Refined four-layer peripapillary OCTA stratification improves anatomical localization and discriminative performance in retinal vein occlusion, though DCP isolation did not increase standardized disease contrast beyond the conventional deep vascular complex.
Key Findings
Results
Refined four-layer peripapillary OCTA stratification substantially improved discriminative performance compared to global quantification for detecting RVO-related microvascular alterations.
AUC for vessel density improved from 0.636 (global) to 0.826 (four-layer model), with P < 0.05
AUC for fractal dimension improved from 0.603 to 0.807 with refined stratification, P < 0.05
AIC was also used alongside AUC to assess discriminative performance
Study included 186 eyes: 72 RVO-affected, 72 fellow eyes, and 42 healthy control eyes
Results
DCP isolation did not increase standardized disease contrast beyond the conventional deep vascular complex (DVC) in RVO.
Standardized disease contrast for VD was -0.57 in the global inner retina, -0.83 in the conventional DVC, and -0.79 in the DCP alone
Corresponding FD contrasts were -0.07 (global), -0.39 (DVC), and -0.39 (DCP), indicating no incremental contrast from DCP isolation
These findings suggest that separating DCP from DVC does not enhance sensitivity for detecting disease-related vascular density differences
Results
FD-based models showed substantially stronger associations with best-corrected visual acuity (BCVA) than global metrics, with DCP-related metrics contributing substantially to model performance.
Adjusted R² for FD-based models improved from 0.021 (global) to 0.211 with refined four-layer stratification
DCP-resolved metrics, particularly fractal dimension, contributed substantially to BCVA model performance
Regression analysis was used to evaluate functional associations with visual acuity
These findings suggest vascular network organization metrics (FD) carry more functional information than simple vessel density metrics
Methods
The study evaluated peripapillary microvasculature across three hierarchical dimensions using five distinct microvascular metrics in treatment-naïve RVO patients.
Three stratification levels assessed: global (Dim 1), two-layer (Dim 2), and refined four-layer model (Dim 3: RPC, SVP, ICP, and DCP)
Five microvascular metrics quantified: vessel density (VD), fractal dimension (FD), perfusion area (PA), vessel length density (VLD), and vessel diameter index (VDI)
Study design was cross-sectional and included only newly diagnosed, treatment-naïve RVO patients
Fellow eyes of RVO patients were included as a separate comparison group alongside healthy controls
Results
Refined four-layer stratification improved anatomical localization of deep-layer abnormalities in RVO beyond what conventional two-layer models provide.
Conventional models combined deep layers into a single deep vascular complex (DVC), obscuring layer-specific abnormalities
Four-layer model (RPC, SVP, ICP, DCP) enabled more specific localization of layer-related abnormalities
Authors propose this framework supports more anatomically specific characterization of RVO-related microvascular abnormalities
Authors describe this as providing 'a methodological framework for future longitudinal and externally validated studies'
Discussion
The study was conducted in a cross-sectional design, limiting causal inference and longitudinal tracking of microvascular changes in RVO.
Authors explicitly note the need for future longitudinal and externally validated studies
Cross-sectional design was identified as a limitation in the context of the study's methodological framework
Sample included 186 total eyes with 72 RVO-affected eyes as the primary disease group
What This Means
This research suggests that using a more detailed, four-layer mapping of the tiny blood vessels around the optic nerve (peripapillary region) provides better information about eye disease in patients with retinal vein occlusion (RVO), a condition where a vein in the retina becomes blocked. The study used optical coherence tomography angiography (OCTA), a non-invasive imaging tool, to measure blood vessel characteristics in 186 eyes — including eyes with RVO, the unaffected fellow eyes of the same patients, and healthy control eyes. By separating the retinal blood vessel layers more precisely into four distinct layers rather than the conventional one or two layers, the researchers were able to more accurately distinguish diseased eyes from healthy ones, with accuracy scores (AUC) jumping from about 0.60–0.64 to over 0.82.
Interestingly, isolating the deepest vascular layer (deep capillary plexus, DCP) on its own did not detect more disease-related vessel loss compared to combining it with adjacent layers in the conventional way. However, a specific measure of how complex and branched the vessel network is — called fractal dimension (FD) — in that deepest layer was strongly linked to how well patients could see (visual acuity), with the statistical explanation of vision differences improving from about 2% to over 21% when the refined layering approach was used. This suggests that the organization of blood vessels, not just their density, carries important information about visual function.
This research suggests that more detailed layer-by-layer analysis of blood vessels around the optic nerve could help clinicians better understand and characterize RVO at a structural level. The findings are preliminary given the cross-sectional (single time-point) design, and the authors call for future studies that follow patients over time to confirm whether these imaging measures can track disease progression or treatment response.