Inner Retinal Microvasculature With Refraction in Juvenile Rhesus Monkeys

Abstract

Purpose: To characterize inner retinal microvasculature of rhesus monkeys with a range of refractive errors using optical coherence tomography angiography.

Method: Refractive error was induced in right eyes of 18 rhesus monkeys. At 327 to 347 days of age, axial length and spherical equivalent refraction (SER) were measured, and optical coherence tomography and optical coherence tomography angiography scans (Spectralis, Heidelberg) were collected. Magnification-corrected metrics included foveal avascular zone area and perfusion density, fractal dimension, and lacunarity of the superficial vascular complex (SVC) and deep vascular complex (DVC) in the central 1-mm diameter and 1.0- to 1.5-mm, 1.5- to 2.0-mm, and 2.0- to 2.5-mm annuli. Pearson correlations were used to explore relationships.

Results: The mean SER and axial length were 0.78 ± 4.02 D (−7.12 to +7.13 D) and 17.96 ± 1.08 mm (16.41 to 19.93 mm), respectively. The foveal avascular zone area and SVC perfusion density were correlated with retinal thickness for the central 1 mm (P < 0.05). SVC perfusion density of 2.0- to 2.5-mm annulus decreased with increasing axial length (P < 0.001). SVC and DVC fractal dimensions of 2.0- to 2.5-mm were correlated with axial length and SER, and DVC lacunarity of 1.5- to 2.0-mm annulus was correlated with axial length (P < 0.05).

Conclusions: Several inner retinal microvasculature parameters were associated with increasing axial length and SER in juvenile rhesus monkeys. These findings suggest that changes in retinal microvasculature could be indicators of refractive error development.

Translational Relevance: In juvenile rhesus monkeys, increasing myopic refraction and axial length are associated with alterations in the inner retinal microvasculature, which may have implications in myopia-related changes in humans.

Introduction

Visual feedback and local retinal control of eye growth and refractive development have been demonstrated in a wide variety of animals, including fish, chicken, mice, guinea pigs, tree shrews, and marmoset and rhesus monkeys.^1,2 These experimental animal models have contributed to the current understanding of emmetropization and myopia.³ Animal models also continue to provide insight on the cellular and molecular mechanisms of ocular growth control. Most ocular components, including axial length, vitreous chamber depth, and retinal and choroidal thickness, have been studied to understand their contributions in the emmetropization process.^2,4,5 However, less is known about how the inner retinal vasculature changes during emmetropization and in myopia. 

Rhesus monkeys have been used frequently in myopia research.³ The rhesus monkey is an attractive translational model to study human eye diseases and to aid in the development of new therapies. The rhesus monkey eye bears a remarkable level of anatomical and developmental similarity to the human eye.⁶ Whereas some animal models of myopia, such as the chicken and guinea pig, have an avascular retina,^7,8 the rhesus monkey has similar retinal vascular anatomy to humans. Specifically, rhesus monkeys and humans both have inner retinal vasculature originating from the central retinal artery and laminated into superficial and deep vascular plexi, with an avascular zone at the fovea.⁹ Additionally, ocular development in humans and rhesus monkeys proceeds in a similar manner, once scaled for age.⁶ The rate of aging in rhesus monkey is considered to be three times that of humans.¹⁰ Rhesus monkeys are categorized as infants when they are <12 months old, juveniles between 12 and 36 months, adolescents from 3 to 8 years, and adults >8 years old. 

Experimental myopia in monkeys typically uses diffusers (form deprivation) or negative lenses (lens-induced defocus) to alter the image on the retina, inducing axial elongation and refractive error development.^11,12 Similarly, experimental hyperopia can be achieved using positive lenses, resulting in a decrease in axial elongation.^11,13 Experimental myopia in monkeys is accompanied by choroidal thinning, and the opposite occurs in hyperopia.^3,14 Overall retinal thickness has been shown to remain unaffected in marmosets with experimental myopia and hyperopia.^15,16 However, experimental myopia in marmosets is accompanied by thinning of the inner retina (retinal nerve fiber layer and ganglion cell inner plexiform layer thinning) and reorganization of retinal vasculature including increased number of parafoveal string vessels and lower peripheral vessel branching.^15,17 Substantial differences in the retinal gene expression in response to experimental myopia and hyperopia in marmosets have been reported, indicating stronger responses in hyperopia than myopia.¹⁸ 

