The Impact of Hypertension on Macular Perfusion in Patients With Referable Diabetic Retinopathy: An OCTA Analysis

Abstract

Purpose: To assess the impact of hypertension (HTN) on microvasculature and vision in clinically referable diabetic retinopathy (DR) eyes using optical coherence tomography angiography.

Methods: We conducted a cross-sectional study of 185 eyes (139 patients) with referable DR (moderate/severe nonproliferative to naïve/quiescent proliferative DR), categorizing patients based on presence/absence of HTN. Optical coherence tomography angiography 3 × 3 mm scans were utilized to quantify geometric perfusion deficits (GPD), vessel length density, and vessel density within the superficial capillary plexus and deep capillary plexus (DCP). Linear regression analysis investigated the association between risk factors and GPD.

Results: Our DR cohort comprised 52 nonhypertensive and 133 hypertensive eyes. After adjusting for age and dyslipidemia, we found significant differences in DCP metrics (GPD and vessel length density) between groups. The mean DCP GPD was higher in DR eyes with HTN compared with those without (7.06 ± 4.33 vs. 5.58 ± 2.85%; P = 0.032). Conversely, DCP vessel length density was lower in hypertensive DR eyes compared to nonhypertensive DR eyes (0.17 ± 0.02 mm⁻¹ vs. 0.18 ± 0.02 mm⁻¹; P = 0.031). Multivariable analysis confirmed a significant independent association between HTN (β = 0.250; P = 0.036) and DCP GPD. Worse vision was moderately associated with worsening DCP GPD (P < 0.001) in hypertensive DR eyes.

Conclusions: In referable DR eyes, HTN is associated with worse DCP nonperfusion and compromised vision. Therefore, heightened surveillance, in addition to blood pressure control, may need to be prioritized for this high-risk population with comorbid diabetes and HTN.

Translational Relevance: HTN is associated with worse vision and ischemia in the deep capillary layer of referable DR eyes, emphasizing the importance of monitoring patients with both comorbidities.

Introduction

Diabetes mellitus and hypertension (HTN) are the most common coexisting chronic vascular morbidities, both of which can independently cause impairment of retinal vessels.^1,2 Diabetic retinopathy (DR), a microvascular complication of diabetes, is one of the leading causes of visual impairment in adults.³ HTN affects 50% to 80% of patients with type 2 diabetes and 30% of patients with type 1 diabetes.⁴ The co-occurrence of HTN in diabetes has been reported to increase the risk of DR, thus posing a more significant threat to the retinal vasculature and vision than either condition on its own.^5,6 

Optical coherence tomography angiography (OCTA), with its depth-resolved capabilities, has revolutionized the noninvasive assessment of retinal vessels at the capillary level across different layers without contrast.^7,8 This rapid microvascular imaging tool has facilitated our understanding the extent of capillary level pathology caused by HTN and diabetes independently. For instance, DR progression has been linked to lower vessel density (VD) and higher geometric perfusion deficit (GPD) in the deep capillary plexus (DCP) in referable nonproliferative DR (NPDR) eyes.⁹ Many cross-sectional studies have shown a correlation between OCTA vascular parameters and DR severity.^10–13 Similarly, research on HTN indicates that nondiabetic hypertensive eyes had lower VD in superficial capillary plexus (SCP) and DCP and larger foveal avascular zone compared with healthy eyes.¹⁴ Thus, HTN and diabetes independently contribute to macular capillary non perfusion, with a potentially negative impact on vision.¹⁵ 

Although much research has focused on the individual effects of diabetes and HTN on retinal microvasculature, there is a limited understanding of their combined impact at the capillary level. Only a few studies have investigated retinal vessel damage in patients with both conditions, but they did not consider eyes with naïve or treated proliferative DR (PDR) in their analysis.^16,17 Eyes with clinically referable DR, defined as moderate or severe NPDR, naïve, and treated PDR, are closely monitored owing to their vulnerability to sight-threatening complications. Although OCTA imaging has already revealed compromised retinal microvasculature in these eyes, the additional impact of HTN on the retinal capillaries remains unclear. 

