Optical Coherence Tomography Angiography for Anterior Segment Vasculature Imaging Marcus Ang, FRCSEd,1,2 Dawn A. Sim, FRCOphth,3 Pearse A. Keane, FRCOphth,3,4 Chelvin C.A. Sng, FRCSEd,1,5 Catherine A. Egan, FRCOphth,3 Adnan Tufail, FRCOphth,3 Mark R. Wilkins, FRCOphth2,4 Purpose: To evaluate the application of an optical coherence tomography angiography (OCTA) system adapted for the assessment of anterior segment vasculature. Design: Cross-sectional, observational study. Participants: Consecutive subjects with normal eyes on slit-lamp clinical examination and patients with abnormal corneal neovascularization. Methods: All scans were performed using a commercially available AngioVue OCTA system (Optovue, Inc., Fremont, CA) using an anterior segment lens adapter and the split-spectrum amplitude decorrelation angiography algorithm. Each subject underwent scans from 4 quadrants (superior, inferior, nasal, and temporal) in each eye by 2 trained, independent operators. Main Outcome Measures: Analysis of signal strength, image quality, and reproducibility of corneal vascular measurements was performed. Results: In our study of 20 normal subjects (10 men, 10 women; mean age, 25.3
7.8 years), we found good repeatability (k coefficient, 0.76) for image quality score and good interobserver agreement for vasculature measurements (intraclass coefficient, 0.94). After optimization of the angiography scan protocol, vascular measurements within the regions of interest were compared in the superior versus inferior quadrants (mean vascular loops, 3.34
1.16 vs. 3.12
0.90 [P ¼ 0.768]; segment-to-loop ratio, 4.18
0.71 vs. 4.32
0.87 [P ¼ 0.129]; fractal dimension [Df] value, 1.78
0.06 vs. 1.78
0.06 [P ¼ 0.94]; vascular loop area, 25.9
14.5 vs. 25.9
10.7  103 mm2 [P ¼ 0.21]) and nasal versus temporal quadrant (mean vascular loops, 2.89
0.98 vs. 3.57
0.99 [P < 0.001]; segment-to-loop ratio, 3.94
0.69 vs. 4.55
0.78 [P ¼ 0.897]; Df value, 1.78
0.06 vs. 1.77
0.06 [P ¼ 0.14]; vascular loop area, 29.7
15.7 vs. 22.1
7.1  103 mm2 [P ¼ 0.38]. We then used the established OCTA scanning protocol to visualize abnormal vasculature successfully in 5 patients with various corneal pathologic features, including graft-associated neovascularization, postherpetic keratitis scarring, lipid keratopathy, and limbal stem cell deficiency. Conclusions: This preliminary study describes a method for acquiring OCTA images of the cornea and limbal vasculature with substantial consistency. This technique may be useful for the objective evaluation of corneal neovascularization in the future. Ophthalmology 2015;-:1e8 ª 2015 by the American Academy of Ophthalmology.

Since its first in vivo use for the retina, optical coherence tomography (OCT) imaging has revolutionized our ability to evaluate the eye and its structures on a microscopic level.1 It also has been established as a useful tool in providing rapid, noncontact evaluation of the cornea and anterior segment.2,3 Further technological developments also have increased the imaging capabilities of OCT in terms of speed, image resolution, and, more recently, evaluation of vascular flow.4 These new OCT imaging techniques have been described for noninvasive evaluation of vessels within the retina and optic disc.4,5 As opposed to Doppler OCT techniques, which depend on axial flow of blood, OCT angiography (OCTA) techniques visualize vessels via motion contrast imaging of erythrocyte movement across sequential B-scans.4 Currently, assessment of the corneal and anterior segment vasculature is constrained to slit-lamp photography

 2015 by the American Academy of Ophthalmology Published by Elsevier Inc.

