Changes of corneal biomechanics in children using orthokeratology and their roles in predicting axial length progression—A prospective 2‐year study

To determine how orthokeratology (ortho‐k) affects corneal biomechanical properties in myopia control and whether corneal biomechanical parameters can predict clinical efficacy of ortho‐k.


| I N T RODUC T ION
Globally, the burden of myopia has risen rapidly and the prevalence is postulated to reach 80-90% among East Asian high school graduates (Morgan et al., 2018;Pan et al., 2012).Myopia is characterized by abnormal axial elongation which lead to inconvenient daily life and loss of productivity and carries an estimated cost of $244 billion globally (Naidoo et al., 2019).Progression to high myopia increases risks of complications like macular degeneration and glaucoma, which are leading causes of irreversible vision loss and blindness (Lakawicz et al., 2020;Praveen et al., 2010;Xu et al., 2007).Slowing myopia progression is a central principle underlying myopia management.
Orthokeratology (ortho-k) is a clinical technique that is used to achieve correction of myopia and control of axial length (AL) progression (Hiraoka et al., 2015;Swarbrick et al., 2015).The central base curves of modern ortho-k lenses, which are flatter than the central corneal curvature, provide shaping pressure and induce migration of epithelium.The migration can lead to a reduction in central epithelial basal cells, an increase in central stromal thickness, changes in the ocular surface microenvironment and so on (Na et al., 2016;Sanchez-Garcia et al., 2021).This may raise concerns and questions about the safety of ortho-k.Corneal biomechanical parameters, potential indicators of corneal health, can also provide insight into how well the cornea respond to external stresses.Till now, no study has reported the changes in adolescents' corneal biomechanical properties during 2 years of ortho-k lens wear.
On the other hand, several landmark studies have investigated the association between corneal biomechanics and the ortho-k's control efficacy on myopia.Winkler and Kame (1995) found that corneal rigidity was associated with astigmatism resulting from ortho-k, while Carkeet et al. (1995) failed to observe this correlation.Recent studies show short-term ortho-k lenses wearingwithin a single night or even 10 min-significantly alters corneal curvature, spherical equivalent refraction (SER) and uncorrected visual acuity (Berntsen et al., 2005;Johnson et al., 2007;Sridharan & Swarbrick, 2003).Li et al. (2022) conducted a cross-sectional study and reported biomechanical after ortho-k lenses wearing could predict 1-years treatment outcomes.However, to data prospective studies on corneal biomechanical properties' role in controlling AL progression for long-term ortho-k using are rare.
Therefore, the purpose of this study was to investigate the change of corneal biomechanical properties during 2-year ortho-k treatment.Furthermore, whether corneal biomechanical parameters could influence long-term AL progression with lenses wearing would also be examined.

| Study design
The study was designed as a prospective 2-year followup study.Participants were enrolled from April 2018 to October 2018 at Shanghai Eye Disease Prevention & Treatment Center in China where baseline and follow up examinations for the study were conducted.Reexaminations were scheduled at 1 week, and 1, 3, 6, 12, 18 and 24 months after commencing ortho-k lenses.The study protocol was approved by the Shanghai General Hospital Ethics Committee (NO: 2018[09]), adhered to the tenets of the Declaration of Helsinki and was registered on Clini calTr ials.gov(NO: NCT03516357).All the children who participated in follow-up were informed and understood the study protocol.Their parents or legal guardians signed written informed consent forms and children gave oral consent.

| Participants
Children 7-15 years of age with low to moderate myopia receiving ortho-k treatment were enrolled.A series of baseline ocular examinations were performed including slit-lamp microscopy of the anterior and posterior eye segments, AL measurement, cycloplegic refraction, corneal topography, corneal biomechanical parameters, un-corrected and corrected visual acuity (VA).Children with history of systemic or ocular pathology, strabismus or amblyopia, obvious astigmatism (>−1.50 diopters, D) or those who used any myopia control treatment strategies previously were not included.Those who were lost to follow-up, did not cooperate with continuous orthok lenses wearing, did not comply with test procedures/ schedule, or had incomplete data were excluded.

