Identifying Progression With Stratus OCT

The application of statistical analyses enables quantitative detection of structural change.

By Joel S. Schuman, MD, FACS

As we're all aware, detecting glaucoma progression is paramount if we are to treat our patients effectively and preserve their vision. However, clinical practice has shown that detecting progression presents challenges. What follows is a summary of these challenges and a preview of the developments in optical coherence tomography (OCT) that can help us overcome them.

Obstacles to Accurate Detection of Progression

A critical challenge we face in the management of our patients is the lack of gold standard criteria for glaucoma progression. Currently, we rely on a variety of devices, tests and observations to piece together the puzzle of whether or not a patient's glaucoma is truly progressing.

We are faced with several other confounding factors, including the fact that healthy eyes, as a function of aging, lose retinal nerve fiber layer (RNFL) thickness at a rate of 0.3% to 0.5% per year. We also face test reproducibility issues. We have device-related measurement variability, and we have variability in sampling location. We are not sampling exactly where we expect we are sampling every single time; discs move relative to the scanning circle even when patients are doing their best to keep their eyes stable.

Environmental factors also can affect test reproducibility. A patient's cornea might be dry on one testing day but not on another. A cataract may have developed, or there may be differences in the positions at which images were acquired. In addition, scan, and therefore image, quality can be different from one visit to the next.

OCT Is Highly Sensitive to Change

Obviously, we must consider all of these issues when evaluating change over time, and a technology that automatically helps us do that would be very desirable. Because it is more sensitive to change than visual fields, Stratus OCT can serve as a solid basis for improvement in this area.

My colleagues and I conducted a longitudinal study published in Archives of Ophthalmology and first authored by Gadi Wollstein, MD. The aim of this study was to evaluate OCT's ability to detect disease progression in glaucoma suspects and patients.¹ The study involved 64 eyes of 37 patients who underwent a median of five usable OCT tests and a median of six usable visual field tests. Median follow-up was 4.7 years.

We found that Stratus OCT is very sensitive to progression changes (Figure 1). Based on OCT alone, disease remained stable in 66% of the patients and progressed in 22%. In comparison, based on visual fields alone, disease progressed in 9% of patients, and based on both OCT and fields it progressed in 3%. While the study results do not tell us anything about OCT's specificity in detecting change, they do indicate that OCT is more sensitive to change than visual fields are.

Figure 1. Results of the study noted here, represented by a Kaplan-Meier curve, indicate that optical coherence tomography is more sensitive than visual fields to changes in glaucoma status.

Building on this OCT capability, Carl Zeiss Meditec has been developing statistical software that may identify changes that we would not be able to identify by simply looking at Stratus OCT-generated images. Such software can make a major difference in our ability to detect change over time in our patients. It is analogous to perimetry in 1982 versus perimetry in 2007.

Enhancements to OCT Capability

Currently, we can use the Stratus OCT's Serial Analysis function to help us follow patients. Basically, it is an overlay of the RNFL thickness at each of the 256 points along the scan. A different color represents each visit date. I think all of us agree that determining if one is different from another is challenging.

A new version of the software, GPA Advanced Serial Analysis, will be available soon (Figure 2). It performs Advanced Serial Analysis with these functions:

■ Provides quantitative values for overall average RNFL thickness as well as for the inferior and superior quadrants

■ Overlays thickness profile plots from multiple exams (as Serial Analysis does)

■ Plots average thickness over time

■ Uses linear regression analysis to report average RNFL thickness as slope, confidence interval!and P value to show if a statistically significant change has occurred.

Figure 2. An illustration of the current Stratus OCT Serial Analysis compared with the newer GPA Advanced Serial Analysis. Key analyses provided by the latter include quantitative values for the average overall, superior and inferior RNFL thickness, the plot of average thickness over time, and the overlay of multiple exams. The analyses shown here indicate thickness deterioration in one quadrant and thickness increase in another. Regression analysis shows the average thickness slope is a loss of 1.8 microns per year as well as a P value of less than 1% at the 95% confidence interval.

In addition, GPA Advanced Serial Analysis displays signal strength. Signal strength is an important quality parameter. If it falls below a certain level, approximately 6, image quality is likely degraded. Analyses with low signal strengths may not be as accurate as those with higher signal strengths.

Our group at the University of Pittsburgh has been working with Carl Zeiss Meditec to further develop GPA for the Stratus OCT. This version of GPA:

■ Takes into account measurement variability, scan quality and the patient's age

■ Displays an overlay of multiple thickness profiles

■ Provides progression analysis of average RNFL

■ thickness by quadrants and clock hours

■ Provides trend (regression) and event analysis

■ Reports slope of change, confidence interval and P values.

Aid to Clinical Decision-making

Because GPA for the Stratus OCT provides quantitative detection of structural change in our glaucoma patients, it will be a welcome addition to our current set of tools for clinical decision-making.

Dr. Schuman is the Eye and Ear Foundation Professor and Chairman of Ophthalmology, the Eye and Ear Institute, University of Pittsburgh School of Medicine, and director of the University of Pittsburgh Medical Center (UPMC) Eye Center. He is also professor of bioengineering at the University of Pittsburgh School of Engineering.

REFERENCE

1. Wollstein G, Schuman JS, Price LL, et. al. Optical coherence tomography longitudinal evaluation of retinal nerve fiber layer thickness in glaucoma. Arch Ophthalmol. 2005;123:464-470.

New Capabilities in Action
In this patient, Stratus OCT scans show nerve fiber layer thinning inferiorly, which corresponds with visual field loss. The thinning appears to worsen from 2003 to 2006; a borderline area has become outside normal limits. The Guided Progression Analysis (GPA) provides much more information for determining if true progression is occurring. It detects a statistically significant rate of change in the inferior temporal quadrant. It also highlights a possible change at clock hours 11 and 2, which might not have been detected by observation of the scan results alone.
In this patient, Stratus OCT scans were acquired from 2004 to 2006. The GPA evaluates signal strength and accepts all of the images. It detects progression by quadrant, clock-hour analysis, trend and event. It finds no statistically significant global rate of change or statistically significant rate of change in any quadrant. However, it does detect a possible rate of change at clock hour 11 and asks for a confirmatory scan. The change event analysis confirms a change in the temporal quadrant and at clock hours 9, 10 and 3.
In this patient, visual fields fluctuated between 2003 and 2006, but mean nerve fiber layer thickness on OCT did not change much. Glaucoma Progression Analysis finds no statistically significant change during that period.