Topographic Localization of Macular Retinal Ganglion Cell Loss Associated With Localized Peripapillary Retinal Nerve Fiber Layer Defect

Abstract

Purpose.: To investigate the topographic relationship between ganglion cell–inner plexiform layer (GCIPL) and peripapillary retinal nerve fiber layer (pRNFL) defects in open-angle glaucoma patients with localized RNFL defects, using spectral-domain optical coherence tomography (SD-OCT).

Methods.: We analyzed 140 eyes of 140 patients showing a localized RNFL defect in one hemifield, the angular width of which was limited to one clock-hour sector. The RNFL and macular GCIPL scans were obtained using SD-OCT. The clock-hour location and width of pRNFL defects on the RNFL deviation map were determined, and their topographic association with corresponding GCIPL defects on the GCIPL deviation map was assessed.

Results.: A “GCIPL deviation frequency map” demonstrating GCIPL defects corresponding to six different clock-hour locations of pRNFL defects was obtained, and it revealed the following specifics: (1) pRNFL defect at 12, 11, and 10 o'clock corresponded to GCIPL defect in the superior macula, as those at 8, 7, and 6 o'clock did to those in the inferior macula; (2) the overall GCIPL defect had an arcuate shape that appeared as a continuation of the pRNFL defect; (3) the temporal macular region was the frequently damaged site in either hemifield, and was larger in the inferior hemifield than in the superior hemifield. Additionally, an interindividual variability of GCIPL defect was noted for patients with the same clock-hour location of pRNFL defect.

Conclusions.: The GCIPL deviation frequency map demonstrating the topographic relationship between pRNFL and GCIPL defects was generated using SD-OCT. Our results indicated the topographic location of retinal ganglion cell death associated with clock-hour location of pRNFL loss in vivo.

Introduction

Spectral-domain optical coherence tomography (SD-OCT), providing higher-resolution images with more precise segmentation of the retinal layers than time-domain OCT, has enabled the measurement of ganglion cell–inner plexiform layer (GCIPL) thickness in the macular region. ^1–3 As progressive retinal ganglion cell (RGC) loss is the key pathology of glaucoma, ⁴ decreased GCIPL thickness can reflect glaucoma-induced macular RGC loss. ^5–8 Previous studies have demonstrated the usefulness of such measurement for glaucoma diagnosis using various parameters such as GCIPL ^9–13 or ganglion cell complex (macular retinal nerve fiber layer [RNFL] + GCIPL) ^6,12,14,15 thicknesses. These studies reported that glaucoma-diagnostic performance of GCIPL parameters was similar to that of RNFL parameters, ^9,10,13,16 which have been widely applied to the diagnosis of glaucoma. 

When both GCIPL and RNFL parameters are used for the diagnosis of glaucoma, detailed knowledge on their spatial relationship may be needed. Hood et al. ^17,18 presented a schematic model of RNFL projection and glaucomatous macular GCIPL damage using SD-OCT to explain the topographic correlation between the two regions. Garway-Heath et al. ¹⁹ manually traced the paths along the RNFL from each visual field test location toward the sector at the optic nerve head (ONH) to produce a map correlating macular function with structural damage in glaucoma. Understanding such an anatomical relationship between two regions would be a prerequisite for topographic prediction of macular ganglion cell damage from the ONH or peripapillary RNFL (pRNFL) damage, or vice versa. 

In this study, accordingly, we investigated the topographic relationship between ganglion cell death in the macular area and axonal death in the peripapillary area using SD-OCT–derived RNFL and GCIPL deviation maps. A “GCIPL deviation frequency map” was obtained, based on all GCIPL defects shown on the GCIPL deviation map and their correspondences with pRNFL defects at different clock-hour locations, thus demonstrating the correlation between the two defects in terms of shape and location. 

