Potential Choroidal Mechanisms Underlying Atropine's Antimyopic and Rebound Effects: A Mediation Analysis in a Randomized Clinical Trial

Abstract

Purpose: To investigate whether choroidal vascularity participates in high-dose atropine's antimyopia and rebound mechanisms.

Methods: A mediation analysis was embedded within a randomized controlled trial. In total, 207 myopic children were assigned randomly to group A/B. Participants in group A received 1% atropine weekly (phase 1) and 0.01% atropine daily (phase 2) for 6 months each. Those in group B received 0.01% atropine daily for 1 year. Four plausible intervention mediators were assessed: total choroidal area (TCA), luminal area (LA), stromal area (SA), and choroidal vascularity index (CVI).

Results: In group A, LA, SA, and TCA increased significantly after receiving 1% atropine for 6 months. The increment diminished after tapering to 0.01% atropine. In group B, those parameters remained stable. TCA mediated approximately one-third of 1% atropine's effect on spherical equivalent progression in both phases. In phase 1, the mediation effect of TCA was shared by LA and SA, while only that of LA remained significant in phase 2. No mediation effect of CVI was found.

Conclusions: One percent atropine induced choroidal thickening by increasing both LA and SA, while 0.01% atropine had little choroidal response. The choroidal changes following 1% atropine treatment diminished after switching to 0.01% atropine. TCA, but not CVI, partially explains atropine's antimyopic and myopic-rebound mechanisms. SA may serve as a potential biomarker to predict the postrebound treatment efficacy of high-dose atropine. (ClinicalTrials.gov number, NCT03949101.)

Myopia has attracted public attention for its increasingly high prevalence in the past few decades, especially in East Asia.^1,2 Mild and moderate levels of myopia increase the risk of retinal detachment, while high levels of myopia can result in sight-threatening conditions.^2,3 Atropine eye drops of different concentrations have been used for myopia control for years.^4–6 Higher doses of atropine are more effective but are overshadowed by lower doses for greater side effects and the myopic rebound after cessation.⁷ A deeper understanding of the mechanisms underlying the antimyopic effect and the posttreatment rebound may help optimize the advantages of high-dose atropine. 

Choroidal changes may be responsible for the development of myopia. Results from animal experiments^8,9 and cross-sectional studies^10,11 revealed a correlation between choroidal thinning and myopia. Further longitudinal studies^12,13 suggested that choroidal thinning could precede axial eye elongation and spherical equivalent (SE) progression. Recently, reductions in choroidal thickness (ChT), total choroidal area (TCA), luminal area (LA), stromal area (SA), choroidal vascularity index (CVI), and choriocapillaris blood perfusion were found in myopic eyes of anisomyopes.¹⁴ These parameters correlated with the severity of myopia, suggesting the presence of disturbances of the choroidal vasculature during myopia development. Moreover, atropine could alter the choroidal structure. Children receiving either 0.05%¹⁵ or 1%¹⁶ atropine eye drops showed an increased choroidal thickness. In another study where guinea pigs were treated with peribulbar injections of 1% atropine, both choroidal thickness and choroidal blood perfusion were increased.¹⁷ 

Several studies described a correlation between choroidal changes and myopia development. However, a causal relationship has not yet been demonstrated. Also, evidence concerning any choroidal change during myopic rebound is still lacking. Furthermore, more details on the exact changes in choroidal components are needed to dig deeper into the possible mechanisms underlying those responses. In this hypothesis-generating study, we aimed at demonstrating the choroidal responses in children with myopia during atropine treatment and myopic rebound, as well as clarifying the causal role of choroidal changes in either situation. We first assessed the longitudinal changes in choroidal parameters following consecutive use of 1% and 0.01% atropine or 0.01% atropine alone over 1 year. Second, we evaluated the relationship between choroidal vasculature and myopic parameters. A series of mediation analyses were conducted to evaluate the causal relationship between choroidal changes and treatment responses to atropine. Notably, owing to the exploratory nature of the current study, the results should be considered only descriptive and hypothesis generating, needing further confirmation. 

