IMAGING SYSTEMS
Lewis: Is there a clear advantage of one imaging system over another, and what is the latest information about their abilities to document progression?
Medeiros: I do not see any clear evidence of any one instrument's superiority over another, but each has advantages and disadvantages. The Heidelberg Retina Tomograph (HRT; Heidelberg Engineering GmbH, Heidelberg, Germany) has been available the longest, so there are many longitudinal studies demonstrating that the device is able to detect progression. For example, the Confocal Scanning Laser Ophthalmoscopy Ancillary Study to the Ocular Hypertension Treatment Study has shown that measurements with the HRT are able to predict the future development of glaucoma in people with ocular hypertension1 (Figure 1). On the other hand, the Stratus OCT and GDx (both manufactured by Carl Zeiss Meditec, Inc., Dublin, CA) can evaluate the retinal nerve fiber layer (RNFL) (Figure 2), which the HRT can only do indirectly. Although these two instruments have not been available as long as the HRT, the technologies are promising, and longitudinal studies evaluating these instruments should occur soon.
Lewis: In terms of diagnosing early glaucoma versus assessing more advanced disease, do you have a preference on technologies?
Medeiros: The assessment of the RNFL is fundamental to the early diagnosis of glaucoma, and I find the instruments that evaluate the RNFL to be helpful in this situation, because they provide complementary information to what we obtain from evaluations of the optic nerve at the slit lamp or with photographs. Because the RNFL is difficult to examine at the slit lamp or with photographs, these instruments can help us to decide whether or not a glaucoma suspect has early damage. It is important to note, however, that we currently have little evidence to determine the utility of these instruments in early versus advanced disease.
Schuman: In the real world, doctors have to get paid for services that they provide, and many payers are saying that these instruments are not useful in late disease and will not pay for their use. It is difficult for clinicians to get paid fairly for the services they provide. I think Felipe's point is important in that we do not really know if these technologies are less valuable in patients with advanced versus early disease.
RNFL imaging devices allow us to quantitatively and objectively measure something that we otherwise cannot. A clinical examination may reveal an area of RNFL dropout, but you do not know how much. These devices let you put a number on that. The HRT is the only instrument that has validated progression software. Data from late-stage-development progression software have already been presented at events sponsored by Carl Zeiss Meditec, Inc., during the 2006 AAO Annual Meeting as well as other more recent conclaves for the GDx and the Stratus OCT. The information presented to date shows promise.
Mills: As validation studies are laboriously conducted, we will accumulate the evidence that will allow us to answer your original question, Rick. We are in a growth phase for this technology. In the meantime, what has proven to be clinically useful is largely an anecdotal exercise, and we just have to accept that.
Schuman: I am not sure that I agree. There are studies showing that the HRT, Stratus OCT, and GDx are at least as good as an expert observer.2-8 Making these technologies available in communities where people rarely have their optic nerves examined could raise the standard to that of an expert observer in a standardized quantitative objective way, which I think has great value.
Brandt: Yet, how many of us have seen patients who were terrified, because they have what we now call red disease? Their test was "positive" and showed numbers in red. The clinicians who saw these patients handed them this printout and said, "You have glaucoma." They are not putting the imaging results into the context of the clinical examination.
Schuman: Clinicians must ensure that the test they have performed is of adequate quality to be evaluated. There are parameters on each of the devices to help determine if the quality of a particular test is good. If it is not, you cannot trust the result.
Weinreb: There are few published data on glaucomatous progression. These technologies are in great flux. The hardware generally has not been stable, and the software has changed even more. By the time a particular configuration of hardware and software is validated, a new instrument has often been introduced that is touted as an improvement. Clinicians are often left with instruments that are out of date, or they have a new instrument with new software that has not been validated.
Lewis: Artifacts are a problem with imaging technologies, not that Humphrey visual fields (Carl Zeiss Meditec, Inc.) do not have their own set of problems. It seems, however, as if these imaging systems are so prevalent among ophthalmologists and optometrists that glaucoma specialists often receive inappropriate referrals.
How often should patients with glaucoma undergo imaging? How soon can you document change?
