Diabetic retinopathy (DR) is the leading cause of blindness in adults younger than 70 years in the Western world.1 It is estimated that 1 in 29 Americans aged 40 years and older has DR (4.1 million persons) and 1 in 132 persons has vision-threatening DR.2 Despite the fact that DR is a common complication of diabetes, many cases are detected only at a late phase where visual acuity is impaired and some irreversible retinal damage has occurred.3 The earliest detectable changes in DR are the morphologic appearance of microaneurysms and capillary occlusions.4
Functional optical imaging of the eye represents a novel noninvasive diagnostic approach for the measurement of retinal blood flow velocities, mapping of vascular network structure, and for obtaining information about the oximetric5,6 and metabolic status of the retina.5-9 The flow velocity modality of the retinal function imager (RFI) used has been described in detail elsewhere.5,6 Briefly, the imaging system used identifies the motion of erythrocytes in retinal vessels by comparing images in a short movie (8–24 frames) of the retina taken under green light. Each series of 8 frames is acquired within a single short interval of <200 milliseconds. To avoid heartbeat pulsation bias of the measured velocities, the timing of a series capture is always triggered on the electrocardiogram. The distance traveled by erythrocytes in a known time is calculated for each of several retinal blood vessel segments using a cross-correlation algorithm6 and thus directly measuring their average simultaneous velocities.
Recently, we reported that flow velocities in the arteries and veins of patients with nonproliferative diabetic retinopathy were significantly lower than those of healthy subjects.10 These changes are the result of the long-term effects of hyperglycemia on blood vessel wall structure and function, as well as changes in rheologic factors observed in diabetes.11,12 Identifying blood velocity changes before clinical morphologic changes would assist in modifying treatment and follow-up and may prevent or delay vision-threatening complications.
In this study, we used the RFI to discover hemodynamic changes in patients with Type II diabetes mellitus (DM) before morphologic changes occur in the retina. We compared the blood flow velocity in the retinal vasculature of diabetic patients with no evidence of DR with that of age-matched control subjects.
Subjects and Methods
We performed a prospective cross-sectional study, recruiting patients with DM and a control group of healthy subjects. Patients in the study group were experiencing adult-onset DM with no evidence of DR in any eye. Subjects in the healthy group showed no evidence of DM. All subjects included had best-corrected visual acuity of at least 20/40 and had a refractive error within ±6.00 diopters (spherical equivalent). We excluded candidates with unclear media opacity or poorly dilating pupils, as well as individuals with signs of any ophthalmic pathology or who had undergone eye trauma or surgery within 6 months before recruitment.
All participants underwent a comprehensive ophthalmic evaluation that included medical history, assessments of best-corrected visual acuity and refraction, and slit-lamp examination before and after pupillary dilation. Finally, 3 sets of images of 20° centered on the fovea were obtained. Pupils were dilated with 1% tropicamide and 2.5% phenylephrine. When eligible, both eyes of a subject were included. Systemic blood pressure (BP) measurements were measured in all subjects before the functional imaging. Heart rate measurements were recorded by the system in parallel to retinal blood flow velocity assessment. In the diabetic group, glucose was measured at the day of examination and glycosylated hemoglobin (HbA1C) levels were recorded from the patients' charts in the same week. Patients' weights and heights were recorded, and body mass index was calculated (weight/height2).
The study was approved by the Institutional Review Board/Ethics Committee of the Tel-Aviv Sourasky Medical Center and adhered to the regulations of the Declaration of Helsinki and Health Insurance Portability and Accountability Act. A written informed consent was obtained from all participants.
