Biomechanical changes occur in myopic choroidal stroma and mirror those in the adjacent sclera

Abstract

Retina-derived growth signals relayed from the choroid to the sclera cause remodeling of the extracellular scleral matrix, resulting in myopic ocular elongation. However, to the best of our knowledge, no studies have assessed changes in choroidal stromal biomechanical properties during myopia progression. Here we utilized 7 µm-resolution scanning acoustic microscopy (SAM) to assess biomechanical properties (bulk modulus (K) and mass density (rho)) of choroidal stroma from guinea pig eyes with form-deprivation (FD) induced myopia. The choroidal stroma had considerable intrinsic strength arising from its biomechanical properties and these were differentially affected by myopia in central and peripheral regions. Choroidal stromal biomechanical values were also highly correlated with those in adjacent scleral regions, and the choroidal stromal-scleral association was stronger in myopic eyes. Biomechanical changes observed in the choroidal stroma of myopic eyes were mirrored to those observed in the adjacent sclera. These findings suggest that choroidal stromal remodeling may accompany myopia and open the door to the source of the signals that cause scleral remodeling in myopia.

Prof Hoang and colleagues used scanning acoustic microscopy to investigate the biomechanical properties of the choroid in myopic eyes. Their biomechanical analytics reveal changes in choroidal stroma from remodeling were mirrored to those in the adjacent sclera. This finding opens the door to the source of the signals that cause scleral remodeling in myopia.

Introduction

Myopia is a common eye disorder affecting approximately 2.3 billion people worldwide with dramatic rises in global prevalence in recent decades^1,2. While mild degrees of myopia are not often a major cause for concern, eyes with high myopia (defined as a refractive error of more than −6.0 diopters (D) of near-sightedness) can progress to pathologic myopia³, with up to 70% of patients developing visual impairment, potentially leading to permanent visual loss and/or blindness^4,5. Myopia is commonly due to excessive axial elongation of the eye. Across species, visual cues underpin the correct matching between growth in the axial length of the eye and its changing focal length⁶. For example, newborn eyes are generally too short for their optical power and thus hyperopic but this refractive error reduces as individuals mature in most species; such as chicks⁷, mouse⁸, tree shrew⁹, marmoset^10,11, rhesus monkey¹², and human^13–18. This visually-guided process is termed emmetropization. It is disrupted when the eye is repeatedly exposed to blur (such as through form deprivation (FD)^19–22 in which a translucent diffuser is worn that deprives the eye of moderate and high spatial frequencies). Disruption of normal vision with FD or a negative lens worn over the eye results in excessive axial elongation of the eye and the development of myopia (chicks^20,23, tree shrew²⁴, guinea pig²⁵, mouse^26,27, marmoset¹⁰, rhesus monkey^24,28, fish²⁹, rabbit³⁰, and kestrel³¹). Specifically in humans, myopia can develop as a consequence of visual disruption from the condition of eyelid droop (ptosis)³² although it is acknowledged this is a rare origin of human myopia.

The development of myopia is associated with changes in multiple layers of the posterior eye coats (retina, retinal pigment epithelium, Bruch’s membrane, choroid, and sclera), and involves signals originating within the neural retina that are translated through the intervening layers to reach the sclera, the outer fibrous layer of the eye^33–36. Specifically, the sclera thins during the development of myopia³⁷, partly due to decreased scleral collagen synthesis and increased collagen degradation^38,39. These changes are particularly noticeable at the posterior pole of the eye. Decreased collagen fiber diameter occurs in the posterior sclera in both mammalian animal models of myopia and in high myopia patients⁴⁰. This remodeling and thinning suggests that the myopic sclera is structurally weaker and less able to constrain the eyeball under the tension of normal intraocular pressure, thus contributing to eye elongation^41,42.

Current theory proposes that myopia develops ultimately because the sclera loses its biomechanical strength^37,42,43, while the role of the retina and other intervening layers is generally thought of in terms of chemical signal initiation and relay^6,44–46. However, it is possible that changes in the structural integrity of the eyeball during myopia development and progression may not be restricted to only the sclera. Situated between the sclera and the retina is the choroid, which is a highly vascularized layer uniquely positioned to relay retinal-derived growth signals to the sclera to affect changes in extracellular matrix remodeling that result in myopic ocular elongation^46,47. Early studies in chicks demonstrated that the choroid plays an active role in emmetropization by modulating its thickness that results in shifting the retina to the focal plane of the eye (choroidal accommodation)^48–50. In humans and other mammals, bi-directional changes in choroidal thickness are also associated with the sign of imposed defocus^48,51. Specifically, the choroid thins in response to hyperopic defocus (which also leads to myopia)^48,52 and thickens in response to myopic defocus in most species studied, including humans^53,54. In humans, these small transient changes generally lie within the depth of focus and do not directly affect refractive error. Nevertheless, they may reflect the choroidal release of factors that have the potential to regulate scleral extracellular matrix remodeling^48,55,56.

However, it is often overlooked that the choroidal matrix itself likely possesses remarkable biomechanical strength. Van Alphen’s classic study reported that the globe elongated axially but kept its shape despite vitreous inflation after removal of the posterior half of the sclera shell⁵⁷. In his report with porcine eyes, the Bruch’s membrane -choroidal complex sustained a substantially high intraocular pressure before rupture, despite the removal of the scleral shell⁵⁷. Additionally, numerical simulation suggests that Bruch’s membrane may have a non-negligible influence on intraocular pressure-induced optic nerve head (ONH) deformations⁵⁸. These reports suggest that the choroid may not just act as a passive component of the posterior pole but may possess biomechanical strength that actively contributes to maintaining the shape of the eye. However, to date, no studies have been performed to directly assess the biomechanical properties of the choroidal stroma or how these may change during myopia development or progression⁵⁹.

To map the choroidal stromal biomechanical properties, the current study used a 250 MHz ultrasound-based scanning acoustic microscopy (SAM) system, the same as previously used to assess biomechanical properties of scleral tissues extracted from myopic guinea pig eyes⁶⁰. The selection of the ultrasound transducer center frequency was based on the choroidal thickness as reported in the previous study⁶¹. SAM spatial resolution decreases (i.e., improves) with increasing ultrasound frequency. Since choroidal thickness ranges from 15 µm to 100 µm in this species⁶¹, a 250 MHz frequency was chosen to achieve a lateral resolution of 7 µm, enabling the resolution of the choroidal layer. SAM allows reconstruction of two-dimensional (2D) maps at 7 µm resolution of two biomechanical properties: bulk modulus (K in GPa) and mass density (rho in g·cm⁻³). K measures the ability of the tissue to withstand compression under pressure, with higher K values meaning the tissue is relatively resistant to compression (akin to greater stiffness)^62–64. That is, K is the inverse of compressibility and tends to be positively correlated with Young’s modulus⁶⁵. Rho measures mass per unit volume. In the present work, SAM was used to assess choroidal stromal biomechanical properties to determine if they are altered by myopia during the early stages of form deprivation myopia (FDM) in a mammalian eye.

