Peripheral Ganglion Cell Complex Thickness and Retinal Microvasculature in Myopia Using Wide-Field Swept-Source OCT

Abstract

Purpose: This study aims to investigate the impact of axial elongation on ganglion cell complex thickness (GCCT) and retinal capillary density (CD) using wide-field swept-source optical coherence tomography angiography.

Methods: A retrospective cross-sectional analysis was conducted involving 506 eyes. Fovea-centered scans were obtained to assess the subregional GCCT and capillary density across the whole retina, the superficial capillary plexus (SCP), and deep capillary plexus (DCP) among three groups: normal control, high myopia (HM) eyes with axial length < 28 mm, and HM eyes with axial length > 28 mm. Regional variations (central vs. peripheral, quadrants difference [superior, inferior, nasal, and temporal]) were analyzed.

Results: In HM eyes with axial length > 28 mm, GCCT and retinal CD exhibit a general decline in most regions (P < 0.05). In HM eyes with axial length < 28 mm, significant reductions were observed specifically in peripheral regions, as in the GCCT beyond the 3 × 3 mm² area and CD in the 9–12 mm whole retina, 9–12 mm superior SCP, and 6–12 mm DCP (P < 0.05). Maximum GCCT and retinal CD reduction with axial elongation was observed in subregions beyond 6 × 6  mm².

Conclusions: GCCT beyond the 3 × 3 mm² area and peripheral retinal CD beyond the 6 × 6  mm² area were more susceptible to axial elongation and are thereby deserving of particular attention.

Translational Relevance: It is necessary to evaluate different regions during the clinical assessment of the effect of myopia on the fundus and pay close attention to the peripheral retina.

Introduction

Myopia poses an enormous threat to global public health,¹ and it is estimated that by 2050, 4758 million (49.8%) and 938 million individuals (9.8%) around the world will be affected by myopia and high myopia (HM), respectively.² If not adequately corrected, myopia can substantially contribute to visual impairment. Additionally, individuals with HM face a significantly higher risk of developing vision-threatening pathologies that cannot be fully prevented by optical correction alone. Pathological myopia may cause various complications, such as chorioretinal atrophy, posterior staphyloma, choroidal neovascularization, rhegmatogenous retinal detachment, myopic maculopathy, cataracts, and glaucoma.³ Axial elongation of eyes with myopia could affect a wide range of the retina; thus a detailed examination from the posterior pole to the peripheral retina is required.⁴ 

The advent of optical coherence tomography (OCT) has enabled rapid, noninvasive, high-resolution examinations of retinal structures.⁵ OCT angiography (OCTA) has recently enabled a detailed retinal microvasculature assessment.⁶ Previous studies using Fourier-domain OCT and OCTA have shown that HM, defined as an axial length (AL) > 26 mm, resulted in macular ganglion cell complex (GCC) thinning^7,8 and lower capillary density (CD) in the superficial capillary plexus (SCP) and deep capillary plexus (DCP) within the 6 × 6 mm² area.^9,10 

Nevertheless, most previous studies had a limited scanning range; thus, the effects of axial elongation on the peripheral retinal structures and microvasculature beyond the 6 × 6 mm² area have not been fully elucidated.^9,10 Therefore we conducted this study to explore the effects of axial elongation on GCC thickness (GCCT) and retinal microvasculature in the posterior pole and more peripheral regions beyond the 6 × 6 mm² area among eyes with varying AL using wide-field (WF) swept-source (SS) OCTA scans. We also aimed to further clarify the importance of assessing the CD of more peripheral retinal areas during the assessment of myopia progression. 

Methods

Study Participants

All participants for this retrospective cross-sectional study were enrolled at the Zhongshan Ophthalmic Center between June 5, 2016, and May 1, 2022. The study was conducted following the guidelines of the Declaration of Helsinki. The study design was approved by the Ethics Committee of the Zhongshan Ophthalmic Center (ID: 2020KYPJ026). Written informed consent was obtained from all participants before study initiation. 