The Imaging System
The RFI 3005 (Optical Imaging Ltd, Rehovot, Israel) is using standard fundus camera optics with extended functions provided by the addition of a stroboscopic flash lamp system and a fast (100 Hz) high-resolution (1 megapixel) digital camera. The system produces 8 flashes of illumination at frequent computer-controlled intervals (typically 17.5 milliseconds between flashes). For the blood flow velocity operating mode, the illumination filter is a green (“red-free”) interference filter, with transmission centered at 548 nm and a bandwidth of 17 nm. To control the effect of heart pulsation phase on flow velocity measurements, a probe is attached to the subject's finger or earlobe, allowing image acquisition to be synchronized to a selected phase of the patient's pulse pattern. For each patient, the velocity is always measured at the same certain offset a fixed fraction (67% after the beginning of the systole) of the heartbeat cycle. The exposure intervals of the digital camera are synchronized to the flash discharge. The digital pictures are captured, stored, and processed by differential imaging9 that directly detects moving erythrocytes in retinal vessels as small as 5 μm in width. The measured velocity (in millimeters per second) in secondary and tertiary branches of arteries and veins is available in table form and by segment velocity labels superimposed on the fundus image as depicted in Figure 1.
Vessel segments were assigned relative densitometric scores that reflect vessel width, allowing subdivision of vessel segments into 2 categories: between 5 μm and 10 μm (“small”) and between 10 μm and 15 μm (“large”). Because these width measurements are not direct, the actual width values are not used but only their category. We checked the accuracy of vessel categorization in five healthy volunteers, repeating each measurement three times over a short time. On average, in 87% of segments, the same category was repeatedly assigned without mistakes.
Each RFI session was repeated three times for each area measured. The coefficient of variation (standard deviation/mean) of the measured velocity in the three series was calculated for each segment. Segments for which the coefficient of variation exceeded 45% were excluded from the analysis. The average deviation from the mean for all the segments was <8.5% of the measured velocity.
Velocity can be calculated only if each series of images is of high quality and was well in focus. Therefore, image quality was evaluated in two ways. Optical resolution, light intensity, and focus of the image were assessed subjectively by one investigator (O.P.). Only those images in which visible flow was seen on most selected vessels were accepted. In addition, images were excluded if >33% of the segments were excluded because of the high coefficient of variation between series, a condition typically found in segments that were near the edge of the imaged region, or were unfocused or poorly illuminated.
Data were collected on a statistical software database (JMP software; SAS Institute, Cary, NC) and analyzed with the aid of SPSS software (SPSS, Inc, Chicago, IL). The characteristics of the study population were compared by Student's t-test for continuous parameters and by chi-square test for categorical parameters. Spearman correlation was used to determine associations between parameters.
Intergroup differences in blood flow velocity were assessed by a mixed linear model, taking into account the repeated measures of velocities in two eyes in some subjects. All models included adjustment for age and gender; more complex models included additional adjustments for heart rate, BP, smoking status, and existence of hypertension. Significance was set at P < 0.05.
The final study population consisted of 23 eyes of 14 DM patients and 51 eyes of 31 control subjects. The average duration of diabetes was 8.2 ± 8.2 years (range, 1–21 years). Average HbA1C level was 7.7 ± 1.7%, and average glucose level was 137.7 ± 48.3 mg/dL in the DM group. There were no significant differences between the average age, gender, systolic BP, and heart rate between the groups (Table 1). The diastolic BP was lower in the DM group (74 ± 11 vs. 82 ± 9 mmHg, P = 0.03). In the DM group, there was a higher prevalence of hypertensive subjects (64 vs. 33%, P = 0.05). However, the results show that these were not confounding factors. We did not detect differences between groups in weight, body mass index, and smoking habits.
Blood Flow Velocity
The average blood flow velocity in the arteries was 4.7 ± 1.7 mm/second in the DM group. This was significantly higher than that in the healthy subjects (4.1 ± 0.9 mm/second, P = 0.03; Table 2). As expected, in both groups, venous velocity was slower than that in the arteries. The DM group additionally had significantly increased venous velocity compared with control subjects (3.8 ± 1.2 vs. 2.9 ± 0.5 mm/second, respectively, P < 0.0001). We repeated the comparison between the DM and healthy subjects, including multiple parameters in the model (gender, age, current smoking status, hypertension, systolic BP, diastolic BP, and heart rate). We found, yet again, an increase in arterial and venous velocities in the DM patients compared with the healthy group (P = 0.045 and <0.001, respectively). The increase in venous velocity in the DM group was evident in all veins. In the arteries, the increase was evident in the smaller size vessels (5–10 μm; Figure 2).