| Ortho-k lenses wearing
All participants' ortho-k lenses were fitted using spherical 4-zone lenses (Euclid Systems Corporation) made of Boston EQUALENS II (oprifocon A), with oxygen permeability coefficient 127 × 10 −11 (cm 2 •ml•O 2 )/(s•ml•mm Hg), diameter 10.6-10.8mm, optical center thickness 0.21-0.23 mm and wetting angle 36°.Final lenses were determined after optimizing fluorescein pattern, topographical evaluation, refractive and visual outcomes.Full correction was targeted for all participants.Participants were instructed to wear ortho-k lenses for a minimum of 8 h per night, and received instructions about maintenance of lenses.Participants were reexamined at 1 week, and 1, 3, 6, 12, 18 and 24 months after commencing ortho-k lenses.Abnormal symptoms were strictly observed at each follow-up.Given corneal biomechanical properties can be affected by the integrity of epithelial barrier, if intense corneal staining (grade 2 and grade 3, Efron Grading Scale) occurred, the participants were discontinued from treatment and excluded from final analysis.

| Measurements
To avoid the influence of diurnal rhythm, all measurements were performed at 8-10 AM for each examination.Cycloplegic refraction and keratometry (KR) were performed by an auto-refractor (KR-8900, Topcon) at baseline after the ciliary muscle paralysed using 0.5% tropicamide.Corneal biomechanical parameters and other ocular biometry were collected at baseline, 1 week, and at 1, 3, 6, 12, 18 and 24 months.AL values were evaluated per eye (IOL Master V5.02, Carl Zeiss) at baseline and at 6, 12, 18, 24 months.Five consecutive measurements were taken and the average value was used for analysis.If the difference between any two measurements was >0.02 mm, the procedure was repeated.
Corneal biomechanical parameters were obtained using Corvis ST (Type 72100, Oculus Optikger¨ate GmbH).The Corvis ST can record the reaction of the cornea to a defined air pulse and provide a detailed assessment of corneal biomechanical properties in vivo.Following the air puff, the cornea undergoes three distinct phases, involving first applanation (A1), then highest concavity and lastly second applanation (A2), respectively (Figure 1).From this, over 30 biomechanical parameters can be produced.Previous studies demonstrated that deformation amplitude maximum (DA), A1 Time (A1 T), A2 Time (A2 T) and central corneal thickness (CCT) showed relatively high intraclass correlation coefficients and small coefficients variations, while corneal velocities, applanation length and time at the highest concavity were not repeatable (Nemeth et al., 2013;Wu et al., 2016).For the reasons above, only DA, A1 T, A2 T and new parameters derived by the new Corvis ST version, including biomechanically corrected IOP (b IOP), deformation amplitude ratio maximum 1 or 2 mm (DA R1/DA R2), integrated inverse concave radius (IR) and Stiffness parameter at A1 (SP A1) were analysed in our study.The definitions of these parameters were listed in Table S1.

| Statistical analysis
Data were analysed using SAS 9.4 software (SAS Institute, Inc.).Only data from the right eyes were analysed.The normality of data was tested using the Kolmogorov-Smirnov test, and repeated measures analysis of variance (RANOVA) was used to compare changes in corneal biomechanical parameters across time-points.Pearson correlation analysis assessed relationships between the AL progression (AL progression = AL at 24 months-AL at baseline) and different baseline biomechanical parameters.Considering the degree of myopia may influence AL progression rate, subjects were divided into low (−3.00D< SER ≤-0.50D) and moderate myopes (−6.00D < SER ≤-3.00D) when studying the effect of biomechanical properties on AL progression.Multiple linear regression model established independent factors for AL progression with ortho-k use.Receiver operator characteristic (ROC) curves used baseline DA and other variables to test their effect on predicting excessive AL progression (>0.35 mm during 2 years).Multiple logistic regression then generated a formula for calculating the probability for excessive AL progression.Since the dimension of DA is small, the change with a gradient of 1 rarely happens in actual clinical practice.In the multiple logistic regression model, DA is rescaled and multiplying by 100 to avoid the values of OR being too large.The probabilities for the occurrence of excessive AL progression in ortho-k lenses wearers based on different related predicting factors were calculated according to the generated formula.Statistical significance was p < 0.05.

