Article Text
PDF
Abstract
Aims To quantitatively measure retinal curvature (RC) in children with myopia and explore its association with refractive status.
Methods This cross-sectional study included participants aged 5–18 years who underwent comprehensive ocular examinations, including cycloplegic refraction and macula 24×20 mm optical coherence tomography (OCT) scans. RC was derived from OCT data using a three-dimensional reconstruction system. Mean RC was assessed in concentric circles (RC I–VI) with diameters of 1, 3, 6, 9, 12 and 15 mm around the fovea, as well as in four orientations (RC S/I/N/T).
Results A total of 443 eyes were included in the analysis. The values from RC I to RC VI were 0.51±0.19, 0.53±0.19, 0.62±0.19, 0.76±0.23, 0.86±0.23 and 0.81±0.18 10−2mm−2, respectively. RC I exhibited the smallest curvature, while RC V displayed the highest (p<0.001). High myopia (HM) group demonstrated larger RC I and smaller RC III/IV/V/VI compared with low myopia (LM) group (p<0.01). Significant differences among RC S/I/N/T were observed in HM group (pairwise comparison, p<0.001), but not in LM group. Multiple regression analysis revealed that age, sex, corneal curvature radius and subfoveal choroidal thickness (SFCT) were associated factors with foveal RC, while age, SFCT and axial length (AL) were associated factors of peripheral RC.
Conclusion RC can quantitatively characterise retinal shape and the morphological changes induced by myopia. Myopia progression results in a bulging macular retina accompanied by a flattening peripheral retina in children, and also increases the irregularity among the four quadrants. Age, AL and SFCT are associated factors of RC.
- imaging
- optics and refraction
Data availability statement
Data are available on reasonable request. Part of our data has been uploaded as supplementary information. Other data are available on reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Myopia, a prevalent ocular condition affecting billions of people worldwide, is closely linked to the shape of the eyeball; however, current assessments of myopia do not sufficiently provide information regarding eyeball morphology.
WHAT THIS STUDY ADDS
Myopia induces significant changes in retinal morphology, even among children.
These changes are manifested by the protrusion of the foveal retina and flattening of the peripheral retina, accompanied by increased asymmetry among different orientations.
Age is the primary factor driving alterations in retinal morphology, while choroidal thinning also leads to a protuberant retina, and axial elongation results in the flattening of the peripheral retina.
Retinal curvature serves as an objective and accurate measure to describe retinal morphology across a wide range.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Retinal curvature measurements offer a convenient means to monitor changes in retinal morphology.
Incorporating optical coherence tomography-based retinal curvature parameters into the monitoring and clinical management of myopia in children should be considered.
Retinal curvature offers a novel method for research into ocular morphological abnormalities such as posterior staphyloma, potentially providing new insights into their diagnosis and follow-up care.
Introduction
Myopia is a widespread ophthalmic disorder affecting billions of people worldwide.1 2 It is predicted that nearly half of the global population will be afflicted with myopia by 2050, with approximately 10% suffering from high myopia.3 As myopia advances, besides a decline in uncorrected visual acuity, the risks of sight-threatening complications increase,4 5 leading to a deterioration in daily life quality.6–8
The progression mechanisms of myopia are still not entirely elucidated; current theories suggest a close association between myopia and the shape of the eyeball.9 10 In this regard, understanding the overall eye morphology is crucial for prognosis and management. Currently, myopia assessments mainly include spherical equivalent (SE) and axial length (AL), which provide information about refractive status but offer limited insight into other myopia-related changes.
Retinal curvature (RC), a quantitative metric characterising retina steepness, provides new insights for deeper research into ocular morphology. This study aims to analyse RC in children with myopia using ultra-widefield swept-source optical coherence tomography (UWF SS-OCT) and an integrated three-dimensional (3D) reconstruction system, quantitatively characterise the retinal morphology in children with myopia using RC and explore its influencing factors.
Materials and methods
Study subjects
Participants were randomly selected from three schools (one primary, one middle and one high school) in Shanghai. Examinations were conducted between October and December 2023. Inclusion criteria included age 5–18 years, best-corrected visual acuity (BCVA) ≤logMAR 0.1 and cycloplegic SE ≤−0.50 D. Exclusion criteria included non-contact optical tension (NCT) ≥21 mm Hg and other ocular or systemic diseases.
Ophthalmic examinations
BCVA was assessed using a standard logarithmic chart. AL was measured with the IOL Master 700 (V.5.02, Carl Zeiss Meditec, Germany). Cycloplegia was induced with topical anaesthesia using 0.5% proparacaine (Alcaine, Alcon), followed by the administration of 1% cyclopentolate hydrochloride (Cyclogyl, Alcon). Cycloplegic refraction was performed using an autorefractor (KR-8900, Topcon, Tokyo, Japan).
