Pattern Electroretinograms: General Discussion

PATTERN ELECTRORETINOGRAMS: GENERAL DISCUSSION Ivan Bodis-Wollner, Moderator Mount Sinai School of Medicine

City University of New York N2w York, New York 10029 VAECAN:The question to which I would specifically like to address myself is: Can we resolve more explicitly where the site of origin of this response is? We have heard Dr. Maffei suggest possibly a ganglionic cell basis for this and Dr. Sherman suggested some of his clinical data were incompatible with the gringlion cell origin, suggesting it was in the more proximal retina, but not the ganglion cells themselves. Dr. Armington and most other workers who have worked with the pattern electroretinogram have had a quite different assumption, which has not always been specifically tested, that is, that the pattern electroretinogram was merely a way of getting the photopic components of the ordinary electroretinogram. This is a response that is generated in the distal retina, and the idea is that if you have an alternating pattern then what you have are little local areas all going bright and dark simultaneously. But you are still recording the same thing that you would be if you summed the on and ofresponses of the ERG, and the ERG is generally believed to be generated by the distal retina. Dr. Armington has already pointed out to you that you can certainly get rid of stray light effects with alternating patterns, and that is a very good reason for using them. But there is another way of getting focal responses. Brindley and Westheimer showed that if you surround your stimulating area by a bright surround you can cut out the very late waves of the ERG and you will finish up with a very small early component, which is much tinier and much earlier than the late wave is usually considered to be. The main B wave, as Dr. Armington pointed out, is a late wave which reflects a scotopic function mainly due to stray light. Brindley and Westheimer usually use about 10%of their stimulus. There you have the late B wave and here is a very early B wave which Brindley and Westheimer and we shall call the focal ERG. All the ERGS have been recorded with bright surrounds a t about the same level of luminance, and all our responses will be ones that are the sums of alternate on’s and o f s , and we have been able to separate this sort of focal ERG from the pattern ERG in two ways: The first is experimental. The most remarkable thing about the pattern electroretinogram is it has an amazingly fixed peak time. W e have been able to manipulate the timed peak of this response in only two ways. One is a relatively trivial way, that is, with luminance, and the other is, 1 will show you later, with spatial frequency, in humans only. They have a very well organized macula with different cell populations. The pattern electroretinogram has a very peaked time as we measure it, and just to give you a difference from Dr. Armington we only measure the conventional A wave to B wave peak, although we do not know if these are actually A and B waves. You can see that statistically there is no difference at high and low contrasts with the pattern electroretinogram. We have done this very carefully. The focal electroretinogram is earlier than the peak time of the pattern electroretinogram. We get the

602

General Discussion

603

same sort of difference with the pigeon except, in the pigeon, the focal electroretinogram moves back by 10 msec. Otherwise the experiment is comparable. So, we have here experimental separation of the two responses. Now, we have some clinical work which seems to suggest, as Dr. Maffei and others have already told you, that there is a ganglionic basis for this response. We have recorded different reductions in the pattern electroretinogram in a large number of diseases and these include alcohol amblyopia, degenerative optic nerve disease, congenital optic nerve disease, and traumatic disruption of the optic nerve. In this last case, we saw the response was entirely absent, but I will not make much of that data because the subject had very poor fixation in both eyes. In early glaucoma there is a high correlation between the degree of glaucoma and the degree of loss. The response was just below 2 pV, where our average is 2%in six of nine. In seven of the eight where the glaucoma is asymmetrical, the smallest response was in the eye with the worst field and there appears to be a 0.9 correlation between both eyes, which is highly significant. These responses are for one degree per side checks, but, we get exactly the same results down to a quarter degree per side. This is not a high enough spatial frequency to see the high spatial frequency Professor Maffei showed. Dr. Sherman’s case would come in about here, but that is because he recorded both eyes separately and looked for nice, clear responses. If he had recorded both eyes simultaneously and taken the average, this case would probably drop down a bit. So, this data shows that the pattern and focal responses, even if there is very long optic nerve damage, are never totally separate, but, there is a component that drops out with degeneration. I would like to stress the origin of the pattern electroretinogram in more detail, but, I think time will prevent me. However, one thing I would like to point out is that the pattern electroretinogram, as I have said, is extremely fixed in its time to peak. This is, according to Dr. Shapley, unlike the behavior of ganglion cells, which are supposed to change the time of peak for up to 20 msec at low contrast. We are not confident this happens with checkerboard pattern reversal. Using microelectrode recordings in pigeon retinas, we have found that, with various contrasts, the time of peak stays absolutely constant with checkerboard phase reversal. So, I am not sure from this whether the experimental evidence is that the pattern responses are ganglionic. The literature will wait for that. It is certain that there may be a bit of luminous component in the pattern and a bit of pattern component in the luminous. Nevertheless, there is something in both that is different and we can get it in both experimental and clinical separations. L. MAFFEI:I am glad to see that most of you in the crowd are in agreement with my physiological data and I would like to point out in a general remark that it is important to distinguish between the physiological clinical use of a given electrical wave and what one can get out in terms of information in the field of physiology. I think that the origin of the ERG of the different waves are of concern. It is very rare in my opinion to reach to the animal experiment where the condition can be more clearly established and repeatable, and in this context I would like to ask Dr. Veagan if he has tried in the pigeon to cut the optic nerve, which is the obvious experiment to do, and see whether the ERG is changing or not, or if he has done other experiments in which it would be very useful to hear the physiological properties of the ERG of the pigeon? VAEGAN: That is a rather opening question. I shall not talk too much about pigeon research because for the very most part there is hardly any difference between the pigeon results and the human. They are extremely close and very similar and this is most unlike most of the features of the pigeon retina and the human retina.