| The basic characteristics of the participants
In all, 153 subjects are eligible at baseline visit and 12 subjects are excluded for poor centration, 9 for ocular health problem, 7 for loss of follow-up.A total of 125 (34 boys, 27.2%) Chinese myopes with age range of 7-15 years (mean ± SD: 10.30 ± 1.75 years) were included for final analysis.Mean SER was −3.45 ± 0.97D (range: −1.75 to −5.75D), and mean AL was 25.02 ± 0.84 mm (range: 23.18-26.84mm).The subjects were divided into low (n = 36) and moderate myopes (n = 89).No significant differences in age were observed between low and moderate myopes (p = 0.926), but significant differences were observed in baseline AL (p < 0.001).
Other baseline biomechanical parameters are presented in Figure 2. According to the result of the Kolmogorov-Smirnov test, parameters of DA, A2 T, CCT, DA R1, DA R2, IR and SP A1 were normally distributed (p > 0.05), while A1T and b IOP were abnormally distributed (p < 0.05).Besides no significant differences in most parameters were observed between low and moderate myopes (p > 0.05), but significant differences were observed in SP A1 (p < 0.05).

| Change of corneal biomechanical properties after receiving ortho-k
Table 1 and Figure  the whole wearing period, indicating the cornea reestablished biomechanical equilibrium soon.

| Relationship between biomechanical properties and AL progression in ortho-k lenses wearers
AL significantly changed over 24 months from 25.02 ± 0.84 mm to 25.38 ± 0.81 mm (p < 0.001), and AL progressed by total 0.35 ± 0.30 mm (0.18 ± 0.15 mm/year).Besides, no significant differences were observed on AL progression between different genders or refractive groups (p > 0.05).As Figure 4 presents, baseline DA correlated most with AL progression (r = 0.37, p < 0.001) compared to other biomechanical parameters.However, AL progression over 2 years was not significantly correlated with corneal biomechanical parameters at 1 week (p > 0.05).
Table 2 demonstrates the multiple regression analysis for low and moderate myopia groups which include age, gender, AL and the nine biomechanical parameters at baseline.For low myopes, baseline DA (β = 3.27, standard β = 0.91, p < 0.001) and AL (β = −0.34,standard β = −0.81,p = 0.024) were independently associated with AL progression.For moderate myopes, baseline DA and AL were not significantly correlated with AL progression (p = 0.104, 0.938), while age was the only significant variable (β = −0.05,standard β = −0.33,p = 0.007).The R square of multiple regression models in low and moderate myopia groups were 78.49% and 21.80% respectively.

| Prediction of excessive AL progression during Long-term Ortho-K
Since DA correlated with AL progression significantly in the Pearson correlation analysis and multiple linear regression analysis, we selected DA to represent corneal biomechanical properties.The ROC curves for predicting excessive AL progression in ortho-k users using baseline DA, age and AL are presented in Figure 5.The AUC (area under the ROC curve) for low and moderate myopes was 0.902 and 0.698.To explore the combination of baseline DA, age and AL as a predictor for the potential excessive AL progression with ortho-k treatment, multivariate logistic models were generated.Table 3 displays higher DA (OR = 1.16, p = 0.047), lower AL (OR = 0.13, p = 0.012) and lower age (OR = 0.34, p = 0.028) were associated with higher risk of excessive AL progression in low myopes after using ortho-k lens.Subsequently, the formula for this was listed below and the probability for excessive AL progression was presented in Table 4.

