Differences of Retinal Nerve Fibre Distributions Between High Myopes and Emmetropes – A Population Based Study

Diagnosing glaucoma in patients with high myopia is still a challenge despite the rapid develop of optical coherence tomography (OCT) in the past 2 decades. One of the reasons was the different geographic anatomies of the retinal nerve bre layer (RNFL) among the high myopes and the emmetropes. This often resulted in errors when RNFL thicknesses were analysed. As diagnosis of glaucoma relied on the detection of focal RNFL thinning in OCT, this could lead to wrong diagnosis. This study aimed to compare the distributions of superotemporal and inferotemporal retinal nerve bre bundles among these 2 groups of population.


Results
One-hundred-thirty-four highly myopic eyes from 74 subjects were included in the myopic group and 188 non-myopic eyes from 94 subjects were recruited as control group. The mean inter-peak distance of the myopic group 8.89 ± 1.00 mm and that of the non-myopic group was 6.81 ± 0.61 mm. The mean interpeak distance was signi cantly larger in the myopic group (p < 0.0001).

Conclusions
Retinal nerve bres arranged differently in high myopes with a tendency of more temporally arranged superotemporal and inferotemporal retinal nerve bre bundles. Direct comparison of RNFL thickness with built-in normative database made by most of the OCT machines might lead to false positive diagnoses of glaucoma in patients with high myopia.

Background
Myopia is one of the most common ocular abnormalities reported worldwide 1 . High myopia, or pathological myopia, is of particular concern as it is not only a refractive problem, but also increases ones' risks of various ocular diseases such as retinal detachment, myopic maculopathy and primary open angle glaucoma 2 . Glaucoma is the commonest cause of irreversible blindness affecting more than Page 3/9 60 million populations 3 . One with high myopia possesses more than 3 times risks of glaucoma 4 . Due to its irreversibility, early diagnosis and close disease monitoring is crucial to prevent permanent visual loss.
Before the era of optic coherence topography (OCT), clinicians relied on an increased in cup-to-disc ratio (CDR) and the characteristic visual eld loss demonstrated by Humphrey perimetry to diagnose glaucoma. The rapid development of OCT over the past 2 decades has revolutionized glaucoma diagnosis and the way glaucoma were managed. OCT allowed high resolution cross-sectional imaging of the retina, enabling segmentation and accurate measurements of different retinal layers. This empowered clinicians to diagnose glaucoma earlier and more accurately monitoring disease progression by detecting focal thinning of peripapillary retinal nerve bres layer (RNFL). Various studies have proven that RNFL thinning, when compared with one's age-matched normative data, has been one of the most sensitive parameters for glaucomatous optic neuropathy (GON) diagnosis and monitoring [5][6][7][8][9] .
Nevertheless, diagnosing glaucoma and monitoring glaucoma progression in patients with high myopia is still a substantial challenge 10 . High myopes usually have a different optic disc morphology compared with their non-myopic counterparts. They tend to have tilted optic discs with more oval con gurations and peripapillary atrophies 11,12 . Therefore, their CDR might not be easily discerned. Disagreements in CDR could occur even between different experienced glaucoma specialists, let alone general ophthalmologists.
On the other hand, the RNFL anatomy could be quite different between myopic and non-myopic people.
Many studies have been done comparing the thickness of RNFL between myopia and non-myopic populations [13][14][15][16][17][18][19][20] . A number of studies found that people with high myopia have thinner peripapillary RNFL 13,16 . However, seldom had studied the discrepancies of the arrangement of the superotemporal and inferotemporal retinal nerve bre bundles between high myopes and emmetropes. Some even suggested that the RFNL arrangement might not be correlated with the axial length and spherical equivalent of the eye. 19 We compared the circumpapillary OCT images from healthy subjects with and without high myopia to study the differences of their RNFL arrangements, particularly on the superotemporal and inferotemporal retinal nerve bre bundles, between these 2 groups.

