Total Retinal Volumetric Blood Flow Rate in Diabetic Patients With Poor Glycemic Control

Transcription

1 Investigative Ophthalmology & Visual Science, Vol. 33, No. 2, February 1992 Copyright Association for Research in Vision and Ophthalmology Total Retinal Volumetric Blood Flow Rate in Diabetic Patients With Poor Glycemic Control Juan E. Grunwald, Charles E. Riva, Joan Baine, and Alexander J. Brucker Total retinal volumetric blood flow rate was measured in 12 normal subjects and 18 poorly controlled diabetic patients with background diabetic retinopathy. Maximum or center-line erythrocyte velocity (V max ) was assessed by bidirectional laser Doppler velocimetry in four to five major retinal veins of one eye of each subject. Venous diameter (D) was measured from monochromatic fundus photographs. Total venous cross-section and measured total retinal volumetric blood flow in the diabetic patients were significantly larger than normal (P = and P = 0.02, respectively). A positive linear correlation was found between V max and D in normal and diabetic eyes. Volumetric blood flow rate, Q, varied with D at a power of 2.87 in normal eyes and 2.67 in diabetic eyes. Total volumetric blood flow correlated with total venous cross-section. It was found that Q in the temporal retina was significantly larger than in the nasal retina in normal subjects {P = ) and diabetic patients (P = ). A significant difference in Q was observed between the superior and inferior retina in diabetic patients (P = 0.03) but not in normal subjects. The retinal vascular regulatory response to % oxygen breathing was reduced (P = 0.019) in diabetic patients and correlated with the level of background diabetic retinopathy. A close estimate of total volumetric blood flow may be obtained from blood flow measurement in one major retinal vein and the determination of total venous cross-section. This may be important for clinical studies in which measurements of all individual retinal veins may not be feasible. Invest Ophthalmol Vis Sci 33:36-363,1992 The retina is one of the major sites of the body affected by diabetic microvascular pathology. Abnormalities of the retinal circulation and its regulatory responses 1 " 12 are known to occur in the eyes of diabetic patients with retinopathy and are probably an important factor in the development of retinal diabetic pathology. Laser Doppler velocimetry studies of retinal hemodynamics in diabetes mellitus have been done previously. 6 " 11 In most of these studies, measurements of one single retinal vessel have been obtained in each eye. Because volumetric blood flow rate is a function of the size of the vessel measured and the area perfused, large variability in blood flow between patients has been reported. 8 " 11 In our study, to overcome this problem, we determined total retinal venous volumetric blood flow rate by adding the flows measured in four to five major retinal veins in each eye. From the Department of Ophthalmology, Scheie Eye Institute, School of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania. Supported by National Institutes of Health grants EY077, EY0322, and RR-0000 (Bethesda, Maryland) and the Vivian Simkins Lasko Retinal Vascular Research Fund. Submitted for publication: November, 1990; accepted September 17, Reprint requests: Dr. Juan E. Grunwald, Scheie Eye Institute, 1 North 39th Street, Philadelphia, PA Increased blood glucose levels and large fluctuations in these levels are associated with an increased incidence of diabetic retinopathy 13 ' 1 and are known to affect retinal hemodynamics in diabetic patients. 9 To characterize these retinal hemodynamic abnormalities further, we determined total retinal venous volumetric blood flow and the vascular regulatory response to hyperoxia in a group of poorly controlled patients with diabetes mellitus and compared the results with those obtained in normal control subjects. Materials and Methods Eighteen patients with insulin-dependent diabetes mellitus (age range, yr; mean ± standard deviation, 29 ± yr) were included in this study. The duration of diabetes ranged from 6-28 yr (mean, 17 ± 7 yr). All patients had background diabetic retinopathy, an otherwise normal eye examination, and glycosylated hemoglobin (GHb) values higher than three standard deviations above the mean for nondiabetic subjects. Average GHb measured by affinity chromatography was 12. ± 3.3% (upper limit of normal range, 8%). Blood glucose was determined from finger capillary blood samples using an Accu-Check blood glucose monitor (Boehringer Mannheim, Indianapolis, IN). The average blood glucose at the time of retinal volumetric blood flow determination was 192 ± mg/dl (Table 1). 36 Downloaded From: on 09/23/18