or angiography techniques using fluorescein or indocyanine green.6 Semiautomatic quantitative techniques to analyze photographic images have been described,7 with standardized methods of quantification of corneal neovascularization established and used in several clinical trials.8e10 However, underestimation of poorly visible blood vessels may occur, especially in the presence of dense corneal scars,11 whereas invasive angiography techniques expose patients to potential adverse reactions.6 Thus, methods of evaluating abnormal corneal neovascularization remain important, given its prevalence and potential sightthreatening effects,12,13 where visual loss may ensue from associated corneal edema, scarring, and lipid deposition.14 As new antiangiogenic treatments for various corneal pathologic features with associated vascular changes are being developed,15 a recent roundtable expert review

http://dx.doi.org/10.1016/j.ophtha.2015.05.017 ISSN 0161-6420/15

1

Ophthalmology Volume -, Number -, Month 2015 identified the development of new imaging techniques for the evaluation of corneal neovascularization as an important, unmet need.16 Therefore, we conducted a proof-of-concept study to evaluate the feasibility of using an OCTA device intended for retinal vessel imaging for the corneal and limbal vasculature. Although this imaging technique has been used to evaluate retinal vascular pathologic features such as choroidal neovascularization,17 to our knowledge, its role has not been reported for use in the anterior segment at the time of this publication. We then used the derived OCTA scanning protocol to evaluate and describe quantitatively abnormal corneal neovascularization in patients with various corneal pathologic features in a preliminary clinical study.

Methods In the first phase of the study, to establish the OCTA scanning protocol for the anterior segment, we performed OCTA in 20 subjects with no ocular history and a normal slit-lamp examination at the Moorfields Eye Hospital from October 1, 2014, through December 31, 2014. Our study followed the principles of the Declaration of Helsinki, with ethics approval obtained from the local institutional review board. All subjects underwent imaging using the split-spectrum amplitude decorrelation angiography algorithm on the on the AngioVue OCTA system (Optovue, Inc., Fremont, CA) intended for retina imaging (AngioRetina mode), but with the anterior segment optical adaptor lens. Each scan was performed with axial resolution of 5 mm and a beam width of 22 mm, with a light source centered on 840 nm. The instrument captures consecutive Bscans containing 304304 A-scans at 70 000 scans per second in a slow transverse direction, which constructs a 3-dimensional scan cube in approximately 3 to 4 seconds.4 Because the default focus of the system is for the retina, the autofocus function was turned off, and manual adjustments to the XYZ focal lengths had to be made until the vessels of interest were seen clearly in focus on the camera image. Because this often led to the anterior segment lens adapter being just 2 to 4 cm from the subject’s anterior corneal surface, care had to be taken to prevent contact with the eye. For the proof-of-concept phase of the study, all subjects underwent OCTA scans (66-mm volume cubes) in 4 quadrants of the cornea limbus (superior, inferior, temporal, and nasal) in both eyes by 2 independent trained operators.

section between 2 branch points or a terminal point),19 fractal dimension (Df) value, and the area enclosed by each vascular loop (square millimeters).19,20 The quality of the scan images were assessed using the signal strength index and image quality score using a recognized system, that is, 0 to 4 (0, no vessel discernible; 1, poor vessel delineation; 2, good vessel delineation; 3, very good vessel delineation; 4, excellent vessel delineation) on 2 scans per quadrant, performed by 2 independent, masked assessors.20 We then used the scan protocol as derived above to perform OCTA imaging in 5 patients with various corneal pathologic features, namely, corneal neovascularization in a corneal graft, pterygium, postherpetic keratitis scar, lipid keratopathy, and limbal stem cell deficiency.

Statistical Analysis To evaluate the scan protocol described, we analyzed all scan images obtained for repeatability, image quality, and vascular measurements between quadrants. We calculated the k coefficient value for the repeatability of scans using the image quality score, where k  0.2 was considered slight, k ¼ 0.21 to 0.40 was considered fair, k ¼ 0.41 to 0.6 was considered moderate, k ¼ 0.61 to 0.8 was considered substantial, and k ¼ 0.81 to 1.0 was considered almost perfect in agreement.21 The intraclass coefficient (ICC) was calculated for vascular measurements performed by both independent assessors, evaluating the mean number of vascular loops and segment-to-loop ratios in the ROI of the scans in each quadrant. Statistical analysis also included descriptive statistics, where the mean
standard deviation was calculated for the continuous variables and compared using the ManneWhitney U test. A P value less than 0.05 was considered statistically significant for comparisons between 2 quadrants. Statistical Package for the Social Sciences software version 17.0 (SPSS, Inc., Chicago, IL) was used to analyze the data.