Weinreb: In an ideal world, the answer is as many times as possible. The more information one obtains, the better one is able to detect disease and progression. Many insurers, however, only reimburse once annually for an imaging session. Moreover, many patients balk at frequent imaging sessions, even if they are possible. Because you might need several sets of images to assess progression, the reality is that, often, you cannot make that evaluation appropriately except over several years.
Schuman: Certainly, imaging patients once a year is helpful, but it is not enough. As Bob implied, it is sometimes important to perform extra testing for which the doctor waives the fee or the patient pays. It is critical to remember that insurers do not tell you what to do, only what they are going to pay for.
Lewis: What provokes you to get another imaging test in a 12-month period? Would it be a change in the visual field, an elevated IOP, the presence of a disc hemorrhage?
Brandt: All of the above.
Mills: You also might obtain another test if you thought you found change in one image. As with visual fields, if you repeat the same test immediately, you may get a different result. A second test can help you separate out a lot of noise.
Medeiros: It is always important to confirm abnormality.
Weinreb: An attractive aspect of visual field technology is that the software has largely been stable. The HRT has had a stable hardware platform, and many of the images obtained years ago can be utilized to determine whether or not there is progression. There has not been the same opportunity with other instruments. One hopes that all manufacturers now consider the stability of their platforms to enable clinicians to use the information that they have already acquired with future instruments.
Brandt: I want to raise the issue of electronic medical records (EMRs). None of the imaging devices talk to each other or to the programs for EMRs. As physicians are pushed to become more efficient in the way they see patients, manufacturers really must give us the tools to allow us to perform progression analyses and integrate all results into the EMRs at the desktop computer in the examination lane.
Lewis: I had to move to a second office to accommodate the new diagnostic equipment I had accumulated. It would be so helpful if these devices could share printers and hard drives and if we could integrate diagnostic tests into the medical record.
Brandt: If I am concerned that a patient's visual field has progressed, I want to be able to perform a glaucoma progression analysis at the desktop computer. I should not have to ask a technician to access the Humphrey perimeters to pull and run all of the historical fields. I should be able to do that while I see the patient.
PERIMETRY
Lewis: Although the platform for standard automatic perimetry has been stable for 15 to 20 years, we still have problems with the reproducibility of results.9 What is the best initial test for diagnosing glaucoma? Is it the Swedish Interactive Threshold Algorithm (SITA), short wavelength automated perimetry (SWAP), or frequency doubling technology (FDT) (all from Carl Zeiss Meditec, Inc.)?
Herndon: I think SITA-Standard is the gold standard for assessing patients' visual fields. SITA-Fast may have some short-term fluctuation, whereas the results with SITA-Standard are more reproducible over time.
Mills: I would point out that SITA-Fast performs fairly well, except for the very earliest glaucomatous defect, where the program seems to assume normality to an unacceptable degree. Those patients are few and far between, but their disease is exactly what you would like to be able to detect. Plus, as Leon mentioned, the noise factors are greater with SITA-Fast. If you plan to use a visual field as a baseline against which to measure future change, SITA-Standard is a better choice. Of course, SITA-Fast has an advantage with an uncooperative or easily tired patient, but it really does not save that much time.
There is a screening platform using FDT that has been adopted by Prevent Blindness America, and there is the Humphrey Matrix (Carl Zeiss Meditec, Inc.). The latter is the threshold version of the FDT, and I think that has a lot more merit for detecting early defects in ocular hypertensive patients. I find that FDT is more generally applicable to a wider patient population than SWAP, although it arguably may be a little less sensitive.
Medeiros: My colleagues and I compared FDT on the Humphrey Matrix with SITA-Standard.10 In our analysis, the Humphrey Matrix performed well in people with early damage, and its results were substantially better than with SITA. It is important to point out that, to adequately compare these tests, one has to evaluate the influence of the disease's severity and control for that. It is possible that a given test performs better in the early stages of disease, whereas another performs better in the late stages. If one does not stratify or adjust for the disease's severity, these differences will be missed when the entire sample is analyzed. When we stratified the results by the disease's severity, we found that the FDT on the Humphrey Matrix outperformed SITA-Standard, especially in the early stages of disease. In advanced disease, as expected, the two modalities performed similarly. It is important to emphasize, however, that we still do not have well-defined criteria for diagnosing abnormalities with the Humphrey Matrix.
