Large Chinese Cohort Yields Model Predicting Vision Loss After Optic Nerve Stroke

Researchers in China have built one of the largest clinical pictures to date of non-arteritic anterior ischemic optic neuropathy, or NAION, a sudden and often devastating stroke of the optic nerve. Drawing on a nationwide multicenter database, the team identified the factors that drive involvement of the second eye and constructed a statistical model that predicts which patients will end up with severely reduced vision. The work, published in Eye and Vision, offers clinicians treating this poorly understood condition a quantitative framework grounded in Chinese patients, a population that previous studies, dominated by Western cohorts, have largely overlooked.

NAION is the most common optic neuropathy in adults over fifty, although it also strikes younger people. It typically arrives as acute, painless, one-sided vision loss accompanied by a relative afferent pupillary defect and sectorial or diffuse swelling of the optic nerve head. The underlying mechanism is thought to involve hypoperfusion of the posterior ciliary arteries combined with intrinsic vascular risk factors, with a structurally crowded, small optic disc acting as a key anatomical predisposition. Systemic conditions such as arterial hypertension, diabetes mellitus, and hyperlipidemia have been implicated, and certain medications, notably phosphodiesterase-5 inhibitors used for erectile dysfunction, have been reported as potential precipitants.

The new analysis drew on the Chinese Neuro-Ophthalmic Diseases Project Database, a bidirectional observational national multicenter cohort study. The researchers retrospectively identified patients diagnosed or confirmed by neuro-ophthalmologists at tertiary referral hospitals across China between September 2020 and December 2024. Diagnosis required acute monocular painless visual loss, a relative afferent pupillary defect, sectorial or diffuse optic disc edema, and arcuate or altitudinal visual field defects. Patients with inflammatory, hereditary, traumatic, or other optic nerve or retinal disease were excluded, as were those with arteritic anterior ischemic optic neuropathy, distinguished through inflammatory markers, temporal artery biopsy findings, and FDG-PET or CT imaging of vasculitis where available.

After exclusions, 895 patients entered the study, contributing 1,128 affected eyes. The group was predominantly male, with 560 men making up 62.57 percent of the cohort, and the mean age at onset was 55.6 years, with a standard deviation of 10.4. Nearly 72 percent of patients were older than fifty, yet roughly 30 percent were diagnosed younger, a reminder that NAION is not exclusively a disease of later life. Bilateral involvement, the feared complication in which the second eye loses vision, was present in 226 patients, or 25.3 percent. Only 36 of these had simultaneous onset within seven days, and the five-year incidence of contralateral eye involvement was 14.5 percent.

To find what pushes the disease into the fellow eye, the team ran univariate and then multivariate stepwise regression on the full cohort. Age, sex, and body mass index did not differ significantly between unilateral and bilateral groups. In the multivariate analysis, three factors emerged with a significant positive association with bilateral occurrence: hyperlipidemia, hematologic disorders, and smoking history. Univariate analysis had also flagged female sex, coronary artery or cerebrovascular disease, and alcohol consumption, but the final multivariate model narrowed the picture. The authors suggest that systemic changes in blood state may alter the quality and quantity of flow through the short posterior ciliary arteries supplying both optic nerves, compounding the anatomical predisposition of crowded discs.

The second half of the study tackled prognosis. Among patients with available follow-up best-corrected visual acuity, 522 patients with 730 eyes were enrolled in modeling. At the last follow-up, at least three months after onset, 369 eyes, or 50.6 percent, had a best-corrected visual acuity of 0.3 decimal or worse, the threshold defining a poor outcome. Nearly 30 percent of eyes ended at 0.1 or worse. The cohort was randomly split, with 75 percent forming a training set of 392 patients and 547 eyes, and 25 percent a validation set of 130 patients and 183 eyes. Least absolute shrinkage and selection operator regression screened candidate variables, and logistic regression built the final predictive model.

The model identified four independent predictors of a poor visual outcome. A lower visual acuity at onset carried the strongest signal, with an odds ratio of 0.72 per unit change, meaning worse starting acuity predicted worse final acuity. Diabetes mellitus roughly tripled the odds, with an odds ratio of 2.97. Hyperhomocysteinemia showed the largest effect, with an odds ratio of 11.59, though with a wide confidence interval reflecting fewer cases. Macular serous detachment, fluid accumulating under the retina, more than six-fold increased the odds, consistent with its role as a marker of severe disc edema that may damage photoreceptors when persistent. Interestingly, peripapillary hemorrhage at baseline was independently associated with better recovery, an odds ratio of 0.19, which the authors interpret as a sign of early, reversible disc edema in patients who presented within seven days of onset rather than irreversible axonal damage.

Performance was strong and consistent across the split cohorts. The area under the receiver operating characteristic curve was 0.87 in the training cohort and 0.86 in the validation cohort, and calibration curves showed predicted probabilities closely tracking observed outcomes in both groups. The authors note that treatment in this cohort was not standardized and reflected clinical judgment, ranging from observation alone in 45.2 percent of patients to intravenous or oral corticosteroids and routine risk-factor modification, with no comparative efficacy analysis performed, so the model reflects the natural and routinely managed course of the disease rather than any specific therapy.

The study carries important limitations. It was retrospective, lacking prospective longitudinal follow-up, and optic nerve head optical coherence tomography and angiography parameters were unavailable because the imaging protocol targeted the macula. Optic disc drusen, a known anatomical risk factor, could not be systematically assessed without enhanced depth imaging. Nocturnal hypotension, measured by ambulatory blood pressure monitoring, was excluded from the final model because such data existed for only 44.1 percent of patients, and unilateral status was classified at the last available follow-up without accounting for censoring, introducing potential time-dependent bias. Generalizability beyond Chinese populations is also unestablished, particularly since population-based studies show Chinese individuals have a larger mean optic disc size than Caucasians, and a small disc is a major NAION risk factor.

Even with those caveats, the scale of the cohort gives the findings unusual weight. Hyperlipidemia affected 48 percent of patients, hypertension 37 percent, and diabetes 25 percent, matching published ranges for NAION comorbidity. As the first and largest nationwide NAION cohort in China, the dataset provides a foundation for the prospective studies with fixed follow-up intervals and comprehensive optic nerve imaging that the authors say are needed to confirm and extend these predictions.

Subject of Research: Risk factors for bilateral involvement and visual prognosis modeling in Chinese patients with non-arteritic anterior ischemic optic neuropathy

Article Title: Development of a predictive model for visual outcomes and risk factors of bilateral involvement for non-arteritic anterior ischemic optic neuropathy (NAION) in Chinese patients

Article References: Xu, X., Sun, M., Cao, S., Pan, C., Wang, Y., Chen, B., Lee, A. G., Jonas, J., Xu, Q., Wei, S., & Zhou, H. (2026). Development of a predictive model for visual outcomes and risk factors of bilateral involvement for non-arteritic anterior ischemic optic neuropathy (NAION) in Chinese patients. Eye and Vision, 13(1), Article 34. https://doi.org/10.1186/s40662-026-00504-1

Image Credits: AI Generated

DOI: 10.1186/s40662-026-00504-1

Keywords: NAION, optic neuropathy, visual prognosis, bilateral involvement, predictive model, hyperlipidemia, smoking, diabetes mellitus, hyperhomocysteinemia, macular serous detachment, Chinese cohort, retrospective study