Changes in Choroidal Thickness and Choroidal Blood Perfusion in Guinea Pig Myopia

Abstract

Purpose: The purpose of this study was to study changes in choroidal thickness (ChT) and choroidal blood perfusion (ChBP), and the correlation between them, in guinea pig myopia.

Methods: The reliability of optical coherence tomography angiography (OCTA) for measuring ChT and ChBP was verified in guinea pigs, after cervical dislocation (n = 7) or temporal ciliary artery transection (n = 6). Changes in refraction, axial length, ChT, and ChBP were measured during spontaneous myopia (n = 9), monocular form-deprivation myopia (FDM, n = 13), or lens-induced myopia (LIM, n = 14), and after 4 days of recovery from FDM and LIM.

Results: The abolition (by cervical dislocation) or reduction (by temporal ciliary artery transection) of ChBP, and of the associated changes in ChT, were verified by OCTA, thus validating the method of measurement. In the spontaneous myopia group, ChT and ChBP were reduced by 25.2% and 31.9%, respectively. In FDM eyes, mean ± SD ChT and ChBP decreased significantly compared with the untreated fellow eyes (ChT fellow: 76.13 ± 9.34 μm versus 64.76 ± 11.15 μm for FDM; ChBP fellow: 37.87 ± 6.37 × 10³ versus 30.27 ± 6.06 × 10³ for FDM) and increased after 4 days of recovery (ChT: 77.94 ± 12.57 μm; ChBP: 37.41 ± 6.11 × 10³). Effects of LIM were similar to those of FDM. Interocular differences in ChT and ChBP were significantly correlated in each group (FDM: R = 0.71, P < 0.001; LIM: R = 0.53, P < 0.001).

Conclusions: ChT and ChBP were significantly decreased in all three models of guinea pig myopia, and they both increased during recovery. Changes in ChT were positively correlated with changes in ChBP. Therefore, it is possible that the changes of ChT are responsible for the changes of ChBP or vice versa.

Myopia is commonly recognized as an ocular disorder that carries significant risks of visually blinding complications.^1,2 In recent decades, the prevalence and severity of myopia have been on the rise, and it is estimated that by 2050 there will be 4.76 billion people with myopia and 0.94 billion with high myopia.^3–6 Meanwhile, the total cost of myopia correction is also increasing, becoming a relatively large economic burden in urbanized countries.^7–9 

With the drastic increase in the public health impact, as well as the socioeconomic burden of myopia, many researchers have focused on investigating the mechanisms underlying myopia development. Twenty years ago, in a seminal study, Wallman et al. found that choroidal thickness (ChT) in chicks significantly increased and decreased in response to positive and negative lens-induced defocus, causing hyperopic and myopic refractive shifts, respectively.¹⁰ On removal of the imposed negative lens defocus, the choroid of the now myopic eye thickened, moving the retina forward toward the defocused image plane. Such bidirectional growth regulation has stimulated researchers to study the choroid as a target tissue for myopia control, and ChT has been investigated as a surrogate marker for the choroidal response to defocus. 

Many clinical studies have reported dramatic decreases of ChT in highly myopic human eyes, the extent of which was significantly correlated with the severity of myopia.^11–14 In chicks^10,15 and guinea pigs,^16,17 similar changes were also demonstrated in negative lens-induced myopia (LIM) and form deprivation myopia (FDM). As a result, it seemed plausible that ChT gradually decreases during the development of myopia. However, what causes the changes in ChT remains unclear. Researchers have suggested that changes in choroidal blood flow (ChBF),^18,19 lymphatics,²⁰ nonvascular smooth muscle,²¹ and osmotic macromolecules^10,22 play roles in the changes. Considering that the choroid is a highly vascularized structure, capable of rapid changes in blood flow, we speculate that changes in ChBF are most likely responsible for the observed changes in ChT. In humans, a significant decrease of ChBF occurs in highly myopic eyes.^23–25 Importantly, both ChT and ChBF increase after administration of sildenafil, a vasodilator,¹⁸ and they both decrease after intravitreal injection of the antiangiogenic drug, bevacizumab.¹⁹ This suggests that the changes of ChT were due to the changes of ChBF. In chicks,²⁶ ChBF decreased significantly after wearing form deprivation goggles for 14 days. In addition, in our previous study, we found that the expression of hypoxia-inducible factor-1α (HIF-1α), a marker of tissue hypoxia, was upregulated in the scleras of both FDM and LIM in guinea pigs, and that antihypoxia treatments suppressed the progression of myopia.²⁷ Therefore, it is plausible that blood flow in the choroid, being the major source of oxygen for the underlying outer retina and the overlying sclera, accompanies or even causes the observed changes in ChT and plays a vital role in the onset and development of myopia. 

