Predilection of the Macular Region to High Incidence of Choroidal NeovascularizationAfter Intense Laser Photocoagulation in the Monkey FREE

Choroidal neovascularization (CNV), the invasion of newly formed bloodvessels from the choroid through a break in Bruch membrane, is a feature ofmany eye diseases, including age-related macular degeneration, ocular histoplasmosis,angioid streaks, high myopia, chronic uveitis, and choroidal rupture.^{1 - 2} Choroidal neovascularization developsunderneath the retina and disrupts the retinal pigment epithelium (RPE) andneurosensory retina, resulting in severe visual loss in most cases.^{1 - 2} The conventional management of CNVby laser photocoagulation has inherent drawbacks, including less dramaticeffect on subfoveal CNV, a high rate of CNV persistence and recurrence, andirreversible damage to the RPE and normal retina.^{3 - 5} Severalnew therapeutic strategies, including photodynamic therapy,⁶ viralvector–mediated gene therapy,^{7 - 8} andnew antiangiogenic drugs,⁹ have been introducedor are being investigated, and all of these need a reproducible and clinicallyrelated model of CNV.

Induction of CNV in animals can be achieved in 2 ways. One method isto create acute breaks in Bruch membrane traumatically^{9 - 10} orenzymatically,¹¹ while the other is to induceCNV without immediate breaking of Bruch membrane. The disease process inducedby the latter method is slow, and the permeability of the newly formed bloodvessels is uncertain.^{12 - 13} Intenselaser photocoagulation has been used to break Bruch membrane to create CNVin the rat,^{14 - 15} rabbit,¹⁶ cat,¹⁷ and monkey.^{9 - 10} The rat and monkey models have beenthe most commonly used in eye research. Previous success rates of laser-inducedCNV of 50% to 80% have been reported in the rat.^{14 - 15} Theincidence of laser-induced CNV in the monkey, however, is only 30% to 40%,and results vary markedly.^{9 - 10}

Previously, clinical and experimental investigations have overlookedthe importance of the macular region in CNV development.^{1 - 2 ,9 - 10} Ryan¹⁰ reported that 39% of laser-induced lesions developedclinical CNV (fluorescein leakage) in the macular region, with 3% in the areanasal to the optic nerve head and 0.3% in the peripheral retina of the rhesusmonkey. However, no information is available to clearly define the anatomicterritory of the macula; thus, no landmark index can be used to preciselydeliver laser lesions and achieve a high incidence of CNV in the primate model.As the retinal-choroidal circulation and macular anatomy are similar betweenhumans and primates, the information obtained from the monkey model of CNVwould be particularly valuable and crucial for assessment of new antiangiogenictherapies before clinical application. In a previous study,⁹ however,the low incidence of CNV in monkeys limited statistical evaluation of an antiangiogenictherapy. Considering the valuable information for clinical application andthe high cost of experimentation on monkeys, it is critical to find key factorsthat would improve the success rate of laser-induced CNV in the monkey model.In this study, we measured the anatomic size of the macular region and analyzedthe incidence of laser-induced CNV within this region, compared with thatoutside the macular region. Our results show that the macular region is predisposedto the creation of laser-induced CNV in the monkey.

Eight eyes from 4 Macaca fascicularis monkeyswere used in accord with the guidelines of the Association for Research inVision and Ophthalmology on the use of animals in research and following theguidelines of the Animal Care Committee at National University of Singapore.The animals were anesthetized with ketamine hydrochloride (20 mg/kg of bodyweight), acepromazine maleate (0.25 mg/kg), and atropine sulfate (0.125 mg/kg).Pupils were dilated with 2.5% phenylephrine hydrochloride and 1% tropicamidedrops. A trained veterinarian monitored the airway, respiration, and pulseduring all procedures.

