Retinal Blood Flow and Retinal Blood Oxygen Saturation in Mild to Moderate Diabetic Retinopathy

Abstract

Purpose: The aim of this study was to evaluate the relationship between retinal blood flow (RBF) and retinal blood oxygen saturation (SO₂) in mild to moderate nonproliferative diabetic retinopathy (NPDR) and in age-matched controls.

Methods: One eye of each of 15 healthy subjects (68 ± 6 years) and 13 subjects with mild to moderate NPDR (67 ± 10 years) was dilated. None of the patients with NPDR had received treatment for their retinopathic changes or had any evidence of sight-threatening characteristics. Doppler Fourier-domain optical coherence tomography blood flow was measured using the prototype RTVue system; six separate measurements each comprising an upper and a lower nasal pupil scan were acquired. Six hyperspectral retinal measurements were acquired using a noninvasive hyperspectral retinal camera (prototype H-8.5 HR Camera).

Results: Total RBF was significantly lower in NPDR when compared to controls (42.7 ± 7.5 vs. 33.0 ± 9.2 μL/min; P = 0.004). Mean retinal arterial and venular SO₂ were higher in NPDR than in controls (94.7 ± 2.4% vs. 92.9 ± 1.6%, P = 0.02; 62.5 ± 5.7% vs. 56.3 ± 4.7%, P = 0.003). This study showed a correlation between RBF and arteriolar SO₂ in both controls (r = 0.58, P = 0.02) and NPDR (r = 0.54, P = 0.05), but no correlation between venular RBF and venular SO₂ in controls (r = 0.24, P = 0.83) or in NPDR (r = 0.23, P = 0.45). The arteriovenous difference (AV difference) was lower in the NPDR group when compared to controls (30.6 ± 6 vs. 36.7 ± 5.3, P = 0.008).

Conclusions: This study found a lower total RBF and a lower AV difference in the NPDR group, suggesting a reduced oxygen uptake from the retina in people with relatively early diabetic retinopathy.

Diabetic retinopathy (DR) is a primary source of visual loss in the world, including North America.¹ Diabetic retinopathy appears to be allied closely to disturbances in the vasculature. Hence, many studies have directly investigated the inner retinal blood flow (RBF). However, the results of inner RBF disturbance in DR have been generally contradictory,^2–11 although many studies indicate some aspect of disturbance. 

Elevated RBF has been classically proposed to ultimately cause the progress of DR, probably because of amplified frictional forces (i.e., shear stress) on the endothelial cells lining the walls of retinal vessels.¹² However, the exact nature of the blood flow disruption is debatable, perhaps due to the variety of methods used to measure retinal hemodynamics, the diverse stages of retinopathy considered, and the diversity of the diabetic groups.¹³ From a clinical perspective, the assessment of ocular hemodynamics proposes an improved DR severity classification and perhaps a flag for new treatment opportunities. In addition, retinal hypoxia is thought to promote the production of vascular endothelial growth factor and consequently neovascularization. Assessment of retinal oxygenation in patients with diabetes may help to evaluate the severity of DR and the response to treatments. 

It has been suggested that change in blood circulation leads to functional damage and broad retinal tissue impairment and disturbance during diseases such as DR.¹⁴ In this respect, it will be critical to study the oxygen distribution or consumption of the retina and its alteration in response to DR, since RBF or retinal blood oxygen saturation (SO₂) alone represents only part of the required information to calculate retinal oxygen utilization. Measurement of these alterations could be used to improve the early detection and management of diseases such as diabetes/DR. Simultaneous RBF measurement and retinal hemoglobin oxygenation are required to assess absolute values of oxygen delivered to the retina,^15,16 which is not noninvasively possible at the moment. 

