Elevated intraocular pressure (IOP) is a major risk factor for

Transcription

1 Glaucoma The Association between Retinal Vessel Diameter and Retinal Nerve Fiber Layer Thickness in Asymmetric Normal Tension Glaucoma Patients Joon Mo Kim, 1 Mo Sae Kim, 2 Hyo Ju Jang, 1 Ki Ho Park, 3 and Joseph Caprioli 4 PURPOSE. The aim of this study was to evaluate the associations between retinal vessel diameter and retinal nerve fiber layer (RNFL) thickness in patients with normal tension glaucoma (NTG). METHODS. This study included 67 untreated patients with asymmetric NTG, with no evidence of glaucoma in the contralateral eyes, and 48 age- and sex-matched normal control subjects. We included patients from 20 to 70 years of age who had no history of hypertension, diabetes, or other vascular diseases. All subjects underwent detailed eye examinations that included red-free photography, stereoscopic optic disc photography, automated perimetry, and measurement of RNFL thickness with optical coherence tomography. We compared RNFL thicknesses, temporal retinal arteriolar diameters (TRAD), and temporal retinal venular diameters (between quadrants with and without RNFL defects) with computerassisted imaging software. RESULTS. The mean diameter of the temporal retinal vessels in the quadrants with RNFL defects were significantly smaller in patients with NTG than in those with quadrants without RNFL defects (P < 0.001) and in control subjects (P < 0.001). TRADs were correlated with RNFL defects and RNFL thicknesses (P < 0.05) in binary logistic regression analysis and in multiple linear regression analysis (P < 0.001). There were no statistically significant differences between vessel diameters in control subjects and those in nondefective quadrants of the affected NTG eyes or the unaffected eyes of patients with NTG. CONCLUSIONS. Our results show that narrower retinal vessels are found in areas of RNFL defects. Considering previous reports and our analysis, this is likely due to the decreased demand for retinal blood flow in damaged areas of the retina. (Invest Ophthalmol Vis Sci. 2012;53: ) DOI: / iovs From the 1 Department of Ophthalmology, Kangbuk Samsung Hospital, Sungkyunkwan University College of Medicine, Seoul, Korea; the 2 Department of Ophthalmology, National Cancer Center, Goyang,Gyeonggi,Korea;the 3 Department of Ophthalmology, Seoul National University College of Medicine, Seoul, Korea; and the 4 Department of Ophthalmology, Jules Stein Eye Institute, University of California, Los Angeles, California. Supported by an Association for Research in Vision and Ophthalmology travel grant. Submitted for publication March 1, 2011; revised May 2, June 13, and July 18, 2012; accepted July 21, Disclosure: J.M. Kim, None;M.S. Kim, None;H.J. Jang, None; K.H. Park, None;J. Caprioli, None Corresponding author: Ki Ho Park, Department of Ophthalmology, Seoul National University College of Medicine, Seoul, Korea Youngeondong, Jongno-gu, Seoul, Korea ; kihopark@snu.ac.kr. Elevated intraocular pressure (IOP) is a major risk factor for primary open-angle glaucoma (POAG). In addition, research suggests that abnormalities of blood flow contribute to the development and progression of POAG, such as in systemic hypertension, 1 vasospasm, 2 diabetes, 3 and migraine. 4 Although normal tension glaucoma (NTG) is considered a form of chronic OAG and IOP is still widely recognized as a major risk factor for progression, 5 whether NTG and high-tension glaucoma differ in their pathogenesis remains controversial. Some patients still suffer glaucomatous optic nerve damage despite IOP within the statistically normal range. Therefore, additional causes of NTG have been suggested, such as vascular factors. 6 Blood flow can be regulated depending on changing metabolic needs, local vascular resistance, and other processes. 