Laser photocoagulation was performed as previously described,^{9 - 10} with modifications. Three ophthalmiclasers, a krypton red (647 nm) (Coherent Radiation System, Salt Lake City,Utah), krypton green-yellow (528-568 nm) (Coherent Radiation System), andan Alcon frequency-doubled diode (532 nm) (Alcon 532 Ophthalas EyeLite Photocoagulator;Alcon, Inc, Houston, Tex), were used in this study (Table 1). With each ophthalmic laser, the levels of energy wereinitially assessed in an area away from the macula to test the ability toproduce a blister without subretinal hemorrhage. The final protocols usedwere 1.0-W power density, 50-µm spot size, and 0.2-second duration forthe krypton red; 1.5-W power density, 50-µm spot size, and 0.2-secondduration for the krypton green-yellow; and 1.5-W power density, 50-µmspot size, and 0.1-second duration for the Alcon frequency-doubled diode lasers.Eight to 13 lesions were delivered in an area between the temporal vasculararcades in each eye using a slitlamp and fundus contact laser lens.

Fundus photographs were taken immediately after laser photocoagulationand at 2 and 4 weeks after laser treatment. Fundus fluorescein angiography(FFA) was performed to monitor the development of CNV at 2 and 4 weeks afterlaser treatment. Angiography was performed by intravenous injection of 10%fluorescein sodium (0.1 mL/kg of body weight). Identification of CNV was basedon fluorescein behavior during the phases of angiography from 10 seconds to10 minutes after dye injection. The status of the lesions was graded in amasked fashion by 2 examiners (W.-Y.S. and C.-M.L.) using reference angiograms.The scores were recorded as follows: 0, no hyperfluorescein staining; +, slighthyperfluorescein staining; ++, moderate fluorescein leakage; and +++, prominentfluorescein leakage or pooling. Choroidal neovascularization was characterizedby the presence of fluorescein leakage or pooling at a late phase of angiography,and only the lesions that scored 2 and higher were considered clinical CNV.⁹ Based on the criteria of the Macular PhotocoagulationStudy Group,¹⁸ CNV on angiography was consideredclassic when well-demarcated boundaries were discerned at an early stage ofFFA, with progressive pooling or leakage in later phases of angiography. OccultCNV was characterized by a stippled area of hyperfluorescence noted within1 to 2 minutes after dye injection, with leakage in this area in the latephase; or by areas of leakage at the level of irregular elevation of RPE inthe late phase, without well-demarcated areas of hyperfluorescence discerniblein the early phase.

All monkeys were humanely killed 4 weeks after laser treatment. Theeyes were enucleated and fixed in 2.5% glutaraldehyde plus 2% paraformaldehydein 0.1M phosphate-buffered saline for a minimum of 24 hours, then processedfor paraffin embedding. Paraffin sections of 5-µm thickness were cut,and serial sections were collected when the first lesion was identified. Hematoxylin-eosinstaining was performed every 4 slides. The primary goals of histological analysiswere to: (1) correlate the incidence of leaky lesions with the number of microscopicCNV lesions detected by histological analysis, (2) determine the associationof broken Bruch membrane with CNV development, and (3) orient histologicalCNV inside or outside the macular region (Table 2).

To evaluate the imbalance of CNV development between the macular andnonmacular regions, the size of the macula was initially determined by histologicalanalysis (Figure 1). The maculawas identified as a multilaminar ganglion layer, with maximum cell densityin the parafoveal retina and a gradual disappearance of ganglion cell andinner nuclear layers at the central fovea (Figure 1A). The ratio of the macula to the horizontal optic discwas measured in 4 samples from 4 monkeys. The ratio was used as an index tofurther determine the number of lesions within and outside the macular regionson angiography.