604

Annals New York Academy of Sciences

We can, with the pigeon, where the recording techniques are much more stable, record a spetial frequency rate per degree which is higher than the behavior we measured in the resolution of the pigeon. We have done depth profiles through the retina in the pigeon retina with microelectrodes and we find exactly the same differences a t the retinal surface. We find that the focal response stays positive and rolls over fairly quickly. But, at a different rate, the pattern response rolls over and changes a t a different rate and a different manner than the focal response so that at different depths in the retina the two responses will be even more different. But, when you get very deep in the retina to the inner nuclear layer, the two responses have become identical and that is found at all positioning of the stimulus. We have also tried pattern manipulation in the pigeon to try to work out whether the response is dominated by spatial frequency components or by edges or whether a stimulating receptive field centers it. It is rather tricky. Basically we tried to match the fundamental frequency components of the checks, bars and sine waves. We find that if the retina was responding to the fundamental spatial frequency components, then after we matched the fundamentals, all these patterns should give the same response. In actual fact, the bars and the sine waves give exactly the same response, which shows that edges are not important. But, nevertheless, the checkerboard response shows a reasonance, and that is a t about a half a degree per side checks. After that there is a consistent difference between the checks, the bars, and sine waves. This is consistent with the predictions you would make if checkerboards are more optimally stimulating the center surround organization than are the sine waves and bars. Other than that, we have not been able to do simple manipulations such as cut the optic nerve because the pigeon head is just too fragile. There would be too much damage. J. SHERMAN: I would like to ask Dr. Maffei if it is possible that the flash ERG and the flicker ERG (which were recorded without a bright surround) might be coming from the periphery or the area outside of the area centralis of the cat whereas the pattern-reversal ERG is coming from the area centralis. Hence, one was present and one was not present. MAFFEI.The experiment was done like that. The flash was tested before and after in the way that I described. So, the flash is a usual flash and the pattern ERG was a total view, but, we did not do any localized stimulus where the luminance is concerned. You implied that only the central part of the retina could be involved, at least, in the early part. I think that is probably true. I have an additional comment: The pattern ERG that we have in the cat and in the human must be definitive. In the cat any recording is all right. In the human, before recording the ERG we were obliged to put the reference electrode in the other eye because if we did not do that and put the reference electrode on the forehead or any other place, in many instances we picked up the visual evoked potential. So, I feel that this is a very important point because for years we tried to record the ERG and always we had interferences with the visual evoked potential. J. C. ARMINGTON: 1 think the position of reference electrodes is always a problem. If you have a subject, such as myself, with alpha waves that are over a hundred microvolts, if you put the electrodes in the back of your head, and if you put an electrode on each eye and record, it looks just like anyone else’s EEG. Out comes the alpha and all those other things. SHERMAN: Some people might want to look at the pattern ERG as being a far-field potential from the cortex. Now, that is certainly possible. I would like to point out, however, that in the seven cases of optic nerve disease that I presented yesterday the