2 No. 2 TOTAL RETINAL DLOOD FLOW IN DIABETES / Grunwold er ol 37 Table 1. Subject characteristics Normal Diabetes Mean ± SD Range Mean ± SD Range Significance* Age (yr) Disease duration (yr) Glycosylated hemoglobin (%) Blood glucose (mg/dl) Mean brachial blood pressure (mmhg) Intraocular pressure (mmhg) 32 ±7 78 ± 17 8 ± 10 1± ±6 17±7 12. ± ± 8 ± 11 1± P = = not statistically significant. Nonpaired student t-test. Excluded from the study were patients who had: (1) previous treatment with three or more daily injections of insulin or an insulin pump; (2) three or more documented episodes of diabetic ketoacidosis requiring hospitalization; (3) history of systemic hypertension or substance abuse; () obesity, denned as body weight larger than 130% of ideal body weight; () presence of intraocular disease other than diabetic retinopathy; or (6) previous laser photocoagulation treatment. Results in diabetic patients were compared with those obtained from 12 normal volunteers (age range, 19- yr; mean, 32 ± 7) whose blood glucose level at the time of retinal volumetric flow measurements was an average of 78 ± 17 mg/dl (Table 1). All eyes studied had a best-refracted visual acuity of 6/7. or better, an intraocular pressure < 21 mm Hg, and a normal slit-lamp examination. A description and statistical comparison of the characteristics of normal subjects and diabetic patients is provided in Table 1. After a detailed explanation of the procedures, all subjects were asked to sign an appropriate consent form approved by the Internal Review Board of our institution. Only one eye, chosen at random, was investigated in each subject. After pupil dilation with tropicamide 1% and phenylephrine hydrochloride 10%, a Polaroid (Cambridge, MA) color fundus photograph of the posterior fundus was obtained to localize the sites of bidirectional laser Doppler velocimetry (BLDV) measurements. These measurements of the maximum, center-line erythrocyte velocity (V max ) were obtained in four to five major retinal veins. Velocity was measured on straight portions of veins at a distance of less than 2 disc diameters from the center of the optic nerve head. We avoided sites close to venous junctions or arteriovenous crossings and those where two vessels lay close to each other. The location of the measurement site was marked on the Polaroid photograph for later reference. We usedflowmeasurements from veins instead of arteries because the minimal flow pulsatility in these vessels permitted a more accurate determination of the average velocity. 81 During the BLDV measurements, an area of the posterior retina (30 in diameter) was illuminated at a wavelength of 70 nm with a retinal irradiance of about 0.30 mw/cm 2. The levels of laser light used during the experiments were within the maximum permissible levels for extended sources. 16 Fundus photographs were taken in monochromatic light at 70 nm using a Zeiss (Oberkochen, Germany) fundus camera and Plus-X pan film (Eastman Kodak, Rochester, NY). Intraocular pressure was measured by applanation tonometry, and brachial artery blood pressure was obtained by sphygmomanometry. Volumetric blood flow rate (Q) was calculated as described previously 89 as V mean TTD 2 /, where mean blood velocity (V mean ) was calculated as C V max. A value for C equal to 1/1.6 was used 17, and the relationship between V max and V mean was assumed to be the same in normal and diabetic subjects. This assumption has been discussed previously. 