Metabolic Dependence of Photoreceptors on the Choroid in the Normal and Detached Retina

Abstract

purpose. This article assesses the hypothesis that the high blood flow rate and low O₂ extraction associated with the choroidal circulation are metabolically necessary and explores the implications of the spatial relationship between the choroid and the photoreceptors for metabolism in the normal and detached retina.

methods. The O₂ distribution across the retinal layers was previously measured with O₂-sensitive microelectrodes in cat. Profiles were fitted to a diffusion model to obtain parameters characterizing photoreceptor O₂ demand. This was a study of simulations based on those parameters.

results. Photoreceptor inner segments have a high O₂ demand (QO₂), and they are far (20 to 30 μm) from the choroid. These unusual conditions require a large O₂ flux to the inner segments, which in turn requires high choroidal oxygen tension (PO₂), high choroidal venous saturation (ScvO₂), low choroidal O₂ oxygen extraction per unit volume of blood, and a choroidal blood flow (ChBF) of at least 500 ml/100 g-min. Movement of the inner segments further from the choroid, which occurs in a retinal detachment, severely reduces the ability of the inner segments to obtain O₂, even for detachment heights as small as 100 μm. Depending on detachment height and assumptions about choroidal and inner retinal PO₂ during elevation of inspired O₂ (hyperoxia), hyperoxia is predicted to partially or fully restore photoreceptor QO₂ during a detachment. conclusions. The choroid is not overperfused, but requires a high flow rate to satisfy the normal metabolic demand of the retina. Because the oxygenation of the photoreceptors is barely adequate under normal conditions, detachment has serious metabolic consequences. Hyperoxia is predicted to have clinical benefit during detachment.

The high blood flow rate (approximately 1400 ml/100 g per minute) ^{1 2 3 4} and low O₂ extraction (<1 vol %) ^{3 5} associated with the choroidal circulation are well known. However, the reasons for the unusually high flow rate and low O₂ extraction have long been a mystery. In a trivial sense, the reason for the high blood flow rate is that the capillaries in the choroidal circulation are unusually large, at least 10 μm in diameter, ⁶ and there are many in parallel. This means that the resistance of the choroidal vasculature is low, and the blood flow rate is therefore high. The question is why this circulation should have these properties. It has been suggested that the high flow rate exists for some purpose other than satisfying metabolic requirements of the outer retina ^{7 8} —possibly, temperature regulation. ⁹ In the present study, the metabolic role of the choroidal circulation is reassessed by linking a mathematical model of oxygen diffusion to the photoreceptors with well-known concepts about total oxygen extraction from a circulation. The diffusion analysis is an extension of a mathematical model that provides an excellent fit to data on oxygen tension (PO₂) as a function of distance through the cat retina under all conditions examined so far, including light and dark adaptation, ¹⁰ normoxia and hypoxemia, ¹¹ hyperoxia, ¹² elevated intraocular pressure, ¹³ and retinal vascular occlusion. ¹⁴ The analysis performed in this study shows that the combination of high choroidal flow rate and low oxygen extraction is necessary to sustain normal photoreceptor metabolism in the dark. 

Recently, Mervin et al. ¹⁵ and Lewis et al. ¹⁶ have provided evidence that hyperoxia can minimize the loss of photoreceptors and minimize reactions of the Müller cells during retinal detachment in the cat. The modeling performed in the current study provides further insight into the metabolic state of the detached retina, a theoretical basis for the observations of Mervin et al. and Lewis et al., and a prediction of the conditions under which hyperoxia would be beneficial during detachment. 

Methods

Figure 1 shows an example of an O₂ profile—that is, PO₂ as a function of distance, across the dark-adapted cat retina. It was recorded with an oxygen-sensitive microelectrode as it was continuously withdrawn from the retina at 2μ m/sec. ^{11 14} Analysis of many O₂ profiles of this type allowed the development of a mathematical model that accurately describes the O₂ distribution in the outer avascular half of the retina, which consists mainly of photoreceptors. ^{10 14} The model fit is also shown in Figure 1 . Oxygen diffusion and consumption are described by a three-layer model of the outer retina in which only the middle layer (layer 2), corresponding to the photoreceptor inner segments, consumes oxygen. This layer is the location of all the photoreceptor mitochondria, with the exception of those in the synapse (at the proximal edge of layer 3). This basic model is also valid for monkey, ¹⁷ toad, ¹⁸ and guinea pig ¹⁹ retinas, although the parameter values are quite different in toad and guinea pig. The model equations for Po ₂ as a function of distance (x) from the choriocapillaris are the solutions to the steady state version of Fick’s Law:  

\[\frac{d^{2}P}{dx^{2}}{=}\ \frac{Q}{Dk}\]