Methods

Subjects

The medical records of 140 eyes of 140 consecutive open-angle glaucoma patients meeting eligibility criteria, all of whom had undergone SD-OCT imaging at the Glaucoma Clinic of Seoul National University Hospital between January 2012 and January 2013, were retrospectively reviewed. This study adhered to the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Seoul National University Hospital. 

All patients underwent a complete ophthalmic examination, including measurement of visual acuity, measurement of spherical equivalent, slit-lamp examination, measurement of intraocular pressure by Goldmann applanation tonometry, gonioscopy, fundoscopic examination with ≥5-mm pupil dilation, stereoscopic color-disc photography (SDP), red-free RNFL photography (RNFLP; Vx-10; Kowa Optimed, Tokyo, Japan), standard automated perimetry using Swedish interactive threshold algorithm 30-2 (Humphrey Field Analyzer II; Carl Zeiss Meditec, Inc., Dublin, CA, USA), and SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec, Inc.). 

All patients had to meet the following criteria for inclusion: best corrected visual acuity (BCVA) ≥ 20/40, refractive error within ±5 diopters (D) in sphere and within ±3 D in cylinder, and a normal anterior chamber and open angle on slit-lamp and gonioscopic examinations. Patients with a history of ocular surgery other than simple cataract surgery, or any history or evidence of retinal pathology, diabetes mellitus, or nonglaucomatous optic neuropathy, were excluded. 

Glaucomatous eyes were defined as those showing glaucomatous optic disc appearances (e.g., neuroretinal rim thinning, notching, and/or RNFL defects) and corresponding glaucomatous visual field defects, as confirmed by at least two consecutive visual fields. The glaucomatous visual field defects were defined as a cluster of ≥3 points with P < 0.05 on the pattern deviation map in at least one hemifield, including ≥1 point with P < 0.01; a pattern standard deviation (PSD) of P < 0.05; or glaucoma hemifield test result outside the normal limits. Only patients with reliable visual field test results (fixation loss < 20%, false-positive errors < 15%, and false-negative errors < 15%) were included. In cases in which both eyes of a patient were eligible, one was chosen randomly for inclusion. 

Stereoscopic Color-Disc Photography and Red-Free Retinal Nerve Fiber Layer Photography

The SDP and RNFLP were obtained after maximum pupil dilation. Sixty-degree, wide-angle views of the optic disc, focused on the retina using a built-in split-line focusing device and centered between the fovea and the optic disc, were obtained and reviewed on a liquid crystal display monitor (FLATRON; LG Display Co., Ltd., Seoul, Korea). ²⁰  

The SDP and RNFLP were evaluated independently by two glaucoma specialists (KEK and KHP) in a masked fashion. The presence of a glaucomatous disc and a localized RNFL defect was confirmed by consensus between the two observers; eyes thereupon were classified into one of the following categories: normal, localized defect, diffuse atrophy, or ambiguous. A localized RNFL defect was defined as a dark wedge-shaped defect, its tip touching the optic disc border and its width larger than that of the major retinal vessels at a one-disc-diameter distance from the edge of the optic disc in the brightly striated pattern of the surrounding RNFL. ²¹ Only cases that both observers classified as having one localized defect in either hemifield of the retina and in which they agreed on the location of the defect were included. 

Spectral-Domain Optical Coherence Tomography

One macular and one optic disc scan were acquired through a dilated pupil. All of the OCT examinations were performed by the same well-trained technician. Only accurate images with a signal strength of ≥8 (10 = maximum) without misalignment, segmentation failure, suspicious motion artifacts, or decentration of the measurement circle were included. Right-eye orientation was used in documenting all of the OCT data. For both the right and the left eye, 12 o'clock corresponded to the superior region; 3 o'clock, the nasal region; 6 o'clock, the inferior region, and 9 o'clock, the temporal region. Mainly, deviation maps from the ganglion cell analysis (GCA) and RNFL algorithms of Cirrus OCT (Carl Zeiss Meditec, Inc.) were used in the study. 

