Effects of Hyperoxia on the Oxygen Distribution in the Intact Cat Retina

Robert A. Linsenmeier*f and Charles M. Yanceyf

Double-barreled oxygen microelectrodes were used to measure the distribution of oxygen within the dark adapted cat retina during systemic hyperoxia (100% O2 inspired). Oxygen tension (PO2) decreased monotonically from the choroid to the vitreous humor in most cases, showing that a greater portion of the retina was supplied by the choroid during hyperoxia than during normoxia. In the proximal half of the retina the PO2 increased during hyperoxia by an average of about 40 mm Hg, while the increase in the distal retina was larger. At the choroid the average increase in PO2 was about ISO mm Hg. Analysis of the oxygen profiles showed that photoreceptor oxygen consumption was unchanged during hyperoxia. Retinal PO2 increased rapidly at the beginning of hyperoxia, and often partially recovered from its peak value during hyperoxia, even in the distal retina, suggesting that the choroidal circulation may have some limited autoregulatory capacity. As in normoxia, retinal illumination led to an increase in PO2 in the distal retina, due to a decrease in oxygen consumption. The light-evoked increase in PO2 was larger during hyperoxia, but the underlying change in oxygen consumption was probably the same as in normoxia. Invest Ophthalmol Vis Sci 30:612-618,1989

Numerous studies have shown that adequate has been controversial. Some investigators suggest amounts of oxygen are essential for normal retinal that diffusion of oxygen from the choroid is essential function, and in higher mammals both the choroidal in causing vasoconstriction of retinal vessels in hy- and retinal circulations are necessary (eg, refs. 1-3). peroxia4 but the m'ost recent data15 suggest that oxy- When the retinal circulation is compromised, it is gen carried by the retinal circulation must be impor- believed that the choroidal circulation may be able to tant. Less is known' about the PO2 in the distal retina supply most or all of the retina, provided that the during hyperoxia, since this cannot be measured with 4 5 arterial oxygen tension (PaO2) is elevated, ' but hy- vitreal electrodes. Choroidal blood flow is generally peroxia can have deleterious effects as well. If in- considered to be passive with respect to changes in 6 1617 spired PO2 is very high, retinal oxidative metabolism PaO2, and since oxygen extraction from the cho- + + 7 218 and Na -K ATPase activity are reduced. The retina roid is small choroidal PO2 would be expected to is also more susceptible to light damage in hyperoxic rise substantially when PaO2 is elevated. This has, in monkeys than in normoxic ones.8 Finally, the best fact, been demonstrated in the pig retina.19 example of the detrimental effect of hyperoxia is that The current work was undertaken to provide data it prevents normal retinal vascular development in on the oxygen distribution across the retina during premature infants (eg, ref. 9). hyperoxia, since this basic information is important Measurements of vitreal PO2 have shown that in several of the contexts mentioned above. The ef- inner retinal PO2 is reasonably well regulated during fect of light on the oxygen distribution has been eval- hyperoxia,10 due to the constriction of retinal ves- uated, since this has even more dramatic effects dur- sels""13 and the consequent reduction of retinal ing hyperoxia than during normoxia.3 Modelling of blood flow.12"14 The mechanism of the constriction oxygen transport has been employed to address issues related to retinal oxygen consumption. From the *Departments of Biomedical Engineering and Neuro- biology and Physiology and the tlnterdepartmental Graduate Pro- Materials and Methods gram in Neuroscience, Northwestern University, Evanston, Illi- nois. The methods for animal preparation, recording Supported by NIH grant EY-05034, Bethesda, Maryland, and and visual stimulation were the same as those pre- the Whitaker Foundation. viously published.320 Fourteen adult cats were used Submitted for publication: February 18, 1988; accepted October in this study. We have adhered to the ARVO Resolu- 26, 1988. tion on the Use of Animals in Research. During nor- Reprint requests: Dr. Robert A. Linsenmeier, Biomedical Engi- neering Department, Northwestern University, 2145 Sheridan moxia the animals breathed air, supplemented when Road, Evanston, IL 60208. necessary with enough O2 to keep arterial PO2 above

612

Downloaded from iovs.arvojournals.org on 09/26/2021 613 No. 4 HYPEROXIC EFFECTS ON RETINAL PO2 / Linsenmeier and Yoncey

85 mm Hg. Hyperoxia was induced with 100% in- Normoxia spired O2. During normoxia.the average PaO2 was 98.6 ± 13.0 mm Hg (mean ± SD; n = 27) and during hyperoxia it was 415.5 ± 72.3 mm Hg (n = 54). Arte- rial pH and PCO2 were 7.42 ± 0.04 and 31.3 ± 5.6 mm Hg during normoxia, and were unchanged (by paired t-test) during hyperoxia. Arterial blood pressure was generally unchanged by hyperoxia. The du- ration of hyperoxia varied from 6 min to about 2 hr depending on the type of experiment being performed.

