Transverse Separation of the Outer Retinal Layer at the Peripapillary in Glaucomatous Myopes

Glaucoma specialists often overlook the outer retinal changes because the glaucomatous optic neuropathy typically involves retinal nerve fiber layer (RNFL). By detailed inspection of the outer retina in myopic eyes, we observed a separation of the inner nuclear layer (INL) from the outer nuclear layer (ONL) at the peripapillary sclera (pp-sclera). Therefore, we conducted a retrospective observation of 108 eyes of 108 Korean subjects with myopia assessed by swept-source optical coherence tomography (SSOCT) and divided into normal and glaucomatous eyes. Mean subject age, refractive error and axial length difference between 2 groups were insignificant, respectively. To quantify the ONL-INL separation, straight-line distance from ONL endpoint to INL endpoint was measured at the center of the optic disc by SSOCT horizontal scan. The glaucomatous group had significantly large ONL-INL separation than the non-glaucomatous group (p = 0.027) but had no significant difference in INL – Anterior scleral canal opening (ASCO) separation. The width of ONL-INL separation were associated with β-peripapillary atrophy (β-PPA), degree of horizontal tilt of the optic disc and worse glaucomatous RNFL defect by Pearson’s correlation analysis (all p < 0.001, respectively). In conclusion, we demonstrate transverse separation of INL from ONL at the peripapillary region, which was significantly associated with glaucomatous optic nerve damage. These observations may be of interest to elucidate the role of PPA in glaucoma pathogenesis and a clinical index to take notice for myopic subjects.

The optic nerve head (ONH) is a major bottleneck of the retinal ganglion cell axons. Since it is densely packed and simultaneously has to pass through between the lamina cribrosa, it is a major weak point of the visual pathway. Therefore, proper evaluation of the ONH is of high importance for the physiology and pathophysiology of the optic nerve.
Histologic investigations reveal that the ONH is a three-layered hole consisting of the Bruch's membrane (BM), peripapillary choroid and peripapillary sclera [1][2][3] . Emmetropic eyes have the three layers gathered straight and well aligned with one another, seen as the peripapillary ring in the conventional ophthalmoscopy 4 . However, the axially myopic eyes shows markedly elongated and thinned peripapillary scleral flange which distorts its even alignment 5 . Consequently, the multilayered peripapillary ring is slanted towards the temporal, resulting in the separation and exposure of each layers. The amount of distortion in the peripapillary are individually varied which requires specific indicators to assess the peripapillary alteration properly.
The elongated peripapillary region has been defined as two separate regions into the beta zone peripapillary atrophy (β-PPA) and the gamma zone PPA (γ-PPA) with respect to the BM 6,7 . Clinical observational studies regarding the PPA persistently suggest that the β-PPA was associated with glaucoma, while γ-PPA was associated with axial myopia [8][9][10] . There has been efforts to explain glaucoma development with increasing β-PPA but these explanations seems inadequate [11][12][13] .
During routine clinical practice, glaucoma specialists often overlook the outer retinal changes because the glaucomatous optic neuropathy typically involves retinal nerve fiber layer (RNFL) and retinal ganglion cell layer (GCL) thinning in the inner retinal layer 14,15 . Although detecting early phase of circumpapillary RNFL deterioration is the gold standard in glaucoma diagnosis and management, assessing the integrity of the retina as a whole unit should not be overlooked as well. By detailed inspection of the outer retina in myopic eyes, we observed a separation of the inner nuclear layer (INL) from the outer nuclear layer (ONL) at the parapapillary region. This separation observed in the PPA region of the myopes seemed to be related with glaucomatous damage and enlarged with accompanying enlargement of β-PPA. On the basis of this observation, we hypothesized that separation of the two retinal nuclear layers may interrupt the integrity of the peripapillary retina, leading to further optic nerve damage. We retrospectively collected clinical data from the glaucomatous myopic subjects who had horizontal scan of the optic nerve head with SSOCT and classified into two groups with respect to the separation width of INL from ONL. The purpose of the present study was to describe the outer retinal separation in the peripapillary of glaucomatous myopes and to investigate factors associated with such changes.

