrow) with a fairly homogeneous pattern and with no choroidal excavation is seen.

(Reproduced with permission from Bressler NM, Bressler SB, Fine SL. Age-related macular degeneration. Surv Ophthalmol 1988;32:375–413.)

Well-defined extrafoveal and juxtafoveal choroidal neovascularization

Patients assigned to observation in the MPS trials provide natural-history information on cases that met criteria for these studies. When lesions have well-defined boundaries, the CNV lesion can be classified according to the location of the most posterior boundary of the lesion with respect to the center of the FAZ on the fluorescein angiogram. CNV lesions located more than 200 µm from the FAZ center are termed extrafoveal; those between 1 and 199 µm from the center are juxtafoveal; CNV lesions extending under the center of the FAZ are termed subfoveal. In contrast to other pathologic conditions predisposing to CNV (e.g., OHS and pathologic myopia) in eyes with AMD, CNV presents more frequently under the FAZ center. In addition, subfoveal CNV from AMD tends to be larger on initial presentation than when seen in other conditions.⁷² In a retrospective review of fluorescein angiograms of AMD ordered at a community hospital over a 9.5-year period, only 8% (19/244) of the CNV lesions were extrafoveal.⁷³ Other retrospective reports not derived solely from a retinal referral practice have suggested that most patients with CNV from AMD present to an ophthalmologist with subfoveal CNV with poorly demarcated boundaries.^72,74 However, there is no information from population-based studies to describe precisely how many choroidal neovascular lesions are subfoveal versus not subfoveal, and for those that are not subfoveal, how many are well demarcated on laser photocoagulation. Furthermore, there are no studies to determine how many cases that are subfoveal are predominantly classic, and if not predominantly classic, how many have presumed recent disease progression after developing.

In the MPS, 62% of untreated eyes with well-defined, extrafoveal CNV (posterior boundary of lesion located at least 200 µm from the FAZ center) lost 6 or more lines of visual acuity by 3 years of follow-up.⁷⁵ By 5 years, this percentage was similar at 64%.⁷⁶ Furthermore, in nearly 75% of the untreated eyes, the CNV extended under the FAZ center.⁷⁷ For juxtafoveal lesions, 49% of untreated eyes had lost 6 or more lines of visual acuity by 3 years of follow-up.²⁶

Subfoveal choroidal neovascularization

For subfoveal lesions, the MPS reported that 48% of untreated eyes with well-defined subfoveal CNV that was predominantly CNV and included a component of classic CNV in which the entire lesion was no greater than 3.5 MPS disc areas lost 6 or more lines of vision from baseline by 48 months.^24,78 Similar outcomes were noted for untreated eyes with recurrent subfoveal CNV with similar features as for new subfoveal lesions except the total size of the lesion plus any area of prior treatment was to be no greater than 6 MPS disc areas.^25,78 A broader group of subfoveal lesions in the Treatment of Age-related macular degeneration with Photodynamic therapy (TAP) investigation described the natural history in subfoveal lesions that were predominantly CNV with evidence of classic CNV and a lesion size no greater than 9 disc areas in which 62% lost 3 or more lines at 2 years after entry, including 30% losing 6 or more lines of visual acuity.⁷⁹ With increasing size of predominantly classic lesions on presentation, the average visual acuity is more likely to be worse by 2 years.⁸⁰ Similar natural history outcomes were noted for untreated eyes with occult with no classic CNV with presumed recent disease progression,⁵⁴ in which almost half of the eyes lost 3 or more lines at 2 years after entry, including 23% losing 6 or more lines of visual acuity.

A variety of studies have shown that cases of predominantly CNV with a composition that is occult with no classic CNV, or minimally classic CNV have a more heterogeneous outcome.^44,80–85 Most of this information is from series in clinical trials with presumed recent disease progression. Some cases may remain stable for years without visual loss, whereas other cases may develop severe visual loss at a rate similar to the deterioration noted for cases of classic CNV only. Furthermore, increasing size of minimally classic or occult with no classic lesions is not associated with a worse natural history outcome.⁸⁰

Up to 50% of the cases with no classic CNV may develop classic CNV within a year of presentation.^44,81,82,86 Cases that develop some classic CNV may be more likely to have severe visual acuity loss.^81,82,86 Results from several clinical trials suggest that the natural history of lesions with classic CNV but no occult CNV is worse than the natural history of lesions with classic and occult CNV or occult with no classic CNV.^82,84,87 It is likely that the natural history of lesions with classic and occult CNV may lie somewhere between the natural course of lesions with classic CNV only and occult CNV only.⁸²

Natural course of large subfoveal subretinal hemorrhage in age-related macular degeneration

Some eyes with subfoveal subretinal hemorrhage associated with AMD have poor outcomes.^88–90 However, the visual acuity of other eyes does not deteriorate or may improve spontaneously.^88,90 Such findings underscore the importance of evaluating the role of therapeutic interventions for these cases, such as surgery^91,92 to remove subretinal hemorrhage and associated CNV, in randomized clinical trials.⁹³ The SST Group B Trial of relatively large, predominantly hemorrhagic subfoveal lesions demonstrated that 41% of untreated eyes remained stable or improved, although 36% had severe visual acuity loss after enrollment into the trial. Furthermore, this trial suggested that 18% of large subfoveal subretinal hemorrhages will progress to a vitreous hemorrhage as blood dissects from the subretinal space through the retina and into the vitreous.²⁷