The introduction of optical coherence tomography angiography (OCTA), a functional extension of OCT, provides in vivo, noninvasive, and three-dimensional visualization of the retinal microvasculature. Previous techniques required either intravenous injection of fluorescein or enucleation with in vitro staining methods. OCTA has an additional advantage of providing three-dimensional information, so that the superficial vascular complex (SVC) and deep vascular complex (DVC) can be isolated. OCTA has been used previously in humans, across a range of refractive errors, including myopia, hyperopia, and emmetropia, to study retinal and choroidal vasculature at different depths.^19–23 Studies in adults have found that superficial and deep retinal perfusion decreases significantly with myopia and axial length.^19–21,24 In contrast, studies in children have shown mixed results related to perfusion and axial length; some suggest an increase,^25,26 a decrease,²⁴ or no association^27,28 of retinal perfusion with axial length. Studies have also found significantly higher retinal perfusion in hyperopes compared with myopes among children.^29,30 However, few studies have evaluated retinal vasculature in children <3 years of age. One study characterized inner retinal vasculature in a population of children ages 9 weeks to 17 years of age.²⁵ Although the authors did not report refractive error, they found that the DVC, but not the foveal avascular zone (FAZ), varied significantly with age and axial length. 

To date, there are only two published studies that have used OCTA to study the retinal microvasculature in rhesus monkeys. These studies included adolescent and adult monkeys that had completed emmetropization.^31,32 No studies have investigated retinal microvasculature in younger juvenile monkeys that are still in the emmetropization period or with a range of experimentally induced refractive errors. This study aimed to characterize retinal microvasculature of juvenile rhesus monkeys with a range of experimentally induced refractive errors using OCTA. Given the similarities between the human and monkey retina, exploring inner the retinal microvasculature with a range of refractive errors in juvenile rhesus monkeys will contribute to a better understanding of retinal vasculature changes in children and myopia development and progression. 

Methods

Animals

Eighteen rhesus monkeys (Macaca mulatta), which were part of three different previously published experimental protocols,^33–35 were included in this analysis. All of these monkeys were reared under dim illumination as part of experiments aiming to induce a wide range of refractive errors. Experimental procedures were approved by the University of Houston's Institutional Animal Care and Use Committee and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Monkeys were acquired at approximately 2 weeks of age and housed in a climate-controlled nursery. The nursery was illuminated on a 12-hour light/12-hour dark cycle using white fluorescent lights (GE Ecolux Starcoat F32T8/SP35/ECO, General Electric Co., Boston, MA). During the daily light phase (7 am to 7 pm), the mean ± standard deviation illumination measured at the junction between the upper and lower cages was 504 ± 168 lux (range, 312–860 lux) with a correlated color temperature of 3170K. 

Experiments commenced at approximately 24 days of age, at which time the monkeys were transferred to another nursery room with reduced (dim) illumination as a part of experiments specifically aiming to induce myopia using dim illumination. Dim illumination was maintained throughout the light phase of the diurnal lighting cycle (7 am to 7 pm) and achieved by closely attaching an aluminium-deposited, polyester film to the fluorescent lighting panels (Grafix Metalized Dura-Lar, Silver, 0.05-mm thick; Grafix, Maple Heights, OH). The average illumination recorded at the junction between the upper and lower cages was 55 ± 9 lux. In the front of individual cages, with the sensor oriented horizontally toward the outside of the cage, illumination varied from 7 to 36 lux. 

Among the 18 monkeys, 4 were reared under dim illumination with unrestricted vision,³³ 4 were reared under dim illumination with concurrent monocular form deprivation,³⁴ and 10 were reared under dim illumination with concurrent monocular lens treatment.³⁵ For monocular form deprivation, a diffuser lens consisting of a light-perception Bangerter occlusion foil (Fresnel Prism and Lens Co., Bloomington, MN) was attached to plano carrier lenses and mounted in front of the right eye using a light-weight helmet. For lens-induced refractive errors, monkeys were fitted with helmets that incorporated either a −3 D (n = 6) or +3 D (n = 4) lens in front of the right eye. For both the form deprivation and lens-induced refractive error groups, a plano lens was mounted in front of the left eye. The helmet was removed at 154 ± 7 days of age; at this point, monkeys had unrestricted vision and underwent recovery from form deprivation or lens-induced myopia. All monkeys were maintained in dim illumination with unrestricted vision in both eyes until 310 ± 21 days of age, at which time they returned to normal colony lighting. 