Materials and Methods
This investigation was a retrospective observational study of 108 subjects who visited the glaucoma clinic of Seoul Saint Mary's Hospital between September 2016 and November 2017. Informed consent for study participation was obtained. The study was approved by the Seoul St. Mary's Hospital Institutional Review Board. It followed the tenets of the Declaration of Helsinki.
To be included in the present study, subjects were required to have myopia with axial length longer than 24.0 mm and to have an apparent temporal PPA on horizontal OCT scan image with a width of 200 µm or more measured by the built-in caliper tool of the SSOCT. NTG was defined as having glaucomatous optic neuropathy, such as rim thinning, notching, RNFL defect, glaucomatous visual field defect, an open iridocorneal angle and by the absence of a history of elevated intraocular pressure (IOP) >21 mm Hg. Glaucomatous visual field defect was defined as (1) outside normal limits on glaucoma hemifield tests; or (2) 3 abnormal points, with a P < 5% probability of being normal, 1 abnormal points with P < 1% probability of being normal by pattern deviation; or (3) pattern standard deviation of 5% confirmed on 2 consecutive reliable tests (fixation loss rate <20%; false-positive and false-negative error rates <25%). The exclusion criteria were: (1) history or evidence of other optic neuropathies or congenital anomalies of the optic disc; (2) signs of pathologic myopia including myopic choroidal neovascularization, lacquer crack, angioid streak; (3) extremely myopic eyes with an axial length >30 mm; and (4) eyes with poor image quality in which the PPA was not delineated clearly on OCT. Eligibility was determined by 2 glaucoma specialists (Y.C.K. and H.L.P.), who evaluated the optic disc appearance on stereoscopic disc photographs, RNFL defects on red-free fundus photographs, and results of VF examinations. Evaluators were masked to all other patient and ocular data, and an eye was excluded from study analyses if a consensus could not be reached. When both eyes were eligible, one eye was chosen randomly per subject for data analysis.
Tomographic images of the peripapillary fundus were taken by using the SSOCT. The detailed specifications of the SSOCT have been described 16 . Briefly, a 3D imaging data set was acquired for each subject with a raster scan protocol of 512 × 256 A-scans per data set. Each 3D scan covered an area of 12 mm × 9 mm which was enough to cover the whole peripapillary region 17 . Each horizontal line scan was scanned 27 µm apart, vertically. DRIOCT triton software provides a measurement tool to draw straight lines. Two observers (YCK and HYLP) that were masked to all other ocular and patient information manually measured two parameters on the raster image. (1) γ-PPA, which was measured from the anterior scleral canal opening (ASCO) to BM endpoint, and (2) β-PPA, which was measured between BM endpoint to the beginning of the retinal pigment epithelium (RPE) with underlying BM at the center of the optic disc 6 .
In addition to measurements of the PPA, optic disc tilt measurements were identified by two different measure, horizontal disc tilt, and vertical disc tilt, respectively. Horizontal and vertical tilt angle was measured using the clinical disc margin as the ONH plane and the imaginary line connecting each BM margin as the reference plane 18 . Degree-of-tilt was defined as the angle between the reference plane and the ONH plane. Angle measurements were performed by two observers (YCK and HYLP) with the software intrinsic angle tool. A positive degree of horizontal tilt indicated tilt towards temporal, and a negative horizontal tilt indicated tilt towards nasal. A positive degree of vertical tilt indicated tilt towards inferior, and a negative vertical tilt indicated tilt towards superior.