Retinal pigment epithelial tears

Patients with RPE tears involving the foveal center may initially maintain good vision but usually develop severe visual loss. However, cases with RPE tears through the fovea and preservation of good visual acuity have been reported.⁹⁴ Unfortunately, there is a substantial risk of AMD-related visual loss in the fellow eye. Schoeppner and associates⁹⁵ reported a cumulative risk of visual loss in the fellow eye of patients who had an RPE tear in the eye as 37% at 1 year, 59% at 2 years, and 80% at 3 years of follow-up. Visual loss usually arose from development of a PED, RPE tear, or CNV.

Laser treatment of well-defined choroidal neovascular lesions

While laser photocoagulation has been shown to be beneficial for well-defined lesions not involving the foveal center or very small lesions involving the foveal center with a component of classic CNV, failure to cover the entire lesion increases the likelihood of recurrent CNV^96–98 and, for extrafoveal and juxtafoveal lesions, of additional visual acuity loss,^96,98,99 and treatment will destroy retinal tissue (and corresponding function).⁹⁸ Therefore, such treatment generally is reserved for cases that are extrafoveal and in which it is judged that a scotoma from the laser is preferred over proceeding with anti-VEGF therapy.

Preparation for laser photocoagulation treatment

Before photocoagulation treatment, patients with a potentially treatable CNV lesion should be informed that laser treatment will create a permanent blank area, or blind spot, corresponding to the area of retina to be treated. The postoperative vision will correlate with the extent of central retina destroyed with laser photocoagulation treatment. Whereas patients with binocular macular vision and a treated lesion away from the center of the FAZ frequently remain unaware of this treatment scotoma unless they cover the untreated fellow eye, patients with a central scotoma in the fellow eye usually are aware immediately of the laser-induced scotoma in the treated eye. Because laser treatment may cause hemorrhage or increase sensory retinal detachment during the immediate postoperative period, all patients should be warned that they may experience increased distortion and decreased vision after treatment and that this may persist for several weeks. In addition, the ophthalmologist must emphasize that laser treatment is not a cure for AMD; rather, laser photocoagulation treatment is designed to obliterate the neovascular complex and thereby reduce the risk of additional severe visual loss in the future. Furthermore, the treating ophthalmologist must recall that, even under the best conditions in the MPS, many of the treated eyes had recurrent CNV in which anti-VEGF therapy likely would be indicated.

Following the treatment protocol established by the MPS (discussed below), a recent fluorescein angiogram (if possible, less than 72–96 hours old) should be used to guide the ophthalmologist during treatment. Because the CNV may grow, a fluorescein angiogram older than a few days may no longer depict the extent of CNV accurately. A suitable frame from the angiogram is displayed near the laser console top. The angiogram thus displayed permits rapid and accurate orientation of critical retinal vascular and CNV landmarks. With a cooperative patient, an ophthalmologist with adequate experience in the treatment of CNV should be able to treat using topical anesthesia.

Macular Photocoagulation Study photocoagulation techniques

Using a 200 µm spot size and duration of at least 0.2 s, a test burn is placed along the lesion’s perimeter to determine the necessary power setting. Then, the lesion is outlined with laser burns. Ideally, one should start at the inferior edge of a lesion and work superiorly, as if case intraoperative bleeding occurs and tracks inferiorly, obscuring retinal landmarks, this area will have been treated. To obliterate the CNV, the burns should be sufficiently intense to produce retinal whitening. Finally, the center of the CNV is obliterated (Fig. 66.20).

Fig. 66.20 Juxtafoveal choroidal neovascularization lesion. (A) Fluorescein angiogram shows choroidal neovascularization (CNV) in which the posterior edge of the CNV is more than 200 µm from the foveal center and is associated with blocked fluorescence on the posterior edge of the lesion within 200 µm of the foveal center. (B) Red-free photograph of same eye demonstrating that, except for a small amount of blood superonasal to the neovascular lesion, the blocked fluorescence is not due to subretinal blood visible on photograph. (C) Drawing of CNV or neovascular membrane (NVM) in which blocked fluorescence is represented by hatched area surrounding the neovascular lesion.

(Reproduced with permission from Macular Photocoagulation Study Group. Subfoveal neovascular lesions in age-related macular degeneration: guidelines for evaluation and treatment in the Macular Photocoagulation Study. Arch Ophthalmol 1991;109:1242–57.)

To ensure adequate treatment, the MPS protocol required that the entire CNV lesion be covered by a uniformly intense area of whitening extending 100 µm beyond the angiographically visible lesion. Excessive treatment can contribute to visual loss. However, failure to cover the CNV in its entirety could lead to increased persistence within 6 weeks of treatment.^96–98

Rarely, choroidal hemorrhage occurs as a treatment complication. Generally, choroidal hemorrhage stops spontaneously. If not, it can usually be controlled by applying pressure to the globe with the contact lens while treating the area of hemorrhage with repetitive laser spots, employing a larger spot size and longer duration than might have been used for CNV treatment. Long burns (0.5–1.0 s) and large spot sizes (200 µm or larger) decrease the chances of additional choroidal hemorrhage.