6 If local blood supply does not meet the demand of local tissue, ischemic damage can result. If the needs of the local tissue require less supply, blood flow will be reduced through autoregulatory mechanisms. Prior studies have examined the relationship between retinal vascular diameter and retinal nerve fiber layer (RNFL) thickness in patients with OAG. Rader et al. 7 reported that the location of proximal constriction in glaucoma corresponded to the sector with the greatest cupping, and a generalized narrowing of the retinal arteries was related to the severity of optic nerve damage regardless of the cause. Similarly, studies by Jonas et al. 8 and Jonas and Naumann 9 demonstrated that parapapillary retinal vessel diameters may reflect the need of vascular supply in the corresponding superficial retinal area, and it decreases with advanced glaucomatous optic nerve damage. Mitchell et al. 10 reported that eyes with OAG had retinal arteriolar diameters that were significantly smaller than those of either normal eyes or eyes with ocular hypertension. Although many studies have tried to determine whether vascular narrowing was a cause or an effect of glaucomatous damage, additional work is required to address this issue. We investigated here the associations between retinal vessel diameters and RNFL thicknesses in affected and unaffected areas of the retina in patients with NTG and in normal controls to help clarify whether retinal vascular narrowing is a cause or an effect of glaucomatous damage. MATERIALS AND METHODS We retrospectively reviewed medical records of patients who visited the outpatient glaucoma clinic of Kangbuk Samsung Hospital in Seoul, Korea, from January 2010 to December At the same clinic during the same period, age- and sex-matched consecutive control subjects were recruited. Control subjects were recruited from among patients referred for mass routine health check-ups, from among those with mild cataracts in whom surgery was deemed unnecessary, and from among patients referred for dry eye care. This study was approved by Investigative Ophthalmology & Visual Science, August 2012, Vol. 53, No. 9 Copyright 2012 The Association for Research in Vision and Ophthalmology, Inc. 5609

2 5610 Kim et al. IOVS, August 2012, Vol. 53, No. 9 FIGURE 1. Dilated 308 stereoscopic disc photograph. Diameters of central retinal arterioles and venules passing completely through a circumferential zone from 0.5 to 1 disc diameter from the optic disc margin were measured in micrometers with an automated program. the Kangbuk Samsung Institutional Review Board, and adhered to the Declaration of Helsinki. All study subjects underwent ophthalmological examinations that included autorefraction, Goldmann applanation tonometry, visual field testing (Zeiss-Humphrey, San Leandro, CA), measurement of retinal nerve fiber layer thickness with optical coherence tomography (OCT, version 3.0; Carl Zeiss Meditec, Inc., Dublin, CA), stereoscopic optic disc photography, and red-free fundus photography (Visucam Pro NM model; Carl Zeiss Meditec, Inc.). NTG was diagnosed by a glaucoma specialist using the following criteria: typical glaucomatous optic neuropathy including rim thinning or notching in the inferior or superior temporal area of the optic nerve head; corresponding visual field loss, including paracentral or arcuate scotomas or a nasal step; a diurnal IOP measurement always below 21 mm Hg with diurnal measurements (without medication); an open anterior chamber angle on gonioscopy; and no secondary cause of glaucomatous optic neuropathy. The diurnal IOP test consisted of measuring IOP every 150 minutes from 9 AM to 4:30 PM at the initial visit. Control subjects and unaffected contralateral eyes of NTG patients were defined as having a normal optic disc appearance, normal Stratus OCT results, no RNFL defects in red-free photographs, and normal visual fields. In eyes of NTG patients, the unaffected quadrant opposite the affected quadrant was defined as being without glaucomatous damage by red-free photography, OCT measurements, and visual field testing. IOP was measured with a Goldmann applanation tonometer, and visual fields were evaluated with the 24-2 program of the Humphrey visual field analyzer (Zeiss Inc., San Leandro, CA), with the Swedish Interactive Threshold Algorithm (SITA) standard algorithm. Exclusion criteria included a history of hypertension, diabetes, or vascular disease; prior glaucoma treatment; myopia greater than 6 diopters; best corrected visual acuity worse than 20/ 40; peripapillary atrophy greater than 0.5 disc diameter; large discs with a 0.6 or greater cup-to-disc ratio in the control group; and previous intraocular surgery. We obtained dilated 308 stereoscopic optic disc photographs centered on the disc with a fundus camera (Visucam Pro NM; Carl Zeiss Meditec). We measured the diameters of and calculated the arteriovenous (AV) ratios of temporal retinal arterioles (TRA) and temporal retinal venules (TRV) that completely traversed a circumferential zone of 0.5 to 1 disc diameter from the optic disc margin with computerassisted software (Visupac/system version software; Carl Zeiss, Pirmasens, Germany). Visupac software was a customized program that provided the real size of an object on the fundus with an adjusted value that incorporated Littmann s correction. 