Conformation of the anterior segment in human myopia

Topography of the in vivo anterior segment is of relevance in understanding its role in myopia and in the development of ocular surgical procedures. Using 3D magnetic resonance (MR) images of the human eye, regional variations in surface area (SA) and bulbosity of four anterior segment regions were investigated in association with refractive status (Rx), axial length (AL) and total ocular volume (OV).

were observed to be greater during accommodation in myopes, while Niyazamand et al. 5 reported that high myopes exhibited greater thickening with increasing levels of convergence. Posterior and equatorial eye shape are also known to alter in myopia, 16,17 but few studies that assessed the anterior surface shape suggested myopic eyes to be less asymmetric 6,18 and demonstrated greater nasal sagittal height. 18 Observations of larger anterior chamber depth and flatter corneal curvature in myopic eyes suggest a putative involvement of the anterior segment in myopia development and progression. [19][20][21] An accurate depiction of the anterior segment shape is of relevance to many areas of clinical work including scleral contact lens design, 22 intraocular lens geometry, 23,24 transscleral drug delivery, 25 intravitreal injections, 26 scleral collagen cross-linking, 27,28 as well as any surgical procedures involving the anterior sclera. 29,30 The accessibility of the anterior aspect of the eye is a major advantage, but technical challenges related to imaging through the sclera and determining suitably valid and sensitive metrics for identifying physical alterations in vivo pose another difficulty. Several investigators have attempted to model the anterior scleral shape via corneoscleral topography 5,18,22,31 and optical coherence tomography (OCT), 18,32,33 but these are limited to assessment of only the exposed scleral and corneal surfaces and imaging along discrete meridia and depth. 22 Restrictions as to how far posteriorly the in vivo anterior segment can be imaged with OCT has constrained assessment of the sagittal height to 15-mm chord lengths 22 and thickness measurements to 6 mm posterior to the scleral spur. 8,13 Moreover, effects of refractive distortion and tissue curvature and tilt further compound OCT imaging. 34 To evaluate comprehensively the topography of the anterior segment, an assessment of the three-dimensional (3D) eye shape in vivo is valuable. 35,36 Magnetic resonance imaging (MRI) is based on high-contrast delineation of the vitreo-retinal interface, and thus provides a determination of the internal surface of the eye. As an in vivo method of non-contact imaging, MRI allows depiction and investigation of the internal ocular globe. The utility of MRI in humans in evaluating ocular features such as retinal and posterior vitreous chamber shape, 18,37 linear dimensions, 38,39 surface area (SA) and curvature 40,41 and ocular and orbital volume. [42][43][44] has been demonstrated. Despite its use in these studies, 45,46 application of MRI has not been extended into the evaluation of structural characteristics within the anterior segment of the globe.
It is well documented that myopia results from the anterior segment components failing to compensate for the concurrent growth of the posterior vitreous chamber. 47,48 Specifically, the observation that the ciliary muscle has been identified as restricting the anterior processes governing crystalline lens flattening with axial length (AL) expansion may be the basis for myopic growth. 49,50 MR imaging of the anterior segment provides a unique opportunity to characterise the internal morphometry of the anterior segment without the effects of the overlying tissues nor the constraints of the anterior-posterior limits of OCT and topography. To encompass the region of the ciliary muscle 43,50 as well as previous work pertaining to the anterior segment structure, [6][7][8][9]18,23,24 a novel metric to assess MRI-derived SAs and bulbosity of four anterior segment quadrants was assessed. The influence of refractive status as well as associated AL and ocular volume (OV) on these structural characteristics was also investigated.

METHOD
Forty-three young adult subjects were scanned using a whole body MRI scanner (3-Tesla Trio, Siemens, sieme nshealt hinee rs.com). Ethical approval was obtained from Aston University Ethics Committee and the study was performed according to the tenets of the Declaration of Helsinki. Written informed consent was obtained from each subject prior to the commencement of the study. Eligibility to take part in the study was confirmed after subjects completed a screening questionnaire. Exclusion criteria were as follows: previous history of ocular surgery, trauma or pathology, ocular medication, astigmatism >1.75 D as well as standard criteria 51 for MR scanning. Individuals suffering from connective tissue-related disorders were also excluded due to their known effect on collagen composition and, hence, its probable effect on scleral biomechanics and eye structure.