Schuman: My colleagues and I published a couple of articles on the Humphrey Matrix versus SITA, and we found that the former is no better than the latter. The Humphrey Matrix performed about as well, although it missed a large number of defects. SITA-Standard found more defects than the Humphrey Matrix FDT did.11-12 I am not convinced that the Humphrey Matrix is any better or worse than SITA-Standard. Because we do not have the analytical software for the Humphrey Matrix that we do for SITA, we pretty much stick with SITA-Standard.
Brandt: In terms of SITA SWAP, I find it most useful for confirming the normality of an eye with an optic nerve that is difficult to interpret such as with high myopia. If two tests with SITA SWAP are reliable and normal, I feel fairly confident about following the patient a little less frequently.
Lewis: When would you repeat testing?
Brandt: If I had two normal SITA SWAP tests but was somewhat concerned about the optic nerve, I might see the patient every 6 months and then decrease to once per year if the tests continued to be normal.
Lewis: In reality, how often should clinicians repeat the visual field test?
Schuman: Whenever there is a new defect.
Lewis: Would you repeat the test at 6 months?
Schuman: It depends on how bad the defect is and whether there is a structural/functional correspondence. If there is a correspondence, it is much easier to accept the early defect on a single test. If not, the defect might be real, but you should repeat the test.
Weinreb: Selective functional testing is often useful for suspected glaucoma. I use it when a patient suspicious for glaucoma has a normal standard visual field, whether he is a myope with an unusual optic disc, has glaucomatous neuropathy (namely, a glaucomatous-appearing disc without standard achromatic visual field loss), or an ocular hypertensive. Selective functional tests are best for detecting damage early in glaucoma's continuum. Later in the course of the disease, one might be better served by SITA-Standard.
Herndon: Many years ago, I was asked whether I would treat a patient if I found defects with selective functional testing but with normal standard testing strategies. Currently, I think we would consider treating patients with some selective functional abnormalities, whereas most doctors would have only followed these patients 5 or 6 years ago.
Mills: Rick, regarding when to repeat visual fields, I think clinicians should do so a lot more often than they do. The specific time I would recommend, however, is when a clinical decision depends on the result. In a case with a structural/functional correlation, the decision does not hinge simply on the visual field but also on the structural result. If only the visual field changed, you need only repeat the field to confirm the change. The most common error that I see is for practitioners to make clinical changes (as minor as an alteration in medication or as major as performing a trabeculectomy) on the basis of a single, unconfirmed visual field change.
Weinreb: Perimetry is most useful if you have the opportunity to confirm the presence of a defect repeatedly. This often is not possible, due to logistics or a lack of resources. It would be useful to have an objective functional test that does not need to be repeated and might have even higher sensitivity and specificity than that of standard perimetry.
Technologies come and go. There was an electrophysiological test a few years ago that was promoted with great enthusiasm, but its sensitivity and specificity could not be confirmed. It is back in development and no longer commercially available. It is not clear when standard perimetry will be challenged as the norm by another functional test.
Mills: My background in neuro-ophthalmology makes me a bit cynical about the so-called objective tests, because they require just as much concentration on the part of the subject as does a subjective one. It is true that you eliminate the variability of pushing the button. If the subject is not paying attention to the stimulus, however, the cortical response changes dramatically. Objective testing still has a subjective element.
Lewis: What are your feelings about progression software packages?
Brandt: They are not perfect, but these tools are better than anything we have had before. That said, they are not particularly user friendly or usable in a busy practice. We need these tools to be usable in real time as we are seeing patients.
Schuman: I think the Glaucoma Progression Analysis (GPA) software on the Humphrey Field Analyzer II (both from Carl Zeiss Meditec, Inc.) is a useful test for my patients. It shows how many points have changed and whether there was a statistically significant change over a number of tests (Figure 3). You still need to take the results with a grain of salt, because we are asking patients to look into a bowl and press a button when they see a light. It is still a very variable test. We need to interpret the results.
Weinreb: One often needs to change the baseline depending on whether there are alterations in therapy or in the patient's condition. It would be best if this could be done at the desktop instead of requiring a return to the machine or the technician. The GPA software is a huge improvement. It is considerably easier to use and is most likely a better test than anything we have had before. It still can be improved, however, in terms of its user friendliness and efficacy.
