Dr. Radcliffe: Well, hello. My name is Nathan Radcliffe. I'm at Cornell, Weill Cornell Medical College in New York City, and my practice is glaucoma. And, this talk is really about how my practice has changed just in the past few years
In fact, when I did my fellowship, we didn't have Cirrus HD OCT, which I'll be talking a lot about, and we managed things differently, and it's just been a few years. So, this talk is about how new technology, new ways of interrogating the optic nerve have changed the way that you manage glaucoma on a daily basis. And, so I'll be talking a little bit about glaucoma, about Cirrus, and about the management, and will present some cases and look at how all these things interact
I think in 2011, we defined glaucoma as an optic neuropathy. At which point we make that diagnosis, that's changing. We're using--thinking much more about the retinal nerve fiber layer today than just a few years ago
Deciding who to treat may be influenced by our sensitivity, and deciding who not to treat I think is equally important. And, I think a device that helps with glaucoma diagnosis can be most useful if it helps you exonerate your glaucoma suspects
And, when I came into my practice with Cirrus, I think one of the greatest patient benefits has been a large number of people with physiologic cupping and completely normal retinal nerve fiber layer for whom I've been able to stop therapy
And, then there's the issue of knowing when our therapy--our chosen therapy is sufficient or insufficient, so we're going to talk about how all these things interact
I think we have a really nice projection system right now, and I'll show this picture of glaucomatous optic neuropathy. What do you see? Well, you know, it's not just about the optic nerve. It's about the peripapillary retina. We know things like peripapillary atrophy are important
In this nerve, do you see two RNFL defects? I mean, the subtlety of this photograph is just beautiful to me, but here's one that's right near the macula, and here's the larger, broader RNFL defect
We know that OCT has evolved in terms of the ability for us to acquire high quality, high resolution scans. We also can acquire those scans more quickly now, so we can get volumes of data rather than just line scans
But, I'll argue--and I'll show you some examples from Cirrus--that the greatest evolution of OCT isn't the acquisition of higher quality data. It's software developments that allow us to do magnificent new things with the acquired data
So, let's look again at this optic nerve. With Cirrus, what you have is a cube of data that is collected, and this cube is 200-by-200 A scans over a 6-millimeter area in the back of the optic nerve
And, this is point number one, is that all of the data concerning the optic nerve is captured. That data is stores. As new software developments come, that data can be reanalyzed. So, unlike some other systems where you're getting a line scan, if you look at that data a day later and say, "You know what, that line scan wasn't center or was at a bad angle," that's not a problem with Cirrus. You could fix that. You could go back and look at the data differently
In fact, one of the things that Cirrus does that's quite unique is when it collects a cube of data, the first thing is does is it reanalyzes--or it determines the plane of data analysis, which may not be the plane with which that data was collected. So, if you have a tilted optic disc, you can realign that cube of data and then cut it perpendicular to the plane of the retina. That's an advantage for tilted discs. It helps make sure that you're not misinterpreting regions of the neural retinal rim
But, in any case, in this optic nerve, two RNFL defects--easy to see if you have a nice monitor, harder to see in different illuminations or different situations, almost impossible to see clinically, in my opinion, and very easy to see on a retinal nerve fiber layer thickness map
It corresponds very nicely to this double arcuate scotoma here. This is related to that macular RNFL defect. And, here's the broader defect related to the super temporal RNFL defect
So, now that we can a little bit better nail down the precise anatomy around the optic nerve, we have better structure/function correlation
When you subtract that RNFL thickness map from the normative database, you get a deviation map, how this patient deviates from the normal--and, again, two RNFL defects
The circle scan we have kept because it's become part of the language of interpreting OCT. In the case of Cirrus, the circle scan is a small fraction of the acquired data. But, since it's become the standard, we keep that data. But, note, you know, in certain cases that an RNFL defect may not be at its greatest at the arbitrary location where the circle scan interacts it, and that can have some clinical consequences
I think I've got an example of that. So, here's the circle tomogram. It's reconstructed in the case of Cirrus. And, again, the alignment is automated, so you don't have to worry about--not only do you not have to worry about the circle scan being misaligned--if it were misaligned, you could reconstruct it. But, thirdly, no matter what that alignment is, it will be the same on subsequent registered data cubes, so you have consistency, and that's important for glaucoma progression. Of course, we can still take that same circle scan and look at it compared to a normative database
And, you know, just to show you how easy it is to see these two defects, they're much more subtle here looking at the RNFL thickness TSNIT plot. And, for me, you know, this happens instantly. I know exactly what I'm dealing with with the deviation map. I can still read a circle scan; it just takes a little more time. It's not quite as intuitive. So, as I said, tighter correlation between structure and function, structure/function analysis
As Cirrus has evolved, new analysis tools have evolved, so about this time last year, we got neural retinal rim normative data for the Cirrus
Actually, when the Cirrus first came out, it was just the data cube, the RNFL thickness map, the deviation map. Then, we had tools to look at the neural retinal rim thickness. Then, we had the normative database. The point is that as all this software has evolved, the data acquisition protocol has been the same
There are some other high-definition OCT platforms out there where they haven't fully evolved their acquisition protocol yet or their glaucoma management tools. Very excellent images from other machines, but no protocol for collecting data. The problem is we have these machines for two years, the new software development comes out, says, "Oh, start collecting a different cube of information. Stop doing line scans, start doing data cubes.