8 The venous diameter at the site of BLDV measurement, D (determined from projected photographic negatives), was an average obtained from six photographs. Total venous cross-section (S T ) was calculated by adding the crosssection of all visible veins observed around the disc. Measurements of V max and D were done on a major retinal vein during room air breathing and during -6 min of breathing % oxygen at atmospheric pressure. The retinal vascular regulatory response to hyperoxia (R), defined as the percentage decrease in Q between air and % oxygen breathing, was calculated using the formula: R = (Q air - Q ox )/Q air. All measurements of D were done by one examiner, and all V max determinations were done by another one. Both examiners were masked with regard to the results obtained by the other and the clinical status of the subject measured. In the diabetic subjects, seven standard field stereo color fundus photographs were obtained as in the Early Treatment of Diabetic Retinopathy Study (ETDRS) protocol. Retinopathy was assessed in a masked fashion at the Fundus Photographic Reading Center of the University of Wisconsin. An overall reti- Downloaded From: on 09/23/18

3 38 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1992 Vol. 33 nopathy level according to the ETDRS grading protocol was assigned to the study eye of each patient. Mean brachial artery blood pressure (BP m ) was calculated as BP d + V3(BP S - BP d ), where BP S and BP d are the brachial artery systolic and diastolic pressures. Paired and unpaired two-tailed student t-tests, correlation analysis based on r 2 values obtained from regression fits of the data, and rank correlation analysis were used in the evaluation of the results. All data in which student t-tests were applied were validated for normality using the Wilk-Shapiro test. A Wilcoxon rank-sum test was used in those cases in which the data were not distributed normally. We used P < 0.0 for statistical significance. Results Table 1 shows the studied subjects' characteristics. There were no significant differences in age, BP m, or intraocular pressure between diabetic patients and normal subjects. Blood glucose determined at the time of BLDV was significantly higher than normal in diabetic patients (P = 0.001, by unpaired student t- test). Figures 1 and 2 illustrate the dependence of V max and Q on the D of each of the four to five individual veins measured in normal subjects and diabetic patients. The D ranged from ixm in normal subjects and from nm in diabetic patients. When all points shown in Figure 1 were considered as independent data points, which they were not, a positive correlation was found between V max and D in normal subjects (correlation coefficient, r = 0.7) and in dia Normals 1 Diabetics Vessel Diameter (/xm) Fig. 1. Relationship between maximum erythrocyte velocity (V max ) and venous diameter (D), based on measurements obtained from four to five largest retinal veins in eyes of normal subjects (squares) and diabetic patients (circles). Correlation coefficients, r, for the linear fit are 0.7, (linear regression equation V max = X D) for normals, and 0.37 (linear regression equation V max = X D) for diabetic patients when all points are considered as independent observations. 'Normals Diabetics Vessel Diameter (/Am) Fig. 2. Relationship between volumetric blood flow (Q) and venous diameter (D), based on measurements obtained from four to five largest retinal veins in eyes of normal subjects (squares) and diabetic patients (circles). Correlation coefficients for the power curve fit are r = 0.96 for normals, and r = 0.83 for diabetic patients when all points are considered simultaneously as independent observations. Q varies as 6.1 X 10" 6 D 287 in normals and as 1. X 10" D 267 in diabetic patients. betic patients (r = 0.37), showing that erythrocyte velocity increases with increasing D. When the relationship between V max and D was investigated in each subject separately, a positive correlation with an average r = 0.88 ± 0.12 (range, ) was seen in normal subjects and an average r = 0.71 ± 0.27 (range, ) in diabetic patients. The average correlation coefficient was not significantly lower than normal in diabetic patients (P > 0.0, by Wilcoxon rank-sum test). The correlation coefficient between Q and D based on a power curve fit was r = 0.96 for normal subjects and r = 0.83 for diabetic patients when all vessels were analyzed as independent data points (Fig. 2). When the relationship between Q and D was investigated in each subject separately, a positive correlation with an average r of 0.98 ± 0.01 (range, ) was observed in normal subjects and an average r of 0.93 ±0.06 (range, ) in diabetic patients. The average r between D and Q was significantly smaller than normal in diabetic patients (P < 0.01, by Wilcoxoh rank-sum test). Q varied as ~ 6 D 287 in normal subjects and as 1. 