Difference in topographic morphology of optic nerve head and neuroretinal rim between normal tension glaucoma and central retinal artery occlusion

Although central retinal artery occlusion (CRAO) has its own defining pathomechanism and clinical characteristics, morphologic feature of the optic nerve head (ONH) during its later stage is not diagnostic, which makes it difficult to differentiate CRAO from other optic neuropathies. This cross-sectional study was performed to investigate the differences in the topographic morphology of the ONH in eyes with normal-tension glaucoma (NTG) and CRAO. Thirty-one eyes with NTG; 31 eyes with CRAO; and 31 healthy fellow eyes of the subjects with CRAO were included. ONH morphology was evaluated by measuring horizontal rim width (HRW), minimal rim width in the selected horizontal image (MRW), and lamina cribrosa curvature index (LCCI) in horizontal B-scan images obtained using enhanced depth-imaging optical coherence tomography. HRW was smaller and LCCI was larger in NTG eyes than in both CRAO and healthy fellow eyes (both P < 0.001), while both were comparable between CRAO and healthy fellow eyes. MRW differed significantly among the three groups, being smallest in NTG eyes followed by CRAO and healthy fellow eyes (P < 0.001). NTG and CRAO eyes with a similar degree of RNFL loss differed in ONH morphology, indicating that mechanisms of ONH damage differ between these two conditions.


Results
This study initially assessed 121 eyes of 121 patients with NTG and 90 with CRAO. Of these, 59 eyes with CRAO were excluded, 31 because they had incomplete type CRAO, five because they had neovascular glaucoma, 11 because they had a tilted or torted disc, and 12 due to poor visualization of the OCT image. In addition, 76 eyes with NTG were excluded, 47 because they had a tilted or torted disc, and 29 due to poor visualization of the OCT image. After matching for age, IOP at the time of the OCT scan, disc area, and global retinal nerve fiber layer (RNFL) thicknesses, 31 eyes with NTG and 31 with CRAO were included in the study, along with the 31 healthy contralateral eyes in the patients with CRAO. There was excellent interobserver agreement in measurements of horizontal rim width (HRW), minimum rim width in the selected horizontal image (MRW), and LC curvature index (LCCI), with intraclass correlation coefficients (ICCs) of 0.996 (95% confidence interval [CI] 0.994-0.998), 0.998 (95% CI 0.997-0.999) and 0.960 (95% CI 0.940-0.974), respectively. The mean follow-up period for OCT scans after CRAO occurrence was 3.0 ± 2.5 years. Table 1 compares the clinical characteristics of the NTG, CRAO, and healthy contralateral eyes of CRAO subjects. Global and sectoral RNFL thicknesses were larger in the healthy contralateral eyes than in both NTG and CRAO eyes (P < 0.001 each), but did not differ significantly between the NTG and CRAO eyes ( Table 1, Fig. 1). Table 2 and Fig. 2 show comparisons of ONH morphology in the three groups of eyes. Disc area and disc ovality did not differ significantly among these groups (P ≥ 0.137). HRW, however, was significantly smaller in NTG eyes than in both CRAO and healthy contralateral eyes (P < 0.001), but did not differ significantly between CRAO and healthy contralateral eyes. MRW differed significantly among all three groups, being smallest in NTG eyes, followed by CRAO eyes and healthy contralateral eyes (P < 0.001). Horizontal-to-minimum rim width ratio in the selected horizontal image (HMR) was smallest in NTG eyes followed by healthy contralateral eyes and CRAO eyes (P < 0.001). LCCI was significantly larger in NTG eyes than both CRAO and healthy contralateral eyes (P < 0.001), but did not differ significantly in the latter two groups. Overall, smaller HRW was associated with larger LCCI (P < 0.001, r 2 = 0.3274; Fig. 3) in NTG and CRAO eyes. However, subanalyses within each group found that these associations were not significant.
To eliminate the influence of large retinal vessels, which are more prominent at the nasal rim, measurements obtained in the temporal area were compared in the three groups (Table 3). The results did not differ substantially from those shown in Table 2. Figure 4 shows eyes with NTG and CRAO matched by age, IOP, optic disc area, and global RNFL thickness, as well as the healthy contralateral eye of the subject with CRAO. HRW and MRW were noticeably small and    www.nature.com/scientificreports/ LCCI noticeably large in the NTG eye. MRW was smaller in the CRAO eye than in its contralateral eye, whereas both HRW and LCCI were comparable in these two eyes.

