Peer Reviewed
Evaluating Retinal Thickness During Pegaptanib Sodium Therapy
RICHARD B. ROSEN, MD, THOMAS O. MULDOON, MD, & PATRICIA GARCIA, MD

Figures 1A (top) and 1B (bottom). 1A. Location of the thickness grids relative to the fundus is demonstrated by placement on the SLO images. Alignment is confirmed by vessels in the SLO. 1B. Retinal thickness grids before and after treatment are shown. The red colors indicate greater thickness, while the greens and blues represent thinner areas. A post-treatment map on the right in 1B showed a blink artifact (gray band) This was not significant enough to invalidate the scan.

With the development and refinement of imaging systems in recent years, physicians are able to diagnose and evaluate their patients' retinal conditions more closely without having to use invasive means. Topographic retinal thickness mapping is one example of an imaging system measuring the progression of macular disease and the response to a variety of therapies. Currently, the 3 commercial systems available for topographic mapping are the Retinal Thickness Analyzer (RTA), (Talia/Marco, Jacksonville, Fla.), the Stratus OCT3 (Carl Zeiss Meditec, Dublin, Calif.), and the OCT/SLO (OTI, Toronto, Ontario).

The RTA is a computerized laser slit lamp, which assembles sequences of optical cross-sections from 5 points surrounding the fovea to generate a map that is overlaid upon a fundus image. The Stratus OCT3 expands a sequence of 6 fixation-centered radial scans into a free-floating thickness map. The OCT/SLO utilizes volumetric analysis for more complete sampling and offers precise registration with macular landmarks.

This article will describe how the OCT/SLO system obtains images and will illustrate image examples of how the technology was utilized to measure macular thickness changes in patients treated with pegaptanib sodium (Macugen, OSI Pharmaceuticals) for exudative age-related macular degeneration (AMD) in a small pilot study.

A DISCUSSION OF RETINAL IMAGING SYSTEMS

The advent of extended pharmacologic therapies for exudative AMD and diabetic macular edema (DME) has stimulated new interest in using such a technique to monitor anatomic response to therapy quantitatively.¹

Conventional approaches involving the RTA and Stratus OCT have various limitations of the respective technologies. The RTA employs the technique of laser slit beam triangulation to measure retinal thickness. The laser slit beam is projected at an angle on the retina while a CCD camera records the backscattered light. As the camera records the reflected image, a thickness algorithm identifies the location of the anterior and posterior retinal borders. The distance between the two light peaks determines the retinal thickness at a given point.

While this technique limits its accuracy and registration in eyes with any media nonhomogeneity such as cataract or pseudophakia, it does offer alignment of the thickness grid to the fundus map by merging the submaps from the 5 fixation points used in the test.

Figures 2A(top) and 2B (bottom). 2A. Pretreatment grid appears following automated alignment sequence. The operator can adjust any mismatch between the two studies that makes the grid appear tilted against the background of the SLO image and check the placement of the grid on the SLO image. 2B. Post-treatment grid appears once the operator confirms acceptance of the first alignment.

The Stratus OCT3 generates nonregistered maps with minimal detail that are intended to link to a central fixation point. Conversely, the OCT/SLO uses one-to-one registration between high-resolution SLO fundus surface features and the stack of OCT coronal planar images. A comparison algorithm features a user-modifiable automated alignment of the retinal vessels from successive SLO images and coronal OCT scans. This alignment ensures the accuracy of subtracting serial studies.

A second concern with the Stratus OCT maps is the heavy reliance upon interpolated data from a small number of points to construct the maps. Six radial scans, from which the fast macula maps are generated, are only 20 μm in thickness each and do not all run through the same point of origin. Relying upon the patient's fixation increases the likelihood of shifting of the point of origin as the central vision improves or worsens. The interpolated wedges decrease in accuracy as they extend into the periphery. The OCT/SLO maps are generated from a large sampling of points (512 x 512 or 256 x 256 pixels and a stack 100 to 200 sets of C-scans) captured within the coronal OCT stacks. This robust data set for each map helps to ensure accurate representation of retinal thickness, which can be compared with previous and future studies.