No visible wavelength appears to have a significant advantage over other wavelengths. Small differences in convenience of achieving the end-point of a uniform white burn might be seen with red or yellow wavelengths when penetrating through the increased yellow color of the lens nucleus in the older age group afflicted with CNV secondary to AMD, but any significant difference of clinical importance has not been shown. When treating CNV that lies under a major retinal vessel, the laser burns should straddle the retinal vessel to reduce the possibility of causing hemorrhage or damaging the vessel by thermal vasculitis. No evidence has suggested that this technique compromised the effectiveness of treatment.

When treating CNV that is contiguous with the optic nerve, it should be kept in mind that laser treatment directly over the optic nerve can cause thermal necrosis of disc tissue and nerve fiber bundle defects. Therefore one should consider refraining from treatment within 100–200 µm of the optic nerve. Similarly, when treating a peripapillary area of CNV, one may want to consider treatment only when at least clock-hours of papillomacular bundle on the temporal side of the disc is uninvolved with CNV so that at least clock-hours of papillomacular bundle can be spared of treatment, as was done in several of the MPS trials. Treatment to nasal or peripapillary lesions that met criteria for MPS trials will likely not lead to severe visual loss from damage to the nerve fiber layer that serves the central macula if the treatment guidelines outlined previously are followed.¹⁰⁰ In these situations, the MPS group reported that severe visual acuity loss was noted after treatment only when recurrent CNV extended through the center of the fovea.¹⁰⁰ This finding suggests that, in the absence of subfoveal recurrence, severe visual loss only from nerve fiber layer damage after this treatment approach must be a rare complication.

On rare occasions, an area of extrafoveal CNV is contiguous to a serous detachment of the RPE in which the serous detachment extends under the center of the FAZ. There have been case reports in which only the extrafoveal CNV in these lesions is treated, resulting in prompt flattening of the RPE detachment with improvement of vision in selected cases.¹⁰¹ Nevertheless, with follow-up, many of these eyes have acquired recurrent CNV with extensive scarring and visual loss. In these situations, one should consider the possibility that the extrafoveal CNV is associated with a subfoveal fibrovascular PED – a pattern of occult CNV in which treatment of the extrafoveal CNV alone has not been shown to be of any benefit. In such cases, anti-VEGF therapy typically would be indicated.

Evaluations following laser photocoagulation

A follow-up evaluation that includes best-corrected vision, biomicroscopy of the fundus, and fluorescein angiography is obtained 2–3 weeks after treatment. Follow-up earlier than 2 weeks is often difficult to evaluate because swelling and leakage from the treatment itself may obscure persistent or recurrent CNV. At follow-up, an angiogram should be scrutinized for the presence of leakage at the periphery of the laser-treated area to identify the presence of persistent or recurrent CNV, while an OCT can be scrutinized for intraretinal or subretinal fluid. Simultaneous projection of the fluorescein angiogram during biomicroscopy may help differentiate areas of atrophy, which stain, from areas of recurrent leakage with subretinal fluid. If no residual or recurrent CNV is noted at this time, a similar evaluation is repeated 2 to 4 weeks later because the risk of recurrent CNV is so high within the first 6 weeks to 12 months after treatment. By 6 weeks after treatment, a patient should be encouraged to monitor the central vision of the treated eye daily for clarity of distance and near vision, as well as for any distortion, blurry vision, or increase in scotoma. These latter symptoms might indicate leakage from persistent or recurrent CNV and are an indication for prompt examination. The ophthalmologist’s office staff should also be aware that these patients may need prompt re-evaluation if such symptoms develop and not necessarily schedule such patients for the “next available opening” 2 or 3 weeks later. Furthermore, although the risk of recurrent CNV after treatment in eyes with AMD is high, results showing the benefits of treating selected cases of recurrent subfoveal CNV make follow-up warranted for identifying these recurrences before they become too large.

Clinical examination probably cannot replace fluorescein angiography in detecting all recurrent CNV after laser treatment or PDT (as described below) or anti-VEGF therapy (as described below). In one prospective evaluation in which recurrent CNV was not suspected on biomicroscopy within approximately 1 year after laser photocoagulation to CNV, definite or questionable recurrent CNV was identified on the fluorescein angiogram 12% of the time.⁴⁰ Most of these recurrent cases that were not identified on biomicroscopy alone were treated promptly after review of the fluorescein angiogram. This study also showed that questionable recurrences that showed focal staining along the edge of the laser lesion and speckled hyperfluorescence were the patterns that were most likely to progress to definite recurrence.¹⁴

Since many recurrences occur within 3–6 months after treatment, an evaluation, including fluorescein angiography, is again repeated at 3–4 and 6–8 months after treatment. Subsequent evaluations, probably with angiography, at 9–12 months after treatment appear to be indicated for CNV secondary to AMD that has been treated because many recurrences develop between 6 and 12 months after treatment. After 2 years, recurrences are unusual; follow-up every 6–12 months without angiography (unless signs or symptoms suggest a recurrence) is probably sufficient.