11 The AV ratio program was used to select vessels of interest from the fundus photograph and gave the measurement of the TRA diameter (TRAD) and TRV diameter (TRVD), which were provided automatically based on the value by Visuapc (Fig. 1). The algorithm calculated the average value of the selected vessel within a zone of 0.5 to 1 disc diameter from the optic disc margin. Vessel diameters were obtained automatically once the vessel was selected by the investigator. Because such an automated process may result in inaccurate measurements if the quality of the image is poor, we included only subjects with high-quality images showing the optic disc margin and vessel borders clearly. Two readings were obtained, and the average value was used for subsequent analyses. Each TRAD and TRVD was measured in four quadrants (superonasal, superotemporal, inferonasal, and inferotemporal). To localize and match the vessels with a RNFL defect, we used TRADs and TRVDs of the superotemporal or inferotemporal quadrant associated with RNFL defects. We compared demographic data (age, sex, IOP, refractive error), RNFL thickness, TRAD, TRVD, and AV ratio between patients with NTG and control subjects without glaucoma, using the independent t-test, chi-squared test, ANOVA, and binary logistic regression analysis. Age, sex, and factors with a P value of <0.2 on univariate analysis were included in the logistic regression analysis model. TRADs, TRVDs, and RNFL thicknesses within the quadrants with RNFL defects were compared with those of each quadrant in the controls as follows: (1) affected and unaffected quadrants in the same eye; (2) affected quadrants and each corresponding area of the superotemporal or inferotemporal quadrants of the unaffected eye; (3) affected quadrants and the same quadrant of the eyes of normal controls; and (4) unaffected quadrants and the same quadrants of unaffected NTG eyes or the same quadrant of the eyes of normal controls. We also used multivariate linear regression analysis to evaluate associations among TRAD, TRVD, and RNFL thickness adjusted for age, IOP, and refractive error. P values less than 0.05 were considered statistically significant. All statistical analyses were performed with PASW version 17.0 software (SPSS, Chicago, IL). RESULTS Subject cohorts consisted of 67 patients (of 87 who were evaluated for inclusion) with asymmetric NTG, and 48 normal control subjects. Of the 20 patients who were excluded, 3 patients had optic disc hypoplasia, 2 had branch retinal vein occlusions, 1 had pituitary adenoma, 1 had a history of optic neuritis, 1 had a papillomacular nerve fiber bundle defect, 4 had disc hemorrhages at the initial visit, and 8 had missing data (including medical history, red-free photography, stereo disc photography, and diurnal IOP before treatment). Table 1 shows demographic and clinical features of the study subjects. There were no significant differences in age, sex, IOP, refractive error, or AV ratio between the groups. RNFL thickness, mean deviation (MD), pattern standard deviation (PSD), TRAD, and TRVD of eyes with NTG were significantly smaller than those of the normal controls (P < 0.001). Multivariate binary logistic regression analysis was performed to evaluate risk factors for glaucoma with a RNFL defect compared to control subjects without glaucoma (Table 2). TRAD data showed significant relationships in the model adjusted for age, sex, TRAD, and TRVD (P ¼ 0.001). We found a statistically significant association between narrower retinal vessel and RNFL defects. The TRAD (P < 0.001) and TRVD (P ¼ 0.005) of the quadrants with RNFL defects in eyes with NTG were significantly smaller than those in the unaffected quadrants of the same eye (Table 3). In

