Correlation Between Optic Disc Peripapillary Capillary Network and Papilledema Grading in Patients With Idiopathic Intracranial Hypertension: A Study of Optical Coherence Tomography Angiography

Original Contribution Section Editors: Clare Fraser, MD Susan Mollan, MD Correlation Between Optic Disc Peripapillary Capillary Network and Papilledema Grading in Patients With Idiopathic Intracranial Hypertension: A Study of Optical Coherence Tomography Angiography Downloaded from http://journals.lww.com/jneuro-ophthalmology by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC1y0abggQZXdtwnfKZBYtws= on 05/04/2022 Yasaira Rodriguez Torres, MD, Patrick Lee, BS, Melanie Mihlstin, MD, Robert L. Tomsak, MD, PhD Background: The continued increase in idiopathic intracranial hypertension (IIH) prevalence has many implications for societal health care. Its potential vision-threatening consequences make ophthalmologists key players in its diagnosis and management. Newer technology such as optical coherence tomography angiography (OCT-A) enables evaluation of the branching complexity of the peripapillary capillary plexus, a region where accurate imaging via fluorescein angiography was previously limited. Methods: A cross-sectional, observational study of 23 (46 eyes) consecutive patients with IIH. Peripapillary total vasculature was recorded using commercial OCT-A en face vessel density mapping. In addition, OCT-A blood flow slab was compared with papilledema grading. OCT-A images were analyzed using a customized image analysis protocol using ImageJ software (v1.51w) and Photoshop software (Adobe Systems, CA). SPSS software version 25 was used for statistical analysis (SPSS Inc, IBM, Chicago, IL). Results: Skeletonized vessel density peripapillary capillary plexus was significantly associated with Frisen papilledema grades, OCT retinal nerve fiber layer (RNFL), and macular ganglion cell layer (GCL) thickness with a P , 0.001, P = 0.022, and P = 0.006, respectively. Every point increase in grade was correlated with a decrease of 9.1 pixels/mm2 in vessel density (R = 0.512, b = 20.115 ± 0.029; P , 0.001). Increased papilledema was correlated with an increased retinal blood flow percentage (R = 0.300, b = 2.114 ± 1.013; P , 0.05) and decreased choroidal blood flow (CBF) percentage (R = 0.300, b = 2.114 ± 1.013; P , 0.05). Every point increase in grade was correlated with Department of Ophthalmology, Kresge Eye Institute, Detroit, Michigan. This project was supported in part by Research for Prevention of Blindness unrestricted/challenge grant to the Kresge Eye Institute. The authors report no conflicts of interest. Y. Rodriguez Torres and P. Lee are the Co-first authors. Address correspondence to Yasaira Rodriguez Torres, MD, Department of Ophthalmology, Kresge Eye Institute/Wayne State University School of Medicine, 4717 St. Antoine, Detroit, MI 48201; E-mail: [e-mail] 48 a decrease in CBF by 47.4%, as calculated using a linear best-fit line inclusive for all of the data points. Conclusions: OCT-A allows for effective visualization and quantification of the peripapillary retinal vasculature. Our results demonstrate a correlation between skeletonized peripapillary density and papilledema grading, OCT RNFL thickness, and GCL thickness. In addition, we show a significant negative correlation between CBF and papilledema grading. These changes provide key findings regarding the pathophysiology of optic neuropathy in papilledema and highlight the potential of OCT-A as a diagnostic tool for papilledema and a clinical marker for detecting early optic nerve damage. Journal of Neuro-Ophthalmology 2021;41:48–53 doi: 10.1097/WNO.0000000000000877 © 2020 by North American Neuro-Ophthalmology Society I diopathic intracranial hypertension (IIH), also known as primary pseudotumor cerebri, is a disorder characterized by increased intracranial pressure of unknown etiology. Recent studies show its links to poverty, obesity, and a rising economical burden (1). Among the stricken are overweight women of childbearing age, most of whom are within the working age group (i.e., between the ages of 20–64 years). Despite how little is known regarding IIH, the estimated economic burden in the United States exceeds $444 million annually, which includes hospital admissions, surgical treatments, and lost wages (2). IIH can present with an array of symptoms and signs, among which papilledema is a key hallmark. Visual loss is usually slow and reversible, but even with appropriate treatment up to 25% of patients may suffer from severe permanent vision loss (3). However, the exact pathophysiology of vision loss due to papilledema in