With maximum levels of laser energy, photocoagulation using the kryptonred and green-yellow and the Alcon frequency-doubled diode lasers did notinduce obvious hemorrhage in most lesions, except for 2 testing spots showingminor hemorrhage where the lesions were delivered on small retinal vessels(data not shown). Four eyes were treated with the krypton red laser (Table 1). Of 34 lesions delivered, 5 (15%)developed fluorescein leakage on angiography. In 1 eye in which all 8 lesionswere placed outside the macular region, none showed clinical CNV by FFA (Figure 2A). In the remaining 3 eyes in which26 lesions were delivered, 5 (38%) of 13 lesions developed clinical CNV insidethe macular region, compared with none showing leakage outside the macula(Figure 2B). Using the krypton redlaser, none of the 21 lesions outside the macular region developed CNV-relatedfluorescein leakage at 2 and 4 weeks after laser treatment. Eleven lesionswere delivered into 1 eye using the krypton green-yellow laser, 3 (27%) ofwhich developed fluorescein leakage within the macular region. In contrast,none of the 8 lesions outside the macular region developed CNV-related fluoresceinleakage (Figure 2C). Using the frequency-doubleddiode laser, 33 lesions were delivered into 3 eyes, 19 (58%) of which developedCNV-related leakage (Figure 2D-F).Of 13 lesions delivered inside the macular region, all developed CNV-relatedfluorescein leakage (Figure 2D andF). Of 20 lesions that were delivered in the boundary or outside the macula,however, only 6 (30%) showed CNV-related fluorescein leakage (Figure 2D and F). In general, regardless of the type of ophthalmiclaser used, 21 (72%) of 29 lesions developed CNV-related leakage inside themacular region, compared with 6 (12%) of 49 lesions outside the macula (P<.001, unpaired t test). Inall cases, CNV-related fluorescein leakage was observed at 2 weeks and at4 weeks after laser treatment. Based on the criteria of the Macular PhotocoagulationStudy Group,¹⁸ most CNV lesions in the laser-inducedmodel showed features of classic CNV, characterized by well-demarcated boundariesdiscernible at an early stage of angiography, with progressive pooling orleakage in later phases of angiography (Figure2E and F). However, occult CNV lesions were also observed, characterizedby poorly defined areas of hyperfluorescence in the early phase and progressivepooling or leakage in later phases of angiography (white arrow, Figure 2E and F).

Histological analysis of 10 lesions outside the macular region in 4eyes treated by the krypton red laser showed broken Bruch membrane in 5 lesions,but all failed to develop histological CNV (Table 2 and Figure 3Aand B). Inside the macular region, however, 9 of 10 lesions showed brokenBruch membrane, and all developed microscopic CNV (Figure 3C and D). The development of CNV was initiated with disruptionof Bruch membrane, and the new vessels originated from the choriocapillariesand extended into the subretinal space, forming neovascular membrane networks.Histological analysis of the macular region in the eye treated by the kryptongreen-yellow laser was not available, because of unexpected tissue destruction.Two lesions outside the macular region showed broken Bruch membrane, but nonedeveloped microscopic CNV. Among the eyes treated by the Alcon frequency-doubleddiode laser, all 29 lesions showed broken Bruch membrane, regardless of theirgeographic location. Within the macular region, 16 (89%) of 18 lesions showedmicroscopic CNV (Figure 4A and B).In contrast, 5 (45%) of 11 lesions outside the macular region developed histologicalCNV, and most of them formed fibrous tissue. Among the few lesions showingmicroscopic CNV in the extramacular region, the number of new vessels waslimited, and most of them were embedded in firmly packed fibrous tissue (Figure 4C-F). Among some lesions, severalelongated fibroblast-like cells and pigment-laden cells enveloped the newvessels, apparently limiting their further development (Figure 4E and F). In general, regardless of the type of ophthalmiclaser used, 25 (89%) of 28 lesions developed histological CNV inside the macularregion vs 5 (22%) of 23 outside the macular region (P<.001,unpaired t test).

Our results demonstrate that the macular region was predisposed to laser-inducedCNV. Fundus fluorescein angiography showed that 72% of lesions developed CNV-relatedfluorescein leakage inside the macula, compared with 12% outside the macula.Furthermore, histological analysis showed that 89% of lesions developed microscopicCNV inside the macula vs 22% outside the macula.