Medeiros: The GPA software is helpful, but physicians should keep in mind that they need to correlate the results with the other clinical findings. Also, they should consider that glaucoma usually progresses slowly, so they have the time to make correct decisions and do not need to take actions without confirmation and a correlation with other clinical findings.
TONOMETRY
Lewis: Tonometry is probably more confusing and prone to error than ever with the introduction of corneal pachymetry, correlations, and all the fudge factors. Does anybody have any advice for the clinician?
Herndon: I do not think that Goldmann applanation tonometry is going anywhere soon. This technology has been here for more than 50 years, and all of the clinical and multicenter trials are based on it. The true range of IOPs that we know are based on some form of Goldmann applanation tonometry, and I think it will be the gold standard for the foreseeable future. It is going to take time before the new tonometers are mainstream.
Brandt: Part of Hans Goldmann's genius was devising a relatively inexpensive device that fit into the workflow of the slit-lamp examination. All of the new devices are going to have to jump that hurdle before they become widely accepted. A $3,000 to $4,000 device that sits on one slit lamp in an office with eight lanes is not going to be widely used.
Lewis: Some physicians routinely use pneumotonometry in their practices, but a long-standing argument has been that it is less accurate than Goldmann applanation tonometry. Any comments?
Mills: We should also insist that tonometers have reproducibility under a variety of clinical circumstances. Once you have a reproducible tonometer, it really does not matter so much whether the number you are getting is actually the pressure inside the eye. I am not talking about a large discrepancy, but the errors we are all aware of with Goldmann applanation tonometry. The main issue is how that pressure changes over time with therapy. Certainly, I think it is important to understand that maybe you have an underestimate relating to corneal thickness or that other factors are in the equation, but, ultimately, therapeutic decisions are based on change.
Medeiros: I think one of the reasons why new tonometers have not been widely used is because the information that currently guides our clinical practice still comes from clinical trials that have used Goldmann tonometry. We need prospective studies that demonstrate the clinical relevance of these new tonometers in managing patients. We also need information on how to incorporate the values that we get from these tonometers into clinical practice.
Brandt: To some degree, it comes down to a push-pull relationship between accuracy and precision. A precise tonometer is highly reproducible and, under the same conditions, will give the same results every time. An accurate tonometer reflects the "true" IOP. The problem with trying to correct IOP measurements with algorithms based on corneal thickness is that the underlying measurement is incredibly imprecise. You are fooling yourself if you think that you are getting a more accurate result. A recent study in the United Kingdom showed that few ophthalmologists checked the calibration of their tonometers. In this study, 50% of the devices were out of calibration by more than 2.5 mm Hg.13 Consider that most of the nomograms that supposedly adjust IOP for central corneal thickness apply "corrections" in that range. You are therefore trying to apply a corrective algorithm to a measurement that has noise of ±2 or 3 mm Hg. It is a fool's errand.
Schuman: I think we use tonometry to identify a patient's baseline measurement and set a target IOP. You consider follow-up measurements in the context of everything else in the case. If you assume that the tonometer is going to measure IOPs within 1 to 2 mm Hg of reproducibility, you can use it as a guide. You must also look at the optic nerve, the visual field, the family history, etc. You are not just treating the IOP.
Brandt: I would suggest that people think of the analogy of glaucoma to diabetes. Technology in the latter field has advanced so that practitioners have transitioned from measuring sugar in urine to analyzing blood and then fasting blood sugars to analyzing hemoglobin A1C levels to obtain a trailing average. Diabetic patients now have portable glucometers and adjust their therapy in real time. In glaucoma, we are stuck with random, individual IOP measurements. A tonometer that measured the true IOP to within 0.1 mm Hg would not be a significant advance in the management of this disease. We need equivalents in glaucoma of both hemoglobin A1C and the Holter monitor for IOP.
Lewis: Bob, the sleep studies conducted at the Hamilton Glaucoma Center, University of California, San Diego, are important because they demonstrated that nocturnal IOP measured while subjects are supine is much higher than we thought when people are sleeping. How can clinicians incorporate that information into our understanding of glaucomatous progression?