the setting of IIH remains unclear (4). Previous studies on ocular circulation Rodriguez Torres et al: J Neuro-Ophthalmol 2021; 41: 48-53 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution have suggested that patients with optic nerve head (ONH) swelling may have impaired ONH blood flow (5,6). Current diagnostic tools such as laser doppler flowmetry, which samples capillary flow over a small retinal area and laser speckle flowgraphy, which samples blood velocity, can show differences in blood circulation but are too variable for diagnostic application (7,8). Fluorescein angiography is a useful qualitative tool; however, it has limited utility when used to evaluate the ONH capillary plexus (5,9). Currently, optical coherence tomography (OCT) is used to quantify swelling of the peripapillary retinal nerve fiber layer (RNFL). OCT angiography (OCT-A) is a newer noninvasive imaging modality which allows us to visualize and quantify ONH vasculature. Studies have shown the potential of OCT-A at detecting abnormalities in ONH perfusion (10–12). We hypothesize that information from OCT structural imaging and OCT-A perfusion measurements could provide complementary information of both pathophysiological and clinical value for understanding and managing papilledema. METHODS Subjects A cross-sectional, observational study of consecutive patients evaluated at the neuro-ophthalmology clinic at the Kresge Eye Institute in Detroit, Michigan, with a diagnosis of IIH or papilledema from January 2018 through August 2018. All subjects underwent a complete ophthalmic examination, including best-corrected visual acuity, intraocular pressure, slit-lamp biomicroscopy, and dilated fundus examination. Inclusion criteria consisted of patients who met modified Dandy criteria for the diagnosis of IIH who presented with papilledema Frisen Grade 1–5. In addition, we included patients with Frisen Grade 0 with inactive IIH, which was defined as stable visual acuity, stable OCT measurements, and a Frisen Grade 0 on ophthalmoscopy. Patients with glaucoma, optic neuritis, nonarteritic anterior ischemic optic neuropathy (NAION), other causes of optic disc edema or optic nerve diseases that may lead to optic disc edema, optic neuropathy, and atrophy were excluded from the study. All patients were evaluated by a fellowship-trained neuroophthalmologist (R.L.T.). Papilledema grading was determined by 2 graders (R.L.T. and Y.R.T.). Reliability analysis using Cohen’s Kappa was performed to determine consistency among graders. Intrarater reliability was found to be kappa of 0.868 (P , 0.001), whereas interrater reliability was found to be kappa of 0.868 (P , 0.001). Optical Coherence Tomography Measurements Zeiss angioplex technology was used to obtain the OCT-A which detects motion of scattering particles such as red blood cells with sequential OCT-B scans performed repeatedly at the same location of the retina. The angioplex maps consist of a 2D reconstruction of the vasculature of the particular area of interest. Both 3 · 3 mm and 6 · 6 mm scans centered on the ONH were captured for blood FIG. 1. Skeletonization of OCT-A images and retinal blood flow analysis. A. OCT-A images captured using Zeiss Angioplex software. B. Skeletonized conversion with 3.4-mm circle delineating area of interest for analysis. C. OCT-A flow image slab used to derive retinal and choroidal blood flow measurements. D. Pixels with a red hue value were selected using Adobe Photoshop Software and are indicated in white. The purple dotted line delineates the boundaries of the retinal layer. Rodriguez Torres et al: J Neuro-Ophthalmol 2021; 41: 48-53 49 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 1. Patient demographics and papilledema grading P Age (years) OD (%) Eyes (n) Papilledema grade 0 1 2 3 4 Race (n) African American Caucasian Hispanic Middle Eastern 30.8 ± 12.0 23 (50.0%) 46 0.628 0.496 14 (30.4%) 10 (21.7%) 11 (23.9%) 7 (15.2%) 4 (8.7%) 0.628 33 (71.7%) 7 (15.2%) 2 (4.3%) 4 (8.7%) flow measurements. Patients with poor image quality including motion artifacts, poor clarity, or signal strength less than 7 were excluded. Optical Coherence Tomography-Angiography Vessel Skeletonization OCT-A images were analyzed using the publicly available ImageJ software (v1.51w) (Fig. 1A, B). Images were converted into skeletonized forms using the Skeletonize function. A circle of 3-mm width was drawn around the ONH to delineated the peripapillary capillary region. The resulting skeleton around the nerve head was then analyzed using the Analyze Skeleton plug-in (imagej.net/AnalyzeSkeleton), without elimination of loops or end-points. This process