All that data you've collected up until that point isn't going to be useful, whereas with Cirrus, started collecting the data around the optic nerve--as the software evolves, it's the same acquisition protocol. You didn't lose any data, and you can apply new analysis tools to the previously collected cube. So, it's a stable platform from that standpoint
So, here I'm showing you a patient with glaucoma. This patient has superior and inferior RNFL loss in both eyes. Let's see if I can get my pointer involved. And, here's what the macula looks like. So, this, again, is the standard macula protocol that's been out since the machine was invested, but recently they came up with a ganglion cell analysis and you can see how glaucoma shows up in the macula, and you can look at how the structure of the RNFL loss makes its way into the macula
And, again, the same language is being applied to macular OCT that we're applying to the RNFL. We have the ganglion cell analysis and then a deviation map where that ganglion cell tissue is being subtracted from a normative database
We have guided progression analysis for Cirrus OCT. You can track the retinal nerve fiber layer thickness over time. And, I'll show you some of the different analyses that you can do with this. But, this also can evolve with no change in the acquisition
So, this was--these are some scans. This is new software that's not FDA approved at this point in time, but it's hopeful that it will be shortly. You can actually look at optic nerve parameters with Cirrus and how they do over time, so you can look at whether the RNFL is thinning. You can calculate that rate. Same with the neural retinal rim
And, you can imagine in the future, perhaps, you know, you'll be able to look at the ganglion cell analysis and how that does over time. Again, stable acquisition protocol with new software to interrogate a cube of data around the optic nerve
And, then there's the combined report, which is available through Forum. This is basically just putting everything together from a structure/function standpoint. You can look just directly at, you know, the visual field defect, and on the same page, you can see how that correlates to the RNFL defects as well
So, how do we put this into the way that we manage glaucoma patients? Well, I'll just start by saying that when I initiate treatment in a glaucoma patient, I'm expecting them to progress at some point. I think glaucoma progression is going to happen even with adequate treatment
This is some recent data from the CIGTS study. You know, it was initially presented that, on average, CIGTS patients didn't have much visual field progression over the course of this study. These are the--CIGTS was a study where patients were randomized to trabeculectomy or medical IOP reduction. Trabeculectomy got to about 45 percent pressure reduction. Medical therapy, half of which had some laser, got about 35 percent pressure reduction
Well, by nine years, 23 percent of the medicine group, 34 percent of the surgery group had what you might consider a clinically significant worsening of the field, three decibels. The authors did not find that maintaining the pressure below 16 or 18 resulted in less visual field loss
Well, what does this mean? It's hard to predict who's going to get worse, and how much pressure reduction you've got may not be that helpful, so you have to have other strategies for determining, you know, which patients the therapy is insufficient
So, I'll show you a little bit about risk factor analysis, which is one tool we use, which I think has some limitations. And, I'll show you how Cirrus actually aides us with more precise disease staging and monitoring for progression with trends and events
So, just a little bit on risk factor analysis. Oh, this is a--this is interesting. So, apparently from Mac to--well, you can see what's going on here. We have different symbols. The concept here is we've looked at risk factors across trials
Age, for example, is one of the best known risk factors for glaucoma progression, but it hasn't been the same in all the different studies. So, for the collaborative normal tension glaucoma study, age wasn't a risk factor for progression. IOP at baseline, not always a risk factor. Of course, when you go from different populations, risk factors determined in one population may not apply to another
Let's look at the concept of the risk calculator. This came out in the ocular hypertension treatment study, and now you can take a patient with ocular hypertension, plug in some variables, and determine what their five-year likelihood of developing primary open-angle glaucoma is
And, just here's an example of a patient--or here's an example of the risk factors. Here's a clinical example--55-year-old, pressure's around 23 or 26, cornea's around 530, 550, 0.4 cups--pretty good fields. Seventeen percent chance of developing glaucoma in five years
That's one way of managing an ocular hypertensive, but think about what you tell that patient. "If I had six of you, one of you would go on to get glaucoma." Does that help you treat that patient with ocular hypertension? It's really hard to know
Let's look at more precisely staging a patient with ocular hypertension. So, here you have full visual fields. Pressures are 26, let's say. And, here are the optic discs. Well, again, I think we have a very nice monitor here, so you may detect a little RNFL defect inferotemporally there, even though there's really--doesn't appear to be any cupping
Look at that left optic nerve and agree with me, perhaps, that there does not seem to be much cupping there clinically
