Relationship Between Macular Ganglion Cell Thickness and Ocular Elongation as Measured by Axial Length and Retinal Artery Position

Abstract

Purpose: We recently reported on the usefulness of retinal artery trajectory in estimating the magnitude of retinal stretch due to myopia. The purpose of the present study was to elucidate the relationship between the peripapillary retinal artery angle (PRAA) and thickness of the macular ganglion cell–inner plexiform layer (GCIPL).

Methods: This r included 138 healthy eyes of 79 subjects older than 20 years of age without any known eye disease. GCIPL thickness was separated into eight sectors according to quadrant and eccentricity from the fovea. The PRAA was calculated as the angle between the superior and inferior retinal arteries. Relationships between whole GCIPL thickness (average and sectorial) and the values of PRAA and axial length (AL) were investigated using a linear mixed model.

Results: Average GCIPL thickness in the whole scanned area decreased significantly with narrowing of the PRAA with and without adjusting for AL. Sectorized macular GCIPL thickness also decreased significantly, with narrowing of the PRAA in seven out of the eight with the adjustment of AL, the exception being the inferior peripheral temporal sector.

Conclusions: Macular GCIPL thickness decreased significantly with narrowing of the PRAA on average and in seven out of eight sectors.

Optical coherence tomography (OCT) has enabled the in vivo observation of retinal layers, which is useful in assessing various pathological statuses of the eye. The macular ganglion cell–inner plexiform layer (GCIPL), the macular retinal nerve fiber layer (RNFL), and the circumpapillary retinal nerve fiber layer (cpRNFL) are elements that are useful to evaluate when determining the degree of glaucomatous retinal change.^1–6 In particular, the assessment of macular GCIPL thickness is important, as they are the cells primarily damaged by the disease; however, inter-individual variations in the thickness of the GCIPL exist related to various ocular factors, such as myopia,^7–9 which can make diagnosing glaucoma using OCT inaccurate.^10–12 Also, myopia is a known risk factor for the development of glaucoma.^13–15 Therefore, it is important to understand the effects of myopia on structural changes, such as those that occur in GCIPL thickness as measured with OCT. 

Previous studies have suggested that macular GCIPL thickness becomes thinner with elongation of the axial length (AL).^16–18 AL is frequently used to estimate the elongation of an eye due to myopia^19,20; however, we recently reported that the magnitude of associated retinal stretch cannot fully be explained by AL alone, probably due to the large variation of AL inherited at birth across individuals. Even when two eyes have identical ALs as adults, if their ALs were different at birth then the degree of elongation must have differed between these eyes during the growth period.²¹ One of the characteristic findings in such eyes is that supra- and infratemporal thick retinal nerve fiber bundles and retinal vessel trajectory shift toward the fovea in a manner associated with myopic retinal deformation.^22–26 Our earlier research revealed that retinal vessel trajectory was significantly correlated with the RNF bundle trajectory (R = 0.92).^22,26 In contrast with AL, the effect of these shifts on the thickness of GCIPL has not yet been investigated; therefore, the aim of the current study was to investigate the relationship between retinal artery angle and macular GCIPL thickness. 

Methods

This study was approved by the research ethics committee of the Graduate School of Medicine and Faculty of Medicine at the University of Tokyo. All patients provided written consent for their information to be stored in the hospital database and to be used for research. This study was performed according to the tenets of the Declaration of Helsinki. 

Participants

Participants were recruited from medical staffs who were working in any of three hospitals (University of Tokyo Hospital, Hiroshima Prefectural Hospital, and Inouye Eye Hospital; n = 56) or subjects who came to each hospital for the purpose of screening for any ocular disease but turned out to be normal based on the results of ophthalmic examinations including slit-lamp biomicroscopy, funduscopy, and intraocular pressure measurements (n = 23). The inclusion criteria for the healthy eyes were as follows: (1) no abnormal findings, except for clinically insignificant senile cataract on biomicroscopy, gonioscopy, and fundoscopy; (2) no history of ocular diseases, such as diabetic retinopathy, that could affect the results of OCT examinations; (3) 20 years of age or older; and (4) intraocular pressure of less than 21 mm Hg. Eyes with anomalous discs and diagnosis of glaucoma were cautiously excluded. The signs of glaucomatous changes were judged comprehensively, such as focal rim notching or generalized rim thinning, large cup-to-disc ratio with cup excavation with or without laminar dot sign, and retinal nerve fiber layer defects with edges at the optic nerve head margin and disc edge hemorrhages, according to the recommendations of the Japan Glaucoma Society Guidelines for Glaucoma.²⁷ Eyes with past ophthalmic surgeries except for uncomplicated cataract surgery were also excluded. 