Image Analysis The best scans were processed automatically to reduce motion artifacts such as transverse saccadic and residual axial motion by the internal software (ReVue version 2014.2.0.15; Optovue, Inc.). Next, images were exported from the system as a portable network graphics image file into ImageJ 1.38X software (National Institutes of Health, Bethesda, MD) for analysis, similar to a previously described method for quantifying corneal neovascularization.18 In brief, we identified the regions of interest (ROIs) in each scan by labeling 5 contiguous squares (100 pixels; 850 mm) spanning a circumferential arc length of approximately 4 mm, aligned along the circumference of the anterior border of the corneal limbus (Fig 1). We then used a selective filter to highlight the linear structures such as blood vessels and to reduce the surrounding noise to export the resulting vascular tree as a binary image for further processing, similar to a previously described method.19 Each ROI was assessed for the number of vascular loops (as illustrated in Fig 1),6 vessel segments (previously defined as the

2

Figure 1. Optical coherence tomography angiograms of the cornea and anterior segment. A, C, D, Examples of whole-depth, split-spectrum amplitude decorrelation optical coherence tomography angiograms with en face maximum projection of the corneolimbal vasculature in various quadrants of the cornea. B, The binary image was processed with a selective filter and regions of interest along the corneal vascular arcade using a method previously described; the white arrow points to an example of a marginal corneal vascular loop.12

Ang et al



Anterior Segment OCT Angiography

Figure 2. Optical coherence tomography angiograms showing neovascularization in a corneal graft. Abnormal corneal neovascularization associated with corneal graft rejection in an eye with a penetrating keratoplasty is much clearer on (B, D) optical coherence tomography angiography scans compared with (A) slit-lamp examination and (C) a red-free image.

Results The first phase of our study was performed in both eyes of 20 subjects (10 men, 10 women; mean age, 25.25
7.87 years) with normal slit-lamp examination results and no evidence of abnormal corneal vessels. Overall, we found substantial repeatability of scans in all quadrants in terms of image quality score (k ¼ 0.76), with good interobserver agreement for vasculature measurements (ICC, 0.94). Superior and inferior quadrant scans showed a similar signal strength index (39.04
15.06 vs. 39.28
16.71; P ¼ 0.89), whereas temporal scans had a higher signal strength index compared with nasal scans (45.23
15.23 vs. 33.09
14.14; P < 0.001). We compared vascular measurements within the ROI of the superior and inferior quadrants (mean vascular loops: 3.34
1.16 vs. 3.12
0.90 [P ¼ 0.768]; segment-to-loop ratio, 4.18
0.71 vs. 4.32
0.87 [P ¼ 0.129]; Df value, 1.78
0.06 vs. 1.78
0.06 [P ¼ 0.94]; vascular loop area, 25.9
14.5 vs. 25.9
10.7  103 mm2 [P ¼ 0.21]), as well as nasal versus temporal quadrants (mean vascular loops, 2.89
0.98 vs. 3.57
0.99 [P < 0.001]; segment-toloop ratio, 3.94
0.69 vs. 4.55
0.78 [P ¼ 0.897]; Df value, 1.78
0.06 vs. 1.77
0.06 [P ¼ 0.14]; vascular loop area, 29.7
15.7 vs. 22.1
7.1  103 mm2 [P ¼ 0.38]). We then used the derived scan protocol to evaluate the OCTA system in various clinical conditions with abnormal corneolimbal neovascularization. Figure 2 depicts the eye in a 55-year-old man who underwent a penetrating keratoplasty for bullous keratopathy and sought treatment 6 months after