Here's HRT. Any ocular hypertension treatment study--there was an ancillary study on the value of HRT, and it showed that if you had an abnormal HRT at baseline, that gave you a higher likelihood of getting glaucoma. I just have to point out that this was a normal HRT, no cupping was detected, and that that RNFL defect, which is actually visible, was not detected by the HRT
Well, you say, "Well, HRT is a topographer of the optic nerve. It's not looking at the RNFL," and I would agree. There is this retinal nerve profile analysis with the HRT seen right here. It's really not retinal nerve fiber layer thickness. It's a reference plane, and then what's the topography of the peripapillary retinal
As you can see here, it did not detect this RNFL defect inferotemporally, but it did detect an abnormal area in the profile of the peripapillary retina in the fellow eye, which does not have an RNFL defect
Here's the Cirrus, and a number of points are to be mere here. Number one, average parameters are still pretty normal with this eye. A pretty small optic disc, which may explain why we don't see that much cupping. And, that's something that you can use with Cirrus. You can use Cirrus to measure the size of the optic nerve--explains why a lot of people with apparent cupping really just have large discs and may not have glaucoma
But, let's look specifically. I mean, again, this RNFL defect is, in my opinion, impossible to miss. It shows up on the deviation map, but note where it falls in relationship to the circle scan. The circle scan, which is, you know, an arbitrarily chosen distance from the optic nerve, just misses that abnormality
In people, this will vary. In some patients, the RNFL defect falls within the circle scan. Theoretically, it should be present in everyone, but it is
But, furthermore, look at the neural retinal rim thickness in the left eye. It's lower. The left eye is the dotted line here. So, there is loss of neural retinal rim tissue. It corresponds to the region of the RNFL defect. This is a normal field
The point that I'm making here is that we don't need to do risk calculation to determine if this ocular hypertensive is going to develop glaucoma, because they have glaucoma
So--and, this is how I manage ocular hypertension now. I used to do the risk calculation, and I--you know, go over, well--and, you know, if you think about it, you have a 1-in-6 chance of developing glaucoma, so if I treat you, my treatment's not 100 percent successful, so I have to treat about 12 people to prevent one case of POAG
If I just get a more sensitive way of staging them, I can pick up pre-parimetric glaucoma. I can have confidence that that's their diagnosis. I treat people whose pressure is creating some level of damage, and I don't treat the people who aren't developing any damage, and it allows me to be much more specific, treat fewer people, and probably also treat the ones with real disease a little more aggressively because I have confidence that I'm treating something--a real disease and not just treating a risk. We don't want to treat risk; we want to treat glaucoma
So, basically just making these points here, the number needed to treat when you're treating risk is going to be high, 12 people perhaps. But, if you stage that patient, you know they have glaucoma, even accepting that maybe, you know, if we lower pressure in two patients, it'll only prevent one of them from progressing, you decrease that number needed to treat to just two. So, more precise disease staging is a great way to resolve ambiguous issues on how aggressively to treat or who to treat
Other problems with risk factors--you know, the disc hemorrhage was found to be a powerful predictor of glaucoma progression in the ocular hypertension treatment study, but most of the people in OATS who had a disc hemorrhage never got glaucoma. In fact, there were more clinical centers in the ocular hypertension treatment study than there were disc hemorrhages that ended up with a POAG end point
So--and, I've spent a lot of time--disc hemorrhage is an area of my research. It's just not that specific a sign for glaucoma. We wish it were. It's a risk factor. The problem with risk factors is they don't give you all the information you'd like to have
Another example--thin cornea is a risk factor for glaucoma progression, but not in every study. It was in OATS, and in EMGT, thin cornea was a risk factor for progression, but only if you had a high pressure, not if you had a lower pressure. So, you get into really complex situations in--where it's not clear how to use a risk factor
Age is another one. What do you do with that? You can't affect anyone's age, and older people may have less time to live with the burden of their disease, so does it cancel itself out
Well, let's consider an alternate strategy. Given that the past is the best predictor of the future, and a person's biology is probably--their personal biology is probably more meaningful than the risk factors we've identified, let's just look at how this person has been doing and decide how aggressive their therapy should be based on their own performance
We have a couple of different types of progression here. And, I just want to make sure I'm not running over. We have events. A glaucoma progression event occurs when the patient's change exceeds the expected test/re-test variability. It answers the question has progression occurred in this patient. And, then we have trends. Trends are determined when a test value is deteriorating at a rate that can be described. It takes about five tests to see a progression trend