In this study, 506 eyes from 506 participants were divided into three groups: the normal control group (AL < 26 mm), HM eyes with AL 26–28 mm, and HM eyes with AL > 28 mm.^11–13 The inclusion criteria were as follows: age 20–70 years, best-corrected visual acuity (BCVA) 0.1 or higher, appropriate binocular fixation, visual field meeting Anderson-Patella criteria for healthy individuals,¹⁴ and intraocular pressure (IOP) in the normal range (12–21 mm Hg). Patients with a history of cataracts, ocular trauma or surgery and diseases that may affect the retinal microvasculature, such as glaucoma; abnormalities detected during fundus examination by ophthalmoscopy and OCT (including chorioretinopathy, macular hole, macular neovascularization, diffuse chorioretinal atrophy, and all pathological myopic fundus changes³), ocular inflammatory diseases or systemic diseases that may affect the retinal microvasculature such as hypertension, diabetes, kidney diseases, etc. were excluded.¹⁵ Only one eye from each patient was randomly included for statistical analysis. 

Ocular Examinations

All participants underwent comprehensive ophthalmic examinations, including BCVA, uncorrected visual acuity, automatic refraction assessment (autorefractor, KR-800; Topcon Co, Tokyo, Japan), slit lamp-based biomicroscopy, central corneal thickness measurement (Pachymetry SP-3000; Tomey Corporation, Nayoga, Japan), IOP measurement by Goldmann applanation tonometry, dilated fundus examination, AL measurements (IOL Master 700; Carl Zeiss Meditec, Jena, Germany), stereo fundus photography (Nonmyd WX3D; Kowa, Aichi, Japan), and 24-2 standard automated perimetry using the Swedish interactive thresholding algorithm (Humphrey Field Analyzer 3; Carl Zeiss Meditec). The blood pressure of each patient was measured during the examination (Omron M7 Blood Pressure Monitor; Matsusaka, Mie, Japan). 

WF SS-OCTA Imaging

A 200 kHz swept-source optical coherence tomography angiography (SS-OCTA) device (VG200; SVision Imaging, Ltd., Luoyang, Henan, China) with a central wavelength of 1050 nm and bandwidth of 100 nm was used for OCTA. Trained technicians acquired 15 × 12 mm² WF SS-OCTA images centered on the fovea. In the scanning pattern of 15 × 12 mm, each B-scan contained 1280 A-scans across the horizontal dimension, whereas 1024 B-scans were repeated twice at each position, achieving a resolution of 11.7 µm per pixel. We carefully evaluated all B-scan images, manually adjusted the segmentation lines, and excluded unwarranted images with poor image quality (signal-to-noise ratio ≤ 30), mis-segmentation, blinking or motion artifacts, decentration, and defocus.¹⁶ 

Image Processing and Analysis

The Built-in software VG200D (version 1.40.0) was adopted for SS-OCTA image processing and analysis. We measured the GCCT and retinal CD (RCD) in the subregions within the 15 × 12  mm² scanning area in the Early Treatment Diabetic Retinopathy Study (ETDRS) grid (Fig. 1), including the GCCT and CD in the whole retina, SCP, and DCP in the annular regions with 1–3 mm, 3–6 mm, 6–9 mm, and 9–12 mm diameters. Each annular was divided into four quadrants: superior, inferior, nasal, and temporal. In our study, GCC was segmented with an inner boundary 3 µm above the internal limiting membrane and an outer boundary set as the junction between the inner plexiform layer and inner nuclear layer, including the retinal nerve fiber layer and ganglion cell-inner plexiform layer. The SCP was segmented consistently with the GCC, supplying blood flow to the GCC. The DCP was segmented with an inner boundary set as the junction between the inner plexiform layer and the inner nuclear layer and an outer boundary set as the junction between the outer plexiform layer and the outer nuclear layer, including the inner nuclear layer and the outer plexiform layer. The whole retina layer was segmented with an inner boundary set as 3 µm above the internal limiting membrane and an outer boundary set as 10 µm below Bruch's membrane. Figure 2 displays the schematic diagram of the layers included in this study. Considering the impact of ocular magnification on retinal blood flow quantification, we also corrected the OCTA-derived parameters including GCCT and CD in the whole retina, SCP, and DCP using Littmann's method and Bennett's formula.^17,18 Furthermore, nasal GCCT and RCD beyond the 6 × 6  mm² area were excluded to avoid the disturbance of ubiquitous large blood vessels near the optic disc. 