OCT angiography examinations
Macula 24×20 mm OCT scans were obtained using a UWF SS-OCT system (BMizar, TowardPi Medical Technology, China), with a scanning speed of 400 000 A-scans per second, an axial resolution of 3.8 µm and a lateral resolution of 10 µm. Each 24×20 mm OCT scan consisted of 1536 A-scans (lateral) by 1280 A-scans (vertical). Subfoveal choroidal thickness (SFCT) measurements were performed using an auto-segmentation tool within the OCT system.
Retinal curvature measurement
Using integrated auto-segmentation and retina reconstruction software, Bruch’s membrane was identified and isolated, resulting in a fitted 3D retinal model based on 1536×1280 data points obtained from OCT scans. Each point’s Gaussian curvature on Bruch’s membrane was calculated as RC. To ensure accurate RC measurements, several steps were taken before scanning, including removing motion artefacts from the OCT images and manually checking and correcting Bruch’s membrane if necessary after auto-segmentation. AL was used to correct the SS-OCT images to the true geometry of the objective eye.
Mean RC was assessed in concentric circles (RC I to RC VI) with diameters of 1, 3, 6, 9, 12 and 15 mm around the fovea. Mean RC of four quarter-circle regions (RC S/I/N/T) with a 9 mm diameter centred on the fovea were also calculated. The retina model and RC analysis of the right eye of a girl aged 12 years are shown in figure 1.
Three-dimensional reconstructed retina model and retinal curvature (RC) measurement of a right eye. (A) Retinal model of the right eye of a girl aged 12 years with low myopia. (B) RC (10−2mm−2) was measured using a distensible Early Treatment Diabetic Retinopathy Study (EDTRS) grid, with diameters of 1, 3, 6, 9, 12 and 15 mm around the fovea. (C) RC I to RC VI represented the mean value of RC of serial concentric circles areas, which had been labelled using different colours. (D) RC S/I/N/T represented the mean value of RC of four quarter-circle areas with a 9 mm diameter centred on the fovea, which had been also labelled using different colours.
All collected OCT images and reconstruction results were inspected by researchers. Images with poor quality (score ≤8) were manually excluded from the analysis.
Repeatability test
For the first 10 eyes, the scanning and modelling process was repeated to assess the repeatability of the RC measurement method. Each subject’s eye underwent four consecutive OCT scans, and one additional scan was performed at the same location 1 day later.
Data analysis
Data analysis was performed using R software (V.4.2.3). Quantitative data were presented as mean±SD, while categorical data were expressed as percentages. The normality of continuous variables was assessed via QQ plots or the Shapiro-Wilk test. All measured quantitative parameters approximated a normal distribution.
Comparisons between categorical and continuous variables were conducted using the χ2 test or t-test. Analysis of variance (ANOVA) and intraclass correlation coefficient (ICC) were used to assess repeatability. Pearson’s correlation and regression analysis were applied to study associations between RC and myopic parameters.
Significance was defined as a p value <0.05 in a two-tailed test.
Results
Demographic characteristics
This study included 251 participants, totaling 489 eyes. Thirteen participants, due to non-cooperation, completed examinations for only one eye, while the remainder examined both eyes. The participants with data from only one eye and those who were excluded did not show significant demographic differences from the general participants. After excluding inadequate images, 443 eyes (90.6%) remained. Of these, 309 eyes were categorised into the low myopia (LM) group (−6.00 D<SE ≤−0.50 D) and 134 eyes into the high myopia (HM) group (SE ≤−6.00 D).11 Further details and characteristics of the HM/LM groups are outlined in table 1.
Baseline characteristics in different groups
Repeatability test
The repeated measurements for 10 eyes are presented in online supplemental table S1, including foveal (RC I) and peripheral (RC V) RC. The largest SD observed was 0.053 10−2 mm−2, not exceeding one-tenth of the average RC. There were no statistically significant differences between the five repeated measurements of RC I or RC V using ANOVA (p>0.99 in both cases), with F values of 0.012 and 0.026, respectively. Consistency analysis revealed ICC (A,1) values of 0.975 and 0.973 for RC I and RC V, with 95% CI of (0.940 to 0.993) and (0.934 to 0.992), respectively.
