Plexus-Specific Retinal Capillary Blood Flow Analysis Using Erythrocyte Mediated Angiography and Optical Coherence Tomography Angiography

Abstract

Purpose: The purpose of this study was to identify and measure plexus-specific absolute retinal capillary blood flow velocity and acceleration in vivo in both nonhuman primates (NHPs) and humans using erythrocyte mediated angiography (EMA) and optical coherence tomography angiography (OCTA).

Methods: EMA and OCTA scans centered on the fovea were obtained in 2 NHPs and 11 human subjects. Scans were also obtained in NHP eyes while IOP was experimentally elevated. Erythrocyte velocity and acceleration in retinal arteries, capillaries, and veins were measured and capillaries were categorized based on location within the superficial vascular (SVP), intermediate capillary (ICP), or deep capillary plexus (DCP). Generalized linear mixed models were used to estimate the effects of intraocular pressure (IOP) on capillary blood flow.

Results: Capillary erythrocyte velocity at baseline IOP was 0.64 ± 0.29 mm/s in NHPs (range of 0.14 to 1.85 mm/s) and 1.55 ± 0.65 mm/s in humans (range of 0.46 to 4.50 mm/s). Mean erythrocyte velocity in the SVP, ICP, and DCP in NHPs was 0.69 ± 0.29 mm/s, 0.53 ± 0.22 mm/s, and 0.63 ± 0.27 mm/s, respectively (P = 0.14 for NHP-1 and P = 0.28 for NHP-2). Mean erythrocyte velocity in the human subjects did not differ significantly among SVP, ICP, and DCP (1.46 ± 0.59 mm/s, 1.58 ± 0.55 mm/s, and 1.59 ± 0.79 mm/s, P = 0.36). In NHPs, every 1 mm Hg increase in IOP was associated with a 0.13 mm/s reduction in arterial velocity, 0.10 mm/s reduction in venous velocity, and 0.01 mm/s reduction in capillary velocity (P < 0.001) when accounting for differences in mean arterial pressure (MAP).

Conclusions: Blood flow by direct visualization of individual erythrocytes can be quantified within capillary plexuses. Capillary velocity decreased with experimental IOP elevation.

The retina is an extension of neural tissue, allowing for cerebral microvasculature dynamics to be visualized and quantified via retinal imaging.¹ Measures of retinal vascular dysfunction may serve as biomarkers for leading causes of blindness, including macular degeneration, diabetic retinopathy, and glaucoma.² Measurement of alterations in vascular dynamics may allow for early diagnosis of disease before irreversible structural changes occur. However, measurement of vascular dynamics is complex. Vascular dynamics may differ between vascular plexuses and be differentially affected in ocular disease.^3–5 Furthermore, retinal capillary perfusion has both temporal and spatial heterogeneity that has yet to be quantified in full due to imaging limitations.⁶ 

Methodologies with high reproducibility and resolution are needed to determine absolute blood flow in retinal capillaries. Fluorescein angiography is limited to two-dimensional structural information of superficial vasculature.^7,8 Optical coherence tomography angiography (OCTA) utilizes sequential axial scans of the retina and motion contrast to identify blood flow. OCTA provides a noninvasive, three-dimensional map of retinal and choroidal circulation with greater depth of resolution compared to fluorescein angiography,⁹ but currently only allows for a static metric of retinal perfusion due to limited temporal resolution for examining blood flow rates or volumes dynamically. VISTA OCTA provides dynamic information in terms of blood flow speeds, but is limited by field of view and use of a surrogate marker for speed.¹⁰ As currently used, OCTA has a minimum threshold for detection of flow which may be limiting in that dynamic findings, such as erythrocyte stasis,¹¹ hypoperfusion,¹² and retrograde flow,¹³ may not be appreciated or quantified. Other imaging techniques, such as laser speckle contrast imaging, allow for reproducible, dynamic blood flow analysis, but is limited to bulk flow data, measured in arbitrary units, and unable to determine retinal capillary flowrates.¹⁴ Adaptive optics-scanning laser ophthalmoscopy (AO-SLO) has the highest resolution of current technologies and is able to determine absolute erythrocyte velocities and image capillaries precisely, but has a limited field of view and has not yet been used to assess plexus-specific capillary velocities.^15–17 

