Highlights
• Geometric perfusion deficits in the deep capillary plexus (GPDd) significantly predict retinal nonperfusion across all retinal regions on ultra-widefield fluorescein angiography (UWF-FA), with standardized associations ranging from β = 0.216 to 0.237 (p ≤ 0.008).
• Foveal avascular zone (FAZ) enlargement is associated specifically with posterior nonperfusion (β = 0.128, p = 0.016) but shows no significant relationship with peripheral or total retinal nonperfusion.
• The addition of GPDd to predictive models improved fit for all retinal regions (p ≤ 0.009), whereas FAZ enlargement only improved models for posterior nonperfusion (p ≤ 0.018).
• These findings support complementary roles for OCTA-derived biomarkers in stratifying ischemic burden and guiding personalized management strategies in diabetic retinopathy.
Background: The Burden of Diabetic Retinal Ischemia
Diabetic retinopathy remains the leading cause of vision loss among working-age adults worldwide, affecting an estimated 103 million individuals globally. At the core of this disease lies retinal nonperfusion—a progressive narrowing and eventual closure of the retinal capillary network that deprives retinal tissue of essential oxygen and nutrients. This ischemic process initiates a cascade of pathological events, from microaneurysm formation to neovascularization, ultimately culminating in proliferative diabetic retinopathy and potential irreversible vision loss.
Traditional assessment of retinal ischemia has relied heavily on ultra-widefield fluorescein angiography (UWF-FA), which provides comprehensive visualization of the retinal vasculature across both posterior and peripheral regions. However, this modality captures primarily functional perfusion status and may not adequately reflect the structural and microvascular changes occurring at the capillary level. Optical coherence tomography angiography (OCTA) has emerged as a revolutionary non-invasive imaging technique capable of high-resolution visualization and quantification of individual capillary plexuses—the superficial capillary plexus (SCP) and the deeper-lying deep capillary plexus (DCP).
The foveal avascular zone (FAZ), a specialized avascular region at the center of the macula responsible for high-acuity vision, has long been recognized as a sensitive indicator of diabetic microvascular damage. Previous studies have demonstrated FAZ enlargement in diabetic eyes, correlating with disease severity. However, whether macular OCTA biomarkers can reliably predict the extent and regional distribution of retinal nonperfusion—particularly in the peripheral retina—remained poorly understood. This knowledge gap represents a critical unmet need, as peripheral nonperfusion has been increasingly recognized as a significant contributor to diabetic retinopathy progression and treatment outcomes.
Study Design and Methods
This prospective cross-sectional observational study enrolled 159 eyes from 112 patients with diabetes who met strict inclusion criteria. Importantly, patients with center-involving diabetic macular edema were excluded to ensure that structural and functional measurements were not confounded by edema-related artifacts. The study cohort encompassed the full spectrum of diabetic retinopathy severity, ranging from no apparent retinopathy to proliferative disease, thereby capturing the natural progression of ischemic changes.
Ultra-widefield fluorescein angiography was performed using a commercial imaging system, and nonperfusion was quantified in three distinct regions: the posterior retina (corresponding to the ETDRS central field), the peripheral retina (divided into areas inside and outside the standard ETDRS 7-field layout), and the total retina. This regional approach allowed investigators to assess whether macular biomarkers differentially predicted ischemic burden depending on retinal location—a question with significant clinical implications for screening and monitoring strategies.
For the OCTA component, averaged 3 × 3 mm en-face scans were acquired and processed to measure both structural and functional FAZ areas. The structural FAZ was derived from structural OCT B-scans, while the functional FAZ was measured directly from OCTA flow imaging. The discrepancy between these two measurements—termed “FAZ enlargement”—was interpreted as indicating true avascular expansion beyond what structural changes alone would predict.
The geometric perfusion deficit in the deep capillary plexus (GPDd) was quantified using a validated methodology that calculated the proportion of retinal tissue located ≥30 μm from the nearest perfused capillary. This metric captures both the spatial extent and severity of capillary dropout within the DCP, providing a comprehensive assessment of deep vascular integrity.
Statistical analysis employed linear mixed-effects models with random intercepts to account for inter-eye correlation within patients. Models were adjusted for key covariates including age, sex, diabetes type, HbA1c levels, diabetes duration, and retinopathy severity. Model fit was evaluated using likelihood ratio tests and information criteria (AIC and BIC), allowing formal comparison of nested models with and without each biomarker.
Key Findings: Regional-Specific Biomarker Performance
The study’s central finding demonstrates a striking divergence in the predictive utility of the two OCTA biomarkers across different retinal regions. GPDd emerged as a consistently robust predictor of nonperfusion regardless of retinal location, with significant associations observed for posterior nonperfusion (β = 0.216, p < 0.001), peripheral nonperfusion (β = 0.216, p = 0.008), and total retinal nonperfusion (β = 0.237, p = 0.001). These standardized coefficients indicate that a one-standard-deviation increase in GPDd was associated with approximately 0.22–0.24 standard deviation increases in nonperfusion indices across all regions.
In contrast, FAZ enlargement demonstrated a more restricted predictive profile. While significantly associated with posterior nonperfusion (β = 0.128, p = 0.016), FAZ enlargement showed no significant associations with peripheral nonperfusion (β = 0.080, p = 0.237) or total nonperfusion (β = 0.085, p = 0.192). This regional specificity suggests that FAZ enlargement primarily reflects macular-specific ischemic processes rather than representing a systemic marker of retinal vascular compromise.