OCT/SLO METHODOLOGY

OCT/SLO imaging employs a single illumination source and multiple detectors designed to integrate the raster scanning technology of the confocal scanning laser ophthalmoscope with the high-resolution capabilities of OCT interferometry. Utilizing transverse raster scans
(T-scans) that run parallel to the retinal surface, the device generates both coronal OCT slices (C-scans) and conventional saggital OCT cross-section (B-scans).² Rapidly acquired stacks of C-scans are aligned looking at retinal vessels and are reconstructed into 3-D volumes for thickness mapping. Scanning at a rate of 100 slices per second, each OCT/SLO slice samples a 512 x 512 or 256 x 256 pixel matrix. Multiple bursts of slices are used to compare sample sets and discard ones with movement artifacts, which are often generated by the cardiac pulse.

METHODS FOR PILOT STUDY

Figure 3. Subtraction of the topographic maps. A map of the subtracted aligned images reveals the differences between corresponding points and shows the overlap between scans. Average thickness change and standard deviation is calculated for the overlapping points. Unchanged areas remain gray and do not affect the average change or standard deviation.

To evaluate the utility of this approach we followed the progress of 20 patients undergoing pegaptanib sodium therapy in the spring of 2005 shortly after it became commercially available. The study was designed as a retrospect case series. Patients were imaged before and after treatments and topographic maps were generated. Automated comparative analysis was performed following alignment of serial maps to evaluate the thickness changes. Here is a discussion of 2 cases from the study.

STUDY DISCUSSION

In this small pilot study of topographic mapping of pegaptanib sodium therapy, several observations were evident. First, OCT B-scan cross-sections frequently do not portray the effectiveness of the therapeutic response accurately. Since each slice is only 20 μm wide, the response seen in one slice often does not represent the response of the larger 10 x 10 area of the macula. This was evident in Case 2, where the OCT slices looked similar but the topography revealed a more extensive response.

Richard B. Rosen, MD, is director of the New York Eye and Ear Infirmary (NYEE) Advanced Retinal Imaging Center, and Thomas O. Muldoon, MD, is director of its vitreo-retinal service. Patricia Garcia, MD, is a research associate. Dr. Rosen can be e-mailed at RRosen@NYEE.EDU. Dr. Rosen has received travel reimbursements from OTI; Dr. Garcia is a consultant for OTI; and Dr. Muldoon does not have any financial interest in the technology discussed in this article.

CASE 1

Figure 4. Serial B-scan OCT slices during pegaptanib sodium therapy. Images taken at 6-week intervals preceding injections are seen in the upper row and in the lower left, and large cystic spaces appeared unresponsive to the therapy. The lower right frame shows resolution of the cystoid spaces and retinal thickening following intravitreal triamcinolone acetonide.

A 74-year-old woman who had a history of choroidal neovascular membrane in her right eye had been treated with verteporfin (Visudyne, Novartis)
1 year prior to presentation. Her visual acuity (VA) was 20/400 in the right eye and 20/30 in the left eye. The patient presented after 2 days of visual disturbance in her left eye with visual acuity reduced to 20/150. Fundus exam revealed a macular hemorrhage. Fluorescein angiogram confirmed the presence of a poorly defined choroidal neovascular membrane and indocyanine green (ICG) angiography showed a hot spot in the same region. Following pegaptanib sodium therapy, performed on the day of presentation, the patient experienced a return of visual acuity to the level of 20/50 measured at 2 weeks post treatment. OCT/SLO was performed pre treatment and two weeks posttreatment with pegaptanib sodium. Figure 3 shows a representative B-scan OCT slice beneath topographic maps taken during the same session. The B-scan OCT images demonstrate significant decrease in cystoid edema and thickness, which is described quantitatively by the topographic maps. In this case the B-scan OCT, topographic map, and VA all demonstrated a positive therapeutic response. This was, in fact, unusual since often there was a disconnect with functional response preceding or lagging behind anatomic change.