The parameters of the outer retinal layer separation were measured at the center of the optic disc using the same section as above. Advances in OCT enable to discriminate the individual retinal layers with high-resolution noninvasive real-time imaging 19 . In the OCT images, the INL of the retina is defined as a relatively hyporeflective zone, internal to hyperreflective outer plexiform layer 20 . The ONL is defined as the hyporeflective zone between the external limiting membrane and the outer plexiform layer 20 . These individual retinal layers assemble and ends at the BM (Fig. 1D) 21 . However, elongated peripapillary region of myopes distorts this configuration, resulting in various endpoints of each layer (Fig. 1E,F). The endpoint of the ONL was defined as the point where the continuous hyporeflective ONL merges at the surface of the BM or sclera. The endpoint of the INL was defined as the point where the continuous hyporeflective INL merges at the surface of BM or RPE. ONL-INL separation was defined as the straight-line distance from ONL endpoint to INL endpoint (Fig. 2). INL-ASCO width was measured as the straight-line distance from INL endpoint to ASCO (Fig. 2). For the sub-analysis, each observer, who was masked with the clinical information of the OCT image, classified each eye into 2 categories with respect to amount of ONL-INL separation: (1)  INL-ASCO width were evaluated by calculating intraclass correlation coefficients 22 . Comparison between 2 groups was performed with the chi-square and Student's t tests. To identify the associated factors with ONL-INL separation, Pearson's correlation analyses were used. P < 0.05 was considered to be statistically significant.

Results
A total 149 eyes of 149 subjects who had more than 1 year of follow-up using OCT were included in the study. Of the initial subjects, 9 eyes were excluded because of a history or evidence of non-glaucomatous optic neuropathy (4) or juvenile glaucoma (5). Of the remaining 140 eyes, 32 eyes with minimal PPA (<200 µm) were excluded, leaving a final sample of 108 eyes of 108 subjects.  The mean subject age and refractive error were 44.34 ± 11.96 years (range, 14-72 years) and −4.57 ± 3.16 diopters (range, −10.75 to + 1.0 diopters), respectively. 41 eyes were assigned as the non-glaucomatous group and 67 were assigned as the glaucomatous group. Table 1 summarizes the comparison of baseline characteristics between the non-glaucomatous group and glaucomatous group. No significant differences were found between groups with regard to age, refractive error, axial length and central corneal thickness. Mean deviation and pattern standard deviation of visual field were more glaucomatous in the non-glaucomatous group and glaucomatous group (all P < 0.001, respectively). Table 2 shows the amount of outer retinal layer separation and peripapillary region characteristics of the two groups. There was excellent interobserver reproducibility in measurement of the ONL-INL separation width, INL-ASCO width, β-PPA width, γ-PPA width, horizontal tilt and vertical tilt (intraclass correlation coefficient = 0.998, 0.972, 0.842, 0.882, 0.996 and 0.987, respectively). INL-ASCO separation, γ-PPA width and total PPA width was similar between the two groups (P = 0.741, P = 0.534 and P = 0.381). However, ONL-INL separation and β-PPA width was significantly larger in the glaucomatous group (P = 0.006 and P = 0.027, respectively). Table 3 compares the non-separated group (separation under 200 µm) and separated group (separation over 200 µm). There were no significant differences in the baseline characteristics but the separated group had worse glaucomatous indications in visual fields and RNFL thickness.
On correlation analysis, ONL-INL separation was significantly associated with axial length, pattern standard deviation of visual field, β-PPA width and horizontal disc tilt (Fig. 3). INL-ASCO separation was significantly associated with axial length, horizontal disc tilt, γ-PPA width but had no significance with visual field parameters or RNFL thickness (Table 4). Linear regression analysis was done regarding the amount of ONL-INL separation. In the univariate and multivariate analysis, amount of β-PPA width and horizontal tilt degree was associated with ONL-INL separation (Table 5).

Discussion
We describe transversely separated ONL-INL at the peripapillary region which was associated with glaucomatous parameters. The separation is associated with the longer β-PPA width, larger horizontal tilt of the disc, worse glaucomatous visual field and RNFL thickness which are all factors associated with development and progression of glaucoma. As far as we know, this is the first documentation of transversely separated outer retinal layer in the peripapillary region of glaucomatous myopic eye.