Our study aims to analyze the concomitant impact of HTN on the retinal microvasculature and vision of eyes with clinically referable DR. This could provide valuable insights to improve monitoring and management strategies for DR in patients with comorbid diabetes and HTN. 

Methods

This prospective cross-sectional study enrolled diabetic patients who visited the retina clinic in the Department of Ophthalmology at Northwestern University in Chicago, Illinois, from October 2021 to September 2023. This study protocol was approved by the Institutional Review Board of Northwestern University, conducted in compliance with the tenets of the Declaration of Helsinki, and followed the Health Insurance Portability and Accountability Act regulations. Written informed consent was obtained from each participant before any study procedure. 

Study Participants

Our study recruited participants with type 1 or type 2 diabetes who had clinically referable DR, diagnosed by a board-certified retina specialist. Eyes with clinically referable DR included moderate NPDR, severe NPDR, naïve PDR, and quiescent PDR. An ultrawide-field Scanning Laser Ophthalmoscope (Optomap Panoramic 200, Optos PLC, Dunfermline, Scotland, UK) was used to obtain color fundus photographs. The grading details followed the International Classification of Diabetic Retinopathy severity scale, as described in a previous study. Two graders independently assessed the photos and any discrepancies were resolved by a third grader.¹⁸ Moderate NPDR was defined as eyes with microaneurysms, intraretinal hemorrhages or venous beading that do not meet severity criteria. Severe NPDR was termed when eyes followed 4:2:1 rule: more than 20 intraretinal hemorrhages in each of the 4 quadrants, 2 quadrants of venous beading, or 1 quadrant of intraretinal microvascular abnormalities. Naïve PDR was characterized by the presence of neovascularization of the disc or elsewhere, or vitreous/preretinal hemorrhage with no prior laser treatment.¹⁹ Finally, quiescent PDR was defined as stable PDR after panretinal photocoagulation with no evidence of active neovascularization in the past 6 months. To ensure high-quality imaging and avoid optical distortions, only eyes with a Q-score, which reflects signal strength of 6 or greater and axial lengths between 22 and 26 mm were included. 

A history of HTN was obtained from participants and their medical records. Although it is well-known that a significant proportion of diabetic patients are also hypertensive, routine blood pressure measurements are not always taken during every eye clinic visit, particularly when HTN is already known and managed with medication. Owing to logistical constraints and the observational nature of this study, direct measurement of blood pressure was not conducted. This approach reflects common clinical practice in ophthalmology settings. As a result, diabetic participants were grouped into two categories: referable DR eyes with a history of HTN and referable DR eyes without HTN. 

The exclusion criteria for the study included eyes with clinical evidence of foveal involving diabetic macular edema (central subfield thickness >300 µm on Heidelberg Spectralis OCT), those who had received intravitreal anti-vascular endothelial growth factor or steroid treatment within the last 6 months, or any condition that could impact retinopathy status or alter visual acuity. Additionally, participants with a history of major ocular surgery in the past 3 months or planned surgery within 6 months of enrollment were excluded. Eyes with a glycated hemoglobin (HbA1c) level of greater than 10.0% and patients participating in any study evaluating investigational medicinal products were also excluded. If both eyes of a participant were eligible, both were included in the study. 

Participant's clinical and demographic data were gathered through history taking and reviewing from medical records. Best-corrected visual acuity (BCVA) was calculated for each patient using Early Treatment Diabetic Retinopathy Study eye charts for refraction and translated to the logarithm of the minimum angle of resolution. 