General Discussion

605

pattern ERG was present, but, the pattern VEP was extinguished. In these instances, there is no way one can argue that the pattern ERG that was present is a reflection of the pattern-reversal VEP, which was extinguished. M.TRICK: Why I have done is attempt to look at the effect of spatial frequency on the human pattern-reversal ERG and at the same time to look at the effect of temporal frequency. Essentially what I have done is replicate Dr. Armington’s study for various temporal frequencies using various spatial frequencies. In doing that I find, first of all, that, as long as we look a t the human at spatial frequencies higher than about 0.25 cycles per degree, we can get, using those points, a very good fit to visual accuity. 1 used 14 emmetropic subjects, and, depending upon temporal frequencies, the range of visual accuities I predicted with a linear regression to noise level (rather than to zero) ranged from 20/60 to 20/25 over these normals with the best prediction being obtained with 1.8 or 3.75 H z temporal frequency. Now, 1 would also like to comment relative to what has been reported as the fall-off or the temporal spatial tuning question relative to the pattern-reversal ERG. In doing my 14 normals, I noted that in some cases I did get a fall-off in amplitude for large checks and in other cases I did not get that fall-off for large checks. It disturbed me immensely that I was getting a lot of differences. I went and repeated these experiments using just one temporal frequency on a number of subjects, and in this case I used a number of different blurs and I blurred the image to the eyes by up to 12 diopters. I found in those cases when there was blur, that as long as I was using checks of higher spatial frequency than 0.25 cycles per degree I got a rapid fall-off in the amplitude of the response as a function of blur, but, for large checks such as 4-degree checks I got no significant difference in the amplitude of the response when I blurred the subject. So, I think this is an interesting point that we need to look at a little more extensively and it may be relevant to one of Dr. Spekreijse’s comments earlier this week about the local luminance changes. Conceivably, in that case we may be picking up some luminance changes. J. NELSON:I would like to address this to Dr. Vaegan. The results that you reported in amblyopia are in many respects similar to a number of recent reports in experimental amblyopia in cats in that ganglion cells are reported to have abnormal spatial resolutions in kittens reared with experimental amblyopia. I wonder if, maybe, you would like to expand on where you think the anomaly in amblyopia lies via the ERG anomalies that you have reported. VAEGAN: The reason that we looked at the deprivation amblyopes in particular, as some of you might be aware, is because a t present in the literature there is some degree of controversy about whether retinal effects occur in animal models of amblyopia, and up until very recently Dr. Ikeda was the only one to promote the idea that there might be a retinal change in amblyopia. However, deprivation amblyopia does not seem to work in animals. We certainly find deprivation amblyopia does not produce retinal change in animals and there are at least five studies where deprivations failed to have an effect. We, certainly, find that the majority of untreatable, amblyopes, or amblyopes failing to respond to treatment, have a retinal change, and it now looks a lot more like these cats and a lot more as if blur might be the important factor. Now, as far as deprivation amblyopes, I believe that they are to a very large degree deprived rather than blurred because they were very rapidly corrected, i.e., in most cases very, very soon after surgery and contact lenses were put in. The optics were always good apart from the time they had cataracts. So, they seem to be totally different than the cats. Most of the cat work to date has been a very short deprivation, and I think that if

