Effect of Focal Lamina Cribrosa Defect on Disc Hemorrhage Area in Glaucoma

Abstract

Purpose: The purpose of this study was to evaluate the association between focal lamina cribrosa defect (FLCD) and the topographic characteristics of disc hemorrhage (DH), including area and location.

Methods: We enrolled a total of 98 primary open–angle glaucoma eyes with DH (98 subjects). In vivo lamina cribrosa (LC) images were obtained by swept-source optical coherence tomography (SS-OCT) immediately following the detection of DH. Two masked graders identified FLCD (laminar holes or disinsertions >100 μm in diameter and >30 μm in depth), defined by a customized protocol using en face images and 12 radial-orientation raster scans of SS-OCT. En face image/stereo-disc photography overlay images were evaluated to determine the spatial relationship between the respective FLCD and DH locations. A method of comparing the disc area and DH area pixel numbers was used to estimate the DH area.

Results: Sixty-eight of 98 eyes with DH (68.4%) had at least one FLCD. Thirty-eight of those 68 eyes with DH and at least one FLCD (55.9%) had a DH corresponding to the FLCD location (within one-half clock-hour distance from the midline). The FLCD-correspondent DHs (39 DHs) showed significantly larger areas (0.092 ± 0.030 mm²; P < 0.001) and more proximally located proximal ends (P < 0.028) than the noncorresponding ones (33 DHs; 0.065 ± 0.024 mm² of area).

Conclusions: The DHs that correspond to FLCD location tend to have larger areas and to be more proximally located than those without correspondence. This suggests that FLCD might affect the topographic characteristics of DH.

Disc hemorrhage (DH) is one of the well-known risk factors for the development and progression of glaucoma.^1–5 Typically, DH related to glaucoma is located in the peripapillary retinal nerve fiber layer (RNFL) or the prelaminar region of the optic nerve head (ONH). A previous study suggested that most DHs are found in the vicinity of the border between localized RNFL defects and relatively healthy-looking RNFL.⁶ Conversely, others have reported that DHs are associated with abnormalities of the lamina cribrosa (LC).^7,8 Disc hemorrhages occur frequently in eyes with focal LC defects (FLCDs)^9–11 and in fact show a considerable spatial relationship with them.^10–12 These findings signify that structural changes at the LC level lead to DH.¹³ However, there remains controversy as to the pathogenesis of DH. 

Our group recently reported that, in glaucoma patients, the topographic characteristics of DH, including area and extent, may be explained in part by the functions of mechanical properties such as intraocular pressure (IOP).¹⁴ The larger DH area in glaucoma eyes with normal baseline IOP was explained in two ways. First, eyes having a relatively lower IOP can produce a larger area of DH owing to the lesser tamponade effect. Second, normal baseline IOP glaucoma may have a relatively more vulnerable blood vessel than high baseline IOP glaucoma, even at a lower IOP. As an advance on the previous findings, we deduced that the topographic characteristics of DH also can differ in cases of various DH origins, such as FLCD, the border of RNFL defect, or both. 

This study investigated the effect of FLCD on the topographic characteristics of DH, focusing on area and location in glaucomatous eyes. We presumed that knowledge of the effect of FLCD on the topographic characteristics of DH could help us to clarify the developmental mechanism of DH and eventually the pathogenesis of glaucoma. 

Methods

This study followed the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Seoul National University Hospital. We performed a retrospective review of primary open–angle glaucoma (POAG) patients' electronic medical records, which had been compiled by the author (K.H.P.) at the Glaucoma Service of Seoul National University Hospital between January 2012 and May 2015. 

Subjects

All study participants underwent a comprehensive glaucoma evaluation, including a central 30-2 threshold test of the Humphrey visual field (HVF) (HFA II; Humphrey Instruments, Inc., San Leandro, CA, USA), RNFL and ONH imaging by Cirrus spectral-domain coherence tomography (SD-OCT; Carl Zeiss Meditec, Dublin, CA, USA), digital color stereo-disc photography (SDP), and digital RNFL photography. Further, in vivo LC images were obtained by DRI OCT-1 Atlantis 3D swept-source optical coherence tomography (SS-OCT; Topcon Medical Systems, Oakland, NJ, USA) from study participants immediately following the detection of DH by SDP. 