Weinreb: The answer is not clear, because we do not know the implications of the increased pressure. Earlier studies largely evaluated IOP while patients were seated for 24 hours. My colleagues, including John Liu, PhD, and I thought studying habitual pressures with patients supine during the nocturnal period and seated during the diurnal period should be investigated. We found that at least two-thirds of patients had peak IOPs during the nocturnal period and not during usual office hours. We wondered if there were a way to measure IOP throughout the day that would enable clinicians to estimate a patient's peak pressure. The best that we have determined is that measuring IOP during the day while the patient is in a supine position better reflects peak pressure than any other in-office IOP measurement. This approach is not effective in all patients, however, and there is no way to determine in advance when it will be successful. Moreover, we do not know the significance of increased IOP during the nocturnal period. Until we understand the full pattern of other concurrent physiologic changes, including changes in perfusion pressure and blood flow, we are not going to make therapeutic recommendations.14
Brandt: Based on your group's data, if a patient's glaucoma is progressing despite good IOP control, should the clinician communicate with that individual's internist to say maybe he is overtreating the individual's blood pressure or to suggest changing the dosing time of his hypertensive medications?
Weinreb: There is considerable evidence that low perfusion pressure and blood pressure (particularly diastolic) are risk factors for glaucoma. Many patients' blood pressures, particularly diastolic, were considerably lower during the evening than we expected. With high IOP, a low diastolic pressure might not be good for a patient with moderate glaucoma. Should we recommend that patients also have their blood pressure monitored? If so, and their blood pressures are low, how do you elevate them? There are many questions for which we have no answer.
Lewis: Has anyone changed his manner of practice based on the sleep study in terms of diagnosing normal-tension glaucoma?
Weinreb: My primary therapy continues to be a prostaglandin analog. In the past, I added a beta blocker to a prostaglandin when an additional medication was needed. Our work showed that timolol has virtually no effect on nocturnal IOP,15 something predicted by Brubaker based on measurements of aqueous flow.16 I therefore am no longer using beta blockers for second-line therapy but rather select agents that are more likely to provide 24-hour pressure lowering.
Mills: Regarding ocular perfusion pressure, Stephen Drance, MD; Andrew Crichton, MD; and I found that prostaglandins have an advantage over beta blockers in terms of the calculated measurement of an optic nerve capillary's perfusion.17
Schuman: Beyond that, a beta blocker does not have a large additive effect to a prostaglandin.
Herndon: If we use beta blockers, obviously, aqueous production is down at night while patients are sleeping. The classic teaching is that you administer a beta blocker in the morning. Should the agent we add to a prostaglandin not affect aqueous production?
Weinreb: I use beta blockers as primary therapy in patients who cannot tolerate a prostaglandin or in the very few patients who do not respond to a prostaglandin. These drugs are effective at lowering pressure when used as first-line therapy. The local ocular effects are minimal, but they have some systemic effect. They just are not effective, however, as additive agents to prostaglandins.
Lewis: Bob, another interesting aspect of the sleep study was that it showed different drugs affecting that curve, a pressure rise in the evening, and surgery as the only factors that seemed to blunt that curve.18 Although we certainly have not proven that in a published study, the implications are that mediations help bring the curve down a bit, but they really do not prevent the rise in pressure. Surgery probably flattens the curve.
Weinreb: Argon laser trabeculoplasty had little effect on nocturnal IOP in patients who were already on maximally tolerated medical therapy.19
THE CURRENT CHALLENGE OF DIAGNOSING GLAUCOMA
Lewis: Most of the patients diagnosed in North America are categorized as having primary open-angle glaucoma (POAG). We have lumped a lot of patients into the category of POAG without being able to differentiate this large group. It would be of benefit to use criteria that split these into genetic or clinical traits to facilitate better understanding of the disease. The problem could lie with the meshwork, Schlemm's canal, or the outflow system. At present, we do not have any way of differentiating these entities. We have the secondary glaucomas, which are not always categorized anatomically. There may be a better way of differentiating a diagnosis. Any thoughts?
Mills: We all vote for genetics.
Weinreb: Macular degeneration was a genetics wasteland, and, overnight, it was revolutionized by the discovery of just a few macular-degeneration genes. We could be close, too, and just do not know it, although it may take a number of years to sort out the information. Most likely, the genetics associated with secondary glaucoma, such as exfoliation, will emerge before that of POAG.