Ocular Measurements

All measurements were performed between 9:00 am and 12:00 pm to minimize the influence of diurnal variation. Refractive error, OCT scans, and axial length measurements were collected every 2 weeks between 24 ± 2 days and 327 ± 10 days of age. Refractive error, axial length, and OCT scans collected at 327 ± 10 days were used for further analysis in the present study. OCTA scans were captured for right eyes at 347 ± 13 days of age (range, 329–371 days). For OCTA measurements, monkeys were anesthetized using an intramuscular injection of 20 to 25 mg/kg ketamine hydrochloride, combined with 0.4 to 0.6 mg/kg xylazine. Eyes were dilated and cyclopleged with 1% tropicamide instilled 25 and 20 minutes before the measurements. Refractive error and ocular dimension measurement protocols have been discussed in detail previously.^33–35 In brief, refractive error was determined by two experienced examiners using retinoscopy and reported as the mean spherical-equivalent of the spectacle-plane refractive correction (spherical equivalent refraction [SER]) for a vertex distance of 14 mm. Axial length was measured with A-scan ultrasonography along the normal to the cornea apex with a 13-MHz transducer (OTI-Scan 1000, Ophthalmic Technologies Inc., Downsview, Ontario, Canada), from the anterior cornea apex to the internal limiting membrane. Ten independent readings were obtained and averaged for each eye. 

Retinal thickness was measured using spectral domain OCT (Spectralis, Heidelberg, Germany). Two high quality (signal strength >30) images consisting of 31 horizontal scans centered on the fovea were captured for the right eye, as described previously.³⁶ Raw data (*.vol files) were exported and analyzed using custom semi-automated MATLAB software (MathWorks, Natick, MA) and were adjusted for individual lateral magnification.³⁷ The retinal pigment epithelium and inner limiting membrane were segmented automatically and corrected manually, when necessary, by an experienced observer. The center of the fovea was identified as the deepest point in the foveal pit, and the foveal thickness was calculated as the axial distance between the posterior boundary of the retinal pigment epithelium and the inner limiting membrane. Total retinal thickness along the horizontal meridian was binned into four zones: central 1.0-mm diameter and 1.0- to 1.5-mm, 1.5- to 2.0-mm, and 2.0- to 2.5-mm regions. 

For OCTA, one 15° × 15° macular scan was captured (768 A-scans per B-scan, and 384 B-scans per volume). The scan was repeated if there were motion artifacts, if signal strength was <30 dB or if the scan was not centered on the fovea.³⁸ The monkeys had a large range of axial lengths. The captured OCTA images were scaled to account for differences in lateral magnification.^37,39 

The SVC and DVC images were exported from the OCTA scans. The SVC is located between the internal limiting membrane and inner plexiform layer, and the DVC is located between the inner nuclear layer and outer plexiform layer.⁴⁰ DVC images were exported after projection artifact removal. Images were analysed using a previously described custom image analysis MATLAB program and ImageJ to extract parameters of interest, including the FAZ area, perfusion density, fractal dimension, and lacunarity.^41,42 The SVC and DVC images were then adjusted for magnification,⁴² followed by binarization and skeletonization. 

For perfusion density analyses, the SVC and DVC images were first binarized using the Otsu auto thresholding method.^43,44 Binarization involves assigning the OCTA image pixels to either black or white. The binarized images were then used to determine perfusion density, as the ratio of white pixels to the total number of pixels in the region of interest. Four circles at 1.0 mm, 1.5 mm, 2.0 mm, and 2.5 mm (Fig. 1) were centered on the fovea to extract perfusion density from the central 1.0-mm region and three annuli of the SVC and DVC images. The FAZ margin was traced on the DVC images, once corrected for magnification, using ImageJ, and the FAZ area and perimeter were extracted.⁴⁵ 

Figure 1.

