Structure-Function Associations Between Quantitative Contrast Sensitivity Function And Peripapillary Optical Coherence Tomography Angiography in Diabetic Retinopathy

Abstract

Purpose: To assess changes in radial peripapillary capillary (RPC) microvasculature and their impact on visual function, measured by visual acuity (VA) and contrast sensitivity, in diabetic retinopathy (DR).

Methods: This was a cross-sectional study in 96 eyes of 67 patients, including controls, diabetes without DR (DMnoDR), nonproliferative DR (NPDR), and proliferative DR (PDR) groups. Participants underwent same-day quantitative contrast sensitivity function (qCSF) and 6 × 6 mm OCT angiography (OCTA) centered on the optic disc. The Peripapillary Nerve Fiber Layer Microvasculature Density algorithm (ARI Network) was used to calculate capillary perfusion density (total area of perfused microvasculature per unit area), capillary flux index (CFI, total weighted area of perfused microvasculature per unit area), and retinal nerve fiber layer (RNFL) thickness surrounding the optic disc. Mixed-effects multivariable regression models, controlling for age, hypertension, and lens status, evaluated associations between RPC OCTA metrics, DR severity, VA, and qCSF.

Results: Significant RPC microvascular changes were observed across DR stages. Capillary perfusion density decreased with DR severity and even before retinopathy onset in DMnoDR versus controls (βavg = −0.42, P = 0.021). PDR compared to NPDR showed a significant decrease in CFI (β = −1.02 to −0.92, P < 0.01) and in RNFL (βavg = −0.71, P = 0.033). CFI had significant associations with qCSF at various spatial frequencies (β = 0.20 to 0.34, P = 0.002 to 0.042), but not with VA.

Conclusions: Radial peripapillary capillary perfusion density worsens with onset of diabetes and increasing severity of DR while capillary flux index is more significantly affected later in disease. Structure-function associations suggest that DR-induced peripapillary microvascular changes are more strongly associated with contrast sensitivity changes than with visual acuity.

Diabetic retinopathy (DR) is a leading cause of preventable blindness among working-aged adults worldwide, with approximately one third of people with diabetes having concurrent DR.¹ The pathophysiology of DR involves both retinal microvascular and neural alterations, with many studies reporting on these alterations within the macula.^2–4 However, emerging evidence suggests that the peripapillary area may also be affected early in the course of DR, with microvascular alterations potentially serving as preclinical indicators of diabetic microvascular disease.⁵ 

The radial peripapillary capillary (RPC) plexus, a network of long, straight vessels with minimal anastomoses, supplies the retinal nerve fiber layer (RNFL) and plays a crucial role in maintaining its structural integrity and function.^6,7 Optical coherence tomography angiography (OCTA) provides a noninvasive method of measuring RPC microvasculature through two key metrics: peripapillary capillary perfusion density and capillary flux index (CFI). Capillary perfusion density is defined as the area of perfused vasculature per unit area and is analogous to the commonly used vessel density (VD), offering a structural measure of perfusion. In contrast, CFI is defined as the area of perfused vasculature per unit area weighted by normalized pixel flow intensity, providing a functional assessment of perfusion and flow dynamics.⁸ 

Previous studies suggest that damage to the RPC microvasculature is associated with increasing DR severity and correlates with changes in thickness of the RNFL.^5,8–10 However, findings on the relationship between RNFL thickness and DR severity remain inconsistent, with some studies identifying significant associations^5,8 and others reporting no clear relationship.^9,10 These discrepancies highlight the need for further investigation into the structural and functional changes in DR. 