3 IOVS, August 2012, Vol. 53, No. 9 Changes in Retinal Arterioles and Venules in Glaucoma 5611 TABLE 1. Comparison of NTG Patients with RNFL Defects and Control Subjects without Glaucoma Variable Patient Values Control Values P Value Sex (male:female) 38:29 27: * Mean age 6 SD (y) Mean IOP 6 SD (mm Hg) Mean refractive error 6 SD (D) Mean deviation 6 SD (db) <0.001 Mean pattern 6 SD (db) Average RNFL thickness 6 SD (lm) <0.001 Mean TRAD 6 SD (lm) <0.001 Mean TRVD (lm) <0.001 Mean AVR 6 SD AVR, arteriovenous ratio; D, diopters. * Chi-squared test. Independent two-tailed Student s t-test. TABLE 2. Comparison of Risk Factors in NTG Patients with RNFL Defects with Those in Control Subjects without Glaucoma Adjusted Factor Odds Ratio 95% Confidence Interval P Value Age per 1 y to Sex (female) to TRAD per 1 lm to TRVD per 1 lm to addition, the TRAD (P < 0.001) and TRVD (P ¼ 0.011) were significantly smaller in eyes with NTG than they were in the same quadrant of the unaffected contralateral eye (Table 4). Multivariate binary regression analysis for the quadrants with and without RNFL defects in the same patient and the quadrants with RNFL defects and corresponding quadrants of contralateral unaffected eye showed significant associations for TRAD (P < 0.05) (Table 5). Multivariate linear regression analysis was used to evaluate the association between several independent variables and the measured RNFL thickness in eyes with a RNFL defect and those in control subjects (Table 6). The model was adjusted for age, IOP, refractive error, TRAD, and TRVD. This model showed statistically significant results (r 2 ¼ 0.250, P < 0.001). TRAD was significantly correlated with RNFL thickness (P < 0.001), but there were no statistically significant associations among RNFL thickness and age, IOP, refractive error, and TRVD. Scatter plots illustrating the relationships between TRAD and RNFL thickness are shown in Figure 2. There were no significant correlations between RNFL thickness and any of the independent variables in the control group (P ¼ 0.114). Multivariate linear regression analysis was also used to evaluate the associations between TRAD and several independent variables (age, IOP, refractive error, and TRVD). We found a positive correlation between the TRAD and the TRVD (P < 0.001). There were no significant associations among RNFL thickness, TRAD, TRVD, or AV ratio among quadrants without an RNFL defect, unaffected contralateral eye, or eyes of normal control subjects (Table 7). DISCUSSION Previous reports have evaluated the retinal vessel diameter in glaucomatous eyes and its association with RNFL thickness and visual field defects. For example, Hall et al. 12 found a strong association between decreased peripapillary arteriole diameter and visual field defects in the corresponding hemifield in eyes with POAG. Rankin et al. 13 reported a higher prevalence of focal arteriolar narrowing in patients with chronic OAG and NTG than in patients with ocular hypertension and normal controls. In addition, the Blue Mountains Eye Study showed that eyes with glaucomatous damage had significantly smaller mean retinal vessel diameters and were at least twice as likely to have generalized arteriolar narrowing than normal eyes. 10 Similarly, Zheng et al. 14 found a significant association between narrower retinal vessel caliber and RNFL thinning in a population-based study of Asian Malays aged 40 to 80 years without glaucoma, although they found no such relationship in patients with glaucoma. Additional work is required to determine whether these differences between retinal vessel diameter are a cause or an effect of glaucomatous optic neuropathy. Our results support the previously reported associations between retinal arteriolar narrowing and NTG. We found a significant association between narrowing of retinal vessels and RNFL defects in eyes with NTG. We compared diameters of retinal vessels in quadrants with RNFL defects in eyes with NTG to those in normal quadrants or in the opposite, still unaffected eye of these asymmetric NTG patients. Specifically, the TRADs of quadrants with RNFL defects in patients with NTG were significantly smaller than those in other quadrants without RNFL defects in the same eye, as well as those in the same quadrant of the contralateral unaffected eye. Multivariate analysis adjusted for age, IOP, refractive error, and TRAD revealed a significant positive correlation between the RNFL thickness and the TRAD in eyes with NTG. The associations TABLE 3. Comparison between Quadrants with and without RNFL Defects in the Same Eye Quadrant Mean RNFL Thickness 6 SD (lm) Mean TRAD 6 SD (lm) Mean TRVD 6 SD (lm) Mean AVR 6 SD With RNFL defect Without RNFL defect P value* <0.001 < * Independent t-test.

