Retinal Vessel Oxygen Saturation under Flicker Light Stimulation in Patients with Nonproliferative Diabetic Retinopathy

Abstract

Purpose.: We investigated the response of retinal vessel diameters and oxygen saturation to flicker light stimulation of neuronal activity in patients with diabetic retinopathy.

Methods.: We included 18 patients with nonproliferative diabetic retinopathy (mean age 62.2 ± 8.3 years, diabetes type 1 in 4 patients and type 2 in 14, hemoglobin A1c 7.7 ± 0.9%, duration of diabetes 24.1 ± 9.3 years) and 20 age-matched healthy controls (age 66.7 ± 10.3 years). Dual wavelength (548 and 610 nm) fundus images were taken before and during luminance flicker stimulation (12.5 Hz, modulation depth > 1:25) for 90 seconds. Diameters (central retinal arterial [CRAE] and venous [CRVE] equivalents) and oxygen saturation (SO₂) were determined, and averaged for all arterioles and venules in an annular area centered at the optic disk.

Results.: Flicker light increased CRAE, CRVE, and venous SO₂ by 0.6 ± 6.6%, 2.7 ± 6.1%, and 2.0 ± 2.4% (P < 0.05), respectively, in the patients as well as 4.7 ± 8.4% (P < 0.05), 8.7 ± 5.2% (P < 0.05), and 4.2 ± 3.5% (P < 0.05), respectively, in the controls. The arterial SO₂ remained unchanged in both groups. The increase of the venous SO2 correlated significantly (P = 0.027) with that of the CRAE. There was a trend (P = 0.06) for lower increase of the venous SO₂ with higher body mass index.

Conclusions.: Our results support the thesis of an impaired regulation of oxygen supply to the diabetic retina. Whereas in healthy subjects the stimulation of neuronal activity increases the vascular diameters and, subsequently, the oxygen supply, this increase is reduced in diabetic retinopathy. This may hint at the role of endothelial dysfunction in the etiology of the disease.

Introduction

Hyperglycemia, as one of the major detrimental conditions in diabetes, may contribute to the development of retinopathy by the formation of advanced glycation end products, via the protein kinase C pathway, and by disturbance of the endothelial function. ¹ The vascular endothelium, on one hand, is the cell layer first exposed to high glucose concentration and, on the other hand, has a pivotal role in local blood flow control by the production of vasoactive factors, such as endothelin and nitric oxide (NO). Endothelial dysfunction, together with structural changes in the retinal microvasculature, results in a loss of the modulatory capacity of the vessels and, subsequently, in hemodynamic changes and tissue hypoxia. An impaired blood flow regulation in patients with diabetes mellitus has been shown recently by a reduced vasodilatation secondary to flicker light stimulation of neuronal activity. ^2–6  

Linsenmeier et al. found a reduced oxygen partial pressure (pO₂) in the inner retina of diabetic cats. ⁷ Whereas the pO₂ in the vitreous cavity was not affected by experimental diabetes in other animal models, ^8–10 it was decreased in patients undergoing vitrectomy, ¹¹ but higher over retinal areas treated with laser photocoagulation than over non-treated areas. ¹²  

In contrast to pO₂, the hemoglobin oxygen saturation (SO₂) can be assessed non-invasively by spectroscopic techniques. ^13–15 An increased venous SO₂ in patients with diabetes mellitus ¹⁶ (Hardarson SH, et al., IOVS 2008;49:ARVO E-Abstract 5366) points to an impaired tissue oxygen supply due to the attenuation of the capillary network and shunting vessels, ¹⁷ as well as diabetic alterations of the vessel wall and hemoglobin oxygen affinity. Performing SO₂ measurements during flicker light stimulation, we found an increase of venous SO₂ along with vasodilatation. ¹⁸  

In our study, we investigated the change of retinal vessel SO₂ and diameter under neuronal flicker light stimulation in patients with nonproliferative diabetic retinopathy and in an age-matched healthy control group. 