The resulting impact of microvascular and RNFL alterations on visual function remains largely unexplored in DR. Traditionally, visual acuity (VA) has been the primary measure of visual function in clinical trials.¹¹ However, growing evidence supports the use of alternative tests for visual function, such as contrast sensitivity (CS), given it is affected earlier in the course of neurodegenerative disorders and may be more reflective of vision-related quality of life and subtle changes in visual function than VA.^12–15 Although limitations in testing modalities have previously prevented the integration of CS into routine clinical practice, the advent of the quantitative contrast sensitivity function (qCSF) computational approach has facilitated the assessment of CS in various retinal conditions.^16–18 Our group has previously demonstrated that DR-induced microvascular changes in the superficial capillary plexus and deep capillary plexus measured by expanded field swept-source OCTA centered on fovea are more strongly associated with changes in CS than VA. Additionally, CS is affected earlier than VA in the course of DR and performs better in discriminating between controls, diabetes mellitus without DR, and across DR stages.¹⁹ 

No studies to date have explored how changes in RPC vasculature correlate with various visual function metrics, including CS and VA, in DR. Clarifying the relationship between RPC alterations and qCSF may help elucidate how visual function in DR is affected in regions beyond the macula. Understanding these associations is crucial for the early diagnosis and effective monitoring of DR progression. Therefore the current study aims to investigate the association between alterations in RPC microvasculature and RNFL thickness in the peripapillary area with functional changes in visual acuity and qCSF in DR. 

Methods

Study Design

This is a cross-sectional observational, single-center study including patients seen in the outpatient retina clinics at Massachusetts Eye and Ear between June 2019 and January 2024. The study was approved by the Institutional Review Board of Massachusetts General Brigham (2019P001863), and informed consent was obtained from all subjects. All procedures were conducted in accordance with the tenets set forth in the Declaration of Helsinki and Accountability Act regulations. 

Participants

Participants were included based on the following criteria: (1) age of 18 years or older; (2) diagnosis of Type 1 or Type 2 diabetes mellitus (DM) using the American Diabetes Association criteria²⁰; (3) current diagnosis of DM without DR (DMnoDR) or DR without diabetic macular edema (DME) by a senior retina specialist (LAK, DH, DGV, NAP, or JBM) on the day of study visit; (4) visual acuity of at least 20/200 as measured by a standard Snellen chart; (5) same-day quantitative contrast sensitivity function (qCSF) testing and imaging with 6 × 6 mm OCTA centered on the optic disc. Exclusion criteria were as follows: (1) other ocular comorbidities (e.g., glaucoma or age-related macular degeneration); (2) significant media opacities including corneal scars and lens opacity greater than grade > 3+ following the Lens Opacities Classification System (LOCS) III²¹; (3) significant artifacts or signal strength less than 7 on OCTA imaging; (4) neovascularization of the disc, due to potential confounding of OCTA metrics. A control group of patients with same-day qCSF and peripapillary OCTA was derived from our previously published normative database, which includes healthy eyes with no ocular pathology other than non-visually significant cataracts.²² 

All subjects underwent a comprehensive ophthalmologic evaluation, including slit lamp examination, measurement of VA using a standard Snellen chart, qCSF testing (Manifold Platform; Adaptive Sensory Technology, San Diego, CA), measurement of intraocular pressure, and dilated fundus examination and photography (Optos California; Optos plc., Dunfermline, UK), and 6 × 6 mm OCTA scans of the optic disc (PLEX Elite 9000; Carl Zeiss Meditec, Dublin, CA, USA). Lens status grading followed the LOCS III and was simplified for the purposes of the multivariable regression analysis as follows: clear lens was graded as “clear”; NO1NC1 graded as 1+ NS (nuclear sclerosis); NO2NC1, NO1NC2 or NO2NC2 was graded as 2+ NS; and NO3NC1-3 and NO1-2NC3 graded as 3+ NS. Staging of DR into nonproliferative diabetic retinopathy (NPDR) or proliferative diabetic retinopathy (PDR) was established by a retina specialist based on the Early Treatment Diabetic Retinopathy Study grading system.²³ 