OCT Measurement

GCIPL thicknesses were obtained using an OCT device (RS 3000; Nidek Co., Ltd., Aichi, Japan). OCT imaging was performed after pupil dilation with combined eye drops of 0.5% tropicamide and 0.5% phenylephrine hydrochloride (Midrin-P; Santen Pharmaceutical Co., Ltd., Osaka, Japan). The raster scan protocol for 30 degrees of visual angle (128 × 512 pixels) was used for all scans, and the data in the central 8.1-mm square were used in the analysis. The magnification effect was corrected according to the manufacturer-provided formula based on Littman's equation,^28,29 using the measured AL value. Data with a signal strength index of greater than 7 were included in the current analysis. Images that were unclear due to eye movements or involuntary blinking were acquired again or carefully excluded. In addition, all of the boundaries were checked, and eyes with segmentation errors were carefully excluded from the study. All 65,536 (128 × 512) A-scan pixels of GCIPL thickness were exported for each eye, and those in the analysis area were divided into 10 × 10 grids, then stratified into eight sectors according to the quadrant and the eccentricity from the fovea (Fig. 1). The data obtained in the left eye were mirror-imaged to those obtained in the right eye for statistical analysis. 

Figure 1.

 

Macular grids and sectors. Macular (8.1-mm square) 10 × 10 grids were stratified into eight sectors according to the quadrant and the eccentricity from the fovea. The left eye was mirror imaged.

The positions of major retinal arteries in the superotemporal and inferotemporal areas were determined by identifying the points where the retinal artery and the 3.4-mm-diameter cpRNFL scan circle overlapped using ImageJ 1.48 (National Institutes of Health, Bethesda, MD, USA). The peripapillary retinal arteries angle (PRAA) was the angle between the superior and inferior retinal arteries (Fig. 2). 

Figure 2.

 

Measurement of the right-eye PRAA. The PRAA was calculated by identifying the angle between the intersecting positions (white dots) of a 3.4-mm-diameter peripapillary scan circle (red) and the supratemporal/infratemporal major retinal arteries.

Statistical Analysis

First, the relationship between AL and PRAA was investigated using a linear mixed model approach in which the random effect was the subject, with and without adjusting for age. Then, the relationship between average GCIPL thickness in the whole scanned area and the values of AL and PRAA were investigated using a multivariate linear mixed model in which the random effect was the subject, with and without adjusting for age. Similarly, at each sector, the relationships between GCIPL thickness (derived from 10 × 10 grids) and the values of AL and PRAA were investigated using a multivariate linear mixed model in which the random effect was the subjects, with and without adjusting for age. The linear mixed model is equivalent to ordinary linear regression in that the model describes the relationship between the predictor variables and a single outcome variable; however, standard linear regression analysis makes the assumption that all observations are independent of each other. In the current study, measurements were nested within subjects and, thus, dependent on each other. Ignoring this grouping of the measurements would result in an underestimation of standard errors of the regression coefficients. The linear mixed model adjusts for the hierarchical structure of the data, modeling in a way in which measurements are grouped within subjects to reduce the possible bias of including both eyes of one patient.^30,31 Benjamini and Hochberg's method was used to adjust for multiple comparisons.³² Statistical significance was set at 0.05. All analyses were performed using R 3.5.2 (R Foundation for Statistical Computing, Vienna, Austria). 

Results

Table 1 presents the demographic details for the subjects. Of the 138 healthy eyes studied, 70 eyes were right eyes and 68 eyes were left eyes. Twenty-three subjects were male, and 56 subjects were female. The mean ± SD age was 42.3 ± 18.4 years. The mean AL was 24.6 ± 1.3 mm, and the mean PRAA was 135.9 ± 20.7°. 

Table 1.

 

Study Subject Demographics

 

Figure 3 shows the relationship between AL and PRAA. Of note, there was a significant relationship between the two values (coefficient = –4.8; P = 0.0021, linear mixed model). This was not largely changed with the adjustment for age (coefficient = –4.7; P = 0.0040). 

Figure 3.

 

Scatterplot of AL and PRAA. There was a significant relationship between AL and PRAA.

The average GCIPL thickness in the whole scanned area was not significantly related with AL (coefficient = –0.37; P = 0.34, linear mixed model) (Fig. 4). This did not change when adjusted for age (coefficient = –0.53; P = 0.19); however, the average GCIPL thickness in the whole scanned area decreased significantly with narrowing of the PRAA (coefficient = 0.050; P < 0.001, linear mixed model) (Fig. 5). A very similar relationship was observed with the adjustment for age (coefficient = 0.052; P < 0.001). Multivariate linear mixed model analysis also revealed that PRAA was significantly related to average GCIPL thickness (coefficient = 0.050; P < 0.001, adjusted for age), but AL was not (coefficient = –0.36; P = 0.35, adjusted for age). 