Results

Oxygen Profiles in the Dark-Adapted Retina

The distribution of oxygen in the dark-adapted retina was measured during electrode penetrations and 0 L withdrawals. During penetrations the electrode was 100 80 60 40 20 0 advanced in 3 /im steps from the vitreal surface to the choroid, and at approximately 20 uta intervals the Retinal depth (%) current was allowed to stabilize and was recorded. Fig. 1. Profiles of retinal oxygen tension during electrode with- Withdrawals were performed with the electrode drawal from choroid (100% depth) to vitreous (0% depth) during moving in 1 fim steps at 1 or 1.5 /xm per sec. The (A) normoxia, and (B) and (C) hyperoxia. Each profile is a single retinal depth of the electrode was assessed with the withdrawal during dark adaptation; 100% depth was the point at local electroretinogram, which was recorded with the which the reference barrel recorded the increase in potential across the retinal pigment epithelium, which coincided with the beginning reference barrel.3 Recordings were usually made in or of the PO2 decrease. Zero was taken to be the point at which the near the area centralis. Figure 1 shows oxygen profiles ERG b-wave changed from a negative intraretinal polarity to a (ie, PO2 as a function of retinal depth) measured dur- positive vitreal polarity. B was recorded in the area centralis (cat ing electrode withdrawals during air breathing (nor- 31). C was recorded during hyperoxia near the superior retinal moxia) (Fig. 1 A) and hyperoxia (Fig. IB) in the same artery and shows a peak in PO2 in the proximal retina (cat 31). animal. The typical normoxic profile320 showed a relatively high PO at the choroid and then a steep gra- 2 profile in most of the retina is more likely to be simi- dient, culminating in a PO of close to zero mm Hg 2 lar to that in Figure IB. between 65 and 85% retinal depth. The PO then rose 2 A substantial variation was observed in choroidal in the inner half of the retina, where the PO2 averaged PO2 during both normoxia and hyperoxia, as shown about 20 mm Hg. During hyperoxia the PO2 increased throughout the retina, but the most dramatic in Figure 2. During normoxia the average choroidal change was in the outer retina, where the minimum PO2 was 77.2 ± 20.7 mm Hg (n = 24), while during was eliminated. Since profiles were never measured hyperoxia it was 230 ± 87 mm Hg (n = 18). The in exactly the same location twice, this type of mea- corresponding arterial PO2s were 98.6 ± 11-0 and 436 surement was not suitable for determining the size of ±57 mm Hg. Thus, in both absolute and relative terms, the difference between arterial and choroidal changes in PO2 at any depth. They did show, however, that in most cases (seven of eight penetrations PO2 was greater during hyperoxia. and 15 of 20 withdrawals in eight cats), PO2 increased monotonically, but not linearly, from the vitreal sur- Oxygen Consumption during Hyperoxia face to the choroid. The slope of the profile in the As described in the previous paper,20 oxygen pro- distal half of the retina was usually at least twice as files can be subjected to a steady-state diffusion analy- great as the slope in the proximal half of the retina. sis that allows estimation of the parameter Q/Dk, Several profiles showed the same features as de- which includes consumption per unit volume of tis- scribed above in the distal retina, but also had a rela- sue per time (Q), oxygen solubility (k) and oxygen tive maximum in PO2 in the proximal retina. The diffusion coefficient (D). As in the previous paper, the largest effect of this type was observed close to the avascular layer was divided into three sublayers superior retinal artery (Fig. 1C). Since this type of whose thicknesses were determined by fitting data to profile was observed only very near large vessels, the the model. The layer closest to the choroid (0 < x

Downloaded from iovs.arvojournals.org on 09/26/2021 614 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1989 Vol. 30

400

6 200

0 L '

0 100 200 300 400 500 600

Arterial PO2 (mm Hg)