Contrast Sensitivity Testing

Contrast sensitivity was evaluated using the qCSF method on the AST platform (Adaptive Sensory Technology, San Diego, CA, USA) as previously described by our group.²⁴ In brief, the qCSF method evaluates the ability to distinguish different contrast ratios across a range of spatial frequencies. Patients are initially presented with three letters at a single spatial frequency and decreasing levels of contrast. A technician records the patient's responses (correct, incorrect, or no answer) for each letter. Using a Bayesian active learning algorithm, the next combination of contrast and frequency is chosen to maximize information gain and reduce testing time, with a total of 25 triplets of optotypes tested. The time required for testing is 2-5 minutes per eye. Various qCSF values derived include: (1) area under the logarithm of contrast sensitivity function (AULCSF) represents CS estimates integrated across spatial frequencies ranging from 1.5 to 18 cycles per degree (cpd); (2) contrast acuity (CA) corresponds with the intersection of CSF curve with the x-axis and is the smallest optotype seen at full (100%) contrast; (3) contrast sensitivity thresholds at 1, 1.5, 3, 6, 12, and 18 cpd provide the lowest contrast visible at each spatial frequency in logCS units. The qCSF method has demonstrated good test-retest reliability as compared to other contrast sensitivity tests that evaluate thresholds at various spatial frequencies.^16,25 

OCTA Imaging and Analysis

Trained research technicians acquired 6 × 6 mm OCTA images centered on the optic disc using a 100-kHz SS-OCTA instrument (Plex Elite 9000, Carl Zeiss Meditec Inc., Dublin, CA). Images were manually assessed for quality, and those with significant artifacts or signal strength less than 7 were excluded. The Peripapillary Nerve Fiber Layer Microvasculature Density algorithm v0.9, a prototype research algorithm available on the ARI Network (Zeiss Portal v5.4–1206) was utilized to analyze imaging. This algorithm performs segmentation of the inner limiting membrane and the RNFL to create an en face image of the RPC vasculature (Fig. 1). Because this algorithm is a prototype, the automated segmentation and all output images and values were reviewed manually for accuracy of the analysis, and manual segmentation was performed in cases with segmentation errors. The algorithm outputs include capillary perfusion density, capillary flux index (CFI), and retina nerve fiber layer (RNFL) thickness in four quadrants (temporal, superior, nasal, inferior) of an annulus with an outer diameter of 6 mm and inner diameter of 2 mm centered on the optic disc, as well as the average of the four quadrants. Capillary perfusion density represents the total area of perfused microvasculature per unit area in a region of measurement, whereas CFI represents the total weighted area of perfused microvasculature per unit area in a region of measurement, with the weight being the normalized flow intensity corresponding to each pixel. RNFL thickness is provided in micrometers for the same analyzed regions of interest. 

Figure 1.

 

Representative images of RPC vasculature enface, structural enface, RPC perfusion heat map, and RNFL thickness map by DR stage.

Statistical Analysis

Statistical analysis was performed using RStudio Version 2023.12.1+402 and Stata/BE 18.0. Demographics and clinical characteristics were described using traditional descriptive methods. Non-normally distributed data were reported as medians with interquartile ranges. Snellen visual acuity was converted to the logarithm of the minimum angle of resolution (logMAR) equivalent. Initial variables for multivariate modeling were selected based on their potential to confound outcomes related to DR microvascular progression and visual function.^26–28 These included age, gender, lens status, duration of diabetes, coexisting hypertension, and type of diabetes. Variables were initially chosen based on their unadjusted statistical significance in univariate analyses and further refined using backward stepwise elimination to develop the final multivariable regression models. Participants were divided into groups including controls, DMnoDR, NPDR (including mild, moderate, and severe NPDR), and PDR. Mixed-effects multivariable regression models controlling for age and hypertension were used to compare RPC microvascular metrics between groups. Mixed-effects multivariable regression models controlling for age and lens status were used to evaluate associations between OCTA metrics and both VA and qCSF metrics. In all models, a random effect for each participant was included to account for the potential correlation between two eyes from the same individual. Associations were reported as standardized beta coefficients by refitting standardized data. Statistical significance was defined as a P value <0.05. 

Results

Demographics and Clinical Characteristics

The study included 96 eyes from 67 participants, comprising 23 control eyes, 27 DMnoDR eyes, 27 NPDR eyes, and 19 PDR eyes. The demographics and clinical characteristics are summarized in Table 1. The median age was 59 (47–67) years, and 40.3% of participants were female. Among the eyes with DM, 61.6% had type 2 diabetes. 

Table 1.

