Characteristics of myopic and hyperopic eyes in patients with antimetropia

ABSTRACT Clinical relevance Antimetropia is a rare type of anisometropia in which one eye is myopic and the fellow is hyperopic, This optical condition condition permits the evaluation of both sides of the emmetropisation process failure in the same individual by minimising genetic and environmental factors. Background This study aimed to evaluate the ocular biometric, retinal, and choroidal characteristics of myopic and hyperopic eyes of antimetropic subjects older than six years. Methods In this retrospective study, myopic and hyperopic eyes of 29 antimetropic patients with a spherical equivalent (SE) difference of at least 2.00D between the eyes were included. Axial length (AL), mean corneal keratometry, anterior chamber depth, the proportion of anterior chamber depth in AL, crystalline lens power, central macular thicknesses, disc-to-fovea distance, fovea-disc angle, peripapillary retinal nerve fibre layer (RNFL) thicknesses, and subfoveal choroidal features were compared between the eyes. The prevalence of amblyopia was determined. Refractive parameters and total astigmatic profile were evaluated in eyes with and without amblyopia. Results The median absolute SE and AL differences between the eyes were 3.50D (interquartile range:1.75) and 1.18 mm (interquartile range:0.76), respectively (p < 0.001). Myopic eyes had lower crystalline lens power and proportion of anterior chamber depth in AL, and longer disc-to-fovea distance. Macular thicknesses, global RNFL, and temporal RNFL were thicker in myopic eyes, and there was no difference in other RNFL quadrants. Despite the increase in the choroidal vascularity index, other choroidal parameters were decreased in myopic eyes. Amblyopia was found in three of the myopic eyes and seven of the hyperopic eyes (p = 0.343). The highest interocular SE and AL difference and the highest frequency of anisoastigmatism were observed in patients with amblyopia in the myopic eye. Conclusion Each ocular structure may respond differently to, or may be affected differently by, ametropic conditions.


Introduction
Emmetropisation is a vision-dependent process in ocular structures that targets to focus the image on the fovea, the mechanism of which has not been fully elucidated. 1 Generally, young infant eyes have a hyperopic refractive error, and the ocular changes that are effective in the emmetropisation process play a role in focusing the image on the retina after birth. 2 Cornea, axial length (AL), and crystalline lens play an active role in this process. 3However, due to environmental and genetic factors, the emmetropisation process may not result as desired and refractive errors, in which the image is not focused on the retina, namely ametropia, may occur. 4metropic conditions, particularly observed after six years old, are considered to be the failure of emmetropisation. 5In hyperopia, the image is focused behind the retina due to the insufficiency of the emmetropic changes.In myopia, the image is focused in front of the retina, as the ocular changes that lead to emmetropia continue even after retinal focus is achieved.Thus, hyperopia and myopia represent two opposite sides of the emmetropisation process anomaly.
Most studies investigating ocular changes associated with refractive errors have included hyperopic and myopic subjects with different genetic and environmental characteristics, which may lead to conflicting results. 6,7[10] In most of these anisometropic subjects, either both eyes have a similar ametropic condition, or the fellow eye is emmetropic.In antimetropia, which is a rare type of anisometropia, one eye is hyperopic, and the fellow eye is myopic. 8ntimetropic individuals may be considered an eligible study population as they minimise the effects of environmental and genetic factors affecting ocular structures and allow for the isolated evaluation of changes in ocular structures caused by hyperopia and myopia.
The aim of this study is to evaluate ocular biometric parameters, as well as retinal and choroidal features, that may be associated with refractive error in the myopic and hyperopic eyes of antimetropic subjects older than six years old.

