Abstract
purpose. To obtain high-quality angiograms of the rat choriocapillaris with continuous laser-targeted angiography (LTA), for the purpose of assessing the choroidal circulation system in vivo by studying the patterns of the images.
methods. A slit lamp was modified to incorporate two kinds of lasers (argon and diode). Carboxyfluorescein was encapsulated in heat-sensitive liposomes and injected intravenously. Encapsulated carboxyfluorescein was released locally by applying a continuous heat beam provided by diode laser (810 nm) with various powers. Video angiograms were generated with excitation illumination provided by argon laser (488 and 514 nm) to observe highly selective images of the choriocapillaris.
results. Three distinct phases (filling, plateau, and draining) were observed in fluorescent images of choriocapillaris by applying the diode laser continuously. In the plateau phase, a lobe-shaped area of choriocapillaris peripheral to the laser site was illuminated, and this finite area did not change in size with continuous laser application to the same spot. When laser power was increased, a larger area of choriocapillaris was illuminated in the plateau phase. The filling and draining phases demonstrated the flow patterns in choriocapillaris lobules, which filled from a central spot and drained along a peripheral ring.
conclusions. This study showed that the rat choriocapillaris is divided into independent functional units and that the choroidal circulation is segmental under normal conditions. The results implied that in LTA, the diode laser warms up a choroidal artery and the released fluorescein flows downstream to an area of choriocapillaris fed by the same artery. LTA appeared to be a powerful method to analyze choroidal circulation in vivo.
The choriocapillaris, a single layer of capillary vessels located below the retinal pigment epithelium (RPE), provides metabolites to the outer retina and RPE. Abnormalities in the choriocapillaris are thought to be linked to several fundus diseases such as age-related macular degeneration. However, little information is available about its normal physiology and involvement in the disease process, mainly because of the difficulty in visualizing it in vivo.
Historically, considerable controversy about the end artery system and segmental distribution of the choroidal circulation has existed. Although some researchers have found that the choroidal arteries have a segmental supply to the choroids indicating that they behave as functional end arteries,
1 2 3 4 5 6 7 8 others have refuted this idea.
9 10 11 12 13 A review of the literature shows that a segmental distribution has been postulated by investigators whose observations are based on in vivo occlusion in animals and humans, whereas a nonsegmental view has been based mainly on postmortem injection studies in animals and humans. This disparity between the two viewpoints appears important. After a corrosion cast study of the microvascular architecture of the rat choroid, Bhutto and Amemiya
14 reported no evidence of a lobular arrangement of the choriocapillaris. To date, few in vivo physiological studies of the rat choroid have supported or disproved this anatomic evidence, because an adequate experimental method was not available.
Zeimer et al.
15 16 have developed a method of laser-targeted delivery that provides controlled local release of pharmacologic substances to the ocular fundus. The method consists of encapsulating the substance in heat-sensitive liposomes (lipid vesicles that can be disrupted by heat), injecting the liposomes intravenously, and then disrupting them, causing a release of their contents by warming the target tissue (retinal or choroidal vessels) with a mild, noncoagulating laser pulse through the pupil. One application of laser-targeted delivery is laser-targeted angiography (LTA), which consists of encapsulating carboxyfluorescein, a derivative of fluorescein, in liposomes. The encapsulated concentration is high enough to cause quenching of the fluorescence, thus making the circulating liposomes invisible in the angiogram. When the dye is released from the liposomes with a local pulse of heat, it is diluted in the plasma and yields a bright fluorescent bolus that selectively highlights either retinal or choroidal vessels as the bolus travels through the bloodstream. Observing this bolus as it travels from the arteries through the arterioles, capillaries, and venules generates selective, high-quality angiograms of the retinal vasculature
17 and choriocapillaris.
18 19 20 The images obtained by this method are totally different from the conventional fluorescent angiograms (FAG) or indocyanine green angiograms (ICG), in that they are highly selective for the choroidal circulation and clearly show the dynamic filling of the choriocapillaris. This original method uses a pulsatile laser beam to disrupt heat-sensitive liposomes.