The current study also assessed the degree of choroidal stromal intrinsic biomechanical strength compared to the neighboring sclera and explored the regional variations in choroidal stromal biomechanical properties and their relationship to myopia and to those observed in the adjacent sclera. It was found that the biomechanical properties of the choroidal stroma were differentially affected by myopia, notably when comparing central to more peripheral zones, suggesting that the choroid may play an essential role in loss of the eye’s biomechanical strength in myopia. Unexpectedly, it was also found that local biomechanical changes in the choroidal stroma tightly correlated with those observed in the overlying neighboring sclera, suggesting that signals transmitted from the choroid to sclera may be biomechanical in nature and/or that possibly a common factor initiated locally in the retina may locally remodel both the choroidal and scleral matrix.

Results

Young guinea pig pups that wore a diffuser (to impose FD) over one eye for one week developed significant relative myopia (mean intraocular difference (FD—control eye) −6.23 ± 1.75 D, range −3.21 to −9.29 D, FD eye −0.34 ± 4.02 D, control eye +5.72 ± 2.65 D, p < 0.001) due to rapid elongation of the FD eye (mean axial length increase (intraocular difference (FD—control eye) 93 ± 32 µm, range 58–137 µm, FD eye 7.97–8.14 mm, control eye 7.85–8.05 mm, p < 0.05).

Both choroidal stroma and sclera have intrinsic biomechanical strength that is affected by myopia

Analysis of choroidal stroma and scleral biomechanical properties in ex-vivo thin sections, cut through the center of the optic nerve and stretching to the limbus in both the vertical and horizontal meridians, revealed that the choroidal stroma showed considerable biomechanical strength, with an average bulk modulus (K) of 2.80 ± 0.14 GPa (vertical) and 2.73 ± 0.14 GPa (horizontal) in untreated control eyes. Although the sclera had a greater mean bulk modulus value than the choroidal stroma (3.13 ± 0.27 vs. 2.78 ± 0.15, respectively, Fig. 1a, Table 1, p < 0.001), the average K in the choroidal stroma was only 11.4% less than that measured in the sclera.

Fig. 1. Choroidal stromal and scleral biomechanical properties differed.

Bar charts comparing bulk modulus (a vertical axis, K in GPa) and mass density (b vertical axis, rho, in g·cm⁻³) in choroidal stroma (filled columns) and sclera (blank columns) averaged across the eyeball in control eyes (left columns) and FD eyes (right columns). Each dot represents the raw data from a given region of interest (ROI) used to calculate the mean, depicted by the bar height, and the standard deviation (error bar). Horizontal line between bars represents the statistical significance of the difference (p < 0.05). P-values were calculated using paired t-test. FD form deprivation.

Table 1.

Sclera vs. choroid bulk modulus by region

Eye	Meridian	Sclera (Mean ± SD) [GPa]	Choroid (Mean ± SD) [GPa]	p-value (sclera vs choroid)
Control	Vertical	3.09 ± 0.27	2.80 ± 0.14	<0.001***
	Horizontal	3.23 ± 0.26	2.73 ± 0.14	<0.001***
FD Eyes	Vertical	3.05 ± 0.21	2.83 ± 0.22	<0.001***
	Horizontal	3.02 ± 0.22	2.77 ± 0.19	<0.001***
FD Eyes-control	Vertical	−0.10 ± 0.20 (p = 0.37)	0.01 ± 0.09 (p = 0.79)	0.32
	Horizontal	−0.22 ± 0.16 (p = 0.04*)	0.04 ± 0.14 (p = 0.55)	0.03*

Scleral and choroidal stromal bulk modulus (in GPa, represented as mean ± standard deviation, SD) for control (top rows), 1-week FD eyes (middle rows), and the intraocular differences (FD-control, bottom rows). “p-value” refers to the result of a paired t-test. Asterisks after the p-value represent the statistical significance. P-values were calculated using paired t-test for sclera vs choroid, and one-sample t-test for the intraocular differences.

FD form deprivation.

*p < 0.05, ***p < 0.001.

Consistent with our prior study⁶⁰, FDM caused the sclera to become more compliant (i.e., more compressible with smaller bulk modulus values) generally (Fig. 1a), although only developing significantly reduced K values in the horizontal meridian (Table 1). While the bulk modulus of the sclera remained larger than the choroidal stroma in myopic eyes (Fig. 1a), relative to control eyes, these differences between the bulk modulus of the sclera and choroidal stroma were not significantly different in the vertical meridian and instead arose primarily from differences in the horizontal meridian (Table 1).

A different pattern emerged in the general difference between the choroidal stroma and sclera in their mass densities (rho) (Fig. 1b, Table 2). The choroidal stroma was significantly denser than the sclera in untreated control eyes, where choroidal stromal rho was greater than that in the sclera by 0.011 g·cm⁻³ (p = 0.01, Fig. 1b). This difference was blunted by induced myopia since the mass density of the sclera and choroidal stroma were no longer significantly different in FD eyes (0.009 g·cm⁻³, p = 0.07).

Table 2.

Sclera vs. choroid density by region

Eye	Meridian	Sclera (Mean ± SD) [g·cm−3]	Choroid (Mean ± SD) [g·cm−3]	p-value (sclera vs choroid)
Control	Vertical	1.03 ± 0.04	1.05 ± 0.03	<0.001***
	Horizontal	1.06 ± 0.04	1.03 ± 0.03	<0.001***
FD Eyes	Vertical	1.04 ± 0.03	1.05 ± 0.03	0.16
	Horizontal	1.03 ± 0.04	1.05 ± 0.03	0.06
FD Eyes-control	Vertical	−0.01 ± 0.04 (p = 0.81)	-0.01 ± 0.01 (p = 0.18)	0.82
	Horizontal	−0.02 ± 0.05 (p = 0.50)	0.01 ± 0.01 (p = 0.03*)	0.23

Scleral and choroidal stromal mass density (in g·cm−3, represented as mean ± standard deviation, SD) for control (top rows), 1-week FD eyes (middle rows), and the intraocular differences (FD-control, bottom rows). “p-value” refers to the result of a paired t-test. Asterisks after the p-value represents the statistical significance. P-values were calculated using paired t-test for sclera vs choroid, and one-sample t-test for the intraocular differences.

