The choroid is a vascular layer between the retina and the sclera¹. Choroidal thickness (ChT) correlates with myopia; eyes with higher degrees of myopia and longer axial lengths tend to have thinner choroids^2–5. Experimental defocus studies align with this biology: myopic defocus thickens the choroid and slows eye growth, while hyperopic defocus thins the choroid and accelerates elongation⁶. Accordingly, ChT is being pursued as a rapid structural biomarker of myopia risk⁷ and, more recently, as an indicator of the effectiveness of myopia control^8–10.

Atropine is a cornerstone of myopia control, with putative mechanisms that span retinal neurotransmission, scleral remodeling, and choroidal perfusion^11,12. Research shows a dose-dependent relationship between atropine efficacy and adverse reactions for myopia control^13,14. Low-dose atropine offers prolonged effects and reduced adverse reactions¹⁵. Yam et al. have shown that nightly use of 0.05% atropine eyedrops resulted in a significantly lower incidence of myopia with rapid myopic shift at 2 years in pre-myopic children¹⁶. Nevertheless, it appears that school-age children in the United States who have low to moderate myopia are insensitive to 0.01% atropine¹⁷. This discovery aligns with the general agreement that the effectiveness of 0.01% atropine still needs to be consistently confirmed and further tested among various ethnic populations¹⁸.

Research findings indicate that, while controlling myopia, atropine delays the thinning of the choroid and may even lead to a thickening trend in ChT¹⁹. For example, Yang et al.²⁰ suggest that 0.01% atropine causes choroidal thickening in myopic children in the short term.

However, most prior work has evaluated atropine’s choroidal effects in already myopic children, typically with infrequent imaging and coarse spatial assessments focused on subfoveal thickness²¹. There is still a significant lack of evidence regarding pre-myopia, a high-risk condition that precedes clinical myopia, in which early intervention could alter growth patterns^22,23.

In order to address these gaps, we carried out a prespecified imaging companion analysis within the randomized AMPP cohort, employing dense 12-month follow-up and 15-sector OCT mapping to detail the temporal–spatial profile of ChT with nightly 0.01% atropine. Recently, we reported that 0.01% atropine reduced myopia onset and axial elongation compared to placebo in a one-year randomized trial in pre-myopes²⁴. Then, using mixed-effects models, we mapped sector-level ΔChT and annular gradients versus placebo, yielding structural evidence that supports refractive and axial findings and informs mechanism-based prevention.

Children were examined at baseline and at 3, 6, 9, and 12 months. To minimize diurnal variation in ChT, all SS-OCT/OCTA imaging (VG200S, SVision Imaging, Henan, China) was scheduled in the mid-afternoon (typically 14:00–16:00), consistent with prior work restricting acquisition to late-morning or afternoon windows^28,29. Cycloplegia (1% cyclopentolate; two drops 5 min apart; 30-min wait) preceded autorefraction (mean of five readings; KR-800, Topcon, Japan) and axial length (AL) by optical biometry (mean of three; Lenstar LS-900, Haag-Streit, Switzerland) at each visit. A complete ophthalmic examination included slit-lamp biomicroscopy, best-corrected visual acuity (BCVA), and intraocular pressure (IOP; non-contact tonometer CT-1, Topcon, Japan). To explore phenotypic patterns in ChT dynamics and their relationship to myopia progression, eyes were additionally categorized as Thinning-Dominant or Stable ChT based on baseline ChT and progression criteria.

Macular imaging was performed using swept-source OCT/OCTA (SS-OCT/OCTA) with enhanced depth imaging. A 6×6-mm macular volume centered on the fovea was acquired for each eye, and the built-in software segmented the retinal pigment epithelium and the choroidal-scleral interface to generate a ChT map. Images with low quality (signal metric ≤ 50), motion artifacts, or segmentation errors were re-acquired. Remaining segmentation errors were corrected by two trained graders, with adjudication by a senior reader.