Erythrocyte mediated angiography (EMA) is a technique in which fluorescently labeled autologous erythrocytes are tracked in vessels in vivo, allowing for a reproducible and precise measure of blood flow.¹⁸ This is similar to leukocyte fluorography, applied to leukocyte imaging in the mouse retina.¹⁹ EMA when used synergistically with OCTA in multimodal imaging can provide high-resolution imaging to capture both temporal and spatial retinal capillary blood flow data. In this study, we use a dual modal approach with OCTA and EMA to identify and measure plexus-specific absolute retinal capillary blood flow velocity and acceleration in vivo in both nonhuman primates (NHPs) and humans. We further assess the role of alterations in intraocular pressure (IOP) on blood flow velocity in NHPs. 

Methods

Human Participants

A total of 11 human participants with and without glaucoma were imaged between June 2017 and October 2021. One participant was imaged twice within the study period. Human subjects were recruited from the Department of Ophthalmology and Visual Sciences at the University of Maryland, Baltimore. All experiments involving human subjects followed the study protocols approved by the Institutional Review Board at the University of Maryland, Baltimore. This study was carried out in accordance with the Declaration of Helsinki and informed consent was obtained from all human subjects prior to enrollment. To be eligible for inclusion, subjects had to be at least 18 years of age, have open angles on gonioscopy, and have best corrected visual acuity of at least 20/200 in the study eye. Control subjects had no clinical signs of glaucoma, had peripapillary retinal nerve fiber layer (RNFL) thickness within the normal range for their age group, had cup-to-disc ratio under 0.5, and had IOP that was never recorded above 21 mm Hg. The diagnosis of glaucoma or glaucoma suspect was made based on Preferred Practice Patterns of the American Academy of Ophthalmology.^20,21 Subjects with known allergy to indocyanine green (ICG), iodine, or shellfish, who were pregnant or nursing, had significant liver disease or uremia, or were participating in any other investigational drug study were excluded from this study. 

Experimental Animals

Two healthy, adult male rhesus macaques aged 14 and 20 years old without known ocular or systemic disease were included in this study. Baseline examinations were conducted to ensure ocular health during the study period. Imaging was conducted from August 2020 to February 2022. Experiments were performed in adherence to the guidelines set by the Association for Research and Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. The methods of this study related to the NHPs were carried out in accordance with the protocol approved by the Institutional Animal Care and Use Committee at the University of Maryland, Baltimore. 

Human Erythrocyte Preparation

Erythrocyte preparation was conducted as described in prior publications.^11,18,22 Briefly, 34 mL of blood was drawn from each study participant and autologous erythrocytes were isolated from whole blood for processing with ICG dye. An osmotic shock technique was used to load erythrocytes with 1 mM of ICG. Up to 3 mL of ICG-loaded cells were intravenously injected during image acquisition. 

NHP Erythrocyte Preparation

Seventeen mL of blood was drawn for processing with 5,6-carboxyfluorescein succinimidyl ester (CFSE; Molecular Probes, Eugene, OR, USA) and reconstituted in anhydrous dimethyl sulfoxide with a method similar to the human erythrocyte preparation, as previously described by Pottenburgh et al.²³ Autologous erythrocytes were isolated from whole blood and loaded with 7.5 mM of CFSE. Up to 1.2 mL of CFSE-loaded cells were intravenously injected during image acquisition. 