Optical coherence tomography angiography (OCTA) is a new noninvasive high-resolution method for imaging retinal and choroidal blood vessels. It detects contrast differences in OCT signals caused by the movement of red blood cell in the perfused vessels.²⁸ The moving red blood cells act as an intrinsic contrast agent that generates blood flow signals and visualizes the vascular networks without the need for dye injection.²⁹ The movement of red blood cell in the imaged vessels indicates perfusion, whereas the absence of red blood cell movement indicates nonperfusion. For assessing the choroidal blood supply with OCTA, which allows detection of blood movement but not measurement of the velocity of flow, the appropriate descriptive term is choroidal blood perfusion (ChBP). 

Importantly, OCTA can measure ChT and ChBP at the same time, which makes it easy to assess the relationship between these two parameters. However, the limitation of instrument is that it cannot precisely identify the choroidal boundaries in the guinea pigs; consequently, it cannot image the choroidal blood vessel very well. In our study, we analyzed the blood flow signals on B-scan images acquired by OCTA to demonstrate the level of correlations between ChT and ChBP and the corresponding refraction and axial length (AXL) in the guinea pig myopia model. These findings could yield important insights into the mechanism of myopia and thereby improve the design and efficacy of myopia prevention and treatment strategies. 

Materials and Methods

Animals and Experimental Design

This research was approved by the Animal Care and Ethics Committee at the Wenzhou Medical University, and all treatment and care of animals were performed according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. In this study, 3-week-old pigmented guinea pigs (Cavia porcellus, the English short-hair stock, n = 77) were randomly assigned to different groups. The illuminance of the animal facility was approximately 300 lux, provided by straight fluorescent lamps under a daily 12-hour light/12-hour dark cycle (starting at 8:00 AM), and room temperature was maintained at 25°C. The animals were provided standard food and fresh vegetables twice a day. The right eyes of all animals were treated, whereas the left eyes served as untreated fellow controls. All eyes underwent biometric measurements, including refraction by a custom-made eccentric infrared photorefractor,³⁰ AXL (A-scan ultrasonography, AVISO Echograph Class I-Type Bat; Quantel Medical, Clermont-Ferrand, France), and ChT and ChBP (Spectralis HRA + OCT; Heidelberg Engineering, Heidelberg, Germany). Four types of experiments were carried out, as described below, to validate and assess the repeatability of ChT and ChBP measurements in the guinea pig eye and to study the ChT and ChBP changes in guinea pig myopia (Table 1). 

Table 1

 

Experimental Design

 

In experiment 1, the reliability of the Spectralis HRA + OCT in guinea pigs was studied by quantifying the changes of ChT and ChBP in two models of impaired choroidal circulation: a cervical dislocation group (n = 7), in which circulation was stopped completely, and a group in which the temporal ciliary arteries were transected (n = 6), in which circulation in the affected region of the choroid was markedly decreased. First, baseline measurements of ChT and ChBP were taken under general anesthesia, by intraperitoneal injection of a mixture of 60 mg/kg ketamine hydrochloride and 9 mg/kg xylazine hydrochloride. Then, either cervical dislocation or temporal ciliary artery transection was performed. 

Cervical Dislocation

This is the most common method of euthanasia for small laboratory animals. The head of the guinea pig was firmly pressed with thumb and forefinger, and the hind legs were grasped with the other hand and pulled forcefully backward and upward, causing separation between the spinal cord and brain, and the guinea pig to die immediately. 

Temporal Ciliary Artery Transection

The superior temporal bulbar conjunctiva was cut along the limbus, and blunt dissection exposed the superior rectus and lateral rectus muscles. Traction was exerted on them, giving full view and access to the optic nerve and its surroundings. Finally, the ciliary arteries around the temporal side of the optic disc were partially transected. The superior and inferior ciliary arteries might have been slightly affected during the procedure; however, the vessels of the contralateral nasal quadrant remained intact. Five minutes after each procedure, ChT and ChBP were measured again. 

In experiment 2, the guinea pigs were divided into a control group (n = 9), with refractions between +3.00 diopters (D) and +8.00 D, and a spontaneous myopia group (n = 9), with refractions between −3.00 and −8.00 D. Differences in ChT and ChBP between the two groups were compared. 

In experiment 3, ChT and ChBP were measured after 7 days of monocular form deprivation to induce FDM (n = 13),^31,32 and again after removal of the myopia-inducing facemasks followed by 4 days of recovery (R-FDM). The control group (n = 8) did not receive any treatment, and ChT and ChBP were measured on experiment days 7 (control 7 days) and 11 (control 11 days). 

