Better Glaucoma Progression Tracking

New technology boosts objectivity in assessing glaucoma progression.

BY DIANE DONOFRIO ANGELUCCI

Accurately detecting glaucoma progression is important in effectively managing the disease. The problem remains, however, that all changes detected with perimetry and imaging technologies do not necessarily indicate true progression.

The good news is that technology is becoming ever smarter in providing more objective information to help you determine whether your patient's condition is progressing and whether your treatment strategy is sufficient. But which technologies will guide you best? Here is what experts have to say about implementing available technology for this challenging task.

Functional Tests

Visual field testing has been a mainstay in helping clinicians assess glaucoma progression. Before the introduction of statistical software, ophthalmologists spread out a series of visual fields on a desktop to subjectively judge whether visual fields were worsening, says Joel S. Schuman, M.D., F.A.C.S., Eye and Ear Foundation professor and chairman, Department of Ophthalmology, University of Pittsburgh School of Medicine and director, UPMC Eye Center. Subsequently, Statpac and Statpac II statistical software was introduced to help ophthalmologists make more objective decisions.

Even with advances, however, changes detected on visual field tests may reflect variations in patient responses, physiology and interpretation. With test fluctuations in a single patient, it can be difficult to pinpoint true progression and make treatment decisions.

"In my experience, a lot of progression gets overcalled just from normal variation — the patient has a bad day, the test giver has a bad day," says Robert J. Noecker, M.D., M.B.A., director of the glaucoma service and vice chairman at the University of Pittsburgh Medical Center (UPMC) Eye Center. "There's a lot of subjectivity that's involved in functional testing."

There is no gold standard to answer the question of whether visual field progression has occurred, says Angelo P. Tanna, M.D., director of the glaucoma service at Northwestern University Feinberg School of Medicine, Chicago. "Think about the clinical trials that have been done, like the Early Manifest Glaucoma Trial (EMGT), Collaborative Initial Glaucoma Treatment Study (CIGTS), Advanced Glaucoma Intervention Study (AGIS), the Normal Tension Glaucoma Study (NTGS). All of these carefully designed clinical trials used different criteria for defining visual field progression," he says. "Studies have shown that when the different criteria are applied to a collection of serial visual fields, there is poor agreement regarding whether progression occurred. For example, if EMGT criteria demonstrate progression in one set of serial visual fields, AGIS criteria may not."

Glaucoma Progression Analysis

To reduce subjectivity in assessing glaucoma progression, several technologies are being used. Carl Zeiss Meditec introduced the Humphrey GPA (Glaucoma Progression Analysis, Carl Zeiss Meditec, Dublin, Calif.) in 2005 for the Humphrey Field Analyzer, based on the Early Manifest Glaucoma Trial. The GPA printout helps clinicians see at a glance whether there is evidence of visual field progression.

"One of the most important challenges in the detection of visual field progression centers around the fact that the threshold sensitivity measured at a particular location in the visual field may vary each time it is measured, even in a normal eye," Dr. Tanna says. "What makes matters worse is that in eyes with glaucoma damage, the magnitude of this fluctuation is even greater. The more severe the glaucoma damage and the farther the testing location is from fixation, the greater the magnitude of anticipated fluctuation. Therefore, if one observes deterioration in threshold sensitivity from test to test, one cannot be immediately certain whether that change represents true progression or normal fluctuation."

In developing the GPA software, researchers generated a database of the normal magnitude of fluctuations that occurs in patients with varying degrees of glaucomatous visual field damage at various locations in the visual field, Dr. Tanna says.

Using this database, the GPA software flags locations in the visual field where the deterioration observed exceeds the amount that was observed in stable glaucoma patients used to generate the database. In the EMGT, visual field progression was defined as deterioration in the same three or more locations in the visual field on three consecutive tests. GPA software is designed to define progression using the same criteria, Dr. Tanna says.

