Comparison of Two Spectral Domain Optical Coherence Tomography Devices for Angle-Closure Assessment

Abstract

Purpose.: To compare two spectral domain optical coherence tomography (SD-OCT) devices for the identification of angle structures and the presence of angle closure.

Methods.: This was a prospective comparative study. Consecutive patients underwent gonioscopy and anterior segment imaging using two SD-OCT devices (iVue and Cirrus). Images were evaluated for the ability to detect angle structures such as Schwalbe's line (SL), trabecular meshwork (TM), Schlemm's canal (SC), and scleral spur (SS), and the presence of angle closure. Angle closure was defined as iris contact with the angle wall anterior to the SS on SD-OCT, and nonvisibility of the posterior TM on gonioscopy. Angle closure in an eye was defined as ≥two quadrants of closed angles. AC1 statistic was used to assess the agreement between devices.

Results.: Of the 69 subjects studied (46.4% male, 84.1% Chinese, mean age 64.0 ± 10.5 years), 40 subjects (40 eyes, 58.0%) had angle closure on gonioscopy. The most identifiable structure on Cirrus SD-OCT was the SS (82.2%) and SL on iVue SD-OCT (74.5%). Angle closure was indeterminable in 14.5% and 50.7% of Cirrus and iVue scans (P < 0.001), respectively. Interdevice agreement for angle closure was moderately strong (AC1 = 0.67), but agreement with gonioscopy was only fair (AC1 = 0.35 and 0.50 for Cirrus and iVue, respectively).

Conclusions.: It was more difficult to determine angle closure status with iVue compared with Cirrus SD-OCT. There was fair agreement between both devices with gonioscopy for identifying angle closure.

Introduction

Gonioscopy is the current reference standard for assessing anterior chamber angle (ACA) structures and their configuration. The identification of regions of apposition of the iris to the trabecular meshwork is the hallmark of the diagnosis of angle closure. Unfortunately, gonioscopy is a highly subjective test, with only moderate agreement reported among observers. ^1–3 The varying annotation of angle findings with gonioscopic grading schemes, ^4,5 variable findings with different gonioscopic lenses, and the alteration of the angle configuration by light, placement of the lens, and/or mechanical compression of the eye often lead to significant variability in gonioscopic assessments. ^2,3,6–11  

Anterior segment optical coherence tomography (AS-OCT) is a noncontact imaging device that acquires high-resolution cross-sectional images of anterior segment structures. Fourier or spectral domain OCT (SD-OCT) differs from time domain OCT, used in AS-OCT, by utilizing a shorter wavelength (830 nm) and having a fixed reference mirror, which allows higher scanning speed and more images to be taken in a single pass. SD-OCT scans at a rate of 26,000 A-scans/s, producing detailed cross-sectional images of structures at an axial resolution of 5 μm and a transverse resolution of 15 μm. ¹²  

The major advantages of both AS-OCT and SD-OCT devices include ease of operation and rapidity of image acquisition. The noncontact method eliminates patient discomfort and inadvertent compression of the globe. The incorporation of automated analysis software allows for rapid estimation of various anterior segment parameters, including corneal thickness, anterior chamber depth, and ACA measurements. 

The Cirrus SD-OCT (Carl Zeiss Meditec, Dublin, CA) and iVue SD-OCT (Optovue Corporation, Fremont, CA) are commercially available SD-OCT devices used for imaging of the retina. In a recent study, Wong and colleagues ¹³ modified the Cirrus SD-OCT device with a 60 diopter lens (increasing its transverse scan range from 6 to 7.2 mm) to acquire high-resolution images of the ACA. In addition to the scleral spur, they were able to identify angle landmarks such as the Schwalbe's line in 93.3% and the trabecular meshwork in 62.2% of images. Although the device showed good correlation with gonioscopic findings, which was better than that obtained by AS-OCT, ¹⁴ Cirrus SD-OCT detected fewer closed angles compared with gonioscopy. The current Cirrus SD-OCT device has inbuilt anterior segment imaging options without the need for modifications. 

The aim of this study was to compare iVue and Cirrus SD-OCT devices to image angle structures and to identify angle closure, using gonioscopy as the reference standard. 

Materials and Methods

This was a prospective, comparative study. Consecutive patients attending the glaucoma clinics at the Singapore National Eye Centre were recruited. Informed consent was obtained from all participants, and the study had the approval of the institutional ethics review board and adhered to the tenets of the Declaration of Helsinki. Eyes with a previous history of previous intraocular surgery, penetrating trauma, or any cornea opacities or abnormalities that precluded gonioscopy were excluded. One eye from each subject was randomly selected. The luminance of the room was standardized at 10.76 lux, as measured by a commercial light meter (L-398A Studio Deluxe III; Sekonic, Tokyo, Japan). All tests, including gonioscopy, were done in the same room. 