Materials and Methods

Study Design

This study is a secondary analysis of the 1-year data from the Atropine for Children and Adolescent Myopia Progression (ACAMP) study, a single-blind, randomized controlled trial¹⁸ of two atropine treatment regimens: (1) receiving 1% atropine weekly (phase 1) and 0.01% atropine daily (phase 2) for 6 months each and (2) receiving 0.01% atropine daily for 1 year. Detailed experimental design of the ACAMP study has been described in our former report.¹⁶ 

Written informed consent forms were obtained from the participants and their parents or guardians. The study protocol was approved by the Ethics Committee of Shanghai General People's Hospital, Shanghai, China (approval number: 201939), and registered at the Clinical Trials.gov PRS (Registration No. NCT03949101). All procedures were conducted in accordance with the tenets of the Declaration of Helsinki. 

Participants

In total, 207 children aged 6 to 12 years with myopic refraction of at least –0.5 diopters (D) and astigmatism of less than −2.0 D in both eyes were enrolled in the ACAMP study. Those with ocular diseases or severe systemic diseases, previous use of myopia interventions (atropine, pirenzepine, orthokeratology lens, etc.), or allergy to atropine and cyclopentolate were excluded. All participants were followed at baseline and 1, 12, 24, and 48 weeks. Compliance data were collected using a questionnaire at every visit. Those with poor compliance (days with administration/follow-up period <0.8) were excluded from the study. The study was conducted at Shanghai Eye Disease Prevention and Treatment Center from March 2019 to April 2020. On account of the epidemic transmission of COVID-19, the 36-week follow-up visit was canceled, and 16 (9.4%) participants were delayed by 3 to 4 weeks for the 48-week follow-up visit. 

Study Procedures

At the beginning of the study, the participants were randomly assigned to two treatment groups in a ratio of 1:1: group A (experimental group) received 1% atropine sulfate eye gel (Dishan; Shenyang Xingqi Eye Hospital Co., Ltd., Shenyang, China) once weekly in both eyes for 6 months (starting with a 1-week loading dose: 1% atropine once daily) and were switched to 0.01% atropine (Myopine; Shenyang Xingqi Eye Hospital Co., Ltd.) once nightly in both eyes for another 6 months; group B (control group) received 0.01% atropine sulfate eye drops once nightly in both eyes throughout 1 year. 

Examination procedures were performed by fixed optometrists who were masked to treatment groups, as reported by former reports.¹⁶ Except for the 1-week visit, the ocular biometrics were obtained following complete cycloplegia. After excluding the contradictions of cycloplegia, one drop of topical 0.5% proparacaine (Alcaine; Alcon, Fort Worth, TX, USA) was administered in both eyes, followed by two drops of 1% cyclopentolate (Cyclogyl; Alcon) at a 5-minute interval. A third drop of cyclopentolate was given 45 minutes later if the pupillary light reflex was still present or the pupil size was less than 6.0 mm. Autorefraction was performed using an autorefractor (Topcon KR 800; Optical Corp., Guangdong, China). SE was calculated as spherical power plus half of the cylinder power. Axial length (AL) was measured using an IOL-Master 700 (Carl Zeiss Meditec AG, Jena, Germany). 

The swept-source optical coherence tomography (OCT) (DRI OCT Triton, Topcon, Tokyo, Japan) was performed after cycloplegia with the exception of the first week’s visit. To minimize the diurnal variation in the choroidal structure, all measurements were conducted between 10 am and 3 pm.^19,20 Follow-ups were prearranged within 2 hours based on the time of the baseline visit. The scanning procedure and instrument parameters have been previously described.²¹ Briefly, 12 radial lines centered on the fovea were obtained at a rate of 100,000 A-scans per second, covering an area of 9 mm. Every radial line was an average of 16 overlapped B-scans, each consisting of 1024 A-scans. The images were binarized according to the method proposed by Sonoda et al.²² Then, a semiautomatic segmentation was conducted using a custom-designed algorithm (Beijing Airdoc Technology Co., Ltd., Beijing, China). After image processing, the mean values of TCA, LA, and SA were calculated in a 6-mm submacular region centered on the fovea. The CVI was defined as the ratio of LA to TCA. 