Schuman: Clearly, there are different phenotypes that we can see, but we do not necessarily know what they mean. There may be more differences than we are appreciating now, and some other variations that we are calling different may not be at all. I think that we still need to be humble about our abilities, and I am not sure that we can discriminate any better than we are now.
Medeiros: Some attempts have been made to classify basic phenotypes, for example, based on the appearance of the optic nerve such as focal ischemic, senile sclerotic, myopic, or concentric enlargement. It is clear that the majority of patients cannot be classified into a single entity.20
Mills: You are asking if we really believe that POAG is a diagnosis, that it is an etiologic description of some sort. It is a wastebasket we put patients in because we have no other place to put them. As long as we do not fool ourselves that we know what is going on here, then I think we are OK.
Lewis: Now that some ultrasound images show the canal closed,21 I begin to wonder what came first. Does Schlemm's canal close and cause problems in the trabecular meshwork, or does the meshwork's failure lead to problems in the canal? If we knew, for example, that the collection system was a problem and causing the IOP to rise, then maybe we could direct our treatment there.
Schuman: In normal eyes, approximately 70% of the resistance to aqueous outflow lies somewhere between the inner wall and the juxtacanalicular tissue, and the rest is downstream.22-24 The ratio in glaucomatous eyes is anybody's guess.
THE FUTURE OF DIAGNOSING GLAUCOMA
Lewis: In the next 5 years, what will help us to diagnose and manage glaucoma?
Herndon: Genetics.
Medeiros: How about new imaging instruments like the next generation of optical coherence tomography? I am not convinced that we will have a genetic tool for diagnosis in 5 years.
Schuman: The current Stratus OCT was introduced 5 years ago, and there are now more than 6,500 units in use around the world, according to the manufacturer (V. M. Patel, oral communication, May 2007). If the new technology works, I think it can catch on quickly. I would certainly think we will see improvements in imaging technologies and, I hope, in functional assessment as well. In 5 years, we may have a greater ability to combine the information we get from structural and functional assessments to create some synergy of the two.
Herndon: The evolving technology is exciting and interesting. Ultimately, we may be looking at the issue of job security, but I think these devices will still require a clinician to put all of the information together.
Brandt: The biggest advance in the next 5 years for managing glaucoma would be figuring out a way to have 24-hour monitoring of IOP or a hemoglobin-A1C equivalent for IOP. I think this would represent a paradigm shift in the the disease's management.
Schuman: How about if there was an intraocular device? Do you think that would catch on? Would people be willing to have an implant?
Brandt: I think that is too invasive for the 90% of glaucoma patients who never need surgery, but adding a monitor to a glaucoma implant for patients with advanced disease will probably be the first step and is likely within 5 years.
Medeiros: I think there is some promise in the development of instruments to image the rate of the death of ganglion cells at a cellular level. Such devices could change the way we diagnose the disease.
Weinreb: Confocal imaging can be used to assess the health and death of retinal ganglion cells in mice. I expect that this research will move forward over the next several years, and we will be able to assess human retinal ganglion cells to diagnose disease or, more importantly, to detect progression. Within 5 years, there will be continuous 24-hour IOP monitors. There is also an opportunity to diagnose glaucoma and its progression without even directly examining the eye and instead, evaluating the central visual pathway. Such an examination would include the lateral geniculate nucleus and visual cortex. Glaucoma is a neurodegenerative disease. With sensitive and specific structural and functional tests that assess the entire visual pathway, we should be able to evaluate progression, to monitor the effects of therapy, and possibly to develop new therapies.
Finally, it is possible that we will not only consider glaucoma as a pressure-dependent disease, but there will also be technologies that will delineate IOP-independent factors. I speculate that this will lead our therapies into new directions. My expectation is that we will find that most patients' have pressure-dependent and pressure-independent components, and they will benefit from new glaucoma therapies. For their development, regulatory agencies will have to be more flexible in their requirements for clinical-trial endpoints, and industry will have to take risks and sponsor the innovative clinical trials.
Mills: We have always been notoriously poor at predicting the state of medicine in the relatively near term. Particularly, we find it difficult to predict which technologies will change how we view a disease. I am sure, however, that there will be no breakthrough technologies if we fail to fund biomedical research adequately.