OCTA Imaging

The 3 × 3 mm OCTA scans centered on fovea were obtained using the RTVue-XR Avanti system (Optovue, Inc, version 2017.1.0.151, Fremont, CA, USA) with split-spectrum amplitude-decorrelation angiography.²⁰ A minimum of five scans were captured on each study eye. This scanner uses near-infrared light with approximately an 840-nm wavelength with a bandwidth of 50 nm and has an A-scan rate of 70,000 per second, achieving a tissue resolution of 5 µm axially and 15 µm transversely. The OCTA performs two repeated B scans at each location in the retina, each constituting 304 consecutive A-scans across 3 mm, with a total acquisition time of approximately 3 seconds. The split-spectrum amplitude-decorrelation angiography algorithm extracts angiographic flow data by quantifying the decorrelation signal between these two successive B-scans. Finally, the motion correction technology of the system registers and merges two orthogonal scans to create a single OCTA scan with reduced motion artifacts.^21,22 

OCTA Image Analysis

The automated AngioVue software segmented retinal microvasculature into a full retinal vascular network, SCP, and DCP. The full retina slab consists of tissue from the inner limiting membrane to 10 µm below the outer plexiform layer. The SCP slab extends from the inner limiting membrane to 10 µm above the inner plexiform layer. The DCP slab extends from 10 µm above the inner plexiform layer to 10 µm below the outer plexiform layer. These angiograms were then exported into Fiji software, an open-source distribution of the program ImageJ.²³ The SCP angiograms were aligned using the register virtual stack slices plugin with a rigid feature extraction model and an elastic registration model. The best-quality angiogram was chosen as a reference for registration. Then, the transformation method was applied to the DCP and full retina slabs with the help of the transform virtual stack slices plugin. Finally, the registered images in each stack were averaged by applying intensity projection in the Z-project to improve the image quality and minimize noise.²⁴ 

A semiautomated FIJI macro was used to measure VD, vessel length density (VLD), and GPD in the averaged scans of each vascular slab. Initially, the average angiograms were transformed into binary images using the Huang2 plugin for capillaries and the Max Entropy plugin for larger vessels. The capillaries were then skeletonized, as the OCTA's transverse resolution was below the size of the capillaries. VD was assessed from the binarized average images by calculating the proportion of white pixels relative to the total area. VLD was evaluated from the skeletonized images and defined as the length of the skeletonized vessels divided by the total area. GPD (Fig. 1) represents the percentage of retinal tissue located more than 30 µm from the nearest blood vessel, as this area is considered to have oxygen levels below the physiological threshold. This is based on lateral oxygen diffusion in animal models and the intercapillary distance observed in OCT systems.^24–26 Based on this hypothesis the calculation of GPD followed a two-step approach: first, for each nonvessel pixel, the capillary perfusion distance, that is, the minimum distance that an oxygen molecule would need to diffuse to reach that pixel, was computed; and, second, based on this distance, the pixel is classified as either adequately or inadequately perfused. The pixels with a minimum distance of more than 30 µm away from the nearest vessel skeleton were classified as inadequately perfused and thus considered as GPD. The GPD percentage was calculated as the total perfusion deficit area divided by the total reference area, multiplied by 100.²⁶ However, before GPD calculation, a large vessel mask was subsequently applied to the skeletonized SCP and full capillary angiograms. We refrained from overlaying larger vessels onto the DCP to avoid eliminating genuine GPD areas. Instead, GPD was measured using the skeletonized DCP angiogram, and areas under large vessels were subtracted from both the GPD and total area calculations to eliminate the effects of projection or shadowing artifacts. The foveal avascular zone area was manually traced on the full retinal slab using the polygon selection tool in ImageJ, and both the foveal avascular zone area and watermark were excluded from the GPD calculation.¹⁸ 

Figure 1.

 

GPD (red areas) of macular region from OCTA 3 × 3 mm scans. (A) SCP and (B) DCP of nonhypertensive quiescent PDR (C) SCP and (D) DCP of hypertensive quiescent PDR. (E) SCP and (F) DCP of nonhypertensive severe NPDR. (G) SCP and (H) DCP of hypertensive severe NPDR. Significant differences in DCP GPD were observed between hypertensive and nonhypertensive DR groups whereas SCP did not differ significantly between groups.