Statistical Analysis

Only the right eyes were included in the statistical analysis. Baseline data were described as mean ± standard deviation. Longitudinal changes in ocular parameters for each group and intergroup differences were examined using linear mixed-effect models (LMMs) with restricted maximum likelihood estimation. The longitudinal changes were defined as the difference between follow-up and baseline values. LMM analysis was conducted for every ocular parameter, with the longitudinal changes set as the dependent variable. Group, time (as a categorical variable), and group × time interaction were included as fixed variables. Random intercepts for participants were included to eliminate the dependence of repeated measurements. Estimated marginal means (EMMs) were obtained from the fitted models for pairwise comparisons and P value acquisition. Correlations among parameters were analyzed using Pearson's or Spearman's coefficients and corresponding P values. 

Mediation analysis is an analytic method to quantitively evaluate the extent to which an intervention may affect an outcome through a causal mechanism.²³ In the current study, mediation analysis was performed using a structural equation model (SEM) framework, wherein multiple regression equations were evaluated simultaneously.²⁴ Baseline characteristics were fitted in mediation models as confounding variables. Atropine treatment (1% vs. 0.01%, group A versus group B) was considered a dummy variable (group A = 1, group B = 0) and evaluated in phase 1 and phase 2, corresponding to the antimyopic (0–6 months) and myopic-rebound (6–12 months) phases, respectively. Outcomes were changes in axial length and spherical equivalent during phases 1 and 2. Four plausible intervention mediators were assessed: LA, SA, TCA, and CVI. LA and SA were regarded as “specific indicators” of the choroidal structure, as they directly reflect the two components of the choroid. TCA and CVI were considered “comprehensive indicators” of the choroidal structure because they were calculated from LA and SA. A total of four mediation models were implemented so that the specific and comprehensive indicators could be evaluated separately in either phase. Models were named with a number (1 or 2, corresponding to phase 1 or phase 2 treatment), followed by a letter (X or Y, corresponding to comprehensive or specific indicators). For example, model 1X evaluated phase 1 treatment (group A versus group B for the first 6 months) and comprehensive indicators of the choroidal structure (TCA and CVI). The four assumed causal models are presented in a directed acyclic graph (Supplementary Fig. S1). Variables inside the solid rounded rectangles were fitted to different models according to the symbol on their left side. In each represented model, variables are connected with arrows representing a path. Treatment might exert its effects via indirect paths (red arrows) through single or multiple mediators or straightly via a direct path (blue arrows). The paths linking atropine effects to either AL or SE are surrounded by a dashed rounded rectangle. For example, in model 1X, phase 1 treatment might exert its effects on AL via indirect paths through either TCA (path a1*b1) or CVI (path a2*b2), or via a direct path (path d1). Path coefficient estimates were obtained with maximum likelihood estimation. The corresponding 95% confidence intervals and P values were derived from bootstrapping with 1000 bootstrap replicates. No calculation of sample size was conducted for mediation analysis as it was not the primary aim of the ACAMP study. A detailed description of sample size calculation was included in our former report.¹⁶ 

Due to the exploratory nature of the study, significant levels were set as P < 0.05 (two-sided), and no adjustment was taken for multiple comparisons. The results should be considered only descriptive and hypothesis-generating, and they need to be confirmed in other studies. Statistical analysis was conducted using R software (version 4.1.2). LMM, EMM, correlations, and SEM analyses were performed using packages lme4²⁵ (version 1.1-29), emmeans²⁶ (version 1.7.3), corrplot²⁷ (version 0.92), and lavaan²⁸ (version 0.6-11), respectively. 

Results

Of the 207 children originally randomized to receive 1% atropine once weekly (group A) or 0.01% atropine once daily (group B) for 6 months, 185 continued in the second phase of the trial, with 98 in group A and 87 in group B, respectively. Finally, 171 children completed the 1-year follow-up, and 1 child was excluded from the final analysis due to loss of OCT images. Among the 37 children excluded from the analysis, 6 (5.8%) in group A and 8 (7.8%) in group B had poor compliance (Supplementary Fig. S2). The baseline characteristics of the 170 children included in the analysis were similar to the 37 children who were excluded (Supplementary Table S1). 

