Local Glaucomatous Defects of the Circumpapillary Retinal Nerve Fiber Layer Show a Variety of Patterns of Progression

Precis: The region of glaucomatous progression, seen on optical coherence tomography (OCT) images of the circumpapillary retinal nerve fiber layer (cRNFL), increases in width and depth in all eyes, but shows a variety of different patterns of loss across eyes. Purpose: The purpose of this study was to examine the patterns of cRNFL loss secondary to glaucomatous progression in a region associated with the superior hemifield of the 24-2/30-2 visual field (VF). Methods: Twenty-four eyes (20 patients) with a diagnosis of glaucoma and evidence of progression on OCT had OCT disc cube scans on at least 3 separate visits (mean follow-up 7.4 y; range: 3.9 to 11.4). Circumpapillary b-scans were derived after enface images were aligned to assure that the study region (ie, 0 to −135 degrees, where 0 degree is 9 o’clock, on a right eye) coincided. Within this region, a region of progression (ROP) was defined based on the loss in cRNFL thickness between the first and subsequent visits. The width of the ROP was determined, along with the locations of its leading (close to fixation) and trailing edges. In addition, for each ROP, the location and depth at the point of maximal loss, total loss, and average remaining retinal nerve fiber layer were measured. Results: The ROP proceeded both toward and away from fixation. Across eyes, the ROP varied widely in width (32 to 131 degrees, mean 82.7 degrees), location, and loss at point of deepest loss (22 to 99 μm, mean 52.9 μm), as well as total cRNFL loss. Conclusions: All eyes showed a widening and deepening of the ROP, but a variety of different patterns of progressive cRNFL loss. Thus, one should expect considerable variation in patterns of VF loss. Furthermore, conventional metrics (global or quadrant cRNFL thickness) do not fully depict the progressive changes that can be appreciated by inspecting OCT images.

T here is no litmus test for the detection of glaucoma or its progression. Thus, most clinicians use both functional [visual field (VF)] and structural/anatomic information. Although traditionally the anatomic test was the fundus examination, typically with fundus photography, clinicians are increasingly using optical coherence tomography (OCT) along with VFs for confirmation of progression.
The most common OCT analysis involves measurement of the thickness of the circumpapillary retinal nerve fiber layer (cRNFL) seen on circumpapillary b-scan images. 1 These b-scan images are either obtained from OCT circle scans or derived from OCT cube scans. In both cases, the cRNFL thickness is measured. Typically, the clinician looks at summary measures, such as the global average or the average of portions (eg, quadrants) of the cRNFL thickness. These summary measures (called metrics here) can miss damage. [2][3][4] Moreover, these metrics ignore the information that can be obtained by looking at the actual circumpapillary b-scan image. 2 We have argued that these images should be examined the way a radiologist would examine a magnetic resonance imaging scan, namely by closely inspecting the details of the OCT b-scan image. 2,4 Examination of the cRNFL defects seen on b-scan images reveal a wide range of patterns of local damage, challenging the usefulness of trying to classify VF patterns into particular groups. 5 In any case, Hood et al 5,6 argued that local defects can vary in location, depth, width and homogeneity of damage. However, relatively little is known about how these local defects change with progression.
The purpose of this study was to better understand how the cRNFL patterns of damage change as glaucoma progresses. In particular, we examined derived OCT circumpapillary b-scan images of eyes with clear evidence of progression over a period of at least 3 years. Furthermore, we focused on the region of the cRNFL associated with the superior portion of the 24-2/30-2 VF, as this allowed us to align the scans obtained at different times in a region known to be prone to progression. [7][8][9] domain OCT (Topcon 3D-OCT 2000) cube scans of the disc obtained between August 2007 and July 2019. In particular, eyes were included if there was evidence of progression in the region of study described below and if they met the following inclusion criteria: 3 or more years between the first and last OCT disc cube scans, and at least 3 separate visits with adequate scan quality. The time between the first and last visits ranged from 3.9 to 11.4 (mean ± SD: 7.4 ± 2.6) years. At the first visit, the age of patients ranged from 32 to 74 (mean ± SD: 57.6 ± 11.5, median 61) years. Thirteen of the 20 patients were women and 7 were men. Mean deviation of a reliable 24-2 VF closest in date with the first and last scans ranged from −8. 19