 

Clinical and Demographic Features of the Studied Cohort

 

Visual Acuity, Contrast Sensitivity, and OCTA Metrics

VA and qCSF outcomes are summarized in Table 1. The median LogMAR VA was 0.04 (0.00–0.16). The median AULCSF was 1.10 (0.91–1.32). Figure 1 illustrates the output of the Peripapillary Nerve Fiber Layer Microvasculature Density algorithm v0.9 across different stages of DR. The RPC OCTA microvascular metrics and RNFL thickness results are summarized across different stages of DR in Table 2. Median CFI averaged across the four quadrants was 0.40 (0.39–0.41) in controls, 0.41 (0.39–0.42) in DMnoDR, 0.40 (0.37–0.41) in NPDR, and 0.38 (0.36–0.39) in the PDR group (Fig. 2). Median capillary perfusion density averaged across the four quadrants was 0.57 (0.56–0.58) in controls, 0.57 (0.56–0.57) in DMnoDR, 0.56 (0.55–0.57) in NPDR, and 0.55 (0.54–0.56) in the PDR group (Fig. 2). Median RNFL thickness averaged across the four quadrants was 85.25 (79.83–93.19) in controls, 87.19 (82.37–91.25) in DMnoDR, 87.20 (78.88–93.13) in NPDR, and 74.32 (71.39–81.50) in PDR. 

Table 2.

 

Radial Peripapillary Capillary Optical Coherence Tomography Angiography Metrics Among All Eyes

 

Figure 2.

 

Peripapillary microvascular metrics across stages of DR. (A) Capillary perfusion density, mean and by quadrant, across DR stages. (B) CFI, mean and by quadrant, across DR stages. *P < 0.05.

Peripapillary Microvasculature Between Group Comparisons

Mixed-effects multivariable regression models controlling for age and hypertension were used to compare RPC microvascular metrics across various stages of DR (Table 3). Eyes with PDR showed a significant decrease in CFI in all quadrants compared to NPDR (β = −1.02 to −0.92, P = 0.003 to <0.001) (Table 3). There were no significant differences in CFI between DMnoDR, NPDR, and control eyes. Eyes with increasing severity of DR were associated with decreased capillary perfusion density in multiple quadrants. Specifically, eyes with DMnoDR showed decreased average, temporal, and inferior capillary perfusion density compared to controls (βavg = −0.42, P = 0.021; βtemp = −0.65, P = 0.001; βinf =-0.63, P = 0.007) (Table 3). Differences in capillary perfusion density were also found between control eyes and NPDR, DMnoDR and NPDR, and NPDR and PDR eyes. The largest effect size was detected for the average capillary perfusion density between eyes with NPDR and PDR (β = −0.89, P = 0.009) (Table 3). Eyes with DMnoDR were associated with an increased RNFL in the superior quadrant compared to control eyes (β = 0.61, P = 0.015). Eyes with NPDR did not show a significant difference in RNFL compared to DMnoDR. However, eyes with PDR compared to NPDR showed a decrease in average, nasal, and inferior RNFL (βavg = −0.71, P = 0.033; βnas = −0.58, P = 0.047; βinf = −0.87, P = 0.006) (Table 3). 

Table 3.

 

Mixed Effects Multivariable Regression Analysis, Controlling for Age and Hypertension, Comparing Radial Peripapillary Capillary OCTA Metrics Across Different Stages of Diabetic Retinopathy

 

Structure-Function Analysis

Mixed-effects multivariable regression analyses controlling for age and lens status were used to evaluate the associations between OCTA metrics and visual function (Table 4). CFI in only the temporal quadrant showed a significant association with decreased logMAR VA (β = −0.22, P = 0.049), suggesting that higher CFI in this quadrant is correlated with better visual acuity. Conversely, CFI in all quadrants showed significant associations with multiple CS metrics, including CA and CS at thresholds of 6–18 cpd. Additionally, CFI on average and in all quadrants with the exception of the nasal quadrant was associated with AULCSF. For temporal CFI, standardized β coefficients for AULCSF, CA, and CS at 6, 12, and 18 cpd were all greater than or equal to effect size for VA (0.22, 0.25, 0.22, 0.25, 0.33 vs. −0.22, respectively) (Table 4). 

Table 4.