Methods

Subjects

The study adhered to the tenets of the Declaration of Helsinki and was approved by the local institutional review board. Informed consent was provided by all subjects before performing any study procedures. We included 18 patients with nonproliferative diabetic retinopathy (4 with diabetes mellitus type 1 and 14 with type 2) and 20 age-matched healthy controls. Exclusion criteria for the control group were any retinal or optic nerve diseases, eye trauma, diabetes mellitus, severe cardiovascular or respiratory diseases, and pregnancy. Subjects with a medical history of systemic hypertension were included if blood pressure was well controlled by antihypertensive therapy. Although it might be argued that antihypertensive medication may affect vascular diameters, the Beaver Dam Eye Study did not find evidence for this assumption. ¹⁹ There was no significant difference (Students t-test) in mean arterial blood pressure (MAP) between the study groups. However, the patients had a higher weight (P < 0.001) and body mass index (BMI, P < 0.001) than the controls. Clinical characteristics as well as medication of the study population are surveyed in the Table. The pupil of one eye was dilated with Tropicamide (Pharma Stulln GmbH, Stulln, Germany) before the investigation. Diabetic retinopathy was assessed from retinal photographs and classified according to the Early Treatment of Diabetic Retinopathy Study (ETDRS) criteria. ²⁰  

Table. 

 

Clinical Characterization as Well as Medication of the Study Groups

	Patients (N = 18)	Controls (N = 20)
Clinical characteristics
Gender (female)	9 (50%)	14 (70%)
Age (years)	62.2 ± 8.3	66.7 ± 10.3
MAP (mm Hg)	105.0 ± 12.9	99.8 ± 9.7
Body height (cm)	168 ± 8	166 ± 7
Body weight (kg)	92.2 ± 21.0	68.6 ± 9.9
Body mass index	32.6 ± 7.2	25.0 ± 2.9
Blood glucose (nmol/L)	10.4 ± 3.2	n./a.
Hemoglobin A1c (%)	7.68 ± 0.78	n./a.
Duration of diabetes (years)	24.2 ± 9.3	n./a.
Creatinine clearance (ml/min)	89.5 ± 25.3	n./a.
Medication
Insulin	16 (89%)	–
Oral anti-diabetics	10 (55%)	–
ACE inhibitors, AT1-blockers	17 (94%)	–
Beta-blockers	16 (89%)	–
Statins	12 (67%)	–
Diuretics	14 (78%)	–
Calcium antagonists	5 (28%)	–
Acetylsalicylic acid	10 (55%)	–
Anti-hypertensive medication	18 (100%)	11 (55%)
Laser therapy	4 (22%)	–

 n./a., not available.

Procedures

The Retinal Vessel Analyzer (DVA; IMEDOS Systems UG, Jena, Germany) was used for digital fundus photography as well as for retinal vessel analysis. The vessel diameters were measured with the software Vesselmap2 (IMEDOS Systems), ²¹ from red-free 30° images. The investigator assigned vessels to arteries or veins, and their diameters were calculated automatically by the software. The calculations of the central retinal arterial (CRAE) and venous diameter (CRVE) equivalents were performed according to the formula by Parr and Spears, ²² modified by Hubbard et al. ²³  

Oxygen saturation measurements were performed using the “oxygen tool” of the vessel map system (IMEDOS) described previously. ¹⁵ Briefly, fundus images (fundus camera FF 450; Carl Zeiss Meditec AG, Jena, Germany, and digital camera KY-F75; JVC Inc., Yokohama, Japan) were taken in a 30° field using a customized dual bandpass filter (transmission bands at 548 and 610 nm, bandwidth 10 nm each, fitting the green and red camera channel). Optical densities of the vessels were measured as the logarithmic ratio of the fundus reflection at the vessel and besides the vessel. To exclude specular reflex from the vessel, pixels with a reflection above 20% over the mean value were excluded. The ratio of the optical densities at 610 nm to that at the isosbestic wavelengths 548 nm is proportional to the vessel hemoglobin oxygen saturation ¹³ after compensation for vessel diameter and fundus pigmentation. ¹⁵ A linear relationship between the optical density ratio and a relative oxygen saturation measure was established by calibration. ¹⁵ Using this technique, the oxygen saturation was measured in all vessels in a peripapillary annulus with an inner radius of 1 and an outer radius of 1.5 disc diameters, and averaged over all arterioles and venules, respectively. Typically, about 20 single measurements in each of 6–8 arterioles and venules were averaged. Full field (30°) luminance flicker stimulation of the retina was induced by inserting an electro-optical chopper into the illumination path of the fundus camera. This provided a rectangular wave irradiation with a frequency of 12.5 Hz and a bright-to-dark contrast >25:1. ²⁴ All measurements were taken before as well as 90 seconds after onset of the flicker. 

Group mean values were compared by Student's t-test if Gaussian distribution of the data was confirmed by Kolmogorov-Smirnov test (SPSS 19.0; SPSS Inc., Chicago, IL). Otherwise, the Mann-Whitney U test was used. 