FD form deprivation.

*p < 0.05, ***p < 0.001.

It may be thought that these changes might be due to differences in choroidal axial thickness. However, when comparing within quadrants, there was no significant difference comparing contralateral control versus FD eyes (64.3 ± 31.2 vs. 59.3 ± 22.5 µm, p = 0.36 for the SUP; 84.3 ± 28.9 vs. 78.8 ± 33.4 µm, p = 0.33 for INF; 39.7 ± 13.2 vs. 37.6 ± 13.2 µm, p = 0.44 for NAS; and 36.7 ± 18.1 vs. 42.1 ± 12.1 µm, p = 0.44 for the TMP quadrant, respectively). Moreover, the proximal to distal variation in neither biomechanical compressibility nor mass density induced by FDM was associated with the thickness of the choroidal matrix (p = 0.45, R = 0.09 with bulk modulus K, and p = 0.69, R = −0.04 with the mass density rho), at least under these ex-vivo conditions.

Regional variation in the bulk modulus (K) of the choroidal stroma emerges in myopic eyes

The mean values of choroidal stromal K as a function of eccentricity are plotted in Fig. 2a for all vertical sections and in Fig. 2c for all horizontal sections. The original values are shown in Supplementary Fig. 1a and c. In control eyes, the mean values of K showed little variation between central and more peripheral regions (for the vertical meridian: proximal, 2.81 ± 0.15 GPa, n = 79 data points; distal, 2.79 ± 0.13 GPa, n = 59, p = 0.61; for the horizontal meridian: proximal, 2.72 ± 0.14 GPa, n = 51; distal, 2.76 ± 0.14 GPa, n = 47, p = 0.35). In contrast, in FD eyes, K values were significantly greater in the proximal zone compared to the distal zone within the vertical meridian (proximal: 2.88 ± 0.23 GPa, n = 59; distal: 2.73 ± 0.15 GPa, n = 33; p < 0.001, Fig. 2e).

Fig. 2. Regional differences in choroidal stromal bulk modulus (a, c, e) and mass density (b, d, f).

Binned scatterplot comparing averaged choroidal stromal bulk modulus (K in GPa) and averaged choroidal stromal mass density (rho in g·cm⁻³) at different eccentricities (mm) from the central axis midline for the vertical (a, b) and horizontal meridians (c, d). Triangle symbols represent control eyes and circle symbols represent FD eyes. Each dot represents the mean from up to 12 regions of interests (ROIs) from 12 animals measured at the given eccentricity. Black solid line is the center of the central axis. Gray shaded area represents the optic nerve head (ONH, a 1.2 mm region). Bar charts comparing choroidal stromal bulk modulus and choroidal stromal mass density averaged across proximal and distal regions (e, f). Grey and red bars represent control and FD eyes, respectively. Each dot on the bar represents each ROI used for the calculation, and the error bar represents the standard deviation. The horizontal line across bars represents the statistical significance (p < 0.05) of the difference between control vs. FD eyes, proximal vs. distal areas, or vertical vs. horizontal meridians. FD form deprivation. P-values were calculated using paired t-test for control vs. FD eyes, and unpaired t-test for proximal vs. distal areas, or vertical vs. horizontal meridians.

Interestingly, compared to control eyes, values of K in the vertical meridian of FD eyes were significantly greater in the proximal zone located near the optic nerve head (i.e., <2 mm away from the central axis, FD: 2.88 ± 0.23 GPa, n = 59 data points vs. control: 2.81 ± 0.15 GPa, n = 79, p = 0.029) but significantly less in the more peripheral distal zones (FD: 2.73 ± 0.15 GPa, n = 33 vs. control: 2.79 ± 0.13 GPa, n = 60, p = 0.037, Fig. 2e left). This implies myopia induced a decrease in choroidal stromal compressibility in regions within 2 mm above or below the central axis and a potentially more elastic and compressible choroid stroma in more peripheral vertical regions. No significant differences in K were detected between the FD and control eyes in the horizontal meridian (Fig. 2e right).

Mass density (Rho) of the choroidal stroma is affected by myopia

The mean mass densities (rho) for the choroidal stroma at each eccentricity are plotted in Fig. 2b for all vertical sections and in Fig. 2d for all horizontal sections. The original values are shown in Supplementary Fig. 1b and d. In control eyes, mass density was significantly greater in the far periphery compared to more central eccentricities closer to the central axis and optic nerve (distal: 1.06 ± 0.02 g·cm⁻³, n = 74 data points; proximal: 1.04 ± 0.03 g·cm⁻³, n = 130; p < 0.001). This trend was consistent both in the vertical and horizontal meridians (vertical (distal): 1.06 ± 0.03 g·cm⁻³, n = 60; vertical (proximal): 1.04 ± 0.02 g·cm⁻³, n = 79; p < 0.001; Fig. 2f left; and horizontal (distal): 1.06 ± 0.01 g·cm⁻³, n = 14; horizontal (proximal): 1.02 ± 0.04 g·cm⁻³, n = 51; p < 0.001; Fig. 2f right).

This central-peripheral variation in mass density was not observed in FD eyes (p = 0.38). Furthermore, within distal regions, overall mass density was significantly reduced in FD eyes compared to control eyes (FD: 1.04 ± 0.04 g·cm⁻³, n = 45 data points vs. control: 1.06 ± 0.02 g·cm⁻³, n = 74, p < 0.001) and this was true in both the horizontal (Fig. 2d) and vertical meridians, particularly noticeable in the inferior region (Fig. 2b and Supplementary Fig. 2b). In contrast, within proximal regions, although rho values were not affected by FDM in the vertical meridian, choroidal stromal mass density tended to be increased by induced myopia within the horizontal meridian (Fig. 2f). This suggests that FDM appeared to reduce choroidal stromal mass density in the periphery while increasing it in more central regions.