Region of Study and Progression Reports
We chose to study a restricted region of the disc for 2 reasons. First, we were interested in variations in regions of the disc associated with the VF tested with the 10-2 and 24-2 patterns. Second, due to head tilt and/or scanning artifacts it was not always possible to align the images of the scans from 2 tests (see below). Thus, we needed to study a smaller region. We chose to focus on the disc region associated with the superior 24-2/30-2 VF, as it is the most common area damaged by glaucoma and includes a region associated with the macula. 5,10-14 Figure 1 shows the derived circular b-scans (A, C) for 2 visits and the enface images with the location of the derived circle (B, D). Using a custom-made MATLAB R2015a (MathWorks, Natick, MA) program, the enface images of disc cube scans from different sessions were aligned using blood vessel landmarks (blue arrows in Figs. 1B, D, corresponding to blue vertical lines in Figs. 1A, C) to assure that the study region (ie, 0 to −135 degrees, white arrow in A) of each scan coincided. This portion of the cRNFL includes the region associated with the superior hemifield of the 24-2 and 30-2 VF. 5,15 Notice in Figure 1 that the blood vessels are reasonably well aligned in the study region (blue vertical lines), while they are misaligned for the superior disc where the yellow vertical lines correctly fall on the shadows of the vessels in the lower panel (red arrows), but miss these in the upper panel (yellow arrows).

Measuring Changes in the Region of Progression (ROP)
The ROP was defined based upon the difference in cRNFL thickness between the first and subsequent visit. This is illustrated in Figure 2 for a different eye than in Figure 1. The upper panels (A, B) show the derived circle scans for the 2 visits and the enface images with the location of the derived circle scan superimposed (right panels of A and B). Figure 2C is a plot of the cRNFL thickness in the study region for the first (solid black curve) and Nth visit (dashed blue curve). The difference between them is shown as the light blue line at the bottom of Figure 2C. This panel is enlarged as Figure 2D. The location and amount of loss at the point of maximal loss (red arrow in Figs. 2C, D) was measured. Then starting from this point of maximal loss and FIGURE 1. Region of study and generating progression reports: An example of the custom-made progression report. A and C, The derived circular b-scans are for the first and Nth visit. The blood vessels are aligned (blue vertical lines) in the study region [ie, 0 degree (9 o'clock) to −135 degrees (between 4 and 5 o'clock on right optic disc)]. They are misaligned for the superior disc where the yellow vertical lines correctly fall on the shadows of the vessels in the lower panel (red arrows), but miss these in the upper panel (yellow arrows). B and D, The enface images of the disc cube scans with the location of the derived circle scan from 2 different sessions were aligned using blood vessel landmarks (blue arrows in B, D, corresponding to blue vertical lines in A, C) to assure that the study region of each coincided. The eye shown corresponds to eye 1 in Figure 3. moving along the difference curve to the left (blue dashed arrow, Fig. 2D) and to the right (orange dashed arrow, Fig. 2D), the leading (close to fixation, blue arrow) and trailing edges (orange arrow) were defined as the first location on the difference curve which fell below −5 μm. The width of the ROP was taken as the distance (in degrees) between the leading and trailing edges.
Total loss in the ROP was also measured (hatched region, Fig. 2D) by calculating the area, in μm 2 , under the difference curve starting and ending at the first points on the curve which fell below −5 μm. The average RNFL thickness at the last scan date was also calculated for each eye.

Variations in ROP Between First and Last Scans
The hatched region in Figure 2D shows the cRNFL loss in the ROP between the first and last scans. This region of loss is shown in Figure 3 for each of the 24 eyes. Each curve represents the difference curve of an eye (ie, light blue curve in Fig. 2D), ordered according to location of the point of maximal loss. Black indicates clinical diagnosis of POAG; red indicates normal tension glaucoma, and blue secondary glaucoma (pigmentary or exfoliation glaucoma).

ROP Width
The width of the ROP, which is the width of the curves in Figure 3, ranged from 32 to 131 degrees (mean 82.7 ± 27.1 degrees). This represented between 24% and 97% of the entire study region. Even after correction for time between the first and last scans, the change in width of the ROP varied over a large range, from 2.6% to 16.9% of the study region per year.

ROP Depth
In all eyes, the ROP widened with concurrent deepening. No eyes showed widening without deepening. However, there was a large variation in the location and depth of the point of maximal loss. The red circles in Figure 3 show the location of maximal loss, which varied from −38 to −112 degrees. The amount of loss at the location of maximal loss ranged from 22 to 99 μm (mean 52.9 ± 20.4 μm). After adjusting for time, the wide range of depths persisted.