The solution, subject to matching PO₂ and O₂ flux at the boundaries between layers is  

\[\begin{array}{cc}P_{1}(x){=}{\alpha}_{1}x{+}{\beta}_{1}&0{\leq}\ \mathit{x}\mathrm{\ {\leq}\ }\mathit{L}_{\mathrm{1}}\\\mathit{P}_{\mathrm{2}}\mathrm{(}\mathit{x}\mathrm{)\ {=}\ }\ \frac{\mathit{Q}_{\mathrm{2}}}{\mathrm{2}\mathit{Dk}}\mathrm{\ }\mathit{x}^{\mathrm{2}}\mathrm{\ {+}\ {\alpha}}_{\mathrm{2}}\mathit{x}\mathrm{\ {+}\ {\beta}}_{\mathrm{2}}&\mathit{L}_{\mathrm{1}}\mathrm{\ {\leq}\ }\mathit{x}\mathrm{\ {\leq}\ }\mathit{L}_{\mathrm{2}}\\\mathit{P}_{\mathrm{3}}\mathrm{(}\mathit{x}\mathrm{)\ {=}\ {\alpha}}_{\mathrm{3}}\mathit{x}\mathrm{\ {+}\ {\beta}}_{\mathrm{3}}&\mathit{L}_{\mathrm{2}}\mathrm{\ {\leq}\ }\mathit{x}\mathrm{\ {\leq}\ }\mathit{L}_{}\end{array}\]

where α _i,β _i are constants in layer i (i = 1, 2, 3) determined from the boundary conditions (see Haugh et al. ¹⁰ for further details); P _i (x) is the PO₂ in layer i as a function of position x from the choriocapillaris; L ₁ is the position of the photoreceptor inner segment–outer segment boundary; L ₂ is the position of the inner segment–outer nuclear layer (ONL) boundary; L is the position of the inner–outer retinal boundary; Q ₂ is the oxygen consumption of the inner segment layer, between L ₁ and L ₂; D is the diffusion coefficient of oxygen in the retina; and k is the solubility of oxygen. 

In the present work, the model equations were not used to fit data, but to perform simulations by computer (Excel; Microsoft, Redmond, WA), in which parameters were varied, and the resultant effects on Q were evaluated. The following assumptions were made in all the modeling. First, the product of oxygen diffusion coefficient (D) and solubility (k) is constant across the tissue. The value of D is 71% of the value in water, ²⁰ and k is taken to be the value in blood, resulting in Dk of 2.8 × 10⁻¹⁰ (cm²-ml O₂)/(ml tissue-mm Hg-sec). Second, during dark adaptation under normoxic conditions, the oxygen consumption in layer 2 (Q ₂) has the maximum possible value—that is, the value that just reduces the PO₂ at some point along the rod inner segment (layer 2) to zero. This is the typical observation in the dark-adapted retina of cat. ^{11 12 13} As noted, oxygen consumption in layers 1 and 3 is zero. In particular, the O₂ consumption of the retinal pigment epithelium is so low that it does not influence the shape of the profile (Robert A. Linsenmeier, unpublished calculations, 1990). Oxygen consumption is reported in terms of mean outer retinal oxygen consumption, Q _OR (equivalent to Q _av in earlier publications). Q _OR is the consumption averaged over the entire outer retina  

\[Q_{\mathrm{OR}}{=}Q_{2}(L_{2}-L_{1})/L\]

The rationale for reporting Q _OR rather than Q ₂ is that we have higher confidence in values of Q _OR, as explained in detail by Haugh et al. ¹⁰ As the equations state, all oxygen diffusion is along the photoreceptors (x direction), and the retina is assumed to be homogeneous with no diffusion parallel to the retinal layers (y and z directions). Other assumptions are discussed in the results when they are relevant to the simulations. 