Acquisition of MR images
Magnetic resonance images were acquired using an eightchannel phased-array head coil, which allows simultaneous scanning of both eyes with a high signal-to-noise ratio. T2-weighted scans were obtained using an imaging protocol that allowed high-definition delineation of the ocular surfaces. These scans were performed using a protocol reported previously. 17,42 Subsequently, the T2-weighted MR images (voxel thickness of 0.5 × 0.5 × 1.0 mm) produced by

Key points
• Topography of the in vivo anterior segment is of relevance in understanding its role in myopia and in the development of ocular surgical procedures. • Using 3D magnetic resonance images of the human eye, regional variations in surface area and bulbosity of four anterior segment regions were investigated. • Significant regional variation in surface area exists across the anterior segment that are modulated by refractive error and axial length.
this sequencing were optimised to allow discrimination of the eye's fluid-filled chambers from the rest of the head scan. Scanning required a distance refractive correction as subjects were asked to fixate on a distance target viewed through an angled mirror mounted on the head coil; any subjects who required such a refractive correction were fitted with best-sphere daily disposable contact lenses to facilitate accurate fixation. The average scanning time was 5 min and 40 s.

Analysis of MR images
Magnetic resonance images were analysed using a specially modified version of the freeware software, mri3dX. 17,52 Analysis of the T2-weighted image voxels was conducted as per previous work to produce 3D depictions of the eye shape. 42,43,52 Vector co-ordinates of the 3D surface polygons were designated as superior-temporal (ST), superiornasal (SN), inferior-temporal (IT) and inferior-nasal (IN) to determine the regional conformation of the globe. These divisions are based on a standardised co-ordinate system where the longitudinal axis of symmetry is projected from a line connecting the geometric centre of the eye to the anterior corneal pole. Although the procedure cannot account fully for the discrepancy between the visual and optical axis (i.e., angle alpha), this discrepancy can be considered relatively small for anterior regions of the globe.
To illustrate the morphological features of the globe, the 3D array of polygons was then collapsed to provide a 2D representation of each quadrant. Data were extracted initially for 15%-100% along the AL for all quadrants; 15% (approx. 3.5 mm posterior to the cornea apex) along the AL allows avoidance of the cornea and assessment of the globe from beyond the corneo-limbal junction while 100% denotes the most posterior aspect of the eye (Figures 1 and  2). This technique provides good correlation with partial coherence interferometry measurement of AL and allows repeatable and reproducible results. 52 To encapsulate the entirety of this region of special interest, the anterior section lies between ~3.5 and 9 mm, equivalent to approximately 15%-40% along the AL ( Figure 2). A second-order polynomial (Equation 1) was fitted to each quadrant for this range ( Figure 2). As AL is used as the reference for extracting the 15%-100% range of MRI data, inter-subject variability in AL influences the width of this range and, hence, affects the size of the anterior region (15%-40%) of interest. The study protocol presently adopted controlled, in part, for this source of error. Since the anterior 3.5 mm (15% along the AL) is likely to vary with AL, maintaining a constant end point at 9 mm (i.e., 40% along the AL) ensured that the extent of the region assessed remained relatively constant between subjects.
Surface area was generated by integrating and rotating the second-order polynomial about the x-axis. Dividing the segment area by four provided a measure of the relative SA for each quadrant (Equation 2). The coefficient of the x 2 polynomial offers a measure of the bulbosity of each respective quadrant.
The values a, b and c represent coefficient values for the polynomial.
The values a, b and c correspond to the polynomial equation values of y = ax 2 + bx + c.
The values l 1 and l 2 (in mm) correspond to the respective 15% and 40% x-axis values.

Biometric measurements
Refractive error was determined after inducing cycloplegia in each eye using 1 drop of tropicamide HCl 1% (Minims®, bausch.co.uk). Sufficient cycloplegia was designated as an amplitude of accommodation <2D. Objective measurements of the refractive error and central corneal curvature were determined with an infrared, binocular, open-view autorefractor/keratometer (Shin-Nippon SRW-5000, grand seiko.com). Five measurements were taken from each eye, which were averaged and converted to mean spherical error (MSE) (sphere power + 0.5 × cylinder power) and central corneal curvature (mm) readings. AL and anterior chamber

R ESULTS
Surface area and bulbosity were assessed in 43 healthy individuals (Table 1).