Statistical Analyses

Statistical analysis was performed using IBM SPSS Statistics, version 26 (IBM SPSS Statistics; IBM Corporation, Chicago, IL, USA) and R (R 4.3.1; R Project for Statistical Computing, Vienna, Austria). The Shapiro–Wilk test was used to determine the normality of the distribution of the continuous variables. To compare continuous variables between groups, an independent sample t test was used for normally distributed variables, while the Mann–Whitney U test was applied for nonparametric data. For categorical data, the Fisher exact test was used. When comparing OCTA measurements between two groups, generalized estimating equations were applied to account for the contribution of two eyes from the same patient and control for covariates such as age and dyslipidemia. 

The association between GPD and risk factors was assessed using linear regression models, with generalized estimating equations to account for intereye correlation. In the univariate analysis, we primarily focused on HTN and diabetes-related variables, including diabetes duration, type of diabetes, and HbA1c, because these were the key variables of interest in our study. In the multivariable model, we included significant variables from the univariate analysis and adjusted for additional covariates as potential confounders, such as demography (age and gender), ocular factors (axial length and Q score), and other systemic factors (dyslipidemia and smoking history), all of which are known to influence vascular metrics. This regression model with confounder adjustment allowed for an evaluation of the independent effect of HTN on GPD, minimizing the potential influence of confounding variables.²⁷ Linear regression analysis using generalized estimating equations was also used when assessing the association between GPD in the DCP and BCVA in referable DR eyes, stratified by HTN status. A P value of less than 0.05 was considered statistically significant. 

Results

Cohort Characteristics

We studied 185 clinically referable DR eyes from 139 patients, which included 133 eyes from 105 patients with HTN and 52 eyes from 34 patients without HTN. The patient demographics and ocular characteristics are shown in Table 1. The median age of patients with concurrent DR and HTN was 64 years (interquartile range [IQR], 52–70 years) and was significantly older than those without HTN (median 55 years; IQR, 44.5–62.0 years; P = 0.008). The median BCVA of hypertensive DR eyes was 86 letters (IQR, 81.5–88.0 letters) and was significantly worse compared with nonhypertensive DR eyes (median, 81 letters; IQR, 76.8–85.0 letters; P < 0.001). The proportion of eyes with dyslipidemia was higher in DR eyes with HTN than those without HTN (P = 0.001), but no significant difference was found in relation to gender, diabetes duration, HbA1c, type of diabetes, DR severity, and smoking between the two groups. 

Table 1.

 

Demographics and Clinical Characteristics of Study Population

 

Quantitative OCTA Vascular Measurements

Table 2 presents the mean measurements of SCP and DCP metrics, including GPD, VD, and VLD, adjusted for age and dyslipidemia. We did not adjust for BCVA in this analysis, as we considered it an outcome of vascular alterations. The percentage of GPD at DCP layer was worse in referable DR eyes with HTN than those without HTN (7.06 ± 4.33 and 5.58 ± 2.85; P = 0.018). Conversely, the VLD of hypertensive DR eyes was significantly lower than nonhypertensive eyes (0.17 ± 0.02 mm⁻¹ and 0.18 ± 0.02 mm⁻¹; P = 0.031). None of the OCTA metrics at the SCP layer differed significantly. 

Table 2.

 

OCTA Metrics in Each Group

 

Association of GPD With HTN and Other Risk Factors

The univariate regression analysis in Table 3 assessed the relationship between primary predictors, including HTN and diabetes-related variables, with GPD. HTN showed a significant positive correlation with GPD in the DCP layer (β = 0.251; P = 0.020), but not in the SCP layer (β = 0.121; P = 0.220). Diabetes-related variables, such as HbA1c (β = −0.082; P = 0.040), severe NPDR (β = 0.328; P = 0.006), treatment-naïve PDR (β = 0.290; P = 0.023), and quiescent PDR (β = 0.570; P < 0.001), were also significantly associated with GPD at the DCP level. We found an unexpected significant inverse relationship between HbA1c and GPD. 