For inclusion, subjects were deemed to have POAG if they met the following criteria¹⁴: presence of typical glaucomatous optic neuropathy with compatible visual field loss and normal anterior chamber angles. Glaucomatous visual field loss was defined as a visual field with at least three adjacent test points having a deviation of ≥5 decibels (dB) and one test point having a deviation of >10 dB; at least two adjacent test points with a deviation ≥10 dB; at least three adjacent test points with a deviation ≥5 dB abutting the nasal horizontal meridian; or a mean visual field defect of >2 dB.¹⁴ Subjects were excluded from analyses if any of the following was true: the presence of a secondary cause of glaucomatous optic neuropathy; a history of intraocular surgery (except cataract operation) or retinal laser photocoagulation; the presence of high (<−6.0 diopters [D]) myopia; and visual field mean deviation worse than −12 dB.¹⁴ 

Discrimination of DH

Two graders (K.H.P. and Y.K.K.), masked to all other information, performed DH evaluation independently. Discrepancies between the two observers' findings were resolved by consensus. A DH was considered to be unrelated to glaucoma if the disc was swollen or otherwise obviously abnormal due to nonglaucomatous optic neuropathy; there were multiple nearby retinal hemorrhages suggestive of diabetic retinopathy or retinal vascular abnormality; or there was acute posterior vitreous detachment that could have caused DH.¹⁴ In cases of recurrent or persistent DH during interval testing, DH detected for the first time was selected for the study. For cases with multiple DHs, simultaneously affecting different sites, average values of all DHs were used in analysis of the study. In cases where a subject had bilateral DH and met all of the inclusion criteria, one eye was selected randomly for the study. 

Measurement of Octant Location and Proximal-End Location of DH

The octant locations (superotemporal, superonasal, nasosuperior, nasotemporal, inferonasal, inferotemporal, temporoinferior, and temporosuperior) of DH were assessed based on the axis connecting the fovea and Bruch's membrane opening (BMO) center (fovea-BMO center axis). The BMOs are automatically defined by the Cirrus internal-specific software.¹⁵ To determine the fovea center and BMO center, superimposed images, constituting an RNFL deviation map and ganglion cell–inner plexiform layer deviation map overlaid onto RNFL photography and aligned by Photoshop software (Version 10.1; Adobe, San Jose, CA, USA) based on vascular landmarks, were used (Fig. 1). Further, the proximal-end locations of DHs were classified,¹⁶ based on the disc margins and the cup margins indicated on the RNFL deviation map, into three types (more proximal to cup margin, disc rim, or peripapillary area; Fig. 1D). 

Figure 1

 

Example of topographical analysis (measurement of octant location and proximal-end location) of DH performed on a study subject. (A) Eye with splinter-shaped DH detected by RNFL photography. (B, C) The superimposed images, constituting (a) an ganglion cell-inner plexiform layer deviation map and (b) RNFL deviation map overlaid onto RNFL photography and aligned by Photoshop software (Version 10.1; Adobe) based on vascular landmarks, were used to determine the fovea center and BMO center. (D) The octant location of DH (ST, superotemporal; SN, superonasal; NS, nasosuperior; NT, nasotemporal; IN, inferonasal; IT, inferotemporal; TI, temporoinferior; TS, temporosuperior) was assessed based on the axis connecting the fovea and BMO center (fovea-BMO center axis; blue dashed line). Further, the proximal-end location (black asterisk) of DHs (red arrow) was classified, based on the BMOs (blue arrow; defined as disc margins in Cirrus-HD OCT algorithm) and the cup margins (yellow arrow) indicated on the RNFL deviation map, into three types (more proximal to cup margin, disc rim or peripapillary area).

Estimation of DH Area

The details of the methodology for the estimation of DH area have been reported previously.¹⁴ In brief, a method of comparing pixel numbers of the disc area and DH area in SD-OCT RNFL deviation map/SDP overlay images, with ImageJ software (version 1.45s; Wayne Rasband, National Institutes of Health, Bethesda, MD, USA), was used to estimate the DH area (Fig. 2), which was calculated by the following formula:  

 

 

Figure 2

 

Example of topographical analysis (measurement of area and length) of DH performed on a study subject. (A) Eye with splinter-shaped DH detected by digital color SDP. (B) An RNFL deviation map derived from RNFL and optic nerve head imaging (Optic Disc Cube 200 × 200) with Cirrus SD-OCT. (C) An RNFL deviation map/SDP overlay image with black line delineating optic disc boundary as defined by SD-OCT. (D) Plotting of optic disc boundary (blue arrow) and DH boundary (red arrow) on RNFL deviation map/SDP overlay image for counting of pixel numbers by ImageJ software (National Institutes of Health). (E, F) Measuring of the LMRE of DH (black arrow) by ImageJ software.