Results

Effect of Choroidal PO₂ on Photoreceptor Oxygen Consumption

From the data in Figure 1 , it can be appreciated that a PO₂ gradient exists, even in layers 1 and 3, where there is no O₂ consumption. PO₂ decreases across the outer segments from the choroid to the inner segments and across the ONL from the retinal circulation to the inner segments. The flux of O₂ through layers 1 and 3, which is proportional to the slopes in these layers, determines the O₂ available to the inner segments. Because the PO₂ is zero at some point along the inner segments, the determinants of the slope through layer 1 are mainly the choriocapillaris PO₂ (P _C), and the distance from the choroid to the inner segments (L ₁). Figure 2 illustrates that if P _C declines, the flux decreases, because the minimum PO₂ in the profile cannot be below zero, and Q _OR must be smaller. In these simulations, parameter values were based on normal values obtained previously ^{10 11 13} for the size and position of layer 2 (L ₁ = 24 μm and L ₂ = 48 μm), the overall thickness of the outer retina (L = 120 μm), choriocapillaris PO₂ (P _C = 62 mm Hg), and the PO₂ at the inner boundary of the avascular region (P _L = 15 mm Hg). 

When P _C is varied while all other parameters (except Q _OR) are held constant, a linear relationship between Q _OR and P _C emerges  

\[Q_{\mathrm{OR}}{=}0.0645{\cdot}P_{\mathrm{C}}{+}0.5\]

The normal value of Q _OR is 4.5 ml O₂/100 g-minute at P _C = 62 mm Hg. A linear relationship was observed in vivo in data obtained during hypoxemia. ¹¹ In the simulation, the relationship has a positive intercept, because approximately 10% of the O₂ demand is met by diffusion from the inner retina, and the inner retinal supply is assumed not to change (i.e., P _L is constant). At the high end, it has been found that Q _OR is not different during normoxia and hyperoxia ¹² ; therefore, the function relating Q _OR to P _C plateaus at some value of P _C, but this value is not precisely known. The essential point of this analysis is that photoreceptor O₂ consumption can only be maintained at normal levels by keeping P _C as close as possible to arterial PO₂. 

To understand how these local oxygen measurements relate to blood flow, we must know the oxygen extraction per unit volume of blood. A high value of P _C implies a high choroidal venous O₂ saturation. A high venous saturation implies a low arteriovenous O₂ saturation difference in the choroidal circulation. The relation between P _C and the choroidal venous saturation (ScvO₂, ml O₂/100 ml blood) is determined by the hemoglobin saturation curve. The cat hemoglobin saturation curve does not fit the standard Hill equation well, and the relation used here was therefore obtained from experimental data. ²¹ In large vessels the hematocrit is approximately 40 in cat, but, as in other microcirculations, the hematocrit in the choriocapillaris, where O₂ exchange occurs, is lower than that in the large vessels. ²² For the simulation, a choriocapillaris hematocrit of 20 was assumed. When PaO₂ is 90 mm Hg and P _C is 62 mm Hg, the arteriovenous saturation difference (SaO₂ − ScvO₂) in the choriocapillaris is only 0.82 volume %. 

Implications for Choroidal Blood Flow

The local values for oxygen consumption and choriocapillaris PO₂ can be related to blood flow by the Fick principle, which recognizes that O₂ consumption for any tissue can be determined by a mass balance on O₂. ²³ For the particular case of the outer retina and choroid  

\[(0.9)Q_{\mathrm{OR}}{=}ChBF\ (\mathrm{SaO}_{2}-\mathrm{ScvO}_{2})\]

where ChBF is choroidal blood flow in ml/100 g-min. The factor 0.9 is included because only approximately 90% of Q _OR is derived from the choroidal circulation under normoxic conditions. ¹¹ In the present application the equation can be rearranged to calculate ChBF, because Q _OR and O₂ extraction are known. Therefore, the required value of ChBF is  

\[ChBF{=}0.9\ (Q_{\mathrm{OR}})/(\mathrm{SaO}_{2}-\mathrm{ScvO}_{2})\]

For the normoxic case, where Q _OR= 4.5 ml O₂/100 g-min, and SaO₂ − ScvO₂ = 0.82 volume %, ChBF is calculated to be 494 ml/100 g-min. This value is nutrient blood flow—that is, flow per 100 g of retina supplied by the choroidal circulation. It is important to realize that this high value was computed solely from considerations about the metabolic demand for oxygen. 