Table 3.

 

Univariate Regression Analyses Evaluating the Association Between HTN and Diabetes-Related Variables to GPD Outcome in Clinically Referable DR Eyes (n = 185 Eyes)

 

Because the significant association between HTN and OCTA metrics was observed only in the DCP, we focused our multivariable analysis on GPD at the DCP level (Table 4). The model was adjusted for potential confounders, including age, gender, axial length, Q score, dyslipidemia, and smoking history, which are known to influence the retinal vasculature,^28–33 in addition to the significant univariate variables. After adjusting for these covariates, HTN remained significantly associated with GPD at the DCP level (β = 0.25; P = 0.036), confirming this relationship. 

Table 4.

 

Multivariable Regression Analysis Including HTN, Significant Univariate Diabetes-Related Covariates, and Other Potential Confounders Affecting GPD in Clinically Referable DR Eyes (n = 185 Eyes)

 

Correlation of GPD With Visual Acuity in DR Eyes With or Without HTN

Figure 2 displays the linear regression analyses for referable DR eyes with and without coexisting HTN. A statistically significant, negative correlation was observed between GPD in the DCP and BCVA in referable DR eyes with HTN (β = −1.11; P < 0.001), indicating that greater DCP perfusion deficits were associated with worse visual acuity in this group. In contrast, no significant correlation was found between DCP GPD and BCVA in referable DR eyes without HTN (β = −0.80; P = 0.064). 

Figure 2.

 

Scatterplots illustrating the association between GPD in the DCP and BCVA in referable DR eyes, stratified by HTN status. Linear regression coefficients (β) and P values are shown for each group. Linear regression analysis revealed a significant negative association between GPD at DCP and BCVA in the hypertensive group (β = −1.11; P < 0.001), indicating that higher GPD at DCP is associated with worse visual acuity.

Discussion

In this study, our findings showed that clinically referable DR eyes with concurrent HTN had significantly more impaired macular microvasculature than eyes without HTN. Specifically, we found higher ischemic deficits at the DCP in hypertensive eyes compared with nonhypertensive eyes with referable DR, but no differences in the SCP layer. HTN exerted a statistically significant effect on DCP GPD in these eyes even after adjusting for potential confounders. Although the effect size is modest, it may still hold clinical relevance, considering HTN's established role as a systemic risk factor in the progression of DR.^5,6 Moreover, visual acuity was significantly correlated with worsening GPD in DCP in the hypertensive group. These results suggest that the macula of hypertensive individuals who also have referable DR may experience greater ischemia, particularly in the DCP, along with impaired vision. Therefore, even modest associations may have meaningful clinical implications and emphasize the importance of prioritizing patients in the referable DR cohort who also have comorbid HTN for closer monitoring. 

We were intrigued that the significant capillary perfusion differences were exclusively within the deep, but not the superficial capillaries, of hypertensive referable DR eyes compared with nonhypertensive ones. This finding diverges from a previous study¹⁶ that identified alterations in both the DCP and SCP layers of hypertensive diabetic eyes. Notably, that prior study focused on eyes with NPDR and well-controlled HTN. We also note that the authors of the prior study did not include PDR eyes or adjust for systemic factors such as dyslipidemia or smoking, which commonly co-exists in the diabetic cohort. We can think of several potential reasons for the particular vulnerability at the DCP. The topographic organization of DCP distinctly differs from that of SCP. Imaging studies suggest that the venular drainage predominates the DCP, potentially rendering it more vulnerable to ischemic damage owing to lower oxygen tension and its proximity to the high oxygen demand of the photoreceptors.^34,35 OCTA studies indicate that DCP loss typically occur first in the early stage of DR, and progress more rapidly as DR advances. This may be associated with progressive dilation and telangiectasia in the SCP, which could lead to blood being diverted more towards the SCP, potentially altering perfusion in the DCP.^12,36,37 HTN disrupts the retinal autoregulatory mechanisms and chronic HTN induces endothelial cell dysfunction, which can lead to progressive retinal ischemia, particularly in the more vulnerable DCP, even when blood pressure is controlled.^38,39 Consistent with our findings and hypothesis, Chua et al.⁴⁰ demonstrated that poorly controlled hypertensive eyes experience reduced capillary density at the DCP compared with well-controlled hypertensive eyes, with no notable differences at the SCP. This finding suggests that the DCP in eyes with referable DR may be more vulnerable to the added insult of HTN. Further research is needed to elucidate the underlying mechanisms that make the DCP more vulnerable in this context. 