The optic disc images were obtained using dataset of 6- × 6- × 2-mm optic disc cube composed of 200 A-scans derived from each of 200 B-scans. From the cube dataset, the built-in software automatically detects the center of the optic disc and then extracts from the dataset a peripapillary circle 3.46 mm in diameter and consisting of 256 A-scans for calculation of the RNFL thickness at each point of the circle. The RNFL thickness at each axial scan was measured, and a RNFL thickness deviation map was generated. The RNFL thickness deviation map comprises 50 × 50 superpixels that appear yellow or red if the RNFL thickness is less than the lower 95th or 99th percentile ranges, respectively. 

The GCA algorithm used in this study, included in the 6.0 software version of Cirrus OCT (Carl Zeiss Meditec, Inc.), performs 200 horizontal B-scans, each comprising 200 A-scans, in over 1024 samplings to detect and measure macular GCIPL thickness within a 6- × 6- × 2-mm cube centered on the fovea. The annulus has inner vertical and horizontal diameters of 1 and 1.2 mm, respectively, and outer vertical and horizontal diameters of 4 and 4.8 mm, respectively. The GCA algorithm identifies the outer boundaries of both the RNFL and the inner plexiform layer, and the difference between these yields the GCIPL thickness. Full details on the mechanism of the GCA algorithm are available elsewhere. ^2,5 The software analyzes the thickness values, compares them with the built-in internal normative database, and generates the GCIPL deviation map. On the deviation map, the defects appear as yellow or red to represent GCIPL thickness less than the lower 95th or 99th percentile, respectively. 

Assessment of Localized Peripapillary Retinal Nerve Fiber Layer Defect

On the RNFL thickness deviation maps, radiating defects from the ONH shown in yellow (<5% level) or red (<1% level) were defined as abnormal; those extending at least 10 contiguous superpixels corresponding to the location on the RNFLP were defined as a localized RNFL defect. Determination of the clock-hour location of the color-coded pRNFL defect on the RNFL deviation map was performed by a blinded observer (BWY) using a computer program written in MATLAB R2011a (The MathWorks, Inc., Natick, MA, USA), and this method has been previously described in detail. ²²  

In determining main pRNFL defect, the MATLAB program (The MathWorks, Inc.) detected areas of contiguous color-coded pixels at least 10 superpixels in length and more than 2 superpixels in width. Detected pixels more than 4 superpixels from the 3.46-mm-diameter circle were selected. To localize the clock-hour location of the pRNFL defect, a clock-hour circle was positioned on an outline of the 3.46-mm-diameter circle (Fig. 1A). The blue lines in the circle divided the area into 12 clock-hour sectors, each of which was equal to 30°. The MATLAB program then drew a concentric inner circle, its diameter corresponding to the margins connecting the two points indicating the end of Bruch's membrane. The main color-coded area was defined, and two red lines bisecting each sector were drawn so as to contain the outermost edges of that color-coded pRNFL defect. If the main color-coded area, indicating the defect width, fit within one clock-hour sector (equal to or less than 30°, indicated by blue lines), the subject was included in the analysis (Figs. 1A, 1B). However, if the main color-coded area covered more than one sector, the MATLAB program automatically excluded the image and its subject from further analyses. 

Figure 1

 

Determination of clock-hour location of main localized peripapillary retinal nerve fiber layer (pRNFL) defect on the SD-OCT deviation map. (A) A clock-hour circle the same as that shown on the SD-OCT clock-hour thickness map was positioned over the 3.46-mm-diameter circle outline. The inner blue lines divided the area into 12 clock-hour sectors, each of which was equal to 30°, and two red lines bisecting each sector containing the outermost limits of the color-coded pRNFL defect were drawn. Then, a concentric inner circle of a diameter corresponding to the margins and connecting the two points (arrows) indicating the end of Bruch's membrane was drawn. The example shows that the main color-coded area is within one clock-hour sector and that the clock-hour location of the pRNFL defect was determined to be 7 o'clock. (B) Another example showing a localized pRNFL defect in the right eye. Note that the main color-coded pRNFL defect lies within one clock-hour sector and that the clock-hour location was determined to be 11 o'clock.