The corrected disc area was calculated in consideration of the magnification factors related to the SD-OCT camera and the eye by substituting the subject's axial length (AL).^17–20 Thus, the formula for disc area correction was as follows:  

 

 

Estimation of Radial Extent of DH

The straight distances from the innermost point to the furthermost point of the DH were defined as the length of maximum radial extent (LMRE) of the DH (Figs. 2E, 2F). The straight distances from the optic disc center to the RNFL calculation circle (provided by SD-OCT RNFL deviation map; 3.46 mm of diameter) were measured and defined as the radius of the RNFL calculation circle. While compensating for the magnification factor, the corrected LMRE of the DH was calculated by the following formula¹⁴:  

 

 

Swept-Source OCT Imaging

A 3D raster scan protocol consisting of 256 × 256 A-scans was conducted in 0.8 seconds over a 3 × 3-mm area centered on the ONH. Additionally, 12 radial orientation raster scans (6-mm scan length, centered on ONH) were obtained for each eye. A total of 32 A-scans were averaged for each of the 12 radial lines. Then, for the 3D data set, serial en face images were reconstructed. These images were then used to observe the laminar structures. 

Assessment of LC Defects

An FLCD was defined as an anterior laminar surface irregularity violating the normal smooth curvilinear U- or W-shaped contour.²¹ To avoid false positives, defects (laminar holes or disinsertions) needed to be >100 μm in diameter and >30 μm in depth.²¹ The obtained SS-OCT image sets (en face images and 12 radial orientation raster scans) were independently reviewed to determine the existence of candidate for LC focal defects by two graders (Y.K.K. and K.H.P.) masked to all other information including the presence or absence of DH. Then, by comparing the en face images to the disc photographs, the graders confirmed that the candidate for LC defects did not correspond to hyporeflectivity caused by vascular shadowing. 

Assessment of Spatial Relationship

In an assessment of the spatial relationship between FLCD and DH, superimposed images, constituting en face images overlaid onto SDP and aligned by Photoshop software based on vascular landmarks, were used. Disc hemorrhages were deemed to correspond to the FLCD location if they were located within the one-half clock-hour from the midline of the FLCD location in the en face image/SDP overlay image. 

Data Analysis

The DH area and the LMRE of the DH were compared between the correspondent and noncorrespondent DHs. The continuous data were compared between groups using the Student's t-test. The categorical data were compared between groups using χ² tests. Univariate and multivariate linear regression analyses with forward stepwise selection were used to evaluate the factors associated with DH area. The κ coefficients for the graders' assessment of the presence or absence, the number of FLCDs, and the measurement of DH area were calculated as a measure of the reliability of interobserver agreement. The κ coefficient adjusts the observed proportional agreement to take into account agreement that would be expected by chance, with κ = 1 indicating perfect agreement and κ = 0 indicating no agreement.²² All statistical analyses were performed using SPSS software version 19.0.0 (SPSS, Inc., Chicago, IL, USA). P < 0.05 was considered significant. 

Results

One hundred nineteen POAG eyes with DH, representing 119 subjects, were initially included in the study. Of these, 21 subjects were excluded: (1) in 17 subjects, there was poor visibility of the anterior LC surface (<70% of anterior LC visible on en face image of SS-OCT); (2) 2 subjects had poor-quality SDP or RNFL photography; and (3) for 2 subjects, the retinal vessel on the RNFL deviation map could not be matched to the SDP. The remaining 98 eyes of 98 subjects were examined in the ensuing analysis (Fig. 3). The subjects' demographic characteristics are summarized in Table 1. As indicated, between the eyes with at least one FLCD and those without FLCD, there were no significant differences in age, sex, systemic comorbidity, intake of acetylsalicylic acid, IOP, HVF mean deviation, refractive error, central corneal thickness, or AL. 

Figure 3

 

Flow diagram of 98 DH eyes of POAG patients in this study. Ninety-eight eyes with DH were classified according to the presence or absence of FLCD, the number (single or multiple) of DHs, or the presence or absence of DH corresponding to the location of FLCD.

Table 1

 

Clinical Characteristics of Glaucomatous DH Patients With at Least One Focal LC Defect and Without Focal LC Defect

 

Interobserver Agreements in Discrimination of LC Defects and Measurement of DH Area

There was an excellent agreement between two masked graders as to whether an eye had an FLCD or not (κ = 0.84; 95% confidence interval [CI], 0.75–1.00; P < 0.001). Also, there was a good overall interobserver agreement regarding the number of detected LC defects (κ = 0.70; 95% CI, 0.65–0.79; P = 0.01). The interobserver agreement was excellent on the measured DH area as well (κ = 0.87; 95% CI, 0.77–1.00; P < 0.001). 