ROP Total cRNFL Loss
There was also a wide range of total cRNFL loss. Total cRNFL loss in the ROP between the first and last scans (hatched area under curve in Fig. 1) ranged from 27,090 to 134,185 μm 2 (mean 67,997.9 ± 32,182.4 μm 2 ). After adjustment for time, the wide variation in total loss per year, which was more than a factor of 6, persisted.

Direction of Progression Relative to Fixation
All 24 eyes had at least 3 scan dates from which we could determine whether progression proceeded toward or away from the location on the disc associated with fixation (∼0 degree, or 9 o'clock on right eye). Figure 4 shows an example of an eye with progressive loss of cRNFL both towards 0 degree (leading edge) and in the opposite direction (trailing edge). The initial edges of the ROP are shown as the gray dashed (leading edge) and solid (trailing edge) vertical lines, and the final edges as the dark blue vertical dashed and solid lines. The dark blue solid line shows extension of the trailing edge past the study region (−135 degrees). The ROPs of 16 of the 24 eyes (67%) increased in width both towards and away from fixation, while 5 increased just towards fixation (21%), and 3 (12%) only away from fixation.

Variations in ROP Patterns of cRNFL Loss
The results above suggest a wide range of variation in patterns of progression in width, loss at location of maximal loss, and total loss in the ROP. To illustrate and confirm this variation, we compared patterns of loss for eyes with similar ROP width, ROP total loss, and retinal nerve fiber layer remaining in the study region at the last scan date. and to the right (orange dashed arrow), the leading (close to fixation, blue arrow) and trailing edges (orange arrow) were determined as the first location on the difference curve which fell below −5 μm. The ROP width was taken as the distance (in degrees) between the leading and trailing edges. Hatched region indicates the area of total loss in the ROP (μm 2 ). The eye shown corresponds to eye 11 in Figure 3.

Similar ROP Width But Different Patterns
The 3 eyes represented in Figure 5A-C all had an ROP with similar width (99.5 to 106.2 degrees), but different patterns of loss. Notice, for example, the variation in the width, number, and location of particularly deep regions of loss (white arrows).

Similar ROP Total Loss But Different Patterns
Eyes with similar ROP total loss also showed a variety of patterns of loss as seen in Figure 5D-F. Figure 5D shows wide loss which becomes deeper away from fixation. Figure 5E shows approximately equal depth across the ROP. Figure 5F shows a single deep area of local loss which is less wide than D or E.

Similar Average Thickness Remaining in Study Region But Different Patterns of Retinal Nerve Fiber Layer
Eyes with similar average cRNFL thicknesses at the last scan dates also demonstrated widely varying patterns of loss as illustrated in Figure 5G-I. The depth of local loss, as well as the width of the ROP varies in these eyes. Figure 5G demonstrates greatest loss (white arrow) towards 0 degree relative to the blood vessel, while Figure 5H shows loss on either side of blood vessels, and Figure 5I shows loss on the side of the blood vessels away from 0 degree.

DISCUSSION
The purpose of this study was to better understand the patterns of change in OCT cRNFL as glaucoma progresses. We have previously argued that at any point in time there was a wide range of patterns of glaucomatous damage, 5 even in eyes with similar 24-2 mean deviations. Here we report that the change over time, progression, also shows a wide range of patterns. In particular, the ROP within the 135 degrees region studied shows marked variation in width, depth, amount and location of maximum cRNFL loss, average cRNFL remaining, as well as heterogeneity of damage. These findings underscore the limitations of conventional metrics (global or quadrant cRNFL thickness) in fully depicting progressive changes seen with OCT.
Although methods for evaluating the glaucomatous progression of cRNFL through average thickness measurements have been well-studied, 1,16-19 patterns of cRNFL progression have not been studied. Studies using global and sectoral average cRNFL thickness measurements to track progression of glaucoma have highlighted that glaucomatous damage in the inferotemporal sector, which is included in our region of study, is most vulnerable to progression. [7][8][9]20 Other studies, which explore patterns of progression through fundus photography and/or OCT, have found an increased tendency of progressive widening of glaucomatous defects towards fixation more than away, and widening with concomitant deepening of loss. 6,17 Our results are consistent with the general framework that progression primarily occurs through widening and deepening of existing defects. 21 Even taking into consideration the uncertainty of defining exactly the start and end of the ROP, it is clear there is a wide variation in width. In addition, we provide evidence for 4 other general features of progression. One, progression occurs over a relatively wide region. It is rarely a very local phenomenon. Two, it occurs both toward and away from the region associated with fixation. Three, there is wide variation in the pattern of loss even in our study region, which covers <40% of the disc. As the ROP can vary in location, width, depth, and heterogeneity of loss over time, simple metrics such loss of global or quadrant or clock hour cRNFL thickness will not capture the nature of the loss in cRNFL. Thus, these observations point to the need for the clinician to closely examination OCT images (b-scans), as well as the thickness and probability maps, to understand the nature and location of progression for any given eye.