In this LTA study, we applied a pulsatile laser beam and a continuous-heat laser beam. Continuous application of the laser allowed the prolonged observation of the movement of the dye front in rat choriocapillaris with high-quality images. We also changed the laser power and application site to see whether different angiographic patterns could be obtained. By studying those various images, we analyzed the physiological choroidal circulation system to inform the controversy about the segmental distribution of choroidal circulation.
The temperature of RPE and choroidal tissue is raised by using a laser with energy that is absorbed by hemoglobin and melanin. This temperature increase corresponds to the phase-transition temperature (40°C) in heat-sensitive liposomes, causing the release of their contents.
We modified a slit lamp (model SL-10L; Topcon, Tokyo, Japan) to incorporate two kinds of lasers. The beam of the argon laser at 488 nm (blue) and 514 nm (green) (Novus 2000; Coherent, Palo Alto, CA) was conducted through a reflection mirror and used as the slit light source to excite and visualize the fluorescent dye. To warm the target tissue, the beam of a diode laser at 810 nm (F-System; Coherent) was fed into a fiberoptic cable and applied to the fundus through the slit lamp. A zoom system of the diode laser allowed a continuously variable spot diameter (50–1000 μm). In addition, the duration and power of irradiation of the diode laser could be adjusted with an attached controller. A continuous diode laser application mode was installed and controlled by a foot switch. The diode laser spot was moved with a joystick. The images taken by the charge-coupled device (CCD) camera (Sony, Tokyo, Japan) mounted on the modified slit lamp are displayed on a monitor after being amplified by a video enhancer (Seprotec, Tokyo, Japan). These images can be recorded in media with S-VHS video recorder (Sony) for analysis. The sequences of video images are digitized by media converter (Sony) and analyzed with Adobe Photoshop (Mountain View, CA).
The actual power of the illumination laser at 488 nm (blue) and 514 nm (green) and the hyperthermic diode laser at 810 nm was measured. The laser power analyzer (Ultima Labmaster; Coherent) was set up just in front of the rat cornea. The actual power of the illumination laser was 0.17 mW at the corneal surface, and it was applied for 1.0 to 10.0 seconds. Under the American National Standards Institute (ANSI) standard,
21 a power of 0.4 mW of continuous argon laser is permissible when the eye is immobilized and the pupil is dilated. The measured intensity of the hyperthermic laser at 810 nm was equal to that indicated by the controller. According to the standard, a power of 10 to 30 mW (the power range that we used) of continuous diode laser is considered class 3 of the Laser Hazard Classification of the ANSI standard.
5,6-Carboxyfluorescein (CF; Molecular Probes, Junction City, OR) was purified on a hydrophobic gel (lipophilic Sephadex LH-20; Sigma-Aldrich, St. Louis, MO) and diluted to approximately 100 mM. The diluted dye was filtered through a 0.22-μm syringe filter (Millex-GV; Millipore, Bedford, MA). Dipalmitoylphosphatidylcholine (DPPC) and dipalmitoylphosphatidylglycerol (DPPG; Genzyme, Liestal, Switzerland) were used without further purification. Liposomes were prepared by a method described by Mayer et al.
22 Briefly, lipids were dried (dissolved in chloroform and methanol) to a thin film by rotary evaporation under vacuum. A 4% solution of CF filtered through a 0.22-μm filter was mixed with the dried lipid film, and the mixture was subjected to five freeze-thaw cycles. Next, extrusion sizing was performed in a thermobarrel extruder (Lipex Biomembranes, Vancouver, British Columbia, Canada) through a stack of two 0.2-μm polycarbonate membranes (Millipore) 10 times to yield large unilamellar vesicles. Unentrapped CF was removed by passing it through a purification column (Sephadex G-50; Pharmacia Biotech, Uppsala, Sweden). In the present study, 40°C liposome (phase-transition temperature, 40°C) was prepared by using DPPC and DPPG with a ratio of 4:1 (mol/mol).
