New Technology for the Diagnosis and Monitoring of Glaucoma

Glaucoma diagnosis and monitoring have relied on sequential documentation of the optic nerve and retinal nerve fiber layer. As technology has become increasingly sophisticated, new imaging devices are available to assist in these examinations to better follow the disease. The clinical importance of these new tools depends on the accuracy and reproducibility of the results, the ability to differentiate between normal and glaucomatous eyes, and the ability of the test to aid in monitoring progression. This article reviews three of these new devices, the HRT II, the OCT, and the GDx, and addresses their practical utility in support of their implementation into standard clinical practice.

WHAT ARE YOUR THOUGHTS ON THE EVOLUTION OF GLAUCOMA DIAGNOSTIC TOOLS?

For years the diagnosis and monitoring of glaucoma has been based on serial documentation of the optic nerve and retinal tissue, measuring subjective functional and/or objective visible structural damage. Multiple forms of perimetry, ^1,2 visual psychophysical, and electrophysiological tests ^3,4 have been developed for subjective functional assessment (example: fusion flicker ^5,6). Both Goldmann and Humphrey perimetry are current clinical standards. ⁷

For an objective evaluation of the optic nerve and retinal tissue, we have until recently relied on direct observation. Over the years, our observation documentation methods have grown increasingly sophisticated as we moved from drawings to film photography to digital photography.

Now, there are optical tools that augment direct visualization, such as the HRT, ^8-10 OCT, ¹¹ and GDx ¹² devices. These represent three different technologies that examine three different aspects of the ocular anatomy. All three technologies provide useful, yet different information.

The Heidelberg Retina Tomograph II (HRT II, Heidelberg Engineering, Vista, CA) (Fig. 1) is a confocal scanning diode laser ophthalmoscope that analyzes three-dimensional images of the optic disc and peripapillary retina. ^13,14 This HRT II uses a fixed 15° field of view with 384 × 384 pixels per image plane, providing a pixel resolution in the x-y plane of about 10 μm. The number of images per millimeter of depth is fixed at 16. The standard scan depth is 2 mm, which will result in 32 images. The scan depth automatically adjusts, depending upon the depth of the cupping. For example, in the case of a patient with advanced glaucomatous cupping, a 4-mm scan depth will enable acquisition of 64 images in the series, ensuring standard axial resolution. As cupping develops or progresses, the resolution remains constant and therefore data can be compared over time.

Figure 1

The Optical Coherence Tomography (Stratus, Zeiss-Meditec, Dublin, CA) (Fig. 2) device is a noninvasive diagnostic imaging device that enables cross-sectional imaging of microstructures within the eye. ¹⁵ The echo delay time of the light (similar to the way sound is used in ultrasonography) backscattered from different layers in the retina is measured using an optical interferometer. The depth resolution of this technology depends on the characteristics of the light source and is approximately 15 μm. ^16,17 The lateral resolution is dependent on both the wavelength and the optics of the eye and is between 30 and 50 μm. A circular scan centered on the optic disc and the peripapillary retina acquires between 128 and 768 A-scans in less than 2 seconds and uses mathematical algorithms to determine the nerve layer thickness. ¹⁸ This information is then used to create a two-dimensional image by performing successive axial range measurements.

Figure 2

The GDx (GDx Access, LDT, San Diego, CA) (Fig. 3) directly measures the nerve fiber layer by means of a confocal scanning laser ophthalmoscope with an integrated retinal polarimeter that measures changes in polarization (retardation). ^19,20 The nerve fiber layer causes a change in the state of polarization of a laser beam when it double-passes through this layer. After being reflected by the deepest layers of the retina, the light is collected and is used to create a retardation map. A baseline image is created and specific patient values can be compared over time. Typical image reproducibility better than 15 μm allows precise, accurate follow-up over time. The field of view of the GDx VCC is 40° × 20° and the resolution is 256 × 128 pixels.

Figure 3

We know that these devices improve our ability to monitor our patients. What we don't know is which patients would benefit most from being monitored by which devices and which will provide the most benefit/cost over time.

HOW DID YOU DECIDE WHICH AND WHEN?

