Graefe's Archive for Clinical and Experimental Ophthalmology

, Volume 247, Issue 8, pp 1025–1030

Diabetic patients with retinopathy show increased retinal venous oxygen saturation

Authors

    • Department of OphthalmologyUniversity of Jena
  • Walthard Vilser
    • Imedos GmbH
  • Thomas Riemer
    • Imedos GmbH
  • Aleksandra Mandecka
    • Department of Internal Medicine IIIUniversity of Jena
  • Dietrich Schweitzer
    • Department of OphthalmologyUniversity of Jena
  • Uta Kühn
    • Department of OphthalmologyUniversity of Jena
  • Jens Dawczynski
    • Department of OphthalmologyUniversity of Jena
  • Fanny Liemt
    • University of Applied Sciences Jena
  • Jürgen Strobel
    • Department of OphthalmologyUniversity of Jena
Retinal Disorders

DOI: 10.1007/s00417-009-1078-6

Cite this article as:
Hammer, M., Vilser, W., Riemer, T. et al. Graefes Arch Clin Exp Ophthalmol (2009) 247: 1025. doi:10.1007/s00417-009-1078-6
  • 298 Views

Abstract

Background

Longstanding diabetes mellitus results in a disturbed microcirculation. A new imaging oximeter was used to investigate the effect of this disturbance on retinal vessel oxygen saturation.

Methods

The haemoglobin oxygen saturation was measured in the retinal arterioles and venules of 41 diabetic patients (65 ± 12.3 years) with mild non-proliferative through proliferative diabetic retinopathy (DR). Twelve individuals (61.3 ± 6.2 years, mean ± standard deviation) without systemic or ocular disease were investigated as controls. Measurements were taken by an imaging oximeter (oxygen module by Imedos GmbH, Jena). This technique is based on the proportionality of the oxygen saturation and ratio of the optical density of the vessel at two wavelengths (548 nm and 610 nm).

Results

Whereas there were no significant differences in the arterial oxygen saturation between controls and diabetic retinopathy at any stage, the venous oxygen saturation increased in diabetic patients with the severity of the retinopathy: controls 63 ± 5%, mild non-proliferative DR 69 ± 7%, moderate non-proliferative DR 70 ± 5%, severe non-proliferative DR, 75 ± 5%, and proliferative DR 75 ± 8%.

Conclusions

The increase of retinal vessel oxygen saturation in diabetic retinopathy points to a diabetic microvascular alteration. This may be due to occlusions and obliterations in the capillary bead and the formation of arterio-venous shunt vessels. On the other hand, hyperglycaemia-induced endothelial dysfunction, with subsequent suppression of the endothelial NO-synthase and disturbance of the vascular auto-regulation, may contribute to retinal tissue hypoxia.

Keywords

Diabetic retinopathyBlood flowOxygen saturation

Introduction

Retinal oxygen tone is tightly controlled by autoregulation of the vessel diameters [29]. Any imbalance in the autoregulation may result in a disturbance of the microcirculation and, finally, in diabetic microangiopathies which can be critical for the maintenance of retinal structure and function by capillary occlusion and poor distribution of blood to the retina [7]. Diabetic retinopathy (DR) seems to be linked closely to disturbances in the vascular system. Thus, a variety of studies have investigated the retinal blood flow, i.e. the product of flow velocity and vessel cross section. The results, however, have been somewhat contradictory: Compared to healthy control groups, elevated [17, 18, 28, 36], reduced [46], or unchanged [13, 15, 16, 20] flow values were found in diabetics with no or non-proliferative retinopathy. In proliferative retinopathy, however, laser treatment reduces retinal blood flow [14, 19, 27, 28]. This may reflect either a reduced oxygen demand by the coagulated tissue, or an increase of oxygen diffusion from the choroid to the inner retina, or both [37]. Taken together, these investigations suggest that the development of retinopathy may be linked to the partial pressure of oxygen (pO2) in the outer retina [1, 10, 25, 32, 34]. This can be measured with oxygen-sensitive electrodes or fiberoptic probes; this, however, is an invasive procedure not suitable for clinical diagnostics. Thus, most studies have been done in diabetic animal models, which, however, do not display retinal pathologies comparable to DR in humans. An investigation in patients with proliferative DR undergoing vitrectomy revealed decreased vitreal pO2 compared to non-diabetics [24]. Stefansson et al. showed an increased pO2 over areas which underwent photocoagulation, showing an improvement of oxygen delivery to the retina by the therapy [33]. Therefore, the investigation of oxygen supply may be indicative as a diagnostic marker as well as for therapy monitoring in diabetic retinopathy.