Complications of laser photocoagulation

Immediate complications from treatment of CNV are uncommon. Rates of choroidal hemorrhage, which can occur when attempting to achieve a white burn, may be minimized by avoiding a spot size less than 200 µm and an exposure time shorter than 0.2 s. Macular pucker, observed with some frequency after argon blue–green laser treatment, is rarely clinically significant and is far less common when employing argon green or krypton red wavelengths.⁷³ Rates of inadvertent treatment of the foveola in extrafoveal or juxtafoveal lesions, the most serious treatment complication, are rare now that anti-VEGF therapy or PDT is used for these lesions.

Delayed perfusion of choroidal vessels, thought to result from vascular spasm, has been reported following the use of the krypton red laser.¹⁰² In most cases, normal choroidal perfusion returns rapidly without any permanent visual sequelae. The krypton red laser has also been implicated in RPE tears arising from treatment of CNV.³³ However, RPE tears may be observed spontaneously and after treatment of CNV using any wavelength, particularly in patients with associated RPE detachments, either serous or fibrovascular.

Visual acuity might decline several years after treatment with late remodeling and enlargement of the area of RPE disturbance. Such occurrences may be less frequent now that anti-VEGF therapy, or less commonly, PDT, is used to treat most CNV, while laser is reserved for lesions that are relatively far away from the center of the macula. These areas of enlargement of RPE abnormalities following thermal laser have been reported by Morgan and Schatz¹⁰³ to increase in size by 50–100 µm over serial follow-up. In their series, 4 (3%) of 174 eyes lost vision when the area of disturbance extended into the fovea. The peripheral zone of atrophy in a laser-treated area may correspond with the runoff or spread of the laser burn during the treatment. “Runoff” refers to the border of less-intense whitening surrounding the area of heavy treatment; it probably represents less severe damage to the RPE, retina, and choroid.

Results of photodynamic therapy treatment

Follow-up after photodynamic therapy

Any decreased vision within 5 weeks after treatment may need prompt evaluation. If a change in the biomicroscopic examination is noted compared with the pretreatment appearance, OCT and fluorescein angiography may be indicated to try to understand why decreased vision has occurred, for example, due to growth of the lesion, additional fluid accumulation, bleeding, choroidal perfusion abnormalities, tears of the RPE, or other causes. Additional treatment for growth of the lesion earlier than 5–6 weeks has not been studied extensively, so that if deterioration occurs, switching to anti-VEGF therapy likely is warranted. Even if no change in vision is noted by 12 weeks after treatment, re-examination should occur at that time. If no fluorescein leakage from CNV is noted at that time, then no additional treatment is needed, although re-examination in another 6–12 weeks is warranted to see if leakage develops subsequently. If fluorescein leakage is noted and visual acuity has not improved, then switching to anti-VEGF therapy likely is warranted. If improvement in visual acuity has occurred but fluorescein leakage persists, then either switching to anti-VEGF therapy or continuing PDT treatment might be considered. If retreatment with PDT is considered, it should be applied using the GLD on the area of leakage and any contiguous blood plus 1 mm when determining the spot size. This follow-up approach should be repeated indefinitely until the situation has stabilized (because there is minimal leakage associated with a relatively flat scar and no change in visual acuity, biomicroscopic appearance, or fluorescein angiographic appearance for at least 6 months). Usually, if therapy is not switched to anti-VEGF treatments, patients will need two to three PDT treatments in the first year, and possibly one to two treatments in the second year.

Efficacy of ranibizumab vs PDT with verteporfin for predominantly classic subfoveal CNV lesions

Trial participants in the ANCHOR study had CNV lesions that were to be predominantly classic CNV (at least 50% of the total area of the lesion was classic CNV).⁵⁵ Unlike eligibility requirements for the MARINA study (see following section), participants were not required to have presumed recent disease progression (usually defined as presence of blood, documented recent loss of visual acuity, or documented recent growth of the choroidal neovascular lesion). Participants received ranibizumab monthly for 24 months.

Mean visual acuity letter score at baseline was 47.1 (approximate Snellen equivalent of 20/125⁺²) in the 0.5 mg ranibizumab group. For the primary outcome (visual acuity decline of fewer than 15 letters from baseline at 12 months), 96.4% of the 0.5 mg group (n = 140) and 64.3% of the verteporfin group (n = 143) avoided this loss (P <0.001).⁵⁵ These outcomes were maintained at 24 months for most subjects, with 89.9% of the 0.5 mg group and 65.7% of the verteporfin group avoiding a loss of 3 lines or greater. Investigators also reported that visual acuity improved by 15 letters or more in 40.3% of the 0.5 mg group, as compared with 5.6% of the verteporfin group (P <0.001) at 12 months⁵⁵ and 41.0% of the 0.5 mg group gained 3 or more lines of visual acuity versus only 6.3% of the verteporfin group at 24 months. Differences in the mean visual acuity from baseline visual acuity were noted as early as 1 month, when 8.4 letters of improvement occurred in the 0.5 mg ranibizumab group compared with 0.5 letters in the PDT group (P <0.001).⁵⁵ At 12 months, mean visual acuity improved by 11.3 letters in the 0.5 mg ranibizumab group but decreased by a mean of 9.5 letters in the verteporfin group (P <0.001).⁵⁵ At 24 months, mean visual acuity improved by 10.7 letters in the 0.5 mg ranibizumab group but decreased by 9.8 letters in the verteporfin group.