Longitudinal Changes in Choroidal Parameters

For group A, after receiving 1% atropine weekly for the first 6 months, there were significant increases from baseline in LA and SA (0.22 ± 0.23 mm² and 0.13 ± 0.10 mm², respectively; both P < 0.001) (Table 1 and Fig. 1). As a result, TCA increased significantly while CVI remained unchanged (0.35 ± 0.30 mm², P < 0.001 for TCA and −0.22% ± 1.55%, P = 0.144 for CVI). In the following 6 months, the dosage of atropine had switched to 0.01% daily. There were decreases in LA of −0.14 ± 0.24 mm², in SA of −0.16 ± 0.10 mm², and in TCA of −0.30 ± 0.29 mm², respectively, while CVI increased by 1.38% ± 1.55% (all P < 0.05). 

Table 1.

 

Longitudinal Changes of Refraction, Axial Length, and Choroidal Parameters in Different Treatment Groups

 

Figure 1.

 

Graphical display of longitudinal changes of choroidal parameters. Longitudinal changes of choroidal parameters during the 1-year period are shown graphically as mean ± 95% confidence intervals. (A) Longitudinal changes of total choroidal area. (B) Longitudinal changes of choroidal vascularity index. (C) Longitudinal changes of luminal area. (D) Longitudinal changes of stromal area. Values on the x-axis correspond to baseline, 1 week, 3 months, 6 months, and 1 year, respectively.

Changes in LA, SA, and TCA were subtler in group B than those in group A (Table 1 and Fig. 1). LA decreased slightly in the first 6 months (−0.05 ± 0.18 mm²; P = 0.026) and remained stable until the end of 1 year. SA increased marginally at the 6-month follow-up and then dropped slightly in the second 6 months (0.03 ± 0.09 mm² for the first 6 months and −0.06 ± 0.08 mm² for the second; both P < 0.05). As a result, TCA remained constant in both phases. An exception was CVI, which decreased by 0.96% ± 1.31% in phase 1 and increased by 1.19% ± 1.43% in phase 2. 

In sum, LA, SA, and TCA increased more in group A than in group B during the first 6 months. After switching to daily administration of 0.01% atropine, the increments diminished in the following 6 months. CVI, however, decreased less in group A than in group B in phase 1. Since the amount of CVI increase in the two groups in phase 2 was similar, the difference between them at 6 months was maintained until the end of 1 year. 

Correlation Among Ocular Parameters and Baseline Factors

During phase 1, changes between TCA and CVI were positively correlated (Pearson's coefficient: 0.37, P < 0.05). A similar correlation was further observed between LA and SA (Pearson’s coefficient: 0.76, P < 0.05). Increases in TCA, CVI, LA, and SA were positively correlated with shorter AL elongation (−0.73, −0.26, −0.71, and −0.67 for TCA, CVI, LA, and SA, respectively; all P < 0.05) and slower SE progression (Figs. 2A, 2B). Similar relationships were found during phase 2 (Figs. 2C, 2D). 

Figure 2.

 

Graphical display of correlations between changes in ocular parameters and baseline characteristics. Correlations between (A) changes of SE, AL, TCA, and CVI through phase 1 and baseline parameters. (B) Changes of SE, AL, LA, and SA through phase 1 and baseline parameters. (C) Changes of SE, AL, TCA, and CVI through phase 2 and 6-month parameters. (D) Changes of SE, AL, LA, and SA through phase 2 and 6-month parameters. In each plot, correlation coefficients are represented in the lower triangle. The same values are shown as colored bubbles in the upper triangle. Red bubbles refer to negative values, and blue bubbles refer to positive ones. The closer the values are to zero, the lighter and smaller their bubbles will be. Significant coefficients are marked as bolded numbers in the lower triangle and white stars in the upper triangle (significance level set at P < 0.05).