 

Mixed Effects Multivariable Regression Analysis Controlling for Age and Lens Status Evaluating Associations Between Radial Peripapillary Capillary OCTA Metrics and Visual Acuity (VA) and Contrast Sensitivity (CS) Outcome Measures

 

Capillary perfusion density on average or in each quadrant individually was not significantly associated with VA and CS metrics (all p-values > 0.05) (Table 4). Additionally, average and nasal RNFL thickness were not associated with VA and CS metrics. Temporal RNFL thickness was significantly associated with decreased CS at 6 cpd (β = −0.22, P = 0.022). Similar negative associations were observed between superior RNFL thickness and CS at 1 cpd (β = −0.21, P = 0.028) and inferior RNFL thickness and CS at 3 cpd (β = −0.24, P = 0.001) (Table 4). 

Discussion

Here, we present the first study investigating the structure-function associations between peripapillary microvascular changes on OCTA and contrast sensitivity in diabetic patients without DR and across various stages of DR. In our cohort, which underwent same-day qCSF-measured CS and OCTA centered on the optic disc, we found significant reductions in peripapillary microvascular metrics, as measured by CFI and capillary perfusion density, with increasing DR severity. Notably, peripapillary microvascular alterations due to DR correlated more with CS than with VA, underscoring the importance of peripapillary microvascular health in regulating visual function, particularly CS, in patients with DR. 

The RPC plexus in the peripapillary region is crucial for maintaining the health of the RNFL by supporting the neural metabolic demands around the optic nerve head. Histologic studies reveal that the RPC plexus consist of long, straight vessels with few anastomoses.⁷ Non-myelinated axons of retinal ganglion cells in the RNFL have high energy demands, making them highly vulnerable to ischemic insults.²⁹ Consequently, damage to the RPC plexus can lead to significant visual impairment. Early changes in peripapillary microvasculature may precede clinically detectable DR, highlighting the importance of advanced imaging and functional biomarkers.⁵ 

Mixed-effects multivariable regression analysis demonstrated progressive reductions in RPC capillary perfusion density after onset of DM and across DR stages. Capillary perfusion density in the temporal quadrant and average of all quadrants was the most affected. Meanwhile, eyes with PDR versus NPDR showed significant reduction in CFI in all quadrants and on average, suggesting that advanced DR stages are associated with substantial RPC functional microvascular compromise. These findings align with other studies that have also demonstrated early changes occurring in the peripapillary microvasculature of patients with DM without DR compared to controls and progressive reduction with increasing DR severity.^5,30–32 Prior data from Mitamura et al.⁸ described changes in RPC CFI occurring at the onset of diabetes and changes in RPC capillary perfusion density occurring at the onset of DR. Although these findings appear to differ from our study, where capillary perfusion density is affected before CFI, several factors may account for this discrepancy. Differences in the OCTA devices (PLEX Elite 9000 in our study vs. Cirrus HD-OCT 5000 with AngioPlex OCT Angiography in Mitamura et al.⁸) and differences in the size of the examined retinal area (6 × 6 mm in our study versus 4.5 × 4.5 mm in Mitamura et al.⁸) may have contributed to the contrasting findings. Aditionally, controlling for concurrent hypertension in our multivariable regression models, which was not done in the study by Mitamura et al.,⁸ may also help explain the differences between the studies.^33,34 Capillary perfusion density, which measures total area of perfused vasculature per unit area, serves as an indicator of microvascular structure. In contrast, CFI, which measures capillary perfusion density weighted by the brightness of flow signal, provides a more qualitative assessment of microvascular function and perfusion effectiveness.^8,35 The initial changes in capillary perfusion density may be reflective of structural damage to vascular cells, including endothelial cells and pericytes. This ultimately leads to reduced perfusion effectiveness in the microvasculature as measured by CFI reduction. 