Supplemental material
Retinal curvature measurement
For children with myopia, the average RC across the entire measurement area was 0.685±0.296 10−2 mm−2. The results of RC for different ranges and orientations are summarised in table 2. RC of various ranges and orientations are presented in figure 2. In comparison with LM group, HM group had a larger RC I (p<0.05) and smaller RC III (p<0.01), IV (p<0.001), V (p<0.001) and VI (p<0.001). Significant differences were observed in RC across all four orientations between HM and LM groups (RC S/I/N: p<0.001, RC T: p=0.011). In the LM group, no significant difference was observed in the mean RC among different orientations. In the HM group, RC S was found to be smaller than RC I/N/T (p<0.001), and RC N was found to be smaller than RC I (p<0.001) and RC T (p<0.001).
RC of different measure ranges and orientations in each group
(A) Retinal curvature (RC) of different measure ranges in total sample. (B) RC of different measure ranges in high myopia (HM)/low myopia (LM) groups. (C) and (D) RC of different orientations in HM/LM groups. The unit of RC was 10–2 mm–2. **P<0.01; ***p<0.001; ****p<0.0001.
Correlation between retinal curvature and myopic metrics
The result of Pearson’s correlation between RC and age, SE, AL, corneal curvature radius (CR) and SFCT are illustrated in online supplemental figure S1. RC I and RC II, closer to the fovea, exhibited a negative correlation with SE and a positive correlation with age, whereas RC III–VI demonstrated a positive correlation with SE and a negative correlation with age. RC I showed a positive correlation with AL, while RC II–VI further away from the fovea exhibited a negative correlation with AL. CR demonstrated a negative correlation with RC I–VI. Regarding SFCT, RC I–III exhibited a negative correlation, while RC IV–VI exhibited a positive correlation.
Scatter plots of AL/age/SFCT and RC across different ranges are presented in online supplemental figure S2–S4.
Influencing factors of retinal curvature
RC I and RC V were selected to represent the foveal RC and peripheral RC for regression analysis. Univariate linear regression was initially conducted to explore the correlation between RC I, RC V and variables including AL, SFCT, total retinal thickness, inner retinal layer thickness (from the retinal nerve fibre layer to the inner plexiform layer), CR and demographic factors such as age, sex and body mass index.
Variables with a p value ≤0.05 were subsequently included in multivariate linear regression, and the results are presented in table 3. The model for foveal RC had an adjusted R2 of 0.166, and p value <0.001. The model for peripheral RC had an adjusted R2 of 0.407, and p value <0.001. Age and SFCT demonstrated significant regression coefficients with both foveal and peripheral RC. Additionally, being female and having a smaller CR were associated with a larger foveal RC, while a longer AL was associated with a smaller peripheral RC.
Multivariate linear regression models of foveal and peripheral RC
Variance inflation factor (VIF) was calculated to examine the multicollinearity of our models. All calculated VIFs for the variables were <2.0, with the highest value being 1.951. This indicates that there is no significant multicollinearity among the variables, and it does not affect the stability of the regression coefficients.
Discussion
Myopia is a complex ocular disorder involving the entire eye, which is closely related to the shape of the eyeball. Currently, there is a lack of effective methods to assess and compare eyeball morphology. This posed great difficulties in the diagnosis and follow-up of diseases such as PS. Previously, measurements of the shape of the posterior eye had been conducted using MRI and relative peripheral refractive errors.12–14 While MRI remains the gold standard for morphology study, it is too costly and time-consuming to conduct on a large scale. Measuring peripheral diopters to estimate the shape of the retina through mathematical formula is indirect and its accuracy is questionable.15 The latest advances in the OCT field provid a safe, accurate and affordable way to study retinal morphology.16–18
RC is a quantitative metrics describing the steepness of the local retina. The steeper the retina, the higher the curvature. Abnormal increases or decreases in RC can be easily detected using curvature maps, indicating a protruding or concave irregularity of the retina.19 Some researchers have employed the radius of RC in earlier literature,13 20 21 which can be mathematically converted into Gaussian curvature22:
(R: radius of curvature; K: Gaussian curvature). Both metrics were initially employed for data analysis; however, Gaussian curvature yielded more satisfactory results. Consequently, Gaussian curvature was selected as the primary marker for retinal morphology in our study.
By calculating ICC values, the repeatability of our OCT-based RC measurement method was preliminarily verified.23 Our method is also highly operable, with OCT scans taking only about 5 min per person, and over 90% of OCT angiography images successfully completed 3D modelling and curvature measurement.
As the distance from the central fovea increases, RC exhibits a noticeable trend of initially increasing and then decreasing. Among the children with myopia, fovea is the flattest region in the eye’s posterior pole, resulted in a mean curvature of 0.51±0.19 10−2mm−2. The retina gradually becomes steeper as the distance from the macula increases, reaching RC’s maximum value (0.86±0.23 10−2mm−2) at approximately 4.5–6 mm away from the fovea. Further outward, towards the equatorial region, the retina becomes gradually flatter and RC begins to decrease.
