This study is a retrospective, non-comparative, observational cohort study performed after obtaining approval from the Institutional Review Board. All study-related procedures adhered to the guidelines of good clinical practice (International Conference on Harmonization of Technical Requirements of Pharmaceuticals for Human Use [ICH] E6), applicable FDA regulations, the Health Insurance Portability and Accountability Act, and the Declaration of Helsinki. The electronic medical record was used to identify patients aged 18 years or older with a documented diagnosis of DMO at the Cleveland Clinic Cole Eye Institute from January 2012 to December 2019.

Patients were included if they met the following conditions: initiation of intravitreal anti-VEGF therapy at the Cleveland Clinic for DMO without prior anti-VEGF treatment, OCT scans at baseline, 3, 6, 9, and 12 months (+/− 2 weeks), and follow-up at 3 years or 5 years after the first injection (+/− 3 months). With regards to the treatment protocol, patients received an anti-VEGF injection administration regimen according to the discretion of the clinician, which included medications such as bevacizumab, aflibercept, ranibizumab, or mixed injections. The medications were administered at varying intervals per the discretion of the clinician, with the intention of treating all fluid until dry. To avoid duplication of data and potential biases, in cases where both eyes were eligible only the eye that first received anti-VEGF treatment was selected. A total of 270 patients with a confirmed diagnosis of DMO who were undergoing anti-VEGF treatment at the Cleveland Clinic were identified and screened for the study (Supp Fig. 1). Among them, 184 patients had follow-up data available at the 3-year mark (86 patients were excluded), and 138 patients had follow-up data available at the 5-year mark (132 patients were excluded). Patients were excluded from the study if their eyes had concomitant maculopathies not related to DMO, or if they received steroid injections or focal laser photocoagulation treatment during the study period. Patients who received other ophthalmologic interventions outside of the exclusion criteria (i.e., cataract surgery) during the study period were included in the study.

At the beginning of the study, participant demographics and medical history were gathered, along with details of the treatment administered during the initial visit. BCVA and central subfield thickness (CST) were measured at the baseline, 3-month, 6-month, 9-month, 12-month, 3-year, and 5-year follow-up visits. All BCVA and CST measurements were taken during the same visit. The BCVA measurements were a combination of VA with or without correction, as well as pinhole VA, following the institutional standard protocols. To determine macular thickness parameters, including CST, CV, and CAT (retinal cube average thickness), the Cirrus High-Definition Spectral Domain-OCT Review (V.9.5.1, Carl Zeiss Meditech, Dublin, CA) was utilised [16, 17]. To assess macular thickness variability, the CST standard deviation (SD), CST amplitude, and CST area under the curve (AUC) were measured using the recorded CST values from the end of the loading phase (3 months after treatment initiation) to 12 months.

R Statistical Software (Version 3.6.1, Vienna, Austria) was used to perform statistical analysis. Categorical variables were described in the form of frequencies and percentages, while continuous variables were described using means ± SD. Eyes were stratified into quartiles based on macular thickness variability, with quartile 4 representing the highest degree of variability. Quartiles were used over continuous analysis in order to further visualise the relationship between macular thickness variability measures and visual outcomes, minimise the impact of outliers, and provide a clearer picture of risk strata for potential clinical applications. VA was converted from Snellen to ETDRS via the formula ETDRS  =  85  +  50 × log10(Snellen) [16]. Continuous variables such as final BCVA, change in BCVA, final CST, and CST change were compared among quartiles by Kruskal–Wallis multiple comparisons with Dunn’s post-hoc analysis, while categorical variables were compared through chi-square tests. Afterward, multiple linear regression analyses were performed to evaluate whether CST SD, choroidal volume (CV) SD, or cube average thickness (CAT) SD are correlated with final BCVA, while accounting for other factors at baseline. A significance threshold was set at p <  0.05.