Our need to know basis grows with the invention of each new diagnostic tool, but which of these tools is truly helpful to our goal of finding and fixing? With extensive experience in laser techniques, both therapeutic and diagnostic, and as an early adapter of new devices, Cited Here... I took a more conservative approach to these new diagnostic tools. ²¹ I decided to wait and watch before jumping on the HRT, GDx, OCT bandwagon. Having been around long enough to have watched the emergence of a clinical standard in perimetry, an evolution from tangent screens to Goldmann to Tuebingen to Octopus to Humphrey, I wanted to await the technology that was not only informative, but user friendly and one that would be adopted by a large number of clinicians to enable not only financial security for the technology to grow but also a large shared database of clinical cases to increase the technology's usefulness.

Each of these devices has advantages and disadvantages, but which could best help me manage my patients? After evaluating the currently available options, and waiting for more relevant clinical studies, I recently added the first of the second generation of these devices, a confocal scanning laser ophthalmoscope, the Heidelberg Retina Tomograph II (HRT II) to my practice. This tool offers another meaningful method of monitoring and understanding the disease process by providing an objective, noninvasive, 3-dimensional image of the optic nerve head structure and papillary retinal tissue.

WHY NOT USE STEREO PHOTOGRAPHS FOR THE STRUCTURAL ASSESSMENT?

Photography, especially stereo photography, has been a staple for monitoring glaucoma, but its important to understand that glaucoma is a multifaceted disease. The availability of more data allows earlier and more sophisticated monitoring of this dynamic, degenerative process. Even though stereo photographs offer high specificity and sensitivity, many busy practices find it challenging to obtain reproducible, quality photos. Due to the difficulties of archiving and retrieving photographic data in a timely and useful manner, it is a rare practice in which comparative photos are available to the physician at the time patients are reexamined. Because of this, these newer technologies, being computer based, are advantageous. However, photography remains a standard of care that I still find useful.

HOW DO YOUR PATIENTS RESPOND TO THE TECHNOLOGY? HOW HAS THIS TECHNOLOGY IMPACTED THE PATIENT FLOW OF YOUR PRACTICE?

I feel it is important that patients know that the practice has incorporated the most current technology to provide them with the best possible care. In general, my patients are an informed group and respond very well to sophisticated technology. Since they are educated to participate in their therapy, graphic representation of their disease status and its level of control enables me to better assist them in their clinical decision-making process. By sharing a 3-dimensional view of their own optic disc, the patients are more capable of understanding the nature of their disease and monitoring their level of control. Therefore, I have found the HRT II to be a useful tool to help educate my patients.

In addition, there is little if any impact on office flow, since the examination itself is often completed in 5 to 10 minutes. Because dilation is not necessary, patients can be examined with the HRT II while waiting to dilate, therefore making better use of clinical time. Also, the required technician training time is minimal compared with that for other devices.

WHAT ROLE DOES THIS TECHNOLOGY PLAY IN DISEASE MANAGEMENT?

I am often asked to assess patients with ocular hypertension and suspect discs and glaucoma suspects. Following a review of risk factors, I rely on traditional tools: biomicroscopy, gonioscopy, applanation, and tonopen tonometry, funduscopy, visual fields, and stereophotographs. With new information forthcoming from studies such as OHTS, ²² we have more to think about in evaluating this group of patients. Additional risk factors such as corneal thinness ^23,24 require us to not only reassess and redefine our terms but also to rethink the traditional paradigms. As I follow these patients for years, I am able to monitor their visual field changes but prefer improving my ability to control their disease by preventing these changes from occurring. Studies from Moorfields Eye Hospital in London ²⁵ and Dalhousie University in Halifax, Nova Scotia ²⁶ have demonstrated that the HRT can detect structural damage prior to visual field loss-sometimes by several years. This type of objective measure gives me confidence to advise beginning treatment even when visual field results appear normal.

WHAT LED YOU TO THE DECISION TO ACQUIRE THIS PARTICULAR TECHNOLOGY?

I researched several instruments both in the marketplace and in research stages before I decided. For almost a decade, I have followed CSLO, read peer-reviewed studies, and studied years of longitudinal data supporting the technology. I did not buy the HRT I-the first generation instrument-because it was very technician dependent. It was research oriented-appropriate for an institutional setting-but not a busy clinical practice because it requires extensive operator involvement and understanding. In contrast, with minimal compromise, the next generation device, the HRT II, is user friendly.