Fig. 2. Choroidal PO2 as a function of arterial PO2 during nor- 30 mm Hg moxia (left group of points) and hyperoxia (right group of points). 100% oxygen The solid line connects the means of arterial and choroidal PO2 5 min under the two conditions. These are given in the text. Fig. 3. Response of PO2 at two locations in the dark-adapted retina to 5 min inspiration of 100% O2. < Li) was assumed to have no oxygen consumption. The second layer (L, < x < L2) had a uniform value of greater than 70 mm Hg. No factor has been applied to consumption (Q2/Dk), and the third layer (L2 < x Qav/Dk to correct for the distortion of retinal thick- < L) again had no oxygen consumption. Fits of the ness observed during the withdrawals,3 but the rela- hyperoxic profiles to this model enabled us to deter- tive normoxic vs. hyperoxic values are meaningful. mine an average consumption parameter for the dis- The consuming layer (Li/L to L2/L) was found to lie tal retina: from 25% to 45% of the way from the choroid to the inner edge of the avascular region during hyperoxia Qav/Dk = Q2/Dk X (L2 - L,)/L (or 78% to 88% retinal depth), and from 21% to 45% which varied less than the values of Q2/Dk. We were during normoxia. The hyperoxic values were not sta- also able to find the boundaries of the consuming tistically different from the normoxic ones. The aver- layer in terms of normalized distance through the age oxygen consumption parameter, Qav/Dk, was avascular retina, L|/L and L2/L. also the same during hyperoxia as during normoxia, The means and standard deviations of the fitted 0.70 X 106 mm Hg/cm2 vs. 0.73 X 106 mm Hg/cm2. parameters for 17 hyperoxic profiles are given in The hyperoxic value also agrees with the value ob- Table 1, along with the corresponding values for the tained from previous normoxic data3 refitted with the normoxic area centralis profiles analyzed in the pre- three-layer model (Haugh LM, Linsenmeier RA, and vious paper20 for which perfusion pressure (mean ar- Goldstick TK, submitted for publication) after a cor- terial pressure minus intraocular pressure) was rection for withdrawal length is applied (L = 186 ± 42 /xm here vs. 164 ± 32 nm in the previous data).

Table 1. Parameters of oxygen consumption model Transients in Retinal PO2 Normoxia Hyperoxia In addition to measurements of steady state oxygen 2 profiles, we were interested in the time course and Qnv/Dk (mm Hg/cm 6 X ICT ) 0.73 ± 0.56 0.70 ± 0.45 magnitude of changes in retinal PO2 at different reti- L,/L(%) 20.8 ± 13.5 24.9 + 15.5 nal depths during step changes in inspired oxygen L2/L(%) 45.2 ± 12.4 45.4 ± 26.5 concentration. Typical responses are shown in Figure PCO2 (mm Hg) 61.6 ± 16.5 263.5 ± 141.8 PLO2 (mm Hg) 11.0 ± 10.2 88.9 ± 15.5 3. For all of the retinal recording locations the PO2 RMS error rose rapidly to a peak and often partially recovered (mm Hg/data point) 1.53 ± 0.54 5.98 + 4.85 during hyperoxia. For quantitative analysis, these Number of profiles 61 17 transients were grouped into those measured at four Parameters (mean and standard deviation) derived from fits of dark- locations: the vitreous humor, the proximal half of adapted oxygen profiles to the model described in text. The first three parameters are defined in the text. PCO2 is choroidal PO2. PLO2 is the PO2 at the the retina, the region of the dark-adapted minimum proximal edge of the avascular layer. RMS error is the error per data point in in PO (65 to 85% retinal depth) and the region distal the fits. The normoxic data are the area centralis data at perfusion pressure 2 greater than 70 mm Hg, from ref. 20. to this minimum. As shown in Table 2, the amplitude

Downloaded from iovs.arvojournals.org on 09/26/2021 No. 4 HYPEROXIC EFFECTS ON RETINAL PO2 / Linsenmeier ond Yoncey 615