Number and Location of LC Defects

In the total of 98 eyes with DH, the proportion with at least one FLCD was 69.4% (68 eyes). Examples of DH eyes with LC defect are shown in Figures 4 and 5, and examples without an LC defect are shown in Figure 6. Sixty-three eyes with DH had one FLCD, four had two FLCDs, and one had three LC defects (Fig. 3). There were 52 FLCDs in the inferotemporal sector, 21 defects in the superotemporal sector, and 1 defect in the superonasal sector. 

Figure 4

 

Representative LC SS-OCT imaging performed on one study subject. (A–C) A DH that had more proximally located proximal-end location to cup margin, as detected in inferotemporal area of the optic disc by color SDP (outline of DH is indicated by white line in [B]). (D) En face images of the optic disc at level of LC as reconstructed from 3D SS-OCT data set (outline of FLCD is indicated by white line in [E]). (F) Radial scan images corresponding to DH on optic disc from the SS-OCT (scan line: long white arrow in [E]; outline of FLCD is indicated by white line in [G]). (H, I) En face image/stereo-disc photograph overlay images determining spatial relationship between respective FLCD and DH locations. (J) Presence of FLCD-correspondent DH (yellow dashed circle).

Figure 5

 

Representative LC SS-OCT imaging performed on one study subject. (A–C) A DH that showed that its proximal-end location was located on the disc rim, as detected in the inferotemporal area of the optic disc by color SDP (outline of DH is indicated by white line in [B]). (D) En face images of the optic disc at the level of LC as reconstructed from the 3D SS-OCT data set (outline of FLCD is indicated by white line in [E]). (F) Radial scan images corresponding to a DH on the optic disc from SS-OCT (scan line: long white arrow in [E]; outline of FLCD is indicated by yellow line in [G]). (H, I) En face image/stereo-disc photograph overlay images determining the spatial relationship between respective FLCD and DH locations. (J) Presence of the FLCD-correspondent DH (yellow dashed circle).

Figure 6

 

Representative cases of LC SS-OCT imaging performed on two study subjects. (A, B) The DHs, which had a proximal-end located on the disc rim, as detected in the inferotemporal area of the optic disc by color SDP (outlines of DHs are indicated by white lines in [C] and [D]). (E, F) En face images of the optic disc at the level of LC as reconstructed from the 3D SS-OCT data set. (G, H) Radial scan images corresponding to the DH on the optic disc from SS-OCT (scan lines: long white arrows in [E] and [F]). The absence of FLCD corresponded with the DH in both cases.

Spatial Relationship of DH and LC Defects

Among the 72 DHs from 68 eyes with at least one FLCD, the DH corresponded to the location of the FLCD (within one-half clock-hour distance from the midline) in 39 DHs (54.2%) from 38 eyes (55.9%). Meanwhile, the DH did not correspond to the location of the FLCD in 33 DHs (45.9 %) from 30 eyes (44.1%) (Fig. 3). Examples of eyes with FLCD-correspondent DHs are shown in Figures 4 and 5. 

Disc Hemorrhage Area and Length

The DH area was larger in eyes with, rather than without, at least one FLCD (0.080 ± 0.030 mm² in 68 eyes versus 0.067 ± 0.027 mm² in 30 eyes; P = 0.045; Table 2). In eyes with at least one FLCD, the FLCD-correspondent DHs showed significantly larger areas (0.092 ± 0.030 mm² in 39 DHs of 38 eyes) than those not corresponding to FLCD location (0.065 ± 0.024 mm² in 33 DHs of 30 eyes; P < 0.001; Table 3; Fig. 7). The LMRE of the DH was significantly greater in the FLCD-correspondent DHs (0.513 ± 0.124 mm in 39 DHs of 38 eyes) than in the others (0.365 ± 0.117 mm in 33 DHs of 30 eyes; P < 0.001; r₁ versus r₂ in Fig. 7). 