10" D 267 in diabetic patients. Values of total measured Q (Q M ), obtained by adding the Q of the four to five vessels in which BLDV determinations were done in each eye, are summarized in Table 2 for normal subjects and in Table 3 for diabetic patients. Average Q M in diabetic patients was approximately % larger than normal (P = 0.02, by two-tailed, unpaired student t-test; Tables 2,3). Average S T of all visible veins in diabetic patients was also Downloaded From: on 09/23/18

4 No. 2 TOTAL RETINAL DLOOD FLOW IN DIADETE / Grunwald er al 39 Table 2. Total venous cross-section (S T ), total measured volumetric blood flow (Q M ), and corrected total volumetric blood flow (Q T ) in normal subjects Subject Mean SD # Veins Q measured %S T S T (cm 2 X W~ ) QM (iil/min) Qr (id/min) %S T : Sum of cross-section of all veins from which volumetric blood flow measurements were obtained expressed as a percentage of the cross-section of all visible veins (S T ). Q T : Corrected value of Q M based on an estimation of flow rates for vessels in which volumetric flow rate was not measured. Q T = Q M /%Sr X. significantly larger than normal, by approximately 30% {P = 0.001; Tables 2, 3). The BLDV measurements of V max, and therefore also of Q, were not obtained in all visible veins in all eyes studied. The percent S T values shown in Tables 2 and 3 represent the sum of the cross-section of all veins from which volumetric blood flow measurements were obtained expressed as a percentage of the cross-section of all visible veins present in each studied eye. To include those vessels in which Q was not determined, a total volumetric blood flow rate (Q T ) was calculated based on an estimation offlowrates for the vessels in which volumetricflowrate was not measured using the formula: Q T = Q M /%S T X (Tables 2, 3). Average Q T was also significantly larger than normal in diabetic patients (P = 0.009; Tables 2, 3). Nine patients with the mildest form of retinopathy (level ) already had an average Q T of.6 ± 6.6 /Lil/min and S T of ± 1.1 cm 2-10" ; these were significantly larger than normal (P < 0.0 and P < 0.01, respectively). Significant positive correlations were observed between S T and Q T in normal subjects (r = 0.88, P = , with a linear-regression equation of Q T = S T ) and in diabetic patients (r = 0.9, P = 0.0, with a linear-regression equation of Q T = S T ). Average D, V max, and Q for the largest retinal vein in normal subjects and diabetic patients are shown in Table. No significant differences from normal in D, Vmax> or Q were observed when the measurements obtained in the largest vein in diabetic patients and normal subjects were compared. We measured R in one of the major temporal veins; it was significantly reduced in diabetic patients (P Table 3. Total venous cross-section (S T ), total measured volumetric blood flow (Q M ), and corrected total volumetric blood flow (Q T ) in diabetic patients Subject Retinopathy level #Veins Q measured %S T S T (cm 2 X 10~ ) QM (til/mm) Qr (nl/min) Mean SD Significance* P = P = P = %Sr: Sum of cross-section of all veins from which volumetric blood flow measurements were obtained expressed as a percentage of the cross-section of all visible veins (ST). Q T : Corrected value of Q M based on an estimation of flow rates for vessels in which Q was not measured. Q T = Q M /%S T X. * Significantly different from normal by nonpaired student t-test. Downloaded From: on 09/23/18