Study population and determination of refractive error
This study was a retrospective analysis of consecutive eligible patients diagnosed with antimetropia at a tertiary ophthalmology centre between September 2018 and January 2022.Antimetropia was defined as cycloplegic spherical equivalent (SE) of ≤ −0.75 D in the myopic eye and ≥ +0.75 D in the hyperopic eye, as well as a SE difference of at least 2.00 D between the eyes.Patients with high (>26.00mm) or pathologic myopia, manifest strabismus, history of previous ocular surgery or trauma, history of neurological or systemic diseases, additional ocular pathologies such as age-related macular degeneration, glaucoma, and lenticular or corneal pathologies that might affect the refractive error were excluded.Ethical approval was obtained from the institutional ethics committee, and informed consent was obtained from all participants/legal guardians.The procedures used in this study adhere to the tenets of the Declaration of Helsinki.
Non-cycloplegic refractive error was measured with the Nidek ARK 560A autorefractor-keratometer. Prior to determining the best-corrected visual acuity (BCVA) for each eye, manifest refraction was performed using a Snellen chart.Cyclopentolate hydrochloride (1%) was administered to all eyes three times with an interval of 5 minutes, autorefraction was repeated after 30 minutes, and cycloplegic refractive error was determined.BCVA measurements were converted to logMAR equivalents, and eyes with a BCVA worse than 0.10 logMAR were considered amblyopic.Cycloplegic SE (spherical refraction + [0.5 × cylindrical refraction]) and total astigmatic error were recorded.Anisoastigmatism was defined as an absolute difference of total astigmatic error between the eyes of ≥1.00 D, and the presence of ≥1.00 D and ≥2.00 D anisoastigmatism in patients was determined.

Measurement of ocular biometric parameters and calculation of crystalline lens power
Axial length, anterior chamber depth, mean keratometry reading and mean corneal radius were measured by noncontact partial coherence laser interferometry (IOL Master 500; Carl Zeiss, Meditec, Germany).The AL-to-corneal radius ratio was calculated as the AL divided by the mean corneal radius.The following formula was used to determine the percentage of anterior chamber depth relative to the AL: Anterior chamber depth x (100/Axial length) Crystalline lens power was calculated with the modified Bennett-Rabbetts formula detailed in Appendix 1. 11

Determination of macular and optic disc parameters
Retinal and choroidal imaging was performed using Heidelberg Spectralis spectral domain-optical coherence tomography (SD-OCT) (Heidelberg Engineering, Inc., Heidelberg, Germany).Prior to the retinal SD-OCT imaging, the mean corneal radius was entered into the built-in Heidelberg Eye Explorer software (HEYEX; Heidelberg Engineering) following patient registration, and the cycloplegic SE was set using the focus knob during imaging.
The mean central macular and the minimum foveal thicknesses were automatically measured with a 25-line raster protocol.Optic disc-to-fovea distance was determined by measuring the distance between the fovea and the temporal border of the optic disc on the near-infrared retinal en-face image by the masked ophthalmologist (CE) using the built-in calliper function.Global, temporal, superior, nasal, and inferior peripapillary retinal nerve fibre layer (RNFL) thicknesses were automatically measured using a disc-centred 12° circumpapillary RNFL scan.The fovea-to-optic disc axis orientation was automatically measured with the built-in fovea-disc alignment function in Spectralis SD-OCT and recorded as the fovea-disc angle.
Centralisation of the optic disc, automatically determined foveal location, and peripapillary RNFL segmentation in the RNFL scanning protocol was checked by a masked observer (CE).The foveal location was re-localised manually in the images in which the foveal location was automatically incorrectly determined.In patients with incorrect disc centralisation, lack of patient compliance during imaging, or RNFL segmentation error, RNFL thickness measurements and fovea-disc angle degrees of the patients, were not included in the analysis.

Measurement of choroidal parameters
A 30° horizontal enhanced depth imaging SD-OCT line scan (100 frames) passing through the centre of the fovea was utilised for choroidal imaging.The outer border of the choroidal stroma was defined as the choroidoscleral interface, and all choroidal measurements were performed by adhering to this border.Subfoveal choroidal thickness was measured from Bruch's membrane to the choroidoscleral interface using a calliper function.The method described by Agrawal et al. was used to determine the structural features of the 1500 μmwide nasal-to-temporal subfoveal choroidal area. 12Briefly, selected choroidal area was binarised by the Niblack autolocal thresholding method.After the binarisation, dark pixels represent the luminal area, and white pixels represent the stromal area (Figure 1).The total choroidal area, luminal area, stromal area (= [Total choroidal area] -[Luminal area]), and choroidal vascularity index (=[Luminal area]*100/[Total choroidal area]) were determined.Choroidal thickness measurements and binarisation processing of enhanced depth imaging SD-OCT images were performed by a masked observer (CE).Intrarater and inter-rater reliability were evaluated for choroidal parameters (detailed in Supplementary Material 1).
Axial length-based ocular correction was applied to all retinal and choroidal measurements except the fovea-disc angle to eliminate axial length-dependent transverse and axial magnification errors.Ocular adjustment using the Littmann-Bennett's formula was detailed in Appendix 2. 13,14