Biomechanical properties of the sclera and choroidal stroma are correlated in myopic eyes

Choroidal stromal and scleral bulk modulus were not generally correlated when averaged across the eye in control eyes (Fig. 3a, R = 0.02, p = 0.81). This was particularly obvious in the horizontal meridian (Fig. 3c, R = −0.13, p = 0.32), but also failed to reach significance across the vertical meridian (Fig. 3b, R = 0.16, p = 0.06) except in the SUP-proximal region (Table 3, R = 0.36, p = 0.016). In contrast, after 1 week of FDM, choroidal stromal and scleral K became correlated (Fig. 3a, R = 0.45, p < 0.001) showing that local stiffness (measured independently of local tissue thickness) was comparable between adjacent choroidal stromal and scleral regions of interest (ROI). In FD eyes, choroidal stromal and scleral K were positively correlated in each meridian (Fig. 3b and c) as well as all quadrants (Table 3). Both SUP-proximal and NAS-proximal regions showed strong correlations (SUP-proximal: R = 0.87, p < 0.001; NAS-proximal: R = 0.58, p < 0.001; Table 3). Additionally, the slopes of the choroidal stromal/scleral relationship were significantly different between control and FD eyes when averaged across the eye (Fig. 3a, p = 0.004) confirming that myopia-induced coordinated changes in the bulk modulus of adjacent sclera and choroidal stromal regions.

Fig. 3. Correlation between choroidal stromal and scleral biomechanical properties before and after myopia inducement.

Scatterplots comparing choroidal stromal K (in GPa) and scleral K (in GPa) (a, b, c) and choroidal rho (in g·cm⁻³) and scleral rho (in g·cm⁻³) (d, e, f) for all data points (a, d); vertical meridian (b, e); and horizontal meridian (c, f). Each graph shows values for control (open circles) and FD eyes (filled circles) from sectioned guinea pig eyes. Each dot represents the local K or rho value for each region of interest (ROI) and the color of the dot indicates its eccentricity ranging from 0.6 mm (near the central axis, blue) to 5 mm in the periphery (yellow). Fitted linear regression lines are shown in red for N = 233 to 251 data points (dashed line for control eyes, and solid line for FD eyes). The diagonal dotted line represents a slope of 1.0 where the sclera and choroidal stroma would have equal values. Formulae on each graph are for the regression lines with asterisks representing the statistical significance; *p < 0.05, **p < 0.01. P-values were calculated using Fisher’s z transformation. FD form deprivation.

Table 3.

Bulk modulus regression analysis by quadrant and region

Quad.	Eye	Lateral position	R	p-value	Slope [GPa]
SUP	Control	all	0.24	0.045*	0.55
		prox.	0.36	0.013*	0.44
		dist.	0.24	0.248	0.73
INF	Control	all	0.02	0.883	0.02
		prox.	0.04	0.801	0.06
		dist.	−0.02	0.914	−0.02
NAS	Control	all	−0.09	0.487	−0.12
		prox.	−0.07	0.625	−0.10
		dist.	−0.22	0.521	−0.18
TMP	Control	all	−0.62	0.073	−5.60
		prox.	−0.67	0.148	−9.22
		dist.	−0.32	0.791	−0.27
SUP	FD Eye	all	0.58	<0.001***	0.42
		prox.	0.87	<0.001***	0.56
		dist.	0.13	0.677	0.11
INF	FD Eye	all	0.27	0.039*	0.30
		prox.	0.09	0.573	0.10
		dist.	0.35	0.123	0.50
NAS	FD Eye	all	0.58	<0.001***	0.58
		prox.	0.58	<0.001***	0.64
		dist.	0.87	0.053	0.74
TMP	FD Eye	all	0.66	0.007**	1.26
		prox.	0.64	0.089	1.41
		dist.	0.86	0.013*	2.04

Details of the correlations in each quadrant is shown between choroidal stromal K and scleral K. Asterisks on the p-value represent the statistical significance. R values were calculated using Pearson’s correlation, and p-values were calculated using Fisher’s z transformation.

FD form deprivation, INF inferior, NAS nasal, SUP superior, TMP temporal, prox proximal, dist distal.

*p < 0.05, **p < 0.01, ***p < 0.001.

Unlike K, mass density was naturally correlated between adjacent choroidal stromal and scleral regions as shown in control eyes (Fig. 3d), and this correlation was strongest in the horizontal meridian (Fig. 3f). After 1 week of FDM, choroidal stromal and scleral rho values remained correlated, and extended into the INF quadrant (Table 4, R = 0.42, p = 0.001). However, the slope of the correlation was reduced in FD eyes compared to control eyes in SUP and TMP quadrants (Table 4) suggesting that myopia disrupted the normal association between scleral and choroidal stromal mass density in some regions.

Table 4.

Mass density regression analysis by quadrant and region

Quad.	Eye	Lateral position	R	p-value	Slope [g·cm−3]
SUP	Control	all	0.59	<0.001***	1.34
		prox.	0.49	0.001**	0.70
		dist.	0.56	0.003**	1.79
INF	Control	all	0.20	0.093	0.18
		prox.	0.17	0.331	0.14
		dist.	0.26	0.145	0.24
NAS	Control	all	0.40	0.004**	0.58
		prox.	0.38	0.015*	0.58
		dist.	−0.21	0.542	−0.30
TMP	Control	all	0.95	0.001**	1.91
		prox.	0.99	0.010*	2.01
		dist.	−0.87	0.333	−1.58
SUP	FD Eye	all	0.27	0.125	0.28
		prox.	0.47	0.030*	0.47
		dist.	-0.22	0.462	−0.25
INF	FD Eye	all	0.42	0.001**	0.47
		prox.	0.57	<0.001***	0.65
		dist.	0.25	0.297	0.29
NAS	FD Eye	all	0.59	<0.001***	0.93
		prox.	0.60	<0.001***	0.96
		dist.	0.05	0.968	0.05
TMP	FD Eye	all	0.52	0.047*	0.52
		prox.	0.67	0.067	0.66
		dist.	0.32	0.488	0.33

Details of the correlations in each quadrant are shown between choroidal stromal rho and scleral rho. Asterisks on the p-value represent the statistical significance. R values were calculated using Pearson’s correlation, and p-values were calculated using Fisher’s z transformation.

FD form deprivation, INF inferior, NAS nasal, SUP superior, TMP temporal, prox proximal, Dist distal.

*p < 0.05, **p < 0.01, ***p < 0.001.

Discussion

The choroid is situated between the retinal and scleral layers and plays an important role in relaying the ocular growth signals to the sclera while contributing to maintaining the shape of the eyeball with its own stiffness. However, the choroidal response to the ocular growth signal is not fully understood. Many reports suggest that either separate visual signals guide the choroidal and scleral responses, or the same visual signal results in differing responses depending on the tissue type (choroid vs. sclera)^33–36. This study with fine-resolution SAM reports two key findings: first, the choroid is not just a bystander during myopic eye elongation, but its fundamental structure is possibly altered, as manifested by changes in its biomechanical properties, and second, the changes in biomechanical properties in the choroidal stroma mirror those in the adjacent sclera in a given location.