Surface area
Axial length was a significant covariate of SA (F 1,41 = 4.96, p = 0.03); hence, its effect was controlled on subsequent analysis. Significant differences were observed between  Analysis showed no main effect for refractive status (F 1,40 = 0.43, p = 0.52). However, a significant interaction effect was noted between refractive status and regional differences (

Correlates of surface area
For the combined myopic and non-myopic groups, (Table 3a) as well as the separate myopic group (Table 3b), a more myopic refractive error and longer AL were negatively associated with SAs for SN, IN and IT (Table 3a and b). In contrast, this relationship was not observed in the nonmyopic group (Table 3c). Although ST SA showed no association with refractive error or AL, it was the only region to show a strong positive association with OV for all groups (Table 3a-c). Among the combined group, mean corneal curvature was positively associated with SAs in the SN, IN and IT quadrants but only in the IT and SN regions for the myopic and non-myopic groups, respectively. No relationship was found between ACD and SAs.

Quadrant bulbosity
The curvature for each region was represented as bulbosity and indexed by the x 2 coefficient of the second-order polynomial over the ≈3.5-9.0 mm section along the axial length (Table 4). Bulbosity failed to show any significant regional differences (

DISCUSSION
This study is the first to explore the utility of MRI-derived measures of anterior segment SA and bulbosity in eyes of different refractive groups. Notably, these novel metrics provide an evaluation of the internal shape of the anterior segment rather than the external surface, thus mitigating the effects of overlying tissues such as the sclera, episclera, Tenon's capsule and conjunctiva. Regional differences in the SA between the ST versus IN, IN versus IT and SN versus IT quadrants were observed. Interestingly, other than IT versus IN, the SA differences seem to lie between the diagonal quadrants, that is, ST versus IN and SN versus IT, suggesting that most adjacent regions, both horizontally and vertically, are congruent. Anterior scleral shape is known to demonstrate rotational asymmetry, with the sagittal height of the temporal sclera being greater than the nasal and the superior region larger than the inferior. 22,53 The insertion site of the extraocular muscles may influence such structural characteristics. 32,33,54 For the temporal (7.0 mm) and superior (7.5 mm) meridia, the distance from the limbus to the insertion points is larger than for the nasal (5.5 mm) and inferior (6.5 mm) meridia. 55,56 It is possible that different biomechanical forces generated by the extraocular muscles across the various meridia may result in regional disparity in anterior segment growth. Heterogeneity of the internal dimensions of the anterior segment has also been observed previously in regards to the magnitude of the vertical and horizontal meridia. 23,24,57 Inconsistency in the internal dimensions of the anterior eye is likely to affect the overall conformation of the anterior segment and may further explain the variation in SAs found here. Such observations suggest that the anterior eye fails to exhibit the commonly assumed spherical shape, but, in fact, shows considerable regional variation in its morphometry.