Changes in ocular parameters in phase 1 seemed to depend slightly on baseline characteristics. Only changes in CVI, LA, and SA were correlated with their baseline values (P < 0.05) (Figs. 2A, 2B). In phase 2, however, changes in all ocular parameters, except for LA, depended on their values at 6 months (baseline for phase 2; P < 0.05) (Figs. 2C, 2D). Additionally, AL changes were negatively correlated with age in both phases. 

Mediation Analysis

Four separate models were tested to examine the mediation effects of two different aspects of choroidal structure (specific indicators of choroidal structure [LA and SA] or comprehensive indicators of choroidal structure [TCA and CVI]) on the antimyopia effect of 1% atropine or the rebound after switching to 0.01% atropine (group A). The effects of 0.01% atropine on these parameters (group B) were treated as a reference. 

Model 1X and model 2X showed that TCA, but not CVI, partially mediated the effects of atropine treatment on AL in both phases. This, in turn, affected SE progression (34% of the total effect during phase 1 and 26% during phase 2; both P < 0.05) (Table 2 and Fig. 3; Supplementary Table S3 and Supplementary Fig. S4). The direct effects of atropine on SE were not significant in both phases (Supplementary Table S4 and Supplementary Fig. S4). Hence, the pathways “treatment → ΔTCA → ΔAL → ΔSE” and “treatment → ΔAL → ΔSE” completely mediated the antimyopic effects of 1% atropine (79%) and the rebound after switching to 0.01% atropine (85%). 

Table 2.

 

Specific Path Coefficients for Model in Figure 2 (Model 1X)

 

Figure 3.

 

Relationships between phase 1 treatment and changes in TCA, CVI, AL, and SE. A path diagram is represented to show the main results of model 1X. Bold arrows indicate significant effects (significance level set at P < 0.05), with unstandardized path coefficients shown as β (95% confidence intervals [CIs]). Multivariate mediation analyses were conducted in an SEM framework. The indirect path of phase 1 treatment → ↑ΔTCA → ↓ΔAL → ↑ΔSE was significant (unstandardized coefficient: 0.19; 95% CI, 0.12–0.27; P < 0.001). Thirty-four percent of treatment's total effect on SE progression was mediated through this pathway. In addition, the indirect path of phase 1 treatment → ↓ΔAL → ↑ΔSE was also significant (0.26; 95% CI, 0.16–0.37; P < 0.001), mediating 45% of the total effect.

Model 1Y showed that the effects of 1% atropine on AL during phase 1 were partially mediated by LA and SA, accounting for 25% and 17% of the total effect, respectively (both P < 0.05) (Supplementary Table S2). AL affected SE, resulting in three statistically significant pathways linking treatment to the prevention of SE progression. The coefficient of the pathway “treatment → ΔLA → ΔSE” was also significant. Overall, a total of four different pathways mediated the prevention of SE progression during phase 1 (91%, P < 0.05) (Supplementary Fig. S3). The total mediation effect was considered complete due to the insignificance of the direct effect. In model 2Y, there were two significant pathways. The pathways “treatment → ΔLA → ΔAL → ΔSE” and “treatment → ΔAL → ΔSE” mediated the effects on SE progression during phase 2 (22% and 60%, respectively, both P < 0.05; direct effect P > 0.05) (Supplementary Table S4 and Supplementary Fig. S5). 

Discussion

Considerable evidence has sprung up in the past two decades concerning the relationship between myopia and the choroid.^{8,10,11,29,30} Recently, a choroid-thickening effect was observed in individuals treated with atropine, which might delay the progression of myopia.¹⁵ However, questions remain: (1) How does atropine exert its choroid-thickening effect? (2) Are these effects reversible during the myopic rebound? (3) What is the causal relationship between the choroidal responses and myopic control? In this study, we demonstrated that the choroidal-thickening effect of 1% atropine was shared by growth in both LA and SA, while 0.01% atropine had little choroidal responses. The choroidal changes following 1% atropine treatment diminished after switching to 0.01% atropine. Second, increases in LA, SA, TCA, and CVI following atropine administration were associated with less axial elongation and SE progression. Third, mediation analyses showed that TCA, but not CVI, contributed to the changes in AL and SE. LA might explain the changes in myopic metrics during both atropine treatment and concentration switchover, while the antimyopic effects mediated by SA might persist after the adjustment of concentration. SA might be a potential biomarker to predict the postrebound treatment efficacy of atropine. 