surgery, with signs of an early rejection of his corneal graft. Upon slit-lamp examination, there was mild anterior chamber inflammation with an epithelial defect, stromal edema with keratic precipitates, and early corneal neovascularization at the graftehost junction from 12 to 3 o’clock (Fig 2A). The OCTA system was able to outline the corneal vessels that were invading the corneal graft, which were not as clearly visible on the slit-lamp images (Fig 2B, D). Figure 3A, B demonstrates the OCTA scans of a 60-year-old woman with a pterygium encroaching on the visual axis, with the abnormal invasive vessels clearly delineated in the fibrovascular tissue. We also documented a case of previous herpetic keratitis in a 28-year-old woman with residual stromal scar and corneal neovascularization and a clear visualization of the abnormal vessels invading the corneal stroma on the OCTA scans that were not as clearly seen on slit-lamp images (Fig 3C, D). We also observed that one of the advantages of using the OCTA for anterior segment pathologic features is that it provides simultaneous assessment of the depth of the lesion, as well as its associated abnormal blood vessels. Figure 4 depicts serial cross-sectional OCTA images at various depths in the same eye with a nasal pterygium from Figure 3, while also showing the incursion of the abnormal fibrovascular tissue into the corneal stroma. The OCTA images also show the abnormal vascular loops, which now form at the head of the pterygium, associated with a relative loss of the normal corneal vascular arcades at the corneal limbus underneath the pterygium wing. Another feature of the OCTA technique is

3

Ophthalmology Volume -, Number -, Month 2015

Figure 3. Slit-lamp images and optical coherence tomography angiograms of clinical examples with aberrant anterior segment vessels. Illustration of the abnormal fibrovasculature structure of a pterygium, where abnormal blood vascular loops at the head of the pterygium are delineated by (B) the optical coherence tomography angiography (OCTA) scan using the 33-mm scan function, which are less apparent on (A) slit-lamp photography. The fibrovascular structure of the pterygium may make it difficult to differentiate stromal fibers from blood vessels in the remaining body of the pterygium, limited by the current built-in image processing software. C, D, A case of postherpetic keratitis scar, where the corneal neovascularization invading the corneal stroma is not visualized clearly on (C) slit-lamp examination but is seen clearly on (D) the OCTA scan. However, although the limited field of view from the scan suggests that vascular loops of the marginal arcade of the adjacent limbus are retained, the OCTA is unable to differentiate the origin of the abnormal neovascularization or the margins or demarcation of the normal and abnormal vessels.

demonstrated in Figure 5: OCTA scans reveal abnormal feeder vessels (Fig 5B), which are obscured by the dense lipid deposits and scarring in the slit-lamp images (Fig 5A), in an eye with lipid keratopathy from previous interstitial keratitis in a 45-year-old man. The ability to detect subtle vascular changes also is seen in Figure 5C, D, where a localized area of corneal neovascularization is outlined within the corneal scar on OCTA, with disruption of the normal corneal marginal vascular arcades inferior to the scar, of an eye of a 62-year-old man with limited limbal stem cell deficiency from long-standing ocular rosacea.

Discussion In this article we describe an application of OCTA technology for noninvasive imaging of the anterior segment vasculature. Because this is a relatively new technology developed for retinal use, we initially conducted a proof-ofconcept study of eyes of healthy subjects to determine the scan protocol, which produced substantially consistent scans of the corneolimbal vessels (k value, 0.76; ICC, 0.94). Although our corneal vascular measurements were similar

4

in terms of consistency with those of a previous corneal indocyanine green angiography (ICGA) study (ICC, 0.67e0.83),19 we recognize that several aspects such as age, gender, or other ocular factors may lead to variation in loop size and branching between subjects, even for similar corneal quadrants.19 However, the aim of this study was not to describe the corneal vasculature in detail, but rather to demonstrate that this OCTA system may be adapted for anterior segment vasculature imaging with adequate consistency. We used the same scanning protocol to image eyes with abnormal corneal vasculature to confirm that the OCTA could successfully image aberrant blood vessels amid various important pathologic features, such as vascularization in a corneal graft that is associated with corneal graft rejection22 and postherpetic keratitis, where its treatment is increasingly being evaluated using new antiangiogenic therapies.16 Normal corneal and limbal vasculature anatomic features have been described using fluorescein angiography (FA) and, more recently, ICGA.23e25 Because abnormal corneal neovascularization may arise from the corneal marginal arcade or

