Choroidal Thickness and Age Predict Ongoing Eye Growth in Adults with High Myopia

In most adults, the eyeball stops growing. But in people with high myopia — severe nearsightedness defined by a refractive error of −6.00 diopters or more, or an axial length of at least 26.5 millimeters — elongation of the eye can quietly continue for decades, stretching the retina and raising the risk of irreversible complications such as myopic maculopathy and retinal detachment. A new analysis from the Wenzhou High Myopia Cohort Study, published in Eye and Vision, reports that two simple measurements — the thickness of the choroid, the vascular layer nourishing the retina, and the patient’s age — can identify which adults are most likely to experience this continued growth, with an accuracy that rivals far more complex models.

The prospective cohort study, conducted at the National Clinical Research Center for Ocular Diseases in the Eye Hospital of Wenzhou Medical University, enrolled 1,025 eyes from 532 adults with high myopia aged 18 to 60 years and followed them for two years. Participants underwent comprehensive examinations at baseline, including axial length measured with optical biometry, best-corrected visual acuity, intraocular pressure, cycloplegic autorefraction, and swept-source optical coherence tomography (OCT) imaging of the choroid. The researchers defined rapid axial elongation as growth of at least 0.10 millimeters over two years, and tested nine candidate predictors drawn from prior literature: age, sex, axial length, spherical equivalent refraction, visual acuity, intraocular pressure, choroidal thickness, choroidal vascularity index, and myopic macular degeneration graded by the META-PM system.

The results quantify just how variable adult eye growth can be. Across the cohort, the mean annual axial elongation rate was 0.020 ± 0.051 millimeters per year, but 24.78 percent of eyes crossed the rapid-elongation threshold. Nearly 58 percent of eyes grew less than 0.05 millimeters over the full two years, while almost 14 percent grew 0.15 millimeters or more — a range that matters clinically, because even fractions of a millimeter of stretching in an already elongated globe increase mechanical stress on the retina and choroid.

Among the nine candidate variables, choroidal thickness stood out decisively. Measured as the axial distance between the posterior border of the retinal pigment epithelium and the anterior surface of the chorioscleral interface, with image magnification corrected for each eye’s axial length, choroidal thickness alone achieved an area under the receiver operating characteristic curve (AUC) of 0.728 (95 percent confidence interval: 0.678 to 0.778), with an optimal cut-off of 138 micrometers determined by the maximum Youden index. Age was the next strongest single predictor (AUC = 0.666), followed by spherical equivalent refraction (AUC = 0.648) and axial length (AUC = 0.630). No other individual variable performed appreciably better than chance.

Using bidirectional stepwise regression with generalized estimating equations to account for the correlation between a patient’s two eyes, the team retained six predictors — age, sex, axial length, best-corrected visual acuity, choroidal thickness, and choroidal vascularity index — in a full multivariable model that achieved an AUC of 0.811 (95 percent CI: 0.769 to 0.852). Removing either age or choroidal thickness significantly impaired performance: excluding age dropped the AUC to 0.735 and excluding choroidal thickness to 0.753 (both DeLong test P < 0.001), with integrated discrimination improvement values of −0.116 and −0.095 respectively (both P < 0.001). By contrast, dropping sex produced only a marginal loss (IDI = −0.008, P = 0.032), and removing axial length, vascularity index, or visual acuity changed performance negligibly.

The striking result was how little the full model gained from its extra variables. A simplified model using only choroidal thickness, age, and sex reached an AUC of 0.794, statistically comparable to the full model (DeLong test, P = 0.084), and a bivariate model with just choroidal thickness and age performed nearly identically (AUC = 0.793, 95 percent CI: 0.749 to 0.836; P = 0.067 versus the full model). Internal validation with 1,000 bootstrap resamples yielded optimism-corrected AUCs of 0.799 for the full model and 0.790 for the simplified model — an optimism bias of only 0.003 — and calibration plots showed close agreement between predicted and observed probabilities.

One of the most clinically useful outputs is an age-specific risk chart. In the bivariate model, the choroidal thickness cut-off signalling elevated risk declined progressively with age, from 187 micrometers at age 20 to just 29 micrometers at age 60. For any given age, the predicted probability of rapid elongation rose as the choroid thinned. This interaction reflects a genuine biological pattern rather than a statistical artifact: locally weighted scatterplot smoothing and a generalized additive mixed model — adjusted for age, sex, baseline axial length, refraction, visual acuity, and myopic maculopathy — confirmed a significant nonlinear relationship between baseline choroidal thickness and annual elongation rate (EDF = 3.27, F = 17.78, P < 0.001), with an inflection point near the 138-micrometer threshold. Below that point, thinner choroids were associated with faster elongation; above it, the association largely disappeared.

The study population itself shapes how these numbers should be read. Of 810 patients screened, 532 were included; the mean age was 36.14 ± 9.83 years, 71.8 percent were female, mean axial length was 27.02 ± 1.15 millimeters, and mean spherical equivalent refraction was −8.46 ± 2.03 diopters. Most participants had relatively healthy fundi and early-stage myopic maculopathy, so the authors caution that generalizability to patients with advanced macular degeneration remains uncertain. The analysis also included only a limited set of demographic and ocular variables, and although internal validation was strong, external validation in independent cohorts has not yet been performed. The risk threshold optimized by the maximum Youden index corresponded to roughly 28.9 percent predicted probability — a deliberately conservative cut-off the authors justify by the already elevated baseline axial lengths in this population.

The findings carry practical weight because both key predictors are already part of routine ophthalmic practice. Choroidal thickness is easily measured on standard OCT platforms, and automated segmentation is now commercially available. The authors suggest a risk-stratified framework: adults whose baseline choroidal thickness falls below their age-specific threshold could receive intensified monitoring and, following personalized risk–benefit assessment, consideration of proactive interventions — options that currently include specialized optical lens designs, low-dose atropine, or posterior scleral reinforcement, though recommendations for adults rest largely on expert consensus and extrapolations from pediatric research rather than direct trial evidence in this age group. Lower-risk individuals, in turn, might safely extend their monitoring intervals.

Important caveats remain before such a framework becomes standard care. The study predicts axial elongation itself, not the development of vision-threatening complications, and the authors note that longitudinal studies with extended follow-up are needed to confirm whether managing patients according to these risk profiles actually prevents sight-threatening outcomes. Still, by converting previously unquantified risk factors into a validated, interpretable two-variable model — and by revealing a threshold below which the choroid appears to signal an eye still actively growing — the Wenzhou cohort offers clinicians a concrete starting point for personalized monitoring, and gives researchers a clear target for the next generation of choroid-centered studies in adult myopia.

Subject of Research: Prediction of continued axial eye elongation in adults with high myopia using choroidal thickness and age

Article Title: Prediction of axial elongation in adults with high myopia: the Wenzhou High Myopia Cohort Study

Article References: Wang, Y., Rui, X., Sun, Y., Li, Y., Ding, S., Zhang, Z., Lu, J., Wu, M., Qu, J., Zhou, X., & Wu, H. (2026). Prediction of axial elongation in adults with high myopia: the Wenzhou High Myopia Cohort Study. Eye and Vision, 13(1), Article 26. https://doi.org/10.1186/s40662-026-00496-y

Image Credits: AI Generated

DOI: 10.1186/s40662-026-00496-y

Keywords: high myopia, axial elongation, choroidal thickness, optical coherence tomography, prospective cohort study, risk prediction, myopic maculopathy, Wenzhou, adult myopia, OCT imaging, predictive model, eye growth