Retinal vessel oximetry [3, 21, 22, 30] provides a non-invasive method for the measurement of haemoglobin oxygenation. Simultaneous measurements in arterioles and venules, together with measurements of the blood flow, provide us with detailed information on the amount of oxygen delivered to the retina. Thus, we determined the oxygen saturation of the retinal vessels in diabetics at different stages of retinopathy, as well as in an age-matched healthy control group.

Methods

Retinal vessel oxygen saturation measurements were taken in 41 patients suffering from type-II diabetes with retinopathy and in 12 age-matched (61.3 ± 6.2 years, mean ± standard deviation) healthy controls. Diabetic patients were classified [11] as mild non-proliferative retinopathy (six patients, mean age: 68 ± 11.4 years), moderate non-proliferative retinopathy (12 patients, mean age: 65.8 ± 10 years), severe non-proliferative retinopathy (12 patients, mean age: 62.8 ± 17.1 years), or proliferative retinopathy (11 patients, mean age: 62.5 ± 12 years). Generally, the left eye was investigated. The right eye was chosen only if the image quality of the left eye was poor due to cataract or other conditions. The study was approved by the local institutional review board, and patients were enrolled after informed consent was given. The subject’s pupils were dilated with Tropicamide (Pharma Stulln GmbH, Stulln, Germany) prior to the investigation.

Oxygen saturation measurements were performed using the “oxygen tool” of the vessel map system (Imedos GmbH, Jena, Germany) described elsewhere [21]. Briefly, fundus images (fundus camera FF 450, Carl Zeiss Meditec AG, Jena, Germany, digital camera KY-F75, JVC Inc., Yokohama, Japan) were taken in a 30-degree field using a customised dual bandpass filter (transmission bands at 548 nm and 610 nm, bandwidth 10 nm each). This optical set-up provides an image at an isosbestic wavelength at the green camera channel and one image at a wavelength giving optimal contrast between oxygenated and de-oxygenated haemoglobin at the red camera channel. Optical densities of the vessels were measured as the logarithmic ratio of the fundus reflection at the vessel and alongside the vessel. In order to exclude specular reflex from the vessel, pixels with a reflection greater than 20% over the mean value were excluded. The ratio of the optical densities at 610 nm to that at the isosbestic wavelength 548 nm is proportional to the vessel haemoglobin oxygen saturation [3] after compensation for vessel diameter and fundus pigmentation [21]. A linear relationship between the optical density ratio and a relative oxygen saturation measure was established by calibration [21]. Vessel tracking and calculation of the oxygen saturation is done by the software of the device automatically. The reproducibility of the measurement was shown to be 2.5% in arteries and 3.25% in veins (mean standard deviation of repeated measurements) [21]. Using this technique, the oxygen saturation was measured in all vessels in a peri-papillary annulus with an inner radius of 1 and an outer radius of 1.5 disc diameters (see Fig. 1) and averaged over all arterioles and venules respectively. Typically, 6-8 arterioles and venules and about 20 single measurements along each vessel were averaged. The mean values of arterial and venous oxygen saturation were compared between the groups of patients with different stage of diabetic retinopathy by variance analysis (ANOVA) and Tukey’s post-hoc test (SPSS 15.0.1, SPSS Inc., Chicago, USA).
https://static-content.springer.com/image/art%3A10.1007%2Fs00417-009-1078-6/MediaObjects/417_2009_1078_Fig1_HTML.gif
Fig. 1

Pseudocolour representation of retinal vessel haemoglobin oxygenation. Right: healthy control subject, aged 60 years. Left: patient suffering from moderate non-proliferative DR, aged 70 years

Results

The system for retinal vessel oximetry used provides us with colour-coded maps of oxygen saturation as well as with numerical data. Figure 1 shows saturation maps of a healthy subject (left) and a patient suffering from diabetic retinopathy (right). Arterial haemoglobin oxygenation was always close to 100%. In the healthy subject, blue colour of branch venules indicate oxygen saturation of some 50%, whereas venules, supplying the macula, reveal saturations clearly higher than 60%. In diabetic retinopathy, all venules show increased oxygen saturations, indicated by the green to orange pseudo-colour.