Ang et al



Anterior Segment OCT Angiography

Figure 4. Serial cross-sectional optical coherence tomography angiography (OCTA) scans of a pterygium at varying corneal and scleral depths allow for simultaneous assessment of the structure, depth of corneal stromal invasion, and the abnormal blood vessels in this fibrovascular lesion. Left and middle columns, The OCTA scans delineating stromal fibers and blood vessels with a coronal or en face view at varying depths. (Right column) Scans show the traditional cross-sectional or B-scan view. The current OCTA built-in software allows the user to examine multiple, consecutive, cross-sectional and en face scans to reconstruct a 3-dimensional evaluation of the lesion and its associated vasculature.

the limbal palisades vessels, objective imaging of the corneal vasculature is important to understand and potentially treat the associated abnormal vascular supply of corneal pathologic features.25,26 Our findings support previous observations that vessels in the periphery of the cornea comprise a fine capillary network consisting of vascular loops with 3 to 4 branches, which are not as evident on slit-lamp images.19 Although a previous study of corneal vascular arcades using ICGA was described in detail, these measurements were from various random quadrants of the corneal limbus.19 In our study, we compared scans from each quadrant and found that there was a higher density of vascular loops in the temporal

marginal corneal arcades compared with those in the nasal quadrants. This may provide interesting insights into the potential significance of these differences in corneolimbal vasculature, particularly with regard to pathologic features, such as pterygia, that more frequently affect the nasal aspects of the anterior segment. Moreover, in comparing the corneal vasculature in normal eyes with various pathologic features such as postherpetic scarring and limbal stem cell deficiency, we observed qualitative changes to the normal corneal marginal arcades, which may have potential for prognostic evaluation in these conditions. However, future prospective studies of corneal and limbal vascular anatomic

5

Ophthalmology Volume -, Number -, Month 2015

Figure 5. Slit-lamp images and optical coherence tomography angiograms of clinical examples of abnormal corneal neovascularization. Demonstration of corneal neovascularization in a case of lipid keratopathy is clearly imaged by (B) the optical coherence tomography angiography (OCTA) system but is not apparent on (A) the slit-lamp image because of the dense stromal scarring and lipid deposits. Note the artifacts in this scan, produced by the intrastromal lipid deposition, which may either obscure or appear as abnormal vessels. This may be discerned from actual vessels by scrolling between scans and various depths in the en face function and correlating with slit-lamp photographs. C, Abnormal corneal and limbal vessels are seen easily in a case of localized limbal stem cell deficiency in this example of an eye with ocular rosacea. D, Early signs of disruption of the normal corneal vascular arcades on the OCTA scans are recognized as being more extensive than on clinical slit-lamp images.

features correlating OCTA with established imaging techniques used in clinical trials of antiangiogenic drugs for corneal vasculature8e10 probably are required to study these observations in greater detail using quantitative assessments. Although FA and ICGA have been useful in visualizing areas of corneal neovascularization,20 these invasive tests are performed infrequently for the anterior segment compared with the posterior segment. Although FA potentially is able to detect abnormal corneal vessel maturity and inflammatory activity, the leakage of dye may affect vessel evaluation and measurements.6 However, indocyanine green is a larger molecule that stays within the corneal vasculature, thereby improving delineation of blood vessels, even in the presence of corneal stromal scars.20 Nevertheless, both FA and ICGA are invasive procedures that can be both time consuming and relatively difficult to perform, with a significant learning curve, exemplified by one corneal ICGA study in which up to 40% of images had to be excluded.20 Moreover, patients may have contraindications to dye injections, and clinicians have to be prepared for rare but serious adverse reactions associated with the intravenous administration of fluorescein or indocyanine green dyes.27,28