Statistical analysis
Statistical analyses were performed using SPSS statistical software (SPSS Inc, version 22, Chicago, IL).The number and percentage of baseline categorical variables of the participants were presented, and all continuous variables were expressed with median and 25th percentile − 75th percentile (Q1 -Q3).Laterality was assessed by Pearson's chi-square test, and presence of amblyopia was assessed by McNemar's test.Wilcoxon signed-rank test was used to compare ocular parameters between myopic and hyperopic eyes.
Patients were classified into three subgroups according to amblyopia status and the ametropia type of the amblyopic eye: amblyopia in the myopic eye, amblyopia in the hyperopic eye (both amblyopic groups were subgroups of Amblyopia (+)), and Amblyopia (-).The SE, AL, and total astigmatic error of the patients were compared with the Kruskal-Wallis test among the amblyopia based-subgroups.Post-hoc analysis was performed with the Mann-Whitney test with Bonferroni's correction.
The frequencies of ≥1.00 D and ≥2.00 D anisoastigmatism in the subgroups were evaluated with Fisher-Freeman-Halton Exact test, and p-values for pairwise comparisons were calculated according to the adjusted standardised residuals (Z-scores).The p-value <0.05 was considered statistically significant.

Results
Fifty-eight eyes (29 myopic, 29 hyperopic eyes) of 29 subjects with antimetropia were included in this study.The demographic and refractive characteristics of all participants have been summarised in Table 1.Of all subjects, 58.6% were female, and the mean age was 21.10 ± 9.57 years.The right eye of 20 subjects was myopic (p = 0.004).The median absolute SE and AL differences between the eyes were 3.50 D (interquartile range [IQR]: 1.75) and 1.18 mm (IQR: 0.76), respectively (p < 0.001).There was no significant difference in total astigmatic error (p = 0.466).
Biometric, adjusted-retinal, and adjusted-choroidal parameters of myopic and hyperopic eyes are summarised in Table 2. Regarding the biometric measurements, AL and AL-  to-corneal radius ratio were higher in myopic eyes, and the proportion of anterior chamber depth in AL and crystalline lens power were higher in hyperopic eyes (p < 0.001, for all).
There was no significant difference in anterior chamber depth and mean keratometry reading between the eyes.
Adjusted minimum foveal and central macular thicknesses were thicker, and disc-to-fovea distance was longer in myopic eyes (p < 0.001, for all).Peripapillary RNFL thicknesses and fovea-disc angle were compared in 50 eyes of 25 patients due to errors/deficiencies in optic disc imaging in four participants.Adjusted global and temporal RNFL thicknesses were found to be greater in myopic eyes (p = 0.017, and p = 0.001, respectively).There was no difference between the eyes in other peripapillary RNFL thicknesses and the fovea-disc angle.
Excellent intra-rater and inter-rater agreements were found for all choroidal measurements (see Supplementary Material 1).Choroidal thickness and total choroidal, luminal, and stromal areas were higher in hyperopic eyes, and choroidal vascularity index was higher in the myopic eyes (p < 0.001, p:0.004, p < 0.001, and p < 0.001, respectively for adjusted measurements).Unadjusted retinal and choroidal measurements have been presented in Supplementary Material 2.
Although amblyopia was detected in the myopic eye of three patients and the hyperopic eye of seven patients; the difference was not statistically significant (p = 0.343).No patients had amblyopia in both eyes.Subgroup analyses between the myopic and hyperopic eyes of patients with amblyopia in the myopic eye, amblyopia in the hyperopic eye, and Amblyopia (-) are summarised in Table 3.Although The interocular difference was calculated by subtracting the hyperopic eye from the myopic eye.§ : Indicates ocular magnification-adjusted measurements using the Littmann-Bennett's formula.
myopic eyes with amblyopia had lower SE than all myopic eyes without amblyopia, hyperopic eyes with amblyopia had higher SE than hyperopic eyes only in the Amblyopia (-) subgroup.
Patients with amblyopia in any eye were found to have a higher absolute median interocular SE difference than patients without amblyopia.In the analysis regarding the presence of anisoastigmatism and the total astigmatic error in amblyopic and non-amblyopic eyes, in all patients with anisoastigmatism in the Amblyopia (+) groups, the dioptric power of the total astigmatic error in the amblyopic eye was lower than in the non-amblyopic fellow eye.≥2.00 D anisoastigmatism was detected in only two patients; both patients had amblyopia in the myopic eye (p < 0.001).