Discussion
The present study analyzed features of the ONH in eyes with NTG and CRAO. Small HRW and MRW and large LCCI were characteristic of NTG eyes, whereas CRAO eyes had larger HRW and MRW and smaller LCCI than NTG eyes with similar amounts of RNFL loss. Comparisons of CRAO eyes and healthy contralateral eyes in the same subjects showed that MRW was decreased in CRAO eyes, whereas HRW and LCCI were comparable. Although these findings may be insufficient to differentiate between the two diseases, they may help to understand differences in the pathophysiology of NTG and CRAO. To our knowledge, this is the first study to compare ONH morphology in eyes with NTG and CRAO.
Despite their similar degree of RNFL loss, NTG and CRAO eyes showed significant differences in rim thickness, as determined by both HRW and MRW. Moreover, HRW did not differ between CRAO eyes and healthy contralateral eyes of the same subjects. This finding is consistent with results in rhesus monkey eyes, showing that optic disc cupping was similar in CRAO and normal eyes 17 . When the primary insult is outside the ONH, the ONH glioarchitecture is not disrupted. In response to axonal loss, hypertrophic astrocyte processes fill the space formerly occupied by axons, maintaining the general tissue architecture in non-glaucomatous descending/ascending optic neuropathies 18,19 . In glaucoma, however, disruption and loss of the glioarchitecture through progressive disorganization can cause rim tissue atrophy 20 . Even eyes at early stages of glaucoma showed slight disorganization of the previously ordered arrangement of glial columns 21,22 . Evaluation of experimental glaucoma in primate eyes with rim thinning has shown the destruction of glial columns, marked disarrangement of glial cells, and loss of prelaminar tissues 23,24 .
Although HRW was comparable, MRW was significantly smaller in CRAO eyes than in healthy contralateral eyes. This may be attributable to the site of the MRW measurement, which was anatomically close to the transition zone between the superficial nerve fiber layer (SNFL) and the prelaminar part that includes components from both areas 20 . The NRR has four zones in the anterior ONH: the SNFL, the transitional zone, the anterior prelaminar area, and the posterior prelaminar area 25,26 . The NRR contains two distinct spatial arrangements of astrocyte processes, one parallel and the other perpendicular, extending towards the course of axon bundles. In the RNFL-SNFL compartment containing parallel axons and glial processes there is no spacer between the upper and lower layers of axon bundles [27][28][29] , resulting in a decrease in rim width that is relatively proportional to the decrease in axons. The compact arrangement of bundles becomes less parallel and disperses when these  www.nature.com/scientificreports/ bundles reach the optic disc, bending at the transition zone 30 . Therefore, the tissue thickness in this area is less affected by the decrease in axons. The combination of a relatively small MRW and a relatively large HRW in CRAO eyes made their ratio (HMR) the most distinguishable parameter characterizing ONH morphology in CRAO. A large HMR indicates that the NRR was relatively well maintained despite the amount of axonal loss. Additional studies are required to determine whether a large HMR could serve as a specific indicator of ONH morphology in eyes with CRAO.
Morphologically, the ONH in eyes with NTG was characterized by a large LCCI. Glaucoma is caused by mechanical stresses on the ONH, with the LC regarded as the principal site of axonal injury of retinal ganglion cells 31,32 . Deformation of the LC is thought to induce damage in retinal ganglion cells by blocking axonal transport, thereby reducing the diffusion of nutrients from the laminar capillaries to the adjacent axons 33,34 , or by connective tissue remodeling 35 . LC deformation has been regarded as an important pathophysiologic manifestation of glaucoma, even when accompanied by low IOP 36,37 . The difference between LC curves of NTG and CRAO eyes with similar IOP provides additional evidence for the difference in their pathomechanisms. These findings also suggest that acute retinal ischemia does not induce morphologic changes in the LC.
Global and sectoral RNFL thicknesses were larger in the healthy contralateral eyes of subjects with CRAO than in both CRAO and NTG eyes, but did not differ significantly between the latter two groups. Reductions in inferotemporal and superotemporal RNFL thicknesses are diagnostic feature of glaucoma 38,39 . Although glaucomatous NRR loss can occur in a diffuse manner, it can also occur sequentially in sectors of the eye. Generally, NRR loss was found to begin in the inferotemporal disc region and to progress sequentially to the superotemporal, temporal horizontal, inferior nasal, and superior nasal sectors. Although this sequential progression may be applicable to early glaucoma 40 , the present study included patients with advanced glaucoma, as one of the factors used to match patients with NTG and CRAO was the amount of RNFL loss. The inclusion of advanced glaucoma patients likely masked the sectoral RNFL thickness distribution characteristic of glaucoma.
Overall, HRW was negatively associated with LCCI, suggesting that HRW decreases as the rim tissue turns off along the deformed LC. However, subanalyses within each group found that the associations between HRW and LCCI were not significant. Most NTG eyes had smaller HRW and larger LCCI, whereas CRAO and healthy eyes differed in HRW while having small LCCI. The relatively small sample size of this study could also have www.nature.com/scientificreports/ contributed to these inconclusive results. Future studies in larger numbers of subjects are needed to evaluate the relationship between HRW and LCCI. This study had several limitations. First, the sample size of each group was relatively small, primarily because the incidence of CRAO is lower than that of NTG. Second, eyes with tilted or torted optic discs were excluded, making the findings of this study inapplicable to eyes with these conditions. Third, HRW and MRW were measured manually using the horizontal disc scans, primarily because many of the patients included in this study were enrolled before the Bruch's membrane opening (BMO)-MRW protocol of Spectralis, which is based on the radial disc scans, became available. Horizontal disc scans were more useful than radial scans for evaluation of the LC curve, because the LC has a relatively regular configuration in the horizontal plane, having a flat or U-shaped appearance despite differences in regional steepness, allowing the measurement of LCCI 41,42 . However, using the horizontal images also means that our MRW measurement may not accurately represent the "minimum" rim width. Therefore, it should be referred to as minimum rim width in a selected horizontal image. Fourth, this study included eyes with NTG, preventing the generalization of these results to all eyes with glaucoma. Because IOP has been associated with LC morphology, the bias resulting from the influence of IOP could be ruled out by excluding eyes with high IOP. Future studies should therefore include eyes with glaucoma other than NTG.
In conclusion, the present study found that HRW was larger and LCCI was smaller in both CRAO and healthy contralateral eyes than in NTG eyes, but that these parameters did not differ significantly between CRAO and healthy contralateral eyes. The differences observed in ONH indices likely reflect differences in ONH morphology and in the pathogenesis of these two diseases.