CASE 2

A 78-year-old woman with a history of unsuccessful treatment with photodynamic therapy (VA=20/400) in her left eye presented with a new onset of macular hemorrhage and decreased VA to 20/70 due to an occult choroidal neovascular membrane in her right eye. She received pegaptanib sodium as part of an open label trial and her visual acuity stabilized in the range of 20/100 to 20/150 on a standard treatment regimen. OCT /SLO B-scans taken in August, October, and December showed persistence of cystoid changes in the macula despite apparent stabilization.

Figure 5. Comparison of topographic maps. The thickness grid pre and post first pegaptanib sodium treatment showed a small reduction in the size of the thickened area. The version of the software at this point did not allow subtraction

Permission for study monitors was granted for the patient to receive an intravitreal injection of 4 mg of triamcinolone acetonide (Kenalog, Bristol-Myers Squibb). B-scan OCT (Figure 4) revealed a dramatic reduction, but not elimination, of the cystic edema. Topographic map comparison documented the full extent of the marked reduction in the thickened area, which is seen in Figure 7B. As impressive as the anatomic and quantitative changes were with the addition of the steroid, the VA improvement was modest to the level of 20/80 and subjectively the patient was not able to notice the improvement.

Second, comparison of OCT changes over time can only be accurately performed if registration between studies can be assured. Even with scans and topographic maps centered on fixation, alignment of blood vessels between corresponding SLO images revealed that fixation-centered scans demonstrated considerable shifts and rotation that were only evident in subtraction overlays. This was seen in Figures 3 and 7 and to a lesser extent in Figure 6. Third, with regard to the relationship between OCT anatomy and VA, topographic maps may be able to define the relationship better. This study, however, was too small and was not designed to answer that question. Finally, the variability of response to treatment even in the same patient contains a variability component. This is encouraging since the risk of choroidal neovascular membranes in the fellow eye is significant, and patients having limited response in the first eye may respond more favorably in the second eye.

SUMMARY

Figures 6A and 6B. Comparison and subtraction of maps shows the changes before the second injection vs. before the third injection. Subtraction reveals some reduction and some increased thickening.

The OCT/SLO retinal topographic mapping offers a potentially more accurate imaging system for measuring anatomic response to treatment with anti-VEGF drugs. The point-to-point registration of OCT/SLO ensures accurate assessment of topographic changes over time. Analysis of patterns of thickness changes may be helpful in predicting which eye will respond more favorably to treatment. Patterns of change also may help with decisions as to when therapy may be discontinued.

The OCT/SLO approach of linking surface anatomy to internal features and thickness changes is a major advance in retinal monitoring since it ensures that measurements used for therapy decisions reflect the changes being monitored. Since this study, other anti-angiogenic agents have been made available and the concept of discontinuing therapy as soon as stability is demonstrated has become increasingly popular. retinal topographic maps may serve as an important tool for documenting a clinical stability.

Having a technology like the OCT/SLO that employs a more comprehensive scanning pattern and a larger field of view can potentially assist physicians in gauging the condition of the macula in response to therapies, and may influence treatment patterns. However, further studies are warranted to confirm these possibilities. RP

REFERENCES

1. Rosen RB, Will D, Garcia P, et al. OCT ophthalmoscope. In: Alfaro DV, Liggett PE, Mieler WF, et al. Age-Related Macular Degeneration: A Comprehensive Textbook. Philadelphia, Pa: Lippincott Williams & Wilkins; 2005.

2. Rosen RB, Dunne S, Podoleanu AG, Garcia PM. OCT ophthalmoscopy. Posterior Segment Imaging of the Eye. Philadelphia, Pa: Elsevier; 2006.

Figures 7A and 7B. Subtraction of topographic maps showing the differences before the third injection and 6 weeks following triamcinolone. Registered subtraction reveals a dramatic response to treatment with marked reduction of thickness.