HM group demonstrated a higher foveal RC and a smaller peripheral RC compared with the LM group. One possible explanation is that the progression of myopia is often accompanied by a localised outward protrusion of the retina near the central fovea. Simultaneously, due to the elongation of the eye axis and the increase of eyeball volume, peripheral retina is stretched to become flatter, especially in the retro-equatorial region. This also causes the distribution pattern of curvature in relation to distance to be more pronounced in emmetropic eyes, as the morphological change induced by myopia tends to diminish the curvature differences between macular and peripheral retina. These findings corroborate the description of myopia-induced morphology changes in previous research.13 24
The result of correlation analysis revealed a correlation between foveal/peripheral RC and age, SE, CR, AL and SFCT. Overall, as age increases and myopia progresses (manifested by a decrease in SE and an increase in AL), the foveal RC increases, while the peripheral RC decreases gradually. However, correlation analysis may not adequately demonstrate the relationship between SFCT and curvature, because SFCT’s direct impact on curvature is overshadowed by the stronger correlation between SFCT and SE or AL.
Regression analysis also presented the association between age and RC as described above. This aligns with the clinical phenomenon where, despite most posterior staphyloma (PS) patients’ SE stabilising in their adolescence, PS often develops many years later. For every 10 years of age increase, the peripheral RC decreases by 0.14 10−2mm−2, while the foveal RC increases by 0.13 10−2mm−2, which amounts to nearly 20% of the mean foveal RC in children with myopia. Thinner SFCT was associated with the steepening of the central or peripheral retina. For every 100 µm decrease in SFCT, both foveal and peripheral RC increase by 0.06 10−2mm−2. Thinning of the choroid can result in localised weakening of the eye wall, making it more susceptible to developing morphological changes. Furthermore, axial elongation was associated with a flatter peripheral retina, with peripheral RC decreasing by 0.09 10−2mm−2 for every 1 mm increase in AL. Interestingly, foveal RC was not found to have a significant regression coefficient with AL, which may suggest that the association between myopia progression and retina protrusion may be mediated by choroidal thinning rather than by direct elongation of the eyeball. The association between female gender/smaller CR and larger foveal RC could be attributed to females or individuals with smaller CR tend to have smaller eyeballs.
Despite the relatively young average age of the participants, significant differences in RC among four orientations were observed within HM group. With myopia progressing, some portions of the retina protrude outwards and the shape of the retina becomes irregular gradually. These protrusions may occur more frequently below and temporally to the fovea due to their larger RC, and least frequently above the fovea.
Future studies may require longitudinal designs to observe how retinal morphology affects the progression of myopia, and whether local changes in retinal morphology result in changes in refractive diopter or lead to other refractive problems. Establishing relevant cohorts is essential to observe the progression of these retinal protrusion over time, which may help to identify which retinal protrusions are prone to eventual development into PS. Additionally, applying curvature measurements to a larger population and establishing standardised percentile data would be valuable in the early identification of patients with abnormal ocular morphology. Our study will provide a reference for further research and application of RC.
This study had several limitations. First, in our sample, there appeared to be a higher prevalence of myopia among older participants. This could potentially impact the conclusions drawn from intergroup comparisons. The study had a relatively small sample size, so the sample might not properly represent the population of Chinese children with myopia. Insufficient sample size and notable internal heterogeneity of the study population led to regression models with small R2 values, which might affect the accuracy of predicting changes in curvature values.
Conclusions
The progression of myopia leads to a steeper macular retina in children, while the peripheral retina becomes flatter, and the disparities between different orientations increase. RC serves as an effective metric for quantitively characterising retinal shape and morphological changes induced by myopia. Ageing, choroidal thinning and axial elongation are contributing factors to increased curvature.
Data availability statement
Data are available on reasonable request. Part of our data has been uploaded as supplementary information. Other data are available on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by Ethics Committee of Shanghai Eye Disease Prevention and Treatment Center (research protocol ID: 2024SQ003). This study adhered to the Declaration of Helsinki. Informed consent was obtained from participants and their parents prior to the study.
Acknowledgments
The authors acknowledge the support of TowardPi Medical Technology for the UWF SS-OCT system used in this study.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Correction notice This paper has been amended since it was first published. The first author's name has been corrected.
Contributors Conception or design of the work: HW, XX, HL, XH. Acquisition of data: HW, BZ, JC, ZQ. Analysis and interpretation of data: HW, JC. Writing the manuscript: HW. Revising the work: HW, XX, HL, XH. Final approval of the version: all authors. Agreement to be accountable for all aspects of the work: all authors. Overall responsibility for the work: XH.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.