Ganglion Cell–Inner Plexiform Layer Deviation Frequency Map

To evaluate the topographic relationship between pRNFL and GCIPL defect, all of the GCIPL defects from Cirrus OCT (Carl Zeiss Meditec, Inc.) deviation maps corresponding to the same clock-hour locations of pRNFL defects were merged with the aid of a computer program written in MATLAB R2011a (The MathWorks, Inc.). In determining main GCIPL defect, contiguous color-coded pixels of at least 20 superpixels in area and more than a boundary of one superpixel away from the inner annulus were detected. If the largest area was less than 20 superpixels, contiguous color-coded pixels of more than 10 superpixels in area were selected; and if the largest area was less than 10 superpixels, the image was excluded so as to discard the artifact. After adding all GCIPL deviation map images, one total GCIPL defect image from each group of patients showing the pRNFL defects at different clock-hour locations (12, 11, 10, 8, 7, and 6 o'clock), defined as a GCIPL deviation frequency map, was obtained (right column in Fig. 2). The map was represented in the color-bar scale from blue to red. The red-colored region indicated GCIPL defects detected at the highest frequency in the pRNFL clock-hour group; the blue-colored region, close to zero frequency. The reference line, based on the fact that RNFL fibers do not cross the horizontal raphe, was considered as the horizontal line of the GCIPL deviation map, passing through the center of the inner and outer annulus. The temporal point of the line was set as 0°, and the angle was assessed in a clockwise direction for right eyes, all left eye images having been converted to the right-eye format. The individual GCIPL defect area, the GCIPL defect area detected in more than 50% of the patients, and the superior and inferior hemiannulus area were assessed by pixel counting using the MATLAB program. 

Figure 2

 

Photographic examples of patients with localized peripapillary retinal nerve fiber layer (pRNFL) defects at different clock-hour locations and corresponding ganglion cell–inner plexiform layer (GCIPL) defects on the SD-OCT deviation map. The left column shows the localized pRNFL defects limited to one 12, 11, 10, 8, 7, and 6 clock-hour sector on the RNFL deviation map, and the middle column shows the corresponding GCIPL defects on the GCIPL deviation map. The right column shows the GCIPL deviation frequency map, localizations of all of the GCIPL defects corresponding to the pRNFL defects at each of the clock-hour locations (12, 11, 10, 8, 7, and 6 o'clock).

Statistical Analysis

The baseline characteristics of patients with different clock-hour locations of pRNFL defect were compared by one-way analysis of variance and χ² test for continuous and categorical variables, respectively. To demonstrate the distribution of GCIPL defect areas, their proportion to the hemiannulus (superior or inferior) area on the GCIPL deviation map was calculated, and the distribution in each clock-hour group was presented as box-and-whisker plots—the box indicating the median and the whiskers indicating the minimum and maximum values. A comparative analysis of the areal proportion of GCIPL defect among the groups was performed by one-way analysis of variance with Tukey post hoc test. All of the statistical analyses were performed with the commercially available statistics software SPSS version 21.0 (SPSS, Inc., Chicago, IL, USA). Statistical significance was defined as a P value <0.05. 

Results

Preparatory to this study, 246 open-angle glaucoma patients with localized pRNFL defects were reviewed. Of these, 106 patients were excluded from the analysis for the following reasons: poor image quality (15), multiple localized pRNFL defects in the same hemifield (32), localized pRNFL defect(s) larger than 30° (29), and not showing concurrent GCIPL defects on the GCIPL deviation map (30). This left a final sample of 140 eyes of 140 patients with a localized pRNFL defect in either hemifield. 