We analyzed the study cohort data dividing the macula into quadrants. We used only quadrants with ≥3 segments of arteries or veins for the analysis. When comparing the average arterial velocity in each quadrant, we did not find a significant difference either in the healthy or DM group (P = 0.17 and 0.45 for healthy group and DM group, respectively, analysis of variance). Similarly, no difference was detected in the quadrant analysis in veins (P = 0.35 and 0.27 for healthy group and DM group, respectively, analysis of variance). Interestingly, when we analyze the data divided into halves, including this time any hemiretina with ≥3 segments of arteries or veins, we found higher arterial velocity in the nasal hemiretina compared with the temporal region in the healthy group (4.38 vs. 3.88 mm/second, P = 0.007). A similar difference was evident in the DM group (5.07 vs. 3.94 mm/second for nasal and temporal sides, respectively, P = 0.003). Venous velocity was also higher nasally than temporally in healthy subjects (3.07 vs. 2.80 mm/second, P = 0.01), but such differences were not observed in veins of DM patients. Dividing the data attitudinally to superior and inferior halves did not show any differences in both groups. Regional comparison between healthy subjects and DM patients according to vertical and horizontal halves showed significantly higher velocities in veins in all areas and in arteries only in nasal and inferior hemiretina (gender and age-adjusted model).
To examine whether the velocity in macular vessels is different from that in the periphery, we compared the velocity of similarly sized vessels in the periphery and the macula. Our sample size (eight experiments in two healthy volunteers) did not allow statistical analysis, although we saw a trend for higher arterial velocity in the macula. We calculated vessel density (the total length of visible vessels in pixels divided by the total number of pixels in the images) and found a marginally lower density in the diabetic patients compared with healthy subjects (P = 0.05).
Correlation Between Blood Flow Velocity to Physiologic and Pathologic Parameters
In the DM group, velocities in neither arteries nor veins correlated significantly with BP, while in healthy subjects, we found a significant correlation between average arterial blood flow velocity and BP (systolic BP: r = 0.3, P = 0.04, and diastolic BP: r = 0.4, P = 0.009). Such a correlation was apparent in all arterial categories (small: systolic BP: r = 0.3, P = 0.04, and diastolic BP: r = 0.5, P < 0.001; large: systolic BP: r = 0.4, P = 0.02, and diastolic BP: r = 0.3, P = 0.05). Venous velocity correlated significantly only to systolic BP in the large veins category (r = 0.3, P = 0.02).
The average heart rate did not correlate with average velocity of either the healthy or the DM group. We assessed the relationship between retinal blood flow velocity and heart rate in individual participants by attempting to correlate the heart rate recorded by the instrument in parallel with each velocity measurement. For each participant, we obtained a series of three separate paired measurements of heart rate and flow velocity. Each value was normalized by the corresponding subject's average. We found a positive correlation between the heart rate and both arterial and venous velocities in the healthy group (r = 0.4, P < 0.0001, for both arteries and veins; Figure 3A). In the DM patients, a small correlation exists only with the arterial velocity and not with the venous velocity (r = 0.4, P = 0.0008, for arteries, and r = 0.06, P = 0.6, for veins; Figure 3B).
In the DM group, we did not detect a significant correlation between velocity values of either arteries or veins and the duration of diabetes. Neither did the glucose and HbA1C levels and the body mass index correlate with velocity values in this group.
