Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Choroidal Blood Flow During Isometric Exercises Charles E. Riva*-\ Patrick Titze*^ Mark Hero,* Armand Movaffaghy* and Benno L. Petrig* Purpose. To investigate the response of choroidal bloodflowin the foveal region of the human eye to increases in mean perfusion pressure (PPm = mean ophthalmic artery pressure intraocular pressure; IOP) induced by isometric exercises. Methods. Using laser-Doppler flowmetry, changes in velocity (ChBVel), number (ChBVol), and flux (ChBF) of red blood cells in the choroidal vascular system in the foveal region of the fundus were measured in both eyes of 11 normal subjects (ages 18 to 57 years) during isometric exercises. Results. During 90 seconds of squatting, PPm increased by an average of 67%, from 46 to 77 mm Hg. This resulted in a significant increase of 12% in ChBFm (the mean of ChBF during the heart cycle), mainly caused by an increase in ChBVelm. A further increase in PPm to a value approximately 85% above baseline resulted in a 40% increase in ChBFm. A significant negative correlation was found between the changes in ChBVelm and ChBVolm during squatting. Conclusions. Previous studies have demonstrated that during isometric exercise, blood pressures in the ophthalmic and brachial arteries rise in parallel. These observations and the current results indicate that an increase in PPm up to 67% induces an increase in choroidal vascular resistance that limits the increase in choroidal blood flow to approximately 12%. This regulatory process fails when PPm is further increased. Invest Ophthalmol Vis Sci. 1997; 38:2338-2343. Isometric exercises increase heart rate, arterial pressure, and sympathetic nerve activity.1"3 The effect of this type of exercises on blood flow in vessels supplying muscle groups, 4 the skin,5 or larger body regions 6 has been studied extensively. Few investigations, however, have been devoted to the ocular circulation. Results of studies in cats and monkeys7"9 have demonstrated an efficient regulation of blood flow through the optic nerve, uvea, and retina in the face of acute increases in arterial blood pressure. In humans, such a regulation has been shown in the ophthalmic artery, iris,10 and retina." The purpose of this work was to investigate the From the *lnstilut de Recherche en Ophtalmologie, Sion and f Medical School of the University of Lausanne, Switzerland. Presented at 1996 Meeting of the Association for Research in Ophthalmology and Vision, Fort Laudeidale, Florida, April 27-May 3. Supported in part by the Swiss National Science Foundation (grant #3200043157.95) and Loterie Rornande, Lousanne, Switzerland. Submitted for publication December 6, 1996; revised May 30, 1997; accepted June 3, 1997. Proprietary interest category: N. Reprint requests: Charles E. Riva, Institut de Recherche en Ophtalmologie, Av. Grand-Champsec 64, Case Postale 4168, 1950 Sion 4, Switzerland. 2338 time course and magnitude of the response of choroidal blood flow in the foveal region of the human fundus to increases in arterial blood pressure induced by the static exercise of squatting. MATERIALS AND METHODS Subjects Measurements were obtained from 22 eyes of 11 healthy male volunteers, ranging in age from 18 to 57 years. Their average systolic/diastolic brachial artery blood pressures at rest were 119/80 mm Hg. The experiment was performed first for measurements in one eye, chosen at random, and were repeated some days later for measurements in the other eye. Subjects had no history of systemic or ocular diseases, and the results of ocular examinations were normal. Spherical refraction ranged from —3 to +3 diopters. Mean intraocular pressure (IOP) was 12.4 ± 2.4 mm Hg. Subjects were asked to abstain from drinking coffee and smoking for 24 hours before the study. The pupils were Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 Copyright © Association for Research in Vision and Ophthalmology Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933419/ on 05/12/2017 2339 Choroidal Blood Flow dilated with one to two drops of 1% tropicamide. Brachial artery blood pressure was measured, using an electronic sphygmomanometer. The procedures were approved by the University of Lausanne Medical Faculty Ethical Committee and followed the tenets of the Declaration