5 360 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / February 1992 Vol. 33 Table. Average venous diameter (D), maximum erythrocyte velocity (V max ), and retinal volumetric blood flow (Q) in the largest retinal vein of diabetic patients and normal subjects D( M m) V max (cm/sec) Q Oil/min) Normal (n = 12) 166 ± 13f 1.77 ± ± 2.6 = not statistically significant. * Nonpaired student t-test. tsd. Diabetes (n = 18) Significance* 179 ± 1.72 ± ±3.9 N.S. N.S. N.S. = 0.019, by unpaired student's t-test). Whereas Q decreased by ± 9% during hyperoxia in normal subjects, it decreased by only 6 ± 11% in diabetic patients. Patients with more advanced retinopathy showed smaller regulatory responses (Fig. 3), as demonstrated by the significant correlation present between R and retinopathy level (Spearman's rank correlation coefficient = 0.2, P < 0.0). Although average R for patients with mild retinopathy (level ) was smaller than normal (9 ± 11%), the difference was not statistically significant. No significant correlation was found between Q T and retinopathy level (Spearman's rank correlation coefficient = 0.01, In 16 diabetic patients, measurements of Q and R in one of the major temporal veins were repeated a few weeks later. The average percentage change in Q between the two measurements was 2.8 ± 12%. The difference in R between the two measurements (R 2 (%) - R,(%)) was 1.9 ± 7.7%. The distribution of Q in different areas of the retina is shown in Table. On average, Q was significantly larger in the temporal retina than in the nasal retina Table. Average of the sum of the venous volumetric flow rate, Q Oul/min), in the temporal, nasal, superior, and inferior regions of the retina Temporal retina Nasal retina Significancef Superior hemiretina Inferior hemiretina Significance! Normal Diabetes (n = 12) (n = 18) Significance* 2.0 ± ±6.2 P = ± ± ± ±7.0 P = ± ±6.3 P = 0.03 P = 0.00 = not statistically significant. * Nonpaired t-test comparing flow in normal subjects and diabetic patients. f Paired t-test comparing flow in the temporal and nasal or the superior and inferior retina. by 9% in normal eyes (P = ) and by 83% in the diabetic eyes (P = ). In addition, Q in the superior retina was significantly larger than that observed in the inferior retina in diabetic eyes (P = 0.03). In normal subjects, however, this difference was not statistically significant (P = 0.8). When normal subjects and diabetic patients were grouped together, Q was also significantly larger in the superior retina than in the inferior retina (P = 0.02). We attempted to estimate total volumetric blood flow rate from blood flow measurements obtained in only one major retinal vein and from the total venous cross-section using the following two procedures: (1) Q was measured in the largest vein and scaled by a factor consisting of the total venous cross-section/ largest vein cross-section (the result was defined as Qnaisest) or (2) Q was measured in the second largest vein and scaled by a factor consisting of the total venous cross-section/second largest vein cross-section (QTsecond largest)- Both Q Tlargesl and Q Ts econd largest Were On average about 1% larger than Q T both in normal and diabetic eyes (Table 6). In addition, a highly significant correlation was found between Q T and Qriargest (r J Retinopathy Level 0 Fig. 3. Relationship between the retinal vascular regulatory response to % oxygen breathing (R) and retinopathy level of the study eye. Spearman's Rank Correlation, r s = 0.2, P < 0.0. Average R in normal subjects was % ± 9%. Table 6. Average corrected total volumetric blood flow (Q T ) and estimated total volumetric blood flow rates from the largest (QriargestX an d the second largest (Q Tsecond largest) retinal vein Normal eyes Diabetic eyes Qr (nl/min) 38.1 ±6.2* 7.0 ± 9.7 \ctlargest (fil/min) 3.1 ±9..6 ± 13.0 \J-Tsecond largest (nl/min) 3. ± 8.9. ± 13.6 QTUIJCSV Estimated total volumetric blood flow from Q measured in the largest vein scaled by a factor Total Venous Cross-Section/Largest Vein Cross-Section. QTsecond iarsrai : Estimated total volumetric blood flow from Q measured in the second largest vein scaled by a factor Total Venous Cross-Section/Second Largest Vein Cross-Section. *SD. Downloaded From: on 09/23/18