of the change in PO2 increased with retinal depth, as Table 2. Changes in PO2 during episodes of hyperoxia expected from the shape of the profile in Figure IB. tJ>02 The average amplitudes of the responses at the two Measurement site (mmHg) n % Recovery n. distal recording locations were not significantly different from each other, but the average amplitude at Vitreous 25 ±20 0 Proximal retina 39 ±25 26 ± 7 both of these locations was greater than that in the Minimum 108 ± 55 19 ± 10 proximal retina (P < 0.05). The vitreal responses Distal to minimum 126 ±96 33 ± 16 were not significantly different in amplitude from the Increase in PO2 during hyperoxia (hyperoxic steady state minus normoxic) proximal retinal ones. Table 2 also shows the average and percent recovery from the peak PO2 observed during episodes of hyper- extent of recovery from the peak as a percentage of oxia, as a function of recording location, n is the total number of episodes of the initial change [100 X (Ppeak - Psteady state)/(Ppeak hyperoxia and n, is the number showing some POj recovery during hyper- - Pnormoxia)] for those measurements in which recovery was observed. Recovery was never observed in the vitreal recordings, but was observed in 27 of 34 recordings within the retina. A significantly smaller illuminations during both conditions. In both cats the percentage recovery was observed at the location of threshold for a detectable light-evoked change in PO2 decreased by about 0.5 log unit, from 6.1 log quanta/ the distal minimum than at the more distal location 2 3 (P < 0.02), but no other differences were significant. deg -sec during normoxia to 5.6 during hyperoxia. The mechanism of this recovery will be considered in Since the light-evoked increase in PO2 reflects a the Discussion. decrease in oxygen consumption, the larger light- evoked responses during hyperoxia might suggest As shown in Figure 3, PO2 increased rapidly at the that the dark-to-light change in retinal oxygen con- beginning of hyperoxia and then approached a peak sumption was greater during hyperoxia. Simulations more gradually. The time to peak was 30.4 ± 16.8 sec of oxygen profiles, however, showed that this was not for the proximal retina, 51.6 ± 18.9 for the PO2 min- necessarily the case. Figure 5 shows oxygen profiles imum (46 ±9.8 after excluding two unusually slow derived from a model of oxygen distribution (see Ap- transients) and 33.8 ± 17.0 for the distal retina. pendix), in which three layers were used for the distal Transients at the minimum were significantly slower than at either of the other locations (P < 0.02). Vit- real responses were slower, in general, than those in the retina, but this was not quantified, since the time course depends strongly on the distance of the electrode from the retina.21 The time course of recovery during hyperoxia was similar at all depths, with the 40 mm Hg PO2 reaching steady state 3 to 4 min after the onset of hyperoxia. The transition from 100% O2 to air was not studied in detail, but appeared to be about as rapid as the transition from air to 100% O2 and to have a very small undershoot in some cases.

Effects of Illumination Because oxygen consumption decreases with illumination, PO2 increases with light in the outer half of the cat retina.3 During hyperoxia, light-evoked changes in PO2 were always larger in amplitude, but had the same time course as during normoxia, as illustrated in Figure 4. The increase in the light- evoked change in PO2 was variable, ranging from a factor of 2 to a factor of 6 (n = 14 pairs of responses to 30 or 60 sec of illumination). These light-evoked 30 sec illumination

changes were elicited by a relatively strong stimulus Fig. 4. Light-evoked changes in PO2 during normoxia (top) and (sufficient to saturate rod responses) and the increase hyperoxia(middle). The stimulus was a diffuse white light at 8.6 log 2 in amplitude suggested that an increase in PO might quanta/deg -sec presented to an initially dark-adapted retina. The 2 responses are superimposed after the normoxic one was increased be detectable at a lower illumination during hyper- by a factor of 2.4 to show that the time course was not altered oxia than normoxia. This was tested in two cats by significantly during hyperoxia. Fluctuations in the records are at recording light-evoked changes in PO2 over a range of the animal's respiratory frequency (cat 19).

Downloaded from iovs.arvojournals.org on 09/26/2021 616 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1989 Vol. 30

250 r larger change in PO2 from dark to light during hyperoxia than during normoxia, even though the change Hyperoxia in consumption was assumed to be the same. This is 200 shown for one depth by the vertical bars in Figure 5. Second, the exact amount of the increase in the light- evoked response between normoxia and hyperoxia depended strongly on position. At 100 /xm, for in- 150 stance, the change during hyperoxia was more than four times that during normoxia, while at 20 /im the hyperoxic response was only about 40% larger. Both 100 of these results mimic features of the microelectrode recordings. Neither of these results were sensitive to the details of the simulation (ie, to the value chosen 50 for net proximal retinal oxygen consumption, to the assumed PO2 at the vitreal border of the retina or to the exact values of distal retinal consumption). It also became clear that during hyperoxia a light-evoked change in outer retinal oxygen consumption would 50 100 150 200 lead to effects that propagate more proximally; the Distance from choroid (|jm) PO2 changed with light in the proximal retina in this Fig. 5. Simulations of oxygen profiles during light and dark adap- model, although the assumed value of oxygen con- tation in normoxia and hyperoxia. For these profiles the avascular sumption in the proximal retina was constant. PO2 region was assumed to extend from 0 to 115 urn. The consuming was held constant in this model at the vitreal surface, portion of the distal retina extended from 35 to 60 pm and the but in fact would be expected to increase with light if 6 2 value of Q/Dk in this region was 8.5 X 10 mm Hg/cm during a vitreal compartment were added to the model. The darkness and 5.0 X 106 mm Hg/cm2 during light. Diffusion, but not consumption, occurs everywhere else in the distal retina. The propagation of the change in distal retinal consump-