Although PDT produced outcomes that likely were better than the natural course of the disease, the 24-month data reinforced the conclusions of the 12-month data previously published and indicated that the increased chance of avoiding 15 or more letter loss and of gaining 15 or more letters with monthly intravitreal ranibizumab persisted through at least 2 years compared with standard applications of PDT.

Efficacy of ranibizumab vs sham treatment for minimally classic or occult with no classic subfoveal choroidal neovascular lesions and presumed recent disease progression

In a phase 3, double-blind, sham-controlled, randomized clinical trial, eligibility criteria included subfoveal CNV with lesion composition on FA that was minimally classic or occult with no classic component accompanied by presumed recent disease progression.⁵⁴ Mean visual acuity letter score at baseline was 53.7 (20/80⁻¹) in the 0.5 mg ranibizumab group.⁵⁴ Using the primary outcome of visual acuity decline of fewer than 15 letters from baseline as a measure of efficacy, 94.6% of those in the 0.5 mg ranibizumab group and 62.2% of those in the sham-injection group met the primary goal (P <0.001) after 12 months of treatment.⁵⁴ At 24 months, 90.0% of those in the 0.5 mg ranibizumab group and 52.9% of those in the sham-injection group met the primary goal (P <0.001).⁵⁴

At 12 months, visual acuity improved by 15 or more letters in 33.8% of the 0.5 mg ranibizumab group, as compared with 5% of the sham-injection group (P <0.001).⁵⁴ Participants in the 0.5 mg ranibizumab group achieved mean increases in visual acuity of 7.2 letters, whereas the sham-injection group lost an average of 10.4 letters (P <0.001).⁵⁴ At 24 months, participants in the 0.5 mg ranibizumab group gained a mean of 6.6 letters of visual acuity, compared with a mean loss of 14.9 letters in the sham-injection group (P <0.001).⁵⁴

In addition to experiencing improvements in objectively measurable visual outcomes, at 12 and 24 months, patients treated with ranibizumab were more likely to improve with respect to self-reported vision-related quality of life (QOL)^114–116 regardless of whether the better- or worse-seeing eye was treated. This information is complementary to the visual acuity outcomes described above because although most physicians and patients make decisions based on functional parameters such as a patient’s ability to read, watch television, and avoid dependence on others because of vision, the primary outcome measure in most recent trials evaluating treatments for neovascular AMD has been visual acuity in the treated eye.^54,55Although changes in visual acuity on an eye chart are important in understanding treatment effects, measures of visual acuity do not always completely capture the degree of visual function. Clinicians may assume that visual function or patient’s perception of visual function improves along with visual acuity; however, this is not always the case.¹¹⁷ Among MARINA participants, scores on the National Eye Institute Visual Function Questionnaire-25 (NEI VFQ-25) relating to activities that require both near vision (such as sewing) and distance vision (such as watching television) were more likely to improve by 10 or more points, a clinically relevant amount, in the ranibizumab group than in the sham-injection group (P <0.001).¹¹⁸ In addition, ranibizumab-treated patients were less likely to perceive themselves as being dependent on others because of their vision.¹¹⁸ Similar results were confirmed in the ANCHOR trial.

Safety of ranibizumab

Safety issues with ranibizumab intravitreal injections include local ocular adverse events (AEs) from the drug or the injection, as well as potential systemic AEs of the drug. Ocular AEs may be categorized as common but not serious and rare but potentially serious. AEs that are common but not serious include subconjunctival hemorrhage, vitreous floaters from medication or vitreous hemorrhage, and discomfort from antiseptic used to prepare the cornea before the injection. Endophthalmitis is a rare but potentially serious ocular AE that may develop after any intravitreal injection, regardless of what is being injected. Just how much preparation is necessary to minimize the development of postinjection infections is controversial. Based on data from the Diabetic Retinopathy Clinical Research Network, there is little evidence to support the need for pretreatment or post-treatment topical antibiotics.¹¹⁹ Some clinicians use sterile gloves and drapes. All clinicians seem to agree that the use of a lid speculum and povidone-iodine treatment of lids, lashes, and the area to receive the injection are generally recommended to minimize postinjection colonization with normal flora. Other rare but serious ocular AEs include inflammation (as opposed to infection), vitreous hemorrhage, retinal tear or detachment, and increased intraocular pressure, which in the trials was listed as a serious AE (SAE) but might not always be considered an SAE by a treating ophthalmologist.

Systemic AEs can be a concern; VEGF inhibitors that cross into the general circulation can compromise functions that rely on VEGF outside of the eye, such as wound healing and the formation of new blood vessels around the heart or brain in cases of ischemia. Patients with AMD already are at higher risk of cardiovascular disease than the general population by virtue of their age, and patients with such risk did not have to be excluded from participating in ANCHOR or MARINA. Consequently, participants in clinical trials of VEGF inhibitors were carefully monitored for possible increases in blood pressure, occurrence of myocardial infarction/stroke, and nonocular hemorrhages.