6 No. 2 TOTAL RETINAL BLOOD FLOW IN DIABETES / Grunwold er ol 361 = 0.89, P = , with a linear-regression equation of Q T = Q Tlargest ) or Q Ts econd u^est (X = 0.87, P = , with a linear-regression equation of Q T = Q Tsecond lareest ). Discussion We attempted to measure total retinal venous blood flow in patients with poorly controlled diabetes mellitus and background diabetic retinopathy. Our results suggest that, in these patients, total retinal volumetric blood flow rate is significantly larger than normal by approximately 23% (Table 6). Measurements obtained in the largest retinal vein in each eye showed an average Q that was only 12% higher than normal, and this difference was not statistically significant (Table ). This was similar to the results in our previous study, 8 in which average Q obtained in a single retinal vessel of patients with background retinopathy was approximately 8% above normal (also not statistically significant). The results of our current study also show largerthan-normal veins in diabetic patients (Table 3), a finding that agreed with previous reports demonstrating vasodilatation in diabetic patients. 818 Both Q T and S T were already significantly larger than normal in the nine patients with the mildest form of retinopathy (level ). However, V max was not significantly different from normal in poorly controlled diabetic patients with background retinopathy (Table ). These results differed from those in our previous study 8 in which a decrease in average V max of 18% was observed in eyes with background diabetic retinopathy. There were important differences between the two studies that might explain the discrepancy in V max results. In our current investigation, most of the patients had mild retinopathic changes (Fig. 3); in our previous study, 8 approximately one half of the subjects had severe background retinopathy. Because V ma x seems to decrease with disease progression, 8 a more prevalent presence of severe background retinopathy could be related to the lower than normal V max value found in our previous study. In our current study, we selected only type I diabetic patients with poor glycemic control. Most of the patients had GHb levels that were three standard deviations or more higher than the mean for nondiabetic subjects. In our previous study, GHb levels were not measured, and no attempt was made to select patients specifically with poor control. It is possible that patients with poor glycemic control could have a higher V max than those with better control. This hypothesis was supported by our previous article, 9 showing that in very poorly controlled diabetic patients during marked hyperglycemia, V max is significantly above normal and that normalization of blood glucose with insulin administration results in significant decreases in Q and V max. We stress, however, that the actual blood glucose levels at the time offlowmeasurements in our current study were, on average, close to those observed in our previous one. 8 Further studies are needed to clarify the importance of the influences on the retinal circulation of chronically elevated blood glucose levels versus acutely elevated blood glucose levels. As shown by our results, total retinal volumetric blood flow was significantly larger than normal in patients with poorly controlled diabetes mellitus and background retinopathy. This increase was caused mainly by vasodilatation, which probably constitutes a regulatory response of the retinal vasculature attempting to normalize a reduced supply of nutrients or an increased concentration of metabolic waste products in the retinal tissue. The average r between Q and D in diabetic patients was significantly smaller than that observed in normal subjects. Moreover, among diabetic patients, there was also a greater heterogeneity in r values, as demonstrated by a larger standard deviation. The reason for this effect was not clear, but the changes in retinal architecture and/or blood rheology, known to occur in patients with diabetic retinopathy, might alter the normal relationship between Q and D, leading to a weaker or perhaps different association between these parameters. We found no correlation, however, between the r values and retinopathy level or disease duration, and therefore, we could not conclude that more advanced diabetic pathology is associated with a weaker correlation between Q and D. In our previous study, in which measurements were obtained in a single major retinal vein of each eye, a large variability in Q was found in diabetic (12.3 =h. iul/min) and normal (11.3 ± 3. ^1/min) eyes. 8 This variability mainly was related to the differences in D of the veins measured and the retinal areas perfused by these vessels. Because of this variability, the sensitivity of the technique to detect differences in Q between groups of patients was approximately 23%, ie, only changes larger than 23% could be demonstrated in that study. In our current study, we estimated Q T by adding the flows measured in the four to five largest retinal veins and estimating theflowsin the few vessels in which Q could not be measured. The variability in Q T between subjects was smaller than that observed for the largest retinal veins (Table ) and than seen in our previous study. 