Results

Flicker light stimulation of neuronal activity induced highly significant arterial vasodilatation by 4.7 ± 8.4% (mean value ± SD, P = 0,004) as well as venous dilatation by 8.7 ± 5.2% (P < 0.0005) in the control group (Fig. 1). In patients with diabetes mellitus, we found virtually no arterial dilatation (0.6 ± 6.6%, P = 0.699) and non-significant venous dilatation was noted in 2.7 ± 6.1% (P = 0.077). The arterial SO₂ remained unchanged at 97.6% (SD between 3.9% and 4.5%) during flicker stimulation in both groups. The venous SO₂ increased by 4.2 ± 3.5% from 66.6 ± 5.0% to 70.8 ± 5.3% (P < 0.0005) in the controls, and by 2.0 ± 2.4% from 68.8 ± 7.3% to 70.8 ± 7.3% (P = 0.003) in the patients group (Fig. 2). Although the mean increase by 2.0 ± 2.4% in the diabetic group still was highly significant, it was significantly smaller (P = 0.034) than that in the controls (4.2 ± 3.5%). The increase of CRVE by the flicker was significantly different between the groups (P < 0.0005), whereas the difference was not significant for the CRAE (P = 0.061). Interestingly, the change of the venous SO₂ correlated significantly (Pearson's correlation coefficient 0.355, P for the trend 0.027) with the arterial vasodilatation (Fig. 3). Furthermore, there was a trend towards lower increase of venous SO₂ with increasing BMI (Fig. 4, correlation coefficient −0.326, P = 0.06). There were no differences between patients with types 1 and 2 diabetes in the SO₂ values, as well as CRAE and their change during flicker. The venous dilation was larger in type 2 than in type 1 diabetes (2.6% vs. 1.8%) with borderline significance (P = 0.045, Mann-Whitney U test). None of the retinal vessel diameter or SO₂ parameters correlated with blood glucose or hemoglobin A1C. In the patients but not in the controls, the venous vasodilatation (change of CRVE) correlated with MAP (P = 0.001). No further correlation of any vascular parameter with the blood pressure was found. 

Figure 1. 

 

Increase of CRAE and CRVE by retinal flicker light stimulation for 90 seconds.

Figure 2. 

 

Venous oxygen saturation before and during retinal flicker light stimulation.

Figure 3. 

 

Change of venous oxygen saturation versus arterial vasodilatation.

Figure 4. 

 

Change of venous oxygen saturation versus body mass index.

Discussion

To the best of our knowledge, this is the first study investigating retinal vessel hemoglobin oxygenation changes under flicker light stimulation in patients with diabetes mellitus. Investigations in young healthy subjects showed an increase of the venous oxygen concentration by 4% along with arterial as well as venous vasodilatation by the flicker. ¹⁸ This is in good agreement with the increase in the control group of elderly healthy subjects (4.2%) found in our study. 

Our study did not measure blood flow. Blood flow measurement in single vessels is possible by laser Doppler technique; ²⁵ however, the measurement suffers from ambiguities regarding the angle between the vessel and the laser beam as well as the turbulent or laminar characteristics of the flow. ²⁶ According to the Hagen-Poiseuille law, at constant blood viscosity, the flow is proportional to the fourth power of the vessel diameter and the blood pressure difference along the vessel segment investigated. As the blood pressure is not altered by the flicker stimulation, ²⁷ we suggest vessel diameter changes to be a good surrogate parameter for blood flow changes. Thus, from the vasodilatation an increase of the retinal blood flow may be concluded, which might be necessary to enhance the oxygen diffusion from the vessels to the tissue by an increase of the concentration gradient. This may suggest a sufficient oxygen supply to the tissue to be a driving force of retinal blood flow regulation trying to maintain a constant tissue pO₂ by an increase of the intravascular pO₂ as a reaction to an augmented oxygen demand of the inner retina upon physiologic stimulation. The correlation between venous SO₂ increase and arterial vasodilatation found in our study gives further support for the hypothesized mechanism of blood flow regulation. Apparently, this regulation is disturbed in diabetic retinopathy. This assumption is supported by the diminished vascular diameter response to the flicker as well as venous SO₂ increase in the patients compared to the controls. 