 

Representative OCTA en face images of the (A) SVC (B) DVC with four circles overlayed at 1.0 mm (blue), 1.5 mm (green), 2.0 mm (red), and 2.5 mm (yellow) diameters centered on the fovea. The FAZ is indicated in green (B) as the filled area in the center in the DVC.

Retinal microvasculature of the SVC and DVC was also characterized by fractal dimension and mean lacunarity in the central 1.0-mm diameter and 1.0- to 1.5-mm, 1.5- to 2.0-mm, and 2.0- to 2.5-mm annuli using the box counting method with FracLac plugin on ImageJ.^46–50 This process requires skeletonizing the binarized OCTA images, that is, reducing all blood vessels into a single pixel width. The skeletonized images were then used to assess the complexity of the vascular networks.⁴⁶ The fractal dimension in a two-dimensional image represents the level of complexity of the branching vascular network, ranging between 1 and 2.⁴⁸ The box counting method overlays the skeletonized OCTA image with square boxes of side length (L) in pixels. The number of boxes (N) that contain parts of the blood vessels is counted, denoted as N(L). The box size is then incrementally increased, and the process is repeated. The number of boxes required to cover the vasculature structures decreases as box size increases. The relationship between box size (L) and the corresponding count N(L) is analyzed using linear least squares regression, to determine the fractal dimension, expressed as a ratio.⁴⁹ Higher values of fractal dimension correspond with a more complex vascular network. 

Lacunarity is a measure of structural nonuniformity and provides information on the distribution and size of gaps between blood vessels. Lacunarity also uses the box counting method and is expressed as a ratio, ranging between 0 and 1. The variance of pixel intensities within boxes of varying sizes overlaying the skeletonized OCTA image is calculated. Lacunarity is then derived from the relationship between box size and variance.⁵¹ Images with greater lacunarity are more heterogeneous in nature and have more irregular distribution of gaps between blood vessels. A value of 0 represents a completely homogeneous zone.^52,53 

Statistical Analysis

Statistical analysis was performed using SPSS (version 29; IBM Corp, Armonk, NY) and Excel (version 2404; Microsoft Corporation, Redmond, WA). Descriptive statistics were calculated for all parameters and provided as mean ± standard deviation unless otherwise noted. OCTA parameters between depths and zones were compared using two-way repeated measures analysis of variance with Bonferroni-adjusted post hoc pairwise comparisons. Pearson correlation was used to study the relationship of OCTA parameters with refraction, axial length, and retinal thickness. A P value of < 0.05 was considered significant. 

Results

Figure 2 shows changes in axial length and SER from 24 ± 2 days to 327 ± 10 days for the monkeys. The mean SER and axial length on the day of helmet removal, age 154 ± 7 days, for right eyes was +0.25 ± 3.60 D (range, −7.81 to +5.94 D) and 17.17 ± 0.86 mm (range, 15.68–18.80 mm), respectively. At 327 ± 10 days (the final refraction and biometric measurement), mean SER and axial length for right eyes was +0.78 ± 4.02 D (range, −7.12 to +7.13 D) and 17.96 ± 1.08 mm (range, 16.41–19.93 mm), respectively. Axial length was significantly correlated with SER (R = −0.78; P < 0.001). The mean FAZ area and perimeter, perfusion density, fractal dimension, and lacunarity of the SVC and DVC and retinal thickness in various zones are shown in Table 1. 

Figure 2.

 

(A) Spherical equivalent refraction (D) and (B) axial length (mm) for individual monkeys from ages 24 ± 2 days to 327 ± 10 days (N = 18). Monkeys were reared under four different paradigms, all under dim illumination: (1) unrestricted vision (N = 4, violet), (2) form deprivation (N = 4, red), (3) positive defocus (N = 4, green), and (4) negative defocus (N = 6, blue). The vertical dashed line indicates helmet removal, and the shaded gray area represents the age when measurements for the current study were collected.

Table 1.