Table 2

 

Comparison of DH Area in DH Eyes With at Least One Focal LC Defect and Without Focal LC Defect

 

Table 3

 

Comparison of Area and Radial Extent of DH Corresponding to the Location of Focal LC Defect and in DH Not Corresponding to the Location of Focal LC Defect

 

Figure 7

 

A DH diagram focusing on area, LMRE, and length of circumferential extent. DH₁ presents the measurements of the DHs (39 DHs of 38 eyes) corresponding to FLCD: (1) DH area = 0.092 mm², (2) LMRE [r₁] = 0.513 mm, and (3) length of circumferential extent [c₁] = 0.208 mm. DH₂ presents the measurements of DHs (33 DHs of 30 eyes) not corresponding to FLCD: (1) DH area = 0.065 mm², (2) LMRE [r₂] = 0.365 mm, and (3) length of circumferential extent [c₂] = 0.177 mm. R, overall diameter of optic disc in 98 eyes (1.678 mm). P (DH area) < 0.001; P (r₁ versus r₂) < 0.001; P (c₁ versus c₂) = 0.858.

Octant Location and Proximal-End Location of DHs

The octant location of the DH was not significantly different between the FLCD-correspondent DHs and those not corresponding to FLCD location (P = 0.896). However, in terms of the proximal-end location, the DH presented relatively near to the disc center in the FLCD-correspondent DHs relative to those not corresponding to FLCD location (P = 0.028; Table 4). 

Table 4

 

Comparison of Octant Location and Proximal End Location of DH Corresponding to the Location of Focal LC Defect and in DH Not Corresponding to the Location of Focal LC Defect

 

Factors Associated With DH Area

Univariate and stepwise multivariate linear regression analyses were carried out to identify the variables associated with DH area. The presence of FLCD was significantly associated with DH area (standardized β value, 0.41; P < 0.001; Table 5). 

Table 5

 

Univariate and Multivariate Linear Regression Analysis with Area of DH as a Dependent Variable and Other Systemic and Ocular Factors as Independent Variables

 

Discussion

This study demonstrated that, in glaucomatous eyes with DH, FLCD can be detected in vivo using SS-OCT and with good interobserver agreement. In the present study, 68 of 98 eyes with DHs (69%) had at least one FLCD detected. Further, in DH eyes with FLCD, there was good spatial agreement between the DH and FLCD locations (38 of the 68 eyes; 56%). Although the etiology of DH is unclear, both mechanical^13,23 and vascular theories^24,25 having been hypothesized, these results suggest that a localized structural LC defect might be a factor predisposing to the disruption of the microvascular structure passing through the damaged sector. 

Focal LC defects occur mostly on the far periphery of the LC at or near the laminar insertion point,^21,26 and similarly, DH occurs preferentially at the disc margin.^1,27 In the same context, several studies have demonstrated that, in glaucomatous eyes, FLCD is spatially correlated not only with neuroretinal rim loss,²⁶ visual field defects,²¹ and RNFL defects,²⁸ but also with DH.^10,12 Further, DH develops commonly in superior and inferior areas of the LC that contain large pores with less connective tissue.²⁹ All of these findings suggest that DH correlates with IOP-related mechanical properties. During shearing, stress on the nerves and capillaries is applied at the level of the LC pores, and the stretching of the anterior capillaries consequent on posterior bowing of the LC causes DH.^13,29 A potentially important finding of the present study was that DHs corresponding to FLCD location have relatively larger areas than do others (Fig. 7). Indeed, this result supports the hypothesis that localized structural defects of the LC are the active sites of microvascular network breakage.¹⁰ That is to say, relatively larger DHs may occur in these circumstances, due to the ongoing mechanical disruption of the capillaries secondary to stretching or degenerative LC change. This result is consistent with the recent finding from a longitudinally designed study that the peripheral LC exhibited a recent alteration in eyes with DH and that the alteration was spatially correlated with DH location.¹⁰ Certainly, it is not possible to establish whether the presence of FLCDs is a predisposing factor for a relatively larger DH area; however, the topographic characteristics of DH, including area, have been reported to significantly depend on the IOP.¹⁴ Likewise, depending on the origin of DH, be it the rupture of the vascular structure inside the periphery of the LC, degenerative changes attendant on RNFL defect, or both, the DH area might be affected as well. 

A larger area of DH in cases of spatial correspondence between DH and FLCD can be explained in two ways. First, the peripheral LC reportedly is a region wherein there is relatively high mechanical strain with disruption of collagen and elastin.^30,31 This can cause significant rupture of the capillaries inside the lamina beams, resulting in a relatively larger area of DH. Second, the absence of overlying neural tissue at the area of FLCD may affect the size of DH due to less tamponade effect. That is to say that more bleeding could occur before stopping, thus resulting again in a large area of DH. Contrastingly, a higher IOP leads to a smaller area of DH.¹³ 

Another interesting finding was that, in terms of the proximal-end location, DHs corresponding to FLCD location presented relatively nearer to the disc center than the others did. This result supports our hypothesis that FLCDs are the sites of microvascular network breakage and are associated with emergence of DHs. Indeed, of 39 DHs corresponding to an FLCD location, only 2 had a proximal-end location in the peripapillary area, and more than half had a proximal-end location close to the cup margin. 