Discussion
The occurrence of hyperopia due to the lack of an emmetropisation process in one eye, and the development of myopia due to the continuation of the change process even after emmetropisation is achieved in the other eye, suggests that each eye has a separate and independent local ocular growth control. 8Clarifying the questions of which ocular structures play a role in the occurrence of ametropia and which compensatory mechanisms act to provide emmetropia at the age of six years and later is crucial in terms of revealing the pathophysiology of refractive errors.The present study contains the largest antimetropic case series in the literature and evaluated biometric, retinal, and choroidal parameters of myopic and hyperopic eyes in antimetropic subjects.
In this study, mean keratometry readings did not differ between eyes, whereas crystalline lens power decreased in myopic eyes.0]15 The other main ocular structures effective in the emmetropisation process are the cornea and the crystalline lens. 3Controversial results have been reported in previous studies comparing keratometry reading in non-anisometropic subjects with different types of spherical errors.][18] Contradictory results have also been reported in anisometropes.
In the present study, there was no significant interocular difference in mean keratometry reading, similar to that in the antimetropic subgroup of the previous anisometropia study. 10These results suggest that the cornea, whose refractive structure changes to provide emmetropisation in the first years of life, does not have a significant impact on the elimination of hyperopia, especially in patients older than six years old, and does not contribute to the mechanisms of emmetropisation in the late process.
A precise determination of the refractive power of the crystalline lens is possible only in-vitro.However, if the refractive error and ocular biometric parameters are known, it can also be estimated in-vivo using appropriate formulae. 11In this study, crystalline lens power was significantly reduced in myopic eyes.Similarly, previous studies have shown that there is a negative correlation between AL and crystalline lens power, and myopic eyes have lower crystalline lens power. 19,20Although it has been reported that crystalline lens power behaves differently in children and adults depending on the type of refractive error, in the current study, crystalline lens power was found to be lower in all myopic eyes under cycloplegic effect. 19This finding suggests that the crystalline lens power changes depending on the effect of stretching of the crystalline lens as a result of the expansion of the globe on the coronal plane due to myopia, rather than attempting to achieve emmetropisation in individuals after six years of age.In previous studies, this alteration of the crystal lens was mentioned as a "passive mechanism". 20,21 common finding of previous studies is that a short AL is associated with a narrow anterior chamber, which is why it is often argued that there will be a narrower anterior chamber in hyperopic eyes.15,17 In this study, although the anterior chamber depth did not show a significant difference between the eyes, the proportion of anterior chamber depth in AL was found to be lower in myopic eyes.These findings suggest that the alteration in anterior chamber depth does not occur proportionally to axial elongation and that elongation of the vitreous cavity plays an essential role rather than the change in anterior chamber depth in the myopic process that develops due to the axial elongation.
In this study, although thicker minimum foveal and central macular thicknesses were obtained in myopic eyes after ocular adjustment, no significant interocular difference was found in unadjusted values.3][24] A published meta-analysis determined that central macular thickness and minimum foveal thickness differed significantly only in high myopic eyes compared to control eyes. 7Wu et al. explain the foveal thickening in high myopic eyes by foveal elevation caused by the tendency of the inner limiting membrane to stretch and flatten and the centripedal force of the posterior vitreous. 25In the interocular comparison of the foveal position, longer disc-to-foveal distance was detected in myopic eyes.In contrast, fovea-disc angle, which indicates the vertical axis position of the fovea relative to the optic disc, did not differ between eyes.Previous studies have shown that fovea-disc alignment angle is independent of axial elongation, whereas disc-to-fovea distance is positively correlated with AL. [26][27][28] The variability of peripapillary RNFL thickness due to refractive error has been assessed in various studies. 7,24,26,29he meta-analysis by Salehi et al. 7 suggested that all peripapillary thicknesses, except temporal RNFL, decreased in moderate and high myopic eyes compared to control eyes, whereas temporal RNFL increased significantly in high myopia.However, there are also studies demonstrating that peripapillary RNFL thicknesses may not be significantly different or may be thicker in myopic eyes after axial length-based ocular magnification compared to emmetropic and/or hyperopic eyes. 30,31n this study, unadjusted global, nasal, and inferior RNFL thicknesses were found to be thinner in myopic eyes, but after ocular adjustment, global RNFL was thicker in myopic eyes, whereas the interocular differences were eliminated in the nasal and inferior quadrants.Although the current study did not include subjects with high myopia, both adjusted and unadjusted temporal RNFL thicknesses were found to be significantly thicker in myopic eyes.Isolated thickening of the temporal RNFL in myopic eyes has been explained by the movement of the inferotemporal and superotemporal retinal nerve fibre bundles to the temporal quadrant as a result of axial elongation. 