| DI SC US SION
In this 2-year prospective study, we followed children wearing ortho-k lenses and observed corneal   biomechanical properties fluctuated greatly within the first week but stabilized throughout follow-up.Furthermore, baseline DA was independently associated with 2-year AL progression in low myopes wearing ortho-k lenses.Based on this, we then constructed a prediction model about the effect of ortho-k on low myopia and its AUC reached 0.902.This study observed rapid changes of corneal biomechanics after 1 week of ortho-k use, which stabilized thereafter until follow-up ceased at 24 months.These rapid changes are consistent with a meta-analysis of 10 studies that observed significantly decreased central corneal thickness over first week of ortho-k use (Li et al., 2016).It also partially agrees with a prospective study of 36 ortho-k users where corneal curvatures and thickness stabilized after 1 week until follow-up ceased at 6 months, although other parameters continued to change (Lam et al., 2019).Ortho-k lenses work by stimulating migration of corneal epithelial cells from the center to mid-peripheral area, causing central corneal thinning and mid-peripheral thickening (Swarbrick, 2006).Matsubara et al. (2004) found cell migration took 7-10 days and epithelium returned to its original state within 7 days after lenses were removed.As shown in our study, most corneal biomechanical indexes also changed within 1 week, in accordance with laboratory research.It is thought epithelial migration may disturb the cell-cell interactions and weaken corneal biomechanics, however, corneal biomechanical parameters remained relatively stable after rapid changes at 1 week.This may indicate cells can rapidly establish a new balance of intercellular forces.
Previous studies demonstrate interindividual variations in ortho-k effect could vary from −0.22 to 0.61 mm in AL progression per year (Cho et al., 2005;Cho & Cheung, 2012;Lee et al., 2017;Wen et al., 2015;Zhu et al., 2014) and the reason for this may be that the mechanism and factors is still not fully understood.Therefore, a series of factors have been proposed to explain the variations including CCT (Lee et al., 2018), magnitude of corneal power changes (Zhong et al., 2014), corneal shape (Santodomingo-Rubido et al., 2013), and ortho-k lens-induced corneal power shift (Hu et al., 2019) are associated with AL progression after long-term ortho-k use.In the current study, we tried to figure out whether corneal biomechanical properties were involved in this, too.We investigated nine corneal biomechanical parameters but DA was the only independent factor for 2-year AL elongation in low myopes (β = 3.27).We assume that a softer cornea with higher DA will show not only a more rapid response ortho-k, but also a more rapid recovery after lens removal which may result in rapid attenuation of peripheral defocus, reduced ortho-k efficacy and excessive AL progression in low myopia.
In moderate myopes, we observed age as an independent factor for AL progression while no significant correlation between baseline DA and AL progression was observed.There was also no statistically significant difference on corneal biomechanical parameters between the low and moderate myopia groups, which is in accordance with previous studies (Wang et al., 2015;Yu et al., 2020).Nevertheless, to correct moderate myopia, ortho-k lenses have to provide stronger shaping stress to induce more redistribution of epithelium and more peripheral myopia defocus, which might influence AL progression during ortho-k treatment (Charman et al., 2006;Kang & Swarbrick, 2011;Santodomingo-Rubido et al., 2012;Walline et al., 2009).As mentioned above, compared to low myopes' corneas, moderate myopes' corneas with similar recovery rates and more corneal deformation are less likely to return their original shape.In this hypothesis, the peripheral defocus effect of ortho-k on moderate myopes is not easily undercut by DA and corneal rebound, and therefore DA and AL progression are not statistically correlated.
Considering gender being statistically insignificant in Xu et al.'s study (Xu et al., 2021), Lee et al.'s study (Lee et al., 2018) and our multiple linear regression models, gender was not included in the final predictor set and further analysis.We combined three significant indicators, baseline DA, age, and AL and conducted ROC analysis to comprehensively evaluate predictive accuracy.After stratifying by the degree of myopia, the ROC curves achieved different AUCs which is consistent with the different adjusted R 2 of the multiple linear regression.Referring to the AUC evaluation criteria, AUC has low accuracy at 0.5-0.7,F I G U R E 4 Correlation heat map between AL progression and baseline age, AL, each corneal biomechanical parameter.A1 T, first applanation time; A2 T, second applanation time; AL, axial length; b IOP, biomechanical corrected intraocular pressure; CCT, central corneal thickness; DA, deformation amplitude maximum; DA R1/DA R2, deformation amplitude ratio maximum 1 or 2 mm; KR, keratometry; IR, integrated inverse concave radius; SP A1, stiffness parameter at first applanation time.some accuracy 0.7-0.9, and high at 0.9 or above.Therefore, this combined set of indicators which reached AUC of 0.902 can be used in clinical practice to predict whether a low myopic adolescent wearing ortho-k lenses will experience AL progression over 0.35 mm during 2 years.
As for the selection of the 0.35 mm as cutoff to predict whether ortho-k lenses would control myopia effectively, we first refer to the demonstration in Wallman & Winawer's study that the refractive development and eye growth are regulated by homeostasis mechanisms which try to eliminate mismatches between the eye's optical power and its length and achieve emmetropia (Wallman & Winawer, 2004).However, genetic or environmental factors may interfere with this emmetropization mechanism and lead to the development of myopia.Besides, Rozema's study provided that the mean AL progression of persistent myopia, myopia onset at age 10, myopia onset at age 11, and persistent emmetropia for 10-year-old children (the mean age of subjects in our study was 10.30 years) was 0.35, 0.37, 0.32, and 0.17 mm/ year, respectively (Rozema et al., 2019).It is noteworthy that the already myopic adolescents in this study had an annual AL progression of 0.17 mm after wearing ortho-k lenses, which is in accord with previous studies (Cho & Cheung, 2012;Hiraoka et al., 2012;Kakita et al., 2011).Based on the above, we believe that when ortho-k lenses work well, the annual growth rate of AL in already myopic adolescents can be controlled to 0.17 mm or even less.Therefore, in this study AL progression over 0.35 mm for 2 consecutive years was defined as excessive AL progression which indicates failure of ortho-k to achieve its anticipated effect.
After constructing ROC curves to evaluate the combined set of variables, we developed a logistic regression model and generated a formula to calculate the probability of excessive AL progression in low myopes.As the ROC curve and regression coefficients were promising, the formula might have good real-world accuracy.Such a formula could apply before ortho-k lens fitting to assist clinician decision-making and choose effective myopia interventions.For example, for a 12-year-low child, AL 24.00 mm and DA 0.8 mm, the probability excessive AL progression after lenses wearing over two years could be 3.6%, then the effect of ortho-k can preliminarily be assumed promising.In the case of a child with the same age and AL, but DA 1.20 mm, the probability could be 92.6% which implies ortho-k treatment may not be suitable in this case.However, a real-world study is needed to test its efficacy in practice.
Our study has several limitations.Firstly, among 125 participants 91 were girls which prevents a gender association with AL from being analysed without bias.Secondly, the study did not include other critical associated factors such as peripheral defocus, time spent outdoors, screen and reading time which may all influence the effect of ortho-k on myopia control.Thirdly, we only investigated relationships between AL progression and corneal biomechanical parameters during the initial 2 years after ortho-k lenses wearing, and more long-term study is still needed.Lastly, this study comprised of Chinese children aged 7-15.Studies comprising multiethnic cohorts are needed to affirm these findings are applicable for other races and geographic regions.
In conclusion, we observed fluctuations in corneal biomechanics during the first week and DA at baseline was significantly associated with long-term Al progression after lens wearing.Although 1-week change of biomechanics is statistically significant but clinically insignificant, and long-term stability suggest that ortho-k is a safe approach for myopia control.Combination of baseline DA and other indicators is promising for its application in ortho-k to predict future excessive AL progression with high accuracy and provide a useful reference about the efficacy of ortho-k for clinicians before lens fitting.More studies