 

OCT and OCTA Parameters (N = 18)

 

Figure 3 shows the perfusion density, fractal dimension, and lacunarity of SVC and DVC for each zone. Repeated measures analysis of variance showed that all three OCTA parameters significantly differ by depth and zone (P < 0.001 for all). Additionally, there was a significant interaction between depth and zone (P < 0.001 for all). 

Figure 3.

 

(A) Perfusion density, (B) fractal dimension, and (C) lacunarity of the SVC (open bars) and DVC (filled bars) for each zone, expressed as ratios. Error bars represent standard error of mean. Significant differences between SVC and DVC from repeated measures analysis of variance Bonferroni post hoc pairwise comparisons are indicated by ***P < 0.001 and *P < 0.05.

Table 2 depicts the significance values for Bonferroni adjusted pairwise comparison for perfusion density, fractal dimension, and lacunarity between zones in the two layers. Both SVC and DVC perfusion density increased with eccentricity from central 1.0-mm diameter to 1.5- to 2.0-mm annuli (P < 0.001 for both), with no difference in perfusion density between 1.5- to 2.0-mm and 2.0- to 2.5-mm annuli in the SVC (P = 0.23) and decrease in perfusion density between 1.5- to 2.0-mm and 2 to 2.5 annuli in the DVC (P = 0.004).The perfusion density in the central 1.0-mm diameter and 1.0- to 1.5-mm, 1.5- to 2.0-mm, and 2.0- to 2.5-mm annuli was significantly higher in the DVC than the SVC (P < 0.001 for all). 

Table 2.

 

Significance Values for Bonferroni-adjusted Pairwise Comparison Between Zones for SVC and DVC

 

Fractal dimension increased between the central 1.0-mm diameter and 1.0- to 1.5-mm annulus (P < 0.001) and then decreased in 1.5- to 2.0-mm (P < 0.001) of SVC. In the DVC, the fractal dimension showed no difference between central 1.0-mm diameter and 1.0- to 1.5-mm diameter (P = 0.09) and then gradually decreased with eccentricity (P < 0.001). Fractal dimension was significantly higher in the DVC than SVC in all four zones (P < 0.001 for all). 

Lacunarity did not change significantly from central 1.0-mm diameter to 1.0- to 1.5-mm (P = 0.33) and to 1.5- to 2.0-mm annuli (P > 0.99), but showed an increase in the 2.0- to 2.5-mm annuli of the SVC (P = 0.004). Lacunarity gradually increased with eccentricity in the DVC (P < 0.001). Lacunarity was significantly higher in the SVC than DVC in all four zones (P < 0.001 for the central 1.0-mm diameter, 1.0- to 1.5-mm and 1.5- to 2.0-mm annuli, and P = 0.02 for the 2.0- to 2.5-mm annulus). 

The FAZ area was correlated with the retinal thickness of the central 1.0-mm diameter (R = −0.53; P = 0.02). SVC perfusion density decreased with increased axial length for the 2.0- to 2.5-mm annulus (R = −0.55; P = 0.02), but not for the other zones (central 1.0-mm: R = −0.39, P = 0.12; 1–1.5 mm: R = −0.40, P = 0.10; 1.5–2 mm: R = −0.46, P = 0.05). SVC perfusion density and SER were not correlated for any zone (P > 0.05 for all). The FAZ area and perimeter and DVC perfusion density for any zones did not exhibit any statistically significant correlations with SER or axial length (P > 0.05 for all). 

The fractal dimension of the SVC and DVC in 2.0- to 2.5-mm annulus decreased with increase in axial length (SVC: R = −0.87, P < 0.001; DVC: R = −0.92, P < 0.001) and increase in hyperopic refraction (SVC: R = 0.69, P < 0.001 DVC: R = 0.74, P < 0.001). In the remaining zones of the two layers, fractal dimension was only correlated with axial length in the 1.0- to 1.5-mm annulus of the DVC (R = 0.58, P = 0.01) and with SER in the 1.5- to 2.0-mm annulus of the SVC (R = 0.47, P < 0.05). 

A statistically significant correlation was found between axial length and lacunarity of the 1.5- to 2.0-mm annulus of the DVC (R = −0.48; P = 0.04), but not in the other DVC zones or SVC (P > 0.05 for all). Lacunarity and SER were not correlated in either of the zones of SVC and DVC (P > 0.05 for all). 