The demographic and baseline characteristics of the study population are summarized in Table 1. The mean age of the included patients, 75 (53.6%) women and 65 (46.4%) men, was 54.8 ± 10.3 years, and the average mean deviation (MD) was −2.9 ± 2.5 decibels (dB). Patients were divided into groups according to the clock-hour location of the pRNFL defect on the RNFL deviation map: 10 (7.1%) patients showed a localized defect at 12 o'clock, 32 (22.9%) at 11 o'clock, 16 (11.4%) at 10 o'clock, 17 (12.1%) at 8 o'clock, 38 (27.1%) at 7 o'clock, and 27 (19.3%) at 6 o'clock, with concurrent GCIPL defect on the GCIPL deviation map. There were no statistically significant differences in age, sex, spherical equivalent, intraocular pressure, and visual field indices (MD or PSD) among the groups (all P > 0.05). 

Table 1

 

Comparison of Baseline Characteristics Among Open-Angle Glaucoma Patients With Different Clock-Hour Locations of Peripapillary Retinal Nerve Fiber Layer Defect

	Clock-Hour Location of pRNFL Defect						P
	12, n = 10	11, n = 32	10, n = 16	8, n = 17	7, n = 38	6, n = 27
Age, y	49.0 ± 10.6	56.5 ± 12.3	58.8 ± 11.5	55.0 ± 14.1	55.8 ± 11.3	55.8 ± 10.0	0.614*
Sex, M:F	4:6	17:15	7:9	7:10	17:21	13:14	0.940†
Laterality, R:L	4:6	20:12	7:9	10:7	22:16	10:17	0.295†
Spherical equivalent, diopters	−2.8 ± 2.2	−3.2 ± 3.4	−1.5 ± 3.5	−2.5 ± 2.8	−1.1 ± 1.8	−0.9 ± 2.5	0.163*
Intraocular pressure, mm Hg	12.5 ± 1.8	13.4 ± 2.6	13.1 ± 2.0	13.7 ± 0.6	13.7 ± 1.9	13.2 ± 2.4	0.788*
Humphrey C30-2 threshold visual field
Mean deviation, dB	−2.2 ± 2.3	−1.9 ± 2.6	−2.0 ± 1.8	−1.0 ± 1.8	−1.3 ± 1.2	−1.7 ± 2.0	0.784*
Pattern standard deviation, dB	2.6 ± 1.4	5.2 ± 2.4	3.7 ± 1.6	4.8 ± 3.1	4.1 ± 3.1	4.2 ± 2.4	0.399*

*  One-way analysis of variance.

†  χ² test.

Topographic Relationship Between pRNFL and GCIPL Defects

The left and middle columns in Figure 2 present photographic examples for each patient, showing a localized pRNFL defect at each of the six clock-hour locations (12, 11, 10, 8, 7, and 6 o'clock) on the RNFL deviation map and corresponding GCIPL defects on the GCIPL deviation map, respectively. The pRNFL defect within one clock-hour sector in either hemifield showed the corresponding GCIPL defect in the same hemifield: Superior GCIPL defect corresponded to localized pRNFL defect at 12, 11, and 10 o'clock, and inferior GCIPL defect corresponded to those at 8, 7, and 6 o'clock. All of the GCIPL defects showed an arcuate pattern that appeared to be in connection with the pRNFL defect. 

Characteristics of GCIPL Defect Corresponding to Different Clock-Hour Locations of pRNFL Defect

In the right column in Figure 2 are the GCIPL deviation frequency maps generated for corresponding localized pRNFL defects at 12, 11, 10, 8, 7, and 6 o'clock, respectively. Among the clock-hour groups, distinctive patterns of GCIPL defect in regard to location, shape, and area were noted. Moreover, GCIPL defects exhibited a variation in the magnitude of deviation from the fovea between groups on the GCIPL frequency deviation map. 