Biomechanical choroidal stromal changes implicate micro-structural alterations induced by myopia

The local stiffness within a 7 × 12 µm area of the choroidal stroma in normal eyes (as measured by the bulk modulus in which a smaller K value suggests that under the same uniform increase in static pressure, the volume of the choroidal stroma will decrease more) was consistent throughout different regions. However, myopia caused the peripheral superior and inferior choroidal stroma to become more compressible, implying greater elasticity at a microscopic level, while there was a tendency to change in the opposite direction and become less distensible in central regions closer to the central axis. It is known that in the sclera, larger K values are correlated with longer collagen fibers⁶⁶. However, the exact contributors to changes in K in the choroidal stroma are unknown. Further studies are warranted, that include detailed structural analysis for myopia-induced changes in the choroidal stroma elements (consisting of collagen, elastic fibers, fibroblasts, and smooth muscle cells).

At the macroscopic level, the stiffness of the choroid could be affected by other factors, notably the effect of fluid flow. Although the thickness of the choroid is known to be modulated by imposed defocus, the changes in mammals⁶⁷ and primates⁶⁸ induced by myopia are also relatively small, and measurement of the changes over the diurnal cycle is generally preferred to detect these small shifts^53,54,69. Nevertheless, it is of interest to know if such thinning and thickening might also change the biomechanical properties of the choroid. In the current study, the eyes were extracted and sectioned at one point in time, well after such dynamic changes are normally initiated⁴⁶, and the biomechanical changes were measured in the remaining sponge like-matrix of the choroid that was rehydrated with saline. Therefore, one would not expect to find, nor did we find, any thickness difference between FD myopic and control eyes. Additionally, the method used in the current study allows the calculation of biomechanical properties independently for each pixel⁷⁰, and independent of any external perturbation, so fluid flow is irrelevant. Moreover, all sections were cut at the same thickness, being 12 µm in the depth. It may be thought that the microscopic biomechanical measures taken would depend on the axial thickness of the choroid. However, this is unlikely since the ultrasound beam penetrates each 12 µm section perpendicular to the axial thickness of the choroid, and neither K nor Rho were correlated with axial choroidal thickness.

Since measurements were orthogonal to the plane of section, all layers were equally sampled in both vertical and horizontal sections and the values provided are the average of all layers within each ROI. Any distinctions observed between these two orientations denote a fundamental disparity in the overall fiber orientation and/or stiffness, as opposed to indicating any disparities among the different layers of the choroid. These layers are known to vary widely in matrix composition and structure and hence would be expected to differ in biomechanical properties.

Another factor that may affect choroidal stiffness is choroidal composition of ground substances such as proteoglycans (PG) and glycosaminoglycan (GAG). Their synthesis is bidirectionally changed when defocus is imposed with spectacle lenses worn by chicks; increasing with myopic defocus and decreasing with hyperopic defocus^48,71, conditions which significantly thicken or thin the choroid respectively⁷¹. Since proteoglycans and glycosaminoglycans are osmotically active molecules, their increased synthesis may lead to greater fluid retention in the choroid, consequently augmenting choroidal thickness⁷². However, thus far, there have been no reports on whether these changes affect the biomechanical properties of the choroid.

The second biomechanical parameter, rho, measures mass per unit volume (i.e., mass density) as determined by the concentration of choroidal components including collagen fibers, melanocytes, and fibroblasts. The average mass density over a fairly large measurement area is expected to be different from conventional measurements due to the fine resolution afforded by SAM. Specifically, given our ability to discern biomechanical properties at a 7-micron resolution, we are able to mathematically omit regions that contain only blood, or empty space, which could artificially drive down values obtained by more conventional methods in which the low or near-zero values for such regions would drive down the calculated value for a given ROI. In our specific study, we were able to assess the biomechanical properties of the stroma of a vascular layer, the choroid, and mathematically omit the choroidal lumen that would be prone to fluctuations from inflation (or deflation) from blood fill. Our current data demonstrated regional differences in the microscopic density of these fundamental elements in normal choroidal stroma, extracted as ex-vivo sections and rehydrated. We uncovered that the distal choroidal stroma normally is consistently denser, in terms of mass per unit volume, in comparison with more central regions. FDM blunted this regional variation resulting in mass density being more uniform throughout the posterior eyewall. Myopia caused greater but equivalent changes in overall choroidal stromal mass density like those observed for bulk modulus, whereby mass density reduced in all four sectors (SUP, INF, NAS, TMP) in distal regions of the choroidal stroma while increasing in more central regions about the central axis.

Taken together, these results could be interpreted as during the early stages of FDM in a mammalian eye, the microscopic structure of the choroidal stroma changes, becoming relatively stiffer and denser around the optic nerve and central axis while becoming less stiff with relatively lower mass density in more peripheral regions. Although it is well known that blood flow in the choroid is reduced in myopic eyes and many molecular factors in the choroid are modified^46,73, the current results demonstrate that myopic signals originating from the retina also affect the biomechanical properties of the choroidal stroma at a microscopic level, suggesting that scleral remodeling is not the only target of these signals.

Implications of the opposing local changes in proximal and distal choroidal stroma

Human myopic eyes elongate primarily at the posterior pole becoming prolate in shape leading to myopia on-axis and relative hyperopia in the periphery⁷⁴. In the same way, like in other species⁷⁵ the guinea pig eye shows opposite changes in eye shape between the periphery and more central regions during the early development of myopia^76,77. The presence of relative peripheral hyperopia in myopic eyes is partly the basis for multiple anti-myopia spectacles and contact lens designs⁷⁵. Therefore, it is interesting that even at the microscopic level, FDM initially induces relatively greater bulk modulus and increased matrix mass density in the choroidal stroma around the optic nerve and central axis, with opposite changes in more peripheral regions towards the eye equator.