was repeated for each individual quadrant, using the Quadrant Picking plug-in (imagej.nih.gov/ij/plugins/quadrantpicking/index.html). Branch lengths of each skeleton from a single image were summed, which was then divided by the area inside the circle to derive the skeletonized vessel density (SVD). These methods are modified from previously published article by Falavarjani et al (13) on quantitative OCTA image analysis. Blood Flow Measurement Cross-sectional images of retinal blood flow (RBF) and choroidal blood flow (CBF) were analyzed using Photoshop software (Adobe Systems, CA) (Fig. 1C, D). The color range function was used to sample all red pixels in both the retina and choroid, and the total pixel count was recorded. The exact hue values of the selected red pixels were maintained between each image for consistency. A count of the total number of red pixels within the retinal or choroidal layers was recorded as total retinal blood flow (RBF) and total CBF, respectively. These values were divided by the total red pixel count within the entire slab and recorded as RBF percentage (RBF%) and CBF percentage (CBF%), respectively. 50 TABLE 2. Correlation of RNFL thickness or skeletonized vessel density (SVD) with papilledema grade b R RNFL Overall SVD Inferior SVD Nasal SVD Superior SVD Temporal SVD 0.677 0.512 0.339 0.457 0.420 0.427 0.019 20.115 20.263 20.388 20.400 20.324 ± ± ± ± ± ± P Slope 0.003 ,0.001 0.02 0.029 ,0.001 20.11 0.110 0.021 20.26 0.114 0.001 20.39 0.130 0.004 20.40 0.103 0.003 20.32 RNFL, retinal nerve fiber layer. Correlation Analysis Linear best-fit lines inclusive of all data points were generated comparing RNFL thickness, SVD, and RBF/ CBF values as the number of pixels or as a percentage of the blood flow to the entire slab to the papilledema grade. To assess the changes in RBF/CBF values with an increase of grade, individual linear best-fit lines were also created for RBF% and CBF% of eyes with Grades 0 and 1, Grades 1 and 2, Grades 2 and 3, and Grades 3 and 4. Statistical Analysis SPSS software version 25 was used for statistical analysis (SPSS Inc, IBM). The Mann–Whitney bilateral test was used for comparison of continuous data. Correlations were tested using the Spearman correlation coefficient. P values ,0.05 were considered statistically significant. RESULTS After excluding 8 eyes for poor-quality OCT-A image, a total of 46 eyes from 23 female patients (mean age, 30.8 ± 12.0 years) were included in the analysis. Table 1 summarizes the baseline characteristics of the study participants. There were no statistically significant differences, as assessed by one-way analysis of variance, in age, right vs left eye, or race. OCT RNFL thickness was found to be significantly associated with papilledema grade (R = 0.677, b = 0.019 ± 0.003; P , 0.001). SVD peripapillary capillary plexus was significantly associated with papilledema grades (Fig. 2). Table 2 shows the correlation between grading and vessel density. Every point increase in grade was correlated with a decrease of 9.1 pixels/mm2 in vessel density (R = 0.512, b = 20.115 ± 0.029; P , 0.001). RNFL thickness was found to be significantly associated with SVD (R = 0.336, b = 22.632 ± 1.113; P = 0.022). Every point increase in RNFL thickness was correlated with a SVD decrease of 0.38 pixels/ mm2. Macular ganglion cell layer (GCL) thickness using retinal ganglion cell-inner plexiform layer thickness was found to be significantly associated with SVD. Decreasing Rodriguez Torres et al: J Neuro-Ophthalmol 2021; 41: 48-53 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution FIG. 2. Correlation of RNFL thickness or skeletonized vessel density (SVD) with papilledema grade. A. Papilledema grade in comparison with RNFL thickness was found to have a positive correlation (R = 0.677). B–F. Papilledema grade was negatively correlated with SVD of the entire peripapillary area, as well as SVD of the inferior, nasal, superior, and temporal quadrants (R = 0.512, 0.339, 0.457, 0.420, 0.427, respectively). Raw data corresponding to these plots are presented in Table 2. RNFL, retinal nerve fiber layer. GCL thickness was associated with decreasing SVD (R = 0.408, b = 21.804 ± 0.623, P = 0.006). RBF/CBF, as a percentage relative to total blood flow to both the retina and choroid, was correlated with increasing papilledema grades (Fig. 3). Table 3 depicts the correlation between grade and blood flow. A worsening papilledema was correlated with an increased RBF % (R = 0.300, b = 2.114 ± 1.013; P , 0.05) and decreased CBF% (R = 0.300, b = 2.114 ± 1.013; P , 0.05). Every point increase in grade was correlated with an increase in 