In experiment 4, ChT and ChBP were measured in guinea pigs wearing a −4-D lens to produce LIM (n = 14)³³ or a control plano lens (Plano, n = 11) over the right eyes for 7 days. The lenses were then removed (R-LIM group and R-Plano group, respectively), and after 4 days of recovery following removal of the lens, ChT and ChBP were measured again. 

Biometric Measurements

Eccentric Infrared Photorefractor

Refraction was measured by a custom-made eccentric infrared photorefractor designed by Schaeffel.^30,34 The operational approaches and conditions of the machine were similar to those described earlier.^35,36 Guinea pigs are compliant and cooperative, allowing the measurements to be made without either anesthesia or cycloplegia. Treatment and control groups had baseline refractions between +3.00 and +8.00 D (mean ± SD: +5.62 ± 1.19 D). The spontaneous myopia group consisted of animals with myopic refraction between −3.00 and −8.00 D (−5.64 ± 2.11 D).³⁶ Animals with anisometropia of more than 2 D were excluded from this study. In experiments 3 and 4, we measured the refraction immediately before beginning treatment (baseline), after 7 days of treatment, and after 4 days of recovery. 

Ultrasonography

A-scan ultrasonography was used to measure the ocular AXL. The ultrasound frequency and the conduction velocity in the eyes were the same as previously described.³⁶ Topical anesthesia (0.5% proparacaine hydrochloride; Alcon, Puurs, Belgium) was administered prior to ultrasound measurement. The final ocular biometric values for each eye were the average of 10 measurements. In experiments 3 and 4, AXL was also measured just before beginning treatment (baseline), after 7 days of treatment, and after 4 days of recovery. 

OCTA Image Acquisition

The Spectralis HRA + OCT operates with a central wavelength of 870 nm, with the lateral and axial resolutions at 6 and 3.9 μm, respectively. The enhanced depth imaging mode and the follow-up mode were selected, and the pattern size was 30° × 15°, according to the instrument specifications for humans, and seven B-scans (the maximum that can be set in OCTA mode) were obtained for the final OCT and OCTA images. The locations of the moving red blood cells are indicated in yellow on the OCTA images. According to information provided by the manufacturer, the OCTA system generates a signal strength index or “quality number” that represents the signal/noise ratio in decibels; therefore, a high value indicates a good scan. Only images with quality numbers >30 were selected for further analysis.³⁷ 

Along the horizontal scan (from the temporal quadrant to the nasal quadrant of the right eye or from the nasal quadrant to the temporal quadrant of the left eye) and vertical scan (from the inferior quadrant to the superior quadrant of both eyes), we selected B-scans that passed through the center of the optic disc (Fig. 1A). The B-scans were exported in three different formats: a structural OCT image (Fig. 1B), an OCTA image (Fig. 1C), and the overlay of structural OCT and OCTA images (Fig. 1D). From the yellow ChBP signal points in the choroid (Figs. 1C, 1D), we calculated the total numbers of movement-positive pixels in the defined regions, which provided a semiquantitative measure of blood perfusion. These three images were imported into a custom program (MATLAB R2017a; MathWorks, Natick, MA, USA) that simultaneously measured both ChT and ChBP. 

Figure 1

 

Representative fundus image and the three different forms of B-scans. (A) The fundus image of guinea pig showing the analysis template. O, optic disc. (B) The structural OCT image showing the defined regions of interest: interior choroid surface (green line), exterior choroid surface (blue line), and two concentric circles (red lines): 600 μm (“in,” near the center of the optic disc), 1050 μm (“out,” away from the center of the optic disc); the region of interest in each quadrant is between the boundaries of the choroid layer and the two concentric circles. (C) OCTA image. Yellow points in the defined regions of interest represent the blood perfusion signal points. (D) Overlay of structural OCT and OCTA images.