These findings of region-dependent changes caused by myopia are also consistent with past reports of regional variations due to localized, partial eye shape changes resulting in localized, regional myopia inducement in macaque monkeys^68,78, chicks^79,80, and guinea pigs⁸¹. However, it might seem odd that the direction of biomechanical changes in the choroidal stroma (and as described below also in the sclera) are the opposite of the well-known findings that scleral collagen remodels⁸² and as a result, the sclera thins and loses strength, becoming mechanically weaker at the posterior pole with advanced myopia development^41,77. However, the reduction in collagen fibral diameter and scleral collagen remodeling is substantially delayed in FD tree shrews until 4-5 weeks after myopia induction, which is after the rapid eye elongation phase moves into the slow elongation phase^83,84. Guinea pig eyes undergoing relatively short-term FD (1 week) may not be expected to show dramatic decreases in collagen fiber diameter at this early stage of myopia development. In support of this, it has previously been reported that although the guinea pig eye rapidly elongates after 6 days of FD, on-axis scleral- and choroidal- thickness did not decrease significantly at this given time point²². Additionally, at least in the sclera, 2 weeks of FD in guinea pigs only reduced the total number of cells in the region between the optic nerve and 10° nasal but not elsewhere and there were no significant changes noted in the number of myofibroblasts at this early stage⁸⁵.

Nevertheless, the present work clearly demonstrates regional differences in choroidal stromal biomechanical properties between myopic and control eyes, suggesting that variation in matrix structure occurs early during myopia development. Specifically, the data here shows an early shift to a more compliant choroid and sclera towards the periphery, suggesting that if similar regional effects are seen in children developing myopia, early treatments for myopia during development may need to target and strengthen this region.

The relationship between the sclera and choroidal stroma biomechanical properties

The bulk modulus in the choroidal stroma was consistently found to be ~10% lower than that found in adjacent sclera in all meridians (regardless of the level of induced myopia). However, the mass density was slightly higher in the choroidal stroma compared to the sclera, at least in control eyes, possibly due to a tighter packing of vascular membranes in the choroid and/or arising from the suprachoroidal layer. Additionally, the fact that the choroidal stroma had a bulk modulus of approximately 3 GPa supports van Alphen’s classic result that the choroid, by itself, can maintain the shape of the eyeball⁵⁷. Tissues such as myocardium (subepicardial region)⁸⁶ and bovine Achilles tendon⁸⁷ are well-known to possess sufficient biomechanical strength to maintain their own shape during rest and with movement, where the average bulk modulus is around 3 GPa (specifically, 2.78 ± 0.04 and 3.08 ± 0.08 for myocardial and tendon tissues, respectively). It also suggests that the structural integrity of the choroid could help maintain the location of the photoreceptor plane and hence defocus compensation in the face of variation in stress forces which may arise from intraocular pressure and/or accommodative forces that are thought to be directed into the choroid and may impact the optic nerve head⁸⁸.

In the myopic signal cascade, aberrant signals are thought to arise in the retina, transported through the choroid, and ultimately lead to changes in biomechanics of the effector tissue, the sclera. Our finding that myopia results in microscopic biomechanical properties of the choroidal stroma that mirror those found in the adjacent sclera is important as it sheds new light on the potential role of the choroid in myopiagenesis, which may be more than only a transport medium. The fact that K values in proximal and distal regions of control eyes were not significantly different (p = 0.46) shows that K values are normally homogeneous for all positions measured, whether horizontal or vertical or distal or proximal. In contrast, in myopic eyes, K values were statistically different (p = 0.02) between proximal and distal in both the horizontal and the vertical direction, directly confirming that myopia caused horizontal versus vertical differences. It is interesting to note that in these small mammals, the superior region which views the ground is naturally myopic while the inferior region which views the sky is naturally hyperopic^61,77,89, suggesting that bulk modulus varies as a function of the degree of natural inherent myopia. Additionally, when myopia was induced with FD, choroidal stromal and scleral bulk modulus became strongly correlated throughout all the different regions measured.

The bulk modulus of the sclera and choroidal stroma were not significantly different in the vertical meridian and instead arose primarily from differences in the horizontal meridian. This may possibly derive from the extraocular muscles (i.e., temporal and nasal recti muscles versus superior and inferior recti muscles)⁹⁰ and the fact that a majority of the eye movements in guinea pigs, be it saccadic eye movements or vergence eye movements, predominately employ the horizontal recti muscles. Although speculative, this raises the intriguing possibility of an effect of stresses associated with eye movements being supported by local biomechanical differences in the underlying tissues. This may arise from possible differences in the orientation of collagen bundles in the horizontal and vertical planes. Such differences have been observed in human sclera where collagen anisotropy differs between horizontal and vertical planes, a pattern disrupted in eyes with high myopia.

The synchrony between the local biomechanical properties of the choroidal stroma and adjacent sclera may arise if both structures respond to the same signals. Numerous eye elongation signal pathways have been suggested to regulate eye growth and result in aberrant eye elongation in myopia^47,91. TGFβ expression is well known to be associated with scleral remodeling during myopia development^92,93 but what its exact effects are on the neighboring choroid remains to be determined. Common local growth factors that mediate scleral remodeling that reside in both sclera and choroid include retinoic acid (RA)⁹⁴, adenosine^95,96, insulin^97–99, and nitric oxide¹⁰⁰. Even atropine has been suggested to act directly on the sclera due to the presence muscarinic receptors^27,101, and if true, it may also influence the choroid. Alternatively, it is possible that retinal or retinal pigment epithelium factors modulated by myopiagenic stimuli can remodel the choroid, and that the induced change in local stress within the choroid could stimulate scleral fibroblasts, ultimately leading to collagen remodeling in the sclera. Elastic components have been detected in the choroid-sclera transition zone in primate and human eyes, which may also be important¹⁰².

Due to the absence of the extracellular fluids, the choroid used in this study was expected to be thinner on-axis than in in-vivo conditions. The thicknesses of the choroid and sclera were 54 µm and 62 µm in control eyes and 52 µm and 66 µm in myopic eyes respectively. This was indeed thinner than that measured in-vivo at a matched age in normal animals. However, the ratio of choroidal and scleral thickness in control eyes from our measurement (87%) was matched well with that measured in vivo (86%) in untreated animals¹⁰³. This suggests choroid and sclera were equivalently influenced by the hydration and pressure.