of Helsinki. Informed consent was obtained from all subjects, after the nature and possible consequences of the study were fully explained to each. Measurement of ChBF The method for measuring choroidal red blood cell flux (ChBF) in the foveal region has been published recently.12 Briefly, subjects were asked to fixate at a diode laser beam (wavelength = 811 nm, 100 fiW at the cornea), which was delivered to the eye through a fundus camera (model TRC, Tokyo, Japan). Light scattered by moving red blood cells in the tissue volume sampled by the incident laser beam was detected at the fundus image plane of the camera by an optical fiber. The diameter of the beam at the fundus of an emmetropic eye was approximately 300 fxm. Relative ChBF was determined by analyzing the laser-Doppler flowmetry signal with a NeXT computer (Redwood City, CA),13 using an algorithm based on Bonner and Nossal's photon diffusion theory.14 This algorithm is equivalent to the one implemented in commercial laser-Doppler flowmetry systems (BPM403A; Vasamedics, Minneapolis, MN; PeriFlux PF3; Perimed, Stockholm, Sweden). It provides, in addition to ChBF, a relative, continuous measurement of the velocity (ChBVel), the volume (ChBVol) of blood within the sampled tissue volume, and the DC level of the photocurrent. The latter is proportional to the total amount of light reaching the detector and is used to monitor the alignment of the camera with the eye of the subject. The algorithm also automatically excludes the Doppler signal during blinks, thereby minimizing potential artifacts. The flow parameters are related to each other through the relationship ChBF = k X ChBVel X ChBVol, wherein k represents an instrumental constant. Pulsatility of ChBF, PUCI,BF, was defined as (1 — ChBF<li:isl/ChBFsysl), where ChBFcli:isl and ChBFsysl are the values of ChBF at end diastole and peak systole, respectively. PUCI,BVCI and Puciiiwoi were defined in a similar way. For determining Pu, each flow parameter was measured every 46 ms. Serial data points were then averaged in phase with the heart pulse, which was continuously recorded. All data points within a chosen time segment, which have the same phase, were averaged together. This procedure was repeated for all phases, producing an average waveform representative of each flow parameter. For calculation of Pu of a given flow parameter, the systolic and diastolic values of this parameter were taken as the maximum and minimum of the average waveform, respectively. Isometric Exercises Squatting was used as the isometric exercise. Systolic and diastolic brachial artery blood pressures were measured by sphygmomanometry, at rest and at various times during the experiment. Mean perfusion pressure, PPin, was calculated as: PPin = 2/3(MAP) — IOP. IOP is the intraocular pressure. Mean blood pressure (MAP) = BPdiMl + 73(BPsysl-BPdiasl), where BPdiM1 and BPsysl are the blood pressures during diastole and systole, respectively. Heart rate was continuously recorded with an earlobe transducer. Intraocular pressure was measured, using a TONO-Pen XL (Mentor, MD), before, at the end of the exercise, and immediately after the end of the exercise. Before the subjects were asked to perform the squatting exercise, two baseline measurements of the flow parameters were performed, each of at least 15 seconds' duration, while the subject was sitting. Sensitivity of the Technique Sensitivity, S, of each flow p a r a m e t e r — t h a t is, the m i n i m u m c h a n g e that can be detected by the techn i q u e — w a s calculated u n d e r n o r m a l conditions, using two successive 10-second baseline (bl) recordings, a n d d u r i n g squatting, using the last two 10-second segments of squatting. T o calculate Sbi u n d e r normal conditions (baseline) for a given flow p a r a m e t e r , x, we d e t e r m i n e d the difference, A 1 = ^ b | 2 — xJbn between the m e a n of both segments in eye i. After det e r m i n i n g the standard deviation <rbi of the A' s , Sbi was obtained using the formula 2Xl00XtXab, N/nX(.Vbl in which n is the number of eyes measured and / is the two-tailed critical value of the ^-distribution for n1 degrees of freedom, at die 0.05 level of significance. The same procedure was used to calculate Ssq at the end of squatting. Statistics. Mean changes in the blood flow parameters, perfusion pressure, and flow pulsatility were assessed for significance by ±95% confidence intervals of the mean, corresponding to a significance level of 0.05. RESULTS Figure 1 shows a sample time course of ChBVel, ChBVol, and ChBF at baseline, during squatting and during the return to normal physiologic conditions. Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933419/ on 05/12/2017 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 2340 MAP(mmHg) 120/80 IOP(mmHg) 11 160/110 185/120 200/140 14 160/95 150/90 9 10 ChBVel ChBVol ChBF Time Constant n 1 sec 10 seconds of recording centered at this time. Average heart rate increased from 73 to 107 bpm. During squatting, PPm increased by as much as 67%, from 46 ± 2.6 mm Hg to 77 ± 7 mm Hg. ChBVelm and ChBFm increased with the duration of squatting. We fitted the changes in both flow parameters with a function of the form a X timeB, where a and b are parameters. A plot of log(ChBVelm-100%) and log(ChBFm-100%) versus log(time) provided significant linear correla- 130 ^ 120C J& FIGURE l. Sample time course of the flow parameters, PQ ChBVel, ChBVol, and ChBF during baseline (0 to <se>65 seconds), squatting {period between vertical, dotted lines) and U recovery from squatting (after <se>160 seconds). Time constants of the recordings and display were 0.05 seconds and 1 second, respectively. The IOP and systolic/diastolic are shown on top of the recordings. Changes in the flow parameters during the heart cycle (bottom) were obtained on the basis of approximately 20 seconds of recordings centered at 25 seconds (baseline) and at 140 seconds (end of squatting). The determination of these changes is explained 43 in Methods. ChBVel, ChBVol, and ChBF = velocity, volume, U and flux, respectively, 110 - 1009080 -40 40 80 120 160 200 240 40 80 120 160 200 240 140120- 8060 -40 0 140130 - Time constant of the recordings was 0.05 seconds. Time constant of the display was 1 second. In this subject, heart rate increased during the exercise from 64 to 110 bpm and systolic/mean/diastolic BP from 120/93/80 to 200/160/140 mm Hg. There was a 3mm Hg increase in IOP. PPm increased from 50 to 92 mm Hg (84%). Average ChBFm, in which the subscript m indicates the mean value of ChBF during the heart cycle, was determined for the 10-second segment centered at 25 seconds: The 10-second segment centered at 140 seconds showed an increased from 9.8 to 12 units (22%). This increase is mainly caused by a 25% increase in ChBVelm from 2.43 to 3.04 units. ChBVolm decreased by =^3%, from 0.35 to 0.34 units. Most of the changes in the flow parameters occurred at systolic/ diastolic blood pressures between 185/120 and 200/ 140 mm Hg. When squatting was stopped, MAP returned to baseline value within approximately 2 minutes (not shown in Fig. 1). ChBVelm and ChBFm returned to baseline in less than 15 seconds. Figure 2 demonstrates the normalized (baseline = 100%) group average of PPm and of the flow parameters, ChBVelm, ChBVolm and ChBFm, as a function of time for the 22 eyes during the first 90 seconds of exercise, the period during which all subjects were able to squat. In each subject, the value of each flow parameter at a given time was obtained by averaging 120- trill f 110- PQ U 1009080 -40 0 40 80 120 160 200 240 200 240 200180 160140120- * 100- 80 -40 0 40 80 120 * 160 * Time (sec) 2. Time course of group average of PPm, ChBVel,,,, ChBVol,,, and ChBF,,, (the mean of each flow parameter during the heart cycle) during squatting and recovery from squatting. They were obtained from 22 eyes in 11 subjects. Each data point is an average of 16 values. It represents a mean during 10 seconds of recording. Error bars are the 95% confidence limits of the mean. The initial point is the average baseline value which was normed to 100% in every subject. PPm = mean perfusion pressure; ChBVel,,,, ChBVol,,, and ChBF,,, = mean values for velocity, volume and flux, respectively, of red blood cells in the choriodal vascular system of the fovea. FIGURE Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933419/ on 05/12/2017 2341 Choroidal Blood Flow 180160w 140120- u 100- 80 80 X \ {I 100 120 140 160 180 200 P P m(°/n\ ^ ' 3. Average ChBFm versus PPm during squatting obtained from 22 eyes in 11 subjects. Dotted line: ChBFm versus PPm with no regulation (constant vascular resistance). Each data point represents an average of 24 consecutive values. Error bas are the ±95% confidence limits of the mean. For some of the points, these limits for the PPm values are within the filled circle ChBFm = mean flux in choroidal red blood cells; PPM, = mean perfusion pressure. FIGURE tion coefficients (r = 0.94, P < 0.01 for ChBVelm and r = 0.86, P < 0.05 for ChBF m ). These fits were better than linear fits. ChBVolm decreased linearly as a function of time ( r = — 0.93, P = 0.01). A significant negative correlation (r = 0.84, P < 0.05) was obtained between ChBVol,,, and ChBVelm during squatting. ChBVel,,, increased linearly (P < 0.02) as a function of PP m , but the change in ChBVolm with PPm was not significant. When squatting was stopped, ChBVelm returned rapidly to a value not significantly different from that at baseline. It occurred before PPm reached baseline. Figure 3 shows the relationship between average ChBFm and PPni, both expressed as a percentage of baseline value, using data obtained during the full duration of squatting, which in some subjects was up to 3 minutes. It thus includes higher PPm values than the graph of Figure 2. Seven or eight pairs of ChBF m PP,n data points were obtained in each eye, for a total of 168 pairs of points. These were sorted according to ascending values of PPm and divided into seven groups of 24 pairs of values. For each group, mean and 95% confidence interval of the ChBFm and PPm values were calculated. ChBFm was significantly above baseline value only at the highest average PPm reached (by analysis of variance, P < 0.003). At this PP m , IOP was 17 ± 2 mm Hg, significandy above the baseline value of 13 ± 1 mm Hg. Immediately after the end of the exercise, it dropped to 11 ± 2 mm Hg. We investigated the correlation between the changes in the flow parameters in the right eye (OD) and those in the left eye (OS). These changes were evaluated in each subject during squatting at values of PP m with increases above baseline differing by no more than 4% between eyes (average difference 0.3%; linear correlation coefficient between PPM1 (OD) and PPm (OS) was r = 0.99). Furthermore, to use the same number of data points in each subject, we considered only the values of PPin < 60% above baseline. We found no significant correlation between the changes in each of the flow parameters in OS and those in OD. In Figures 4A and 4B, we plotted the time course of the mean resistance, Rm = PP in /ChBF m during exercise and after the subjects stopped squatting, respectively. During squatting, Rm increased linearly with time (r = 0.95, P < 0.01). After cessation of squatting, Rm decreased rapidly and significandy (P < 0.01) to baseline. A three-parameter exponential fit of the form y = c + a-exp(-bt), starting immediately after subjects stopped squatting, provided a time constant of the return of ChBFm to baseline of 19 seconds. During this time, PP m nearly reached baseline value. Figure 5 shows that PQIBK decreased significantly (linear fit, r= 0.76, P < 0.01) when PPm increased during squatting. Immediately after the subject stopped squatting, PuChBK increased significantly above baseline (0.24 versus 0.21, paired Student's Hest, P < 0.05) to return rapidly to a value not significantly different from baseline. A plot of PuChBF versus pulsatility of the systemic blood pressure, PuBP = (1—BP(|iasI/BPs>,s,), measured during squatting and recovery from squatting, showed no significant correlation. For the subjects used to obtain the data in Figure 3, the values of Sbi and Ssq for ChBVelm, ChBVol,,, and ChBFm were 4%, 3%, and 4%, and 4%, 3%, and 6%, respectively. For this calculation, we conservatively used n = 11, although the lack of significant correla1.6 1.6 1.4 1.4 1.2 1.0 1.0 20 0.8 40 60 80 100 120 Time (sec) 0 40 80 120 160 Time (sec) FIGURE 4. Change in average choroidal vascular resistance, Rm, versus time (a) during squatting (start at t = 0 sec) and (b) immediately after subjects stopped squatting (t = 0 sec). The correlation coefficient of the linear fit: of Rm during squatting is r = 0.95, P < 0.01). The linear correlation coefficient of Rm during recovery from squatting, calculated from log(Rm — c) = loga — bt, where a, b, c are parameters, is r = 0.96 (P< 0.01). Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933419/ on 05/12/2017 Investigative Ophthalmology & Visual Science, October 1997, Vol. 38, No. 11 2342 0.30 0.25 - 0.20 - U 0.15 - 0.10 - 0.05 - 0.00 40 60 70 80 90 PPm (mmHg) FIGURE 5. Pulsatility PUCI,BF versus PPm during squatting. Correlation coefficient of linear fit r = 0.76 (P < 0.01). PPm = mean perfusion pressure. tion between the changes in ChBVelm, ChBVol and ChBF,n during squatting in OS and OD would have warranted using n = 22. DISCUSSION Results of previous work have demonstrated that during isometric exercise, the blood pressure in the brachial artery rises in parallel with that in the ophthalmic artery and that the IOP does not change significandy.11 These findings, together with the results of the current investigation (Fig. 2), indicate that the maintenance of constant ChBFm, in spite of increased perfusion pressure, is achieved through an increase in choroidal vascular resistance (Fig. 4A). The decrease in ChBVolm during squatting suggests that some of die increase in resistance takes place in the volume sampled by the laser, which contains mainly the choriocapillaris.12 To estimate it, we assumed that the change in ChBVolm was caused by vasoconstxiction of the capillaries of the choroid and that blood flows in these capillaries according to Poiseuille's law. In this case, die 6% reduction of ChBVolm occurring during squatting corresponds to a 3% change in vessel diameter, which would increase the resistance by 12%. Although Poiseuille flow is unlikely in these capillaries, this estimated value is well below the ~70% increase needed to keep flow constant in the presence of a 60% increase in PPm. Thus, the changes in resistance during exercise and during recovery from squatting (Figs. 4A, 4B) must occur predominandy in the branches of the ophthalmic artery feeding the foveal region of die choroidal circulation. This interpretation of local action is supported by the finding of a lack of correlation between PuChBF and PuBp under these conditions. Aim and Bill7'8 have revealed the protective role of ocular sympathetic vasomotor nerves in acute arterial hypertension in cats and monkeys. The investigation of the effect of systolic arterial hypertension after ocular sympathectomy by Ernest15 has further emphasized the role of the sympathetic nervous system in the vasoconstriction of die choroidal circulation in response to arterial hypertension. Our results suggest a similar regulatory mechanism in humans. The highly significant increase in heart rate during squatting is expected, in that it is primarily responsible for die elevation in blood pressure.16 In previous studies, Robinson etal,11 in three subjects, and Marcus et al,17 in seven subjects, found no significant changes in IOP during squatting. In the current study in 22 eyes, a 4-mm Hg increase in IOP was observed. This significant change can be explained by die larger number of eyes we studied and also the larger increase in PPm we reached, compared with that seen in die study of Marcus et al.17 The small increase in IOP and the nonsignificant change in end-tidal Pco2 during exercise1117 rule out the possibility that the blood flow response to acute hypertension was related to changes in IOP or Pco2. The nearly constant ChBF, in spite of increases in PPm up to «62%, does not represent an auto regulatory response in the classic sense, which consists of a local change in the vascular resistance when the organ has been isolated from systemic influence. To study autoregulation in the choroid in response to raised arterial pressure, the neural input to the choroidal vessels would have to be removed. Our measurements provide only information on the hemodynamics in the foveal region of die fundus. Whether die ChBF regulation during isometric exercise is as efficient in the peripheral region of die fundus remains to be investigated. A regional variation of ChBF versus PPm cannot be excluded, because spatial variations in the changes in choroidal Po2 and pH in die subretinal space have been found in cats in response to step increases in IOP.1819 In this investigation, the laser-Doppler flowmetry technique has been applied in normal volunteers. Because the probing laser beam is also the fixation target, the position of the laser beam at the fovea can be maintained easily. The main task of the operator of the equipment is to insure that the input pupil of die laser-Doppler flowmetry optical system remains approximately centered in the pupil of the eye measured. The continuous monitoring of the DC level of die Doppler signal, which is proportional to the amount of laser light reaching the detector, allows the operator to insure that this is die case. We believe, therefore, that measurements of ChBF in the region of the fovea during isometric exercises can be applied in studies in cooperative patients. Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933419/ on 05/12/2017 2343 Choroidal Blood Flow Key Words choroidal blood flow, isometrics, laser Doppler flowmetry, regulation, sympathetic nervous control 10. Acknowledgments The authors thank Nicole Michellod for help in preparing the manuscript. 11. References 1. Lind AR, Taylor SH, Humphreys PW, Kennelly BM, Donald KW. The circulatory effects of sustained voluntary muscle contraction. Clin Sci. 1964;27:229-244. 2. Delius W, Hagbarth K-E, Hongell A, Wallin BG. Manoeuvers affecting sympathetic outflow in human nerves. Acta Physiol Scand. 1972;84:82-94. 3. Mitchell JH, Schmidt RF. Cardiovascular reflex control by afferent fibers from skeletal muscle receptors. In: Shepherd JT, Abboud FM, eds. Handbook of Physiology. Sec. 2, The Cardiovascular System. Vol 3. Bethesda, MD: American Physiological Society; 1983:623-658. 4. Wesche J. The time course and magnitude of blood flow changes in the human quadriceps muscles following isometric contraction. / Physiol (Lond). 1986; 337:445-462. 5. Taylor WF, Johnson JM, Kosiba WA, Kwan CM. Cutaneous vascular response to isometric handgrip exercise. fAppl Physiol. 1989;66:1586-1592. 6. Bystrom SEG, Kilbom A. Physiological response in the forearm during and after isometric intermittent handgrip. Eur fAppl Physiol. 1990; 60:457-466. 7. Aim A, Bill A. The effect of stimulation of the sympathetic chain on retinal oxygen tension and uveal, retinal and cerebral blood flow in cats. Acta Physiol Scand. 1973;88:84-96. 8. Aim A. The effect of sympathetic stimulation on blood flow through the uvea, retina, and optic nerve in monkeys. ExpEyeRes. 1977;25:19-25. 9. Bill A, Linder M, LinderJ. The protective role of ocular sympathetic vasomotor nerves in acute arterial hy- 12. 13. pertension. Proceedings of the Ninth European Conference on Microcirculation. Antwerpen, Belgium, 1976. BiblAnat. 1977; 16:30-37. Michelson G, Groh M, Griindler A. Regulation of ocular blood flow during increase of arterial blood pressure. Br fOphthalmol. 1994; 78:461-465. Robinson F, Riva, CE, Grunwald JE, Petrig BL, Sinclair SH. Retinal blood flow autoregulation in response to an acute increase in blood pressure. Invest Ophthalmol Vis Sci. 1986;27:722-726. Riva CE, Cranstoun SD, Grunwald JE, Petrig BL. Choroidal blood flow in the foveal region of the human ocular fundus. Invest Ophthalmol Vis Sci. 1994; 35:42734281. Petrig BL, Riva CE. Optic nerve head laser Doppler flowmetry: Principles and computer analysis. In: Kaiser HJ, Flammer J, Hendrickson P, eds. Ocular Blood Flow. New Insights into the Pathogenesis of Occult Diseases, 1995. Basel, Switzerland: Karger; 1996:120-127. 14. Bonner RF, Nossal R. Principles of laser-Doppler flowmetry. In: Shepherd AP, Oberg PA, eds. Developments in Cardiovascular Medicine: laser-Doppler Blood, 15. 16. 17. 18. 19. Flowmetry. Boston: Kluwer Academic Publishers; 1990:73-92. Ernest JT. The effect of systolic hypertension on rhesus monkey eyes after ocular sympathectomy. Am J Ophthalmol 1977;84:341-344. Shepherd JT, Blomqvist CG, Lind AR, Mitchell JH, Saltin B. Static (isometric) exercise. Retrospection and introspection. Circ Res. 1981;48(suppl):l79-188. Marcus DP, Edelhauser HF, Maksud MG, Wiley RL. Effects of sustained muscular contraction on human intraocular pressure. Clin Sci Mol Med. 1974; 47:249257. Yancey CM, Linsenmeier RA: Oxygen distribution and consumption in the cat retina at increased intraocular pressure. Invest Ophthalmol Vis Sci. 1989;30:600-611. Yamamoto F, Steinberg RH. Effects of intraocular pressure on pH outside rod photoreceptors in the cat retina. ExpEyeRes. 1992;55:279-288. Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933419/ on 05/12/2017