This study's finding that eyes with LC defect tend to have a larger area of DH has two clinical significances. First, in eyes with relatively large DH, the clinician has to consider the possibility of the presence of underlying LC defect. Given the wide acceptance of DH as one of the most important risk factors for development and progression of glaucoma,^2,32,33 as well as the recent studies reporting a significant association between the LC defect and glaucoma,^26,28,34 the identification of the DH area might be meaningful in terms of glaucoma management. That is, in a glaucomatous eye with a large DH, closer observation and stricter IOP control might be needed, because having two factors (DH and LC defect) could tend to accelerate glaucoma progression. The second clinical significance of this study's findings is that the FLCD-correspondent DHs were larger in area and longer in length than the noncorrespondent ones. This suggests the possibility that previous studies' reportedly higher DH prevalence in eyes with FLCD^9–11 was partially affected by these topographic characteristics of DH, which make DHs more easily detectable. 

Several points need to be considered when interpreting the results of the current study. First, DHs are transient phenomena, usually lasting only 6–12 weeks.^14,35 Although we conducted the examinations (SDP) in regular 3-month intervals and obtained SS-OCT LC images immediately following DH detection, theoretically it is possible that the timing of DH development could have varied, from minimally just 1 day to maximally 3 months before SDP was taken. Therefore, depending on the time at which the SDP were obtained, the values of DH area could have differed, and some DHs might even have been missed. However, in this study, (1) there was a significant difference in DH area between the two groups (a 1.42-fold larger area in DHs corresponding to FLCD location than in noncorrespondent DH), and (2) when considering that there are differences not only in area but also in length and proximal-end location, it can be a critical evidence that supports our hypothesis that FLCD affects the topographic characteristics of DH. Second, we used manual alignment of SDPs and en face SS-OCT images to measure the DH area and to declare the spatial correspondence between the DH and FLCD locations. Using the retinal vessels as alignment reference structures can be somewhat subjective and imprecise. Third, small LC defects might have been missed. To explain, in the current study, the size criteria of FLCDs (to be greater than 100 μm in diameter and 30 μm in depth) were chosen to allow comparison with previously published results.^21,28 However, a histologic study³⁶ suggests that LC defects can be much smaller than those defined by the criteria used in the present study. Fourth, 14.8% (17 eyes) of eyes with DH were excluded from the study due to poor visibility of the anterior LC surface. Although the quality of SS-OCT imaging has been improved by means of an averaging mechanism, this modality was inadequate for consistent delineation of the LC in all of the subjects. Fifth, in 77 (78.6%) of 98 eyes, the part of the nasal half of the LC was poorly visualized due to large retinal vessels. Even though only two DHs were found in the nasal half of the optic disc, the possibility of FLCD involving the nasal half should be considered when interpreting the results of the current study. Sixth, the disc margin reference might differ between OCT and SDP. Although we endeavored to standardize the reference as the line connecting the BMOs by using superimposed images constituting an RNFL deviation map overlaid onto SDP, there might still have been a discrepancy in the disc margin reference between OCT and SDP. Last, with regard to our method for determining the BMO center, segmentation error (e.g., erroneous detection of the BMO) theoretically could have been incurred, although there was no such case confirmed. 

In conclusion, we demonstrated that DH corresponding to FLCD location tends to have a larger area, a longer length, and a more proximally located proximal-end than noncorrespondent DH. This signifies that FLCD, as the site vulnerable to microvascular disruption, might have an influence on the topographic characteristics of DH. Further studies to identify the cause-and-effect relationships between FLCDs and the topographic characteristics of DH are needed. 

Acknowledgments

The authors thank Hee Chan Kim (Department of Biomedical Engineering, College of Medicine and Institute of Medical and Biological Engineering, Medical Research Center, Seoul National University, Seoul, Korea) and Byeong Wook Yoo (Interdisciplinary Program, Bioengineering Major, Graduate School, Seoul National University, Seoul, Korea). 

Disclosure: Y.K. Kim, None; J.W. Jeoung, None; K.H. Park, None