Limitations

It is important to note that SAM measurements were done ex vivo. Under in vivo conditions, a rich vascular network supplies blood and extracellular fluids that fill the choroidal stroma. Our ex vivo tissue sections attempted to replicate the in vivo condition as much as possible by performing enucleation as soon as possible after euthanasia, with immediate freezing to −80 °C, however, it is impossible for our tissue sections to contain the same amount of intrachoroidal blood as when in vivo. In relation to the subsequent thawing from −80 °C with rehydration prior to imaging with SAM, there is evidence indicating the potentially concerning phenomenon of thaw loss is greatly minimized with rapid freezing¹⁰⁴. In order to address any potential effects from hydration status, all sections were rehydrated with saline immediately before SAM imaging. Hence the microscopic mechanical properties measured in the choroidal stroma here were not affected by either the overall axial choroidal thickness or coronal sectioning thickness changes during the tissue processing. Nevertheless, it would be expected that at least at the macroscopic level, the resulting biomechanical properties of the choroid could also be influenced by any temporal variation in blood flow present in vivo. Moreover, it is important to stress the present work focused on choroidal stroma, in that statistical analysis was performed only on non-zero values contained within each ROI, which effectively and mathematically omits all choroidal lumen. This was done to allow the current study to exclude relatively large spaces with near-zero values for biomechanical measures (that would have artificially driven down averaged values within ROIs) since the choroidal lumen could contain variable volumes of blood, including eyes void of blood. For example, the bulk modulus of fluid is constant at 2.25 GPa but varies based on temperature. The biomechanical properties of the full choroidal layer, including the dynamically-changing choroidal lumen (in terms of volume and temperature) may differ from our findings that focused on specifically the choroidal stroma.

It should be acknowledged that bulk modulus is calculated from the acoustic impedance and speed of sound estimated from the very high-frequency compressional wave, not a low-frequency shear wave, and that no deformation response is being measured in the current study. Thus, the biomechanical properties measured here are not the same as other viscoelastic measures such as that extracted from strip extensiometry. Nor do we yet know exactly how viscoelastic properties, such as linearity, might relate between the micro and macro scales.

It should also be noted that the current results relate to a relatively brief period of FD selected here because it was thought that choroidal signals occur prior to longer-term changes in the sclera. However, we have demonstrated that fundamental changes in the choroid matrix may commence relatively early in the myopia development process and may prove to be a myopia correction treatment target. It would be of future interest to determine how the choroid biomechanical properties change after longer periods of myopia progression.

Conclusion

Scanning acoustic microscopy (SAM) employed with a 250 MHz ultrasound transducer allows the two-dimensional quantification of the choroidal biomechanical properties at 7-µm resolution. Taking advantage of this imaging tool, we have shown that the choroid is not a passive tissue neighboring the tectonic sclera, but itself possesses stiffness. Moreover, choroidal stiffness is influenced by location and/or level of myopia (including both inherent regional myopia and induced myopia). Most importantly, the result suggests that the biomechanical changes in the choroid mirror that in the adjacent sclera at a given location. These important findings may eventually contribute to unveiling the role of the choroid in myopia progression and identifying the potential treatment target.

Materials and methods

Animals and Ocular Measures

To induce myopia, young guinea pig pups (n = 12, tri-colored aged between 4 and 7 days) were raised with a translucent diffuser over their right eye for 1 week. Throughout this period, young animals were raised with their mothers under white-light LED illumination⁸⁹ on a 12 h/12 h day-night cycle. At day 7 of life, refractive error was measured in cyclopleged eyes using a Nidek autorefractor (ARK-30, Nidek Co., Japan) as previously described⁹⁴. Cycloplegia was induced with 1% cyclopentolate 1.5–2 h prior to measurement. This was followed by measurement of ocular distances on axis using high-frequency ultrasound in animals anesthetized with isoflurane in oxygen⁹⁴. Following each ocular measurement, the translucent diffusers were replaced on the eye. All procedures were approved under Australian animal ethics legislative requirements and adhered to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.

Sample preparation

One day after the ocular measurements, light-adapted animals were euthanized by intracardiac pentobarbitone injection (0.3 ml of Lethobarb, Virbac@ 325 mg per ml) after anesthesia was induced with isoflurane in oxygen. Eyes were rapidly enucleated and placed in medium (Tissue-Tek optimum cutting temperature, O.C.T.) and then frozen using liquid nitrogen and stored at −80 °C until used. Eyes were serially cryosectioned as previously described⁶⁰. Briefly, 12 μm thick unfixed cryosections were obtained from serially-sectioned FD-treated and contralateral control eyes sectioned across the whole posterior pole. Specimens were sectioned in either the vertical (SUP-INF) or horizontal (NAS-TMP) plane (Fig. 4a). Serial sections spanning within 4.6 mm of the vertical or horizontal midline of the ONH were mounted onto slides. Selected sections (2–3 sections per eye for each orientation) that contained the central axis, ONH midline, or the ONH nasal edge among vertical sections, or the ONH midline or inferior edge of the ONH (that also contained the central axis) among horizontal sections were analyzed. Prior to imaging with SAM, the slides were placed in a saline bath for 10 min at room temperature, ensuring that the samples were thawed and rehydrated. Degassed water was used as coupling medium between the transducer and the sample^60,105.

Fig. 4. Image segmentation and definitions of regions of interest (ROIs).

a Posterior view of an enucleated whole eye. The vertical and horizontal lines represent the section planes of superior-inferior and nasal-temporal meridians. b Definition of the regions. Red dot represents the midpoint of the optic nerve head. Black dot represents the central visual axis of the eye which was used as the origin of the quadrants. Proximal region (green area) was defined from the origin to 2.0 mm from the central axis. The distal region (yellow area) was from 2.0 mm to the end of the section (~5.0 mm) from the central axis. ONH was located approximately 0.9 mm from the NAS-TMP equatorial and 0.2 mm from the SUP-INF primary meridian. Inset represents the peripapillary area excluded from analysis (gray shaded area) that incorporates the ONH and the surrounding area within 0.6 mm in radius that is typically void of choroidal and scleral tissue layers. c ROIs were positioned on vertical (SUP and INF) and horizontal (TMP and NAS) SAM amplitude images starting ~0.5 mm from the optic nerve head midpoint. All tissues that were not sclera or choroid were manually segmented out and excluded from the amplitude images. ROIs were semi-automatically placed as described in the literature⁶⁰. ROIs inside the peripapillary area (grey circle) were excluded from analysis (detailed below). Scale bar = 500 µm. d Example of segmentation of boundaries in the INF in a control eye—SAM amplitude map with overlaid image of the same tissue section stained with hematoxylin and eosin (H&E). Green line is the outer extent of the sclera, the yellow line is the boundary between choroid and sclera, and blue line indicates Bruch’s membrane. e Cross-sectional image of the H&E histologically-stained section cut in the superior-inferior plane showing the direction of ultrasound interrogation with schematic illustration depicting key anatomic features contained within. Statistical analysis was performed on non-zero values contained within each ROI to mathematically omit all white regions, focusing our analysis on the non-luminal, stromal choroid. Pictorial illustration is based on Fig. 2 below stained section from Nickla and Wallman⁴⁶. INF inferior; ONH optic nerve head; NAS nasal; SAM scanning acoustic microscope; SUP superior; TMP temporal.