Figure 1.

 

Representative images of 15 × 12 mm² wide-field SS-OCTA scanning. (A) SS-OCTA image; (B) corresponding en-face image; (C) B-scan image; (D) measurement of the capillary density in the annular regions of 1–3 mm, 3–6 mm, 6–9 mm, and 9–12 mm in diameter, which were further divided into four quadrants. All images are from the left eye. I, inferior; N, nasal; S, superior; T, temporal.

Figure 2.

 

Schematic figure of GCC and RCD stratification. (A) The boundaries of GCC, which was segmented with an inner boundary 3 µm above the internal limiting membrane and an outer boundary set as the junction between the inner plexiform layer and inner nuclear layer, including the retinal nerve fiber layer and ganglion cell-inner plexiform layer. (B) The boundaries of CD in the whole retina, which was segmented with an inner boundary set as 3 µm above the internal limiting membrane and an outer boundary set as 10 µm below Bruch's membrane. (C) The boundaries of CD in SCP, which was segmented consistently with the GCC, (D) The boundaries of CD in DCP, was segmented with an inner boundary set as the junction between the inner plexiform layer and the inner nuclear layer and an outer boundary set as the junction between the outer plexiform layer and the outer nuclear layer, including the inner nuclear layer and the outer plexiform layer. All images are from the left eye.

Statistical Analysis

R, version 4.1.2 (R Foundation for Statistical Computing) was used for the statistical analysis. Data distribution was assessed using the Shapiro–Wilk normality test. Normally distributed continuous variables are reported as mean ± standard deviation values, and non-normally distributed continuous variables are reported as median and interquartile range values. Categorical variables are reported as frequency (percentage). Continuous numerical data were analyzed using one-way analysis of variance; Bonferroni correction was used for multiple comparisons.¹⁹ Categorical variables were analyzed using the χ² test. 

One of the aims of our study was to evaluate the GCCT and RCD variation patterns in different regions with different ALs; notably, the GCCT and RCD varied considerably in different regions. Therefore direct comparison using regression coefficients of linear regression was inappropriate. To address this issue, we standardized all variables with Z-score standardization before incorporating them into the multivariable linear regression models.²⁰ Initially, we incorporated target factors, as well as all potential confounding variables with a P value ≤ 0.05 from univariate linear regression analyses as preliminary independent variables in our model. To mitigate the confounding effects in the model,^21–25 we used a backward stepwise elimination procedure to refine the selection of independent variables. The final model retained age, AL, and systolic blood pressure (SBP) as the conclusive independent variables. The magnitude of reduction was presented as the β coefficient of AL in the multivariable linear regression models while controlling for potential confounders. P < 0.05 was considered statistically significant. 

Results

Baseline Characteristics

In total, 169, 184, and 153 eyes were included in the normal control, HM eyes with AL < 28 mm, and HM eyes with AL > 28 mm groups, respectively. The demographic characteristics of all participants are listed in Table 1. There were no statistically significant differences in age, gender, IOP, central corneal thickness, history of systematic diseases, or nonocular surgery among the groups (all P > 0.05). However, the differences in body mass index, SBP, and diastolic blood pressure among the three groups were statistically significant (all P < 0.05). As expected, AL, BCVA, and spherical equivalent differed statistically among the three groups (all P < 0.05). 

Table 1.