boundary conditions were a choroidal PO2 of 85 during normoxia tion into the proximal retina probably accounts for and 250 mm Hg during hyperoxia, and a vitreal PO2 of 15 during the ability of Stefansson et al22 to detect a light- normoxia and 45 during hyperoxia. All of these parameter values evoked increase in PO2 in the vitreous during hyper- are close to the averages actually observed. The proximal retina was assumed to have a constant PO2 during normoxia and a uniform net consumption of 0.4 X 106 mm Hg/cm2 during hyperoxia. The

vertical bars show the changes in the PO2 during illumination that would be expected at a depth of 50 fim under this set of conditions. Discussion These values are 22 mm Hg during normoxia and 38 mm Hg during hyperoxia. Inspiration of 100% O2 rapidly caused a large increase in retinal PO2, particularly in the distal retina. The average choroidal PO2 increased by about 150 half of the retina as described above. The parameters mm Hg, but was still 200 mm Hg below PaO2 during used for the distal retina were, as far as possible, aver- hyperoxia. The average choroidal PO2 measured in ages of the values obtained from the fitting. The same cat, 230 mm Hg, is similar to that previously reported change in oxygen consumption was assumed to occur for miniature pig, about 190 mm Hg.19 Choroidal from darkness to light in both normoxia and hyper- PO2 does not approach PaO2 more closely because oxia. The extension of the curves into the proximal dissolved rather than hemoglobin bound oxygen is half of the retina required an approximation, since used by the distal retina during hyperoxia. A simple the geometry of the retinal circulation is complicated calculation shows that if the choroidal arteriovenous and the diffusion equation can not be applied in an saturation difference were one volume percent18 and 5 exact way. During normoxia, PO2 in the proximal if oxygen solubility were 3.15 X 10~ m O2/ml blood- retina was assumed to be constant. During hyperoxia mm Hg, then the expected choroidal venous PO2 the shallow gradient through the proximal retina (Fig. would be 315 mm Hg below PaO2 during hyperoxia. IB) was simulated by adding a layer to the model The wide range of values of choroidal PO2 observed, which had a homogeneous net consumption that was and an average value 200 mm Hg less than PaO2, the sum of a term representing utilization and a term would be expected, since the electrode's position rela- of opposite sign representing supply. This was as- tive to the arterial and venous ends of the chorioca- sumed to be independent of illumination. pillaris might vary. During normoxia, average choroi- The modelling provided two important results. dal PO2 is closer to arterial PO2, because hemoglobin First, at every point in the distal retina there was a bound oxygen is used and an arteriovenous satura-

Downloaded from iovs.arvojournals.org on 09/26/2021 No. 4 HYPEROXIC EFFECT5 ON RETINAL PO2 / Linsenmeier and Yancey 617

tion difference of one volume percent is accompanied moxia. This substantiates the conclusion drawn from by a relatively small change in PO2. These consider- isolated mammalian retina that hyperbaric oxygen ations also suggest that during hyperoxia the oxygen reduces ATPase activity7 and oxygen consumption,6 distribution in the distal retina may be more heterog- but oxygen at atmospheric pressure does not.25