Previous investigations have demonstrated that the presence of fluoresceinleakage or pooling on angiography correlated with microscopic CNV, but thatnot all microscopic CNV developed fluorescein leakage on angiography.^{10 ,14 ,19 - 20} Inthe present study, 27 (35%) of 78 lesions developed fluorescein leakage onangiography, but 30 (59%) of 51 lesions on histological analysis showed microscopicCNV. This observation is similar to findings in a previous study¹⁴ ina rat model showing that 28% of lesions developed fluorescein leakage, withsubsequent histological analysis revealing microscopic CNV in 60% of lesions.In primate models, Ryan¹⁰ and Miller et al²⁰ reported that the incidence of lesions showing fluoresceinleakage on angiography was much lower than that of microscopic CNV. In humans,Gass²¹ reported that, while FFA failed to showfluorescein leakage, subsequent postmortem examination demonstrated pathologicCNV development. The reason why some histological CNV membranes remained inactiveduring FFA is unclear. In the present study, most lesions inside the maculadeveloped classic CNV. However, we observed that certain lesions inside themacula developed occult CNV. By histological analysis, most CNV membranesoutside the macula contained fewer new vessels, embedded in firmly packedfibrous tissue, compared with the microscopic CNV inside the macula, and somenew vessels were enveloped by elongated fibroblast-like cells and RPE-likecells. The limited number of microscopic CNV lesions and the absence of afluid-filled space may account for the inactivation of fluorescein leakagefrom lesions outside the macula. It could be possible that the CNV lesionswithin the macular region mainly developed classic CNV, while the lesionsoutside the macular region predominantly formed occult CNV, which is difficultto interpret on angiography. In addition, the differences in endothelial cellfenestration and pericyte maturation between leaky and nonleaky new vessels—andthe local molecular mechanisms, such as the level of vascular endothelialgrowth factor that is known to initiate and maintain angiogenesis and to increasevascular permeability, may also affect the activity of lesions during angiography.^{15 ,19 - 20 ,22}

Previous studies^{9 - 10} haveshown that the occurrence of hemorrhage at the time of laser photocoagulationdoes not affect CNV development. In contrast, massive subretinal hemorrhageusually leads to scar tissue formation.^{9 - 10} Becausethe intensity of laser burns varies depending on treatment protocols (powerdensity, spot size, and duration), properties and conditions of ophthalmiclasers, fundus pigmentation, and magnification factors of laser lens used,we initially delivered laser spots far away from the macular region for energytesting. The final laser protocols were designed to promote laser burns withheating blasters but without obvious hemorrhage. McAllister and colleagues²³ showed that an ophthalmic laser with a green wavelengthspectrum requires at least a 1.5-W power density to reliably break Bruch membranein humans when a 50-µm spot size and a 0.1-second duration were applied.In this study, we used the most powerful laser burns to break Bruch membrane.For the krypton red laser, a 1.0-W power density was used because of the evidencethat rupture of Bruch membrane requires less power density compared with thegreen spectrum lasers.²⁴ In the present study,of 91 lesions delivered in 8 eyes, only 2 showed minor hemorrhage in whichthe lesions hit small vessels. With the krypton red laser, rupture of Bruchmembrane occurred much more frequently inside the macular region comparedwith outside the macular region.

Based on the histological analysis, CNV only occurred in lesions inwhich Bruch membrane was broken. However, a break in Bruch membrane was notalways sufficient to induce CNV. In the laser-induced CNV model, factors affectingthe success rate of CNV may include the wavelengths and energy levels of ophthalmiclasers and the locations where lesions are delivered. Moreover, a recent study²⁵ demonstrated age to be an independent risk factorfor severity of laser-induced CNV in rodents. With the 3 different types oflaser used, the success rate of CNV ranged from 15% to 58% in the presentstudy. Given the fact that different power densities and durations were usedfor different types of lasers, it is hard to draw any conclusions about therelative efficiency of these lasers in terms of their production of CNV, althoughit appeared that the Alcon frequency-doubled diode laser produced the highestincidence of CNV when the highest energy level was used.