There was no evidence that ranibizumab 0.5 mg was associated with increases in either diastolic or systolic blood pressure. In fact, treatment-related hypertension emerged in a larger proportion of PDT patients (12 of 143, or 8.4%) than among ranibizumab 0.5 mg patients (9 of 140, or 6.4%).⁵⁵ Among participants in the MARINA trial, approximately 16% in both the ranibizumab 0.5 mg and sham-injection groups developed hypertension.⁵⁴

Nonocular hemorrhages include events such as cerebral or gastrointestinal bleeding. In the ANCHOR trial, nonocular hemorrhage was more frequent in the 0.5 mg ranibizumab group (6.4%) than in the PDT group (2.1%). In the MARINA trial, the cumulative frequency of nonocular hemorrhage by month 24 was 5.5% (13 of 236) in the sham-injection group, compared with 8.8% (21 of 239) in the 0.5 mg ranibizumab group.

Since ranibizumab is a recombinant monoclonal antibody (mAb) that contains both mouse- and human-derived segments, the human segments are engineered into the monoclonal antibody to minimize the chance that the patient’s immune system will react against the mAb. Nevertheless, some patients treated with ranibizumab may develop antibodies to it. Thus, ranibizumab trials included routine testing of participants for antiranibizumab antibodies using an electrochemiluminescent assay.

In the ANCHOR trial, 8% of ranibizumab 0.5 mg subjects and 1.5% of subjects with PDT had low levels of antibodies to ranibizumab at baseline. At trial conclusion, 3.9% of ranibizumab 0.5 mg subjects had developed antibodies to ranibizumab, compared with 0% in the PDT group. In the MARINA trial, none of the subjects in the 0.5 mg ranibizumab group and 0.5% of those in the sham-injection group had antibodies to ranibizumab at baseline. After 24 months, 6.3% of subjects treated with ranibizumab 0.5 mg and 1.1% of those in the sham-injection group developed antibodies to ranibizumab.

The clinical implications of baseline and postexposure immunoreactivity to ranibizumab are unclear. Although the numbers were small, subjects in the ANCHOR trial who were immunoreactive were more likely to develop inflammation than those who were not.

With respect to cardiovascular or cerebrovascular events, during the ANCHOR trial, 1 subject in the PDT group (0.7%) and 3 subjects in the ranibizumab 0.5 mg group (2.1%) developed nonfatal myocardial infarctions, although the events occurred at times that were unrelated to treatment. The frequency of stroke (1 in each group) and cerebral infarction (0 in each group) in the ANCHOR trial were too low to draw meaningful conclusions.

At 24 months, the overall frequency of cardiovascular systemic events in the MARINA trial was similar in the 0.5 mg ranibizumab and sham-injection groups. There were only small differences in the frequency of thromboembolic events between the sham-injection group (3.8%) and the ranibizumab 0.5 mg group (4.6%). The frequency of death (2.5%) was the same in the ranibizumab 0.5 mg and sham-injection groups. Two individuals in each group died of stroke.

Although event rates for these cerebrovascular or cardiovascular events appear to be low with ranibizumab, ophthalmologists should ensure that patients understand the theoretic potential for these risks. Additional studies over time may help to refine understanding of the magnitude, if any, of this risk.

Impact of noninferiority results on frequency of treatment and the role of aflibercept or bevacizumab in place of ranibizumab

The Comparison of Age-related Macular Degeneration Treatments Trial (CATT) protocol was designed to answer two important questions relevant to treatment of CNV in AMD: (1) does bevacizumab (Avastin) injected every 4 weeks provide visual acuity results which are equivalent (strictly speaking, which are noninferior) to ranibizumab (Lucentis) injected every 4 weeks with an acceptable safety profile; and (2) does either bevacizumab or ranibizumab when provided as needed result in visual acuity outcomes equivalent to those following ranibizumab provided every 4 weeks. Here, “as needed” means that an initial injection is given with subsequent examinations of the eye every 4 weeks; reinjection is given at the time of any examination at which any disease activity is noted. Activity includes subretinal or intraretinal fluid or thickening on OCT, new or persistent hemorrhage from CNV, persistent fluorescein leakage from CNV or growth of CNV on fluorescein angiography, or visual acuity loss since the most recent injection.