Although referable DR eyes exhibit compromised capillary density and nonperfusion,^9,11,18 our study demonstrates that the added burden of HTN is reflected in the GPD but not the VD. GPD is a robust OCTA biomarker that quantifies ischemic retina areas based on the principle that oxygen supply to retinal tissue depends on its proximity to capillaries. We used the theoretically estimated oxygen diffusion limit of approximately 30 microns to define the ischemic retina in GPD.^26,41 By integrating oxygen diffusion principles, GPD metric inherently provides a more meaningful link between vascular structure and tissue oxygenation. In contrast, VD simply measures the area occupied by vessels, without considering the functional consequences of that geometry. High VD can therefore be misleading in areas with poor oxygenation owing to abnormal vessel arrangements. Furthermore, VD may overestimate perfusion, particularly owing to the inclusion of large vessels that are not skeletonized in its calculation. In contrast, the GPD is determined after the skeletonization process, which avoids overestimating large vessels and potentially provides a more accurate representation of retinal perfusion.^26,41 Along the same lines, our study revealed differences in VLD but not in VD between the study groups. Similar to the GPD, VLD, calculated using a skeletonization process, minimizes the influence of larger vessels, and increases sensitivity to microvascular changes at capillary-level, where oxygen exchange predominantly occurs.^42,43 Hence, we hypothesize that GPD and VLD may be more sensitive to localized nonperfusion that may not be reflected in the VD. Furthermore, GPD is less susceptible to artifacts like vessel discontinuities caused by speckle noise or motion, which can significantly affect vessel diameter measurements.²⁶ Regression analysis revealed that, although DR severity was the primary driver of DCP perfusion deficits, HTN also significantly contributed to the exacerbation of the DCP non perfusion after adjusting for potential confounders. Therefore, GPD could be a valuable parameter for assessing HTN-induced nonperfusion in eyes with referable DR. 

Although we excluded eyes with macular edema, we were intrigued by the finding that visual acuity was worse in the hypertensive compared with nonhypertensive referable DR eyes. This observation aligns with prior research demonstrating decreased visual acuity in diabetic eyes with coexisting HTN.^16,17 In our analysis, a significant negative correlation between GPD and BCVA was observed in the HTN group, but this relationship was not evident in the non-HTN group. This discrepancy may be attributed partly to the smaller sample size in non-HTN group, potentially limiting the statistical power to detect a significant association. Larger datasets may help to clarify whether this relationship is also present in nonhypertensive individuals. Previous OCTA studies have established a moderate correlation between DCP perfusion deficits and vision loss in DR eyes.⁴⁴ Our study further suggests that the damage to the DCP is aggravated by the added burden of HTN in referable DR eyes, potentially contributing to the worse BCVA observed in our cohort with both comorbidities. It is possible that the impact of GPD on visual acuity becomes clinically apparent only after a certain threshold of DCP nonperfusion is surpassed—an effect that may be amplified in the presence of HTN. While photoreceptors, crucial for vision, primarily depend on diffusion from the choroidal circulation, the DCP partially contributes to nourishing the inner segments of photoreceptors, particularly during dark adaptation.⁴⁵ Evidence has shown that, during systemic hypoxia, the retinal vascular contribution to meeting the metabolic needs of photoreceptors becomes even more critical, as the choroidal vasculature does not autoregulate its blood supply in the hypoxic environment. Consequently, DCP nonperfusion could contribute to photoreceptor dysfunction^46,47 and explain worse vision in eyes with comorbid DM and HTN. 