enous than during normoxia, and raise the possibility The response of retinal PO2 to step changes in in- of a lateral component to PO2 gradients in the distal spired gas confirmed the observations, made also retina as well as the dominant radial component. during withdrawals, that proximal retinal PO2 was Oxygen diffuses much further from the choroid much better regulated than choroidal or distal retinal during hyperoxia, but, since the retinal circulation PO2. The recovery of PO2 during hyperoxia is ex- was not occluded in this study, and because the value pected in regions supplied by the retinal circulation, of inner retinal oxygen consumption is unknown, it is where flow rate would be high at the beginning of still difficult to predict whether the entire retina could hyperoxia and then gradually decrease. A partial re- be supplied by choroidal oxygen during hyperoxia. covery of brain PO2 during hyperoxia has been ob- 26 The profiles in Figure 5 cannot be taken as evidence served previously. For the choroid, however, auto- 1617 for the ability of the choroid alone to supply the ret- regulation has not been demonstrated, so the re- ina, since the model assumed that the retinal circula- covery of distal retinal PO2 during hyperoxia was tion was present. Certainly those profiles do indicate surprising. Several explanations for this phenomenon that the possibility of supplying the inner retina with can be proposed. First, a partial recovery of arterial choroidal oxygen is greater in the light-adapted eye, PO2, due to some slow regulatory effect in the lung, since photoreceptor oxygen consumption is reduced would explain this result, but repeated sampling of by light, allowing more oxygen to diffuse into the arterial blood during the first 4 min of hyperoxia in 5 inner retina. Landers showed, by measuring vitreal two cats failed to show any tendency for blood PO2 to PO2, that after occlusion of the retinal circulation the recover. Continuous measurements of PaO2 in min- 15 choroid could supply the entire retina in cats and iature pigs during hyperoxia also show no recovery. monkeys. This has led to confusion as to why Flower Second, retinal PO2 would decrease if oxygen con- and Patz,23 who occluded the retinal circulation with sumption increased with time during hyperoxia. a laser, were able to restore the ERG b-wave to only While not impossible a priori, the estimates of oxygen 45% of control by administration of 100% oxygen at consumption derived from hyperoxic profiles appear atmospheric pressure. It is possible that the laser to rule this out. A third possibility would be a move- damaged the choroidal circulation, but fluorescein ment of the electrode relative to the retina, such that angiograms demonstrated patency of the choroidal the electrode recorded from a more vitreal location. vessels.23 A difference between these studies that may The recovery would then simply represent sliding be at least as important is that Flower and Patz did down the gradient of Figure IB. One way to assess their experiments in dark adaptation while Landers electrode movement is to determine whether the in- probably worked in light adaptation. Clinically, oxy- traretinal b-wave changed in amplitude during hy- gen therapy in cases of retinal occlusion appears to be peroxia. Since the b-wave peaks at about 50% retinal of limited value,5 but the current work suggests that it depth and decreases more distally, an electrode mov- might be more beneficial if the patient were kept ing in the vitreal direction from a position initially at light-adapted. or beyond 50% depth would be expected to record a slightly larger b-wave. A small increase in the b-wave When circulatory status is normal, oxygen cer- during hyperoxia was occasionally observed, but no tainly reaches far enough from the choroid to make more often than a small decrease or no change in the some contribution to regulation of the diameter of b-wave. In addition, on at least one occasion, the retinal vessels during hyperoxia. Even during nor- electrode clearly bumped the RPE during hyperoxia, moxia choroidal oxygen probably reaches the inner 3 24 indicating directly that the electrode had moved more retina, - but only in the light-adapted eye. From the distally. Why any movement at all should occur is profiles measured here (Fig. IB), it appears that cho- not clear, unless a fourth hypothesis has some merit. roidal oxygen could play a role in constriction of 15 This would be that the choroidal circulation does small retinal vessels, but, as Riva et al have pointed have a limited capacity for oxygen autoregulation out, and as shown by Figure 1C, the excess oxygen and that its flow rate decreased during hyperoxia. derived from retinal vessels themselves must also be This could be mediated locally or perhaps by a cen- important, perhaps more important for the vasocon- 11 3 tral neural mechanism. A decrease in flow would lead striction seen ophthalmoscopically. ' to an increase in oxygen extraction per unit volume Oxygen does not appear to be toxic to the retina in of blood, and to a partial recovery of PO2. The con- the short term at the levels used here, since outer comitant decrease in choroidal volume might lead to retinal oxygen consumption was the same as in nor-