The major outcomes during the first year which are relevant to these important questions included the following: (1) visual acuity outcomes when using bevacizumab every 4 weeks were equivalent (not inferior) to those when using ranibizumab every 4 weeks; (2) visual acuity outcomes when using ranibizumab as needed, based on examinations every 4 weeks, were equivalent to those when ranibizumab was given every 4 weeks; (3) visual acuity outcomes when bevacizumab as needed was compared with ranibizumab every 4 weeks were inconclusive, i.e., possibly equivalent but also possibly inferior; (4) CATT participants assigned to ranibizumab as needed received an average of approximately seven treatments in the first year and those assigned to bevacizumab as needed received an average of approximately eight treatments; (5) although no difference between ranibizumab and bevacizumab participants was identified with respect to proportions of participants who had myocardial infarction, cerebrovascular accidents, or endophthalmitis, the combined bevacizumab groups had a higher rate of systemic serious AEs than the combined ranibizumab groups (24% for bevacizumab, 19% for ranibizumab) – of note, the point estimates for these systemic serious adverse events was higher in the bevacizumab as-needed group than the bevacizumab every-4-weeks group; and (6) the average cost of drug per patient for the first year was $23 400 in the ranibizumab every-4-weeks group, $13 800 in the ranibizumab as-needed group, and $595 in the bevacizumab every-4-weeks group. Two-year data confirmed that every-4-week bevacizumab was equivalent to ranibizumab, the increased risk of serious systemic adverse events with bevacizumab was sustained and still requires further study, and the average visual acuity in both the bevacizumab-as-needed and ranibizumab-as-needed groups continued to decline from year 1 to year 2.¹²⁰

The 1-year results from another trial comparing bevacizumab and ranibizumab, and continuous and discontinuous treatment¹²¹ confirmed that the bevacizumab group appeared to have a slightly smaller mean gain in visual acuity at 1 year compared with the ranibizumab group (when combining both continuous and discontinuous treatment groups), and the discontinuous group (which required three consecutive monthly injections whenever treatment was resumed) appeared to have only a slightly smaller gain in mean visual acuity when compared with the continuous group (combining the ranibizumab and bevacizumab groups). Two-year data are anticipated in 2013.

Yet another set of trials confirmed that aflibercept (Eyelea) given every 4 weeks for three doses followed by every 8 weeks through 48 weeks gave equivalent outcomes to ranibizuamb-every-4 weeks.⁵⁷ In the second year of these trials, when either aflibercept or ranibizumab was given at least every 12 weeks as well as when retreatment was judged to be indicated based on clinical parameters, including OCT, evaluated every 4 weeks, showed a small decline in the average visual acuity for each of the groups.

Potential implications of anti-VEGF noninferiority trials results on clinical practice

Based on the information summarized above and published from other randomized controlled trials, the following implications should be considered.

If treatment with bevacizumab is to be initiated in an eye with CNV due to AMD, the patient should be informed that when using bevacizumab, treatment every 4 weeks for at least the next year, without dependence on OCT or other imaging, has been shown by CATT to be equivalent to the current “gold standard” of ranibizumab every 4 weeks. If treatment with bevacizumab every 4 weeks is to be given, the patient also should be informed that the 2-year CATT results show an increase in serious systemic AEs with intravitreal bevacizumab for which the causes are not yet understood fully. The ophthalmologist should keep in mind that comparison of visual acuity outcomes when bevacizumab or ranibizumab from treatment initiation, or aflibercept starting 1 year after treatment initiation, was given as needed, with examinations every 4 weeks, visual acuity outcomes appeared, on average, to be slightly worse. While a slightly worse outcome may not be clinically relevant one year after treatment initiation, it becomes relevant if the decline in the average visual acuity continues, as it appears to do at least from 1 to 2 years after treatment initiation. Very importantly, the ophthalmologist and patient must keep in mind that from the societal perspective, use of bevacizumab leads to very large cost savings in the management of this condition.

Ranibizumab as needed with examinations every 4 weeks, including OCT, to detect features that suggest the need for retreatment, provides equivalent visual acuity outcomes to the current gold standard of ranibizumab every 4 weeks at 1 year but not necessarily by 2 years, or thereafter. Aflibercept every 4 weeks for three doses followed by every 8 weeks to 1 year also provides equivalent results to ranibizumab-every-4 weeks; when the group given this regimen received an as-needed regimen starting at 1 year, the average visual acuity declined by 2 years, although not by a clinically relevant amount. However, if this decline were to continue during treatment 3 or 5 years after initiating therapy, the decline could be clinically relevant.

The results should be interpreted with the finding that as-needed ranibizumab or aflibercept costs about $13 000 more at least for the first year than every-4-week bevacizumab, and every-4-week ranibizumab costs about $10 000 more than ranibizumab as needed. Whether the greater costs of ranibizumab or aflibercept are worth the possibility that a small percentage of patients might have substantially better vision or that a larger number of patients might have slightly better vision or a slightly higher risk of systemic AEs, or clusters of endophthalmitis when not compounded properly, are beyond the scope of this chapter. However, these trials provide important data for both private (commercial) and government-sponsored insurances to make decisions on utilization of funds from the perspective of the public health and societal costs for patients who cannot pay for the costs of these very beneficial treatments which, when indicated and available, could reduce at least 75%^2,122 of the prevalence of blindness from what was the leading cause of blindness just a decade ago.

Follow-up after deciding to initiate anti-VEGF therapy for neovascular AMD

Figure 66.21 provides one approach¹²³ to consider if monthly anti-VEGF therapy with bevacizumab or ranibizumab, or 3 monthly doses of aflibercept followed by every-2-months aflibercept, is not employed. The approach considers repeating therapy as long as improvement is noted on subjective symptoms, visual acuity, biomicroscopic examination, OCT, or fluorescein angiography. The approach errs on the side of treatment since failure to treat when VEGF is present could result in irreversible visual acuity loss from fibrosis involving the retina or RPE.