Figure 4.

 

Scatterplots of AL and average GCIPL thickness in the whole scanned area. There was not a significant relationship between AL and average GCIPL thickness.

Figure 5.

 

Scatterplots of PRAA and average GCIPL thickness in the whole scanned area. There was a significant relationship between PRAA and average GCIPL thickness (linear mixed model).

Tables 2 and 3 show the relationship between GCIPL thickness and the values of AL and PRAA for each sector, using the multivariate linear mixed model, without (Table 2) or with (Table 3) adjustment for age. GCIPL thickness was not significantly related to AL at all sectors (P > 0.05, multivariate linear mixed model with adjustment for multiple comparisons), with and without the adjustment for age. In contrast, GCIPL thickness decreased significantly with the decrease in the PRAA in all sectors, except for sector 7 (P < 0.05, multivariate linear mixed model with adjustment for multiple comparisons), with and without the adjustment for age. 

Table 2.

 

Relationship Between Sectorial GCIPL Thickness and AL and PRAA Values, Not Adjusted for Age

 

Table 3.

 

Relationship Between Sectorial GCIPL Thickness and AL and PRAA Values, Adjusted for Age

 

Discussion

In the current study, the relationships between macular GCIPL thickness and the values of AL and PRAA were investigated in 138 eyes of 79 participants without ocular pathology. The results indicate that macular GCIPL thickness decreased significantly with narrowing of the PRAA in seven out of eight sectors. This finding was not observed for GCIPL thickness and AL. Interestingly, AL was not significantly related to macular GCIPL thickness, either in total area or in each sector, in the study. This finding contrasts with many other previous reports suggesting that GCIPL thickness decreases with elongation of AL.^{7,8,16–18,33} This may be because most of the previous studies did not account for the magnification effect,^7,8,16–18 and more peripheral retina was analyzed in eyes with long AL. Of note, Mwanza et al.⁷ estimated the influence of the magnification effect on the thickness of GCIPL in detail and suggested that this influence was large enough to even reverse the effect of AL on GCIPL thickness from negative to positive. In contrast, a previous study has suggested a significant effect of AL on GCIPL thickness, even after correcting for the magnification effect. Araie et al.³³ investigated this association in a circular retinal area with a diameter of 0.6 mm (corresponding to approximately 2 degrees of visual angle) with correction for the magnification effect and observed that the GCIPL thickness decreased with a decrease in spherical equivalent refractive error. This contradictory result compared with the current study may be attributed to differences in the study settings. For example, the area of analysis was much wider in the current study (8.1-mm square), because the purpose of the study was to investigate the association between PRAA and GCIPL thickness. By comparison, in the study by Araie et al.,³³ the purpose was to evaluate the correlation between retinal sensitivities of the innermost four test points with the Humphrey Field Analyzer (HFA) 24–2 test and corresponding GCIPL thickness in normal eyes. In addition, spherical equivalent refractive error was used in the previous study, whereas the AL value itself was used in the current study. 

In contrast to AL, the current study suggested that there was a significant negative relationship between PRAA and macular GCIPL thickness, agreeing with our previous reports.^22,26,34 In our previous work,²² the distribution of cpRNFL fibers through the thickest peaks of cpRNFL coincided well with that of the retinal artery trajectory (R = 0.92). A clinical merit of utilizing PRAA or retinal artery trajectory versus the cpRNFL peak angle is that, in glaucoma patients, the cpRNFL thickness profile changes during the disease process, whereas the PRAA does not. Indeed, we have previously reported that adjusting the cpRNFL profile using the retinal artery angle significantly improved the structure–function relationship,³⁵ suggesting that it is useful to consider the retinal artery trajectory when assessing the cpRNFL profile in eyes with glaucoma. However, the current study found that macular GCIPL thickness was also significantly decreased with narrowing of the PRAA. This suggests that careful consideration of the PRAA is also key when assessing GCIPL thickness in eyes with glaucoma. In a future study, the effects of PRAA on the structure–function relationship and diagnosis should be investigated in patients with glaucoma. Furthermore, we previously reported the usefulness of applying a random forests machine learning method to macular-grid GCIPL thicknesses when diagnosing early-onset glaucoma.³⁶ It would be of interest, as well, to investigate whether adding PRAA information to this further improves the diagnostic accuracy. 