An increase in retinal thickness with an increase in perfusion density was found only in the central 1.0-mm diameter of SVC (R = 0.54; P < 0.01), but not in DVC (R = 0.42; P = 0.08). Retinal thickness was not correlated with fractal dimension or lacunarity of either the SVC or DVC in any zone (P > 0.05 for all). SER showed statistically significant correlations with retinal thickness in the 1.0- to 1.5-mm, 1.5- to 2.0-mm, and 2.0- to 2.5-mm annuli (R = 0.64, P < 0.01; R = 0.67, P < 0.01; and R = 0.65, P < 0.01, respectively), but not in the central 1.0-mm diameter (R = 0.46, P = 0.06). Retinal thickness and axial length were only correlated in the 1.5- to 2.0-mm annulus (R = −0.48; P < 0.05). 

Discussion

This study characterized the inner retinal microvasculature of juvenile rhesus monkeys with a range of refractive errors using OCTA, including the FAZ, perfusion density, fractal dimension, and lacunarity in the macular region. The perfusion density, fractal dimension, and lacunarity differed significantly by depth and zone. The OCTA parameters were also examined with respect to refractive error and axial length. Findings show that increasing axial length is associated with a decrease in perfusion density of the SVC, but not the DVC. In addition, results show that for the superficial and deep vascular network, fractal dimension (complexity) decreased with increasing axial length and myopia whereas lacunarity (deep vascular network heterogeneity) decreased only with increasing axial length. 

OCTA produces noninvasive high-resolution angiograms of the retina at different depths, making it valuable to study in vivo the microvasculature characteristics of the retina. In the past decade, OCTA has been used in adults and children to investigate retinal perfusion.^{19–21,24,26} The present study quantified the perfusion, complexity, and homogeneity of the retinal microvasculature in juvenile rhesus monkeys with a range of refractive errors. The mean perfusion density of the SVC and DVC at various eccentricities ranged between 13% and 52%, similar that reported in human studies.^28,54,55 

Studies in animal models have used OCTA to examine various aspects of retinal perfusion. A recent study captured wide field images of the retinal microvasculature in seven common lab species, including mouse, rat, pig, rabbit, guinea pig, chicken, and cynomolgus macaque monkeys.⁵⁶ The cynomolgus macaque exhibited highest level of perfusion and fractal dimension, whereas rabbits exhibited the lowest perfusion and fractal dimension. Another study compared the perfusion density from OCTA in adult cynomolgus macaques (4.91 ± 0.43 years) and humans (25.11 ± 6.21 years).³¹ The perfusion density in both the SVC and DVC for the foveal (central 1 mm) and parafoveal (1- to 2-mm annulus) regions were significantly higher in humans (SVC: 16.15 ± 5.02% and 43.73 ± 4.49%; DVC: 30.13 ± 8.25 and 53.11 ± 5.99%, respectively) than the cynomolgus macaques (SVC: 7.38 ± 3.80% and 38.72 ± 3.38%; DVC: 23.99 ± 4.65% and 58.02 ± 4.87%, respectively). 

Li et al.³² reported perfusion in adult healthy rhesus monkey (mean age, 6.5 years; range, 5.5–7.5 years) in terms of flow index, which is the average decorrelation value and related to blood flow velocity. They found a higher flow index of the SVC (0.044 ± 0.011) compared with the DVC (0.036 ± 0.011). The present study did not measure the flow index, but rather perfusion density, which is a different measure and denotes the area occupied by blood vessels; however, both metrics represent a direct or indirect measure of blood flow. In general, both a high flow index and perfusion density imply an increase in blood flow; however, this may not always be true. The present study in juvenile rhesus monkeys showed higher perfusion density in the DVC compared with the SVC. Considering the rapid increase in retinal thickness in rhesus monkeys between infancy and adulthood,⁵⁷ changes in retinal perfusion in the different layers over a span of 5 years may occur. A study in 4-year-old children,⁵⁸ comparably similar in age with the juvenile monkeys in the present study, also showed higher perfusion in the DVC compared with the SVC. 