Submacular surgery

Studies have shown that visual acuity outcomes with submacular surgery are no different compared with observation for subfoveal CNV in patients with AMD in which a majority of the lesion is CNV and there is evidence of classic CNV.²⁷ However, for lesions that are predominantly hemorrhagic (at least 50% of the lesion is blood), submacular surgery can reduce the risk of additional severe visual acuity loss compared with observation, although many patients undergoing surgery lost at least two lines of visual acuity.²⁰ Specifically, 21% of patients undergoing removal of hemorrhage lost 6 or more lines of visual acuity compared with 36% assigned to observation. Almost half of the patients who were phakic at the time of undergoing submacular surgery had cataract surgery by 2 years. Furthermore, there was a relatively high risk of rhegmatogenous retinal detachment, especially when the lesion was greater than 16 disc areas or the visual acuity was very poor after enrollment. For this reason, submacular surgery for predominantly hemorrhagic lesions could be considered for lesions that were no greater than 16 disc areas if identified before visual acuity was very poor. More details on the findings of the trials investigating surgery for CNV in patients with AMD are included in the surgery volume of this book.

Although there is no strong evidence to support benefits of submacular surgery,^{27,93,104,133} except for reducing the risk of additional severe visual acuity loss in selected cases of predominantly hemorrhagic lesions, alternative surgical approaches under investigation for managing CNV and its neovascular complications include macular translocation and mechanical displacement of relatively large submacular hemorrhages using intraocular gas injection. Macular translocation can be accomplished by vitrectomy combined with retinotomy or by vitrectomy combined with external sclerochoroidal foreshortening.¹³⁴ After translocation has been accomplished and the CNV lesion is no longer subfoveal, laser photocoagulation is applied with sparing of the foveal center from photocoagulation. The American Academy of Ophthalmology published an Ophthalmic Procedure Preliminary Assessment of this technique,¹³⁵ highlighting the need for large-scale case series and randomized trials before widespread acceptance of this approach should be considered.

When AMD-related CNV is associated with a large submacular hemorrhage, intraocular injection of gas and face-down positioning may allow for mechanical displacement of the blood and at least temporary recovery of visual acuity.¹³⁶ The improvement may not have long-term benefit if the underlying neovascular lesion and its accompanying destruction of the retina, not the blood, determines the ultimate visual outcome. More experience is needed to evaluate the safety and benefit of these surgical approaches.

Indocyanine green angiography

Indocyanine green is a dye that is more highly protein-bound than sodium fluorescein and that fluoresces in the near-infrared wavelength. These properties were suggested to be useful in the evaluation and management of CNV (see Chapter 2, Indocyanine green angiography). Three basic patterns of fluorescence have been reported in indocyanine green angiography of CNV judged to be occult on fluorescein angiography: a small, focal “hot spot” (a bright area of fluorescence more than one disc area that usually shows by the mid-phase of the angiogram), a plaque (a well-demarcated area of fluorescence more than one disc area in size that emerges relatively late in the angiogram), and ill-defined fluorescence.^137,138 Because the hot spot is small and may be extrafoveal in location, attention is drawn to it as a possible treatment site. Pilot studies utilizing photocoagulation of this site have reported improvement or stabilization of visual acuity, as well as resolution of “exudative features.”^137–140 However, similar outcomes have been observed in eyes with occult CNV that have not received any treatment.^82,139 For example, a reanalysis of data used in the MPS juxtafoveal trial suggested that eyes with occult only CNV may have a more favorable natural course than eyes with classic CNV only. It is possible that the outcome of treating hot spots is no better than that of the natural course of these lesions. As described previously, eyes with both classic and occult CNV did not benefit from photocoagulation of the classic component alone. Treatment of the entire neovascular complex appears to be necessary. Treatment of a “hot spot” alone, identified on indocyanine green and associated with a larger area judged to be occult CNV on fluorescein angiography, may be similar to the unsuccessful strategy of treating just the classic CNV in an eye with both classic and occult CNV on fluorescein angiography. Until carefully designed, randomized clinical trials are conducted that show that indocyanine green-guided laser therapy of AMD-related CNV results in a better visual outcome than no treatment, one cannot know for sure if this particular intervention is beneficial.¹⁴¹ The role of ICG angiography in the setting of polypoidal choroidal vasculopathy patterns of CNV also is discussed further in Chapter 71 (Polypoidal choroidal vasculopathy).

Radiation therapy

The use of radiation therapy for CNV in AMD is based on the possibility that radiation can damage rapidly proliferating neovascular tissue. Studies have been inconsistent in documenting a benefit to this approach, either using external beam, plaques, or use of a probe with local irradiation.^142–144

Other pharmacologic therapies and combination therapies

A variety of other pharmacologic therapies, delivery devices for pharmacologic therapies, and combinations of therapies are under investigation,^145,146 but given the public health importance and biologic variability of visual acuity outcomes with CNV in AMD, changes in the management of CNV in AMD should be influenced strongly by large-scale, appropriately designed randomized clinical trials with adequate follow-up that show benefit before new therapies are incorporated into standard practice.