4 5612 Kim et al. IOVS, August 2012, Vol. 53, No. 9 TABLE 4. Comparison between Quadrants of Eyes with an RNFL Defect and Corresponding Quadrants of Contralateral Unaffected Eyes Quadrant Mean RNFL Thickness 6 SD (lm) Mean TRAD 6 SD (lm) Mean TRVD 6 SD (lm) Mean AVR 6 SD Eyes with RNFL defect Same quadrant of the contralateral unaffected eye P value* <0.001 < * Independent t-test. TABLE 5. Comparison between Quadrants in the Same Patients with and without RNFL Defects and Quadrants with RNFL Defects and Corresponding Quadrants of Contralateral, Unaffected Eyes* Quadrant Arteriolar or Venular Diameter Odds Ratio 95% Confidence Interval P Value With vs. without RNFL defect in the same eye TRAD per 1 lm to TRVD per 1 lm to Eye with RNFL defect vs. unaffected eye TRAD per 1 lm to TRVD per 1 lm to * Multivariate binary logistic regression analysis. TABLE 6. Comparison of RNFL Thickness and Other Variables between Eyes of NTG Patients and Those of Control Subjects* Adjusted Factor Regression Coefficient (B) 95% Confidence Interval P Value Age per 1 y to IOP per 1 mm Hg to Refractive error per 1 D to TRAD per 1 lm to <0.001 TRVD per 1 lm to * Multivariate linear regression analysis (r 2 ¼ 0.247, P < 0.001). uncovered in this study may be more reliable than previous reports because we objectively measured the vessel caliber. The use of a trained grader, even when masked to patient data, results in subjective measurements that may affect the reproducibility of the study. 10,12 14 In contrast, our study used a human data collector only to select vessels of interest, which we then evaluated with an automated program to determine their diameters. As such an automated process may result in inaccurate measurements if the quality of the image is poor, we used only high-quality images. The vascular cause of glaucoma is not completely understood, but clinical studies have shown an association between glaucoma and retinal arteriolar narrowing. 8,9,12 14 This theory considers glaucomatous optic neuropathy is the result of insufficient blood supply due to either increased IOP or other factors that reduce ocular blood flow and perfusion pressure, leading to ischemic or reperfusion damage. 15 Koh et al. 16 reported localized nonprogressive RNFL defects associated with cotton wool spots. Cotton wool spots are thought to be caused by acute swelling of the nerve fiber layer due to acute infarction of a retinal arteriole. 17 The pathogenesis of this type of RNFL defect probably differs from that of glaucoma. This phenomenon suggests that an RNFL defect can be induced by retinal vascular insufficiency, but very localized, acute vascular insufficiency may not be the cause of progressive glaucomatous damage. Our multivariate analysis results revealed a significant positive correlation between retinal arteriolar diameter and TABLE 7. Comparison among Quadrants without an RNFL defect and Those of Unaffected Contralateral Eyes and Eyes of Control Subjects Variable Group Mean Value 6 SD P Value RNFL thickness (lm) QsD/UN / QsD/CON / UN/CON / TRAD (lm) QsD/UN / QsD/CON / UN/CON / TRVD (lm) QsD/UN / QsD/CON / UN/CON / AVR QsD/UN / QsD/CON / UN/ CON / CON ¼ control subject s eye; QsD ¼ quadrant without defect; UN ¼ unaffected eye.