8 This intersubject variability in Q T may be explained by the variability in S T because both quantities correlate significantly in diabetic patients (P Downloaded From: on 09/23/18

7 362 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / February 1992 Vol. 33 = 0.0) and normal subjects (P = ). Although total retinal flow was significantly larger than normal in the diabetic eyes, the difference in flow in the larger retinal vein was not statistically significant. From determinations of Q obtained a few weeks apart in each of 16 diabetic subjects, we calculated that our technique had a sensitivity of approximately 10%. Thus, we could detect an average change in Q of 10% or larger {P < 0.01, by two-tailed paired student t-test). Two determinations of R done a few weeks apart in 16 diabetic patients showed an average difference in R (R 2 (%) - R,(%)) of 1.9 ± 7.7%. From these values, we determined that our technique had a sensitivity to detect average changes in R of approximately 6% (P < 0.01, by two-tailed paired student t-test), when applied as described in a group of 16 diabetic patients. These values are important for the design of experimental protocols to study the effects of disease or treatments on retinal hemodynamics. We found R was significantly reduced in diabetic patients, as reported previously. 6 In addition, the significant decrease in R with progression of background diabetic retinopathy (Fig. 3) further strengthened our previous hypothesis that the decreased vascular regulatory response to hyperoxia may be related to the degree of retinal hypoxia. 6-9 " 11 Measurements of Q T are laborious and time consuming because determinations of V max must be made separately in each individual major retinal vein. In large-scale clinical studies, it may not be possible to do all these measurements. We attempted to estimate Q T from the measurement of Q obtained in the largest or second largest retinal vein and S-r (Table 6). This estimation was close to the actual value (higher by 1%, on average), and furthermore, there was a strong correlation between these quantities, suggesting that this method may provide a rough estimation of total retinal flow when actual total flow cannot be measured. Average total retinal volumetric bloodflowsin normal subjects were close to the average of 3 ± 6.3 /zl/min reported previously 1 in seven normal eyes. Retinal volumetric bloodflowrates in individual major retinal veins were also similar to those reported previously by our laboratory. 8 These values of total volumetricflow,however, are about 0% smaller than those reported by others 19 using laser Doppler velocimetry. The exact source of this discrepancy has not been identified and probably stems from differences in the way the technique was used. The significantly larger Q observed in the temporal retina than in the nasal retina of normal subjects agreed with previous results. 119 A similar difference also was found in our study in diabetic patients (Table ). These differences can be explained by the larger area of the temporal retina. In addition, increased retinal thickness and perhaps an increased metabolic rate in the macular region also may explain this finding. By comparison with the inferior retina, the superior retina showed a significantly larger Q when normal subjects and diabetic patients were grouped together. The inferior visual field is larger than the superior visualfield. In addition, the fovea is located slightly below the center of the optic nerve head. 21 These two facts suggest that there may be a physiologic explanation for the differences in flow observed between the superior and inferior retina. Our results suggest that Q varies with D at a power of 2.87 in normal eyes and 2.67 in eyes of diabetic patients (Figs. 1, 2). The difference between these two values was not statistically significant, and therefore, we cannot conclude at this time that diabetes mellitus affects this factor. A vascular tissue in which Q varies with D at a power of 3 is the ideal system in terms of the amount of energy needed to push blood through the vessels. 