Investigative Ophthalmology & Visual Science, Vol. 31, No. 2, February 1990 Copyright © Association for Research in Vision and Ophthalmology

Vitrectomy Prevents Retinal Hypoxia in Branch Retinal Vein Occlusion Einar Srefansson, Roger L. Novack, and Diane L. Harchell Vitrectomy has been shown to halt diabetic retinal neovascularization, but the mechanism of this process is unknown. We propose that vitrectomy improves the oxygen supply to ischemic inner retina by way of fluid currents in the vitreous cavity. In order to test this hypothesis, we induced branch retinal vein occlusion in cats and measured preretinal oxygen tension before and after branch retinal vein occlusion in ten nonvitrectomized and five vitrectomized eyes. Branch retinal vein occlusion caused a significant decrease in preretinal oxygen tension in nonvitrectomized eyes, in which the oxygen tension fell from 20 ± 7 to 6 ± 5 mmHg (P = 0.001). Conversely, in vitrectomized eyes the oxygen tension was not significantly reduced after branch retinal vein occlusion. The data demonstrate that branch retinal vein occlusion causes retinal hypoxia in nonvitrectomized eyes, whereas after vitrectomy the hypoxic effect of branch retinal vein occlusion is reduced. The relief of retinal hypoxia that follows vitrectomy may be responsible for halting retinal neovascularization after vitrectomy in diabetic patients. Invest Ophthalmol Vis Sci 31:284-289, 1990

fellow eyes before and after branch retinal vein occlusion. In five additional cats the preretinal oxygen tension was measured in normal eyes before and after branch retinal vein occlusion. The cats (2-4 kg, and of either sex) were anesthetized with intramuscular ketamine hydrochloride (20-30 mg/kg) and acepromazine maleate (2-3 mg/kg), repeated as needed. The right pupil of each animal was dilated with topically applied 0.25% tropicamide and 5% Phenylephrine, and the eye prepared for surgery under sterile conditions. A lid speculum was placed; a 180° conjunctival peritomy was performed temporally; and two sclerotomies were made in the pars plana area, supero- and inferotemporally, respectively. A cannula was sutured in place in the inferotemporal sclerotomy for irrigation with lactated Ringer's solution. The vitrectomy instrument (Minivisc, OMS) was placed through the superotemporal sclerotomy into the vitreous cavity under direct observation through the operating microscope, using the coaxial light and a flat corneal contact lens. A pars plana vitrectomy was performed where the vitreous gel was completely removed except for the peripheral vitreous gel adjacent to the vitreous base. The sclerotomies were closed with 6-0 Dexon, the conjunctiva sutured, and 10 mg Gentamicin injected subconjunctivally. The cats were observed for 2-4 weeks after vitrectomy, prior to oxygen tension measurement.

Blankenship and Machemer' studied the long term results of vitrectomy in patients with proliferative diabetic retinopathy. They found that retinal fibrovascular proliferations usually do not progress after vitrectomy, and so they raised the question: What is the mechanism by which vitrectomy affects proliferative diabetic retinopathy? We wanted to pursue this question, and hypothesized that fluid currents in the vitreous cavity after vitrectomy could deliver oxygen and other nutrients from well oxygenated (perfused) areas to ischemic (nonperfused) areas in the retina. This new oxygen distribution relieves retinal hypoxia, reducing the hypoxic stimulus to fibrovascular proliferation. In order to test this hypothesis, preretinal oxygen tension was measured in vitrectomized and nonvitrectomized cat eyes before and after branch retinal vein occlusion induced by transvitreal diathermy.

Materials and Methods In five cats the preretinal oxygen tension was measured simultaneously in normal and vitrectomized From The Duke University Eye Center and Durham Veterans Administration Medical Center, Durham, North Carolina, and The University oflceland, Reykjavik, Iceland. Supported in part by The Veterans Administration (Career Development Award (ES) and Medical Research Funds) and by Research to Prevent Blindness, Inc., The American Federation for Aging Research, Inc., and the National Eye Institute (grants EY-07001 andEY05722). Submitted for publication: April 7, 1989; accepted July 14, 1989. Reprint requests: Einar Stefansson, MD, PhD, University oflceland, Landakotsspitali, Reykjavik, Iceland.