The detected GCIPL defect characteristics that were common to more than 50% of patients in each clock-hour group, indicated by the red tones on the GCIPL frequency deviation map, were as follows. In patients with 12, 11, and 10 o'clock pRNFL defects, the corresponding GCIPL defects were in the superotemporal macula; and in those with 8, 7, and 6 o'clock pRNFL defects, the corresponding GCIPL defects were in the inferotemporal macula. The proportion of the area and the location of GCIPL defects detected in more than 50% of patients in each group are provided in Table 2. The defects in the inferior hemifield (correlated with 6, 7, and 8 o'clock pRNFL defects) were larger and closer to the fovea than those in the superior hemifield (correlated with 10, 11, and 12 o'clock pRNFL defects). In either hemifield, for the more temporally (from 12 to 10 o'clock) or inferotemporally (from 6 to 8 o'clock) located pRNFL defects, larger magnitudes of macular ganglion cell damage expanding toward the ONH were observed. In addition, the characteristics of GCIPL defects detected in a maximum proportion of patients are presented in Table 2. 

Table 2

 

Proportion of Area and Location of GCIPL Defect on GCIPL Deviation Frequency Map Detected in More than 50% and Maximum Percentage of Patients in Each Group Sharing the Same Clock-Hour Location of Peripapillary Retinal Nerve Fiber Layer Defect

Clock-Hour Location of pRNFL Defect	GCIPL Defect in >50% of Patients		GCIPL Defect in Maximum Percentage of Patients
	Proportion of Area, %	Location, °	Maximum Percentage, %	Proportion of Area, %	Location, °
12, n = 10	4.3	0–22	60.0	1.1	72–98
11, n = 32	8.6	0–38	66.7	0.4	7–13
10, n = 16	19.9	0–91	89.5	0.2	3–11
8, n = 17	70.4	191–360	94.7	2.5	298–334
7, n = 38	33.8	227–360	91.3	3.2	318–355
6, n = 27	17.0	246–360	76.2	0.4	308–318

 The GCIPL deviation frequency map is an image representing the total GCIPL defect corresponding to different clock-hour locations of pRNFL defect, and is generated by merging all of the GCIPL defect images from a spectral-domain optical coherence tomography GCIPL deviation map. The temporal point of the line was set as 0°, and the angle was assessed in a clockwise direction for right eyes.

Wide variation in the size and angular extent of GCIPL defect was observed not only between groups but also within each group. Examples of different GCIPL defects in patients having pRNFL defects at 7 o'clock are presented in Figure 3. The proportion of the area of GCIPL defect (with thickness less than the lower 95th percentile) to the hemiannulus area in the inferior hemifield was 49.8 ± 24.0%, which was significantly larger than that in the superior hemifield (26.5 ± 18.4%, P < 0.001). In a separate hemifield analysis, the areal proportion of the GCIPL defect for the 8 o'clock group (64.3 ± 25.0%) was significantly larger than those for the 7 (47.3 ± 22.2%, P = 0.028) and 6 (38.6 ± 18.1%, P = 0.002) o'clock groups, respectively, in the inferior hemifield (Fig. 4A). The areal proportion of the GCIPL defect for the 10 o'clock group (34.9 ± 18.9%) was significantly larger than that for the 12 o'clock group (13.9 ± 11.8%, P = 0.008), but the difference was not significant relative to the proportion for the 11 o'clock group (24.3 ± 17.4%, P = 0.125). The distribution of the area of the GCIPL defect with thickness less than the lower 99th percentile followed a similar pattern (Fig. 4B). Additionally, the angular extent of GCIPL defects in the inferior hemifield was relatively wider than those in the superior hemifield, and the angular extent of GCIPL defects corresponding to 8 and 10 o'clock pRNFL defects were wider than others in the same hemifield. 

Figure 3

 

Examples of interindividual variation in the pattern of corresponding ganglion cell–inner plexiform layer (GCIPL) defects on GCIPL deviation maps in patients with peripapillary retinal nerve fiber layer (pRNFL) defect at 7 o'clock on the RNFL deviation map of spectral-domain optical coherence tomography.