6

Optical coherence tomography angiography provides a promising rapid and noninvasive alternative to FA and ICGA that already has useful clinical indications such as detecting and monitoring inflammation, vascularization, or areas of damage that are not clinically visible.25 However, one of the added benefits of OCTA is that it acquires high-resolution cross-sectional images, simultaneously providing information on the depth of both the corneal pathologic features and the penetration of its associated abnormal vessels. This potentially aids preoperative surgical planning for corneal lesions, such as deciding on the depth and location of feeder vessel diathermy before anterior lamellar keratoplasty for postherpetic scars or a quick assessment of an invasive pterygium to evaluate the depth of the fibrovascular incursion into the corneal stroma (Fig 4). With the use of image processing and objective assessment of areas of vessel invasion,6 OCTA also may be useful for monitoring vascular changes before and after treatment when assessing new therapies for abnormal corneal neovascularization.16 Furthermore, the potential clinical applications of anterior segment OCTA may not be limited to the cornea; for example, evaluating conjunctival and scleral vasculature may add further

Ang et al



Anterior Segment OCT Angiography

information to bleb morphologic features after glaucoma surgery assessed by OCT.29 The OCTA system used in this study uses decorrelative techniques to detect phase variations and OCT signal intensity that result from blood flow, which are independent of flow orientation over a larger range of flow velocities.4 Using the scanning protocol described above, we found that corneal vessels were outlined in the presence of superficial corneal scarring or lipid deposition. However, deeper vessels in corneal opacities may not be delineated as clearly by the OCTA system, and this requires further study in a larger series of cases with corneal pathologic features, with comparisons of FA and ICGA. Clinicians also should note that current OCTA technology may produce scans that contain artifacts that appear as abnormal vessels (Fig 5). These artifacts may be distinguished from actual abnormal corneal vessels by examining multiple consecutive scans in the en face function and correlating with slit-lamp photographs. Other limitations of this OCTA system include an image resolution that is insufficient for distinguishing normal vessels from abnormal vessels, the inability to demonstrate vessel leakage, and a limited field of view compared with current angiography techniques. Thus, in its current iteration, the system may not be useful for evaluating large areas of neovascularization compared with standardized slit-lamp photography techniques established for clinical trials.9 Moreover, current OCTA systems are optimized for retinal imaging; thus, we had to make adjustments to the scanning technique while using the anterior segment adapter. Because the lens had to be relatively close to the surface of the cornea for the vessels to be in focus, image acquisition was relatively easier in the temporal quadrants compared with the nasal quadrants, as illustrated by the differences in signal strength index. Nonetheless, our initial experience with the AngioVue OCTA system revealed a relatively easy learning curve, rapid acquisition (3e4 seconds per scan) of images of the cornea, and good overall reproducibility. As a noninvasive imaging technique that is acquired with relative ease, it has the potential to be implemented easily as part of routine clinical evaluation. This is encouraging because further streamlining and development of OCTA technology and its software, in particular for the anterior segment, may improve image resolution, shorten acquisition time with a larger field of view, or even measure variations in oxygen saturation within the vessels.30 We recognize limitations of our preliminary crosssectional study, adapting the use of this OCTA system for the cornea and anterior segment. A large prospective study of various corneal pathologic features with comparisons with slit-lamp photography, invasive angiography techniques, or histologic images would have been ideal to evaluate this OCTA technology specifically for the anterior segment. Moreover, the existing OCTA image analysis software comes with inherent limitations, such as inability to analyze serial scans at fixed landmarks on the cornea (registration or labeling is not available currently), assumptions of the vascular plane, and the underestimation of vessel density or smaller vessels because of the threshold

method used.19 Future challenges to improve the clinical application of OCTA for corneal vasculature include the need to develop standardized techniques with filters or algorithms to highlight linear structures and reduce surrounding noise in the images, with analysis and monitoring of changes in neovascularization over time for various corneal pathologic features. Nevertheless, we present unprecedented study findings that suggest this rapid, noninvasive OCTA system has the potential to produce consistent images of the anterior segment vasculature, which may aid in further understanding of the role of abnormal neovascularization in various corneal, conjunctival, and scleral diseases.24 In summary, we describe results from a preliminary evaluation of an OCTA system, potentially useful for the imaging of both normal and abnormal corneal vessels. Our study suggests that the prospects of this rapid, noninvasive angiography technique for wider clinical applications in the anterior segment are promising. Further studies are required to confirm whether this imaging technique is useful for the objective evaluation of corneal neovascularization compared with other angiography techniques.