Implications for VF Changes
Relatively little is known about the patterns of VF loss as glaucoma progresses. Suh et al 22 found patterns of VF progression which were compatible with patterns of cRNFL loss seen on fundus photography, namely deepening, widening of an existing scotoma, and appearance of a new scotoma. Boden et al 21 report that most common VF progression patterns are expansion alone, deepening alone, and combination of expansion and deepening. Other studies have reported various rates and patterns of symmetric and asymmetric VF changes in glaucoma patients and identified the superonasal quadrant as most vulnerable to glaucomatous damage, 14,23-25 a region of the field associated with our region of study. There have also been recent attempts to categorize discrete patterns of VF progression into identifiable archetypes. 12 Our results, in conjunction with studies showing good correlation between cRNFL and VF progression, 26,27 argue against discrete VF archetypes, as well as a simple deepening of a defect without a widening. What should the changes in the pattern of damage seen on VFs look like based upon the variation in ROP describe here? First, recall that our region of study includes the portion of the cRNFL associated with the entire superior region of the 10-2 and 24-2/30-2 VF patterns. 3 Second, there is excellent topographical agreement between cRNFL loss and VF loss, 9 as long as both 10-2 and 24-2/30-2 tests are performed. 28,29 Thus, we predict that the pattern of cRNFL loss should, on average, be reflected in the pattern of VF loss in the associated region. In particular, wider and deeper VF defects are to be expected. However, in addition we should expect wide variations in the spatial patterns of local VF loss within the affected regions, and these variations should, on average, be predictable from the changes in the OCT ROP. Consistent with this prediction are our earlier studies 30,31 which showed that eyes with regions of nearly complete cRNFL loss had VF losses worse than −15 dB in the corresponding field locations.
Directions for further study include comparing patterns of cRNFL progression with changes in the thickness of the OCT ganglion cell layer, as well as changes in VF sensitivity. [32][33][34] Although a thorough investigation of the correlation between cRNFL and VF progression are out of the scope of this project, examples from available data in this project corroborate what has been reported, that cRNFL progression is associated with expansion and deepening of VF defects.

Limitations
The most obvious limitation is the sample size of 24 eyes, which was restricted by the inclusion criteria. To be selected, an eye had to progress, and have scans that could be aligned from at least 3 test days over a period of > 3 years. However, a larger sample is not likely to lead to less variation. On the other hand, a larger sample would allow us to ask if the type of glaucoma (eg, POAG vs. secondary) affect the pattern of damage.
A second limitation is the impact of blood vessels on the pattern of loss in the ROP. Clearly, blood vessels influence this pattern and more sophisticated methods will need to be devised to understand this influence. We can, however, rule out the simple model that would invalidate our conclusions. This model assumes that the loss is the same everywhere, say 25% of cRNFL in the region of study. Because glaucoma has a relatively small effect on blood vessel size 35 this model predicts variation in patterns of ROP even with a homogenous loss of cRNFL. However, the examples supplied in Figure 5 clearly rule out this simple model. Furthermore, a wide variety of patterns of local damage can be seen on either or both sides of vessels, or in regions without prominent blood vessels (Fig. 5).

CONCLUSIONS
The region of the cRNFL studied included the 135 degrees portion of the disc that is associated with the superior 24-2/30-2 VF locations. The loss of cRNFL within this region proceeds both toward and away from the location of the cRNFL associated with the macula and fixation. Although all eyes showed a widening and deepening of the ROP, this region also showed a wide variety of patterns of cRNFL loss over time across eyes. Thus, one should expect considerable variation in patterns of VF loss from eye to eye. In addition, conventional metrics (global or quadrant cRNFL thickness) do not fully depict the progressive changes that can be appreciated by inspecting OCT images.