To detect change accurately, one needs to have two reliable baseline visual field tests after the patient has scaled the steep portion of the learning curve. These should be obtained early after diagnosis, Dr. Tanna says. Then changes observed should be verified with two, or preferably, three, visual field tests (Figure).

In his practice, Donald L. Budenz, M.D., M.P.H, professor of ophthalmology and epidemiology and public health, Bascom Palmer Eye Institute, University of Miami Miller School of Medicine, has used a beta version of the latest upgrade of the software, which combines event analysis with trend analysis. It looks at a new global index, the visual field index (VFI), over time. It also factors out the effect of cataract. "It was very helpful in showing rapid rate of decline in some patients, where maybe you wouldn't have picked that up," he says. "It can help you focus on the rate of glaucomatous loss as a whole rather than just a few individual points that may not have as much impact on a patient's quality of life over time."

"The VFI allows you to look at the mathematical probability of the state of the visual field up to 5 years hence," Dr. Schuman says. He believes this is a useful parameter for the clinician, providing a sense of the patient's current and potential status based on regression statistics and the patient's age. In addition, it is a valuable teaching tool in talking with the patient. "If you're having a discussion with the patient about advancing their therapy and they can see that here's the state of their visual function now and here's where it could be as predicted by the mathematics 5 years from now, that can be a powerful motivator for a patient who might be hesitant to move onto the next therapeutic step, especially if the next therapeutic step involves surgery," he says. However, Dr. Budenz says, this assumption has not yet been tested.

Figure. Sixteen visual field tests were performed over an 8.5-year period in this patient. The two baseline visual field tests (p. 58) and the two most recent visual field tests are summarized in the GPA analysis. The intervening pages are not shown. The most recent visual field test contains three locations in which threshold sensitivity is substantially worse compared with the mean of the two baseline visual field tests on three consecutive tests (solid black triangles). An additional two locations are worse than baseline on two consecutive tests (half-black triangles). Three other locations are worse only on the most recent test (open black triangles). This series of visual field tests meets the minimum progression criteria for the EMGT. Accordingly, the GPA software has labeled this most recent field as disclosing "likely progression."

In addition, although GPA reduces variability, it provides only a piece of the puzzle. "The danger I think is looking at some of these functional tests in a vacuum and making a decision on just that information. Because the studies have shown us that, from OHTS on down, that a lot of times you come back another day and it's back to where it was at baseline a year ago or whatever the interval might be," Dr. Noecker says.

Dr. Tanna agrees that it is important to take the entire clinical picture into account. Because there is no gold standard, he says, the true sensitivity and specificity of GPA to detect progression are unknown. There clearly are times when a patient who seems stable in every other regard appears to have progressed using GPA criteria. "Additionally, in some circumstances, glaucoma progression can be diffuse, and if that is the case, it is possible for the total deviation or the mean deviation to deteriorate without deterioration in the pattern deviation," he says. "That's one situation in which true glaucomatous progression can be entirely missed by the GPA software since it evaluates pattern deviation data only."

Moorfields Progressor

To assist in detecting progression, researchers at Moorfields Eye Hospital in London developed a software program several years ago called Progressor that performs point-wise linear regression analysis, which provides a color-coded analysis of change at individual points in the entire visual field over time and the statistical significance of that change. "It's different from the GPA in that it gives you more information about individual points over time," Dr. Budenz says.

Because the system is not incorporated into current visual field machines, however, the clinician needs to take data from the visual field machine, download it to a desktop, and run it through the Progressor program, which can be cumbersome in a busy practice, Dr. Budenz warns.

Octopus

Last year the Octopus 900 Series was released (Haag-Streit USA, Mason, Ohio), which offers standard automated perimetry, short-wavelength automated perimetry (SWAP), Goldmann kinetic perimetry and Flicker perimetry.

The Octopus G1 Glaucoma program has test locations in sensitive spots to prevent misinterpretation, with 59 test locations in the central 30 degrees, as well as points outside the central 30 degrees to help detect nasal defects. Information gathering is split in four groups called phases; the technician can stop at any point if the patient becomes tired. The Octopus G2 program tests only the central 30 degrees.