References 1. Swanson EA, Izatt JA, Hee MR, et al. In vivo retinal imaging by optical coherence tomography. Opt Lett 1993;18:1864–6. 2. Ang M, Chong W, Tay WT, et al. Anterior segment optical coherence tomography study of the cornea and anterior segment in adult ethnic South Asian Indian eyes. Invest Ophthalmol Vis Sci 2012;53:120–5. 3. Ang M, Chong W, Huang H, et al. Comparison of anterior segment optical tomography parameters measured using a semi-automatic software to standard clinical instruments. PLoS One 2013;8:e65559. 4. Spaide RF, Klancnik JM Jr, Cooney MJ. Retinal vascular layers imaged by fluorescein angiography and optical coherence tomography angiography. JAMA Ophthalmol 2015;133:45–50. 5. Jia Y, Wei E, Wang X, et al. Optical coherence tomography angiography of optic disc perfusion in glaucoma. Ophthalmology 2014;121:1322–32. 6. Kirwan RP, Zheng Y, Tey A, et al. Quantifying changes in corneal neovascularization using fluorescein and indocyanine green angiography. Am J Ophthalmol 2012;154:850–858e2. 7. Bock F, Onderka J, Hos D, et al. Improved semiautomatic method for morphometry of angiogenesis and lymphangiogenesis in corneal flatmounts. Exp Eye Res 2008;87:462–70. 8. Bock F, Matthaei M, Reinhard T, et al. High-dose subconjunctival cyclosporine a implants do not affect corneal neovascularization after high-risk keratoplasty. Ophthalmology 2014;121:1677–82. 9. Cursiefen C, Viaud E, Bock F, et al. Aganirsen antisense oligonucleotide eye drops inhibit keratitis-induced corneal neovascularization and reduce need for transplantation: the ICAN study. Ophthalmology 2014;121:1683–92. 10. Cursiefen C, Bock F, Horn FK, et al. GS-101 antisense oligonucleotide eye drops inhibit corneal neovascularization: interim results of a randomized phase II trial. Ophthalmology 2009;116:1630–7. 11. Cursiefen C, Kuchle M, Naumann GO. Angiogenesis in corneal diseases: histopathologic evaluation of 254 human corneal buttons with neovascularization. Cornea 1998;17:611–3.

7

Ophthalmology Volume -, Number -, Month 2015 12. Lee P, Wang CC, Adamis AP. Ocular neovascularization: an epidemiologic review. Surv Ophthalmol 1998;43:245–69. 13. Resnikoff S, Pascolini D, Etya’ale D, et al. Global data on visual impairment in the year 2002. Bull World Health Organ 2004;82:844–51. 14. Menzel-Severing J. Emerging techniques to treat corneal neovascularisation. Eye (Lond) 2012;26:2–12. 15. Chang JH, Gabison EE, Kato T, Azar DT. Corneal neovascularization. Curr Opin Ophthalmol 2001;12:242–9. 16. Cursiefen C, Colin J, Dana R, et al. Consensus statement on indications for anti-angiogenic therapy in the management of corneal diseases associated with neovascularisation: outcome of an expert roundtable. Br J Ophthalmol 2012;96:3–9. 17. de Carlo TE, Bonini Filho MA, Chin AT, et al. Spectraldomain optical coherence tomography angiography of choroidal neovascularization. Ophthalmology 2015;122: 1228–38. 18. Sharma A, Bettis DI, Cowden JW, Mohan RR. Localization of angiotensin converting enzyme in rabbit cornea and its role in controlling corneal angiogenesis in vivo. Mol Vis 2010;16: 720–8. 19. Zheng Y, Kaye AE, Boker A, et al. Marginal corneal vascular arcades. Invest Ophthalmol Vis Sci 2013;54:7470–7. 20. Anijeet DR, Zheng Y, Tey A, et al. Imaging and evaluation of corneal vascularization using fluorescein and indocyanine green angiography. Invest Ophthalmol Vis Sci 2012;53:650–8. 21. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74. 22. Bachmann B, Taylor RS, Cursiefen C. Corneal neovascularization as a risk factor for graft failure and rejection