So, let me show you some Cirrus here. These are the RNFL thickness maps, and we have two baseline tests and then we go forward, two follow-up tests. Instead of looking at a deviation map that I was describing earlier, how does this patient deviate from the normal, we now have a map that's going to tell us how this patient deviates from their two baseline exams
And, when you think about how variable the optic nerve head is in all the patients, who better to use as your so-called normative database than your own performance? It allows people who are unique, who have unique anatomy, who have tilted discs, peripapillary atrophy to be compared to themselves, which is probably who you'd really want to compare them to
Of course, we're familiar with the Humphrey tools. Here, we have the guided progression analysis alert, so you see several points where this patient has deteriorated
And, then finally, we have the VFI plot, visual field index, where we can look at a linear deterioration of a global measure of the visual field performance over time
So, I'm showing you a patient with a series of visual fields here that have deteriorated over time, and look at this patient. His risk factors are identical between his right and left eye. The pressure is the same between the right and left eye, the corneal thickness is the same, but the two eyes are doing very differently
In the top eye, his right eye, based on this visual field index, he's likely to be parametrically blind at age 50. It's a young person, age 36, I think, or age 27. In the left eye, though, he's progressing much more rapidly and could be blind in six years
The point is, risk factor analysis won't help you differentiate these two eyes, but looking at his performance over time lets you know which eye needs the surgery first
Here's a 31-year-old with treated ocular hypertension, thin corneas, but doing well on therapy. And, again, I'm just showing you the power of these analyses--two baseline exams, a number of follow-up exams. It's a beautiful case because this patient's really been followed consistently, and I think this is important
Four patients in whom were concerned about progression, developed good habits, get adequate numbers of parametric evaluations, repeat OCT. For me, twice a year is a good number in a patient like this, but it can be done more often for patients where you're more suspicious
And, what you see over time is that this--his deviation from his own baseline becomes significant, and, you know, this is exactly where glaucoma happens inferotemporally. It's very characteristic
Note that he has a good amount of peripapillary atrophy and that the OCT handles this perfectly fine. And, this is my consistent experience with Cirrus, that Cirrus handles peripapillary atrophy beautifully, and it does that because it defines the termination of the optic disc through a signal that isn't affected by peripapillary atrophy. And, here you see he's evolving a superior arcuate scotoma
You can also look at these circle scans and look for specific regions that have deteriorated, and that's a progression event. So, you have the baseline 1 and baseline 2, and then his follow-up, the C exam here, has deteriorated and he has lost some RNFL. And, this is just his third Cirrus OCT, but you were already suspicious about progression
Linear trends of the RNFL over time can be determined. And, again, he's losing about three microns of RNFL a year. It's not a tremendous amount, but it's useful given how young he is, and you can look at the amount of deterioration and determine how aggressive the therapy needs to be. Maybe this particular patient just needs an adjunct medicine
Again, visual field, multiple examinations over time. Not a dramatic rate of progression, but one that is statistically significant. He's losing 1 percent--0.8 percent of his vision overall per year
Another example--this is a 72-year-old man, primary open-angle glaucoma, and I would argue that this is someone that most of us would think might need surgery. He's--his pressure is 21. He's on three different drops. He has thin corneas. He's had disc hemorrhages in both eyes, and he's got a good bit of damage in the right and left eye
Well, does he need surgery? His pressure is borderline. He's got all the risk factors. How's he been doing? Well, you know, over the past three or four years, not much has been happening. He is progressing a little bit. Now, it's not statistically significant, and that's going to be a function of how much repeatability he has, of how many tests you get, of how long you've been following him
To me, though, when I look at this, it changes the conversation I'm going to have with him. "You know, I think you're getting a little bit worse. It doesn't look like you're going blind in the next couple of years." Maybe we'll talk about laser instead of incisional surgery and see how he responds to that information
But, I think the day of looking at those risk factors and say, "I think you're going blind, but I don't have any information on how quickly"--I think that day is gone, and I think that's a good thing because patients need more specific information so they can help decide whether or not to have surgery
So, in concluding here, I hope I have convinced you that glaucoma management's shifting and that we now have a lot more information and the way that we interrogate that information can help us make better informed decisions. We can better characterize the disease stage. Patients who aren't great at taking visual fields--we have something that can really help us. We can determine whether progression has occurred with event analysis, determine how quickly it's happening with trend analysis, and these strategies can provide useful, powerful prognostic information to guide glaucoma management
And, I'm happy to take any questions