Figure 4

 

Box plots showing the distribution of proportion of ganglion cell–inner plexiform layer (GCIPL) defect area to hemiannulus area on the GCIPL deviation map for patients with different clock-hour locations of peripapillary retinal nerve fiber layer (pRNFL) defect. Note the wide variation within and between the clock-hour groups. The area of GCIPL defect increased in more temporally and inferotemporally located pRNFL defect groups in the superior and inferior hemifields, respectively. (A) Proportion of GCIPL defect area with GCIPL thickness less than the lower 95th percentile. (B) Proportion of GCIPL defect area with GCIPL thickness less than the lower 99th percentile.

Discussion

The importance of macular damage in glaucoma has driven the effort to develop a means of directly assessing the RGC layer and its correlation with other structural parameters in the diagnosis of glaucoma. Several studies have focused on the spatial correlation between damage in the peripapillary and macular regions. ^{17–19,23,24} However, we supposed that more direct plotting of RNFL and GCIPL maps from SD-OCT would help clinicians to combine results from two different algorithms. Therefore, we investigated the SD-OCT deviation maps–based topographic relationship between the macular damage shown by the ganglion cell body and the extramacular damage shown by its axon, the retinal nerve fibers. We merged all of the GCIPL deviation map images from patients sharing the same clock-hour location of localized pRNFL defects, thus obtaining a GCIPL deviation frequency map of six different pRNFL defect locations revealing the areas of RGC death and corresponding pRNFL death. 

The topographic relationship shown in our GCIPL deviation frequency maps can be summarized as follows. The RGCs in the superior or inferior temporal macula project their axons in an arcuate pattern toward the superotemporal or inferotemporal portions of the ONH. ^25,26 Our results, correspondingly, reflected a similar structural relationship between RNFL damage and corresponding macular damage. In our GCIPL deviation frequency map, localized defects showed corresponding GCIPL defects in the same hemifield, without crossing the horizontal raphe, and the overall shape of the GCIPL defect had an arcuate pattern that seemed to be an extension of the RNFL defect. Although the six GCIPL deviation frequency maps showed different areas of the frequently damaged site, they all had the temporal macular area in common. These findings are probably associated with the pattern of RNFL loss in early glaucoma, which typically involves arcuate damage and the inferotemporal area, the site most vulnerable to glaucomatous damage. ^{1,9,18,27–29}  

Our results are consistent with those of Hood et al. ^17,18 with regard to the relationship of glaucomatous lesions in the macula and peripapillary regions. They reported that the probability of glaucomatous damage at the disc increased from the center of the temporal quadrant toward the superior and inferior poles. In our study, the number of consecutively included patients having pRNFL defects in the temporal quadrant (8 and 10 o'clock) was almost half the number of those with defects in the temporal parts of the superior and inferior quadrants (11 and 7 o'clock, respectively). Although our patients were not prospectively enrolled, such difference could have been associated with the difference in the susceptibility to glaucomatous damage between regions. Moreover, the Hood et al. ¹⁸ schematic model of glaucomatous macular damage represented the inferior macula as being more vulnerable than the superior macula. In the present study, the area of superior GCIPL defect was smaller than that of inferior GCIPL defect. The superior GCIPL defects, relative to the inferior ones, also showed considerable sparing of the cecocentral region. 

In the present study, distinctive interindividual variability in the GCIPL defect pattern was noted. Specifically, wide distributions of GCIPL defect area, location, and shape were observed within each group. Garway-Heath et al. ¹⁹ reported that ONH location had a significant effect on 53.8% of the corresponding visual field test points. Other than ONH position relative to the fovea, the disc area, axial length, spherical equivalent, disc shape, disc orientation, and disc tilt were also found to be associated with mapping the retinal locations (functional loss) to the ONH. ^18,30 Our results for interindividual variations in macular damage can be explained by these factors, but further studies may be warranted for the investigation of the possible contributing factors. 