Choroidal Responses of Atropine Treatment

Animal^9,17 and human^15,31 studies supported the choroidal-thickening effects of atropine, but the precise choroidal responses remain to be clarified. We found that the administration of 1% atropine increased both LA and SA. They returned to baseline levels after switching to 0.01% atropine, indicating that the choroidal thickening following treatment with 1% atropine could be explained by increases in both vascular and stromal components. The choroidal-thickening effects of 1% atropine¹⁷ as well as several other antimyopic strategies, including apomorphine,¹⁷ intense light,¹⁷ and orthokeratology,³² have been mainly attributed to the vascular components. However, we provided evidence for the participation of both components. 

At the same time, no meaningful LA or SA changes were observed in participants who received 0.01% atropine for the entire study period. This is not surprising. Sizable increases in choroidal thickness have been described in studies using atropine with concentrations higher than 0.01%, such as 0.025%,¹⁵ 0.05%,¹⁵ and 1%.¹⁶ But individuals treated with 0.01% atropine reported only negligible³³ or no¹⁵ choroidal-thickening effects. 

The CVI was higher in group A than in group B during the entire study period. Of interest, a slight decrease in CVI was found in group B during the first 6 months. In the following 6 months, the CVI returned to baseline values. This pattern was most probably due to minor changes in LA and SA that reversed over time. It is difficult to clarify whether the changes in CVI were a true pharmacologic effect. Further investigations with a longer follow-up period are needed to clarify the true nature of these changes. 

Association Between Choroidal Changes and Myopic Progression

We showed that 1% atropine administration was associated with antimyopic effects and choroidal changes, but whether there is a causal relationship between choroidal changes and retardation of myopic progression is unclear. Previous animal studies,^8,9 cross-sectional studies,^10,11 and longitudinal studies^12,13 have revealed a strong or even causal relationship between choroidal thinning and myopia. The structural changes underlying choroidal thinning have been widely studied with the application of binarization techniques to OCT images.²² Some observational studies described that the severity of myopia was accompanied by reductions in either the luminal³⁴ or the stromal^35,36 areas of the choroid, while others reported decreases in both components.^14,37 In our study, increases in LA and SA were associated with less axial elongation and less SE progression, which may support an overall decline of choroidal components during myopia progression. We have also shown that an increased CVI correlated with less progression in SE and AL. However, previous studies reported positive,³⁷ negative,¹⁴ or no^34,38 correlations. So, it is still doubtful whether the CVI is a proper parameter to evaluate the changes in choroidal blood flow during myopia. 

Considering the relationships among choroidal parameters, LA and SA were highly correlated at baseline (r = 0.9). Moreover, a strong association was found between the changes in the two parameters (r = 0.76), suggesting that vascular and stromal tissues in the choroid might be finely interrelated. Alternatively, the SA might account for both stromal and choriocapillaris areas due to some limitations in image resolution. The choriocapillaris area might have been affected by changes in choroidal blood flow, thus influencing SA. Large-scale longitudinal studies are needed to confirm the relationship between LA and SA. 