GCIPL thickness decreased with narrowing of the PRAA in seven sectors but not in one sector in the inferior temporal area (sector 2) (Tables 2 and 3). The reason for this result is not clear, but it does suggest that a different tendency occurs in terms of macular retinal deformation associated with elongation of the eye in the superior and inferior hemiretinae, particularly near the optic disc. In the paper by Hood et al.,³⁷ the RNFL stream was traced and the corresponding angle on the optic disc was identified. The authors suggested that the RNFL from the inferior temporal optic disc angle runs closer to the macula as compared to that from the superior temporal optic disc angle because of the position of the optic disc superior to the macula. We also validated this result with regard to the structure–function relationship,³⁸ as the PRAA superior to the papillomacular bundle (70.6°) was larger than the PRAA inferior to the papillomacular bundle (64.6°) in the current study (data not shown in the Results section). This difference may be related to the various relationships between the PRAA and GCIPL thickness between the superior and inferior hemiretinae. Indeed, peripapillary atrophy and choroidal thinning are other findings that develop in association with the elongation of an eyeball,^39–41 and they are usually predominantly observed in the inferior hemiretina near the optic disc.^42–44 This also suggests that different macular retinal deformation profiles are associated with the elongation of an eye between the superior and inferior hemiretinae. These differing macular retinal deformation profiles between the superior and inferior hemiretinae are important to consider not only when investigating the mechanisms of retinal deformation due to myopia but also when elucidating the pathological mechanisms of glaucoma in myopic eyes, because a visual field defect is usually predominant in the inferior hemifield in myopic glaucomatous eyes.^45–47 It should be noted, however, that the association between GCIPL thickness and PRAA was significant, but not very strong, as suggested by the relatively large P value (from 0.0032 to 0.027). Thus, PRAA has an effect on the GCLIPL thickness, but it does not explain all of the variation of GCIPL thickness in healthy subjects. 

Araie et al.³³ reported a correlation between visual field sensitivity and macular GCIPL thickness in normal eyes. In addition, previous studies have suggested that cpRNFL thickness tended to increase as standard automated perimetric sensitivity increased in normal eyes, although the slope did not reach a level of significance.^3,48 On the other hand, there are conflicting reports suggesting that spatial^49–51 and temporal^50,52 contrast sensitivity is normal or reduced in myopic eyes.⁵³ It would be of interest to investigate the effects on visual function of GCIPL thinning associated with narrowing retinal arteries. 

We previously reported that a narrow PRAA is related to poor damping capacity of an eye as measured with the Ocular Response Analyzer (Reichert Technologies, Depew, NY, USA) or OCULUS Corvis STL (Oculus, Inc., Menlo Park, CA, USA).^54,55 Such damping capacity of an eye has been reported to be associated with the development and progression of glaucoma and other retinal diseases such as angioid streaks.^56–58 We have speculated that eyes are deformed even in daily life activities, such as postural changes, eyelid blinking, ocular pulsatility due to ocular hemodynamics, performing the Valsalva maneuver, and regular eye movement.^59–61 Eyes with poor damping capacity (poor hysteresis) are not able to absorb these external strains, which could contribute to the development of ocular diseases. Such poor damping capacity and thin GCIPLs in eyes with narrow angles could raise the likelihood of developing glaucoma, which may, at least in part, explain why myopia is a risk factor for the development of glaucoma.⁶² 

Recently, another independent determinant of macular layer thickness was reported: disk fovea distance. ^63,64 We investigated the effect of this variable on GCIPL thickness, but we did not find a significant association utilizing whole field and sector-wise analyses (data not shown in Results). This may be due to differences in the populations analyzed; the previous studies either analyzed total retinal thickness (GCIPL in this study)⁶⁴ or excluded myopic eyes with spherical equivalent less than –6.0 diopters.⁶³ 

One of the limitations of the current study was the small number of highly myopic eyes. A more evident tendency than shown in this study could perhaps be observed in a larger population. On the other hand, the current results suggest that thinning of the GCIPL thickness associated with narrowing of the artery angle is significant even in a non-highly myopic cohort. 

In conclusion, macular GCIPL thickness decreased significantly with narrowing of the PRAA in addition to elongation of the AL on average and in seven out of eight sectors. 

Acknowledgments

The authors thank Enago (www.enago.jp) for the English language review. 

Supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (25861618 to HM; 19H01114, 18KK0253, 26462679 to RA; 20768254 to YF; 00768351 to MM) and by the Suzuken Memorial Foundation and Mitsui Life Social Welfare Foundation. 

Disclosure: T. Omoto, None; H. Murata, None; Y. Fujino, None; M. Matsuura, None; T. Fujishiro, None; K. Hirasawa, None; T. Yamashita, None; T. Kanamoto, None; A. Miki, None; Y. Ikeda, None; K. Mori, None; M. Tanito, None; K. Inoue, None; J. Yamagami, None; R. Asaoka, None