 

Demographic and Clinical Characteristics of the Study Participants

 

Effects of Axial Elongation on GCCT

As shown in Table 2, the average GCCT of HM eyes with AL < 28 mm was lower than that of normal controls (all P < 0.05), except for the GCCT in the 1–3 mm annular region (P = 0.90). The average GCCT of the 1–3 and 3–6 mm annular regions were significantly lower in HM eyes with AL > 28 mm than that in HM eyes with AL < 28 mm (both P < 0.05), whereas the average GCCT of the 6–9 mm and 9–12 mm annular regions was comparable in HM eyes with AL < 28 mm and HM eyes with AL > 28 mm (both P > 0.05). 

Table 2.

 

Comparisons of the GCCT and the CD in the SCP and DCP Between HM Eyes and Normal Controls

 

Effects of Axial Elongation on RCD

As shown in Table 2, the SCP CD did not significantly differ between HM eyes with AL < 28 mm and normal controls in all annular regions (all P > 0.05), except for the SCP CD of the superior quadrant in the 9–12 mm annular region (P = 0.03). The average SCP CD in all annular regions of HM eyes with AL > 28 mm showed a significant reduction compared with those of the other two groups (all P < 0.05). The quadrant SCP CD of HM eyes with AL > 28 mm was mostly lower than those of the other two groups (in most regions, P values were < 0.05). 

After controlling for age and SBP, we observed that the magnitude of the average SCP CD loss gradually increased from the 1–3 mm to 6–9 mm annular regions and peaked in the 6–9 mm annular region (β = −0.131, P < 0.001) (Table 3, Figure 3). The region where axial elongation affected the SCP CD most significantly was the superior quadrant of the 9–12 mm annular region (β = −0.156, P < 0.001). Nasal SCP CD within the range of 1–6 mm was not affected by axial elongation (1–3 mm annular region: P = 0.25; 3–6 mm annular region: P = 0.63), which was in accordance with the GCCT. 

Table 3.

 

The Multivariable Linear Regression Models Adjusted for Age and Systolic Blood Pressure With Axial Length as the Independent Variable

 

Figure 3.

 

β values for changes in GCCT and CD in the whole retina, SCP and DCP are shown in ETDRS grids and subfields. 1–3, 3–6, 6–9, and 9–12 represent ETDRS annular regions with the 1–3 mm, 3–6 mm, 6–9 mm, and 9–12 mm diameters, each annulus is divided into four quadrants: I, S, N, T. The unsegmented rings around the periphery of each ETDRS annulus show the average β values in the corresponding regions (A represents for average). Data within the 0–1mm and 6–12 mm nasal area were not included. A, average; I, inferior; N, nasal; S, superior; T, temporal.

As shown in Table 2, compared with normal controls, the average DCP CD of HM eyes with AL < 28 mm was only significantly reduced in the 9–12 mm annular region (P = 0.01). However, HM eyes with AL > 28 mm showed significantly decreased average DCP CD values in all annular regions compared with those of the other two groups (all P < 0.05). 

After controlling for age and SBP, we observed that the magnitude of the average DCP CD loss gradually increased from the 1–3 to 6–9 mm annular regions and peaked in the 6–9 mm annular region (β = −0.366, P < 0.001), which was similar to the variation pattern of the average SCP CD (Table 3, Figure 3). Axial elongation affected the DCP CD most significantly in the temporal quadrant of the 6–9 mm annular region (β = −0.473, P < 0.001). Additionally, regions with higher regression coefficients were mostly located in the temporal and superior quadrants. 

In HM eyes with AL < 28 mm, the average CD in the whole retina only showed a significant reduction in the 9–12 mm annular region compared with that in normal controls (P = 0.02), with a gradually increasing magnitude of reduction from the 1–3 mm to 9–12 mm annular region; the magnitude was greatest in the temporal quadrant, followed by the superior and then inferior or nasal regions (Supplementary Tables S1 and S2). 