Oxygen Tension Measurements Oxygen tension measurements were performed with polarographic oxygen electrodes.2 The cats were ORA

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933576/ on 01/24/2017

VITRECTOMY AND RETINAL HYPOXIA / Srefonsson er ol

No. 2

BRVO No Vitrectomy

Vitrecfomy

Fig. 1. The experimental set-up. The polarographic oxygen electrodes are placed transvitreally in both eyes to measure preretinal oxygen tension. The oxygen tension was measured simultaneously in the vitrectomized eye and the nonvitrectomized fellow eye, before and after branch retinal vein occlusion.

anesthetized with intramuscular ketamine hydrochloride (20-30 mg/kg) and acepromazine maleate (2-3 mg/kg) and maintained under general anesthesia with alpha-chloralose (80 mg/kg), repeated as needed. A catheter was placed in one femoral artery for continuous measurement of arterial blood pressure and measurement of arterial blood oxygen tension, carbon dioxide tension, pH, and hematocrit. The blood gases were measured with a blood gas analyzer (model 1302; Instrumentation Laboratories, MA). The cat was incubated and ventilated mechanically, and the ventilation was adjusted to keep arterial blood carbon dioxide tension between 25 and 35 mmHg. The cats normally were ventilated with 21 % oxygen and 79% in nitrogen, and in some cases with 100% oxygen. The pupils were dilated with 5% phenylephrine and 0.25% tropicamide. The cat was placed in a stereotaxic head-holder. In both eyes the sclera was exposed temporally by a lateral canthotomy and conjunctival peritomy. A sclerotomy was performed in the pars plana, approximately 3.5 mm behind the limbus, with a myringotomy blade. A Teflon cannula was placed in the sclerotomy, and the polarographic oxygen electrode (model 760; Diamond Electrotech, Ann Arbor, MI) was advanced into the vitreous cavity. The oxygen tension was measured in both simultaneously in both eyes of each cat, 0.1 mm over the retinal area drained by the superior vein (Fig. 1). A silver-silver chloride reference electrode was placed subcutaneously. The entry sites were watertight, and no fluid was infused into the eye. All eyes maintained normal shape throughout the experiments. The intraocular pressure was not measured, but all eyes were soft to the touch. After the preretinal oxygen tension had been measured for at least 20 min with polarographic electrodes, these electrodes were withdrawn for 3-5 min. A bipolar radio frequency diathermy probe was intro-

285

duced through the sclerotomy and advanced to the superior retinal vein under visual control through the operating microscope and a flat contact lens. Diathermy was applied to the vein for approximately 5 sec. The application was repeated if necessary. The vein was seen to constrict in the area of the burn and dilate distal to that area, and the blood column became interrupted and immobile. In selected cases, fluorescein angiography was performed by injecting 10% fluorescein sodium intravenously (0.1 ml/kg) and to taking photographs through appropriate filters of a fundus camera (Fig. 2). Once the branch retinal vein occlusion had been created, the polarographic electrodes were reintroduced into the eyes, and preretinal oxygen tension was measured over the ischemic retina in both eyes. Electrodes were calibrated before and after each experiment with pure N 2 , 5% O2/95% N 2 , and 21% O2/79% N2 in a calibration cell at 37°C (calibration cell model 1251, Diamond Electrotech). The difference in calibration before and after each experiment was less than 20%. The oxygen electrodes were advanced to the preretinal vitreous in the area drained by a superior vein in each eye. The electrodes were placed approximately 0.1 mm above the retina, away from any visible retinal vessels. The electrodes were advanced until a subtle concave mirror effect on the retina could be seen through the operating microscope. The electrodes were then withdrawn approximately 0.1 mm. The preretinal oxygen tension was measured simultaneously in both eyes of each cat. The light intensity originating from white fluorescent ceiling lights during the measurements was 35 footcandles measured at the cornea level (Photometer I; Quantum Instruments, Garden City, NY).

Fig. 2. Afluoresceinangiogram showing the fundus of a cat after branch retinal vein occlusion induced with transvitreal diathermy. The superior retinal vein is occluded and is not filled with fluorescein. The angiogram is in the arteriovenous phase.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933576/ on 01/24/2017

286

21%0 2 \i00°%\ 21%O2

Breathing Mixture

lOOr

Electrode moved to Non-occluded Retina

Branch Retinal Vein Occlusion

80

60

40

I

Vol. 31

INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / Februory 1990

20

20

40

60

80

Fig. 3. An experimental tracing showing time versus preretinal oxygen tension. The breathing mixture is indicated above the tracing. A stable oxygen tension baseline was established, and the superior branch retinal venule subsequently was occluded. The oxygen tension fell rapidly. Breathing 100% oxygen raises the preretinal oxygen tension to 100 mmHg. At 80 min the electrode was moved to an intact area of the same retina, where the oxygen tension was only slightly lower than at the initial baseline.