The amount of CF released was assayed by measuring the fluorescence with a spectrofluorophotometer (Shimadzu, Kyoto, Japan), at 488 nm (excitation) and 514 nm (emission). As samples, a 30-μL liposome suspension was mixed with 3 mL of 50% human serum. Triton X-100 (0.1 mL; Sigma-Aldrich), used to disrupt the vesicles and release the entrapped CF, was added to the control sample at room temperature. The case samples were placed in a water bath and heated at different temperatures for 5 minutes. The release yield was calculated from the ratio between the fluorescence level of the case samples and that of the control sample. The release yield was 5% at 38°C (approximately physiological body temperature) and it increased up to 95% at 41°C
(Fig. 1) .
Male Long-Evans rats, weighing 180 to 200 g each, were used for the study. LTA was performed in 15 rats. The animals were treated in compliance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The rats were anesthetized with intramuscular ketamine (10 mg/kg) and xylazine (4 mg/kg). Topical 1% tropicamide and 2.5% phenylephrine hydrochloride were instilled for mydriasis during LTA. Afterward, the rats were killed with an overdose of intravenous pentobarbital sodium.
Immediately after 1.0 mL/kg of CF-liposome suspension was injected intravenously into the tail veins, LTA was performed with the modified slit lamp through a handheld 78-D lens (Folk, Mentor, OH) placed in front of rat cornea. The 810-nm laser diode with a diameter of 500 μm was applied to various locations on the rat fundus to release a bolus of dye from the liposome. The diode laser remained turned on while a foot switch was operated. The laser application site was controlled with a joystick, and the laser power was adjusted from 10 to 30 mW. The argon laser at 488 nm (blue) and 514 nm (green) with a power of 2.9 mW/cm2 was used as the slit light to excite fluorescence. A long-pass filter (Omega Optical, Tokyo, Japan) was used to block any wavelength shorter than 530 nm. Because 488-nm argon blue light and 514-nm argon green light were used as slit light source, dim background pseudofluorescent images of the fundus were observed without the presence of fluorescent material, facilitating focus and orientation on the fundus.
In one animal, during LTA, 120 seconds of diode laser with a power of 20 mW was applied to one fundus location at two disc diameters from the optic disc in the 12 o’clock direction. The animal was killed with an overdose of pentobarbital sodium. The eyes were enucleated, placed in fixative (phosphate-buffered 2.5% glutaraldehyde) for 24 hours, dehydrated with graded alcohols, and embedded in epoxy resin. Serial sections 2.5 μm thick were obtained in the treated area. Every fifth section was stained with hematoxylin-eosin and examined under a light microscope.
LTA was performed in 15 rats, providing a highly selective visualization of the rat choriocapillaris. Immediately after application of the diode laser and warming of the choroidal tissue, a bolus of dye was released from heat-sensitive liposomes and detected by its intense fluorescence with excitation illumination provided by the argon laser slit lamp. The threshold power of the diode laser (810 nm), at which choriocapillaris started to be illuminated, was 10 mW. A particular area of choriocapillaris was illuminated only when the diode laser was applied to certain locations of fundus. Illumination suddenly stopped when laser application was moved to adjacent locations (even 100 μm). The bolus of dye released in the choroidal vasculature was not accompanied by a release in the retinal vasculature. This is illustrated by the unchanged fluorescence of the retinal vessels.
Three distinct phases of fluorescent images of choriocapillaris were observed by applying the diode laser continuously. The first was a dynamic filling phase
(Fig. 2A 2B 2C 2D) , which was observed immediately after the start of laser application to the fundus. The second was a plateau phase
(Figs. 2D 2E 2F 3) , in which, no matter how long the laser energy was applied to the same location, the same area continued to be illuminated. The third phase was a draining phase
(Fig. 2G 2H 2I) . It began immediately after laser application was stopped. The dynamic filling phase lasted approximately 0.8 to 1.0 second. The draining phase was approximately 1.0 to 2.0 seconds. The plateau phase was controlled by how long the foot switch was kept on.
In the plateau phase, a lobelike or palmlike area of choriocapillaris was uniformly illuminated and stretched peripherally from the laser site in most cases
(Fig. 3) . This illuminated area did not change in size during laser application. The dye did not diffuse beyond a certain border, and the finite area surrounded by this border continued to demonstrate a uniform level of fluorescence. We compared this finite area of fluorescence when the laser site was moved to various proximate positions
(Fig. 4) . Laser application illuminated each corresponding finite area of choriocapillaris, and each finite area was located side by side with almost no gap or overlap. It fitted just like a piece of a jigsaw puzzle.
In the filling phase
(Fig. 2) , discrete spots of fluorescence, away from the laser delivery site, appeared and expanded rapidly. These discrete spots of fluorescence were located widely across an area of choriocapillaris in a patchy fashion. From each spot of fluorescence, dye spread to the surrounding area of choriocapillaris radially, and subsequently each dye front coalesced to illuminate the choriocapillaris uniformly. Immediately after the diode laser application was discontinued, blood flow with fluorescent dye was replaced gradually by nonfluorescent blood flow, showing a honeycomb pattern
(Fig. 2) . Nonfluorescent blood flow entered from a central spot, which corresponded very well with the discrete spot of fluorescence that first appeared in the filling phase. Subsequently, the nonfluorescent spots expanded radially and drained along a peripheral ring. These filling and draining patterns were clearly consistent with the flow in choriocapillaris lobules, which filled from a central spot and drained along a peripheral ring.
The three distinct phases as described were observed in all the animals we studied. As we proceeded, we found that unique patterns of choriocapillaris were illuminated in the plateau phase. The unique patterns shown in
Figure 5 were observed in 18% of 50 plateau-phase images that were obtained from five animals (five images for each eye).
Figure 5A 5B 5C shows a patch with no fluorescence within the illuminated choriocapillaris.
Figure 5A shows several illuminated patches of choriocapillaris were separated from each other. In the filling phase when the laser application began, each patch demonstrated several points of dye being released. Dye spread radially from the releasing points and eventually coalesced to form a patch of illuminated choriocapillaris.
Figures 5B and 5C show two lobes of choriocapillaris connected by a narrow bridging strip. When the laser was applied to the strip, no area of choriocapillaris was illuminated. Instead, when the laser energy was moved slightly away from the strip, the distal lobe was illuminated.
When the diode laser power remained below 13 mW, the illuminated area (in the plateau phase) was fairly small. As the laser power was increased toward 30 mW, a larger area of choriocapillaris stretching from more centrally applied laser site to the periphery was illuminated
(Fig. 6) . If the diode laser power was maintained below 20 mW, no laser burn was observed, even when continuous irradiation was applied to the same location of the fundus for more than 10 seconds. However, when the power surpassed 20 mW, a thermal fleck remained after the laser application
(Fig. 6C) .
Under light microscopy, no histologic abnormalities were observed in the choroid, RPE, or neurosensory retina in all the sections prepared by cutting across the areas exposed to the diode laser
(Fig. 7) .
Laser-targeted angiography (LTA) provided a method to study the physiology of the choriocapillaris in different regions of the posterior pole and in different animal species. Asrani et al.
18 19 and Kiryu et al.
20 developed this method and applied it to rats and primates. These investigators indicated that LTA had several advantages over conventional FAG or ICG. First, the local release of a fluorescent bolus permits visualization of selected vascular beds, such as the choriocapillaris, without interference from the fluorescence of overlying or underlying beds. Second, the short-pulsed release, accompanied by rapid washout, ensures that the dye does not accumulate outside the vessels perfused by the bolus. Third, the angiograms can be repeated for at least 45 minutes (i.e., as long as the liposomes are circulating in the blood). In their original method, they used a pulsatile laser beam (200 ms) to heat liposomes and release the bolus of dye. In the first stage of our experiment, we tested the pulsatile laser to demonstrate the dynamic phase of the perfusion of choriocapillaris and showed the advantages of LTA just described. However, the obtained images were so short-lived that it was difficult to analyze them, and we were interested in seeing long-term movement of the dye front. We therefore modified the instrument to achieve the continuous diode laser application. The modification resulted in higher quality images, and we discovered that there were three distinct phases (filling, plateau, and draining) of fluorescence in images of the choriocapillaris.
Many morphologic studies of choroidal tissue have been performed in humans,
2 3 23 primates,
4 8 24 and rats.
14 25 26 Olver
27 described the functional anatomy of the normal choroidal circulation in humans by scanning electron microscopic examination of microvascular casts. Bhutto and Amemiya
14 performed a corrosion cast study of the microvascular architecture of the rat choroid and reviewed extensively the anatomy of the arterial system and choriocapillaris and venous drainage. They described two long posterior ciliary arteries along the horizontal meridian. In the posterior choroid, these arteries form five to seven branches on each side, supplying the adjacent choriocapillaris. Viewed from the retinal side, the choriocapillaris appears as a nonhomogeneous network of capillaries with different diameters. This monolayer vascular network was located just below the RPE and consisted of a dense honeycomb pattern of capillaries. Bhutto and Amemiya reported no lobular arrangement of the choriocapillaris (i.e., no segmental distribution). The network of choriocapillaris was supplied by feeding arterioles and drained by the collecting venules. These arterioles and venules formed the medium-sized vessels of the choroid in the layer below (i.e., more to a scleral side) the choriocapillaris layer. The large arteries that branch into small arteries and arterioles are located closer to the scleral side and near the optic disc.
In the plateau phase, a lobelike or palmlike area of choriocapillaris was uniformly illuminated and stretched peripherally from the laser site in most cases
(Fig. 3) . This illuminated area did not change in size during laser application. We compared this finite area of fluorescence when the dye laser site was moved to various proximate positions
(Fig. 4) . Laser application to each position created a corresponding area of illuminated choriocapillaris, or lobe. Lobes were located side by side, with almost no gap or overlap although the borderline of each lobe was sometimes highly distorted and complex. They fitted like pieces of a jigsaw puzzle. Within each lobe, several lobules were observed in the filling and draining phases. This finding revealed a functional unit (lobe) in the choriocapillaris that is larger than a lobule and appears to be fed by one artery branch. It is thought that blood does not flow beyond the border of each lobe because there are no anatomic anastomoses between lobes, or a definite pressure gradient is present within each lobe.
There was a strong similarity between the lobar patterns illuminated in the plateau phase by LTA and the three-sided appearance of acute triangular syndromes, which are caused by acute choroidal ischemia due to thrombosis in the posterior ciliary arteries (PCAs) or their arborizations.
27 28 It is thought that in LTA the laser diode warms a choroidal artery and the released fluorescein flows downstream to the area of choriocapillaris fed by the same artery. Hayreh and Baines
24 performed experimental studies involving occlusion of the PCAs in rhesus monkeys and revealed that PCAs do not anastomose at any level with any neighboring artery and are functional end arteries. Our observation that the illuminated lobe did not alter throughout the plateau phase supports the concept that each lobe is an independent unit and that an artery that feeds each lobe is a functional end artery.
Figure 8 shows a representation of the mechanism of choroidal arteries and lobes.
The analysis of the LTA images of the filling and draining phases demonstrated that choriocapillaris lobules are filled from a central spot and drain along a peripheral ring. Hayreh
4 performed in vivo studies on choroidal vascular bed in rhesus monkeys and humans, using fluorescein angiography. His work revealed that the entire choriocapillaris bed is composed of independent small lobules. Each lobule is supplied by a terminal choroidal arteriole situated in the center, and its drainage is by venous channels situated in the periphery of the lobule. Each choriocapillaris lobule is an independent unit, with no anastomoses with adjoining lobules. In the literature,
9 10 11 12 13 postmortem cast preparations have revealed that the choriocapillaris is arranged in one plane as a continuous layer of wide-lumened capillaries forming an uninterrupted anatomic mesh in the entire choroid. This disparity between the in vivo and postmortem studies appears to be important. Our LTA studies demonstrated that each lobule is a functionally independent unit.
The lobules have two possible flow patterns
(Fig. 9) . In pattern 1 each lobule was fed by a corresponding arteriole. No communication existed between lobules. In pattern 2 the lobules were functionally connected with each other. We believe that only pattern 1 is realized. In the filling phase
(Fig. 2) , discrete spots of fluorescence (center of each lobule) appeared almost simultaneously, and the dye expanded radially from each spot. The dye front did not expand gradually from the diode laser spot. As shown in
Figure 5 , we observed in the plateau phase a patch with no fluorescence within the illuminated choriocapillaris. This patch became illuminated when the diode laser was moved to a proximate position. Sometimes several illuminated patches of choriocapillaris were separated from each other in the plateau phase
(Fig. 5A) with several points of dye releasing within each patch. Also, the illuminated choriocapillaris sometimes showed two patches connected by a narrow bridging strip
(Figs. 5B 5C) . When the laser was applied to the strip, no area of choriocapillaris was illuminated. Only when the laser energy was moved slightly away from the strip, was the distal patch illuminated. These findings support the theory that the distal patch is not functionally connected with the central patch but is fed by an independent arteriole.
The independence of each lobule is possible if no anatomic anastomoses exist between them or if a definite pressure gradient exists within each lobule (the highest at a central spot and the lowest along a peripheral ring) so that no blood flows beyond the border of each lobule. Because peripheral resistance is lower in the collecting venules of the lobules than in adjacent lobules, the blood moves into the veins rather than into the contiguous capillaries. Because postmortem cast preparations have revealed that the choriocapillaris is arranged in one plane as an uninterrupted anatomic mesh, the latter appears to be a plausible explanation.
Our findings demonstrate that ciliary arteries and their branches have a segmental distribution in the choroid, that the choriocapillaris is divided into an independent functional unit, and that the choroidal circulation is functionally an end-artery system. This territorial circulation pattern, however, was observed only under physiological conditions. Pathologic conditions such as obstructed arteries or veins and high intraocular pressure could change these patterns, especially if the territorial circulation is not due to the anastomosis-free state between each territory but to the working pressure gradient within it. For example, Ernest et al.
29 performed fluorescein angiograms on rhesus monkey eyes and reported that sectioning the temporal short posterior ciliary artery branches causes a segmental filling of the nasal half of the choriocapillaris in the arterial phase, followed by temporal filling in the late venous phase. Also, ligation of the superior and inferior nasal vortex veins in the animals with their temporal short posterior ciliary artery branches ligated and sectioned resulted in increased filling of the submacular choriocapillaris.
Although Asrani et al.
18 19 in their original method used an argon laser to cause the release of a bolus of dye from the liposomes, we used a diode laser at 810 nm to warm up target tissue so that the heat could penetrate more deeply into the choroidal tissue. We believe that the heat created by the diode laser penetrated into the choroidal arteries or arterioles and released the bolus of dye because an area of choriocapillaris was illuminated only when the light from diode laser was applied to appropriate fundus locations. This illumination was not present when the laser light was aimed at adjacent locations. If the choriocapillaris had been the site of fluorescein release, the choriocapillaris should have remained illuminated no matter where the laser energy was applied. Also, unique patterns of choriocapillaris were illuminated in the plateau phase
(Fig. 5) . When the laser power was increased, very large areas of choriocapillaris were illuminated
(Fig. 6) . It is assumed that, with the increased laser power, the heat penetrated more deeply through the choroidal tissue and reached a large-caliber artery located near the optic disc and that several lobes fed by this large-caliber artery were illuminated.
LTA provided a powerful method for analysis of the choroidal circulation in vivo. It has provided evidence that the choroidal circulation under normal conditions is territorial and it is divided into physiological segments of lobes and lobules. We could also demonstrate a correlation between in vivo choroidal circulation and its anatomic architecture by analyzing the various patterns of the images achieved with the continuous mode of LTA.