While the HRT II is a second-generation device, the data sets are comparable to the HRT. The fact that any data collected on the HRT can be used on the HRT II and with a growing user base and evolutionary software advancements and support, I felt more assured that the instrument would not likely become a doorstop in short order.

HAS THERE BEEN A LEARNING CURVE FOR YOUR TECHNICIANS? IS SPECIAL TRAINING REQUIRED?

No, the basic operation is very straightforward. The HRT is menu driven. The technician merely

1. adjusts the instrument position in relation to the eye.

2. ensures that that patient fixates on the internal light

3. adjusts the focus within one diopter.

As the patient is examined, the technician focuses the patient's image and, if necessary, places astigmatic corrective lenses onto the instrument. The technician must learn to document the disc margins by drawing a circle or contour line. This is a potential error source, so I monitor these drawings carefully. I was concerned about this one subjective aspect of the examination: the drawing of the contour line. This line is drawn on the baseline examination, but may be changed at any time in the future. Also, once saved, it remains consistent and automatically transfers to future examinations. Most often, this issue need not be revisited.

Tracing the outline of the optic nerve head, the contour line, defines the space that will be measured. The technician defines the line by positioning markers along the desired margin. Color-coded displays are generated with red representing the cup, green and blue defining the neural rim tissue, and blue indicative of the more sloped regions. These volumetric measures allow quantification of the change in the optic nerve head as neuropathy develops and progresses, a more objective monitoring of change over time than serial observation or photography.

Defining the cup and rim involves the definition of a reference plane. To do this, the contour line along a 6° wedge in the papillomacular bundle of the temporal side of the optic nerve is used. This area is used because it is most likely to remain stable for the longest period.

A reference plane is positioned 50 μm below the average height of the wedge; tilt across the image is also corrected. One can follow parameters used to describe the optic nerve over time for signs of progression. Along with the intuitive parameters, such as rim and cup volume and contour line height variation, there are others. The frequency distribution, the distribution of depth within a cup, can be summarized in a single number called cup shape. Classifying individual eyes using individual parameters offers poor sensitivity because of the variability of values across the population. Instead, parameters can be combined using multivariate analysis. This offers reasonable sensitivity and specificity for determining whether the disc can be classified as glaucomatous.

Let me review some other features of this device. The software allows various image displays. These include a 3-dimensional reconstruction, in which you can change the viewing orientation, and an interactive view, which lets you analyze profiles across the image in both the x and y planes.

The device also provides refined image analysis. If you define an outer limit, or contour line, of the optic nerve head, you can look at various volumetric parameters that describe aspects of the nerve head's appearance. One important measurement in glaucoma detection is the height of the nerve fiber layer around the optic nerve head. Several instruments have used the double hump data display. The humps correlate to the superior and inferior poles, where the nerve fibers entering the optic nerve are thickest. As neuropathy progresses, this distribution flattens out. The HRT II provides parameters that describe the height profile around the optic nerve.

Published in 1998, Moorfields regression analysis is based upon on work by Dr. Wollstein and colleagues. ²⁵ It is a valuable addition to the armory of diagnostic tools and offers good sensitivity for defining glaucomatous damage. Moorfields regression analysis divides the optic nerve head into 6 sectors, and the rim and cup areas are analyzed. Moorfields regression can differentiate cup size versus disc size to indicate when sufficient neuroretinal rim is not present to support a larger cup. For example, a 0.6 cup in a 1.8-mm disk would be considered pathologic, whereas a 0.6 cup in a 3.4-mm disk would probably be considered normal due to the relative amount of rim present. This stresses the importance of looking at the disk size as well as the cup-to-disc ratio. However, smaller disks do not fit their schema as well, which is important when considering the Moorfields regression analysis.

The HRT II printout offers two different report options. The first report is the initial report printout. This initial report (Fig. 4) details the color-coded topographic image with a vertical and horizontal cut through the center of the optic nerve head. Superimposed on the topographic image is the cup, rim, and disc area. Furthermore, it features the false-color fundus view separated into 6 sectors according to the Moorfields regression analysis. Each sector is marked with either a red cross (denoting possible glaucomatous damage), a yellow exclamation point (denoting suspect area), or a green check mark (denoting a normal appearance). A separate image also shows the retinal nerve fiber layer contour height along the circumference of the optic nerve head in the classic T-S-N-I-T order. A separate graph on the initial report details the Moorfields regression analysis by sector. The graph features vertical green and red bars, which represent the total disc area for a given sector (green for the neuroretinal rim area and red for the cup). The stereometric numerical values are also given in a table. The secondary report (Fig. 5) is the follow-up examination report. If three or more follow-up examinations are available, this type of report is quite helpful in deciding whether a structural change of the optic nerve head has been documented. The upper left image of the report shows the gray-scale fundus view, and superimposed in red and green areas of increased excavation over time (red) or areas of anterior movement (green) over time.