In MATLAB, the upper and lower boundaries of the choroid layer were manually drawn using spline interpolation. The choroid was defined as extending from the external surface of the retinal pigment epithelium (Fig. 1B, green line) to the internal surface of the sclera (blue line).³⁷ To make measurements consistently in a well-defined location in the fundus of the eye, we chose an area defined with reference to the optic disc. Taking the optic disc as the center, two concentric circles were drawn (Fig. 1B, red lines) with radii of 600 and 1050 μm (Fig. 1B). The procedure was based on the lateral magnification correction described by Howlett et al.³⁸ and Jnawali et al.³⁷ ChTs were measured in each quadrant, along the horizontal and vertical radius lines at the inner circle (marked as “in”; Fig. 1B) and at the outer circle (marked as “out”; Fig. 1B). The average ChT values were calculated for the inner and outer circles, at the eight positions of four quadrants (intersections of red and green lines, Fig. 1A). The defined region of interest in each quadrant was between the boundaries of the choroid layer and within the two concentric circles (Fig. 1B). The mean intensity of signals from the vitreous, which was taken as background noise, served as the threshold for identifying meaningful signals from the choroid. The blood perfusion signal points were those with intensity greater than the threshold criterion in the entire region of interest. The numbers of pixels containing ChBP signals in the region of interest in each quadrant were measured, along with the total number of ChBP pixels of the four quadrants. 

Intra- and Interexaminer Repeatability

The intra- and interexaminer repeatability of the ChT and ChBP measurements of the FDM group (n = 13), control 7 days group (n = 8), R-FDM group (n = 13), and control 11 days group (n = 8) were measured to determine the overall reliability of measurements in this study. To assess intraexaminer repeatability, the images were analyzed two times by the same examiner on two different days. To assess interexaminer repeatability, identical images were analyzed independently by two different examiners. 

Statistical Analysis

We verified that all data in this study followed a normal distribution and that data in all groups had the same variance. Thus, our data are presented as means ± SDs, and we used parametric statistical analyses. Linear correlation analysis was used to analyze both the repeatability of the intraexaminer and interexaminer measurements and the correlations among biometric parameters. Paired t-tests were used to compare pre- and posttreatment ChT and ChBP in the cervical dislocation group. Independent-samples t-tests were used to compare the control and spontaneous myopia groups. Comparisons of pre- and posttreatment ChT and ChBP in temporal and nasal quadrants in the temporal ciliary artery transection group were made by repeated-measures ANOVA, with the time and treatment as the repeated measures. ChT, ChBP, and intergroup differences of refraction and AXL in the FDM and LIM groups were also compared by repeated-measures ANOVA, with groups as factors and with time as the repeated measures. Additionally, when analyzing ChT and ChBP, we also included eye as the factor for repeated measures. When data did not meet the Mauchly's test of sphericity, Greenhouse-Geisser correction was used. Interocular differences were calculated as the value for the treated eye minus the value for the fellow eye. Bonferroni corrections were applied in post hoc analyses. Statistical significance was defined as having P < 0.05. SPSS software (Version 16.0) was used for the statistical analysis. 

Results

Intra- and Interexaminer Repeatability

The intra- and interexaminer repeatabilities of the OCTA data were assessed first to validate the methodology for measuring ChT and ChBP. While using the custom software program, both repeatabilities were high (ChT intraexaminer: R = 0.98, P < 0.001, Fig. 2A; ChT interexaminer: R = 0.97, P < 0.001, Fig. 2B; ChBP intraexaminer: R = 0.99, P < 0.001, Fig. 2C; ChBP interexaminer: R = 0.98, P < 0.001, Fig. 2D). 

Figure 2

 

Intra- and interexaminer repeatabilities of measurement. (A) Intraexaminer repeatability of ChT. (B) Interexaminer repeatability of ChT. (C) Intraexaminer repeatability of ChBP. (D) Interexaminer repeatability of ChBP. First, first reading; Second, second reading; ChT, choroidal thickness; ChBP, choroidal blood perfusion.

Cervical Dislocation and Temporal Ciliary Artery Transection Reduced ChT and ChBP

Atlhough OCTA is widely used for analysis of human eyes, its use to measure the choroid of guinea pigs, especially for ChBP, has not been reported previously. Therefore, cervical dislocation and temporal ciliary artery transection were performed to reduce ChBP and to determine whether OCTA could provide sufficient resolution for measuring the changes in our experiments. 

We found that cervical dislocation and temporal ciliary artery transection reduced ChT and ChBP. The main and interaction effects of temporal ciliary artery transection on ChT and ChBP, as determined by repeated-measures ANOVA, are shown in Supplementary Table S1. ChT was thicker before cervical dislocation (65.42 ± 10.41 μm) than afterward (34.77 ± 10.02 μm, P < 0.001; Table 2). Similarly, temporal ciliary artery transection caused ChT to decrease from 77.08 ± 19.13 to 57.34 ± 22.05 μm (P < 0.01; Table 2). The overlay of OCT and OCTA images before and after cervical dislocation and temporal ciliary artery transection illustrates the reduction of ChBP (Figs. 3A–3D). The specificity of temporal ciliary artery transection was evident on comparing the ChBP of the untreated and treated quadrants. For the nasal quadrant, where ChBP remained uninterrupted, the number of blood perfusion signal points was 9.51 ± 3.19 × 10³, but in the perfusion-interrupted temporal quadrant, it was reduced to 3.16 ± 2.71 × 10³ (P < 0.001; Table 2). In contrast, there were no significant changes in either ChT or ChBP in the nasal quadrant following transection of the temporal ciliary artery (P = 0.99 and 0.66, respectively; Table 2). These results indicated that OCTA could measure reliably the changes of ChT and ChBP, in the guinea pig eye in situ. 

