Cell Density Mapped Retinal Nonperfusion and Low-Luminance Visual Acuity in Diabetic Retinopathy
Ophthalmology Science · 2026-12 · 6 (12) · 101435
DOI: 10.1016/j.xops.2026.101435 ↗Kurz gefasst
Fachartikel in der Zeitschrift Ophthalmology Science, 2026-12. Verfasst von Kallista Zhuang und 5 weiteren Autor:innen. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Institutes of Health, Japan Society for the Promotion of Science, Boehringer Ingelheim.
Abstract (Originalsprache)
Objective
To determine whether cell density-weighted retinal nonperfusion, derived by mapping ultra-widefield fluorescein angiography (UWF-FA)-defined nonperfusion to rod, cone, and retinal ganglion cell topography, is associated with low-luminance visual acuity (LLVA) in eyes with diabetic retinopathy (DR) and provides incremental explanatory value beyond conventional area-based retinal nonperfusion (RNP).
Design
Cross-sectional observational study.
Participants
213 eyes from 140 adults with diabetes across the spectrum of DR severity.
Methods
Retinal nonperfusion was manually delineated on UWF-FA images.Using normative retinal cell density maps, rod, cone, and retinal ganglion cell affected ratios (Rod-AR, Cone-AR, and RGC-AR) were calculated within the Early Treatment Diabetic Retinopathy Study (ETDRS) 7-field region.LLVA was measured using ETDRS letter scores with a 2.0-log unit neutral density filter.Linear mixed-effects models with a subject-specific random intercept accounted for inter-eye correlation.Models evaluating LLVA were adjusted for age, sex, diabetes duration, hemoglobin A1c, lens status, and hypertension.Rod-AR, Cone-AR, RGC-AR, and conventional RNP were each added separately to the clinical base model.Models incorporating RNP or Rod-AR were compared, and the incremental explanatory value of adding Rod-AR to the model containing RNP was evaluated. Main Outcome Measures: Statistical associations between Rod-AR, Cone-AR, andRGC-AR and LLVA letter score and model explanatory performance. J o u r n a l P r e -p r o o f4 Results: Mean LLVA was 76.3 ± 6.9 letters.RNP and affected cell ratios generally increased with greater DR severity.In separate adjusted models, greater Rod-AR was associated with worse LLVA (β, -1.72; 95% confidence interval [CI], -2.55 to -0.89; P < 0.001), as was greater Cone-AR (β, -1.51; 95% CI, -2.34 to -0.67; P < 0.001).RGC-AR was not associated with LLVA (P = 0.747).Models adding RNP or Rod-AR showed similar performance (AIC, 1341.1 vs 1340.9;marginal R², 0.377 vs 0.378).Adding Rod-AR to the RNP model did not significantly improve explanatory performance (likelihood ratio test P = 0.607).
Conclusions
Greater Rod-AR and Cone-AR were associated with worse LLVA after adjustment for clinical covariates.However, Rod-AR did not demonstrate incremental explanatory value beyond conventional RNP.These findings support an association with LLVA, but not an explanatory advantage of rod-density weighting over area-based nonperfusion.
Förderung
- National Institutes of Health
- Japan Society for the Promotion of Science
- Boehringer Ingelheim
- Boehringer Ingelheim Corp USA
- Illinois Society for the Prevention of Blindness
Zitieren
APA
Zhuang, K., Kakihara, S., Busza, A. M., Robertson, G., Zhu, I., & Fawzi, A. A. (2026). Cell Density Mapped Retinal Nonperfusion and Low-Luminance Visual Acuity in Diabetic Retinopathy. Ophthalmology Science, 6(12), 101435. https://doi.org/10.1016/j.xops.2026.101435
Vancouver
Zhuang K, Kakihara S, Busza AM, Robertson G, Zhu I, Fawzi AA. Cell Density Mapped Retinal Nonperfusion and Low-Luminance Visual Acuity in Diabetic Retinopathy. Ophthalmology Science. 2026;6(12):101435. doi:10.1016/j.xops.2026.101435
BibTeX
@article{Zhuang2026,
author = {Zhuang, Kallista and Kakihara, Shinji and Busza, Anna M. and Robertson, Gavin and Zhu, Ivy and Fawzi, Amani A.},
title = {Cell Density Mapped Retinal Nonperfusion and Low-Luminance Visual Acuity in Diabetic Retinopathy},
journal = {Ophthalmology Science},
year = {2026},
volume = {6},
number = {12},
pages = {101435},
doi = {10.1016/j.xops.2026.101435}
}
EUVIMED