Time (min)

The experimental animals were handled in accordance with the ARVO Resolution on the Use of Animals in Research. In addition to the ten cats reported, three cats had to be excluded from the study due to vitreous hemorrhages. Two of these cats had undergone vitrectomy. Results In nonvitrectomized eyes the preretinal oxygen tension fell 14 ± 7 mmHg (mean ± 1 SD, n = 10) when the branch retinal vein was occluded. A significantly smaller change was seen in the vitrectomized eyes, in which the oxygen tension fell only 3 ± 2 mmHg (n = 5, P = 0.004) when the branch retinal vein was occluded. The preretinal oxygen tension in normal nonvitrectomized eyes was 20 ± 7 mmHg (mean ± 1 SD, n = 10). After branch retinal vein occlusion, the preretinal oxygen tension fell to 6 ± 5 mmHg (n = 10, P = 0.001; paired student t-test) in the nonvitrectomized eyes (Fig. 3). The preretinal oxygen tension in vitrectomized eyes was 19 ± 11 mmHg (n = 5). The oxygen tension did not fall significantly after branch retinal vein occlusion in the vitrectomized eyes and remained at 16 ± 11 mmHg (n = 5) (Fig. 4). The preretinal oxygen tension in normal nonvitrectomized eyes and vitrectomized eyes is similar prior to the inducing of branch retinal vein occlusion. After branch retinal vein occlusion, the preretinal oxygen tension in nonvitrectomized eyes (6 ± 5 mmHg) is significantly lower than in vitrectomized eyes (16 ± 11 mmHg, P = 0.04; unpaired student t-test) (Fig. 5). Pulling the electrode back from the

retinal surface into the vitreous cavity did not change the measured oxygen tension in vitrectomized eyes. In contrast, in nonvitrectomized eyes a shallow oxygen gradient existed, sloping from the retina into the vitreous cavity. Arterial blood oxygen tension was 114 ± 18 mmHg; carbon dioxide tension was 26 ± 7 mmHg;pH was7.45 ±0.12; hematocritwas24 ±6%; mean arterial blood pressure was 100 ± 17 mmHg; and the body temperature was 26.6 ± 1.2°C. Discussion The data demonstrate that in nonvitrectomized cat eyes, branch retinal vein occlusion causes hypoxia at the inner retinal surface in the area of the occluded vessel. Conversely, in vitrectomized eyes, branch retinal vein occlusion does not cause a significant decrease in preretinal oxygen tension. The preretinal oxygen tension was significantly higher in the vitrectomized eyes than in the nonvitrectomized eyes after branch retinal vein occlusion had been induced in both groups. This indicates that vitrectomy reduced the hypoxic effect of branch retinal vein occlusion. After vitrectomy, oxygen enters the fluid in the vitreous cavity from arterial blood in the ciliary processes and from well perfused areas of the retina. We hypothesize that the fluid in the vitreous cavity of a vitrectomized eye circulates freely and thus moves dissolved oxygen from well perfused to nonperfused hypoxic areas of the retina (Fig. 6). Since oxygen tension gradients could not be measured in the vitrectomized vitreous cavity, this indicates that mixing is almost complete. In contrast, the vitreous gel of a nonvitrectomized eye prevents major fluid circulation, and thus the oxygen moves mainly by diffu-

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933576/ on 01/24/2017

287

VITRECTOMY AND RETINAL HYPOXIA / Srefonsson er ol

No. 2

2 1 % O,

100%

80

6

HU

•I £

20

90

135

180

Time (Min)

NORMAL

VITRECT

BEFORE

BRVO

NORMAL AFTER

VITRECT BRVO

Fig. 5. Time versus the preretinal oxygen tension (mmHg) in a vitrectomized eye (interrupted line) and in a nonvitrectomized eye before and after the induction of a branch retinal vein occlusion in each eye. The graph represents simultaneous recordings from the two eyes of the same cat by means of polarographic electrodes. The oxygen percentage in the inspired air is shown at the top.

40 -

D >

z

30 -

UJ

z

111

O

8 9

20-

o

o o o 0

B Q

8

B

D

UJ

8

cr UJ

cc o.