The mechanisms of laser-induced CNV and the imbalance of CNV developmentbetween the macula and extramacula are unclear. It has been proposed thata balance between factors that stimulate or inhibit vessel growth controlsneovascularization.^{26 - 27} In mostnormal tissues, inhibitory factors are active and vessels remain quiescent.²⁶ In contrast, in different pathologic states, suchas tumor growth and neovascular forms of age-related macular degeneration,neovascularization occurs because of the disruption of the balance betweenangiogenic factors and angiogenic inhibitors.^{26 - 28} Recentinvestigations have shown that sole overexpression of vascular endothelialgrowth factor, one of the most potent angiogenic growth factors, is sufficientto induce ocular neovascularization.^{22 ,29} Furthermore,vascular endothelial growth factor is involved in the maturation and stabilizationof newly formed vessels.²⁸ With a disruptionof Bruch membrane, the effect of laser photocoagulation on the developmentof CNV is often attributed to the up-regulation of angiogenic factors. Previousstudies^{15 ,30 - 32} havedemonstrated that expression of several angiogenic factors, including vascularendothelial growth factor, basic fibroblast growth factor, and matrix metalloproteinase2, is up-regulated in activated RPE and infiltrating cells in laser-inducedCNV. Most recently, Renno et al³³ demonstratedthat down-regulation of pigment epithelium–derived factor, a potentendogenous inhibitor for vascular endothelial cell proliferation and migration,also enhances laser-induced CNV. In the normal retina, pigment epithelium–derivedfactor is expressed most intensely in the outer nuclear layer (photoreceptornuclei).³³ After laser treatment, however,immunostaining for the factor within the outer nuclear layer was absent ordecreased for up to 3 weeks, which seemed to parallel the up-regulation ofvascular endothelial growth factor.^{15 ,30 - 31 ,33} Inthe present study, the macular region was predisposed to the incidence oflaser-induced CNV. Anatomically and physiologically, the macula is differentfrom the extramacula by its denser distribution of photoreceptors and RPEcells and more abundant blood supply from the choroidal circulation.^{34 - 35} It is possible that these differenceslead to the imbalance between the macula and extramacula in overexpressionof angiogenic factors and in down-regulation of angiogenic inhibitors afterlaser photocoagulation. Immunostaining for angiogenic factors and angiogenicinhibitors on macular and extramacular new vessels could clarify these speculations.However, all enucleated eyes in the present study were fixed in 2.5% glutaraldehydeplus 2% paraformaldehyde for a minimum of 24 hours, which challenges the immunostainingtechniques. Our study was initially designed to increase the incidence ofCNV in the primate model for preclinical evaluation of antiangiogenic therapies,rather than to investigate the mechanisms regulating CNV formation. When freshtissue is available in the future, further investigations will be directedto elucidate the molecular mechanisms that could explain the imbalance oflaser-induced CNV between the macular and extramacular regions.

The predilection of the macular region to a high incidence of CNV isalso of clinical significance. Iatrogenic CNV is a common complication oflaser photocoagulation for treatment of subfoveal CNV.³⁶ Previousstudies^{12 ,37 - 38} demonstratedthat chronic cellular processes following moderate-intensity laser treatmentresulted in gradual dissolution of Bruch membrane. Clinically, comparablelaser intensities unable to immediately break Bruch membrane induce choriocapillarybudding, with subsequent digestion of Bruch membrane and histological CNVformation.^{12 ,37 - 38} Thecurrent observation, the predilection of the macular region to a high incidenceof laser-induced CNV, may account for the high recurrence rate of subfovealCNV after laser treatment in humans.

Corresponding author: P. Elizabeth Rakoczy, PhD, Department of MolecularOphthalmology, Lions Eye Institute, 2 Verdun St, Nedlands, Western Australia6009 (e-mail: rakoczy@cyllene.uwa.edu.au).

Submitted for publication October 25, 2002; final revision receivedApril 23, 2003; accepted August 19, 2003.

We thank Tammy Zaknich and Benjamin Rae for their assistance in histologicalpreparations; Paul Pineda, Barathi Amutha, and Robert Ng for animal anesthesia;and Joseph Ho for fundus photography and fluorescein angiography.