As this study was based on patients showing concurrent GCIPL defect on a GCIPL deviation map, careful interpretation of our data is needed. A total of 30 patients were excluded for not showing concurrent GCIPL defects: 3 patients had pRNFL defect at 12 o'clock, 7 at 11 o'clock, 3 at 10 o'clock, 4 at 8 o'clock, 7 at 7 o'clock, and 6 at 6 o'clock. These patients could have had GCIPL defects outside the elliptical scan area of the SD-OCT, which might have affected the characteristics of the size and extent of the GCIPL defects according to the corresponding clock-hour locations of pRNFL defects. On the other hand, this might partly imply that initial glaucomatous damage is not presented in the macula at the same time as in the peripapillary region. Different results on the issue of the causal connection between death of the axon and of the RGC body have been reported depending on the experimental conditions. ^31–33 Some reported that RGC body death is the consequence of axonal damage by retrograde degeneration, ^34–36 while others explained that RGC apoptosis proceeds in advance of axonal damage due to deprivation of neurotrophic factors or excitotoxic injury. ^37,38 However, as this was a retrospective cross-sectional study, the sequence of damage could not be determined. Moreover, since the GCA and pRNFL algorithms cannot perfectly indicate the ganglion cell body and its axons, respectively, we might not have been able to provide clear answers to these questions. 

Several limitations of our study need to be considered. First, included patients were all Koreans. Since we used the software–built-in internal normative database of OCT, the detection rate of the deviation map might have been affected due to its ethnic composition. We minimized this effect, however, by confirming the location of localized pRNFL defects using both RNFLP and Cirrus OCT deviation maps. Second, the clock-hour location of pRNFL defect on the deviation map may be different from the actual location. Owing to intervariability in the location of the optic disc relative to that of the fovea, clock-hour sector determinations could vary if the standard was only RNFLP. Therefore, a clock-hour circle positioned over the 3.46-mm-diameter circle outline on the pRNFL deviation map was used to ensure the objectivity and accuracy of the location of pRNFL defect. Third, we assessed the GCIPL defects detected in more than 50% of patients in each clock-hour group. The characteristics of GCIPL defect areas selected by other cutoff points (e.g., 75%, 90%, 95%) also should have been assessed. However, because the maximum detection frequencies among the clock-hour groups all differed due to the wide GCIPL defect variation (Table 2), the cutoff point satisfying all groups was set at 50%. Lastly, because we selected consecutive patients with one single localized pRNFL defect in either hemifield, a larger number of patients with defects in the inferotemporal (n = 38) and superotemporal (n = 32) areas, which are more susceptible to glaucomatous damage, were included compared to other groups. Also, none showed an isolated pRNFL defect at 9 o'clock, the region less susceptible to glaucomatous damage, consequently limiting the implications of the topographic relationship between the two defects. All of this notwithstanding, our results included all commonly detected pRNFL defects. 

In conclusion, using SD-OCT–derived GCIPL and pRNFL deviation maps, we could generate the GCIPL deviation frequency map, which related the locations of localized pRNFL defects to the corresponding regions of macular ganglion cell loss. In thus demonstrating the regional correspondences between pRNFL and GCIPL defects, the map can provide supporting evidence for the prediction of glaucomatous structural damage from one to the other. Furthermore, it can establish a basis for investigation of the relationship between the two algorithms from SD-OCT. We hope that our results can aid in addressing the question of how the ganglion cell body and its axons are correlated with each other topographically and in understanding the damage occurring in the macular and extramacular region in glaucoma. 

Acknowledgments

The authors appreciate the statistical consultation provided by the Medical Research Collaborating Center at Seoul National University College of Medicine/Seoul National University Hospital. 

Disclosure: K.E. Kim, None; K.H. Park, None; B.W. Yoo, None; J.W. Jeoung, None; D.M. Kim, None; H.C. Kim, None