ChT quantification followed a modified ETDRS layout: the central 1-mm circle, an inner annulus (1–3 mm radius), and an outer annulus (3–6 mm), each subdivided into superior, inferior, nasal, and temporal sectors, yielding a total of 15 subfields. The procedures replicate established pediatric ChT mapping and quality-control techniques for the same platform. A visual representation of the imaging procedure and ChT map segmentation can be seen in Fig. 2A-C, which illustrates the areas of measurement, including the macular sectors and segmentation for ChT analysis.

The primary outcome was the change in choroidal thickness (ΔChT), defined as ChT at each follow-up minus baseline (ΔChT = ChT_visit − ChT_baseline). ChT was quantified in 15 macular subfields at 3, 6, 9, and 12 months, and compared between the atropine and placebo groups. Key secondary outcomes were the 12-month ΔChT for each subfield and ring-averaged ΔChT across three concentric zones (central 0–1 mm, inner 1–3 mm, outer 3–6 mm). Inter-ocular symmetry was evaluated by correlating ΔChT between right and left eyes. Structure–function analyses examined the relationship between ChT and myopia progression: cross-sectional correlations related baseline ChT to 12-month changes in SER and AL, and longitudinal mixed models tested whether time-varying ChT (or ΔChT) was associated with trajectories of AL and SER.

ChT exhibited a characteristic temporal pattern over the 12-month study period. In both the atropine and placebo groups, ChT initially decreased from baseline, reaching its maximum mean reduction between 6 and 9 months, followed by a mild rebound at 12 months (Figure S1 in Supplementary Material). The most significant thinning (negative ΔChT) was observed in the majority of the 15 macular subfields during either the 6- or 9-month check-up. For example, in the subfoveal region (0–1 mm), the mean ΔChT reached its maximum at 9 months (approximately − 8.7 µm from baseline), with some recovery by month 12 (mean ΔChT of -5.8 µm). By the 12-month visit, ChT was generally less than at the 9-month visit, suggesting a mild late rebound, with the inferior outer sector as the exception(Table S1 in Supplementary Material).

Cumulative analyses corroborated these findings, revealing a spatially localized pattern of ChT change. In the 15 ETDRS subfields, atropine-treated eyes experienced less cumulative thinning than those given a placebo, notably in the central macula, with the greatest distinction in the central 0–1 mm subfield(Fig. 3). The differences in the 1–3 mm and 3–6 mm rings were less pronounced, with the greatest separation between groups occurring in the central and inner regions. In addition, the inferior and temporal sectors of the 1–3 mm and 3–6 mm rings showed more significant differences than the superior and nasal sectors, indicating that atropine provided greater protection along the inferior and temporal meridians.

In children who are pre-myopic, using 0.01% atropine at night reduced macular choroidal thinning over 12 months, particularly in the central area, leading to a smaller overall loss and a slower rate of decline compared to a placebo. Both groups exhibited progressive thinning across sectors, but the atropine group demonstrated less cumulative loss and a slower rate of decline, consistent with the sector-wise and time-resolved maps. In addition to average effects, unsupervised trajectory clustering identified two consistent phenotypes: a Thinning-dominant group showing significant mid-year thinning with some recovery by month 12, and a Stable ChT group with little overall change. Importantly, eyes in the Thinning-dominant cluster showed faster axial elongation than Stable eyes in longitudinal models, supporting a structure–function link between choroidal remodeling and ocular growth.

The observation that atropine slows, but does not abolish, ChT is consistent with a modulatory rather than binary action on growth signaling³⁰. Administering atropine nightly for six months also prevents the choroidal thinning that occurs in response to short-term hyperopic defocus when atropine is not used⁶. The group differences emerged mainly at 6–9 months, consistent with cumulative tissue remodeling rather than an acute drug effect³¹. One hypothesis is that atropine gradually upregulates choroidal perfusion or extracellular matrix deposition, thereby counteracting the thinning that usually precedes rapid eye growth^32,33. The localized response, where the inferior and temporal macular areas exhibit the most thickness preservation, could indicate differences in choroidal structure or vulnerability to stretching^34–36. The strong interocular symmetry in ΔChT (r ≈ 0.5–0.7) indicates a stable, subject-level signal rather than random measurement noise, supporting the use of single-eye readouts in subsequent analyses. Ultimately, the difference between Thinning-dominant and Stable paths suggests an underlying biological diversity—possibly genetic, behavioral, or anatomical at baseline—that influences how effective low-dose atropine is on the choroid and axial growth.