Table 2

 

Effect of Cervical Dislocation and Temporal Ciliary Artery Transection on ChT and ChBP

 

Figure 3

 

The overlay of OCT and OCTA images. (A, B) Comparison of the changes of the overlay images of ChBP pre- (A) and postcervical dislocation (B). (C, D) Comparison of the changes of the overlay images of ChBP pre- (C) and posttemporal ciliary artery transection (D).

ChT and ChBP Changes in Spontaneous Myopia

ChT and ChBP were compared in normal 3-week-old guinea pigs (refraction: +4.89 ± 1.60 D; AXL: 8.10 ± 0.27 mm) and in guinea pigs that developed myopia spontaneously (i.e., without any experimental manipulations; refraction: −5.64 ± 2.11 D; AXL: 8.19 ± 0.13 mm). Both parameters were decreased in the myopic group (control nonmyopia ChT: 67.06 ± 10.26 μm versus myopia ChT: 50.15 ± 9.30 μm, P < 0.01, Fig. 4A; control nonmyopia ChBP: 34.30 ± 5.56 × 10³ versus myopia ChBP: 23.35 ± 5.58 × 10³, P < 0.001, Fig. 4B). 

Figure 4

 

Comparison between control and spontaneous myopia eyes of the average ChT and the total ChBP of the four quadrants. (A) ChT in control and spontaneous myopia eyes. (B) ChBP in control and spontaneous myopia eyes. (C) Correlation between ChT and ChBP in the control and spontaneous myopia eyes. *P < 0.05, **P < 0.01, and ***P < 0.001, from independent samples t-test. ChT, choroidal thickness; ChBP, choroidal blood perfusion.

In each position of each quadrant, except the nasal (out) position, ChT was significantly thinner in the spontaneously myopic eyes than in the controls (Supplementary Figs. S1A–1H). ChT tended to reduce in the nasal (out) position of myopic eyes, but the difference from that in control eyes was not statistically significant (P = 0.12). ChBP in each quadrant was significantly less in spontaneously myopic eyes than in control nonmyopic eyes (Supplementary Figs. S1I–1L). ChT and ChBP values were highly correlated (R = 0.95, P < 0.001; Fig. 4C). 

Changes in Refraction, AXL, ChT, and ChBP in FDM Eyes

There were no significant differences in the baseline refractions and AXLs between the 7 days control and FDM groups. However, the results indicated that form deprivation led to myopia, longer AXLs, thinner ChTs, and lower ChBP. The main and interaction effects of FDM on refraction, AXL, ChT and ChBP, as determined by repeated-measures ANOVA, are shown in Supplementary Tables S2A, S2B. 

Interocular differences in refraction (see actual values in Supplementary Table S3A) for the FDM group increased over the baseline values (baseline: −0.06 ± 0.80 D versus FDM: −3.25 ± 2.01 D, P < 0.001; Fig. 5A) and larger than in the control 7 days group (control 7 days: −0.03 ± 1.18 D versus FDM: −3.25 ± 2.01 D, P < 0.001; Fig. 5A). In the R-FDM group, the differences recovered somewhat (FDM: −3.25 ± 2.01 D versus R-FDM: −1.25 ± 0.89 D, P < 0.01; Fig. 5A), but not back to the control level (control 11 days: −0.69 ± 0.65 D versus R-FDM: −1.25 ± 0.89 D, P < 0.001; Fig. 5A). 

Figure 5

 

Refraction, AXL, ChT, ChBP, and correlations among them, for control and FDM group. (A, B) Comparison of the interocular differences in refraction (A) and AXL (B). Interocular differences were calculated as the value for the treated eye minus the value for the fellow eye. (C, D) Comparison of the averages of ChT and the total ChBP over all four quadrants. (E–H) Correlations of the interocular differences in ChT, ChBP, refraction, and AXL. *P < 0.05, **P < 0.01, and ***P < 0.001, repeated-measures ANOVA with Bonferroni correction. AXL, axial length; ChT, choroidal thickness; ChBP, choroidal blood perfusion.