F I G U R E 3 Estimated marginal means (± standard error bars) for
all four quadrants in the non-myopic group, myopic groups and for all subjects. Estimated marginal means provide the adjusted means after controlling for the covariate axial length for each group hence the effect of axial length was statistically removed.
Measurements of the internal conformation may provide a more accurate depiction of growth changes in the myopic eye, which may be a precursor to the observed external scleral shape 6,18 and thickness changes. 12,13 Refractive status was found to have a significant interaction with the regional differences, with smaller temporal and larger nasal SA in myopic eyes. Aside from the ST quadrant, all SAs demonstrated reduction with longer AL and myopic refractive error. It is unclear whether these structural characteristics are a precursor to or a consequence of myopia. There is little evidence in humans to suggest a genetic or environmental factor that may result in such asymmetry with myopia. Differential levels of anatomical restriction offered by the medial and lateral walls during ocular expansion anteriorly, especially near the orbital margin, may partly explain the findings. 58,59 Assuming a constant tissue volume, 60 the smaller SA with increasing AL and myopic refraction may be a subsequent effect of the predominant equatorial and posterior expansion, 17,61,62 which may lead to reduced tissue mass across the anterior regions. The possible propensity for greater globe expansion temporally 63-65 may partly explain why the temporal SA was smaller in myopic eyes. The ST SA uniquely failed to show any association with AL and refraction but demonstrated a consistent, positive relationship with OV, suggesting that this region may be a better indicator of overall globe expansion rather than longitudinal axial growth. Interestingly, the ST quadrant was observed to have the largest retinal SA and the higher incidence of retinal breaks. 41,66 Nagra et al. 44 noted that inter-quadrant differences in the retinal SA existed independent of AL, suggesting that the susceptibility for retinal tears may be more associated with the inter-quadrant differences rather than the absolute magnitude of the retinal surface. It is unclear how the anterior and posterior SA are related, but a biomechanical association may exist which might be modulated differently in myopic eyes.
Despite the findings of regional variation in anterior segment SA, no significant differences were observed between the bulbosity of different quadrants. Bulbosity is an index of the degree of protrusion of a particular element of shape, and hence, is independent of SA and volume. The present study did not assess curvature differences per se since bulbosity is a characteristic of curvature. Visualising the configuration of the anterior eye as an ellipse, having the same vertex curvatures yet different horizontal and vertical dimensions, with asphericity quantified as Q values may help to understand the dissociation with SA. Therefore, with regards to bulbosity, it is possible that anatomical constraints imposed by the surrounding tissue ensure uniformity between the different quadrants assessed. These observations suggest that despite the SA reducing with increasing AL, the actual bulbosity of the quadrant remains constant. Topographic observations of a more symmetrical anterior segment in myopic eyes support these findings. 5,18 Neither ethnicity, gender nor age group were found to have a significant influence on SA and bulbosity.
There are several limitations to this study. The MR image analysis protocol used here provides quadrant reference points based on the geometric (optical) axis, rather than the visual axis. The discrepancy between these two axes is known as angle α. While this angle is present along both the horizontal and vertical meridia, its magnitude is smaller vertically. Angle α is notoriously difficult to assess and T A B L E 3 Univariate correlations (Pearson coefficients) between regional surface area and ocular biometrics: (a) combined group (both myopic and non-myopic subjects) (b) only myopes (n = 23) and (c) only non-myopic subjects (n = 20). account for 67,68 but it can be assumed that for the anterior segment region assessed in this study, its effects should be minimal. Inter-subject variability in AL is likely to influence the initial 15%-100% of the reference data set extracted to determine the size of the anterior segment region (15%-40%). Although a fixed end point comprising 40% of the AL was used, there is a likelihood that the size of the anterior portion varied with AL. In an attempt to account for this potential variable, the statistical analysis controlled for AL as a covariate, hence minimising its influence on the outcomes.

CONCLUSION
Modelling of the eye with MRI provides an invaluable opportunity to investigate how eye shape and growth, both anteriorly and posteriorly, vary in myopic eyes. Observations of regional differences in anterior SAs that are modulated by refractive status suggest that structural and biomechanical changes may be evident across the anterior segment. It is unclear if these changes are an indicator or consequence of myopia, but further work is warranted to determine this. The anterior segment provides a clinically accessible site for the administration of scleral collagen cross-linking for myopia as well as surgical procedures relating to glaucoma and cataract. If refractive status affects the anatomical characteristics of the anterior eye, then improved knowledge of the optimal site for treatment is necessary. Herein, the novel application of anterior segment SA and bulbosity derived from MRI data provides an avenue for future exploration, specifically in relation to the posterior segment eye growth.

AC K N O W L E D G E M E N T S
Hetal Buckhurst received PhD fee waiver from Aston University. Surface area was calculated using a formula developed in collaboration with Dr. Bill Cox, Aston University. Dr. Manbir Nagra is acknowledged for assistance in the MRI and biometry data collection.

FU N D I N G I N FO R M AT I O N
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

CO N F L I C T O F I N T E R E S T S TAT E M E N T
The authors report no conflicts of interest and have no proprietary interest in any of the materials mentioned in this article.