606

Annals New York Academy of Sciences

one starts to look carefully a t long deprivation one will find retinal changes even in cats. My feeling is that a lot of the amblyopia that we are seeing may well have a retinal basis. There may be some cortical things as well and I am not prepared to say that you cannot have both. S. FRICKERWe have approached this from rather a crude clinical viewpoint, but we have been doing ERGS on a somewhat different basis using the cross-correlation method. This allows you to consider the frequency of stimulation in a little different manner. For instance, if you want to get photopic ERGS you stimulate the system rather rapidly, and by doing the cross-correlation process you get a waveform that looks rather like a photopic ERG, but, you are putting i n fairly high frequencies. We usually do it with flashes with a maximum of 50, an average of 25, and so on. So, in the course of a minute's test we get on the order of 1600 flashes in. We get a lot of signals and noise improvement, and we get nice clean signals. This is just what we want clinically because we want clean signals with minimum testing time and the ability to work through a lot of noise. This is fine for flash VERs too. When we come to alternating checkerboards, we have to slow down a little bit because we are constrained, as most of you are I guess, by the TV screen. So, we have adopted a rather sloppy method in which our maximum rate of alternation is 25; average, 1272; and it can be lower. It can be anywhere basically from nothing up to 25. This is further compounded by the fact that you do not have the basic timing for the cross-correlation method linked to the 60 Hz or the TV set or you get into trouble. The TV change is liable to have a bit of uncertainty in it-an average k 9 msec compared with the way you are doing the timing for your correlation. In addition, when the checks change, of course, we have different periods of time in which something is black and, then, later on when it is white. We have the subject sitting there looking a t this and the subject is probably, as most of our subjects are, moving around. He is not absolutely fixed, some of them less fixed than others. The net result of all of this is that the stimulus on the retina is a very variable sort of thing. It is variable in time. It is not ofequally as on for any given element. The elements themselves may change, and we have this added factor of frequency put into it. When we first started doing the pattern VERs we looked a t the pattern ERGS, and, I must say, on listening to same of the papers I would like to believe that there is a strong ganglion cell component. I am rather more inclined to agree with Dr. Spekreijse that there is a luminous component. We have obtained quite variable results. We get small noisy responses which do not have fixed latencies, but, which seem to vary quite a bit. In spite of our noise suppression with the system, which is really rather effective, the responses tend to be variable and noisy. Sometimes we get them and a t other times we do not. W e wrote them off, perhaps rather too quickly, as being luminance variations. Possibly, that is a mistake and we should look at them some more. But the implications from our results with a fairly large number of clinical patients is that these are slippery things to get one's hands on. But, if you want to put in the added factor of timing of frequency, or temporal variations, consider the cross-correlation method because this factor is built in. You will have low and high frequencies, and you can change it and you can also get a lot of noise reaction too. NELSON.In regard to where these new B waves are coming from, I think those of you who have not seen or read the Nelson, Zrenner, and Gouras paper, which was in the Morioka 16th ISCEV Symposium, should do so. That paper shows that if you take a profused cat eye cup and sprinkle it with little dots completely covering it and do an ERG, you will get what looks like a classical clinical ERG. Then, you take your dots

General Discussion

607

and you make them larger and make the spaces between the dots less, but, keeping the luminance completely constant, the ERG B wave slowly diminishes. When you reach a critical dot size, the ERG B wave disappears completely. If you keep the process up, you reach a point finally where a small much faster little B wave appears. Now, this is certainly in line with some of the results that we have been hearing today, and this paper is about two years old. BODIS-WOLLNER: What did you imply for those of us who might have read the paper, but, did not quite get the point? NELSON: Gouras’s notion is that there is spatial interaction a t the B wave level, but, in a completely different part of the retina than we have recognized, that is, at the ganglion cell level. BODIS-WOLLNER: In other words, those interactions would cancel out the B waves? NELSON:Yes. He has a model for receptive field antagonism. BODIS-WOLLNER: Are you suggesting, therefore, that there are certain spatial patterns which if they produce a pattern ERG would not produce a B wave and, therefore, the origin of the pattern ERG could be separated by this method? NELSON: That is right. It is certainly something to think about in terms of the different B waves we all seem to be dealing with now. SHERMAN: The first time I started to record these simultaneous pattern-reversal ERGs and VEPs in a small series of patients with bilateral optic nerve disease with acuities of 20/400 and worse, I could not get the ERGs. You have to remember that this response is typically 2 or 3 p V . These patients with poor vision have poor fixation. The poor fixation itself is going to wipe out the 2 or 3 pV response. You really have to be very, very cautious to say that the patternreversal ERG is not there just because you do not get it the first time you try. Now, in the seven patients that we reported yesterday we were very careful to make sure the patients were fixating carefully. Certainly, if you have a patient with monocular optic nerve disease it is much easier. The patient can fixate with the good eye through a mirror at a point out in space and you can stabilize fixation in the poor eye with the stimulus falling on the central retina. In that last case I presented yesterday, the blind eye was stimulated by the checks, while the good eye fixated. The normal or near normal pattern ERG was recordable from the blind eye using this simple trick. VAEGAN:If I could make a point about technique too for those of you who are interested in doing this type of thing. The practice in our lab is to use the Arden gold foil electrode, obviously because it is Dr. Arden’s lab. This is a very comfortable electrode to use, produces no rotations to the subject, and is very sensitive. The other important aspect of our data collection is that we always record from both eyes simultaneously and take the ratio between the eyes as the critical variable. Finally, our averaging techniques depend heavily on very good artifact rejection programs, and even those artifact rejection programs do not work unless we turn the gain of our system up to a point where we are rejecting somewhere between 20% to 50% of our traces. You have to make sure that your artifact rejection system is really working well and is being pushed to get rid of all of the noise.