FDM also induced changes in AXL that were consistent with the changes of refraction (see actual values in Supplementary Table S3B). In the FDM group, interocular differences in AXL increased over baseline (baseline: 0.02 ± 0.02 mm versus FDM: 0.09 ± 0.05 mm, P < 0.001; Fig. 5B) and also larger than in the control 7 days group (control 7 days: 0.02 ± 0.03 mm versus FDM: 0.09 ± 0.05 mm, P < 0.01; Fig. 5B). In the R-FDM group, the increase decreased compared with the FDM group (FDM: 0.09 ± 0.05 mm versus R-FDM: 0.04 ± 0.04 mm, P < 0.01; Fig. 5B). 

In FDM, ChT was significantly thinner in the treated eyes than in the fellow eyes (fellow: 76.13 ± 9.34 μm versus FDM: 64.76 ± 11.15 μm, P < 0.001; Fig. 5C). Four days after removal of the mask, ChT had increased (FDM: 64.76 ± 11.15 μm versus R-FDM: 77.94 ± 12.57 μm, P < 0.001; Fig. 5C); and ChBP was lower in FD-treated eyes than in fellow eyes (fellow: 37.87 ± 6.37 × 10³ versus FDM: 30.27 ± 6.06 × 10³, P < 0.001; Fig. 5D). Furthermore, like ChT, ChBP had increased 4 days after facemask removal (FDM: 30.27 ± 6.06 × 10³ versus R-FDM: 37.41 ± 6.11 × 10³, P < 0.001; Fig. 5D). 

The trend in ChT and ChBP changes in each quadrant was similar to that of overall changes in all four quadrants. Values for fellow and FDM eyes in some quadrants were not significantly different, however, each quadrant showed a tendency of decreased or increased ChT and ChBP, according to the treatment. (Supplementary Figs. S2A–S2L). 

The interocular differences in ChT, ChBP, refraction, and AXL in the control 7 days group, FDM group, control 11 days group, and R-FDM group were all significantly correlated: ChBP and ChT (R = 0.71, P < 0.001; Fig. 5E), AXL and refraction (R = −0.58, P < 0.001; Fig. 5F), ChBP and refraction (R = 0.57, P < 0.001; Fig. 5G), and ChT and refraction (R = 0.46, P < 0.01; Fig. 5H). 

Changes in Refraction, AXL, ChT, and ChBP in LIM Eyes

There were no significant differences in baseline refractions and AXLs between the 7 days Plano and 7 days LIM groups. However, the negative lens treatment led to myopia, longer AXLs, thinner ChTs, and lower ChBP. The main and interaction effects of LIM on ChT and ChBP, as determined by repeated-measures ANOVA, are shown in Supplementary Tables S4A, S4B. 

The interocular differences in refraction (see actual values in Supplementary Table S5A) in the LIM group increased over the baseline value (baseline: −0.02 ± 0.60 D versus LIM: −4.04 ± 1.10 D, P < 0.001; Fig. 6A) and also increased over the Plano group (Plano: −0.73 ± 0.84 D versus LIM: −4.04 ± 1.10 D, P < 0.001; Fig. 6A). In the R-LIM group, the interocular differences partially decreased (LIM: −4.04 ± 1.10 D versus R-LIM: −1.41 ± 0.93 D, P < 0.001; Fig. 6A), but they did not reach the level of those in the R-Plano group (R-Plano: −0.63 ± 0.72 D versus R-LIM: −1.41 ± 0.93 D, P < 0.05; Fig. 6A). 

Figure 6

 

Refraction, AXL, ChT, ChBP, and correlations among them for Plano and LIM group. (A, B) Comparison of the interocular differences in refraction (A) and AXL (B). Interocular differences were calculated as the value for the treated eye minus the value for the fellow eye. (C, D) Comparison of the average of ChT and the total ChBP over all four quadrants. (E–H) Correlations of the interocular differences in ChT, ChBP, refraction, and AXL. *P < 0.05, **P < 0.01, and ***P < 0.001, repeated-measures ANOVA with Bonferroni correction. AXL, axial length; ChT, choroidal thickness; ChBP, choroidal blood perfusion.

The AXL changes (see actual values in Supplementary Table S5B) were consistent with the changes in refraction. In the LIM group, the interocular differences in AXL increased over baseline (baseline: 0.01 ± 0.03 mm versus LIM: 0.10 ± 0.05 mm, P < 0.001; Fig. 6B) and also were larger than the Plano group (Plano: 0.02 ± 0.02 mm versus LIM: 0.10 ± 0.05 mm, P < 0.001; Fig. 6B), but the differences decreased in the R-LIM group (LIM: 0.10 ± 0.05 mm versus R-LIM: 0.03 ± 0.04, P < 0.001; Fig. 6B). 