A preliminary investigation confirmed that the effect of the tissue freeze-thawing process on the acoustic properties of the tissue is statistically negligible. Briefly, a separate study investigated the effects of freezing and thawing on acoustic scattering parameters of whole, intact guinea pig eyeballs. This experiment evaluated acoustic scattering parameters which are partly associated with spatial variations of acoustic impedance; acoustic impedance is used for the calculation of bulk modulus and mass density¹⁰⁶. The eyeballs were scanned three times under different conditions: frozen and thawed, thawed at room temperature, and refrozen. Statistical analyses showed no significant differences in quantitative acoustic scattering parameters between scans, indicating minimal impact from the freeze-thaw process.

Scanning acoustic microscopy (SAM)

SAM data were obtained using a customized system as described previously^{70,105,107,108}. The SAM system employed a 250 MHz transducer (Fraunhofer IBMT, Sulzbach, Germany) with F-number 1.16, a 160 MHz bandwidth, 7 μm of lateral resolution, and a 72 μm depth of field. A 300-MHz monocycle pulser excited the transducer (GEOZONDAS, Vilnius, Lithuania), while radio-frequency (RF) echo signals were amplified (MITEQ, Hauppauge, NY, USA) and digitized at 2.5 GHz using a 12-bit oscilloscope (HDO6104, Teledyne Lecroy, Chestnut Ridge, NY, USA). The specimens were raster scanned in two dimensions at 2 μm-wide steps. The temperature was monitored using thermocouples (pre and post scan measurements). Custom LabVIEW (National Instruments, Austin, TX, USA) software controlled the SAM system.

Signal and data processing

2D maps of quantitative biomechanical properties were formed using a frequency-domain and model-based approach¹⁰⁸. Briefly, the Fourier transform of RF echo signals at each location was normalized by a reference signal. A model-based inverse method^108,109 operated on the normalized spectra to estimate sample section thickness (d), speed of sound (c), acoustic attenuation (α), and acoustic impedance (Z). Derived parameters, mass density (rho in g·cm⁻³), and bulk modulus (K in GPa) were calculated from direct measures of Z and c using first principles¹¹⁰, where rho = Z·c⁻¹ and K = Z*c. A larger rho signifies a denser tissue while a larger K value is associated with a less compliant (more stiff) tissue. In addition, 2D maps of signal amplitude (Fig. 4c, d) were formed by taking the maximum of the envelope of the RF signals at each scan location to provide a good visualization of tissue morphology.

Image segmentation

After the SAM measurement, specimens were stained with H&E (Fig. 4e) to allow digital co-registration with the SAM amplitude maps to differentiate tissue layers. The 2D parameter maps were divided into FD and contralateral fellow control eyes (control). In separate animals, measurements were taken in either the vertical (SUP and INF) or horizontal (NAS and TMP) meridian relative to the center of the optic nerve as previously described⁶⁰.

Two independent image-segmentation methods were applied to the 2D SAM images of the vertical and horizontal sections. First, the SAM image was automatically divided into regions of interest (ROIs) following to our previous study (Fig.4c) where 200 µm-wide sequential ROIs were automatically defined relative to the center of the central axis which approximately aligns with the optic axis (Fig. 4b). Second, the boundaries were manually drawn for the retina/choroid along the choroid basement membrane, and the sclera/choroid and outer sclera boundaries (Fig. 4d). After SAM imaging, cryosections were then stained with H&E, and the SAM amplitude maps were overlaid and co-registered on the corresponding histology image to differentiate the choroid from the sclera and retina and to draw the segmentation boundary lines. All manual segmentation lines were drawn by two qualified clinician-scientist graders and independently checked by another qualified senior grader (QVH) blinded to eye origin. Following manual segmentation, ROIs were split into choroidal and scleral layers for subsequent data extraction performed within MATLAB R2022a (The MathWorks, Inc., Natick, MA, USA) using the Image-processing toolbox.

Within ROIs, areas that contained sectioning artifacts or regions under the ONH (approximated as a 1.2 mm-diameter circle centered at the ONH center) were excluded from analysis based on the clinician-scientist’s evaluation. Since the choroidal layers consist of choroidal vessels (i.e., choriocapillaris), and the fill of choroidal vessel lumen may alter biomechanical properties, areas within ROIs that contained visible lumen (viewed at a 7 µm spatial resolution) were also excluded to focus on the biomechanical properties of the choroidal stroma.

Subsequently, each ROI was sorted into SUP, INF, NAS, and TMP regions defined relative to center of the central axis (Fig. 4b). Each of the parameters measured in the SUP, INF, NAS, and TMP ROIs were also averaged into those that fell into the proximal (≤2 mm from the center of the central axis) or distal (>2 mm from the center of the central axis) eccentricities in accordance with the previous report (Fig. 4b).

Data analysis

Within each ROI, the mean and standard deviation of each parameter value were calculated and plotted as a function of their eccentricity (distance along the eye wall curvature) from the center of the central axis. Note that because each sample was sectioned perpendicular to the posterior eyewall, the estimated parameters (K and rho) were independent of choroidal/scleral thickness of the eye.

Statistics and reproducibility

Independent t-tests were performed to determine statistical differences between the values among the various combinations of the following groups: FD versus control eyes, choroid versus sclera tissue, proximal versus distal regions, and vertical versus horizontal meridians. Paired t-tests were performed for the choroid versus sclera in control eyes, FD eyes, and FD Eyes-Control. One-way analysis of variance (ANOVA) was performed to determine statistical differences between quadrants (NAS, TMP, INF, SUP). Post hoc analysis was performed using Tukey-Kramer’s least significant difference (LSD). Correlation significance analysis was performed by Fisher’s z transformation. All statistical analysis were executed using MATLAB R2022a (The MathWorks, Inc., Natick, MA, USA) with Statistics and Machine Learning Toolbox.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Kazuyo Ito designed the study, conducted data pre- and post-processing, and wrote the manuscript draft. Cameron Hoerig co-designed the study, analyzed the data, reviewed the results, and wrote the manuscript. Yee Shan Dan wrote the manuscript. Sally A. McFadden, Jonathan Mamou, and Quan V. Hoang provided the initial project direction, co-designed the study, provided research funding, and wrote the manuscript. All authors contributed to the final manuscript.

Peer review

Peer review information

Communications Engineering thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Ros Daw and Mengying Su.

Data availability

The experimental data are available for research purposes from corresponding authors upon request. Requests will be fulfilled within 2 months.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s44172-024-00280-7.