47.4% RBF% and decrease in 47.4% CBF %, as calculated using a linear best-fit line inclusive for all of the data points. Therefore, increasing papilledema Grade 0–3 was correlated with an increased RBF and FIG. 3. Correlation of retinal blood flow (RBF) or choroidal blood flow (CBF) with papilledema grade. A and B. Blood flow area in the retina as a fraction of blood flow area in the entire slab was calculated as RBF%, with CBF% calculated in a similar manner. Papilledema grade in comparison with RBF% was found to have a positive correlation (R = 0.300), whereas papilledema grade in comparison with CBF% was found to have a corresponding negative correlation (R = 0.300). C and D. Total RBF was quantified as the total number of pixels representing blood flow area in the retina, with total CBF representing the area in pixels of blood flow in the choroidal layer. Papilledema grade in comparison with total RBF was found to have a nonsignificant correlation (R = 0.025), whereas papilledema grade in comparison with total CBF was found to have a significant negative correlation (R = 0.477). Raw data corresponding to these plots are presented in Table 3. Rodriguez Torres et al: J Neuro-Ophthalmol 2021; 41: 48-53 51 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution TABLE 3. Correlation of OCT-A blood flow slab with papilledema grade RBF CBF RBF CBF (%) (%) total total R b P Slope 0.300 0.300 0.025 0.477 2.114 ± 1.013 22.114 ± 1.013 ,0.001 ,0.001 0.043 0.043 0.867 0.001 2.11 22.11 0.00001 20.0002 OCT-A, optical coherence tomography angiography; RBF, retinal blood flow; CBF, choroidal blood flow. a corresponding decreased CBF. Further analysis using linear best-fit lines inclusive of only 2 sequential grades showed that CBF% decreased by 2.04%, 2.04%, and 2.22%, with a change in papilledema grade from 0 to 1, 1 to 2, and 2 to 3, respectively (Fig. 4). Correlation between the number of pixels representing blood flow to the retina or choroid (RBF Total or CBF Total, respectively) relative to papilledema grades was also examined. The total number of pixels representing blood flow to the choroid was found to be significantly correlated with increasing papilledema grades, with every point increase in grade correlated with a decrease in 5,000 pixels representing choroid flow. The total number of pixels representing blood flow to the retina correlated with RNFL thickness, but not with papilledema grades. An increase in RBF was correlated with an increase in RNFL thickness, with an increase in 1.34% RBF associated with an increase in one 1 mm in RNFL thickness (R = 0.303, b = 74.692 ± 35.443; P , 0.05). DISCUSSION Our study demonstrates a significant decrease in SVD, which is significantly associated with papilledema grading. These results are in concordance with previous studies noting significant evidence of decreased peripapillary capillary vessel density with papilledema and other forms of disc edema (14–16). Previous studies reported that vessel density in lower grade (i.e., 1 or 2) papilledema did not differ from healthy eyes, whereas there was a significant decrease in vessel density with high-grade papilledema.14 In addition, studies comparing papilledema with acute NAION reported differences in the degree of vessel density loss along with different pattern defects (14,16). These findings show the potential of OCT-A as a diagnostic tool for distinguishing papilledema from other forms of disc edema. We showed a decrease in the skeletonized peripapillary density that correlated with increase in papilledema grading and OCT RNFL thickness. In papilledema, impairment of intra-axonal fluid results in swelling of the axons along with leakage into the extracellular space of the optic disc. The ciliary arterial circle is susceptible to pressure changes, which suggests that the observed decreased peripapillary vessel density could be due to the compression phenomenon. These results are of importance as it supports the hypothesis that optic neuropathy in papilledema could be due to a relative ischemic state which over the long term could cause damage to the peripapillary retina and optic disc resulting in atrophy and gliosis. Analysis of the OCT-A slab blood flow provided further information for understanding the vascular pathophysiology FIG. 4. Average change in OCT-A blood flow slab with change in papilledema grade 0–3. Change in choroidal blood flow percentage (CBF%) was found to decrease by 2.04%, 2.04%, and 2.22% with a change in papilledema grade from 0 to 1, 1 to 2, and 2 to 3, respectively. OCT-A, optical coherence tomography angiography. 