Few changes in peripapillary RNFL were present across stages of DR. Our findings show significant thinning of the RNFL in eyes with PDR compared to NPDR, particularly in the nasal and inferior quadrants, as well as the average of all quadrants. Potential mechanisms of RNFL thinning in eyes with PDR may include primary neural degeneration, secondary effects after panretinal photocoagulation,³⁶ or thinning secondary to RPC microvascular changes, as indicated by decreases in capillary perfusion density and CFI in the corresponding quadrants. Changes in RPC hemodynamics may result in decreased oxygen and nutrient delivery to retinal ganglion cells, leading to axonal loss and RNFL thinning.³⁷ Interestingly, the superior quadrant RNFL was thicker in DMnoDR eyes compared to controls, contrary to previous studies that have demonstrated thinning of the RNFL in patients with early stages of diabetes without retinopathy.^38,39 The current study did not account for other variables such as ethnicity, which has been associated with changes in RNFL thickness and may explain differences in our findings.⁴⁰ Moreover, studies on RNFL thickness in the progression of DR have demonstrated mixed results, with findings ranging from thinning, to no change, to increased RNFL thickness with increased severity of DR.^5,9,31,41 Our findings suggest dynamic changes in RNFL thickness in the context of both DM and DR. Although not designed as a longitudinal study, these results indicate a more complex progression of RNFL changes in DR, with an initial increase in the early phases followed by subsequent thinning. 

Our study also assessed the relationships between DM-induced microvascular changes using peripapillary OCTA and functional changes in CS and VA. After controlling for age and lens status, mixed-effects regression models revealed a stronger effect of peripapillary OCTA metrics on CS, as evidenced by the larger standardized beta coefficients observed for CFI to predict various qCSF metrics compared to VA. Although CFI was associated with VA only in the temporal quadrant, CFI across all quadrants demonstrated robust associations with most qCSF metrics. Significant metrics included AULCSF, CA, and CS values at 6–18 cpd, with greater standardized beta coefficients observed for CS at higher spatial frequencies (6–18 cpd). Interestingly, capillary perfusion density was not associated with any measurements of visual function. This finding may be due to the fact that capillary perfusion density reflects the overall area of perfused microvasculature but does not capture specific functional impairments as sensitively as other metrics. We also found that RNFL thickness was associated with fewer qCSF metrics and showed smaller effect sizes. Specifically, RNFL thickness in the superior, inferior, and temporal quadrants were inversely associated with CS at thresholds of 1, 3, and 6 cpd, respectively. The clinical implications of these findings are significant. Early detection of microvascular changes is crucial for managing DR and improving outcomes. The sensitivity of capillary perfusion density and CFI to microvascular and functional changes across stages of DR highlight their potential as biomarkers for DR. These OCTA-derived metrics could be combined with functional testing such as qCSF to identify subtle abnormalities at earlier disease stages. This could guide earlier intervention strategies, such as tighter metabolic control or closer monitoring, to delay DR progression and visual decline. In addition, these findings further support the incoporation of qCSF as an endpoint in clinical trials, as it provides a more significant correlation with microvascular markers than visual acuity. 

The clinical implications of these findings are significant. Early detection of microvascular changes remains a cornerstone in managing DR, and our results suggest that integrating peripapillary OCTA metrics, such as CFI and capillary perfusion density, with functional testing like qCSF may enhance the ability to identify early structural and functional abnormalities in DR. 

To our knowledge, no other studies have evaluated associations between peripapillary OCTA and CS in DR. However, in glaucoma, RPC density has been associated with multiple qCSF metrics, including AULCSF, CA, and CS at 1–18 cpd.⁴² The lack of association of CS with capillary perfusion density, despite robust associations with CFI, could be attributed to the fact that although capillary perfusion density is an indicator of structural microvascular change, it is not as sensitive as CFI in measuring functional changes in perfusion effectiveness. Reduced capillary perfusion density may indicate early structural damage to the microvasculature, whereas reductions in CFI, which accounts for the brightness of the flow signal, reflect more advanced and functionally significant impairment in blood flow and perfusion.⁸ Although our findings of some associations between RNFL thinning and qCSF metrics are intriguing, some studies on other types of neurodegeneration have shown that AULCSF remains unchanged, despite progressive loss of RNFL, until neuronal loss surpasses a certain threshold.⁴³ If such a threshold effect exists in the context of DR, our regression analyses may not capture the full impact. Additionally, increased RNFL thickness may reflect inflammation, glial cell activation, axonal swelling, or other secondary neuronal changes in response to underlying pathology that already compromises visual function.^44,45 The findings of our study differ from those of Dong et al.,⁴⁶ who reported associations between CS and greater RNFL and GCC thickness but found no correlations with angiographic features in a population of older adults. Several factors may explain these discrepancies, including differences in the methods used to measure CS (MARS chart in Dong et al.⁴⁶ vs. qCSF in our study) and variations in patient demographics (the study by Dong et al.⁴⁶ had a patient population with 46% of the participants identifying as Black). Additionally, the study described by Dong et al.⁴⁶ only assessed angiographic features surrounding the macula. Whereas Dong et al.⁴⁶ focused on older adults without significant microvascular compromise, our cohort included diabetic patients across various DR stages, where there may be disruptions in the normal structural and functional neurovascular coupling within the retina. Further understanding of the influence of RNFL on CS in DR may require studies examining these outcomes and their correlation with microvascular metrics, given that disruptions in the neurovascular coupling, or the interactions between the neurosensory retina and its blood vessels, are known to play a key role in the development of DR.^37,41,47 