Potential Choroidal Mechanisms Underlying the Antimyopic and Myopic-Rebound Effects of Atropine

Both TCA and ChT are partial displays of the choroidal volume. So, the changes in either parameter can be easily interconverted once the scan region is fixed. We revealed that TCA mediated 34% of the antimyopic effect gained by 1% atropine over 0.01% atropine. This mediating role still existed for the reverse myopic changes after the concentration switchover. In accordance, the Low-Concentration Atropine for Myopic Progression (LAMP) phase 2 study showed that changes in subfoveal choroidal thickness mediated 18.45% of the antimyopic effects of 0.05% atropine over 0.01% atropine during a 2-year period.¹⁵ Despite distinct atropine concentrations, different follow-up periods, and separate scan regions, we provided further evidence for the contribution of increased choroidal volume in atropine's antimyopic mechanism. Interestingly, in this study, the degrees of choroidal thickening (around 20 µm) and SE progression (around 0.55 D) in 1% atropine for 6 months were similar to those in the LAMP 2 study using 0.05% atropine for 2 years. In both studies, the choroidal thickness increased very early and remained stable until the end of the treatment. Assuming the existence of a similar trend for 2 years with greater (possibly twice⁵) antimyopic effects, the TCA would mediate less than 34% of the antimyopic effects of 1% atropine, possibly around 17%, as described with 0.05% atropine (18.45%). The predicted antimyopic effects of 1% atropine over 0.05% atropine at 2 years might be achieved through mechanisms not involving the choroidal volume. 

A greater mediation effect of TCA was described during the first 6 months with 1% atropine compared to the following 6 months with 0.01% atropine (unstandardized coefficient: 0.19 vs. −0.10). This means that 6 months of treatment with 1% atropine led to a 0.19-D advantage over 0.01% atropine through TCA. After the concentration switchover, the magnitude of the rebound along this pathway was only 0.1 D. Considering the mediation analysis for the specific indicators, LA and SA contributed to the antimyopic effect of atropine treatment by 0.11 D and 0.08 D, respectively. However, only the mediation effect of LA remained significant in the myopic-rebound phase (−0.08 D). The asymmetric effects of TCA might be explained by the involvement of LA in both phases and the absence of SA in the myopic-rebound phase. It is possible that the effects of SA on myopia can persist after a concentration switchover toward 0.01% atropine. The myopic rebound is one of the greatest concerns during the clinical administration of high-dose atropine. SA might be a biomarker to predict the postrebound efficacy of atropine. 

We also found that CVI was completely independent of either treatment effects or the myopic rebound. Considering the controversial relationship between CVI and myopia, we speculated that CVI may not be a proper parameter to evaluate the choroidal responses to atropine. 

Limitations

This study is affected by some limitations. First, we administered a 1-week loading dose of 1% atropine in group A. Although the consistency of the intervention was not fully achieved, the 1-week loading dose was necessary to demonstrate the early choroidal responses to atropine. Second, a period of 6 months may not be enough for assessing the antimyopic effects of atropine or the rebound after concentration switchover. Third, we did not implement a placebo group. However, no difference in choroidal thinning was previously found between 0.01% atropine and a placebo group, making 0.01% atropine a proper treatment control.¹⁵ Last, due to the exploratory nature of the study, no adjustment for multiple testing was taken. The results of this study should be considered hypothesis generating and in need of further confirmation. 

Conclusions

The 1% atropine treatment led to significant growth in ChT by increasing LA and SA. On the contrary, 0.01% atropine was associated with only a few choroidal responses. The choroidal changes induced by 1% atropine reversed after switching to 0.01% atropine. TCA, but not CVI, accounted for antimyopic and rebound mechanisms of 1% atropine. LA may underlie the myopic changes during atropine treatment and concentration switchover, while the effects of SA may persist even after the adjustment of atropine concentration. SA might be a biomarker to predict the postrebound efficacy of atropine. 

Acknowledgments

The authors thank Shanghai Medoo Tech Co., Ltd. for the technological support of this trial. 

Supported by Shanghai Health Committee, Clinical Research (Project No. 2019240241), Shanghai Shenkang Hospital Clinical Research Program (Project No. SHDC12019X18, SHDC12020127), National Key R&D Program of China (Project No. 2019YFC0840607), and National Science and Technology Major Project of China (Project No. 2017ZX09304010). 

Disclosure: H. Xu, None; L. Ye, None; Y. Peng, None; T. Yu, None; S. Li, None; S. Weng, None; Y. Huang, None; Y. Chen, None; Y. Fan, None; H. Zou, None; J. He, None; J. Zhu, None; X. Xu, None