Methods
Study subjects. This cross-sectional study included patients with NTG who were enrolled in the Investigat- NTG was defined as the presence of glaucomatous optic nerve damage (i.e., NRR thinning/notching, and an RNFL defect in the corresponding region), a corresponding glaucomatous VF defect, an open iridocorneal angle on gonioscopic examination, a maximum IOP ≤ 21 mmHg without glaucoma medications, no prior history of long-term use of steroid medications and no identifiable secondary cause of glaucoma. A glaucomatous VF change was defined as the fulfillment of two or more of the following criteria: (1) outside normal limits on the glaucoma hemifield test; (2) three abnormal points with a < 5% probability of being normal, including one abnormal point with a < 1% probability of being normal by pattern deviation, or (3) a pattern standard deviation < 5%. These VF defects were confirmed on two consecutive reliable tests, defined as tests with a fixation loss rate ≤ 20% and false-positive and false-negative error rates ≤ 25% each.
CRAO was diagnosed as the occurrence of classic clinical findings of sudden, painless vision loss, and funduscopic findings indicative of retinal hypoperfusion, as confirmed by FA 43 . CRAO was categorized as incomplete, subtotal, or total 44 . Incomplete CRAO is characterized by diminished VA, slight retinal edema with indefinite cherry-red spots, and mildly delayed retinal arterial perfusion on FA. Subtotal CRAO is identified as severe reduction in VA, distinct retinal edema with cherry-red spots, and severely delayed retinal arterial perfusion. Total CRAO is characterized by severe retinal ischemia and massive retinal edema in the macula, often accompanied by choroidal perfusion delay and no light perception 44 . The present study included patients with subtotal or total CRAO involving one eye, and excluded those with branch retinal artery occlusion. Eyes with maximum IOP > 21 mmHg, family history of glaucoma, or history of using glaucoma medication were also excluded to eliminate any effect of IOP on ONH morphology. Patients with CRAO and NTG were matched 1:1 by age, IOP at the time of OCT, disc area, and global RNFL thickness. The control group consisted of unaffected eyes in subjects with CRAO.
Patients were also excluded if they had a spherical equivalent of < − 8.0 D or > + 3.0 D, a cylinder correction of < − 3.0 D or > + 3.0 D, a history of intraocular surgery except for uneventful cataract surgery, or any retinal disease, such as diabetic retinopathy or retinoschisis or neurological disease such as a pituitary tumor. Eyes were also excluded if they had optic disc tilt, defined as a tilt ratio (ovality index) of the longest to the shortest diameter > 1.3 45,46 ; or torsion, defined as a torsion angle, or deviation of the long axis of the optic disc from the vertical meridian, of > 15°4 6,47 . Eyes were excluded when good-quality images (i.e., quality score > 15) could not be obtained for more than five sections of enhanced depth imaging (EDI) SD-OCT disc scans. When the quality  48 . EDI has been shown to yield images with a stronger signal and better image contrast in the deep ONH tissue than conventional imaging techniques 49 . Patients with NTG underwent OCT scans of the ONH prior to the initiation of ocular hypotensive treatment to eliminate any potential effects of IOP on LC morphology 11,50,51 . Patients with CRAO underwent OCT scans at least 6 months after the onset of CRAO, after reduction and stabilization of the acute swelling of the ONH. Patients were imaged through undilated pupils using a rectangle subtending 10 degrees × 15 degrees of the optic disc. This rectangle was scanned with approximately 75 B-scan section images that were separated by 30-34 µm (the scan line distance was determined automatically by the machine). Approximately 42 SD-OCT frames were averaged for each section. This protocol provided the best trade-off between image quality and patient cooperation 49 . Potential magnification errors were avoided by entering the corneal curvature of each eye into the Spectralis OCT system before scanning.
ONH parameters, including LCCI and NRR width on horizontal B-scan images, were measured by two experienced observers (J.A.K. and E.J.L.), who were masked to subjects' clinical information. Two observers used the same set of images for the measurement. The mean of the measurements made by the two observers was used for analysis.