All the parameters in these equations (ChBF, Q _OR, P _C, ScvO₂) are linked, and it is therefore possible to derive other relationships among them. One function of interest is the dependence of Q _OR on ChBF (Fig. 3) . This closely matches experimental observations in cat. ² If ChBF were to decrease, more oxygen would be extracted per unit volume of choroidal blood, but this would lower P _C, which would reduce Q _OR. In the choroid, an increased oxygen extraction can never fully compensate for a decreased flow rate, although there is a relatively flat region in Figure 3 over which changes in flow have relatively little impact on consumption. 

Effect of Varying the Distance between the Choriocapillaris and the Inner Segments

One reason that ChBF must be high is the large metabolic demand of the photoreceptors. What makes this situation different from that in other tissues, however, is the large spatial separation between the microcirculation and the metabolically active region. If the inner segments were physically closer to the choroid, P _C would not have to be so high. This is illustrated in Figure 4 , where the inner segments have been moved 12 μm closer to the choroid. In this simulation, it was possible to reduce P _C from 62 to 40 mm Hg while maintaining the normal value of Q _OR. A lower P _C implies that O₂ extraction could be greater and therefore ChBF could be lower (equation 4) . Under this condition it would be possible to reduce ChBF from 494 to 168 ml/100 g-minute. 

Similarly, if the inner segments were further from the choroid, they would not be able to obtain enough oxygen to maintain normal oxidative metabolism. This situation, which occurs in retinal detachment, is shown in Figure 5 for relatively small detachments of 100 and 500 μm. The excess fluid under the retina does not impose a barrier to diffusion, but it adds to the thickness of layer 1, decreasing the O₂ flux to layer 2. Again, it should be noted that the PO₂ decreases across the fluid under the retina, even though this layer consumes no oxygen. The fraction of normal consumption that could be maintained for different detachment heights, if P _C were constant at 62 mm Hg, is shown in Figure 6A (lower curve). The effect is dramatic, so that with a 100 μm detachment, Q _OR declines to only 34% of normal. The lower curve in Figure 6A asymptotes to the consumption derived from the retinal circulation (approximately 10% of Q _OR in the attached retina), which was assumed not to change. Increasing ChBF would be of little value during detachment. At best, when ChBF is infinite, the arteriovenous saturation difference could approach zero, and P _C would be as high as PaO_2. Even in that case, when P _C = PaO₂ = 90 mm Hg, normal metabolism could not be maintained for detachments greater than approximately 15 μm (not shown). 

Effect of Hyperoxia during Retinal Detachment

Although an increase in ChBF would not restore Q _OR in the detached retina, systemic hyperoxia elevates P _C and would partially preserve Q _OR. Inspiration of 100% O₂ increases choroidal PO₂ to approximately 230 mm Hg ^{12 14} or more ²⁴ in cat and to approximately 250 mm Hg in pig. ²⁵ Figure 7 shows the effect of increasing P _C to 230 mm with a detachment height of 500 μm. The value of P _L has been assumed to be 89 mm Hg, based on experimental observations in cat during hyperoxia. ¹²  

The effectiveness of hyperoxia in preserving Q _OR for the conditions of an increase in P _C alone and an increase in both P _C and P _L are shown by the upper curves in Figure 6A . For detachments of more than approximately 500 μm, elevation of P _C alone would leave Q _OR at less than 30% of normal, but elevation of P _L as well would allow a substantial improvement for all detachment heights. 

Light adaptation reduces photoreceptor QO₂, and would be expected to provide some protection for the detached retina. The maximal decrease in QO₂ is approximately a factor of two when illumination is at or above the level sufficient to saturate rod responses. ^{10 11} A simulated oxygen profile during light adaptation is shown by the dashed line in Figure 5 . Until the detachment height is large enough that the PO₂ at the trough of this profile decreases to zero, complete protection of retinal metabolism can be achieved, but this occurs at a detachment height of only approximately 50 μm, as shown in the lowest curve of Figure 6B . For larger detachments, light allows a larger percentage of the normal QO₂ to be maintained than during darkness, but QO₂ is still reduced to approximately 30% of normal. Hyperoxia is beneficial, just as in dark adaptation. In fact, if the predicted increase in PO₂ in the retinal circulation is achieved, this alone would be sufficient to supply the light-adapted photoreceptors (top curve), which would make protection independent of detachment height. 