CON F L IC T OF I N T E R E ST STAT E M E N T
No conflicting relationship exists for any author.

F
The corneal response to an air puff observed with Corvis ST.(a) A precisely metered air pulse forces the cornea to move inward through an applanation.(b) The cornea then continues to move inward until it reaches the highest concavity.(c) The cornea rebounds to its normal convex shape.During this phase, the cornea again passes through an applanation.A1 T, first applanation time; A2 T, second applanation time; CCT, central corneal thickness; DA, deformation amplitude maximum.
3 show the mean or the quartiles of corneal biomechanical parameters at baseline and each follow-up visit (1 week and 1, 3, 6, 12, 18 and 24 months).All biomechanical parameters changed significantly within the first week after wearing lenses (p < 0.05), except A2 T and SP A1.In this week under review, DA, DA R1, DA R2 and IR trended upward, while A1 T, CCT and b IOP experienced a statistically significant decline.However, after the first week, DA, A1 T, CCT, b IOP, DA R1, DA R2 and SP A1 soon regained stability and remained so until the 2-year follow-up ceased (p > 0.05), with only A2 T increasing and IR fluctuating throughout F I G U R E 2 The distribution in corneal biomechanical parameters at baseline.A1 T, first applanation time; A2 T, second applanation time; b IOP, biomechanical corrected intraocular pressure; CCT, central corneal thickness; DA, deformation amplitude maximum; DA R1/DA R2, deformation amplitude ratio maximum 1 or 2 mm; IR, integrated inverse concave radius; SP A1, stiffness parameter at first applanation time; SD, standard deviation.

F
Line charts of biomechanical parameters before and during ortho-k lenses wearing (medians, 25th and 75th percentiles).The red dots and error bars represented the values of 1st week.A1 T, first applanation time; A2 T, second applanation time; b IOP, biomechanical corrected intraocular pressure; CCT, central corneal thickness; DA, deformation amplitude maximum; DA R1/DA R2, deformation amplitude ratio maximum 1 or 2 mm; IR, integrated inverse concave radius; SP A1, stiffness parameter at first applanation time.

F
I G U R E 5 Receiver operating characteristic (ROC) curves for the prediction of excessive AL progression in different refractive groups by combination use of baseline DA, age and AL after wearing ortho-k lenses.AUC, area under the ROC curve; AL, axial length; DA, deformation amplitude maximum.T A B L E 3 Logistic regression models for the prediction of excessive AL progression in different refractive groups.
Multiple linear regression models of AL progression in different refractive groups.