In LIM eyes, ChT was thinner than that in the fellow eyes (fellow: 72.17 ± 10.04 μm versus LIM: 64.00 ± 9.58 μm, P < 0.001; Fig. 6C), but 4 days after lens removal, ChT had increased (LIM: 64.00 ± 9.58 μm versus R-LIM: 71.99 ± 7.62 μm, P < 0.001; Fig. 6C). ChBP also decreased compared with the fellow eyes (fellow: 41.06 ± 5.32 × 10³ versus LIM: 35.08 ± 5.58 × 10³, P < 0.001; Fig. 6D) and the Plano eyes (Plano: 39.62 ± 5.05 × 10³ versus LIM: 35.08 ± 5.58 × 10³, P < 0.05; Fig. 6D). Like ChT, 4 days after lens removal, ChBP increased (LIM: 35.08 ± 5.58 × 10³ versus R-LIM: 38.36 ± 3.05 × 10³, P < 0.01; Fig. 6D). Additionally, in the Plano group, ChBP in the treated eyes was lower than that in the fellow eyes (fellow: 42.17 ± 6.41 × 10³ versus Plano: 39.62 ± 5.05 × 10³, P < 0.05; Fig. 6D). 

The trend in ChT and ChBP changes in each quadrant was similar to that of overall changes in all four quadrants. Although the changes were not significant in some quadrants, there was a consistent tendency for ChT and ChBP in each quadrant to decrease or increase, according to the treatment (Supplementary Figs. S3A–S3L). 

The interocular differences in ChT, ChBP, refraction, and AXL in the Plano group, LIM group, R-Plano group, and R-LIM group were all correlated: ChBP and ChT (R = 0.53, P < 0.001; Fig. 6E), AXL and refraction (R = −0.75, P < 0.001; Fig. 6F), ChBP and refraction (R = 0.36, P < 0.05; Fig. 6G), and ChT and refraction (R = 0.35, P < 0.05; Fig. 6H). 

Discussion

OCTA Was Effective in Measuring Guinea Pig ChT and ChBP

The choroidal blood supply of mammals arises from the short and long posterior ciliary arteries.³⁹ Through dissection, we found that the anatomy of these vessels in the guinea pig is similar to that in other mammals, but it was difficult in our present experimental conditions to distinguish clearly between these two branches. Our purpose was to observe the partial decrease in ChBP and determine whether it can be detected by OCTA. Therefore, the ciliary arteries around the temporal side of optic disc were transected, with minimal damage to the contralateral vessels supplying the nasal quadrant. Because the retinas of guinea pigs are avascular,^21,40 interference of blood flow signals from retinal blood vessels with the ChBP signals was absent, making the guinea pig an especially good animal model for the study of ChBP patterns by OCTA. 

The validity of the custom program for analyzing the intra- and interexaminer repeatability of the OCTA data was confirmed, with optimal R values ranging from 0.97 to 0.99. These results indicate that the OCTA instrument and measurement methodology produced highly reliable measurements of changes in ChT and ChBP in the guinea pig. 

We then showed that OCTA could reliably evaluate changes in ChT and ChBP, by using it to measure those parameters before and after cervical dislocation or transection of the temporal ciliary artery. Theoretically, ChBP should be completely absent, and ChT significantly thinner than normal, after cervical dislocation, as the blood flow to the target area was halted. Similarly, in the blood vessel transection group, ChBP should have been reduced in the treated temporal quadrant, whereas there should have been no changes in the nasal contralateral quadrant. Our findings were generally consistent with these expectations. 

Changes in ChT and ChBP in Guinea Pig Myopia Were Strongly Correlated

In our guinea pig myopia models, we found that ChT was thinner in eyes with myopia than in fellow eyes. Furthermore, increased 4 days after removal of the myopia-inducing form deprivation or negative lens stimulus. Parallel changes were observed in ChBP. 