The primary result of this study is finding an increased retinal blood flow velocity in secondary–tertiary arterial and venous branches in patients with adult-onset DM and apparently normal retina compared with healthy subjects. Previous studies presented controversial results regarding the existence and nature of hemodynamic changes in patients with preretinopathy DM using different measurement instruments and modalities.13 These instruments differ from the RFI not only in the method of measuring the retinal blood flow velocity but also in the location and type of vessels studied. Intravessel measurement location also varies; some techniques measure centerline velocity (laser Doppler velocimetry, RFI, and blue-field simulation technique), while video fluorescein angiography measures flow of dye in the plasma that tends to flow in the periphery of the vessels.13 Scanning laser Doppler flowmetry (Heidelberg retina flowmeter) measures flow and velocity in capillaries and has found an increase in velocity and flow in the papillomacular region in patients with both Type I and Type II diabetes14,15 similar to our findings. Blue-light entoptic phenomenon enables measurement of retinal blood flow velocity in vessels in a similar location to the RFI but is tracing leukocyte movement and not erythrocytes. In one study, measurements in preretinopathy Type I diabetic patients showed, like in the current study, higher velocity in macular vessels16; however, another study did not find differences from healthy subjects.17 Measuring velocity in major retinal vessels, much larger in caliber than the RFI, with bidirectional laser Doppler velocimetry found an increase in the total retinal blood flow in veins of Type I diabetic patients18 but no change in velocity or flow in arteries in both Type I and Type II diabetes.19,20 Increased blood flow velocity has also been detected by shorter fluorescein mean circulation time in preretinopathy diabetic patients,21 although other studies did not detect such a change.22 In insulin-dependent DM, changes in the opposite direction were detected with this technique.23 It may be that the diabetic process have a different effect in central vessels compared with the peripheral retina or that the cause of early vascular disease in Type 2 diabetes is different than that in Type 1 diabetes.
In principle, the increase in blood velocity reported here might be secondary to immediate hyperglycemic state or because of long-standing retinal damage. About the first possibility, previous studies demonstrated increased retinal blood flow velocity during acute elevation in blood glucose (to 300 mg/dL) in diabetic patients without retinopathy.23,24 In contrast, others have shown negative correlation of actual glucose levels with blood flow.14 In our study, we did not detect a correlation between the velocity and glucose levels, probably because of comparatively well-controlled glucose levels (average, 137 ± 48.3 mg/dL). The patients' blood glucose levels were maintained comparatively low, according to modern guidelines established since the Diabetes Control and Complications Trial25,26 showed that intensive therapy reduces the risk of developing retinopathy. When we directly compared the velocity between patients with glucose level <130 mg/dL and those above that level (6 eyes of 4 patients), there was no significant difference in neither the arterial nor the venous velocity. This seems to rule out the possibility that increased blood glucose level as the sole or primary reason for the higher velocities we found.
Considering other causes, the increased velocity found in the DM group might reflect counteracted perfusion abnormalities in diabetic patient retina, stimulated, for example, by changes in blood rheologic properties or increased vascular resistance. In diabetic patients, there is increased aggregation and reduced deformability of erythrocytes, with increased plasma viscosity,11,12 translating to increased capillary resistance. Vascular resistance can result also from multiple molecular changes associated with long-term hyperglycemia and endothelial dysfunction. Many of these pathways are interrelated and may be simultaneously activated in retinal cells.27 Some known vasoconstrictor effectors are related to diabetic changes, like increased expression of endothelin-128 and overactivation of protein kinase C.13 Other vasodilatory mechanisms were identified as well, like endothelin-1 resistance, inhibition of calcium influx channel in smooth muscle cells, tissue hypoxia,29 and increased activity of nitric oxide synthase.30 In addition, in diabetes, there is increased leukocyte adhesion to endothelium, which is caused by increased expression of adhesion molecules31 and is associated with endothelial dysfunction.32 Indeed, in vivo studies found elevated levels of markers of endothelial dysfunction in patients with DR (soluble intercellular adhesion molecule-1 and soluble vascular cell adhesion molecule-1).33 However, studies mimicking retinal capillary obstruction by leukocytes did not detect an effect on retinal blood flow.32 The most physiologically plausible scenario consistent with the findings reported here is that arteries widen in response to impaired capillary perfusion, while venous diameter remains relatively constant. An increase in the arterial:venous diameter ratio is implied by our finding of a greater relative increase in venous velocity (31%) compared with arterial velocity (15%). Excluding an increase in BP, this also implies increased flow volume. Either loss of feedback linearity or capillary resistance inhomogeneity could produce this overcompensation. These changes could join a vicious cycle, according to the hemodynamic hypothesis34,35 that increased blood flow in diabetic patients induces further endothelial damage because of increased shear stress.36 The decreased vessel density in early diabetes that was found here was reported previously.37
In both groups, a higher velocity was found in nasal arteries compared with arteries in the temporal half of the macula. This can be a result of more proximal location manifesting in higher pressure and velocity. The same phenomenon was found in the veins of the healthy group but not in the diabetic group, which probably reflects the higher intersubject variability in the diabetic group related to different stages of the disease. Regional comparison between healthy and diabetic patients was significant in any hemiretina analyzed. In arteries, the regional difference between the groups was observed in the nasal and not in the temporal hemiretina. The nasal hemiretina is the faster half in both groups, which can highlight differences. The intergroup differences were also apparent in the inferior but not in the superior half, which may be related to the small region of interest when imaging with 20°, which can cause asymmetrical number of segments that increase the data variability when viewing small sections. Regional differences in velocity, as shown here, might contribute to the contradictory results obtained by different instruments. It may be that physiologic and pathologic processes affect these central vessels in a different way than the more peripheral retina.