52 Rodriguez Torres et al: J Neuro-Ophthalmol 2021; 41: 48-53 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited. Original Contribution of the decreased vessel density. Our results were significant for a negative correlation between CBF and papilledema grading. These results suggest that accumulation of extracellular retinal fluid could lead to a mechanical impairment of CBF. In active papilledema, it is difficult to assess permanent damage to optic nerve axons based on RNFL measurements alone. However, thinning of the GCL is a useful measure of optic nerve injury, even in the presence of a swollen optic disc. Therefore, we studied the effects of papilledema grading on GCL thickness and SVD. We found a strong correlation between decreased vessel density and decreased GCL thickness. Furthermore, this evidence shows the potential of OCT-A SVD as a clinical marker for detecting early optic nerve damage. Limitations of our study are the small sample size, the retrospective nature of our study, and the possibility of shadowing artifacts from the peripapillary fluid on the OCT-A. In addition, we only had 4 eyes with a papilledema grading of 4 which limits our possibility to detect statistically significant differences in that population. Despite these limitations, we did find significant differences in the peripapillary capillary plexus of patients with papilledema. Overall, OCT-A has the potential to quantitatively analyze vessels density, identify areas of nonperfusion, and detect early ischemic changes in patients with IIH before optic atrophy and gliosis become apparent. Our results suggest that further studies of retinal and CBF in patients with papilledema caused by IIH may be useful in understanding the visual complications in this disease. REFERENCES 1. Mollan SP, Aguiar M, Evison F, et al. The expanding burden of idiopathic intracranial hypertension. Eye (Lond). 2019;33:478–85. 2. Friesner D, Rosenman R, Lobb B, Tanne E. Idiopathic intracranial hypertension in the USA: the role of obesity in establishing prevalence and healthcare costs. Obes Rev. 2011;12:e372–e80. Rodriguez Torres et al: J Neuro-Ophthalmol 2021; 41: 48-53 3. Corbett JJ, Savino PJ, Thompson HS, et al. Visual loss in pseudotumor cerebri: follow-up of 57 patients from five to 41 years and a profile of 14 patients with permanent severe visual loss. Arch Neurol. 1982;39:461–474. 4. Bruce BB, Biousse V, Newman NJ. Update on idiopathic intracranial hypertension. Am J Ophthalmol. 2011;152:163– 169. 5. Arnold AC, Badr MA, Hepler RS. Fluorescein angiography in nonischemic optic disc edema. Arch Ophthalmol. 1996;114:293–298. 6. Collignon-Robe NJ, Feke GT, Rizzo JF III. Optic nerve head circulation in nonarteritic anterior ischemic optic neuropathy and optic neuritis. Ophthalmology. 2004;111:1663–1672. 7. Riva CE. Basic principles of laser Doppler flowmetry and application to the ocular circulation. Int Ophthalmol. 2001;23:183–189. 8. Sugiyama T, Araie M, Riva CE, et al. Use of laser speckle flowgraphy in ocular blood flow research. Acta Ophthalmol (Copenh). 2010;88:723–729. 9. Spaide RF, Klancnik JM, Cooney MJ. Retinal vascular layers imaged by fluorescein angiography and optical coherence tomography angiography. JAMA Ophthalmol. 2015;133:45–50. 10. Jia Y, Morrison JC, Tokayer J, et al. Quantitative OCT angiography of optic nerve head blood flow. Biomed Opt Express. 2012;3:3127–3137. 11. Hata M, Oishi A, Muraoka Y, et al. Structural and functional analyses in nonarteritic anterior ischemic optic neuropathy: optical coherence tomography angiography study. J Neuroophthalmol. 2017;37:140–148. 12. Wang X, Jia Y, Spain R, et al. Optical coherence tomography angiography of optic nerve head and parafovea in multiple sclerosis. Br J Ophthalmol. 2014;98:1368–73. 13. Falavarjani KG, Tian JJ, Akil H, et al. Swept-source optical coherence tomography angiography of the optic disk in optic neuropathy. Retina. 2016;36:S168–S77. 14. Fard MA, Jalili J, Sahraiyan A, et al. Optical coherence tomography angiography in optic disc swelling. Am J Ophthalmol. 2018;191:116–123. 15. Gaier ED, Wang M, Gilbert AL, et al. Quantitative analysis of optical coherence tomographic angiography (OCT-A) in patients with non-arteritic anterior ischemic optic neuropathy (NAION) corresponds to visual function. PLoS One. 2018;13:e0199793. 16. Gandhi U, Chhablani J, Badakere A, et al. Optical coherence tomography angiography in acute unilateral nonarteritic anterior ischemic optic neuropathy: a comparison with the fellow eye and with eyes with papilledema. Indian J Ophthalmol. 2018;66:1144. 53 Copyright © North American Neuro-Ophthalmology Society. Unauthorized reproduction of this article is prohibited.