Discussion

The shapes of retinal PO₂ profiles may be counterintuitive, emphasizing the need for measurements and simulations. The physics of diffusion is such that both the PO₂ and the slope of PO₂ with distance must be continuous across an interface. Therefore, PO₂ changes even in layers of the retina (1 and 3) where there is no oxygen consumption. For this reason, the thickness of layer 1 is important. Another counterintuitive feature is that the slope of the PO₂ gradient is larger across the outer segments, where there is no consumption, than across the inner segments, where consumption is large. There would also be a gradient across the retina, however, even if there were no consumption anywhere. In the case of diffusion with zero consumption, the PO₂ profile would be a straight line connecting P _C and P _L. In the consuming case, the PO₂ across the retina would be expected to be lower. This is exactly what occurs when the diffusion-only case is compared with the case of diffusion plus consumption. The largest differences occur in layer 2, the consuming layer, as might be expected. 

High Choroidal Blood Flow

The simulations performed here allow a better understanding of the performance of the choroid as the oxygen supply to the outer retina. Starting from information about the oxygen demand of the photoreceptors, we have shown that the ChBF rate should theoretically be approximately 500 ml/100 g- minute, more than 10 times the flow rate in the retinal circulation. This predicted value is below the 1020 to 2600 ml/100 g-minute measured with microspheres and other techniques, ^{1 2 3 26} but our assumed values of choroidal PO₂ and outer retinal oxygen consumption (Q _OR) are conservatively low. If we had assumed Q _OR to be 5 ml O₂/100 g-minute, a value that has been observed in vivo, ¹¹ P _C would have had to be 70 mm Hg, and ChBF would have been 1250 ml/100 g-minute, within the range of measured choroidal blood flow. 

The shape of the function relating oxygen consumption to blood flow is relatively flat (Fig. 3) . There may be two ways to interpret this. First, to bring O₂ consumption to maximal levels, flow values must be very high. Achieving the highest possible O₂ consumption may not seem necessary but may be beneficial, because high rates of oxidative metabolism decrease the necessity for anaerobic glycolysis, resulting in less acid production. Alternatively, the flatness of this curve may be seen as protective. Choroidal blood flow is not tied directly to the inner segment metabolism, because it does not increase during hypoxemia, ^{27 28} and it would be undesirable for changes in blood flow to influence photoreceptor metabolism. Relative independence of metabolism from flow rate is only possible when the blood flow rate is high. 

The actual and predicted values are similar enough that we conclude that the choroidal circulation is not substantially overperfused, as has often been suggested. ^{7 8} This does not rule out the possibility that the high blood flow may have additional functions, such as heat transfer ^{9 29} and supply or removal of substances in addition to oxygen. 

The simulations were all performed with the value of Q _OR observed in the dark-adapted retina. During light adaptation Q _OR is only half that in dark adaptation, ^{10 11} and the metabolic requirement for high blood flow is not as great, although it would still be higher than in most circulations. During light adaptation P _C does not change and the PO₂ is well above zero in the outer retina, ¹¹ indicating that the flow does not decrease in response to the decreased metabolic demand. We can speculate that during illumination the high flow rate may assist in heat removal. ⁹  

Our analysis used parameter values specifically for cat retina, but we believe that the principles can be extrapolated to humans. Measurement of intraretinal PO₂ is not feasible in humans, but the oxygen distribution in the monkey retina is similar to that in cat, ¹⁷ and ChBF in the primate retina is comparable to that in cat. ^{29 30}  

For the choroid, a high flow rate–low oxygen extraction system is necessary, because choriocapillaris PO₂ must be kept high to maintain normal photoreceptor oxidative metabolism. A low flow rate–high extraction system, such as those in the inner retina and the brain, would not be adequate. The essential factor controlling the requirement for high flow rate is the distance between the choroid and the photoreceptor inner segments. The relatively long diffusion distance requires a high PO₂ at the choroid so that there is a steep enough O₂ gradient (high enough flux) between the choroid and the inner segment layer. If the distance between the choroid and inner segments was smaller, choriocapillaris PO₂ could be lower, oxygen extraction from the choroid could be larger, and ChBF could, therefore, be lower, even if photoreceptor QO₂ remained the same. 