5 IOVS, August 2012, Vol. 53, No. 9 Changes in Retinal Arterioles and Venules in Glaucoma 5613 inter- and intraphotographic variability of the measurements are not precisely known. In conclusion, we found statistically significant associations between central retinal vessel diameters and quadrants of RNFL defects in patients with NTG. Furthermore, a narrow retinal arteriolar diameter was significantly associated with diminished RNFL thickness. There were no significant associations between RNFL thickness, TRAD, TRVD, or AV ratio among quadrants in NTG patients without an RNFL defect, the unaffected contralateral eye, or eyes of the normal control subjects. Considering previous reports and our analysis, it is a plausible that vessel narrowing is an effect that results from less demand for blood flow in the areas of the retina damaged by glaucoma. FIGURE 2. Scatter plots of the relationship between TRAD and RNFL thickness are shown (Pearson coefficient, 0.244; P < 0.001). RNFL thickness in patients with NTG. The loss of retinal ganglion cells may cause a decrease in the retinal vessel caliber via autoregulatory mechanisms by responding to a decreased regional demand for oxygen. 7 9,18,19 This hypothesis is supported by the observation that retinal vessel narrowing is found in various forms of optic nerve damage and is not specific to glaucoma. 7,14,19 22 In our study, multivariate analysis simultaneously adjusted for TRAD and TRVD revealed a significant association between retinal arteriolar narrowing and decreased RNFL thickness, but retinal venular caliber was not significantly associated with RNFL thickness. Similarly, previous glaucoma studies have shown stronger associations with retinal arteriolar caliber than with retinal venular caliber. 7,23 Leung et al. 24 suggested that retinal arteriolar narrowing was greater than venular narrowing in patients with hypertension. However, this was caused by active microvascular damage with elevated blood pressure. We excluded patients with systemic hypertension. In the retinas of glaucoma patients, a reduced ganglion cell population would demand less blood flow. If there was systemic vascular dysregulation, the unaffected eye, at least the unaffected quadrant of the same eye, should also manifest some vascular narrowing. Considering that the vessel calibers in the unaffected quadrants of glaucomatous eyes did not differ significantly from those in normal controls and that the RNFL thickness change was associated with TRAD, this explanation seems plausible. A study by Arend et al. 25 showed no differences between retinal arterial diameters in patients with early NTG and those in normal controls, suggesting an effect rather than a causal relationship. As this is not a longitudinal study, it is not possible to form any conclusions about the relationship between arteriole and venular diameters. Our study has several limitations. First, because of its retrospective nature, we cannot exclude the possibility of selection bias. Prospective validation of our results will be required. Second, we did not perform complete medical examinations of every patient. Instead, we took detailed histories regarding underlying vascular diseases, dyslipidemia, and diabetes and excluded those with underlying diseases. Third, a potential limitation is the smaller number of control subjects than patients with NTG. Fourth, we used a time domain OCT to measure RNLF thickness. A spectral domain OCT might have provided more accurate results. Finally, the Acknowledgments The authors thank Fei Yu (Department of Biostatistics, UCLA Fielding School of Public Health, Jules Stein Eye Institute, University of California Los Angeles) for statistical advice and guidance. References 1. Tielsch JM, Katz J, Sommer A, Quigley HA, Javitt JC. Hypertension, perfusion pressure, and primary open-angle glaucoma. A population-based assessment. Arch Ophthalmol. 