A survey on the mean values and standard deviation of the venous oxygen saturation is given in Fig. 2a. The saturation in venules of the healthy controls (N = 12) was found to be 63 ± 5%. In diabetics, we found 69 ± 7% (mild non-proliferative DR, N = 6), 70 ± 5% (moderate non-proliferative DR, N = 12), 75 ± 5% (severe non-proliferative DR, N = 12), and 75 ± 8% (proliferative DR, N = 11). The difference to the control value was statistically significant (p < 0.05) for moderate and severe non-proliferative DR as well as for proliferative DR. The overall p-value for the ANOVA was p < 0.0005; the p-values for pair-wise comparison of all groups are given in Table 1. There was no significant difference in the arterial oxygen saturation between the groups, but a slight trend to increase with increasing severity of the DR: controls 97 ± 4%, mild non-proliferative DR 98 ± 7%, moderate non-proliferative DR 98 ± 3%, severe non-proliferative DR 100 ± 5%, and proliferative DR 101 ± 4% respectively. Calculating the individual differences between arterial and venous saturation (Fig. 2b), we found statistically significant differences (see Table 1) between the controls and all groups of DR patients but mild non-proliferative DR: controls 34.0 ± 7%, mild non-proliferative DR 29 ± 5%, moderate non-proliferative DR 27 ± 3%, severe non-proliferative DR 25 ± 4%, and proliferative DR 25 ± 6%.
https://static-content.springer.com/image/art%3A10.1007%2Fs00417-009-1078-6/MediaObjects/417_2009_1078_Fig2_HTML.gif
Fig. 2

Mean values and standard deviations of venous oxygen saturation (a) and arterio-venous oxygen saturation differences (b) versus stage of diabetic retinopathy. * indicates statistical significant (p < 0.05) difference to the controls, ** indicates highly significant (p < 0.001) differences

Table 1

P-values for pair-wise comparison of patient groups with Tukey’s post-hoc test in ANOVA (NPDR = non-proliferative diabetic retinopathy)

 

Controls

Mild NPDR

Moderate NPDR

Severe NPDR

Proliferative DR

Controls

 

0.290

0.046

0.000

0.000

Mild NPDR

0.245

 

0.996

0.305

0.280

Moderate NPDR

0.012

0.958

 

0.307

0.280

Severe NPDR

0.001

0.558

0.848

 

1.000

Proliferative DR

0.001

0.627

0.901

1.000

 

Upper right part of the table, values for venous oxygen saturation; lower left part of the table, values for arterio-venous saturation differences

Discussion

We found an increase in the venous haemoglobin oxygenation with the progression of diabetic retinopathy up to the stage of severe non-proliferative DR. There was no further increase in proliferative DR. It can not be stated conclusively whether this is an effect of autoregulatory arterial vasoconstriction secondary to photocoagulation or not. In our patient with proliferative DR, we found higher venous oxygen saturations in subjects not yet treated (78 ± 8%, N = 6) than in those who underwent photocoagulation previously (73 ± 9%, N = 5). The difference, however, was not statistically significant (p = 0.352). The venous oxygen saturation in patients with mild non-proliferative DR was 6% higher than that in the healthy controls. Although this was the highest difference between two successive groups of increasing severity of the disease, indicating a clear trend, it was not significant. One reason for that may be the low number of cases in the mild non-proliferative DR group. In general, our findings are in good agreement with that of Hardarson et al. [23], who reported a saturation of 60% in healthy subjects and of 67% in non-proliferative DR irrespective of the stage of the retinopathy. The arterio-venous oxygen saturation differences in that study were the same as in ours for the diabetic group (28%), but somewhat lower for the control group (32%). Schweitzer et al. [31] found non-significant increase of venous oxygen saturation and decrease of arterio-venous oxygen saturation difference in type I diabetics. They, however, included only patients with mild or moderate non-proliferative retinopathy, and they used a different oximetry technique based on spectrally resolved fundus reflectometry, which enabled measurements at single points at the vessels only and did not allow for averaging over the vascular tree. The main reason for the lack of significance in this paper, however, may be inclusion of patients in the age of 19 up to 71 years. These authors reported an age-dependence of the oxygen saturation in patients with diabetic retinopathy. This necessarily increases the standard deviation, possibly abolishing the significance of the difference between diabetics and controls.

At a first glance, the increased venous oxygenation reported here may contradict the hypoxia of the retina found in experimental [25] as well as clinical studies [24, 33]. Enhanced venous saturation (at constant flow rate), however, means a reduced oxygen release to the tissue in the capillary bed, finally resulting in tissue hypoxia. Various mechanisms may contribute to the decreased oxygen delivery: from anatomical studies, capillary occlusion and obliteration in conjunction with the formation of shunt vessels in the diabetic retina are known [7]. This results in areas with decreased capillary density and, thus, hypoxia. On the other hand, there may be a faster blood flow through the shunts. This might result in a shortened arterio-venous passage time and, subsequently, in a reduction of oxygen exchange. That, furthermore, would correspond with alterations in blood flow velocities as found by various authors. Depending on the extent of the obliteration and closure of capillaries as well as the resistance of the capillary shunts, the velocity in the arterioles may be increased [9, 12] or decreased [2, 4]. Thus, the elevation of the venous oxygen saturation, which is more closely related to the retinal metabolism and refers to the oxygen supply and consumption of the retina, might be more indicative of diabetic changes of the capillary system than the blood flow velocity. Better, however, would be the simultaneous measurement of blood flow and haemoglobin oxygenation. This could give absolute values of oxygen delivered to the tissue. Without information on blood flow, it is difficult to decide whether changes in the vascular oxygen saturation are caused by haemodynamic changes or by the rate of oxygen exchange with the tissue. This, however, would require the availability of three techniques in one lab: the oximetry, the measurement of vessel diameters (e.g. with Retinal Vessel Analyser), and the measurement of the blood flow velocity (e.g. laser Doppler velocimetry). That the last of these was not available is a limitation of the current study. Vessel diameters were measured along with the oxygen saturation; however, this did not show significant differences between the patient groups.

In addition to vascular damage, oxygen delivery to the tissue may be influenced by rheological parameters of the blood and by its content of glycated haemoglobin (HbA1c) which has a ten-fold higher affinity to oxygen than normal haemoglobin [8]. Although in well-controlled diabetes HbA1c is only a few percent higher than normal, this, eventually, may explain the slightly higher oxygenation values in arterioles. Furthermore, hyperglycaemia may accelerate protein glycation. The accumulation of advanced glycation end products (AGEs) may impair cell function and survival of pericytes, affecting the control of blood flow regulation, but it may also trigger the expression of vascular endothelial growth factor, which is a key player in inner blood retinal barrier dysfunction and neo-vascularisation [35]. AGE immunoreactivity, however, was also found to be related to retinal capillary basement membrane thickening. Thus, subsequently, glycation may decrease the diffusion rate of oxygen out of the vessels [35]. On the other hand, hyperglycaemia may affect the vascular endothelial function, resulting in a decreased release of nitric oxide (NO) by the endothelial NO-synthase. This results in an impaired auto-regulation of the retinal microcirculation [26], which may contribute to retinal hypoxia. Correspondence of increased venous oxygen saturation and diminished capability of blood flow auto-regulation, however, needs further investigation. It is a clear limitation of the current study that data on blood glucose, HbA1c, and blood pressure were not available.

Possibly the best discrimination between patients with DR and controls is provided by the individual arterio-venous oxygen saturation difference, which decreases by 7% in mild non-proliferative DR compared to controls, although significance is lacking. This may be due to the small number of subjects with mild non-proliferative DR. The arterio-venous oxygen saturation difference may indicate very early micro-angiopathy. This could contribute to diagnosis of vascular complications in diabetes mellitus before a retinopathy is clinically evident.

In conclusion, we were able to show that the venous haemoglobin oxygen saturation in DR is increased significantly for moderate non-proliferative DR and more severe stages, and non-significantly in mild non-proliferative DR. Although detailed investigations considering additional clinical parameters such as blood glucose, HbA1c, and rheological factors are necessary, this increase indicates deterioration in the microvascular system, possibly resulting in tissue hypoxia. The arterio-venous oxygen saturation difference was found to be reduced, with statistical significance even in the earliest stage of retinopathy. Whether this parameter may be a predictive marker for the development of diabetic complications has to be addressed in further investigations.

Acknowledgements

The authors are grateful to Dr. Einar Stefansson for valuable discussion.

Copyright information

© Springer-Verlag 2009