The oxygen hypothesis for the effect of vitrectomy on proliferative diabetic retinopathy complements our earlier theory regarding the mechanism of the effect of panretinal photocoagulation on proliferative

(CO

10-

D

NORMAL VITRECT BEFORE BRVO

•

NORMAL VITRECT AFTER BRVO

Fig. 4. Histogram (A) and scatterplot (B) showing preretinal oxygen tension (mmHg) in vitrectomized and nonvitrectomized eyes before and after branch retinal vein occlusion (BRVO). In (A), the mean and standard error of the mean (segment above bar) are indicated. The oxygen tension in the nonvitrectomized eyes after branch retinal vein occlusion was significantly lower than the value before branch retinal vein occlusion. The branch retinal vein occlusion did not result in significant hypoxia in the vitrectomized eyes.

sion.3 This process is too slow to supply the hypoxic areas of retina with enough oxygen. The finding that vitrectomy improved oxygen supply to ischemic areas of the retina suggests an explanation for the inhibitory effect vitrectomy has on retinal neovascularization.1 If we accept the hypothesis that the hypoxic retina stimulates retinal neovascularization,4 it follows that improved oxygenation and relief of the hypoxia would halt the neovascular process.

Fig. 6. Retinal ischemia, showing fluid fluxes in the vitrectomized eye and the diffusion of oxygen from the ciliary body and the well perfused areas of the retina and into the vitreous cavity fluid. Oxygen-rich fluid flows by the ischemic areas of the retina, and oxygen diffuses from the fluid into the ischemic retina to relieve the tissue hypoxia.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933576/ on 01/24/2017

288

INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / Februory 1990

Occluded blood vessels

Vitreous

Retina

Choroid

Fig. 7. Oxygen delivery to the ischemic hypoxic retina after panretinal photocoagulation or vitrectomy. After vitrectomy, oxygen diffuses from the fluid in the vitreous cavity and into the retina. Oxygen diffuses through the photocoagulation scar in the outer retina and into the ischemic inter retina. Both treatment modalities improve the oxygen supply to the inner retina. PE, retinal pigment epithelium.

diabetic retinopathy; this theory stated that after panretinal photocoagulation, the ischemic inner retina receives oxygen from the choroid (Fig. 7).2'5 Both treatment modalities are known to halt the progress of proliferative diabetic retinopathy.1'6 We propose that improved oxygenation of ischemic and hypoxic retina is the common mechanism of action for both vitrectomy and panretinal photocoagulation. While retinal neovascularization is uncommon in the posterior pole after vitrectomy,1 neovascularization may develop from the peripheral retina, which remains covered with remnants of anterior hyaloid vitreous gel not removed during the vitrectomy. The observation of anterior hyaloidal fibrovascular proliferation7'8 offers additional support for our theory. After vitrectomy, retinal neovascularization arrests in the posterior pole. However, the peripheral retina is still covered with vitreous gel and receives too little oxygen from the vitreous cavity to relieve its ischemic hypoxia. It has been observed that the eye with total posterior vitreous detachment rarely develops proliferative diabetic retinopathy.9"" Histopathologic studies of the vitreoretinal relationship in proliferative diabetic retinopathy by Tagawa and associates910 and by Foos and associates" showed that they had either no or partial posterior vitreous detachments. This indicates