Downloaded from iovs.arvojournals.org on 09/26/2021 618 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1989 Vol. 30

a small retinal movement that would make the recov- difference and mean circulation time. Circulation 33:302, ery appear larger. Few measurements have been 1966. 1617 14. Riva CE, Grunwald JE, and Sinclair SH: Laser doppler velo- made of choroidal blood flow during hyperoxia, cimetry study of the effect of pure oxygen breathing on retinal and it is possible that they have been too insensitive blood flow. Invest Ophthalmol Vis Sci 24:47, 1983. to measure small changes in flow. One could also 15. Riva CE, Pournaras CJ, and Tsacopoulos M: Regulation of speculate that flow through the chorocapillaris de- local oxygen tension and blood flow in the inner retina during creases during hyperoxia, but that arteriovenous hyperoxia. J Appl Physiol 61:592, 1986. 16. Bill A: Aspects of physiological and pharmacological regula- shunting increases so that total choroidal flow is un- tion of uveal blood flow. Acta Soc Med Upsal 67:122, 1962. changed. 17. Friedman E and Chandra SR: Choroidal blood flow: III. Ef- fects of oxygen and carbon dioxide. Arch Ophthalmol 87:70, Key words: retina, oxygen, hyperoxia, oxygen consump- 1972. tion, cat 18. Aim A and Bill A: Blood flow and oxygen extraction in the cat uvea at normal and high intraocular pressures. Acta Physiol Scand 80:19, 1970. Acknowledgment 19. Pournaras CJ, Riva CE, Strommer K, Tsacopoulos M, and Gilodi NA: O2 gradients in the miniature pig retina in nor- We thank Laura M. Haugh for assistance with the model- moxia and hyperoxia. In Ocular Circulation and Neovascular- ling. ization, BenEzra D, Ryan SJ, Glaser BM, and Murphy RP, editors. Netherlands, Martin Nijhoff/Dr. W. Junk, 1987, pp. References 31-35. 20. Yancey CM and Linsenmeier RA: Oxygen distribution and 1. Brown K.T, Watanabe K, and Murakami M: The early and late consumption in the cat retina at increased intraocular pres- receptor potentials of monkey cones and rods. Cold Spring sure. Invest Ophthalmol Vis Sci 30:600, 1989. Harbor Symp Quant Biol 30:457, 1965. 21. Linsenmeier RA, Goldstick TK, Blum RS, and Enroth-Cugell 2. Tornquist P and Aim A: Retinal and choroidal contribution to C: Estimation of retinal oxygen transients from measurements retinal metabolism in vivo: A study in pigs. Acta Physiol Scand made in the vitreous humor. Exp Eye Res 32:369, 1981. 160:351, 1979. 22. Stefansson E, Wolbarsht ML, and Landers MB III: In vivo O2 3. Linsenmeier RA: Effects of light and darkness on oxygen dis- consumption in rhesus monkeys in light and dark. Exp Eye tribution and consumption in the cat retina. J Gen Physiol Res 37:251, 1983. 88:521, 1986. 23. Flower RW and Patz A: The effect of hyperbaric oxygenation 4. Dollery CT, Bulpitt CJ, and Kohner EM: Oxygen supply to the on retinal ischemia. Invest Ophthalmol 10:605, 1971. retina from the retinal and choroidal circulations at normal 24. Feke GT, Zuckerman R, Green GJ, and Weiter JJ: Response and increased arterial O2 tensions. Invest Ophthalmol 8:588, of human retinal blood flow to light and dark. Invest Ophthal- 1969. mol Vis Sci 24:136, 1983. 5. Landers MB III: Retinal oxygenation from the choroidal cir- 25. Craig FN and Beecher HK: The effect of low oxygen tension on culation. Trans Am Ophthalmol Soc 76:528, 1978. tissue metabolism (retina). J Gen Physiol 26:467, 1943. 6. Baeyens DA, Hoffert JR, and Fromm PO: A comparative 26. Whalen WJ, Ganfield R, and Nair P: Effects of breathing O2 or study of oxygen toxicity in the retina, brain and liver of the O2 + CO2 and of the injection of neurohumors on the PO2 of teleost, amphibian and mammal. Comp Biochem Physiol cat cerebral cortex. Stroke 1:194, 1970. 45A:925, 1973. 7. Ubels JL and Hoffert JR: Ocular oxygen toxicity: The effect of Appendix hyperbaric oxygen on retinal Na+-K+ ATPase. Exp Eye Res 32:77, 1981. The equations used to generate the hyperoxic curves of 8. Ruffolo JJ, Ham WT Jr, Mueller HA, and Millen JE: Photo- Figure 5 were: chemical lesions in the primate retina under conditions of 2 2 elevated blood oxygen. Invest Ophthalmol Vis Sci 25:893, d P,/dx = 0 0< x

Downloaded from iovs.arvojournals.org on 09/26/2021