Human In Vivo Imaging

All eyes were pharmacologically dilated with one drop of tropicamide 1%. Baseline infrared autofluorescence images were taken before injecting a bolus of fluorescent erythrocytes. Ten-second angiograms centered on the macula were obtained with the Heidelberg Retinal Angiogram 2 (Heidelberg Engineering, Heidelberg, Germany) using a 15-degree horizontal × 7.5-degree vertical field of view taken at 24.6 frames per second. OCTA scans centered on the fovea were taken using 10 × 5-degree and 10 × 10-degree protocols, consisting of 512 a-scans × 512 b-scans with 5 to 10 microns between b-scans and 5 to 7 frames averaged per b-scan location. En face images of the superficial vascular plexus (SVP), intermediate capillary plexus (ICP), and deep capillary plexus (DCP) were generated using the segmentation algorithms and slab definitions of the Heidelberg Eye Explorer software (version 1.10.3.0; Heidelberg Engineering, Heidelberg, Germany). En face images were processed using projection artifact removal (PAR).²⁴ Plexus segmentation was confirmed manually by a trained observer (authors O.J.S. and S.C.). Subjects’ IOP, blood pressure, heart rate, respiratory rate, temperature, and oxygen saturation (SpO₂) were recorded using standard clinical instruments. Demographic data and medical history including medications were also recorded. 

NHP Anesthesia and Experiment Protocol

Prior to the experimental session, the animal was sedated with ketamine (5–10 mg/kg) and intubated by trained veterinary technicians with an endotracheal tube. General anesthesia was maintained with 1.5% to 3% isoflurane with 100% oxygen. Vecuronium (40–60 ug/kg, followed by 0.35–45 ug/kg/min) was used to limit eye movement. Body temperature was maintained at physiologic levels using a thermal blanket and blood pressure was monitored using a blood pressure cuff on the arm. Imaging was conducted with the animal in a prone position with the head extended and upright with a wire lid speculum to keep the eyelid open. One drop of tropicamide 1% was administered for pupil dilation. Sterile Balanced Salt Solution (Alcon Surgical, Inc., Geneva, Switzerland) was used to maintain corneal hydration for image clarity. Baseline blue autofluorescence images were taken prior to administering any fluorescent cells. EMA scans were taken with a 30-degree horizontal × 30-degree vertical field of view at 15.4 frames per second, and OCTA scans were acquired and analyzed with the same specifications used for human subjects. Blood pressure, heart rate, respiratory rate, temperature, oxygen saturation, and end tidal carbon dioxide levels were recorded during imaging by a trained veterinary technician. 

Intrasession and Intersession Repeatability

To assess intra- and intersession repeatability of measurements with EMA, the NHPs underwent repeated imaging sessions and had multiple angiograms taken during each session. NHP-1 was imaged at 5 sessions and NHP-2 was imaged at 4 sessions. Each NHP had a minimum of 2 weeks of recovery between visits. Coefficient of variation (CV) was used to determine intravisit variability for both humans and NHPs in overall erythrocyte velocity measurements for each session that had two or more tracked angiogram sequences. In the NHPs, CV was additionally used to determine the intravisit and intervisit variability of measurements through the same capillary over time. 

Assessing IOP and Mean Ocular Perfusion Pressure Changes on Erythrocyte Velocity Measurements

Elevated IOP was induced experimentally in animals using external pressure with a tight-fitting lid speculum. Angiograms were taken at both baseline and high IOP conditions. IOP was measured with a Tono-Pen AVIA tonometer (Reichert, Inc., Depew, NY, USA) at baseline, immediately after placement of the tight speculum, every 5 minutes after placement of the speculum, and after removal of the speculum. Mean ocular perfusion pressure (MOPP) was calculated for each condition using the measured IOP, systolic blood pressure (SBP), and diastolic blood pressure (DBP).²⁵ 

Image Analysis for Erythrocyte Velocity Measurements

All angiograms were processed using intensity-based image registration to correct for eye movement in MATLAB (MathWorks version R2022a, Natick, NY, USA), then semi-automated erythrocyte tracking was done as previously described and validated.¹⁸ When analyzing flow through the vascular circuit (artery to capillary to vein), a single cell was tracked only if it was present continuously across frames, beginning with the frame where the cell was first visualized in an arteriole to the last frame the cell could be identified in a venule (Fig. 1). Three or more consecutive frames were required to permit velocity and acceleration calculations for an individual cell. To calculate average velocities for each type of blood vessel, the last measured arteriole velocity, the first and last capillary velocity, and the first vein velocity were excluded from analysis to avoid potential vessel misclassification at these areas of overlap. Acceleration was defined as the slope of instantaneous velocities over time. 