These changes in ChT and ChBP were consistent with those found in previous studies. In humans, dramatic reductions in ChT and ChBF were observed in high myopia.^{12,23–25,41–44} Additionally, in highly myopic eyes some studies found the decreases in the large choroidal vessels⁴⁵ and choriocapillaris vessels,^46,47 as well as thinning of the layer of medium-sized vessels,⁴⁸ which might also be correlated with a decrease in ChBP. In our previous study,⁴⁹ we found that ChT was decreased when human subjects received a 6-D accommodation stimulus (a factor that might contribute to the development of myopia) compared with zero-accommodation controls. Animal studies also have consistently found that ChT^10,15–17 and ChBF²⁶ were decreased in experimentally induced myopia and that both parameters increased during recovery. Considering the strong correlations between ChT and ChBP under the different conditions of the present study, it is possible that the changes of ChT are responsible for the changes of ChBP, or vice versa; alternatively, both changes might be due to a common causal factor or mechanism. In addition, it appeared that the changes of ChT and ChBP in the FDM group were statistically more significant than those in the LIM group. These findings indicate that the changes in ChT and ChBP, found in FDM and LIM, may be regulated by different biological mechanisms, as suggested by the previous studies.⁵⁰ 

Anatomically, the choroid is composed mainly of two major vascular layers, Haller's and Sattler's layers, plus the terminal choriocapillaris, which is supplied by them.¹⁰ Some studies have reported that the choroid can autoregulate its own blood flow in humans^51–53 and animals^54–56 in response to rapid changes in IOP or ocular perfusion pressure. We postulate that ChBP in guinea pigs is also autoregulated and that the changes in flow actively subserve emmetropization. Depending on these changes, the choroid would either thin or thicken, consequently moving the retina to the focal plane when defocus-related signals are produced. 

Pathologic Effects of Decreased ChBP on the Retina and Sclera During Myopia Progression

The principal role of maintaining ChBP is to provide sufficient oxygen to the outer retina. Although the outer retina lacks its own vasculature, it consumes the highest amount of oxygen in the eye and is extremely susceptible to hypoxia.⁴⁰ Hypoxia causes increased expression of VEGF, which can promote choroidal neovascularization and cause irreversible visual impairment.^57,58 Epidemiologic studies indicate that choroidal neovascularization is present in 5.2% to 11.3% of high myopia cases,⁵⁹ and a 10-year follow-up study reported that visual acuity dropped to 20/200 or less within 5 and 10 years after the onset of myopic choroidal neovascularization.⁶⁰ Myopic eyes with choroidal thinning will have decreased ChBP and, consequently, a relatively hypoxic environment. The damage caused by chronic hypoxia is likely to cause dysfunction of the retinal pigment epithelium and loss of photoreceptors, and ultimately, visual impairment. In our previous study,²⁷ we found that the expression of HIF-1α in the sclera of guinea pigs was increased significantly after induction of FDM and LIM for 2 weeks. Scleral HIF-1α can promote the trans-differentiation of myofibroblasts and the degradation of type I collagen in human scleral fibroblasts. Importantly, two antihypoxia drugs, salidroside and formononetin, were found to delay the progression of myopia.²⁷ Finally, a reduction in amount of collagen in the extracellular matrix of the sclera can accelerate axial elongation and promote the progression of myopia.^61–63 

In summary, we found that the changes in ChT and ChBP in guinea pig models of myopia are similar to those in human myopia. The choroidal circulation is critical for meeting the oxygen demands of the retina and sclera, and decreases in ChBP can lead to a relatively hypoxic environment, which in turn could induce axial elongation and myopia progression. Although the correlations between changes in ChT and ChBP are strong, the cause-and-effect relationship between them, if any, remains to be determined. 

Conclusions

OCTA is a valuable and reliable tool for measuring ChT and ChBP in guinea pig myopia models. Using this tool, we found that ChT and ChBP decreased significantly during the development of myopia and increased during the recovery from experimentally induced myopia. We propose that choroidal thinning, with concurrent reduction in ChBP, renders the adjacent sclera and retina relatively hypoxic and thus triggers a series of changes in those layers, leading to the onset and progression of myopia and axial elongation. Considering the relatively rapid onset of changes in ChBP and ChT, compared with the much slower development of changes in refraction and AXL, effects on ChBP and ChT may be used as surrogate markers or early predictors for the efficacy of treatments to prevent and control human myopia. 

Acknowledgments

The authors thank Yue Liu (Center for Eye Disease & Development, School of Optometry, University of California, Berkeley, CA, USA) and William K. Stell (Cumming School of Medicine, University of Calgary) for helping with the data analysis and providing editorial support for improving the manuscript. 

Supported by National Natural Science Foundation of China Grants 81670886, 81422007, and 81830027. 

Disclosure: S. Zhang, None; G. Zhang, None; X. Zhou, None; R. Xu, None; S. Wang, None; Z. Guan, None; J. Lu, None; N. Srinivasalu, None; M. Shen, None; Z. Jin, None; J. Qu, None; X. Zhou, None