7,32In addition, the absence of interocular differences in RNFL and macular thicknesses or the thinner measurements in the myopic eye can also be explained by differences in the scan size produced by the Spectralis SD-OCT between the eyes. 31,33umerous studies have reported that myopia is associated with thin choroidal thickness. 8,24,34Similarly, a significantly thin choroid was observed in myopic eyes in the present work.In studies examining choroidal structures, conflicting results have been reported regarding the refractive errors.6][37] This study revealed that luminal, stromal and total choroidal areas were higher in hyperopic eyes than in myopic eyes, consistent with previous findings.However, the choroidal vascularity index was lower in hyperopic eyes than in myopic eyes due to a relatively more significant decrease in stromal area than luminal area.
Although the strength of this study was the evaluation of the ocular effects of different refractive error types by comparing both eyes of the same individual, statistical comparisons could not be performed between the amblyopic and non-amblyopic eyes of the same subjects due to the insufficient number of patients with amblyopia.However, to present the basic refractive characteristics in the myopic and hyperopic eyes of subjects with and without amblyopia in the study population, comparisons of SE, AL, and total astigmatic error were performed between the amblyopia-based subgroups.
Previous studies have reported that the risk of amblyopia may increase with the increase in the difference in interocular refractive errors. 38Consistent with previous studies, it was determined that the absolute interocular SE difference in both amblyopia (+) subgroups was higher than in the amblyopia (-) group.9][40] Consistent with previous findings, the frequency of amblyopia was higher in hyperopic eyes, and the median SE of the hyperopic and myopic eyes with amblyopia were 2.38 (2.19-3.25)D and −3.63 (−4.69 -−3.19)D, respectively.
While the results of these subgroup analyses with a small number of amblyopic patients should be interpreted with suspicion, the present findings seem compatible with previous studies.It has been previously reported that anisoastigmatism may also cause amblyopia. 40,41According to the results of this study, the higher frequency of anisoastigmatism in patients with amblyopia in the myopic eye and a lower dioptric power of total astigmatic error in the myopic eyes with amblyopia suggest that anisoastigmatism accompanying anisomyopia may be associated with an increased risk of amblyopia.
This study has some limitations, most notably the small sample size; however, antimetropia is a rare ocular entity owing to its low prevalence (up to 0.1%). 8Additionally, the magnitude of astigmatism in each group was not presented.
Therefore, the components of astigmatism (e.g., corneal, internal) in the groups could not be identified, and a vectorial analysis of astigmatism could not be performed.
Another limitation is that data on subjective symptoms, such as aniseikonia, have not been evaluated.Previous studies of anisometropia have commonly reported that the left eye is hyperopic, and the right eye is myopic. 17,42,43Ocular dominance is highlighted as an elucidating mechanism for this situation.Although ocular dominance was not evaluated in this study, myopia was significantly higher in the right eye, consistent with previous publications.
Various ocular differences and similarities have been demonstrated in this study; nevertheless, it is not clearly known whether the ocular features detected in antimetropes reflect the ocular characteristics of patients with ametropia encountered in the general population.Further studies evaluating retinal focus/defocus-mediated ocular growth in patients with antimetropia may help clarify this suspicion.

Conclusion
In the evaluation of ocular parameters of myopic and hyperopic eyes of antimetropic subjects over six years old, it is concluded that the mean corneal power has not undergone any change for the elimination of the refractive error, and the crystalline lens power decreased in myopic eyes with the effect of possible passive mechanisms.Moreover, it has been revealed that refractive errors might cause differences in the distribution of the peripapillary RNFL, choroidal structure, and foveal location.Further studies with larger case series are needed in which characteristics such as astigmatic profile, aniseikonia and dominant eye laterality are also evaluated.

Figure 1 .
Figure 1.Enhanced depth imaging SD-OCT image of an antimetropic patient with hyperopia in the right eye (A) and myopia in the left eye (C).Overlay images (B, D) of the region of interest (with a width of 1500 µm fovea-centred total choroidal area) on the enhanced depth imaging SD-OCT scan according to the horizontal direction.Following the binarisation, white pixels (the area bordered by white pixels in Figure 1B and 1D) indicate the luminal area.

Table 1 .
Demographic and refractive characteristics of antimetropic subjects.