The oxygen distribution across the degenerating retina shifted from nearly normal in the early stages of the disease to markedly abnormal in the advanced stages. We were able to study only a few animals in these later stages, but the decrease in metabolism appeared to be progressive and steady. Incorporating more animals in the study would have allowed us to refine the shape of the function relating
Q OR to disease stage, which might become slightly more concave or convex than the linearity shown in
Figure 4 . In this study, the control animals were not Abyssinians, but domestic short-haired cats that were presumably genetically diverse. Previous work has shown that the ERG of such standard cats and unaffected Abyssinians is the same,
18 and we show in the accompanying paper
13 that their photoreceptor numbers are also the same. We have no reason to suspect that the metabolism of unaffected Abyssinians would be substantially different from the control animals we used.
Decreases in
Q OR altered the oxygen gradients, allowing oxygen from the choroidal circulation to reach the inner retina. This, in turn, allowed the inner retinal P
o 2 to remain normal when the retinal circulation was attenuated. These points are emphasized by simulations of oxygen gradients, based on average parameters obtained from the diffusion model
(Fig. 6) . The simulations indicate that the normal oxygen gradient in the ONL, which usually brings oxygen from the retinal circulation to the photoreceptor IS, reverses in stage 2 of the disease. This gradient reversal demonstrates that choroidally derived oxygen, which is normally all used by the photoreceptors, can now reach the inner retina. From stage 2 forward, the choroid provides more and more oxygen to the inner retina. Because the central retina did not become substantially thinner until stage 4 in the cats used in the oxygen study, the distance from choroid to inner retina was taken to be the same in all but the stage 4 simulation. In stage 4, thinning of the retina contributed to increasing the transport to the inner retina beyond what would have been observed if the photoreceptors had been totally dysfunctional, but still present. Some thinning occurred, on average, in stage 3 (see
Fig. 7Hof Ref.
13 ), and so the simulation may slightly underestimate the effect of choroidal oxygen at this stage.
Another conclusion that is emphasized by
Figure 6is that, with the elevation of the minimum P
o 2 well above zero, or loss of the minimum completely, the P
o 2 at the choroid is not the limiting factor for metabolism, as it is in the normal retina.
15 19 Therefore, in retinas with degeneration, any variation of
P C between cats cannot explain the variation of
Q OR in the different disease stages.
The observation that vessels constrict severely in later stages of the disease provides one piece of information that oxygen autoregulatory capabilities remain largely intact throughout disease progression, even though the vessels are small. This argument is further strengthened by direct measurements of vascular reactivity to hypoxia and hyperoxia (Linsenmeier RA, et al. IOVS 2003;44:ARVO E-Abstract 543).
Our conclusions on the mechanism of vascular attenuation are similar to those in previous work. Noell
20 first suggested that vascular attenuation after loss of photoreceptors was dependent on oxygen. After destroying rods in monkey and cat retina with iodoacetic acid, Noell observed vasoattenuation essentially identical with that in RP. He reasoned that without photoreceptors, the metabolic demand of the retina would be lower, and so it would no longer need two circulations and would adapt. It made sense to him that the retinal rather than choroidal circulation would be lost, because retinal vessels were already known to be more responsive to metabolic factors, of which he mentioned high O
2 and low CO
2, than the choroidal vessels. Penn et al.
6 emphasized the importance of choroidal oxygen in the degeneration of retinal capillaries in a transgenic mouse retina, in which they found that systemic hypoxia for several days was able to prevent the loss of capillaries. Yu et al.
10 made intraretinal oxygen measurements in the RCS rat retina and reached some of the same conclusions that we did about the flux of choroidal oxygen, but they did not report Q
o 2 or the correlation with the ERG (see below). The thinning of the retina in the RCS rat appeared to proceed at about the same rate as changes in the oxygen gradients, whereas the percentage of change in thickness was smaller than the oxygen change in cats. Although there may be subtle differences among animal models, the feature of attenuation and loss of the retinal vasculature is common to all animal models of RP,
2 3 4 5 6 is prevalent in human RP
1 as well, and probably has fundamentally the same cause.
An alternate explanation for some of these data is that in the absence of photoreceptors, inner retinal metabolism decreases, and, because the retinal circulation is sensitive to metabolic demand, it shuts down, whether or not oxygen is coming from the choroid.
1 The fact that hypoxia preserves the retinal circulation in animals with photoreceptor degeneration
6 argues against this explanation and indicates that the inner retina still needs oxygen from some source. In addition, the present work provides the only direct evidence of any kind that the absence of photoreceptors does not change inner retinal metabolism to any great extent. This is somewhat tentative, because it is based on comparing inner retinal Q
o 2 in one stage 4 cat to inner retinal Q
o 2 in normal cats with retinal artery occlusion,
17 but it is based on multiple measurements in that one cat. It may initially be surprising that inner retinal metabolism would not be affected by the loss of photoreceptors, but consider what happens in the outer plexiform layer. Either all the photoreceptors will behave as if they are in darkness and release glutamate more of the time, or else they will become damaged and not release as much glutamate. In either case, they would produce less modulation of glutamate. In either case, because of the presence of parallel ON and OFF pathways, roughly half of the bipolar cells and other inner retinal neurons would experience steady depolarization, and the other half would experience steady hyperpolarization. The metabolic effects of more depolarization of some cells, which would be expected to increase their metabolism, and more hyperpolarization of other cells, which would decrease their metabolism, could well cancel out. It is more difficult to analyze the situation if and when the bipolar cells’ glutamate receptors disappear.