Octopus sells a software system, the Field Analysis (OFA), which includes the Octopus polar graph — the bridge between structure and function — and a cluster analysis. OFA supports the evaluation of white/white and blue/yellow visual fields based on the G1 and G2 programs.

"Haag-Streit states the new cluster analysis has proved superior to global and local progression analysis," E. Randy Craven, M.D., director, Glaucoma & Cataract Consultants of Colorado, Littleton, and associate clinical professor of ophthalmology, University of Colorado School of Medicine, Denver, says. "The combination of cluster trend analysis together with the Octopus G pattern even helps determine if there is a future risk of progression. This is similar to the GPA from Humphrey."

Additional Methods

Other techniques for assessing progression using visual fields include point-wise analysis, where the clinician looks at the threshold printout on serial fields to see whether two individual points are worsening by 6 to 8 dB from prior fields on two subsequent fields. "That's the criteria used by the NTGS and has been found to be very good, and you can do that without software help," Dr. Budenz says. However, it does not adjust for cataract effects, which may be misleading, and there are no statistical parameters, he says.

Clinicians also can look at the mean deviation over time; however, changes will not indicate whether visual fields are declining because of a cataract and findings may be affected by other diseases. If there is no other reason for the decline, it's assumed that glaucoma is the cause. "That's probably the least sensitive of diagnosing glaucoma progression, but it's something we use," Dr. Budenz says.

Imaging Devices in Detecting Progression

In assessing structure, three commercial technologies — the Heidelberg Retinal Tomograph (HRT) (Heidelberg Engineering, Heidelberg, Germany), scanning laser polarimetry (GDx) (Carl Zeiss Meditec) and optical coherence tomography (OCT) — have been shown to be roughly equivalent in discriminating between health and disease while speeding up the process in determining if a patient has glaucoma, Dr. Schuman says.

However, fewer data are available on how effectively these devices measure progression. "With imaging we have much younger technology. The progression assessments with these technologies seem to be more sensitive than functional assessment, but they don't necessarily overlap with functional assessment or even with each other," says Dr. Schuman. That may indicate that the technologies are measuring something different, he says.

GDx Analysis

Although Dr. Noecker finds the GDx useful for imaging the retinal nerve fiber layer (RNFL) to identify defects, he says it is limited in assessing progression because it assesses only the RNFL. "The good thing about the variable corneal compensator is that it's taken away a lot of the artifact that the original GDx was associated with in terms of axis or polarization errors due to the cornea, lens, vitreous, etc.," Dr. Noecker says.

However, Carl Zeiss Meditec has developed new guided progression analysis software for the GDxVCC, says Dr. Craven. "You can detect if there are changes in a region of the nerve fiber layer that represent progressive loss much better than any other instrument that is out there right now for detecting progressive thinning or loss of the nerve fiber layer," he says. Dr. Craven, who has not yet used the software upgrade, anticipates that it will enable the clinician to assess change in several ways with the GDx.

Heidelberg Retinal Tomograph

Using confocal scanning laser ophthalmoscopy, the HRT creates a three-dimensional image of the optic nerve.

"If we're starting to detect progression before we see the visual field change, which is really what the key is in glaucoma, I think the strongest software nerve analysis programs that are out there are with the HRT," Dr. Craven says. "The HRT has really improved, especially with the HRT 3, where they have software now that aligns the optic nerve head, number one, so there is no rotational or torsional factors that are interfering with the ability to look at the nerve head. And then it uses the reference to realign things so you end up detecting if there is a topographical change in the optic nerve head over a period of time that is statistically significant."

The printout also shows both optic nerves on one page, and the progression analysis presents information in two ways, Dr. Noecker says. Areas of the optic nerve losing mass or thinning are shown in red, which darkens if more thinning occurs compared with baseline, which makes it easy to check whether damage is worsening.