Figure 1.

 

A stack of EMA images (left) and velocity measurements (right) for a single erythrocyte tracked through the entire vascular circuit. The scale bar shown represents 200 micrometers.

To precisely identify cell locations, we overlaid sequential EMA frames onto OCTA images of the corresponding macular location using ImageJ (National Institutes of Health, Bethesda, MD, USA). En face OCTA images of the SVP, ICP, and DCP were then used to cross-reference the path of a given erythrocyte and identify its axial location within the retina (Fig. 2). Cells with capillary paths that could not be identified in this manner due to being outside of the frame of the OCTA image or poor segmentation leading to inadequate visualization of the capillary path in the OCTA image were excluded from any analysis stratified by vascular layer. Four total human eyes were excluded from this analysis due to the inability to perform the EMA-OCTA overlays. Two excluded eyes had poor quality OCTA scans. Two eyes had angiograms in which the larger vessels were not visible, which did not allow for accurate overlay and registration with the corresponding OCTA scan. 

Figure 2.

 

A stack of consecutive angiogram frames showing the movement of individual erythrocytes through macular capillaries. En face OCTA images were superimposed on the stack of angiogram frames to identify the exact axial location of the erythrocyte. Blue circles highlight examples of fluorescent erythrocytes as they move through NHP and human retinal vascular plexuses. The scale bars shown represent 200 micrometers each.

Statistical Analysis

Analysis of variance (ANOVA) with post hoc Tukey's tests or paired t-tests were used to compare erythrocyte velocity and acceleration between vascular plexuses as well as between baseline IOP and high IOP conditions. Given the multiple observations within each animal, each analysis was completed within each NHP dataset and labeled as NHP-1 and NHP-2. Generalized linear mixed effect models were used to account for repeated measures in the analysis of factors associated with erythrocyte velocity and acceleration and pseudo-R² was calculated to determine the association of erythrocyte velocity or acceleration. Pseudo-R² was calculated for the generalized linear mixed models within RStudio (PBC; RStudio, Boston, MA, USA) using the LME4 package. SPSS version 27.0.1.0 software (IBM, Armonk, NY, USA) was used for the remainder of the statistical analysis. 

Results

Participant Characteristics

We imaged 4 eyes of 2 NHPs across 4 to 5 sessions per animal and 18 eyes of 11 human subjects. Table 1 summarizes the demographic characteristics and relevant clinical characteristics of the imaged subjects and Supplementary Table S1 summarizes the clinical characteristics during the experimental animal imaging sessions. 

Table 1.

 

Human Subject Characteristics

 

Baseline Human and NHP Retinal Capillary Erythrocyte Velocity

The average IOP for NHPs and human subjects in baseline conditions was 14.13 ± 2.03 mm Hg and 15.63 ± 2.79 mm Hg, respectively (mean ± SD). Capillary erythrocyte velocity at baseline IOP was 0.64 ± 0.29 mm/s in NHPs (range of 0.14 to 1.85 mm/s) and 1.55 ± 0.63 mm/s in humans (range of 0.46 to 4.50 mm/s). We further stratified mean erythrocyte velocity by vascular plexus. In NHPs, mean erythrocyte velocity in the SVP, ICP, and DCP at baseline was 0.69 ± 0.29 mm/s, 0.53 ± 0.22 mm/s, and 0.63 ± 0.27 mm/s, respectively. The differences between layers were not statistically significant for either animal (P = 0.14 for NHP-1 and P = 0.28 for NHP-2). Mean erythrocyte velocity in humans did not differ significantly among the SVP, ICP, and DCP (1.46 ± 0.61 mm/s, 1.61 ± 0.58 mm/s, and 1.60 ± 0.79 mm/s, respectively; P = 0.09 between SVP and ICP, P = 0.47 between SVP and DCP, and P = 0.92 between ICP and DCP). In humans, capillary erythrocyte velocity in participants with glaucoma or glaucoma suspects was not significantly different compared to that of control participants (Table 2). IOP, mean arterial pressure (MAP), and MOPP also did not differ significantly between control participants and participants with glaucoma or glaucoma suspects. 