23. 24. 25.

26. 27.

28.

29.

30.

after keratoplasty: an evidence-based meta-analysis. Ophthalmology 2010;117:1300–1315e7. Talusan ED, Schwartz B. Fluorescein angiography. Demonstration of flow pattern of anterior ciliary arteries. Arch Ophthalmol 1981;99:1074–80. Easty DL, Bron AJ. Fluorescein angiography of the anterior segment. Its value in corneal disease. Br J Ophthalmol 1971;55:671–82. Nieuwenhuizen J, Watson PG, Emmanouilidis-van der Spek K, et al. The value of combining anterior segment fluorescein angiography with indocyanine green angiography in scleral inflammation. Ophthalmology 2003;110:1653–66. Watson PG. Anterior segment fluorescein angiography in the surgery of immunologically induced corneal and scleral destructive disorders. Ophthalmology 1987;94:1452–64. Kwiterovich KA, Maguire MG, Murphy RP, et al. Frequency of adverse systemic reactions after fluorescein angiography. Results of a prospective study. Ophthalmology 1991;98: 1139–42. Stanga PE, Lim JI, Hamilton P. Indocyanine green angiography in chorioretinal diseases: indications and interpretation: an evidence-based update. Ophthalmology 2003;110:15–21. quiz 2e3. Sng CC, Singh M, Chew PT, et al. Quantitative assessment of changes in trabeculectomy blebs after laser suture lysis using anterior segment coherence tomography. J Glaucoma 2012;21: 313–7. Liu W, Schultz KM, Zhang K, et al. Corneal neovascularization imaging by optical-resolution photoacoustic microscopy. Photoacoustics 2014;2:81–6.

Footnotes and Financial Disclosures Originally received: April 4, 2015. Final revision: May 9, 2015. Accepted: May 12, 2015. Available online: ---.

Author Contributions: Conception and design: Ang, Sim, Keane, Sng, Egan, Tufail, Wilkins Manuscript no. 2015-555.

1

Department of Cornea and External Eye Diseases, Singapore Eye Research Institute, Singapore National Eye Center, Singapore.

2

Department of External Eye Diseases, Moorfields Eye Hospital NHS Foundation Trust, London, United Kingdom.

3

Department of Medical Retina, Moorfields Eye Hospital NHS Foundation Trust, London, United Kingdom. 4

Institute of Ophthalmology, University College London, United Kingdom. 5 Department of Ophthalmology, National University Hospital, Singapore. Financial Disclosure(s): The author(s) have made the following disclosure(s): A.T.: Nonfinancial, technical support e Optovue, Inc., Fremont, CA.

8

Analysis and interpretation: Ang, Sim, Keane, Sng, Egan, Tufail, Wilkins Data collection: Ang, Sim, Keane, Tufail, Wilkins Obtained funding: Overall responsibility: Ang, Sim, Keane, Sng, Egan, Tufail, Wilkins Abbreviations and Acronyms: Df ¼ fractal dimension; FA ¼ fluorescein angiography; ICC ¼ intraclass coefficient; ICGA ¼ indocyanine green angiography; OCT ¼ optical coherence tomography; OCTA ¼ optical coherence tomography angiography; ROI ¼ region of interest. Correspondence: Marcus Ang, FAMS, FRCSEd, Singapore National Eye Centre, Singapore Eye Research Institute, 11 Third Hospital Avenue, Singapore 168751, Republic of Singapore. E-mail: [email protected].