Methods
This is a retrospective case-control study. Circumpapillary OCT images were taken from normal Chinese populations who participated in a community eye health screening program conducted in the Southern District of Hong Kong. Subjects with spherical equivalent (SE) of ≤ -6 diopters and axial length (AXL) ≥ 26 mm were classi ed as the myopic group, whereas similar number of subjects with SE within 2 diopters and AXL of 24 ± 0.5 mm would be randomly selected from the database as non-myopic control group. This study was approved by the Institutional Review Board of the University of Hong Kong / Hospital Authority Hong Kong West Cluster (HKU/HKWC IRB) and in accordance with the tenets of the Declaration of Helsinki.
Circumpapillary OCT images of the subjects were taken using Spectralis OCT machine (Heidelberg Engineering GmbH, Heidelberg, Germany). Cross-sectional OCT images of a circle of 3.45 mm diameter centred at the optic disc were obtain for each subject by a trained and experienced operator. The 2 humps in the circumpapillary OCT images were identi ed and the distances between the peak of these 2 humps were measured and compared by an independent investigator.
The inter-peak between the 2 retinal nerve bre bundles of the 2 groups are compared. Each eye of the subjects was analysed separately. Only the quali ed eye would be included for subjects with unilateral high myopia. Independent sample t-tests were performed to compare the inter-peak distances and angular widths between the myopic and non-myopic groups. Statistical signi cance was de ned as p ≤ 0.05.

Results
Circumpapillary OCT images of 134 highly myopic eyes from 74 subjects and 188 control from 94 subjects were included in this study. The demographics of the subjects were shown in Table 1. The mean AXL of the myopic group was 27.64 ± 1.12 mm and that of the non-myopic group was 23.95 ± 0.35 mm (p < 0.0001). The mean SE of the myopic group was − 8.97 ± 2.38D and that of the non-myopic group was 0.12 ± 1.11D (p < 0.0001). The mean inter-peak distance was 8.89 ± 1.00 mm for the myopic group and 6.81 ± 0.61 mm for the non-myopic group. The inter-peak distance was signi cantly larger in the myopic group (p < 0.0001). Table 2 showed the results of independent t-test. Inter-peak distance (mm) 8.89 ± 1.00 6.81 ± 0.61 < 0.0001 The inter-peak distances represented the circumferential distance from the superotemporal bundle to the inferotemporal bundle through the nasal retina. Therefore, the 2 retinal nerve bre bundles were more temporally arranged in the retina.

Discussion
In the modern era of ophthalmology, OCT has been an indispensable tool in diagnosing glaucoma.
Thinning of the peripapillary RNFL and the ganglion-cell-internal-plexiform-layer (GC-IPL) complex are 2 most reliable indicators in OCT images to suggest GON 20,21 . They have been proven to be sensitive to diagnose early glaucoma and useful in monitoring glaucoma progression as well 7,21−23 . They have been heavily relied on for these purposes as perimetry is notorious in its subjectivity and variability 24 , while intraocular pressure (IOP) could not thoroughly re ect the nerve status and observed worsening of CDR would have been too late to intervene.
Nevertheless, diagnosing glaucoma in patients with high myopia has been a challenge despite the rapid advancements of OCT technology in the past 2 decades 10 . There are many OCT manufacturers in the market but most of them produce similar reports by comparing the peripapillary RFNL thickness with their own age-match normative database stored in their systems. The comparisons were made directly by quadrants or clock hours of the optic nerve head (ONH).
Our study demonstrated that the superotemporal and inferotemporal retinal nerve bre bundles are arranged differently in subjects with high myopia and a long AXL. This could explain the phenomenon that high myopes were often over-diagnosed with glaucoma. Since the retinal nerve bre bundles, together with the major retinal vessel arcades tend to attain a more temporal distribution, directly compare the RNFL thickness quadrant by quadrant, or by clock hours, could result in misalignment during analysis. Therefore, the OCT systems might mistakenly compare the areas outside the retinal nerve bre bundles which are usually thinner to the thicker area of bre bundles in the normative data. This results in false positives if clinician did not interpret the reports very carefully. As most of OCT manufacturers label area of RNFL thinning using red colour, it is also called the "red disease" 25 . On the other hand, if the retinal neve bre bundles of a high myope with early GON, comparing his RNFL thickness to the supposed thinner non-bundle areas could lead to false negative, which might result in delayed diagnosis and treatment.
To further nd out how common are false positive and false negative diagnosis of glaucoma in high myopes, a much larger number of highly myopic subjects with longitudinal observations and thorough assessments would be necessary. In order to solve this potential problem, normative database speci c for high myopes would be needed.

Conclusions
For population with high myopia, their retinal nerve bre bundles were arranged differently when compared to the non-myopic population, with a tendency of more temporally arranged superotemporal

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.