Table 2.

 

Healthy Control and Glaucoma Subjects

 

Intra- and Intersession Repeatability

One animal imaging session was used to assess intrasession repeatability. CV for repeated velocity measurements in this visit was 37.6% for arteries, 7.0% for capillaries, and 16.7% for veins between 2 angiograms. The intervisit CV of velocity measurements with both NHPs was 25.7 ± 13.1% in arteries, 26.2 ± 3.5% in capillaries, and 33.3 ± 16.8% in veins (mean ± SD). In human subjects, overall intravisit CV for velocity was 7.1 ± 6.4% in arteries, 9.9 ± 9.2% in capillaries, and 5.8 ± 4.2% in veins (mean ± SD). 

In NHPs, to assess intrasession repeatability within individual capillaries, velocity was measured in 3 individual capillaries during the same imaging date in baseline conditions and CV was calculated at 19.24%. Twelve capillaries had repeat velocities measured across at least 2 separate imaging dates in baseline conditions, which demonstrated intersession CV of 31.21%. 

Assessing IOP and MAP Effects on Blood Flow Measurements

After inducing high IOP in the experimental animal sessions, average IOP was 28.67 ± 6.31 mm Hg. Mean retinal capillary erythrocyte velocity was measured within the elevated IOP condition (Table 3). Erythrocyte velocity differed significantly across baseline and high IOP conditions (P < 0.01 for NHP-1 and P = 0.03 for NHP-2). Erythrocyte velocity did not differ among the SVP, ICP, and DCP with high IOP conditions (P = 0.76 for NHP-1 and P = 0.17 for NHP-2). 

Table 3.

 

Average Capillary Velocity in NHP (Baseline and High IOP) and Human Subjects

 

On multivariable analysis using a generalized linear mixed model, IOP was significantly associated with velocity when accounting for the random effects of study date, eye, and animal and the fixed effect of MAP. For every 1 mm Hg increase in IOP, velocities in arteries decreased by 0.13 mm/s (95% confidence interval [CI] = range –0.08 to –0.18, P < 0.001; pseudo-R² of 19% for the model), decreased by 0.01 mm/s in capillaries (95% CI = –0.01 to –0.02, P < 0.001; pseudo-R² of 25%), and decreased by 0.10 mm/s in veins (95% CI = –0.06 to –0.15, P < 0.001; pseudo-R² of 48%). MAP was not significantly associated with velocity in the above models. 

Individual capillaries visualized both at baseline and high IOP were identified, as illustrated in Figure 3. Within this subset of vessels, NHP-1 showed significant decrease in capillary erythrocyte velocity with increased IOP (P = 0.01), whereas NHP-2 did not show a significant change in capillary erythrocyte velocity with increased IOP (P = 0.60). 

Figure 3.

 

Average velocity of consecutive EMA frames in the same capillary in baseline and high IOP conditions. A heatmap is shown demonstrating velocity measurements as erythrocytes move through three different capillaries (rows) in baseline and high IOP conditions. The average velocity across the capillary is shown in the top left of each frame. The scale bars shown represent 200 micrometers each.

When IOP was increased to a critical pressure of 39 mm Hg in NHP-1 at an MAP of 57 mm Hg, retrograde flow in a large artery was observed (Supplementary Fig. S1). 

Erythrocyte Acceleration

Erythrocyte acceleration within arteries, capillaries, and veins was calculated in all experimental animal sessions and in a subset of four human subjects. For human subjects, mean acceleration was –0.08 ± 0.11 mm/s² in arteries, 0.003 ± 0.02 mm/s² in capillaries, and 0.04 ± 0.12 mm/s² in veins (P < 0.001). 