"With confocal scanning microscopy, we have the most mature progression-detection software," Dr. Schuman says. He explains that the HRT image analysis may indicate an abnormality undetected on the physical examination, which will prompt him to re-examine the patient.

Despite the progression-detection software, it can be difficult to detect change because it can be subtle. Therefore, Dr. Schuman looks for a functional correlate if he sees an area of change on an imaging analysis. "If there's progression showing up on both, then that pretty much cinches the likelihood that there is true progression and the patient needs to have more aggressive management of their disease," he says.

Dr. Craven agrees that additional information is necessary if a nerve head change is observed. "There's still some noise that we see in all these nerve head images, and I'll tend to try to get some other early visual field detector if it's in people that have pre-perimetric glaucoma," he says. In these cases, he uses SWAP or a frequency-doubled perimetry to find a correlation with the imaging results.

Optical Coherence Tomography

OCT provides quantitative results and gives the mean thickness of the nerve fiber layer in different areas, which can be useful in assessing progression, Dr. Noecker says. However, the most significant problem in using it to assess progression is that it is a manual process, he explains. "We basically have to compare nerve fiber layer thicknesses from one time point to the next," he says. "If you're in a place where the tests are done very well and they're very consistent, that may be okay. You have to be very careful with the OCT to center the images exactly the same way from test to test. Otherwise, as you move the imaging ring closer to the optic nerve, the nerve fiber layer at that location will appear thicker and the corresponding area on the opposite side will appear artificially thinner."

"We're working on developing progression software for the OCT, and we're providing Zeiss with the test/retest variability of Stratus OCT in hope that they will develop software that can detect change," Dr. Budenz says. "They're working on a trend analysis for OCT mean retinal nerve fiber layer thickness, which is analogous to trend analysis with mean deviation, which is something that we've been using in visual fields for 20 years. So we're at a very rudimentary stage in terms of our progression analysis with OCT because we're just looking at mean RNFL thickness rather than focal areas of structural damage. We're looking at regression line and trend analysis at this point. So I think that progression analysis with OCT is currently in its infancy."

Dr. Schuman is also working on the development of progression software for the OCT. "I think that in future iterations we will be able to have focal assessment of change, looking at the overall trend as in regression, but also looking at event analysis, analogous to the GPA on visual field," he says. He believes future generations of software will address whether glaucoma is worsening, whether there is a specific trend and whether there is a statistically significant difference between assessments.

"With spectral domain OCT having been introduced into the market, we're starting to see newer ways of assessing the optic nerve or the retinal nerve fiber layer in glaucoma and those technologies are rapidly evolving," Dr. Schuman says. "We are likely to see different ways of assessing the ocular structure both for discrimination of health versus disease and also for detection of change over time using three-dimensional OCT imaging."

Conclusion

"I think assessing progression is largely a software issue," Dr. Noecker says. "I think in a relatively short period of time we've already seen great improvements in software for both the functional and structural machines and we've made great progress."

"At present, the clinician should use all available information — visual field data, optic disc evaluation, and imaging data — to determine whether progression is occurring. And even if progression is detected, the decision to intensify therapy must be based on clinical judgment," Dr. Tanna says. "Glaucoma management is still a combination of art and science." OM

Editor's Note: Dr. Tanna is a consultant for Carl Zeiss Meditec, from which he also receives honoraria. Dr. Budenz has receives research support and honoraria from Carl Zeiss Meditec. Dr. Noecker receives speaking honoraria from Heidelberg and research funding from Carl Zeiss Meditec. Dr. Craven has no financial interests related to his comments. Dr. Schuman is an inventor of OCT and receives intellectual property licensed from Massachusetts Institute of Technology to Carl Zeiss Meditec. He also receives speaking honoraria and research funding from Heidelberg, Carl Zeiss Meditec and several pharmaceutical companies. He also receives research support and honoraria from Optovue.