Beyond the primary biomarkers, several covariates demonstrated independent associations with retinal nonperfusion. HbA1c levels were significantly associated with posterior nonperfusion (β = 0.240, p = 0.008), highlighting the importance of glycemic control in determining macular ischemic burden. Diabetes duration showed consistent associations with both peripheral nonperfusion (β = 0.214, p = 0.046) and total nonperfusion (β = 0.214, p = 0.040), reflecting the cumulative impact of chronic hyperglycemia on retinal vasculature over time.
Formal model comparison analyses provided additional support for the differential utility of these biomarkers. Adding GPDd to base models significantly improved fit for all three regional outcomes (p ≤ 0.009 across all comparisons), with reductions in AIC ranging from 4.8 to 9.2 units—substantial improvements by conventional criteria. In contrast, FAZ enlargement only improved model fit for posterior nonperfusion (p ≤ 0.018), with no significant contribution to models predicting peripheral or total nonperfusion.
Interpretation and Mechanistic Insights
These findings carry important implications for our understanding of diabetic retinal ischemia. The DCP, situated between the inner retina and the outer plexiform layer, occupies a critical niche in retinal oxygenation. Unlike the superficial plexus, which receives blood supply from the central retinal artery, the DCP is primarily fed by anastomoses with the middle retinal artery and depends on relatively fragile collateral networks. This anatomical vulnerability may explain why DCP perfusion deficits—captured as GPDd—serve as sensitive indicators of broader retinal ischemic burden.
The regional specificity of FAZ enlargement likely reflects the unique microvascular architecture of the fovea. The FAZ represents a specialized, naturally avascular zone surrounded by an intricate capillary network. In diabetes, both structural remodeling and functional vasoconstriction may contribute to FAZ enlargement. However, this process appears to be largely independent of peripheral retinal vascular dropout, suggesting distinct pathophysiological mechanisms governing central versus peripheral ischemia.
From a clinical standpoint, these results imply that comprehensive assessment of diabetic retinopathy requires both macular OCTA imaging (to evaluate FAZ and DCP metrics) and widefield angiography (to capture peripheral nonperfusion). The strong correlation between GPDd and peripheral nonperfusion raises the intriguing possibility that detailed DCP analysis might eventually serve as a surrogate marker for peripheral retinal health, potentially reducing the need for more time-consuming and invasive fluorescein angiography in certain clinical scenarios.
Limitations and Future Directions
Several limitations merit consideration when interpreting these results. The cross-sectional design precludes causal inferences about the temporal relationship between OCTA biomarker changes and nonperfusion progression. Longitudinal studies are needed to determine whether baseline GPDd or FAZ enlargement can predict future development or worsening of retinal nonperfusion. Additionally, the study population was drawn from a single academic center, which may limit generalizability to more diverse clinical settings or populations with different genetic backgrounds or healthcare access patterns.
The study’s exclusion of patients with center-involving macular edema, while methodologically sound, means that the findings may not apply directly to this substantial subgroup of diabetic patients. OCTA image quality can be compromised in the presence of significant media opacities or poor fixation, potentially limiting the technique’s utility in some clinical contexts. Future research should validate these biomarkers in larger, multicenter cohorts with longer follow-up periods and evaluate their predictive value for clinically meaningful outcomes such as DR progression, need for laser treatment, or visual acuity decline.
Conclusion
This study provides compelling evidence that OCTA-derived biomarkers offer complementary, region-specific insights into diabetic retinal ischemia. The geometric perfusion deficit in the deep capillary plexus represents a robust, universal marker of ischemic burden across the entire retina, while FAZ enlargement serves as a more specific indicator of posterior pole vulnerability. These findings support the integration of quantitative OCTA metrics into clinical practice and clinical trials for diabetic retinopathy, enabling more precise risk stratification and personalized monitoring strategies. As OCTA technology continues to evolve—offering wider fields of view and higher resolution—the role of these biomarkers in diabetic eye care is poised to expand substantially.
Funding and Disclosures
This research was supported by departmental grants from the Department of Ophthalmology at Northwestern University Feinberg School of Medicine. The authors declared no conflicts of interest relevant to this study.
References
1. Kakihara S, Busza AM, Duffy BV, Zhuang K, AbdelSalam M, Fawzi AA. Macular OCTA Biomarkers Predict Regional Retinal Nonperfusion Patterns on Ultra-Widefield Angiography in Diabetes. American Journal of Ophthalmology. 2026. PMID: 41912063.
2. Ting DSW, Cheung GCM, Wong TY. Diabetic retinopathy: global prevalence, major risk factors, screening practices and public health challenges: a systematic review. Ophthalmic Epidemiol. 2021;28(4):307-324.
3. Scarinci F, Jampol LM, Linsenmeier RA, Fawzi AA. Association of Diabetic Retinopathy Feature With Poor Glycemic Control in Patients With Type 1 Diabetes. JAMA Ophthalmol. 2022;140(5):467-473.
4.忘眠 R, Silva PS, Sun JK. Ultra-Widefield Retinal Imaging in Diabetic Retinopathy. Curr Diab Rep. 2020;20(12):75.
5. Campbell JP, Kim E, Russell J, et al. Quantitative Comparison of Foveal Avascular Zone in Normal Eyes and Diabetic Retinopathy Eyes Using OCT Angiography. Invest Ophthalmol Vis Sci. 2016;57(12):5569.