In NHPs at baseline IOP, mean acceleration was –0.14 ± 0.25 mm/s² in arteries, 0.00003 ± 0.01 mm/s² in capillaries, and 0.04 ± 0.09 mm/s² in veins. In high IOP, acceleration was –0.06 ± 0.17 mm/s² in arteries (P = 0.004 for NHP-1 and P = 0.97 for NHP-2 compared to baseline), –0.0005 ± 0.002 mm/s² in capillaries (P = 0.46 for NHP-1 and P = 0.52 for NHP-2 compared to baseline), and –0.00008 ± 0.06 mm/s² in veins (P = 0.02 for NHP-1 and P = 0.04 for NHP-2 compared to baseline). 

On multivariable analysis using a generalized linear mixed model with a target of acceleration, accounting for subject, eye, and study session as random factors and both MAP and IOP as fixed factors, in capillaries, neither MAP nor IOP had a statistically significant association with acceleration (P = 0.27 and P = 0.96, respectively). In arteries, both IOP and MAP were significantly associated with acceleration (0.07 mm/s² per unit change in IOP, 95% CI = 0.01 to 0.14, P = 0.02; –0.07 mm/s² per unit change in MAP, 95% CI of –0.02 to –0.12, P = 0.01; pseudo-R² of 5%). For veins, IOP (–0.037 mm/s² per unit change in IOP, 95% CI = –0.01 to –0.06, P = 0.01) but not MAP (P = 0.97) was significantly associated with acceleration in veins (pseudo-R² of 6%). 

Discussion

We present a novel technique for quantifying erythrocyte velocity and acceleration in retinal capillaries across plexuses. This multimodal technique allows for precise representation and quantification of erythrocyte dynamics in retinal capillaries in vivo in four dimensions (three dimensions accounting for time). In our experimental animal model, we observed a significant decrease in erythrocyte velocity in retinal capillaries with IOP elevation. 

We found that IOP had a negative association with erythrocyte velocity in NHPs in our multivariable analysis. Our findings are consistent with prior studies evaluating the effect of IOP changes on retinal blood flow using OCT angiography.²⁶ In a study by Patel et al., elevated IOP via cannulation resulted in decreased capillary density in superficial vessels of the optic nerve head in NHPs.²⁷ In a study of human subjects receiving intravitreal injections, both superficial and deep macular OCTA macula scans demonstrated decreased capillary density in association with IOP elevation.²⁸ Thus, decreased vessel density following acute IOP elevation may be related to erythrocyte velocity falling below the minimum threshold of visibility on OCTA scans. Liang et al. studied the effect of IOP on optic nerve head blood flow and similarly found that autoregulation was deficient at low blood pressure, resulting in decreased optic nerve head blood flow. Our work differs in that we focused on retinal capillary flow, but adds additional evidence that elevated IOP can overcome autoregulation and result in decreased blood flow. 

Various studies using adaptive optics techniques have quantified retinal vessel velocity. Prior studies have found pulsatile variations in blood flow in human capillaries and across the mouse vascular tree.^29,30 Capillary velocity in NHPs at baseline in our study ranged from 0.14 to 1.85 mm/s, and, in humans, from 0.46 to 4.50 mm/s. Other studies using adaptive optics techniques have also reported ranges of erythrocyte and leukocyte capillary velocities from 0.26 to 4.24 mm/s.^31–34 The maximum theoretical velocity which could be visualized by our EMA approach as used in this study ranges from 52 mm/s to 104 mm/s, which captures all physiologic blood flow for retinal vasculature of this scale when compared to prior blood velocity estimates.^18,31 Despite lower temporal resolution than adaptive optics methodologies, EMA captured a similar range of velocity measurements within retinal capillaries and offers a larger field of view, facilitating its use for multimodal imaging with OCTA, which we share in a previously published dataset.³⁵ 

Variability of flow was notably higher in this study compared to our prior study on EMA in human subjects. Intravisit CV was 7.1% in arterioles, 9.9% in capillaries, and 5.8% in venules for human subjects, compared to previously reported intravisit CV of 3.57% and intervisit CV of 4.85%.¹⁸ Capillaries may intrinsically demonstrate higher variability as erythrocytes flow single file through capillaries as compared to bulk flow in arterioles and venules and there may be regional variation in capillary hemodynamics.³⁶ Other studies have noted CV ranging from 19% to 50% for capillary velocity measurements.^31,37 Factors such as vessel diameter, cardiac cycles, and variable flow dynamics depending on cell position within vessels were not accounted for which may have increased CV as well.^29,38 The rate of frame acquisition was manually validated as well, ensuring that frame rates used to estimate velocity were accurate. 