Over the years, there have been speculations regarding the pathogenesis of PPA. PPA has been hypothesized as an atrophic change of the RPE-BM complex and subsequent photoreceptor and choriocapillary atrophy 23 . In contrast, it has been regarded as a resultant of scleral stretching associated with development of myopia 24 . Our data clearly demonstrate that in the stretching process of the peripapillary sclera, the retinal configuration can be disorganized and separated transversely. The association of ONL-INL separation with amount of myopia suggests that peripapillary outer retinal separation is the result of scleral stretching associated with axial elongation of the eyeball. The endpoint of each retinal layer merges together at the BM opening of optic disc. As the eyeball grows axially, the temporal peripapillary sclera becomes elongated. In this process, each retinal layer endpoint that sits on the peripapillary sclera is also elongated and eventually results in bilateral separation. The width of ONL-INL separation had significant association with the amount of refractive error and the axial length of the eyeball. Our observation has implications that the properties of the peripapillary sclera stretching should be taken into consideration as well. So far, PPA has been hypothesized to change equivalently throughout the whole stretched area. However, our data suggest that peripapillary sclera stretches non-uniformly within the elongated scleral flange. As the peripapillary sclera undergoes elongation in myopic eyes, the optic disc margin and the temporal margin of the PPA should be pulled with a same force from the either side (Newton's third law of motion) 25 . The peripapillary scleral tissue has a characteristic arrangement which transforms from the circumferential orientation adjacent optic disc to the lattice orientation of the posterior pole 26 . Thereby, the temporal and the nasal side of the PPA each undergoes uneven alteration according to the different characteristics and arrangement of each side. This hypothesis is supported by our OCT imaging of the region. In Fig. 4, we compared the OCT images from two subjects, one with minimal ONL-INL separation (Fig. 4B) and the other with large amount of ONL-INL separation (Fig. 4D). With small ONL-INL separation, most of the PPA alterations came from the nasal side of the peripapillary scleral flange. On the other hand, with large ONL-INL separation, most of the stretching came from the temporal side of the peripapillary scleral flange, which may alter the INL endpoint position that sits above. While this finding is not proven longitudinally, this theoretic framework may provide an additional explanation for the association of PPA width with glaucoma 27,28 .  The amount of transverse separation width between ONL-INL was associated with β-PPA width but not with γ-PPA width. Our data shows that the ONL-INL separation had significant association with β-PPA and the INL-ASCO separation had significant association with γ-PPA. In the measurement process, it showed in numerous times that the ONL endpoint matched beginning point of RPE and the INL endpoint matched BM endpoint. Considering the strong association, one hypothesis is that the BM may have some kind of attachment with the INL or the inner retina which moves along with another. However, there is no histologic reference on this speculation and further evaluations should be done.
Manjunath et al. 29 and Lee et al. 30 recently examined the appearances of PPA and retinal morphologic changes with OCT imaging. They reported PPA characteristics as photoreceptor loss and RPE disruption. Additionally, they documented retinal changes such as the RNFL thickness plaque, RNFL cystic spaces and abnormal retinal sloping, but did not examine the changes of individual retinal layer specifically. With detailed investigation, we describe the transverse separation of the outer retinal layer endpoint that seems to be connected with glaucomatous damage.
The question of how the changes in large ONL-INL separation induce glaucomatous optic nerve damage remains to be addressed. In eyes with large ONL-INL separation, disjointed retinal anatomy may induce axonal stress besides the elevated intraocular pressure and its successive mechanical stress to the retinal ganglion cells in the lamina cribrosa level. The pathogenesis of this finding should be addressed in the near future.
Potential limitations of the present study should be discussed. First, all patients were referred to a glaucoma clinic in a tertiary hospital. Further prospective study is needed in subjects with healthy eyes. Second, we are unable to demonstrate conclusively that there is a relationship between the ONL-INL separation and the degree of glaucomatous damage because of the retrospective study design. Only the association discovered with assumption of causality can be reported. The association between changes in ONL-INL separation, glaucomatous damage, axial length and disc change should be evaluated in future prospective studies. Third, our study included only a selected group of individuals who had temporal PPA larger than 200 µm on horizontal OCT scan image. It has remained unclear whether the observations made in this group of individuals can be transferred to groups of  patients with small PPA. Fourth, this finding is observed using a relatively novel SSOCT instrument. Whether this finding could be measured with other instruments should be evaluated in future studies.
In conclusion, myopic eyes may develop transverse separation of ONL-INL, which was associated with worse glaucomatous parameters. These observations may be of interest to elucidate the role of PPA in glaucoma pathogenesis and a clinical index to take notice for myopic subjects.