A second independent finding of this study is that the retinal blood flow velocity values in the diabetic group were not correlated to BP, whereas in the healthy population, most values were. Although we found a reduced correlation in the diabetic group compared with the healthy group, this does not imply that a fundamental dependency is lost. One possibility is that the dependency relationship itself changes as diabetes develops, so that statistical significance is obscured by uncontrolled factors between patients, such as the progress of the disease. In addition, the correlation of blood flow velocity to heart rate was evident in the healthy group, but only arterial velocity was correlated to heart rate in the DM group. The correlation of velocity with heart rate was calculated for each subject comparing three sets of images. The primary confounding factor that remains is blood vessel diameter changes that are not directly measured. If blood flow volume increases, the velocity in veins, which are relatively static, should be more affected than that in arteries. This also explains how the correlation between venous velocity and heart rate may be lost in the diabetic group, while arterial velocity remains correlated to heart rate. The range of blood velocity variations may be larger or at least more irregular in diabetic patients, causing loss of correlation between venous velocity and heart rate in this group.
There was a higher proportion of subjects with systemic hypertension in the DM group. Nevertheless, when controlling for the effect of differences in patients' characteristics by the statistical model, the velocity comparison was consistent, showing significantly higher velocities in diabetic patients compared with control subjects. Diabetic patients, even with coexisting hypertension, were found to have similar caliber19,38 or even wider39 vessels compared with healthy subjects. Despite the higher rate of hypertensive patients, the actual diastolic BP was lower in the DM group; therefore, the same correlation observed between BP and blood flow velocity in healthy subjects could not explain the velocity rise found in the diabetic group.
In this study, we found increased velocity in preretinopathy patients compared with healthy subjects, while we previously found decreased velocity in nonproliferative diabetic retinopathy patients.10 Thus, the patient to healthy blood flow velocity relationship reverses during the development of morphologic alterations in the retina, as arteries reach the end of their compensating range, or capillary resistance assumes dominance in determining flow volume. In longitudinal studies,40,41 decreasing blood flow velocity over time was found in some but not in all diabetic patients. Extending this cross-sectional study to follow this cohort of patients over time with periodic noninvasive evaluations is warranted to study how such retinal blood flow velocity changes are related to the rate of progression and risk of blind-threatening consequences. Additionally, the RFI gives information on velocity only and not on flow volume, which is a weakness of the current study. Improvements in resolution and algorithm are under way and should allow getting volume information.
The ability to measure blood flow velocity simultaneously in a large number of vessels allows us to compare simultaneous changes in the arterial and venous compartments. Another unique characteristic of this method is the ability to measure velocity from secondary and tertiary branches of the retinal vasculature. The measurements are noninvasive, and therefore, frequent monitoring can be performed to detect transition to DR. The instrument used in the study offers measurements of the speed of blood cells moving in the smaller arterioles and venules of the retinal vasculature. It does not, however, measure the diameters of these vessels, so that the actual blood flows are not known. Nevertheless, there is clinical importance of finding a parameter with abnormal values in patients, especially in patients with no apparent retinal pathology.