In a previous study, the FAZ area in adult rhesus monkeys was quantified in vitro on ADPase-stained retinas and was found to be 0.21 ± 0.11 mm².⁹ Here, in juvenile monkeys using in vivo OCTA, we found the FAZ area to be 0.09 ± 0.02 mm², suggesting that the area may increase with advancing age and growing eye size. Chen et al.⁵⁷ investigated inner retinal structures in juvenile (approximately 1 year old) and adolescent (approximately 4 years old) rhesus monkeys using in vitro techniques and concluded that the structures are not mature until adolescence. Interestingly, the area reported previously in adult monkeys is similar to that reported in human adults (0.24 mm²; interquartile range, 0.18–0.32 mm²).⁵⁹ 

Comparatively, the SVC network was less complex and more heterogeneous than the DVC network in juvenile rhesus monkeys examined here, which could be attributed to anatomical differences between the two layers.^60–62 For example, greater heterogeneity could be attributed to the presence of both larger and smaller blood vessels in the superficial layer, creating a more irregular distribution of gaps than in the deep layer. In contrast, the deep layer is composed of numerous small radial and horizontal interconnections of vessels, creating a more complex vascular geometry than the superficial layer, similar to what has been reported in humans.^63–65 Both superficial and deep vascular perfusion densities increased significantly with eccentricity in the macular regions (from the center of the fovea to the 2-mm annulus) similar to previous human studies,⁶⁶ suggesting the two layers are not a homogenous network of blood vessels. Similarly, the fractal dimension and lacunarity beyond the FAZ (outside of the central 1 mm) of the two layers gradually decreased and increased toward the periphery, respectively, indicating less complexity and homogeneity in the outer annuli. The characteristics of the vasculature in the central 1 mm is largely influenced by the FAZ, for instance, an increase in FAZ area is associated with a decrease in perfusion in the central 1.0-mm region.⁶⁷ It is speculated that the heterogenous nature of the retinal metabolic demand could be a reason for the regional variations in the vascular network.⁶² 

It has been reported that myopic refraction and longer axial length are associated with narrower and straighter retinal vessels and diminished complexity of overall retinal vascular network, which may in turn affect ocular blood flow.^68–71 In contrast, eyes with hyperopic refraction and shorter axial lengths exhibit a more complex vascular network.⁶⁹ In the present study, increasing axial length was found to be associated with decreased perfusion density of the superficial layer in the 2.0 to 2.5 mm, but was not associated with perfusion density of the deep layer. Conflicting results have been reported on the association of myopia and axial length with perfusion of the superficial and deep vascular capillaries in children, with some studies reporting a significant association in the superficial layer only,^24,72 in deep layer only,^29,73 or in both the superficial and deep layers.^25,30 It is likely that the contrasting vascular morphology of the superficial and deep layers can contribute to the varying results.⁶⁰ The mechanism behind why axial length might affect the two retinal layers differently is not well-understood. With increasing axial length and change from hyperopic to myopic refraction, it is speculated that retinal vessels may spread over a larger area due to stretching of the retina. Stretching of the retina may compromise the vasculature, resulting in the reduction of perfusion density.²⁰ 

It has been shown that a healthy retinal vascular network is generally more homogenous, with regular distribution of gaps between vessels, than unhealthy retinas.⁵¹ In adults, fractal dimension has been shown to reduce with increased axial length and myopic refraction, in both the SVC and DVC.^21,64,74 Hyperopes have been shown to have higher fractal dimension of retinal vascular structure than myopes in adults.⁷⁰ However, information on lacunarity and fractal dimension with axial length and refractive error in children is lacking. In the present study, reduced fractal dimension was significantly associated with longer axial lengths and myopic refraction in the 2.0- to 2.5-mm annulus of both SVC and DVC. However, fractal dimension in the 1.0- to 1.5-mm annulus of the DVC increased with increasing axial length. Increasing axial length was also associated with decreased lacunarity of DVC, but not of the SVC. These findings may reflect continued changes in the microvascular geometry of the deep retina with axial eye growth of the 1-year-old monkeys. 