Measurement of neuroretinal rim widths.
HRW and MRW were measured based on BMO, as had been detailed by Reis et al. (Fig. 5) 52 . HRW was defined as the distance between the projection of BMO to the BMO reference plane and the internal limiting membrane (ILM), along the BMO reference plane, whereas MRW was defined as the minimum distance between the BMO and the ILM. HMR was determined by dividing HRW by MRW (HRW/MRW). If overlying large vessels prevented interpretation of the structures, an adjacent image was used; thus, the thicknesses of these vessels were not included in the measurements. HRW and MRW were measured at three locations (central, and superior and inferior mid-periphery) equidistant across the vertical optic disc diameter using the built-in manual caliper tool in Heidelberg Eye Explorer. The average of three scan values separated by two scan intervals was calculated for each location.

Measurement of lamina cribrosa curvature index.
To quantify the posterior LC curve on the SD-OCT B-scan images, we defined the LCCI as the inflection of a curve representing a section of the LC. LCCI has been recognized as a robust parameter representing the glaucomatous LC deformation 41,[53][54][55] and was shown to predict progressive RNFL thinning 42,56 .
The method used to calculate LCCI has been described previously 42,53 . In brief, the width (W) of the BMO was measured on each B-scan, followed by measurement of the LC curve depth (LCCD). The BMO width was defined as the width of the line connecting the temporal and nasal termination points. Lines were drawn from each BM termination point perpendicular to the BMO reference line, until they met the anterior LC surface. The line connecting the two points on the anterior LC surface was the reference line for measuring the LCCD. The LCCD was determined as the maximum depth from this reference line to the anterior surface (Fig. 5). The LCCI was then calculated as (LCCD/W) × 100. Since the curvature was thereby normalized according to LC width, LCCI represents the posterior curvature of the anterior LC surface independent of the actual size of the ONH. Only the LC within the BMO was considered because the LC was often not clearly visible outside of the BMO. LCCI was measured at the three locations on the same B-scans used to measure HRW and MRW, with the built-in manual caliper tool in Heidelberg Eye Explorer.  Data analysis. Except where stated otherwise, data are presented as mean ± standard deviation. The interobserver agreements for measuring the HRW, MRW, and LCCI were assessed by calculation of ICCs and 95% confidence intervals (CIs). Comparisons between three groups were performed using ANOVA or Kruskal-Wallis test depending on the assumption of normality using Shapiro-Wilk test. Post-hoc analysis of ANOVA was performed using the Tukey test. Comparisons between two groups were analyzed by t-tests, Mann-Whitney U-tests, paired t-tests, or Wilcoxon's signed-rank tests, as appropriate. All statistical analyses were performed using the Statistical Package for the Social Sciences

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.