Discussion

In this study, we assessed the differences and variation patterns of GCCT and retinal microvasculature among patients with different degrees of myopia using WF SS-OCTA. Our study suggests that the peripheral retinal GCC and microvasculature are more susceptible to the effects of axial elongation. Specifically, HM eyes with an AL shorter than 28 mm displayed a predominant reduction in GCCT beyond the central 3 × 3 mm² and a primary decrease in RCD beyond the 6 × 6 mm², with the alterations in the temporal peripheral DCP being particularly pronounced. Conversely, HM eyes with an AL longer than 28 mm exhibited a widespread reduction in both GCCT and RCD. To our knowledge, this is the first study to report these findings, highlighting the differential impact of axial elongation on the peripheral retinal structures across varying degrees of high myopia. 

Previous OCTA studies have demonstrated that HM caused a reduction in macular retinal microvasculature.^9,10 Moon et al.²⁶ reported a progressive decrease in the vessel density and vessel skeletonized density in the whole retina, SCP, and DCP in patients with increasing severity of myopia within the 12 × 12 mm² scanning coverage area. Conversely, Yang et al.²⁷ conducted a study on a cohort of 268 individuals from the Han Chinese population, exhibiting varying degrees of myopia. They analyzed the macular vessels within a scanning area of 3 × 3 mm² and discovered that myopia did not affect retinal microvasculature. Min et al.²⁸ incorporated 52 highly myopic eyes and 52 normal eyes from the Korean population into their study, using a scanning area of 3 × 3 mm². They observed that the superficial parafoveal vascular density was significantly reduced in the myopic eyes compared to the normal eyes. However, no significant difference was found in the deep parafoveal vascular density between the two groups. Similarly, Su et al.²⁹ conducted a study on 75 eyes of varying degrees of myopia in the Han Chinese population, using a scanning range of 6 × 6 mm². They found that the vessel density in the SCP was significantly reduced in highly myopic eyes compared to normal eyes, whereas the vessel density in the DCP exhibited no significant changes. However, our study suggested that myopia affected retinal microvasculature in most regions in the superficial and deep layers. These inconsistencies may stem from the scanning coverage, study populations and the range of AL of the participants included. 

Our study observed a preferential effect of axial elongation on CD in the temporal DCP. Increasing evidence suggests that the DCP is more susceptible to HM than the SCP, which may be related to the mechanical stretching caused by axial hyperextension in HM eyes, and the vulnerable small-diameter vessels in the DCP are more likely to be straightened and disrupted compared with the large retinal vessels in the SCP.^30–33 A longitudinal study by Lin et al.³⁰ showed that the rate of CD loss in the DCP was significantly faster in the HM group than that in the normal control group, whereas the rates of CD loss in the SCP were similar between the two groups, which is in agreement with our findings. Furthermore, for quadrant analysis we found the CD in the whole retina and DCP generally showed the most remarkable temporal alterations, followed by superior and then inferior or nasal regions, which was in line with the study by Moon et al.²⁶ Moreover, in our study, the 1–6 mm nasal GCCT and SCP CD were not affected by myopia. Liu et al.³² noted that the inferior nasal retinal microvasculature was the last to be affected by myopia, which was also in agreement with our results. Regarding why the retinal microvasculature on the temporal side is more susceptible to the influence of axial elongation, we speculate that the underlying mechanism may be due to the presence of larger vessels in proximity to the nasal side of the optic disc. However, further studies are needed to elucidate the mechanism of quadrant preferences. 

In our study, reduction in the retinal microvasculature in myopic eyes occurred mainly in the peripheral retina beyond the 6 × 6 mm² area among patients with AL > 28 mm. Jonas et al.³⁴ observed that peripheral retinal thinning was associated with longer ALs, but macular retinal thinning was not; this supported the notion that myopic enlargement of the globe occurs predominantly with axial elongation of the globe walls in the midperiphery. However, it is crucial to note that the study by Jonas et al.³⁴ did not specifically explore capillary density. Therefore the findings of Jonas et al.³⁴ might suggest that the peripheral retina is more sensitive to axial elongation, which could complement the current discoveries. Further investigation is required to elucidate the more detailed mechanisms involved. 