This finding is corroborated by a variety of studies. Direct measurements of retinal vessel diameters by the Dynamic Vessel Analyser all showed a reduced response to flicker light in patients with diabetes. ^4–6,28 Using a bolus injection of hydrogen and a polarographic measurement method, Cringle et al. showed increased retinal blood flow in streptozotocin-induced diabetes in rats. ²⁹ In the same animal model, Alder et al. found a markedly decreased arteriovenous pO₂ difference compared to control rats, but no difference in the vitreous oxygen tension. ¹⁰ However, the vitreous pO₂ increase upon pure oxygen inhalation, measured by magnetic resonance imaging, was higher in patients with type 1 diabetes than in controls. ³⁰ This may be explained by a dysfunction of the blood flow regulation, which is unable to reduce the blood flow sufficiently in case of hyperoxia. Measuring the pO₂ by an oxygen-quenchable phosphorescent porphyrin probe, Blair et al. found an increase of the arterial pO₂ as well as the arteriovenous pO₂ difference in rats, which was abolished by making the rats diabetic with streptozotocin. ³¹ Taken together, all these data indicate a disturbed blood flow regulation in diabetic retinopathy. One reason for that might be a dysfunction of the vascular endothelium. 

On the other hand, venous SO₂ is known generally to be increased in diabetic retinopathy (Hardarson SH, Stefansson E. IOVS 2011: ARVO E-Abstract 1275). ¹⁶ This may be explained by a reduced retinal blood passage time due to shunt vessels, ¹⁷ by a decreased oxygen diffusion through the diabetic vessel wall resulting from a thickening of its basal membrane, or to a higher oxygen affinity of the hemoglobin A1c. However, in agreement with findings of increased vessel diameters ^32,33 and blood flow ³⁴ in diabetes, it also may indicate that blood flow regulation, even in rest, is at the upper limit of its control range in the patients to prevent tissue hypoxia. This also would explain why hardly any further increase of the SO₂ is seen in the patients during flicker stimulation. Interestingly, the flicker response of the venous SO₂ increase was correlated inversely with the BMI, showing that a high BMI may be a risk factor for the distorted regulation of retinal oxygen supply in diabetes. 

Although glucose uptake by the cells is affected in different ways in types 1 and 2 diabetes (insulin deficiency or insulin resistance, respectively), the major impact of the disease on retinal vessels seems to result from long-term hyperglycemia. ¹ This might be the reason for similar responses of vessel diameters and SO₂ on flicker stimulation in both types of diabetes. Insulin itself may dilate vessels by endothelium-dependent NO secretion. ³⁵ This may be a confounding factor in our experiments. We had no control on the insulin concentration during the measurements. However, as all patients with type 1 and 11 of 14 patients with type 2 diabetes were under insulin therapy, we do not assume dramatic changes of the insulin concentration during the experiments. This does not rule out the influence of insulin, but the vasodilatation observed under flicker stimulation seems to reflect neuron-vascular coupling as mechanism of blood flow control rather than changes in the insulin concentration. 

Dorner et al. showed a decrease of the flicker-induced vasodilatation under hyperglycemia in young healthy subjects. ³⁶ In contrast, we did not find a correlation of vasodilatation or changes in SO₂ and blood glucose concentration. Whereas Dorner et al. increased blood glucose in an interventional study up to the 3-fold of normoglycemia using hyperglycemic insulin clamps, ³⁶ glucose concentration was relatively well controlled in our patients (mean 10.4 mmol/mL postprandially). This might be the reason for the independence of vessel parameters and blood glucose in our measurements. 

The major limitation of our study is the lack of direct measurement of the blood flow. A further limitation is the large variation of the baseline CRAE, CRVE, and venous oxygen saturation, which may be attributed to inter-individual differences. Thus, we restricted our investigation to the changes of vessel diameter and oxygenation. That way, each subject served as his or her own control. 

In conclusion, the combination of retinal vessel oximetry with flicker light stimulation of neuronal activity is a novel technique for the investigation of retinal metabolic demand. Oxygen supply as well as consumption can be estimated from the vascular hemoglobin saturation. In diabetic retinopathy, a decreased change of venous SO₂, as well as a decreased vasodilatation upon flicker possibly indicates a disturbance of the regulation of blood flow and oxygen supply. Further investigations are needed to clarify the underlying pathologic mechanisms. 

Footnotes

 Disclosure: M. Hammer, P; T. Heller, None; S. Jentsch, None; J. Dawczynski, None; D. Schweitzer, None; S. Peters, None; K.-U. Schmidtke, None; U.-A. Müller, None