In our NHP animal model, multiple experimental factors may have affected blood flow measurements and contributed to higher variability and lower velocities compared to human subjects. We note a relatively low average MAP in the NHPs of 49.93 mm Hg, which may have lowered velocities. Systemic hyperoxia as induced by 100% supplemental oxygen administered as part of the anesthesia protocol has been shown to have variable effects on retinal blood flow. Prior work using EMA demonstrates an increase in erythrocyte velocity with hyperoxia in healthy control patients, whereas other studies have shown an overall decrease in blood flow or plexus-specific changes in flow.^39–41 Additionally, ketamine and isoflurane are commonly used in NHP studies, but may cause hemodynamic changes from baseline, such as a reduction in heart rate and may reduce IOP compared to pre-anesthesia levels, thus serial IOP, blood pressure, and heart rate measurements were taken during imaging to best account for any dynamic changes.⁴² In regard to vecuronium, which was used to limit eye movement in this study, studies have shown minimal hemodynamic changes with its use.⁴³ 

Limitations of this study include a modest sample size, constraints due to noninvasive experimental methods, and measurement errors due to the technology being used. Lid speculums are known to elevate IOP, allowing for noninvasive IOP modulation as opposed to anterior chamber cannulation which poses greater infection risk.⁴⁴ Keratometry values did not change with lid speculum usage, as confirmed by manual keratometry and supported by prior evidence.⁴⁵ Axial lengths of NHPs were comparable to human subjects. Although repeated IOP measurements could induce a reduction in IOP, serial measurements were necessary given the dynamic nature of IOP changes induced by the lid speculum and anesthetic use.⁴⁶ Continuous IOP monitoring could have provided greater precision. Similarly, blood pressure was not measured continuously or altered systematically, limiting our assessment of the role of MAP on retinal capillary blood flow. Additionally, we acknowledge that due to the technical limitations of the scanning laser ophthalmoscope, artifacts within each frame may decreases the accuracy of velocity measurements.⁴⁷ Such artifacts due to raster scanning may lead to overestimation or underestimation of velocity measurements in the direction of the device's slow scan, which is vertical. By averaging measurements taken from all vessels of varying directionality, the error induced by this effect was minimized. Nonetheless, the maximum error due to raster scanning artifact in our study is less than 3%, as we consider the maximum estimated vertical distance traveled by a cell divided by the total vertical distance per frame, then shown as a percent error based on the formula by Tam et al.⁴⁷ Further, despite the potential nondifferential effect of this error, our data demonstrate a statistically significant effect of IOP modification on blood flow. 

In conclusion, our present study demonstrates decreased blood flow in all capillary plexuses with elevated IOP. We also measured acceleration in capillary plexuses, which presents an opportunity to study local, dynamic forces on erythrocytes. This establishes that multimodal imaging with EMA and OCTA may permit the study of plexus-specific differences in blood flow, thus enhancing our understanding of dynamic retinal blood flow and may be applied to study the differential effects of retinal pathology, insults, or medications on different plexuses. The observed disruption of capillary blood flow related to increased IOP enhances our understanding of the relationship between glaucoma and blood pressure and elucidates the role that impaired blood flow plays in the pathogenesis of that disease. 

Acknowledgments

Supported by the University of Maryland Strategic Partnership: MPowering the State, a formal collaboration between the University of Maryland College Park and the University of Maryland Baltimore, and the National Institutes of Health (R01 EY031731) (O.J.S.). 

Disclosure: V.Y. Chen, None; J.A. Pottenburgh, None; S.-E. Chen, None; S. Kim, None; L. Mayo, None; A. Demani, None; M. Cruz, None; A. Park, None; L. Im, None L. Magder, None; O.J. Saeedi, Heidelberg Engineering (F)