To conclude, here we found an increased flow velocity through arteries and veins in the diabetic retina that otherwise appears normal. The question of whether these hemodynamic changes are an early marker of pathology in the vascular function of the eye or even contributing factors for microangiopathy warrants further explorations. Identifying such changes, and administrating modifying factors like meticulous glycemic control, might reverse these changes and prevent or defer irreversible sight-threatening changes to occur.
1. Aiello LP, Gardner TW, King GL, et al. Diabetic retinopathy
. Diabetes Care 1998;21:143–156.
2. Kempen JH, O'Colmain BJ, Leske MC, et al. The prevalence of diabetic retinopathy
among adults in the United States. Arch Ophthalmol 2004;122:552–563.
3. Aiello LM. Perspectives on diabetic retinopathy
. Am J Ophthalmol 2003;136:122–135.
4. Apple DJ, Rabb MF. Ocular Pathology: Clinical Applications and Self-Assessment. St Louis, MO: Mosby; 1985.
5. Abramoff MD, Kwon YH, Ts'o D, et al. Visual stimulus-induced changes in human near-infrared fundus reflectance. Invest Ophthalmol Vis Sci 2006;47:715–721.
6. Grinvald A, Bonhoeffer T, Vanzetta I, et al. High-resolution functional optical imaging: from the neocortex to the eye. Ophthalmol Clin North Am 2004;17:53–67.
7. Hanazono G, Tsunoda K, Shinoda K, et al. Intrinsic signal imaging
in macaque retina reveals different types of flash-induced light reflectance changes of different origins. Invest Ophthalmol Vis Sci 2007;48:2903–2912.
8. Nelson DA, Krupsky S, Pollack A, et al. Special report: noninvasive multi-parameter functional optical imaging
of the eye. Ophthalmic Surg Lasers Imaging
9. Grinvald A, Lieke E, Frostig RD, et al. Functional architecture of cortex revealed by optical imaging
of intrinsic signals. Nature 1986;324:361–364.
10. Burgansky-Eliash Z, Nelson DA, Pupko-Bar-Tal O, et al. Reduced retinal blood flow velocity
in diabetic retinopathy
demonstrated on the retinal function imager. Retina 2010;30:765–773.
11. McMillan DE. Deterioration of the microcirculation in diabetes. Diabetes 1975;24:944–957.
12. McMillan DE. Rheological and related factors in diabetic retinopathy
. Int Ophthalmol Clin 1978;18:35–53.
13. Grunwald JE, Bursell, SE. Hemodynamic changes as early markers of diabetic retinopathy
. Curr Opin Endocrinol Diabetes 1996;3:298–306.
14. Cuypers MH, Kasanardjo JS, Polak BC. Retinal blood flow changes in diabetic retinopathy
measured with the Heidelberg scanning laser Doppler flowmeter. Graefes Arch Clin Exp Ophthalmol 2000;238:935–941.
15. Ludovico J, Bernardes R, Pires I, et al. Alterations of retinal capillary blood flow in preclinical retinopathy in subjects with type 2 diabetes. Graefes Arch Clin Exp Ophthalmol 2003;241:181–186.
16. Sinclair SH. Macular retinal capillary hemodynamics in diabetic patients. Ophthalmology 1991;98:1580–1586.
17. Fallon TJ, Chowiencyzk P, Kohner EM. Measurement of retinal blood flow in diabetes by the blue-light entoptic phenomenon. Br J Ophthalmol 1986;70:43–46.
18. Grunwald JE, DuPont J, Riva CE. Retinal haemodynamics in patients with early diabetes mellitus
. Br J Ophthalmol 1996;80:327–331.
19. Guan K, Hudson C, Wong T, et al. Retinal hemodynamics in early diabetic macular edema. Diabetes 2006;55:813–818.
20. Lorenzi M, Feke GT, Cagliero E, et al. Retinal haemodynamics in individuals with well-controlled type 1 diabetes. Diabetologia 2008;51:361–364.
21. Kohner EM, Hamilton AM, Saunders SJ, et al. The retinal blood flow in diabetes. Diabetologia 1975;11:27–33.
22. Blair NP, Feke GT, Morales-Stoppello J, et al. Prolongation of the retinal mean circulation time in diabetes. Arch Ophthalmol 1982;100:764–768.