Strengths and Limitations

This study has several key strengths. We focused on referable DR eyes, ranging from moderate or severe NPDR to naïve and quiescent PDR eyes, a key population that require higher vigilance and frequent follow-up. By evaluating the impact of hypertensive status, our study addressed an important knowledge gap by demonstrating that eyes with referable DR and coexisting HTN are more ischemic at the DCP layer, and experience worse vision. Another strength of our study lies in the adjustment for baseline DR severity and the consideration of confounding factors, such as dyslipidemia and smoking, that also influence retinal vascular health in the multivariable regression model.^28,29 Our OCTA image analysis used a semiautomated method, averaging multiple high-quality OCTA scans (approximately five) to ensure accurate vascular metrics. Additionally, the 3 × 3 mm OCTA scans offered enhanced lateral resolution compared with larger scans, enabling more precise assessment of macular capillaries.⁴⁸ 

Nevertheless, we acknowledge several limitations of our study. First, HTN was defined based on a history of HTN and the use of antihypertensive medications; however, ambulatory blood pressure was not measured in this study. This may have introduced misclassification bias and excluded patients with undiagnosed HTN. Additionally, we did not distinguish between controlled and uncontrolled HTN, which may influence vascular parameters. Future studies are warranted to investigate the impact of HTN control status on these outcomes. The timeline of HTN diagnosis (whether it occurred before or after the onset of diabetes) or the class of antihypertensive medications was also not considered, which may influence vascular parameters differently. Second, the restricted range of HbA1c (limited to patients with levels under 10%) may also have introduced selection bias.⁴⁹ The observed negative association between HbA1c and DCP GPD was unexpected and appears to be novel. This finding warrants further investigation to elucidate the underlying mechanisms and assess their potential clinical implications. Third, owing to the unavailability of data on other visual function tests such as contrast sensitivity, microperimetry, and color vision testing, we could not assess the impact of retinal vascular impairment on overall visual performance in DR eyes with comorbid HTN. Further research is needed to explore the functional implications of our findings. Fourth, although our analysis was hypothesis driven and adjusted for key confounders, we did not apply a formal correction for multiple comparisons. Future validation studies with larger datasets will be helpful to confirm our findings. Fifth, we excluded eyes with center-involving diabetic macular thickness larger than 300 µm at baseline to minimize the effects of edema on OCTA image segmentation. However, smaller cystoid spaces that could affect segmentation and artificially reduce flow signals may have been missed. Finally, the retrospective design introduces inherent selection bias, highlighting the need for a prospective, longitudinal study to quantify OCTA metric changes over time. 

Conclusions

Given the established link between HTN and DR progression, OCTA offers a crucial tool for understanding the underlying microvascular alterations when both conditions coexist. Our study demonstrates that referable DR eyes with HTN have significantly higher perfusion deficits in the DCP and worse vision than those without HTN. Among the OCTA metrics, GPD and VLD, rather than VD, emerged as key parameters for identifying DCP ischemia induced by HTN in these eyes. This underscores the importance of higher vigilance for patients with comorbid HTN and DR. Longitudinal studies are needed to track OCTA changes and establish strategies that optimize the management of this particularly vulnerable population. 

Acknowledgments

Supported by NIH Grant R01EY31815 (A.A.F), Japan Society for the Promotion of Science (#202460005; S.K.) and a collaborative grant agreement from Boehringer Ingelheim. 

Disclosure: J. Shah, None; S. Kakihara, None; A. Busza, None; A.A. Fawzi, Consulting fees—Regeneron, Roche/Genentech, Boehringer Ingelheim, RegenXbio, 3Helix