Figure 4

Figure 5

The standard HRT II printout provides important information on image analysis, profiles, height variation, and the Moorfields regression analysis. In addition, one particularly helpful piece of information is the image's standard deviation-the lower the number the better the image quality, which provides technicians with a form of instant feedback and encourages quality control in real time, eliminating the need to repeat testing due to technician error.

As with each of these new devices, there are downsides. The HRT II can be noisy during the examination, annoying both patients and technicians. Comparison difficulties may arise from optical changes within the eye following cataract, glaucoma, or refractive surgical procedures, necessitating the establishment of a new baseline. OCT and GDx also have their own particular problems. The OCT III samples only a very limited number of points in the retinal plane (less than 1000) compared with the 147,456 points acquired with the HRT II. Furthermore, the current version of the OCT does not provide for retinal scale correction of ametropia or axial length. Thus, data cannot easily be compared between different patients, and refractive or surgical intervention may invalidate follow-up. The GDx measurements are dependent on accurate compensation of the corneal birefringence, which has been improved in the newest version. ^27-29 Additionally, the current instrument requires a calibration measurement in the macula. In the presence of macular diseases that cover a large geographic area, this calibration measurement may be difficult to perform.

CAN THE HRT II BE USED AS AN EFFECTIVE GLAUCOMA SCREENING DEVICE?

I have mixed feelings about the word screening. While some may suggest that it is possible to accurately classify early glaucomatous damage in a single examination, this is not realistic.

Early glaucoma may not be easy to diagnose on a first visit because of the vast variability among patients and its progressive nature. Normative databases are useful to some extent, but limited because there is a physiological variation in what can be classified as normal and because too many variables often are not segregated, such as and cross-over in races-for example, databases that separate African-American patients and white patients by self-determination.

Thus, it is difficult to immediately define a patient as normal. It often takes several years before the disease has progressed enough to trigger an abnormal reading on one test. It is more important to be able to determine change over time, which is known to be real, and base a diagnosis on the change that is documented earliest. This objective analysis of change, or progression, in optic disc cupping is an important asset of CSLO technologies like the HRT II. The GDx, as with the HRT II, has a normative database that attempts to provide a quick and somewhat objective analysis. The GDx can demonstrate specific areas of disease progression by means of a serial analysis printout and using test parameters that do not rely on questionable risk factors for glaucoma.

Besides, we know that a patient may be functionally normal for many years while the structure of the optic nerve head and nerve fiber layer may be deteriorating. Practically speaking and depending on risk factors, I start with a baseline, a first follow-up examination at 6 months, and second at 12 months. At that point it is possible to comfortably follow the progression data on a patient.

However, as I indicated before, the software does have a normative database that is growing, and some valuable insights might be gleaned even from a first or second examination as we learn more.

In summary, one must always remember that for a new instrument to be accepted and implemented into standard clinical practice, it must be superior to existing methods in clinically relevant ways. This new generation of diagnostic devices has enabled physicians caring for glaucoma patients to find the disease and to follow the disease by augmenting their clinical acumen. However, they have not replaced direct viewing or photography in many clinical practices. Time and additional experience may change this picture. As our understanding of the disease process grows, we will need better tools to determine whom to treat and when to treat, how much to treat, and how to benchmark our treatment decisions. At the same time, we need to be cost conscious and cost effective. I have no regrets about my decisions to wait until the technology came close to fulfilling my clinical and intellectual needs before becoming a user. However, I am still on a learning curve.

FOOTNOTES

*Editor's note: Dr. Michael S. Berlin is a pioneer in diagnostic and therapeutic ophthalmic laser technologies. He is known for having developed the excimer laser trabeculostomy (ELT) procedure, which has strong support in Europe, but is not yet available in the United States. Cited Here...