These results carry several implications for clinical practice and the future of myopia prevention. First, ChT emerges as a promising real-time biomarker to guide therapy in children at risk. Since the choroid reacts within weeks to months, which is quicker than changes in refractive error or axial length, regular OCT could serve as a 'structural feedback' to assess the effectiveness of treatment by monitoring ChT. For example, an initial rise or stabilization in ChT following the start of low-dose atropine might suggest a favorable response ('target engagement'), while ongoing choroidal thinning despite treatment might indicate a less effective response. This data might lead to timely changes, like raising the atropine dosage, enhancing compliance, or incorporating additional treatments (such as orthokeratology or peripheral defocus contact lenses) for those who respond poorly⁶. If significant choroidal stabilisation or mild reduction occurs in the first 6–9 months, this reassures clinicians about the current treatment. Thus, ChT might act as a substitute endpoint in initial myopia control studies and tailored treatments, speeding up decision-making before notable myopic advancement happens. Secondly, our findings underscore the value of early intervention at the pre-myopic, often preschool stage. The ability of 0.01% atropine to delay choroidal thinning and possibly myopic shift suggests that treating high-risk children before they become myopic can alter ocular growth trajectories. This preventive strategy might lead to a significant decrease in new cases of myopia or at least delay its onset to later ages, thereby reducing the lifetime risk of severe myopia and related health issues. Finally, the concept of a 'choroidal thickness reserve' emerges: children who maintain a thicker choroid may be more resilient against myopic elongation, hinting that future therapies might aim to bolster choroidal structure (e.g., red-light therapy has also been shown to sustain choroidal thickening and slow myopia³⁷). Overall, our research advocates for a change in approach towards utilizing imaging biomarkers such as ChT to tailor myopia prevention – initiating treatment sooner, keeping a close watch, and making smart adjustments based on each child's structural response.

Despite its strengths, this study has limitations that temper the interpretation of our findings. The sample size was modest (n = 54), conferring limited statistical power to detect small between-group differences or rare adverse effects. Certain trends, like the marginally reduced axial elongation in eyes treated with atropine and the faster growth in the thinning phenotype, were not statistically significant, necessitating a larger study to confirm these effects conclusively. The follow-up duration of one year, while clinically relevant, is relatively short in the context of myopia development. Many placebo-group children remained pre-myopic after 12 months, so the full impact of atropine on preventing actual myopia onset may only manifest over 2–3 years or longer. Another factor to consider is generalizability: our study involved ethnically Chinese children from a single center, and the findings might vary in different populations or climates. Whether 0.01% atropine is effective in all contexts is still uncertain, with recent U.S. data suggesting it may not work as well for older school-age children¹⁷. Further studies should investigate if increased concentrations, such as 0.025% or 0.05%, or combined treatments provide greater choroidal and refractive advantages for children at risk of myopia. Additionally, while our clustering analysis revealed intriguing phenotypes, the underlying causes of these divergent responses are unknown. Genetic influences, variations in behavior (such as near work and time spent outdoors), or initial eye measurements might play a role and deserve further study in larger datasets. Choroidal thickness is only one facet; adding functional metrics (accommodation, peripheral refraction) and imaging biomarkers (scleral thickness, choroidal perfusion) would yield a fuller view of low-dose atropine’s effects on eye growth^35,38. Large, multi-center, long-term studies in diverse populations are needed to validate ChT-guided early interventions and pave the way for mechanism-driven, personalized myopia prevention in children.

In conclusion, nightly application of 0.01% atropine in pre-myopic children reduced macular choroidal thinning rather than preventing it, and eyes with greater thickness preservation generally showed slower axial elongation.Two choroidal response phenotypes, Thinning-dominant and Stable ChT, were identified, which aid in understanding individual variability in results. These findings advocate for early intervention and suggest that choroidal thickness can serve as a useful structural biomarker for guiding and personalizing myopia prevention through regular OCT. To confirm these results and implement ChT-guided strategies in clinical practice, larger, longer, and more varied studies are necessary.