Retinal Detachment

In retinal detachment, the distance from the choroid to the inner segments increases. Even though there is no O₂-consuming tissue under the retina, the fluid is an unstirred layer that reduces O₂ flux from the choroid to the inner segments. Debris in this space tends to make the situation slightly worse, but is not likely to have much of an effect, because O₂ readily diffuses through cells. Any stirring of subretinal fluid would make the situation slightly better, because convection would enhance O₂ transport to the inner segments. After a detachment, some convection may occur during eye movements, particularly if the detached retina moves. Convection would also tend to enhance the effectiveness of oxygen therapy. 

Experimentally, photoreceptors in the detached retina lose outer segments and then undergo apoptosis, and Müller cells proliferate and become hypertrophic. ³¹ Both photoreceptors and Müller cells exhibit biochemical changes. All these effects were reduced in animals with retinal detachments when they were made hyperoxic with 70% inspired oxygen. ^{15 16} Diffusion of other substances and metabolites would certainly also be compromised in a detachment, but the work by Mervin et al. ¹⁵ strongly suggests that oxygen plays the key role, and they suggested that hyperoxia could have a clinical benefit. The simulations reported here explain why oxygen is beneficial. Hyperoxia elevates both choroidal PO₂ and inner retinal PO₂, ^{12 14 24 25} both of which allow greater oxygen delivery to the photoreceptor inner segments. 

Theoretically, the effect of increased choroidal O₂ in maintaining photoreceptor metabolism is most pronounced for small detachment heights (<500 μm). Increased O₂ carried by the retinal circulation during hyperoxia is predicted to be important in maintaining photoreceptors, for all detachments, however, and this may be especially important during light adaptation. Little is known about the physiology of the retinal circulation during detachment, so quantitative predictions of this effect are difficult to make. The benefit of hyperoxia probably lies between the two upper curves in Figure 6A or 6B. 

The minimal level of QO₂ needed to prevent photoreceptor apoptosis is unknown. At least two factors probably help photoreceptors survive in the absence of oxygen therapy. First, all photoreceptors deconstruct outer segments during a detachment, ¹⁵ which should reduce the major energy-consuming processes—pumping of Na⁺ that usually enters through the light-dependent channels in the outer segments and cyclic nucleotide turnover in the outer segments. ^{32 33} Second, after some photoreceptors are lost, the available oxygen would have to be shared by fewer of them, so the QO₂ of each remaining photoreceptor should be closer to normal. 

Our analysis has focused on increased retinal oxygenation through the retinal and choroidal circulations. Attempts have been made to increase ocular oxygenation by blowing oxygen across the cornea as well. In the absence of oxygen breathing, this was effective in oxygenating the aqueous, ^{34 35} but preretinal PO₂ in rabbits was elevated only after lensectomy and vitrectomy. ³⁵ We cannot model this case exactly, but it is likely that the O₂ gradient from the cornea to the retina would be too shallow to allow significant oxygen flux to the retina. When combined with oxygen inspiration, elevated corneal oxygen would probably provide little benefit. Because of the rich vasculature of the iris and ciliary body, an increase in inspired oxygen should eventually result in an increase in PO₂ in the vitreous, regardless of what is happening at the cornea. 

Clinical Issues

In many retinal detachments, the defect in the retina is peripheral, but the detachment gradually spreads centrally, eventually reaching the macula. ³⁶ When macular involvement begins, the detachment height in that region must be small, and therefore elevated choroidal O₂ would be most beneficial during this time. Choroidal PO₂ can be elevated by inspiration of increased concentrations of O₂. Of course, the foveal photoreceptors benefit very little from increased oxygenation through the retinal circulation. Therefore, it seems important to treat foveal detachments with inspired O₂ and timely surgical repair before the detachment height is too great. Similarly, after surgery, elevated choroidal O₂ could maintain photoreceptor function until the pumping of subretinal fluid by the retinal pigment epithelium has allowed the retina to reattach fully. Although the macula is the most important region for vision, any detached area could benefit from O₂ therapy before and immediately after reattachment surgery. 