The FAZ area and SVC perfusion density decreased with increasing retinal thickness in central 1.0-mm region, similar to findings to children.^26,75 Other OCTA parameters investigated here, including perfusion in the other zones, fractal dimension, and lacunarity, were not related to retinal thickness. It was also observed that retinal thickness exhibited a significant correlation with refraction and axial length. A longitudinal study in young marmosets, ages 2 to 5 months, also reported that the retina thinned in eyes undergoing myopic axial elongation.¹⁵ 

As demonstrated here, the inner retinal microvasculature of juvenile rhesus monkeys is relatively similar to that of humans in terms of vascular geometry and perfusion density. However, there is a lack of information on how early refractive development in humans, that is, <4 years of age, affects the retinal microvasculature. Although the juvenile monkeys in the present study did not undergo normal emmetropization, the findings provide some understanding of how axial eye growth and changes in refraction affect the microvasculature, which may be translatable to children.⁷⁶ The present study captured retinal perfusion in juvenile rhesus monkeys, when eyes were still undergoing emmetropization. A recent study reported that associations between myopia and retinal perfusion changes are different in children compared with adults, suggesting that the vasculature continues to undergo remodeling throughout childhood.⁷⁷ Here, changes in retinal perfusion in monkeys undergoing experimental myopia might represent early biomarkers of refraction related changes. Given that this was a cross-sectional study, future studies should explore inner retinal microvasculature longitudinally in infant rhesus monkeys undergoing emmetropization and experimentally induced myopia, which may widen our understanding of retinal perfusion, fractal dimension, and lacunarity as biomarkers of refractive change in humans. 

Limitations of the present study include the following. The analysis was limited to the inner retinal vasculature, including the SVC and DVC. Future studies could also explore the characteristics of the outer retinal vasculature, being the choriocapillaris, and deep choroid perfusion with experimentally induced refractive errors., which would add to the existing literature related to blood flow and myopia. Another limitation of this study was that refractive errors in these monkeys were experimentally induced in dim illumination, which was intended to disrupt the normal emmetropization process,³³ and that various paradigms were used to induce refractive error, including positive and negative defocus and form deprivation. However, it is worth noting that the rearing protocols used here provided a wide range of refractive errors, which allowed us to explore associations between refractive error, axial length, and inner retinal microvasculature. Given that all monkeys were undergoing various experimental protocols to provide a range of refractive errors, none of the monkeys included here experienced unrestricted vision and underwent normal emmetropization. It is possible that these experimental conditions may have influenced the retinal microvasculature. Future studies are planned to perform OCTA imaging and ocular measurements in monkeys undergoing emmetropization with unrestricted under normal lighting conditions to expand on results presented here. Additionally, incorporating larger sample sizes would allow further comparison across experimental myopia paradigms and increase the statistical power of the study. Over the course of recovery period (Fig. 2), refractive errors showed a small hyperopic shift of approximately 0.50 D. Axial length showed a gradual increase throughout the helmet wear and recovery periods. Axial length demonstrated an increase of 0.80 mm during recovery period, demonstrating that the eye was still growing and developing. The present study analyzed associations of OCTA parameters with refraction and axial length post recovery. Future studies could aim to investigate OCTA parameters and their association with refraction and axial length longitudinally, throughout the helmet and recovery periods. It should be noted that refraction and ocular biometry measures were captured approximately 20 days before OCTA scans. However, as seen in Figure 2, refraction and biometry did not significantly change over this time period.⁷⁸ 

In conclusion, the present study characterized the inner retinal microvasculature with refractive error and axial length in juvenile rhesus monkeys. Increasing axial length was associated with reduction in SVC perfusion. Increasing axial length and myopic refraction were related to decreasing SVC and DVC complexity, whereas DVC heterogeneity decreased only with increasing axial length. Taken together, these findings may be indicative of alterations in microvasculature surrounding the fovea in myopia. 

Acknowledgments

Supported by NEI R01EY003611, NEI R01EY033743, and NEI P30EY007551. 

Disclosure: B. Lal, None; Z. She, None; K.M. Beach, None; L.-F. Hung, None; N.B. Patel, Heidelberg Engineering (E); E.L. Smith, III, Treehouse Eyes (C), SightGlass Vision, Vision CRC (C); L.A. Ostrin, Zeiss (C), Vyluma (C), Meta, LLC (F)