Limitations of the current study include its relatively small sample size and single-center setting, which may have masked subtle changes between different stages of DR. The cross-sectional design precludes causal inferences; therefore longitudinal studies are needed to elucidate the temporal relationships between microvascular changes, RNFL thinning, and functional decline. Additionally, although the study controlled for age and hypertension, other potential confounders, such as macular status and cerebrovascular disease, which may affect peripapillary microvascular status, were not considered because of limitations of our study database.^48–50 Further studies incorporating other imaging and molecular biology evidence may also provide a more comprehensive understanding of the mechanisms underlying microvascular structural and functional changes. 

The absence of axial length measurements is a critical limitation of this study, because individual differences in axial length can alter the scaling of OCTA images, resulting in discrepancies in the actual physical dimensions of the scanned retinal area. This scaling variability may introduce systematic inaccuracies in the quantitative OCTA measurements, because the nominal scan size may not accurately represent the actual retinal area being analyzed.^51–53 Future studies should incorporate direct axial length measurements or surrogate metrics, such as spherical equivalent or corneal curvature, to correct OCTA and RNFL data, ensuring the accuracy and comparability of these measurements.⁵⁴ Alternatively, axial length could be estimated from macular optical coherence tomography images using a deep learning-based approach,⁵⁵ or a Monte Carlo simulation approach could be used to adjust for image scaling given the known distribution of axial length.⁵⁶ 

In summary, the novel structure-function relationships examined in this study by combining peripapillary OCTA with a comprehensive measurement of CS provide insights into the underlying mechanisms of how peripapillary retinal dysfunction affects visual function. This study reveals that, in addition to macular alterations, the microvascular structure of the peripapillary region may also significantly impact CS in DR. Additionally, qCSF shows greater sensitivity to early peripapillary structural changes caused by DR compared to VA, suggesting that incorporating qCSF testing into routine clinical practice could facilitate a more comprehensive assessment of visual function in the clinical management of DR. 

Acknowledgments

Disclosure: S.K. Ha, None; X. Ding, None; F. Romano, None; K.M. Overbey, None; F. Vingopoulos, None; I. Garg, None; C.F. Bennett, None; I. Stettler, None; I. Ploumi, None; G. Baldwin, None; M.J. Finn, None; P. Razavi, None; D.G. Vavvas, Olix Pharma (C), Valitor (C), TwentyTwenty (C), Sumitomo/Sunovion (C), Cambridge Polymer Group (C), National Eye Institute (F), Research to Prevent Blindness (F), Loefflers Family Foundation (F), Yeatts Family Foundation (F), Alcon Research Institute (F); D. Husain, Allergan (C), Genentech (C), Omeicos Therapeutics (C), National Eye Institute (F), Lions Vision Gift (F), Commonwealth Grant (F), Lions International (F), Syneos LLC (F), the Macular Society (F); N.A. Patel, Alimera Sciences (C), Alcon (C), Allergan (C), Genentech (C), Eyepoint (C), Lifesciences Guidepoint (C), GLG (C); L.A. Kim, National Eye Institute (F), CureVac AG (F), Pykus Therapeutics (F); J.B. Miller, Alcon (C), Allergan (C), Carl Zeiss (C), Sunovion (C), Topcon (C), Genentech (C)