We also aimed to explore the effect of myopia on the GCCT. Previous studies involving an RTVue-100 system that covered a 7 × 7 mm² scanning region generally implied macular GCC thinning in HM.^7,8,35,36 In contrast, Wu et al.³⁷ reported no statistical difference in the GCCT between HM and non-HM eyes within the 3 × 3 mm² scanning range. Rezapour et al.³⁸ found that GCC thinning was not associated with AL, although this discrepancy may be attributable to the scanning range and study participants since all participants in this study had glaucoma. In the present study, GCC thinning in most regions correlated with increasing myopia. Our study suggested that the GCCT beyond the 3 × 3 mm² area decreased more significantly during myopia progression, which may partially explain these discrepancies. The variations in the trends of GCCT and RCD associated with axial elongation may stem from the differing resistances that neural and vascular tissues exhibit towards axial stretching. Moreover, distinct layers within the retinal structure and microvasculature may demonstrate diverse patterns in response to axial elongation.^39,40 On further stratified analysis, it becomes apparent that the trends of change in GCCT and CD in SCP are quite similar. 

By comparing the results of GCCT and SCP CD, we assumed that in regions beyond the 3 × 3 mm² area, GCC thinning preceded the reduction of SCP CD. As in most regions beyond the 3 × 3 mm² area, GCC thinning in HM eyes with AL 26–28 mm occurred without affecting the SCP CD. Only HM eyes with AL > 28 mm demonstrated a significant decrease in overall SCP CD. The sequential relationship between thinner GCC and lower SCP CD remains controversial, and the exact sequence must be investigated further because of the lack of longitudinal data in our study. 

This study has certain limitations. First, this was not a longitudinal investigation. Therefore its conclusions were based on the assumption that the data obtained in the cross-sectional analysis would reflect longitudinal development. Only a prospective follow-up study can conclusively address the sequence of GCC thinning and SCP CD reduction. Second, all participants were Chinese, limiting our findings' generalizability. 

In conclusion, our study revealed that peripheral GCCT beyond the 3 × 3 mm² area and RCD beyond the 6 × 6 mm² area generally showed marked changes with axial elongation. Moreover, in HM eyes with AL < 28 mm, a significant reduction in GCCT could be observed only beyond the 3 × 3 mm² area, and a significant reduction in RCD could be observed only beyond the 6 × 6 mm² area. In contrast, GCCT and RCD remained unchanged in most central regions in HM eyes with AL < 28 mm. Decreases in GCCT and RCD could be seen in most regions in HM eyes with AL > 28 mm. Therefore it is necessary to evaluate different regions during the clinical assessment of the effect of myopia on the fundus and pay close attention to the peripheral retina. We hope that our study will contribute to understanding the role of the peripheral retinal microvasculature in the pathophysiological mechanisms of myopia and facilitate further advances in the clinical management of myopia. 

Acknowledgments

The authors thank SVision Company (Shanghai, China) for its support and for providing the imaging software, as well as all the technicians and clinical research collaborators involved in this study. The authors also thank the staff members who have continually supported this work. We are particularly appreciative for Yu Chen and Xiaohong Liang from Zhongshan Ophthalmic Center for their assistance in data collection and Xin Li from SVision for providing essential technical support. 

Supported by the National Key Research and Development Program of China (2022YFC2502805); the High-level Hospital Construction Project, Zhongshan Ophthalmic Center, Sun Yat-Sen University (303020104), Guangdong Natural Science Foundation (2021A151501173), and the National Natural Science Foundation of China (82070955). The funding organizations did not participate in the design or conduct of the study. 

Disclosure: D. Wang, None; Y. Zhang, None; F. Lin, None; Y. Song, None; L. Jin, None; Z. Wang, None; M. Chen, None; J. Jiang, None; P. Wang, None; Z. Yang, None; S. Yuan, None; X. Zhang, None