22 Any change from 3 represents a departure from the ideal situation. Key words: retinal volumetric bloodflow,laser Doppler velocimetry, diabetic retinopathy, vascular regulation, erythrocyte velocity, hyperoxia Acknowledgments The authors thank Sharon Grunwald and Jim Gowan for statistical consultation and analysis of the data and Dolly A. Scott for preparation of this manuscript. References 1. Hickam JB and Sieker HO: Retinal vascular reactivity in patients with diabetes mellitus and with atherosclerosis. Circulation 22:23, Kohner EM, Hamilton AM, Saunders SJ, Sutcliffe BA, and Bulpitt CJ: The retinal blood flow in diabetes. Diabetologia 11:27, Blair NP, Feke GT, Morales-Stoppello J, Riva CE, Goger DG, Collas G, and McMeel JW: Prolongation of retinal mean circulation time in diabetes. Arch Ophthalmol :76, Sinclair SH, Grunwald JE, Riva CE, Braunstein SN, Nichols CW, and Schwartz SS: Retinal vascular autoregulation in diabetes mellitus. Ophthalmology 89:78, Yoshida A, Feke GT, Morales-Stoppello J, Collas GD, Goger DG, and McMeel JW: Retinal blood flow alterations during progression of diabetic retinopathy. Arch Ophthalmol 101:22, Grunwald JE, Riva CE, Brucker AJ, Sinclair SH, and Petrig BL: Altered retinal vascular response to % oxygen breathing in diabetes mellitus. Ophthalmology 91:17, Feke GT, Tagawa G, Yoshida A, Goger DG, Weiter J, Buzney SM, and McMeel JW: Retinal circulatory changes related to retinopathy progression in insulin-dependent diabetes mellitus. Ophthalmology 92:117, Grunwald JE, Riva CE, Sinclair SH, Brucker AJ, and Petrig BL: Laser Doppler velocimetry study of retinal circulation in diabetes mellitus. Arch Ophthalmol 10:991, Downloaded From: on 09/23/18

8 No. 2 TOTAL RETINAL DLOOD FLOW IN DIABETES / Grunwold er ol Grunwald JE, Riva CE, Martin DB, Quint AR, and Epstein PA: Effect of insulin-induced decrease in blood glucose on the human diabetic retinal circulation. Ophthalmology 9:161, Grunwald JE, Brucker AJ, Petrig BL, and Riva CE: Retinal blood flow regulation and the clinical response to panretinal photocoagulation in proliferative diabetic retinopathy. Ophthalmology 96:118, Grunwald JE, Brucker AJ, Schwartz SS, Braunstein SN, Baker L, Petrig BL, and Riva CE: Diabetic glycemic control and retinal blood flow. Diabetes 39:602, Riva CE, Petrig BL, and Grunwald JE: Retinal blood flow. In Laser Doppler Flowmetry, Shepherd AP and Oberg PA, editors. Boston, KJuwer Academic, 1990, pp Chase HP, Jackson WE, Hoops SL, Cockerham RS, Archer PG, and O'Brien D: Glucose control and the renal and retinal complications of insulin-dependent diabetes. JAMA 261:11, Klein BE, Moss SE, and Klein R: Longitudinal increase of glycemic control and diabetic retinopathy. Diabetes Care 10:273, Riva CE, Grunwald JE, Sinclair SH, and Petrig BL: Blood velocity and volumetricflowrate in human retinal vessels. Invest Ophthalmol Vis Sci 26:112, Sliney DH and Freasier BL: Evaluation of optical radiation hazards. Applied Optics 12:1, Damon DN and Duling BR: A comparison between mean blood velocities and center-line red cell velocities as measured with a mechanical image streaking velocimeter. Microvasc Res 17:330, Skovborg F, Nielsen A, Lauritzen E, and Hartkopp D: Diameter of retinal vessels in diabetic and normal subjects. Diabetes, 18:292, Feke GT, Tagawa H, Deupree DM, Goger DG, Sebag J, and Weiter J: Bloodflowin the normal human retina. Invest Ophthalmol Vis Sci 30:8, Anderson DR: Perimetry With and Without Automation. St. Louis, CV Mosby, 1987, pp Bishop PO: Binocular vision. In Adler's Physiology of the Eye, Moses RA, editor. St. Louis, CV Mosby, 197, p Mayrovitz HN and Roy J: Microvascular bloodflow:evidence indicating a cubic dependence on arteriolar diameter. Am J Physiol2:H1013, Downloaded From: on 09/23/18