Vol. 31

that new vessels grew onto the vitreous gel while it was attached to the retina. Our theory of oxygen delivery from the vitreous cavity offers an explanation for this finding. After a posterior vitreous detachment, a fluid-filled compartment is present in front of the retina. As this fluid circulates, it can carry oxygen in dissolved state from well perfused to poorly perfused retinal areas and thus relieve hypoxia and suppress the neovascular stimulation. Our oxygenation theory is not the only possible explanation for the inhibition of preretinal neovascularization after vitrectomy. It has been suggested that the neovascularization stops because the vessels have no "scaffold" to grow on. This theory has survived because it has never before been challenged by an alternative theory. It seems to be based only on the clinical fact that absence of the vitreous gel inhibits neovascularization. However, ocular neovascularization can grow on surfaces other than the vitreous humor. Iris neovascularization grows on the surface of the iris, exposed to fluids similar to those that fill the vitreous cavity after vitrectomy. In vitrectomized diabetic eyes with silicone oil in the vitreous cavity, we have observed preretinal neovascularization on the surface of the retina. Similarly, in proliferative vitreoretinopathy, fibrous proliferation uses the retinal surface as scaffold, in the absence of vitreous. Ernest and Archer12 measured preretinal oxygen tension in monkeys 30 min and 3-6 months after branch retinal vein occlusion. They did not find the retina to be hypoxic. However, reperfusion and tissue atrophy may have raised the preretinal oxygen tension in the months after the vein occlusion. Pournaras et al13 produced branch retinal vein occlusion in miniature pigs. The pigs developed retinal hypoxia, a result that agrees with our findings in the cat. The preretinal oxygen tension levels in the cats in the current study agree well with previously reported values.3'5 The thesis of supplying nutrients and oxygen to the retina from the vitreous cavity may also have a role in acute retinal ischemia, perhaps by oxygenation of the vitreous14 or by vitreoperfusion.15 Over 30 years ago, Wise suggested that retinal hypoxia may stimulate neovascularization in diabetic and other proliferative retinopathies.4 Since that time, panretinal photocoagulation and vitrectomy have been found empirically to suppress retinal diabetic neovascularization. We propose that both treatment modalities affect retinal neovascularization by relieving retinal hypoxia (Fig. 7). This conclusion is a direct extension of Wise's Theory. Key words: diabetic retinopathy, vitrectomy, retinal vein occlusion, oxygen, neovascularization

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933576/ on 01/24/2017

No. 2

VITRECTOMY AND RETINAL HYPOXIA / Srefonsson er ol

References 1. Blankenship GW and Machemer R: Long-term diabetic vitrectomy results: Report of 10 years' follow-up. Ophthalmology 92:503, 1985. 2. Stefansson E, Landers MB, III, and Wolbarsht ML: Increased retinal oxygen supply following pan retinal photocoagulation and vitrectomy and lensectomy. Trans Am Ophthalmol Soc L XXXIX: 307-334, 1981. 3. Aim A and Bill A: The oxygen supply to the retina: I. Effects of changes in intraocular and arterial blood pressures and in arterial PO2 and PCO2 on the oxygen tension in the vitreous body of the cat. Acta Physiol Scand 84:261, 1972. 4. Wise GN: Retinal neovascularization. Trans Am Ophthalmol Soc 54:729, 1956. 5. Stefansson E, Hatchell DL, Fisher BL, Sutherland FS, and Machemer R: Panretinal photocoagulation and retinal oxygenation in normal and diabetic cats. Am J Ophthalmol 101:657, 1986. 6. Diabetic Retinopathy Study Research Group: Preliminary report on effects of photocoagulation therapy. Am J Ophthalmol 81:383, 1976. 7. Lewis H, Aaberg TM, Abrams GW, Han DP, and Williams GA: Current causes of failure following diabetic vitrectomy. ARVO Abstracts. Invest Ophthalmol Vis Sci 29(Suppl):220, 1988.

289

8. Lewis H, Abrams GW, and Foos RY: Clinicopathologic findings in anterior hyaloidal fibrovascular proliferation after diabetic vitrectomy. Am J Ophthalmol 104:614, 1987. 9. Tagawa H, McMeel JW, Furukawa H, Quiroz H, Murakami K, Takahashi M, and Trempe CL: Role of the vitreous in diabetic retinopathy. Ophthalmology 93:596, 1986. 10. Tagawa H, McMeel JW, and Trempe CL: Role of vitreous in diabetic retinopathy: II. Active and inactive vitreous changes. Ophthalmology 93:1188, 1986. 11. Foos RY, Kreiger AE, Forsythe AB, and Zakka KA: Posterior vitreous detachment in diabetic subjects. Ophthalmology 87:122, 1980. > 12. Ernest JT and Archer DB: Vitreous body oxygen tension following experimental branch retinal vein obstruction. Invest Ophthalmol Vis Sci 18:1025, 1979. 13. Pournaras CJ, Ilic J, Tsacopoulos M, Leuenberger PM, and Gilodi N: Experimental branch vein occlusion: Modifications of preretinal PO2 in the affected territory. ARVO Abstracts. Invest Ophthalmol Vis Sci: 26(Suppl):246, 1985. 14. Ben-Nun J, Alder VA, Cringle SJ, and Constable IJ: A new method for oxygen supply to acute ischemic retina. Invest Ophthalmol Vis Sci 29:298, 1988. 15. Shaw WE, Blair NP, Dunn R, and Flora C: Reversal of retinal ischemic injury by vitreoperfusion. Invest Ophthalmol Vis Sci 30(Suppl):137, 1989.

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933576/ on 01/24/2017