1995;113: Drance SM, Douglas GR, Wijsman K, Schulzer M, Britton RJ. Response of blood flow to warm and cold in normal and lowtension glaucoma patients. Am J Ophthalmol. 1988;105: Mitchell P, Smith W, Chey T, Healey PR. Open-angle glaucoma and diabetes: the Blue Mountains eye study, Australia. Ophthalmology. 1997;104: Wang JJ, Mitchell P, Smith W. Is there an association between migraine headache and open-angle glaucoma? Findings from the Blue Mountains Eye Study. Ophthalmology. 1997;104: Jonas JB, Gründler AE, Gonzales-Cortes J. Pressure-dependent neuroretinal rim loss in normal tension glaucoma. Am J Ophthalmol. 1998;125: Flammer J, Orgul S, Costa VP, et al. The impact of ocular blood flow in glaucoma. Prog Retin Eye Res. 2002;21: Rader J, Feuer WJ, Anderson DR. Peripapillary vasoconstriction in the glaucomas and the anterior ischemic optic neuropathies. Am J Ophthalmol. 1994;117: Jonas JB, Nguyen XN, Naumann GO. Parapapillary retinal vessel diameter in normal and glaucoma eyes. I. Morphometric data. Invest Ophthalmol Vis Sci. 1989;30: Jonas JB, Naumann GO. Parapapillary retinal vessel diameter in normal and glaucoma eyes. II. Correlations. Invest Ophthalmol Vis Sci. 1989;30: Mitchell P, Leung H, Wang JJ, et al. Retinal vessel diameter and open-angle glaucoma: the Blue Mountains Eye Study. Ophthalmology. 2005;112: Littmann H. Zur Bestimmung der wahren Gröbe eines Objektes auf dem Hintergrund des lebenden Auges. Klon Monatsbl Augenbeilked. 1982;180: Hall JK, Andrews AP, Walker R, Piltz-Seymour JR. Association of retinal vessel caliber and visual field defects in glaucoma. Am J Ophthalmol. 2001;132: Rankin SJ, Drance SM. Peripapillary focal retinal arteriolar narrowing in open angle glaucoma. J Glaucoma. 1996;5: Zheng Y, Cheung N, Aung T, et al. Relationship of retinal vascular caliber with retinal nerve fiber layer thickness: the

6 5614 Kim et al. IOVS, August 2012, Vol. 53, No. 9 Singapore Malay Eye Study. Invest Ophthalmol Vis Sci. 2009; 50: Flammer J, Orgul S, Costa VP, et al. The impact of ocular blood flow in glaucoma. Prog Retin Eye Res. 2002;21: Koh JW, Park KH, Kim MS, Kim JM. Localized retinal nerve fiber layer defects associated with cotton wool spots. Jpn J Ophthalmol. 2010;54: Schmidt D. The mystery of cotton-wool spots - a review of recent and historical descriptions. Eur J Med Res. 2008;13: Papastathopoulos KI, Jonas JB. Follow up of focal narrowing of retinal arterioles in glaucoma. Br J Ophthalmol. 1999;83: Jonas JB, Fernandez MC, Naumann GOH. Parapapillary atrophy and retinal vessel diameter in nonglaucomatous optic nerve damage. Invest Ophthalmol Vis Sci. 1991;32: Frisén L, Claesson M. Narrowing of the retinal arterioles in descending optic atrophy. A quantitative clinical study. Ophthalmology. 1984;91: Rankin SJ, Drance SM. Peripapillary focal retinal arteriolar narrowing in open angle glaucoma. J Glaucoma. 1996;5: Papastathopoulos KI, Jonas JB. Focal narrowing of retinal arterioles in optic nerve atrophy. Br J Ophthalmol. 1995;102: Hall JK, Andrews AP, Walker R, Piltz-Seymour JR. Association of retinal vessel caliber and visual field defects in glaucoma. Am J Ophthalmol. 2001;132: Leung H, Wang JJ, Rochtchina E, et al. Relationship between age, blood pressure, and retinal vessel diameters in older population. Invest Ophthalmol Vis Sci. 2003;44: Arend O, Remky A, Plange N, Martin BJ, Harris A. Capillary density and retinal diameter measurements and their impact on altered retinal circulation in glaucoma: a digital fluorescein angiographic study. Br J Ophthalmol. 2002;86: