Effects of Photocoagulation on Intraretinal Po ₂ in Cat

Abstract

purpose. To test the hypothesis that intraretinal Po ₂ increases after photocoagulation.

methods. Anesthetized cats underwent retinal argon laser photocoagulation. At least 4 weeks after treatment, Po ₂-sensitive microelectrodes were used to record intraretinal Po ₂ profiles from healed photocoagulation lesions in anesthetized cats breathing air. Histopathologic examination of the retinas was used to confirm that the photoreceptors were destroyed and that the inner retinal layers were preserved, though somewhat disorganized, as in human panretinal photocoagulation (PRP).

results. The retina and tapetum were thinner in the lesioned retina than in the nonphotocoagulated retina. Average Po ₂ across the inner 50% of the retina was higher (22 ± 10 mm Hg) in photocoagulated retina than in untreated retina (14 ± 7 mm Hg; P < 0.01; n = 13 cats). The minimum Po ₂ was also significantly higher, whereas choroidal Po ₂ was significantly lower in the photocoagulated retina than in untreated retina. No significant difference was found in the preretinal vitreous. After lesions, inner retinal Po ₂ could also be maintained above zero, even in the absence of retinal circulation.

conclusions. Previous measurements showed increased Po ₂ in the preretinal vitreous of rabbits and pigs (but not cats) after photocoagulation of the outer retina. These intraretinal measurements in cats provide further evidence for a chronic increase in inner retinal Po ₂ in lesioned areas during air breathing.

Diabetic retinopathy is the leading cause of blindness in working Americans. ¹ A few treatments help slow the progression of the disease. ^{2 3} The most effective is panretinal photocoagulation (PRP), which stops neovascularization or causes it to regress. ^{2 3} This treatment consists of applying widespread grids of laser photocoagulation in the peripheral retina with the goal of destroying large areas of photoreceptors. The rationale is that parts of the midperipheral retina are damaged to salvage the central retina and prevent total vision loss. Anti-VEGF drugs, which have proven to be successful in age-related macular degeneration, ^{4 5} are now being tested in patients with diabetic retinopathy, ^{6 7} but PRP is likely to remain a treatment for some time. 

A leading hypothesis ^{8 9 10} of the mechanism for PRP is that the laser damages photoreceptors, which normally consume large amounts of oxygen. ^{11 12 13 14} This leaves more of the oxygen from the choroid available to diffuse to the inner half of the retina, presumably relieving the inner retina of a possible hypoxia that is thought to be associated with capillary loss ¹⁵ and that may be linked with regression of neovascularization. The second hypothesis is that photocoagulation brings to balance levels of growth factors involved in either promoting or inhibiting angiogenesis. ¹⁵ During the development of diabetic retinopathy, this balance is upset and leads to neovascularization. Microarray studies of changes in gene expression in the mouse retina after laser photocoagulation show that this treatment affects genes involved in the regulation of multiple retinal processes, including those involved in angiogenesis. ^{16 17} Of course, it is possible that the altered levels of growth factors are a direct result of the changing Po ₂ levels. 

The evidence supporting the first hypothesis is indirect. To date, most studies have measured only preretinal Po ₂ after PRP. One might expect vitreal Po ₂ to increase, but studies in the cat retina have demonstrated increased Po ₂ only after 100% O₂ breathing, not during air breathing. ^{8 18 19} Increased preretinal Po ₂ was measured during air breathing in rabbits after PRP, ^{20 21} but the rabbit retina has limited circulation and is metabolically different from retinas with a more complete retinal circulation. ^{22 23} Therefore, the relevance of this finding to human PRP is unclear. Increased preretinal Po ₂ was also measured after photocoagulation in the eyes of miniature pigs with ²⁴ and without ²⁵ venous occlusion. 

Only one intraretinal oxygen profile has been published from the cat retina after PRP, and it was recorded during hyperoxia. ¹⁸ Intraretinal Po ₂ was also found to increase in rabbit retina in acute PRP-like lesions. ²⁶ These measurements were made on the same day the lesions were placed; hence, the long-term effects of PRP after healing of the lesion, which are more relevant for the clinical situation, were not observed. No studies have been conducted of intraretinal Po ₂ after PRP under normoxic conditions in vascularized retinas. 

Better understanding of the mechanisms of PRP action could allow optimization of the procedure and could lead to development of less invasive and more specific treatments. In this study, we addressed whether inner retinal Po ₂ is increased after lesions develop that destroy the outer retina. Cat retina is similar to human midperipheral retina, where PRP is performed. Similar to the human midperipheral retina, the cat retina is rod dominated and has a moderately high neuronal density, and its nutritional needs are supplied by a combination of the choroidal and retinal circulations. We used oxygen-sensitive microelectrodes to make Po ₂ measurements several weeks after producing lesions in or near the area centralis, and we conducted histologic studies to verify that the lesions were similar to those made in humans. 

Methods

The experiments required the production of lesions in the retina with an argon laser, followed by a recovery period of several weeks in which the retina eliminated cells that were destroyed by the lesion. At the end of this time, animals underwent a terminal procedure in which retinal Po ₂ was measured, and the eyes were removed for histologic examination. Sixteen animals were used in these experiments. In three cases, oxygen measurements were not made and the eyes were analyzed histologically only. In three other cases, the retinal artery supplying one of the photocoagulation lesions was occluded to study intraretinal Po ₂ under this condition. 

Photocoagulation Procedure

All experimental procedures were conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Adult cats of both sexes were used in the experiments. Before the laser procedure, cats were preanesthetized with ketamine (2.2–4.4 mg/kg) and butorphanol (0.4 mg/kg) and anesthetized intravenously with 0.35 mL/kg 5% sodium pentothal, or they were anesthetized intramuscularly with 11 mg/kg ketamine and 0.4 mg/kg butorphanol. Sodium pentothal (5%) was then administered intravenously as needed, and the animal was hydrated with 0.9% saline at approximately 20 mL · kg⁻¹ · h⁻¹. The cat’s temperature was monitored and maintained with a heating blanket. The pupil of one eye was dilated with topical ophthalmic solutions of 2.5% phenylephrine (Neo-Synephrine; Abbott Laboratories, Abbott Park, IL) and 1% tropicamide (Bausch & Lomb, Rochester, NY). 

Retinal lesions were produced with an indirect ophthalmoscope using one argon laser (Novus Verdi; Coherent Inc., Santa Clara, CA) at Evanston Hospital or with a slit lamp using another argon laser (Ultima 2000; Coherent Inc.) at Northwestern University. Given that the damage produced with the same energy level varied from retina to retina and even with location on the same retina, we started each procedure with a low power and then increased the level until a change in retinal color was observed after a single laser pulse. This setting was then used to produce all the other lesions in the same retina. The lesions produced in this experiment were histologically similar to PRP lesions seen in human retinas. Specifically, they largely preserved the inner retina and destroyed the photoreceptors. ²⁷ We could not assess the functional state of the remaining inner retina in these experiments, and there is no evidence about its condition after PRP in humans. The laser power varied from 200 to 500 mW, and the duration was 0.1 seconds for the first photocoagulator (Novus Verdi; Coherent Inc.). For the second, (Ultima 2000; Coherent Inc.), the power varied from 500 to 800 mW, the duration varied from 0.2 seconds to 0.5 seconds, and the spot size was 500 μm. 

Initially, lesions smaller than 500 μm in diameter were placed on the retina (four cats) because this is the lesion size frequently used clinically; however, in later experiments we placed lesions to cover an area of at least 2 mm in diameter because our electrodes penetrated the retina at a 45° angle. The electrode had to be far enough from the edge of a lesion that there would be only one-dimensional Po ₂ gradients perpendicular to the retinal surface and that the Po ₂ in the lesion would not be affected by the normal retina at the edge of the lesion. Lesions measuring approximately 2 mm in diameter were made by placing smaller lesions close but separated by a small distance. The goal was to produce a large confluent lesion with uniform damage. Lesions of this type were produced in 14 cats. 

Color fundus photographs were taken immediately after the procedure. Animals recovered from anesthesia and were returned to the animal facility until the time of the terminal experiment, which took place at least 4 weeks after the laser procedure. 

Po ₂ Measurements and Histology

Data were obtained from the center of the lesion and from nonphotocoagulated retina on the same section of the same eye in each animal. Intraretinal Po ₂ was measured in 16 animals, 13 of which were used for analysis. Data from one animal were excluded because we could not visually verify the location of the electrode. Data from two animals were excluded because photocoagulation damaged all the retinal layers. 

After Po ₂ measurements, histologic analysis was performed on all eyes. Nine of the 18 eyes were used for measurements of retinal and tapetal thickness. Qualitatively, the retina was similar in all eyes, but nine were excluded because the histologic processing caused folding or detachment of the retina and the thicknesses were considered unreliable. 

The animal preparation for Po ₂ measurements has been described previously. ²⁸ Briefly, the cats were initially anesthetized by intramuscular injection of ketamine (25 mg/kg) and acepromazine (0.12 mg/kg), or they were preanesthetized with intramuscular injection of butorphanol (0.4 mg/kg) and anesthetized by intravenous 5% sodium pentothal (22 mg/kg). Long-term anesthesia was provided by urethane. The cat’s head was immobilized in a head holder, and the eye was attached to an eye ring. The cat was paralyzed with pancuronium bromide (Pavulon; 0.2 mg · kg⁻¹ · h⁻¹) after surgery and was artificially ventilated. The temperature was kept at 38°C to 39°C. Electrocardiography was performed, and arterial blood pressure and blood gases were monitored throughout the experiment. Blood gases were kept in the normal range (P_aO₂ >90 mm Hg, P_aCO₂ ∼30 mm Hg, 7.35 <pHa <7.45) by adjustments of tidal volume, and blood glucose was kept between approximately 80 and 120 mg/dL by intravenous infusions of insulin and glucose. 

The pupil was dilated with 1% phenylephrine and 1% atropine. Flurbiprofen ophthalmic solution (0.03%) was applied to block prostaglandin-mediated pupillary constriction. A calibrated Po ₂ microelectrode was inserted into the eye through a 15-gauge needle and was sealed with a system that allowed movement of the electrode. Two 20-gauge needles containing Ag/AgCl wires were inserted into the vitreous for the recording of the vitreal electroretinogram (ERG) and as a reference for the oxygen electrode. 

Double-barreled O₂-sensitive microelectrodes were fabricated as previously described. ²⁹ One barrel was used to record oxygen, and the second, voltage-sensitive, barrel was used to record the intraretinal ERG. 

Intraretinal Po ₂ measurements were made in normal and lesioned areas of each retina. The retina was penetrated in 3-μm steps. The intraretinal ERG was recorded every 30 to 60 μm in response to diffuse light flashes 2.5 seconds in duration and at an illumination just below b-wave saturation. The choroid/retinal boundary was identified when a transepithelial potential across the retinal pigment epithelium (RPE) was recorded (a negative shift in the voltage trace). In lesioned areas of the retina, the transepithelial potential was usually not present, probably because of the damage to the RPE, so the choroid/retinal boundary was considered to be the point at which Po ₂ no longer increased. After reaching the choroid, the electrode was withdrawn from the dark-adapted retina at 2 μm/s, generating a Po ₂ profile across the retina. During withdrawal, the vitreoretinal interface was identified as the location where the ERG recorded by the microelectrode was the same as the vitreal ERG. In the lesioned part of the retina, the ERG was absent, and the end of the profile was determined as described in the Data Analysis section. 

Vitreal and microelectrode voltage signals were recorded with unity gain amplifiers (M4A and M707A; WPI, Sarasota, FL). They were referenced to an Ag/AgCl scalp electrode. Current from the oxygen electrode was recorded with a picoammeter (model 614; Keithley Instruments, Cleveland, OH). The output voltage from the picoammeter was fed through an amplifier to a chart recorder and a computer. 

In three animals, a retinal artery was mechanically occluded at the time of the acute procedure. A 0.7-mm diameter glass probe with a 0.5- to 0.7-mm ball at the tip was pressed on a small superior branch artery, supplying a lesioned part of the retina, where the vessel arose from the optic disc. ³⁰ The occluder was inserted into the eye through a 19-gauge needle, which was held in a manipulator that allowed the positioning of the occluder over the desired vessel. The occluder was advanced with a hydraulic microdrive (model 1207S; David Kopf Instruments, Tujunga, CA). The occlusion was confirmed by visual observation of gaps in the blood in the vessel and by absence of the ERG b-wave in a nonlesioned area. 

Before occlusion, profiles were collected from nonphotocoagulated and lesioned parts of the retina. After the artery was occluded, another set of profiles was obtained from the same area of the lesion. 

Experiments were terminated by injection of pentobarbital (120 mg/kg), and eyes were fixed in 10% formalin or Davidson fixative and embedded in paraffin. Four-micrometer sections were stained with hematoxylin and eosin, visualized with light microscopy, and photographed. The sections were separated by 50 to 100 μm. 

Data Analysis

Histology.

Fundus photographs were used to localize the lesions in histologic sections. All the lesions were analyzed for damage to the retina; in some, the lesion width, thickness of the tapetum, and thickness of the retina were measured. The same parameters were also measured for the nonphotocoagulated retina adjacent to the lesion on the same slide for comparison. On each slide, each parameter was measured three times. Parameters were averaged for each slide, and then the values from all slides were averaged for each lesion. No corrections were made for tissue alterations resulting from processing because paired t-tests were used for statistical analysis. 

Po ₂ Profiles.

The length of the Po ₂ profile was based on two boundaries, the choroidal/retinal boundary and the retinal/vitreal boundary. The choroidal/retinal boundary was designated where the Po ₂ changed during withdrawal from its constant value in the choroid to a steep decrease in the outer retina. The retinal/vitreal boundary was based on the ERG, as described. When the electrode was in the lesioned retina, the intraretinal ERG was absent, and it was not possible to determine the retinal thickness by this method. For those profiles, histologic data were used to determine retinal thickness. The thicknesses of the nonphotocoagulated retina (T_NH) and the lesioned retina (T_LH) were measured. The ratio T_LH/T_NH then described the relative thicknesses of the two regions. To determine the thickness of the lesioned retina in Po ₂ profiles, T_LP, the above ratio was multiplied by the average thickness of the nonphotocoagulated retina determined from the profiles from the normal part of the retina, T_NP, as T_LP = T_NP × T_LH/T_NH. 

Average inner retinal Po ₂, minimum Po ₂, and choroidal Po ₂ were calculated for each profile, and the values for all profiles corresponding to a particular lesion were averaged. The average inner retinal Po ₂ was calculated for the inner 50% of the nonphotocoagulated retina. For photocoagulated retina, the inner retina was more than 50% of the total, but, because of the debris or scar layer, the inner retina was generally less than 100% of the retina. To bracket the value for the actual inner retina, the average Po ₂ was calculated for the inner 50% and for 100% of the lesioned retina. 

Paired t-tests were conducted to compare control and lesioned areas. Differences were considered significant for P < 0.05. Data collected from small and large lesions were tested together and are presented in the same figures. 

Results

Fundus Photographs

Fundus photographs of the lesions were taken immediately after treatment and on the day of the terminal experiment. On the day of treatment, the lesions usually appeared brown in the center, surrounded by a greenish-blue ring (Fig. 1A) . Sometimes they were more uniformly greenish-blue (Fig. 1B) . Brown probably signified more damage. The same energy level produced more damage to regions where the tapetum was green than to regions where it was yellow. 

After healing for 4 weeks, the lesions usually appeared black (Fig. 1) . Sometimes small lesions were shiny and yellow, and large lesions could also contain shiny spots (Fig. 1D) . The appearance of the lesion at this time did not appear to correlate with the extent of histologic damage to the retinal layers because the outer retina was destroyed in shiny and black regions. The shininess might have reflected less complete damage to the tapetum. After healing, the lesions typically appeared smaller in diameter than they appeared immediately after treatment (Fig. 1) . 

Histology

Most of the photocoagulation damage was confined to the outer retina (Fig. 2) , which was destroyed. In some cases, the outer retina was replaced by a fibrous glial scar of variable thickness. This glial scar was usually thicker in smaller lesions (Fig. 2A) . In most cases, regardless of whether there was a glial scar, the inner retina was closer to the choroid than in undamaged retina. The inner retina was primarily preserved, but sometimes the inner nuclear layer was disorganized. Because the outer retina was destroyed, the retinal thickness decreased significantly in the photocoagulated areas (141 ± 33 μm) compared with the nonphotocoagulated areas (192 ± 40 μm; P = 0.00002; n = 20 lesions from 9 cats, paired t-test; Fig. 3A) . Only in cat 345 did it appear that the retina increased in thickness after photocoagulation because the glial scar was thicker in this cat. The absolute retinal thickness might have been distorted by processing, but the relative values are reliable because the measurements for control and lesioned retina were made on the same slide. 

The tapetum was also frequently severely damaged or destroyed in photocoagulated areas (Fig. 2B) . The average tapetal thickness was significantly smaller in the photocoagulated retina (18 ± 14 μm) than in the nonphotocoagulated retina (45 ± 17 μm; P = 0.0005; n = 11 lesions from 7 cats; Fig. 3B ). Because the tapetum was damaged, the choriocapillaris running through it must also have been damaged, though the choriocapillaris could not be distinguished clearly during histologic examination. 

Intraretinal Po ₂

Po ₂ Profiles with Intact Retinal Circulation.

In nonphotocoagulated retina, Po ₂ was highest in the choroid, which is an O₂ source (Fig. 4A) . Oxygen diffuses through the photoreceptor outer segments and reaches a minimum around the photoreceptor inner segments, which are the region of highest O₂ consumption. Frequently, O₂ also diffuses toward the outer retina from the inner retina, which in general supplies approximately 10% of the oxygen for the photoreceptors in the dark-adapted retina. ³¹ Consumption in both the photoreceptor outer segments and the outer nuclear layer is negligible. ¹³ As a result, the Po ₂ profile is a straight line through both these layers. In the inner retina, local Po ₂ maxima represent the retinal circulation. 

A range of profile shapes was observed in the center of the lesion (Fig. 4B) . Some profiles, like those from cat 362, had a slope across the retina from the choroid to the vitreous, but the slope was smaller than that in the control profiles. The minimum, if there was one in the retina, was further from the choroid. Other profiles, such as those from cat 364, were relatively flat and did not show an oxygen gradient from the choroid across the retina. In all the profiles from lesioned retina, the minimum Po ₂ was higher. 

Several parameters were obtained from the profiles and compared across lesioned and normal retina. The average Po ₂ was calculated for the inner 50% of normal retina and for both 100% and the inner 50% of the lesioned retina (Fig. 5) . Based on histologic examination, the thickness of the inner retina in the lesion typically fell between those two values. 

In most cats there was a higher average inner retinal Po ₂ in lesioned retina. Sometimes this increase was very small, and in two cats there was a slight decrease in Po ₂. The average Po ₂ for the inner 50% of the retina was 14 ± 7 mm Hg for nonphotocoagulated tissue and was significantly higher, 22 ± 10 mm Hg, for lesioned tissue (P = 0.006; n = 13 cats). The average Po ₂ across 100% of the lesioned retina (25 ± 10 mm Hg) was also higher than that in the inner half of the normal retina (P = 0.004). 

The minimum Po ₂ for each animal, averaged across profiles, is shown in Figure 6 . In untreated retina, the minimum was low because of the high oxygen consumption rate of the inner segment layer. In the lesioned part of the retina, the photoreceptors were destroyed, so the minimum was significantly higher (17 ± 8 mm Hg) than in the untreated part of the retina (3 ± 4 mm Hg; P ≪ 0.001; n = 13 cats). 

Choroidal Po ₂ was also analyzed because in most lesions the tapetum was damaged, suggesting that the choriocapillaris, which lies between the tapetum and the RPE, was probably also damaged. Choroidal Po ₂ decreased significantly in the photocoagulated part of the retina (40 ± 16 mm Hg) compared with the untreated part of the retina (46 ± 14 mm Hg; P = 0.007; n = 13 cats). On average, choroidal Po ₂ from the photocoagulated part of the retina was 84.92% of choroidal Po ₂ from the untreated part of the retina (Fig. 7) . Choroidal Po ₂ varied from cat to cat even in control areas, but the variability was lower within individual cats, as can be observed from the error bars. 

Po ₂ at the vitreal/retinal boundary and in the preretinal vitreous 100 μm away from the retina was also analyzed from the profiles because vitreal measurements have been made in some previous studies. There was no significant difference between the photocoagulated and control regions at the vitreal surface (P = 0.09) or in the preretinal vitreous (P = 0.31). For all the cats, the O₂ gradient was out of the vitreous into the inner retina in the control retina. This was expected because our measurements were made in intravascular regions. ³² In the retina close to arterioles, O₂ flowed from the retina to the vitreous, but it circled back into the retina from the vitreous in parts of the retina further away from arterioles. The Po ₂ gradient was from the vitreous to the inner retina in 6 of 12 cats for the photocoagulated retina. This meant that for the photocoagulated retina, vitreal Po ₂ tended to be less than inner retinal Po ₂, and that for the control retina, vitreal Po ₂ tended to be greater than inner retinal Po ₂. Therefore, the difference between photocoagulated and control parts of the retina was minimized when recordings were made in the vitreous. 

Po ₂ Profiles with Occluded Retinal Circulation.

In clinically advanced diabetic retinopathy requiring PRP, widespread vascular damage can lead to significant capillary dropout. ^{33 34} We tried to mimic the clinical situation of reduced inner retinal circulation by investigating the more severe case of occluding a retinal artery that supplied part of the retina with a photocoagulation lesion. 

We attempted to make measurements in the lesion after arterial occlusion in three animals and were successful in two. We had to produce lesions in an area of the retina supplied by a single small retinal artery that was not paired with a vein at the optic disc so that we could cut off the retinal circulation without producing the hemorrhages that usually occur with a venous occlusion. This was difficult to achieve because the artery decreased in diameter after placement of lesions, presumably because the metabolic needs of the lesioned tissue were supplied to a greater extent by the choroid and because the blood flow through this retinal arteriole was reduced. 

Figure 8shows profiles from different parts of the retina with and without occluded circulation. The profiles from the lesion without retinal circulation had a linear slope at the beginning of the profile, probably corresponding to the debris layer. They also had more curvature than the profiles collected from the lesioned part of the retina with intact retinal circulation. Diffusion theory predicts that this curvature corresponds to the consumption of oxygen. This was not visible in the profiles with intact retinal circulation because the supply and consumption of oxygen overlapped in space. The minimum Po ₂ in the profiles from the lesion without retinal circulation was lower than in the profiles from the lesioned part of the retina with intact circulation, but it was higher than 0 mm Hg, the value that has been observed previously in nonphotocoagulated retina with an occluded circulation. ³⁰  

Figure 9shows the average Po ₂ for the inner 50% of the retina in the two animals with vascular occlusion. The average Po ₂ in the lesion profiles after occlusion was higher (10 ± 3 mm Hg; n = 2 cats) than the Po ₂ in control profiles (7 ± 3 mm Hg), but it was lower than the average for the profiles collected in the lesion with intact retinal circulation (22 ± 0 mm Hg). 

When the retinal circulation is occluded, oxygen consumption of the inner retina can be determined by fitting a mathematical model to the data. ³⁵ For this study, where the photoreceptors were replaced by scar tissue, we used a two-layer model in which the more distal part of the profile, presumed to be scar tissue, did not consume oxygen but in which the more proximal part of the profile, the inner retina, did consume oxygen. The model was fitted to obtain the oxygen consumption of the proximal layer and it determined the boundary between the nonconsuming layer and the consuming layer. This two-layer model has previously been used to fit data at the end stage of photoreceptor degeneration, when the retinal circulation is almost completely attenuated. ³⁶ We found that inner retinal O₂ consumption in the lesioned part of the retina during retinal artery occlusion was 0.9 ± 0.3 mL · 100g⁻¹ · min⁻¹ (n = 5 profiles) in one cat and 0.5 ± 0.5 mL · 100g⁻¹ · min⁻¹ (n = 3 profiles) in the other. The values reported have been increased by 30% from those obtained from the model to account for the tendency of the retina to appear thicker during electrode withdrawal than electrode penetration. ³⁵ This difference artifactually lowers the value of inner retinal O₂ consumption. The 30% correction is approximate in this case and is based on previous work ³⁵ ; an exact correction could not be made because there was no ERG to definitively mark the location of the vitreal-retinal border. 

Discussion

PRP is the most widely used treatment for diabetic retinopathy. At the same time, it is a destructive procedure, and its mechanism is still unclear. The goal of this study was to investigate the mechanism of PRP in a cat model. The results are consistent with the hypothesis that PRP increases intraretinal Po ₂ in patients with diabetes. ^{8 9 24 25} Increased supply of O₂ from the choroid would be expected to contribute to the attenuation of intraretinal neovascularization by relieving inner retinal hypoxia, thought to exist in patients with late-stage diabetic retinopathy. ³⁷ Retinal hypoxia is also implicated by studies that show an improvement in contrast sensitivity ³⁸ and a reduction of macular edema ³⁹ in patients with diabetes during inspiration of elevated Po ₂, which presumably relieves tissue hypoxia. This study shows that intraretinal Po ₂ in a healed lesion is greater than in control areas in a vascularized retina in animals breathing air. The intraretinal data are consistent with previous studies showing increased Po ₂ in the preretinal vitreous of rabbits ^{20 21} and miniature pigs ^{24 25} and increased Po ₂ intraretinally immediately after photocoagulation in rabbit retinas. ²⁶ Neither we nor previous investigators ^{8 18 19} could detect the increased Po ₂ in preretinal recordings from cats breathing air, which suggests that little, if any, choroidal oxygen diffuses into the vitreous and then back to the adjacent retina. Increased retinal oxygenation is also consistent with studies showing decreased retinal blood flow velocities after PRP. ^{40 41 42 43 44} It was hypothesized that the decrease in retinal blood flow after PRP was a response to the relief of inner retinal hypoxia by oxygen diffusing from the choroid, but there had been no direct evidence that Po ₂ actually increased. Our qualitative observation of a decrease in the diameter of the arteriole supplying the lesion also supports this. 

Results from the measurements during occlusion provide further support for the hypothesis that PRP relieves inner retinal hypoxia because in lesioned retina, the choroidal supply alone was able to keep average inner retinal Po ₂ at approximately 10 mm Hg. The presence of a lesion is required for this. Without a lesion, the inner retinal Po ₂ is essentially zero after occlusion. ³⁰  

Previous studies, in which intraretinal Po ₂ ^{18 45} or preretinal Po ₂ ^{8 18 45} was measured in monkeys and cats, found increased Po ₂ over the photocoagulation lesion only when the animals were breathing 100% O₂ and no difference between lesion and control under air breathing. There are several possible explanations for the difference in these studies and our findings. First, the preretinal and retinal Po ₂ vary across the retina and preretinal vitreous. The measurements reported or used for statistics in previous studies were made at a single point within the retina ⁴⁵ or vitreous, whereas our inner retinal values are averages across the whole thickness of the inner retina. Second, preretinal O₂ comes from the retinal circulation, but it reflects oxygenation of the innermost part of the retina, whereas our profiles and the values used for statistical analysis are spatial averages across the retina. The effect of choroidal oxygenation when an animal is breathing air should be largest in the deeper part of the inner retina and smallest at the vitreous. As shown in the results, measurements in the preretinal vitreous in our data did not show significant differences between control and photocoagulated retina. Third, the studies cited were not always successful in recording data from photocoagulated and nonphotocoagulated regions of the retina within the same animal; hence, the number of animals was small. ^{8 45} In one study, the number of measurements rather than the number of animals was used for statistical analysis, ¹⁸ whereas all our statistics were calculated using paired t-tests. Features of the present study allowed us to detect differences that might have been present in earlier studies but that were obscured by the measurement techniques. 

Histologic analysis showed that photocoagulation damaged the photoreceptors and often decreased the distance between the choroid and the inner retina. Both factors contributed to improved oxygen delivery to the inner retina. It is possible that a further increase in inner retinal oxygenation resulted from decreased inner retinal metabolism after the lesion The inner retinal oxygen consumption in two cats for which it was possible to measure this was <1 mL · 100g⁻¹ · min⁻¹, which was smaller than that the average value of 3.5 ± 1.7 mL · 100g⁻¹ · min⁻¹ reported for nonphotocoagulated retina with occluded retinal artery. ³⁵ This is suggestive of damage to the inner retina, but that conclusion is tenuous for two reasons. First, the number of animals in which modeling could be performed was small (n = 2). Second, the statistics are further weakened because we were unable to collect profiles from a region with occlusion but without lesions in the same retinas; therefore, the comparison is with data obtained previously. Based on evidence in models of photoreceptor degeneration, we would not have expected the metabolism of the inner retina to change solely because the photoreceptors were missing. ^{36 46 47} However, the loss of photoreceptors in a retinal degeneration model and the destruction of the photoreceptors with a laser, which may cause some damage or disorganization of the inner retina, are different situations. If the values obtained here are indeed lower than inner retinal oxygen consumption in the normal retina, it is likely that the lesioned human retina is similar. No other information is available about any of the functional aspects of the inner retina in humans with PRP or in animal models of PRP. 

Another important finding of this study was that photocoagulation consistently decreased choroidal Po ₂ in the lesion. Although one case study reported damage to the choroid with PRP, ⁴⁸ in general little attention has been paid to the effect of PRP on the choroid. This is surprising because after capillary loss in diabetic retinopathy, the choroid must be the main source of O₂. The damage to the choriocapillaris could be inferred from the histologic sections, where the tapetum, through which the choriocapillaris runs, was damaged. Investigation of the effect of photocoagulation on choroidal blood flow by means of indocyanine green dye and microsphere imaging in a separate study (Linsenmeier RA, et al. IOVS 2007;48:ARVO E-Abstract 4167) revealed directly that photocoagulation severely damaged the choriocapillaris. The loss of choriocapillaris in that study was so great that we think the oxygen in the lesioned retina must have come, at least partially, from the larger vessels of the choroid, which remained intact. 

Although the cat area centralis is a good model of human peripheral retina, a few important differences might affect the retinal absorbance of laser energy. The cat has a tapetum, and the retinal pigment epithelium (RPE) in this region is nonpigmented, whereas the human retina has no tapetum and a pigmented RPE, which is thought to absorb the energy delivered by the laser. ^{49 50} In cat, this energy is not absorbed by the nonpigmented RPE, but it is probably absorbed by the choroidal melanocytes or by the photoreceptors. For the same laser energy, this might result in different damage to the tissue and different intraretinal conditions, but we have reported data only for lesions in which the outer retina was destroyed and the inner retina was largely intact; this is the same histologic appearance as in human PRP. Because of the differences in cat and human retina, it is possible that the same retinal lesions in cat and human would affect the choriocapillaris differently, but we do not believe this is likely. If the RPE is the major absorber in human retina, it seems likely that the adjacent choriocapillaris is damaged in humans as well. 

Most of the results presented here were from measurements made in lesions much larger than those produced in clinical PRP. It is possible that lesion size affects intraretinal Po ₂ by producing different damage to the tissue; however, we made some measurements in smaller lesions that were consistent with the findings in larger lesions, except for cat 345. In this cat, however, histologic analysis showed that the inner retina was further away from the choroid than in most other animals because of the presence of a thick glial scar (Fig. 2A) , which decreased the amount of oxygen delivered to the inner retina. 

In this study we focused on inner retinal oxygenation directly over the lesion. If one conceives of a lesioned region as being a cylinder, with the laser-destroyed photoreceptors at one end of the cylinder and a relatively undamaged inner retina at the other end, it can be seen that there is increased oxygenation within the cylinder. An important question is whether there is significant diffusion of oxygen out of this cylinder that might increase oxygenation in adjacent, undamaged tissue. We could not address this question experimentally because it was difficult to locate the electrode precisely at the edge of a lesion. Such measurements would have been difficult to interpret in any case because they provide only one-dimensional information about O₂ gradients, whereas at the edge of the lesion the O₂ gradients are expected to be two dimensional. We addressed the question of O₂ diffusing out of the lesioned region by mathematical simulations that will be reported separately. However, modeling suggests that the beneficial effect of the lesion extends for only approximately 70 μm lateral to the photocoagulated retina. ⁵¹  

This study shows that PRP increases Po ₂ levels in the retina, but it does not exclude the possibility that some of the benefit of PRP results from the destruction of cells producing angiogenic factors. It is possible that destruction of those cells combined with increased intraretinal Po ₂ is responsible for the effectiveness of the treatment. Increased Po ₂ should decrease production of angiogenic factors. Additionally, no studies have addressed what happens to substrates or metabolites other than oxygen after PRP. Delivery of other substrates and clearance of waste is likely to be impaired in diabetic retinopathy and could be improved after PRP by bringing the inner retina closer to the choroidal circulation. 