Assessing oxygen tension in the retina has been achieved using O₂-sensitive microelectrodes introduced into the eye.^17–21 While this technique is precise and can accurately define oxygen utilization, the invasive nature of the technique limits its usage to animal models and disqualifies it from clinical purpose. Alternative methods using the introduction of a phosphorescent dye have been utilized to derive oxygen concentration in the optic nerve head and retinal vessels.^19,22 Nonetheless, the use of such a dye in humans is not accepted due to toxicity concerns. Imaging techniques established on oxygenated hemoglobin and deoxyhemoglobin and their spectral changes have been used in humans to noninvasively assess oxygen saturation in retinal vessels.^18,23–27 Oxygen saturation in retinal vessels has been found to be greater in DR in contrast to control subjects.^15,28–31 Assessment of oxygenation in the retina may facilitate the early detection of DR. The aim of this study was to investigate total RBF and retinal blood oxygen saturation in NPDR. 

Methods

Sample

This study received approval by the University of Waterloo Office of Research Ethics and the Research Ethics Board of the University Health Network, University of Toronto, Canada. Informed consent was obtained from each subject after explanation of the nature and possible consequences of the study according to the tenets of the Declaration of Helsinki. The sample consisted of 28 volunteers in two groups of subjects with mild to moderate NPDR (n = 13; mean age 67; SD 10 years; Table 1) and age-matched healthy controls (n = 15; mean age 68; SD 6 years; Table 1). One eye of each subject was selected for the study. All subjects had a corrected visual acuity of 20/40 or better. Subjects were excluded if they had a family history of any ocular disease apart from DR. None of the patients with NPDR had received treatment of any kind for their retinopathic changes or had any evidence of diabetic macular edema or any other sight-threatening characteristic. Subjects taking medications with known side effects on hemodynamics and subjects with rheumatologic diseases were disqualified to enter the study. None of the subjects smoked or had any respiratory diseases. The NPDR participants had early signs of retinopathy such as microaneuryms and a few scattered dot hemorrhages, but there was no presence of macular edema, ischemic-related signs such as venous beading, deep blot hemorrhages, or intraretinal microvascular abnormalities or imminent neovascularization. Subjects abstained from consuming caffeine 12 hours before the study. 

Table 1

 

Group Mean Age, Male-to-Female Ratio, and Glycosylated Hemoglobin

 

Assessment of Retinal Blood Flow

Doppler Fourier-domain optical coherence tomography (Doppler FD-OCT) is used to measure RBF. This novel imaging method offers RBF assessment using a physical phenomenon called Doppler phase shift.³² The principle of the Doppler phase shift has been incorporated into the commercially available Doppler FD-OCT (Optovue, Inc., Freemont, CA, USA). Doppler FD-OCT produces high-resolution cross-sectional images of the retina. This instrument uses a laser light source of 841 nm with bandwidth of 49 nm and an incident power of 500 μW on the cornea. These factors deliver an axial resolution of 5.4 μm in tissue.³³ System transverse resolution was 20 μm.³⁴ 

In contrast to morphologic FD-OCT systems that generate structural images, the prototype Doppler FD-OCT evaluates the Doppler phase shift between two consecutive A-scans. Light reflected from moving particles undergoes Doppler phase shift. 

Flow velocity is determined by  

 

 

where v is the flow velocity in an OCT voxel, and ΔΦ = ϕ1 − ϕ2 is the Doppler phase shift. Φ1 and Φ2 are the phase of voxels in the same position in consecutive OCT axial scans, λ₀ is the source center wavelength, n is the refractive index of the medium, T is the time interval between consecutive scans, and θ is the Doppler angle defined by the OCT beam axis relative to the line perpendicular to blood vessel flow axis.³⁵ 

Assessment of Retinal Blood Oxygen Saturation

The principle of the oximeter has been described elsewhere,³⁶ but briefly, a prototype custom-built fundus camera (H-8.5 HR Camera; Optina Diagnotics, Inc., Montreal, QC, Canada) is the foundation of the hyperspectral imaging system that incorporates a tunable laser source (TLS) as the light source. The TLS, built on Bragg grating filtering technology (Photon ETC, Montreal, QC, Canada), makes it possible to transfer wavelengths with a half peak bandwidth of 2 nm (range, 400–1000 nm). The TLS permits rapid wavelength presentation from the stable and powerful super-continuum light source (Leukos-SM-30-OEM; Leukos Innovative Optical Systems, Limoges, France). The imaging system is administered by using PhySpec (Photon ETC), a software program that operates the Bragg tunable filter (BTF) and charge-coupled device camera to permit factors such as operator-defined wavelength range and wavelength interval. A low-power white light source (delivering approximately 100 mW over 420–2400 nm) and the BTF are utilized to eliminate the use of conventional flash lamp, and consequently all images are obtained at low light levels, thus decreasing any potential change of the ocular metabolites and photopigment status.³⁶ Multispectral retinal images were taken using specific wavelengths (i.e., 548, 569, 586, 600, 605, and 610 nm) at an exposure time of 80 ms for each wavelength. However, the notion of the optical density ratio introduced by Beach et al.²³ of the two monochromatic wavelengths (586 and 605 nm) was performed to calculate retinal blood oxygen saturation. 

Image Processing

The imaging system is based on using PhySpec (Photon ETC), and the cubes are normalized and corrected for wavelength-dependent optical scaling and slight involuntary eye movements by means of this software. The 570 nm was selected as reference image for scaling since the vessels were most distinct at this wavelength. The optic nerve head (ONH) portion of the image was chosen as the reference region, and all images in the cube were aligned correspondingly. Furthermore, a Gaussian filter was utilized for boosting the details of each cube. 

Procedures

Refraction, logMAR visual acuity, Goldmann applanation tonometry, and resting blood pressure were assessed prior to dilation of the study eye. The pupil of the study eye was dilated using Mydriacyl 1% (Alcon, Mississauga, ON, Canada) at the beginning of each visit to achieve an adequate view of the fundus for the RBF image acquisition and retinal blood oxygen saturation measurements. A minimum of six separate FD-OCT Doppler measurements (i.e., each separate measurement comprising an upper nasal pupil scan and a lower nasal pupil scan) and six blood oxygen saturation measurements were acquired on the first-degree arterioles and venules. 

Analysis

Differences in outcome variables between controls and NPDR subjects were assessed with Student's t-tests. A Pearson correlation analysis was performed to examine the relationship between RBF and retinal blood oxygen saturation. In all analyses, 2-tailed P < 0.05 indicated statistical significance. 

Results

The results of this study showed that the total RBF was significantly lower in diabetic patients (33.0 vs. 42.7 μL/min; P = 0.004, Fig. 1). 

Figure 1

 

Box plots representing total retinal blood flow in control and NPDR patients (42.7 vs. 33.0 μL/min; P = 0.004). The asterisk represents extremes, and the small circles represent outliers. The error bars show standard deviation.

The mean SO₂ values in the arterioles and venules for the control participants and NPDR subjects are shown in Table 2. 

Table 2

 

Group Mean Values for Blood Oxygen Saturation and Total Retinal Blood Flow for Nonproliferative Diabetic Retinopathy and Aged-Matched Controls

 

The results of this study showed that mean retinal arteriolar blood oxygen saturation was higher in diabetic patients when compared to the healthy controls (94.7 ± 2.4% vs. 92.9 ± 1.6%, P = 0.025). There was also a significant difference in venular blood oxygen saturation between NPDR and control subjects (62.5 ± 5.7% vs. 56.3 ± 4.7%, P = 0.003). The arteriovenous difference (AV difference) was significantly lower in NPDR subjects when compared to controls (30.6 ± 6.0% vs. 36.7 ± 5.3%, P = 0.008). 

This study revealed a correlation between total RBF and arteriolar oxygen saturation in both controls (r = 0.58, P = 0.02; Fig. 2) and NPDR (r = 0.54, P = 0.05; Fig. 2); however, the study revealed no correlation between total RBF and venular oxygen saturation in either controls (r = 0.24, P = 0.83; Fig. 3) or NPDR (r = 0.23, P = 0.45; Fig. 3). 

Figure 2

 

Arteriolar retinal blood oxygen saturation as a function of total retinal blood flow (retinal venular blood flow) in control (r = 0.58, P = 0.02) and NPDR group (r = 0.54, P = 0.05).

Figure 3

 

Venular retinal blood oxygen saturation as a function of total retinal blood flow (retinal venular blood flow) in control (r = 0.24, P = 0.83) and NPDR group (r = 0.23, P = 0.45).

Figure 2 represents retinal venular blood flow as a function of retinal arteriolar blood oxygen saturation in both control and NPDR groups, and Figure 3 illustrates retinal venular blood flow as a function of retinal venular blood oxygen saturation in both control and NPDR groups. 

Discussion

It has been shown that DR is associated with early retinal vascular dysregulation. Therefore, many studies have investigated the inner RBF. Nevertheless, the results for inner RBF disturbance in DR have been contradictory,^2–11 although most indicate some aspect of disturbance. It has been reported that NPDR patients or patients without retinopathy have disturbances in RBF at different times of the day due to change in their plasma glucose, which was suggested as a possible cause of microvascular changes in DR.³⁷ To the best of our knowledge, this is the first study that investigates the relationship between RBF and retinal oxygen saturation. The results of this study showed lower total RBF in people with NPDR when compared to controls (P = 0.004). The AV difference was also significantly lower in NPDR subjects when compared to controls. The lower total RBF and the lower AV difference exhibited by the NPDR group suggest a reduced oxygen uptake from the retina in people with relatively early DR. The study also showed an association between total RBF and arteriolar oxygen saturation; however, it did not reveal any correlation between total RBF and retinal venular blood oxygen saturation. Bearing in mind the small sample size, the study does not eliminate the possibility of an association between retinal venular blood oxygen saturation and total RBF. The lack of correlation between venular RBF and venular oxygen saturation might be explained either by an absence of any effect or by insufficient study power. Therefore, further studies are required to substantiate any correlation between total RBF and arteriolar and venular oxygen saturation. 

The combination of retinal oxygen saturation and RBF can be used to extract more information about retinal metabolism, since  

 

 

Theoretically, each gram of hemoglobin binds 1.39 mL oxygen. The oxygen saturation curve represents a PaO₂ (arterial oxygen tension) as a function of SaO₂ (arterial SO₂); 0.003 stands for the oxygen solubility coefficient in human plasma (TRBF = Total RBF [μL/min]). 

The direction of the change in total RBF in DR is controversial.^5,38–40 The present study showed a reduction in RBF when comparing NPDR subjects with controls. Our oximetry data are in agreement with Hardarson's thesis results.⁴¹ Hardarson interpreted this result to occur due to impaired diffusion of oxygen to the adjacent tissue through the thickened arteriolar walls of diabetic subjects or because of damaged capillary network. Hardarson also suggested several other possibilities, such as increased total RBF, decreased oxygen consumption due to cell death, and technical artifact. Higher blood flow would be inclined to reduce the release of oxygen from hemoglobin and therefore would result in reduced diffusion of oxygen across the arteriolar walls.⁴¹ 

Measured changes in total RBF might represent contributions from different parts of the retina with both decreased and increased RBF. Lack of association between RBF and SO₂ might be as a result of the fact that both retinal and choroidal circulations provide oxygen and nutrient supply to the retina. The retina could take oxygen from choroid; hence the RBF and the retinal blood oxygen saturation are not associated. A limitation of the current oximetry techniques is that tissue oxygen saturation measurements are not feasible yet. Another challenge is that simultaneous measurement of RBF and retinal oxygen saturation is not possible yet. 

In summary, the results of this work revealed a lower total RBF and a lower AV difference exhibited by the NPDR group when compared to age-matched controls, suggesting a reduced oxygen uptake from the retina in people with relatively early DR. The work also revealed a relationship between retinal arteriolar blood oxygen saturation and RBF. 

Acknowledgments

The authors thank Susith Kulasekara, Ayda Shahidi, and Sunni Patel for their general assistance. 

Supported by the Ontario Research Fund for Research Excellence (RE-04-034) and an anonymous donor. Optina, Inc. provided financial and in-kind support and student stipend funding. 

Disclosure: F. Tayyari, None; L.-A. Khuu, None; J.G. Flanagan, None; S. Singer, None; M.H. Brent, None; C. Hudson, Optovue, Inc. (F), Optina (F, R)