23. Bursell SE, Clermont AC, Kinsley BT, et al. Retinal blood flow changes in patients with insulin-dependent diabetes mellitus
and no diabetic retinopathy
. Invest Ophthalmol Vis Sci 1996;37:886–897.
24. Grunwald JE, Riva CE, Martin DB, et al. Effect of an insulin-induced decrease in blood glucose on the human diabetic retinal circulation. Ophthalmology 1987;94:1614–1620.
25. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus
. The Diabetes Control and Complications Trial Research Group. N Engl J Med 1993;329:977–986.
26. The effect of intensive diabetes treatment on the progression of diabetic retinopathy
in insulin-dependent diabetes mellitus
. The Diabetes Control and Complications Trial. Arch Ophthalmol 1995;113:36–51.
27. Schmetterer L, Wolzt M. Ocular blood flow and associated functional deviations in diabetic retinopathy
. Diabetologia 1999;42:387–405.
28. Takagi C, Bursell SE, Lin YW, et al. Regulation of retinal hemodynamics in diabetic rats by increased expression and action of endothelin-1. Invest Ophthalmol Vis Sci 1996;37:2504–2518.
29. Gardiner TA, Archer DB, Curtis TM, Stitt AW. Arteriolar involvement in the microvascular lesions of diabetic retinopathy
: implications for pathogenesis. Microcirculation 2007;14:25–38.
30. do Carmo A, Lopes C, Santos M, et al. Nitric oxide synthase activity and L-arginine metabolism in the retinas from streptozotocin-induced diabetic rats. Gen Pharmacol 1998;30:319–324.
31. Miyamoto K, Hiroshiba N, Tsujikawa A, Ogura Y. In vivo demonstration of increased leukocyte entrapment in retinal microcirculation of diabetic rats. Invest Ophthalmol Vis Sci 1998;39:2190–2194.
32. Abiko T, Abiko A, Clermont AC, et al. Characterization of retinal leukostasis and hemodynamics in insulin resistance and diabetes: role of oxidants and protein kinase-C activation. Diabetes 2003;52:829–837.
33. van Hecke MV, Dekker JM, Nijpels G, et al. Inflammation and endothelial dysfunction are associated with retinopathy: the Hoorn study. Diabetologia 2005;48:1300–1306.
34. Parving HH, Viberti GC, Keen H, et al. Hemodynamic factors in the genesis of diabetic microangiopathy. Metabolism 1983;32:943–949.
35. Zatz R, Brenner BM. Pathogenesis of diabetic microangiopathy. The hemodynamic view. Am J Med 1986;80:443–453.
36. Kohner EM, Patel V, Rassam SM. Role of blood flow and impaired autoregulation in the pathogenesis of diabetic retinopathy
. Diabetes 1995;44:603–607.
37. Arend O, Wolf S, Jung F, et al. Retinal microcirculation in patients with diabetes mellitus
: dynamic and morphological analysis of perifoveal capillary network. Br J Ophthalmol 1991;75:514–518.
38. Gilmore ED, Hudson C, Nrusimhadevara RK, et al. Retinal arteriolar diameter, blood velocity, and blood flow response to an isocapnic hyperoxic provocation in early sight-threatening diabetic retinopathy
. Invest Ophthalmol Vis Sci 2007;48:1744–1750.
39. Nguyen TT, Wang JJ, Sharrett AR, et al. Relationship of retinal vascular caliber with diabetes and retinopathy: the Multi-Ethnic Study of Atherosclerosis (MESA). Diabetes Care 2008;31:544–549.
40. Konno S, Feke GT, Yoshida A, et al. Retinal blood flow changes in type I diabetes. A long-term follow-up study. Invest Ophthalmol Vis Sci 1996;37:1140–1148.
41. Rimmer T, Fallon TJ, Kohner EM. Long-term follow-up of retinal blood flow in diabetes using the blue light entoptic phenomenon. Br J Ophthalmol 1989;73:1–5.