These simulations suggest that hyperoxia probably cannot allow completely normal amounts of oxidative metabolism in the photoreceptors during most detachments. It may be preferable to restore QO₂ partially rather than fully, to minimize acidosis and buildup of other metabolites. Oxygen cannot be stored, and brief periods of O₂ therapy are therefore not likely to be useful. Instead, we recommend that continuous or nearly continuous inspiration of elevated O₂ be instituted as soon as a detachment begins to affect the macula, if not earlier. Elevated O₂ levels should be maintained until the macula reattaches after surgery, rather than ended immediately after surgery. There should be no concern about the vasoconstriction of the retinal circulation caused by hyperoxia, because the inner retina should be at least as well oxygenated during hyperoxia as during normoxia. ^{12 25} The illumination during oxygen therapy should be considered. We predict that moderate illumination can be beneficial because it decreases photoreceptor metabolism, but it is also known that oxygen can potentiate light damage in response to strong illumination. ³⁷ Inspiration of 100% O₂ is not feasible, because of its toxicity to the lung, but 50% to 60% O₂ can be tolerated for long periods. ^{38 39} This level is not likely to cause permanent injury to the attached portion of the retina, based on the limited available literature. ^{40 41} We anticipate that such a regimen would preserve photoreceptors and may thereby improve visual outcome after reattachment surgery. 

 Supported by National Institutes of Health Grants R01 EY05034 and 1 T32 EY07128.

 Submitted for publication January 24, 2000; revised April 14, 2000; accepted April 26, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Schematic of the retina (top) and O₂ profile through the retina (bottom). An O₂ profile through the dark-adapted retina is shown by the data points. In the outer half of the retina, the profile is fitted to the model described in the text. Layer 1, from x = zero to L ₁, corresponds to the retinal pigment epithelium and outer segments; layer 2, from x = L ₁ to L ₂, is the only layer that consumes O₂ and corresponds to inner segments; layer 3, from x = L ₂ to L, corresponds to the ONL.

Figure 2.

 

Theoretical O₂ profiles through the outer half of the retina computed as described in the text, during normoxia (P _C = 62 mm Hg) and when choroidal PO₂ has been reduced to half this value. The O₂ supplied by the retinal circulation was assumed to be constant. The average oxygen consumption through the outer retina decreases by almost a factor of two.

Figure 3.

 

Expected dependence of outer retinal O₂ consumption (Q _OR) on choroidal blood flow (ChBF).

Figure 4.

 

Effect of moving the inner segments closer to the choroid. The solid line shows the inner segments located between L ₁ = 24 μm and L ₂ = 48 μm from the choroid to a new position between 12 and 36 μm from the choroid (dashed line). In this case, choroidal PO₂ can be reduced to 40 mm Hg without changing Q _OR.

Figure 5.

 

Effect of retinal detachment on outer retinal O₂ profiles and O₂ consumption. The horizontal bar shows the position of the outer retina when the retina is attached and when it is detached by 100 and 500 μm. Oxygen consumption decreases as the detachment size increases. In all cases, the inner segments were assumed to be 24 to 48 μm from the distal edge of the retina. The curves are labeled with the predicted oxygen consumption as a percentage of the normal value in dark adaptation. The dashed line shows the normal profile during light adaptation, when Q _OR is assumed to be 50% of its value during dark adaptation. Profiles in the detached retina during light adaptation are superimposed on the profiles during dark adaptation, but the reduction in Q _OR, as a fraction of the normal value in light adaptation, is not as great (see Fig. 6B ).

Figure 6.

 

(A) Effect of breathing 100% O₂ on photoreceptor oxygen consumption (Q _OR) in the detached retina during dark adaptation under different assumptions. The lowest curve assumes that the animal is breathing air. The middle curve assumes that choroidal PO₂ (P _C), but not inner retinal PO₂ (P _L), is increased during hyperoxia (100% O₂). The upper curve assumes that both P _C and P _L increase during hyperoxia. This relies on the retinal circulation’s behaving normally in the detached retina. (B) Curves similar to those shown in (A) but the animal has been light adapted.

Figure 7.

 

Oxygen profiles during normoxia and hyperoxia with a detachment of 500μ m. The hyperoxic choroidal PO₂ was assumed to be 230 mm Hg.

The authors thank Juan Grunwald, David Weinberg, and Marian Macsai for useful discussions and comments on the manuscript.