 Supported by National Institutes of Health Grant R01EY05034 (RAL), and an unrestricted grant to Northwestern University from Research to Prevent Blindness.

 Submitted for publication January 20, 2007; revised April 19, August 10, and September 6, 2007; accepted November 20, 2007.

 Disclosure: E. Budzynski, None; J.H. Smith, None; P. Bryar, None; G. Birol, None; R.A. Linsenmeier, None

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

 

Figure 1.

 

Fundus photographs of lesions in cat retina. (A, B) Lesions on the day of treatment. (C, D) Healed lesions 5 and 6 weeks after lesion placement. (A, C) Cat 351. (B, D) Cat 378.

Figure 2.

 

Histologic sections of photocoagulated retina stained with hematoxylin and eosin. (A) Section from a small lesion approximately 250 μm in diameter (cat 375, slide L4). The photoreceptors are absent and the inner nuclear layer is disorganized, but much of the inner retina is preserved. The overall thickness of the retina is similar in the lesioned and normal areas. There is mild disruption of the innermost part of the tapetum, which is the laminar structure between the retina and the pigmented choroid. (B) Section from large confluent lesion of the type shown in Figure 1 . The photomicrograph shows approximately half the extent of the lesion, which continues to the left. The lesion ends near the right side of the micrograph, where the outer nuclear layer appears, though at this location it is still reduced in thickness compared with control retina. The inner retina is preserved with only mild disruption of its architecture, and the tapetum was destroyed (cat 366, slide L4). Both micrographs were taken at a total magnification of 100× and are reproduced at equal magnification.

Figure 3.

 

(A) Average retinal thickness of 20 lesions in 11 cats. (B) Average tapetal thickness of 11 lesions in 9 cats. Each symbol represents an average from a single lesion or a control area from the same histologic section close to the lesion. Open symbols: small lesions; closed symbols: large lesions. Values to the left and right are mean ± SD.

Figure 4.

 

Intraretinal Po ₂ profiles. One hundred percent depth represents the choroidal-retinal boundary, and 0% depth represents the vitreal-retinal boundary. (A) Control profile from nonphotocoagulated retina of cat 345 with a schematic of the retina (top). (B) Examples of profiles from photocoagulated retinas.

Figure 5.

 

Average intraretinal Po ₂ for each cat in control and lesioned parts of the retina. Error bars represent SD across all the profiles of one type in each cat. Values for lesioned parts of the retina were computed for the inner half of the retina and for the entire retina. Inset: each pair of points connected by a line represents average values for one cat. Averages for control represent 50% of the retina; averages for the lesion represent 100% of the retina. Open symbols: small lesions; closed symbols: large lesions; inset, symbols: averages ± SD for all cats.

Figure 6.

 

Minimum intraretinal Po ₂ averaged across profiles for each cat for control and lesioned retina. Open symbols: small lesions; closed symbols: large lesions. Values to the left and right are mean ± SD for all cats.

Figure 7.

 

Choroidal Po ₂ for profiles from control and lesioned parts of the retina, averaged for each cat. Error bars represent SD. Inset: percentage change in choroidal Po ₂. Each point represents the average for an individual cat. Open symbols: small lesions; closed symbols: large lesions.

Figure 8.

 

Examples of Po ₂ profiles from normal retina, lesioned retina, and lesioned retina with an occluded circulation. One hundred percent depth represents the choroidal-retinal boundary, and 0% depth represents the vitreal-retinal boundary. Profiles on the left are from cat 376, and those on the right are from cat 380.

Figure 9.

 

Average inner retinal Po ₂ for control areas, lesioned areas, and lesioned areas after arterial occlusion, for each of two cats. Averages were calculated over the inner 50% of the retina. Error bars represent SD of average inner retinal Po ₂ across profiles in that animal.

The authors thank Christina Enroth-Cugell, Shufan Wang, and Norbert Wangsa-Wirawan for their assistance during the experiments, Lissa Padnick-Silver for establishing the collaboration with Evanston Hospital and assistance during experiments, Lori Gubernat for histologic preparations, and Yun Kim for technical assistance during the experiments.