Gonioscopy

Each subject underwent slit-lamp biomicroscopy and gonioscopy by an ophthalmologist (AKN) with previous glaucoma fellowship training. Static gonioscopy was performed in the dark using a Goldmann two-mirror lens (Haag-Streit, Koeniz, Switzerland), at a magnification of ×16, with the eye in primary position of gaze. A 1-mm light beam was reduced to a very narrow slit, and care was taken to avoid light falling on the pupil and accidental indentation during examination. The angle in each quadrant was graded using the modified Shaffer grading system, based on the predominant anatomic structures observed during gonioscopy (grade 0 = no angle structures visible, grade 1 = visible Schwalbe's line only, grade 2 = visible posterior trabecular meshwork, grade 3 = visible scleral spur, grade 4 = visible ciliary body). A quadrant was considered “closed” if the posterior trabecular meshwork could not be seen in the primary position without indentation (Shaffer grade 0 or 1). Angle closure in an eye was defined as the presence of two or more closed quadrants. 

SD-OCT Imaging

Anterior segment imaging was performed by a single trained examiner (TAT) masked to clinical findings. For angle imaging, the patient's fixation was directed to the side of the instrument, using an external fixation light so that the iridocorneal angle concerned was centered in the instrument's field of view. Four scans of the angle of each eye (at the 12-, 6-, 3-, and 9-o'clock positions) were obtained with each SD-OCT device in the dark. 

Image Analysis

All SD-OCT scans were analyzed separately by two examiners with glaucoma subspecialty training (AKN and DTQ) masked to other test results. The following angle structures were identified: 

1.  
   Schwalbe's line (SL) as the termination of the Descemet's membrane.
2.  
   The trabecular meshwork (TM) as a triangular structure with the base attached at the scleral spur.
3.  
   The scleral spur (SS) by its peaked outline and the slight projection into the anterior chamber, where the internal contours of the cornea and sclera meet.
4.  
   Schlemm's canal (SC) was seen as a curvilinear lucent area external to the trabecular meshwork. This lucent area extended from the scleral spur to the anterior tip of the trabecular meshwork located at the end of the Descemet's membrane. This lucent area corresponding to Schlemm's canal was distinct from the lucency of the corneal stroma ¹⁵ (Figs. 1, 2).

Figure 1. 

 

iVue SD-OCT image showing angle structures.

Figure 2. 

 

Cirrus SD-OCT image showing angle structures.

The angle was considered closed on SD-OCT scans if there was any contact between the iris and angle wall anterior to the scleral spur. Angle closure in an eye was defined as the presence of two or more closed quadrants. 

The intraobserver reproducibility for the assessment of angle structures and angle closure in SD-OCT images was assessed in a random subset of 20 images. Images of these quadrants were graded for visibility of angle structures and angle closure on two separate occasions separated by an interval of 1 week by a single examiner (AKN) masked to other findings. 

Statistical Analysis

Statistical analysis was performed using a commercial analytic software package (SPSS version 18; SPSS Inc., Chicago, IL). Parametric and nonparametric tests were used to compare continuous variables according to data distribution. The χ² or Fisher's exact test was used to compare categorical data. The McNemar test was used to compare differences in distribution of a categorical variable between two dependent samples. Statistical significance was set at P < 0.05. The kappa and AC1 statistic were used to assess the agreement between categorical variables. Strength of agreement was described to be very strong if kappa or AC1 values were >0.8, moderately strong if they were between 0.6 and 0.8, fair if they were between 0.3 and 0.5, and poor if they were <0.3. 

Results

A total of 69 eyes of 69 patients were included. The majority of subjects were female (53.6%) and Chinese (84.1%), with a mean (±SD) age of 64.0 ± 10.5 years. 

This study included 69 eyes. On gonioscopy, 29/69 (42.0%) had open angles: 27 were normal eyes and 2 had primary open angle glaucoma. Angle closure disease accounted for 58.0% (40/69) of eyes: 28 were primary angle closure suspects, 6 had primary angle closure, and 6 had primary angle closure glaucoma (Table 1). On gonioscopy, 107/276 (39%) quadrants had angle closure. The closed quadrants had the following distribution: 34/107 (32%), 19/107 (18%), 26/107 (24%), and 28/107 (26%) of superior, inferior, nasal, and temporal quadrants, respectively, were “closed.” 

Table 1. 

 

Gonioscopic Classification of All 69 Included Eyes

Angle Status	No. (%)
Open angle	29 (42)
Normal	27 (93.1)
Primary open angle glaucoma	2 (6.9)
Narrow angles	40 (58.0)
Primary angle closure suspect	28 (70)
Primary angle closure	6 (15)
Primary angle closure glaucoma	6 (15)

Identification of Angle Structures

Cirrus SD-OCT.

Using Cirrus SD-OCT, SL, TM, SC, and SS were identified in 213/276 (77.2%), 95/276 (34.4%), 28/276 (10.1%), and 227/276 (82.2%) of images, respectively. Among quadrants, in the superior and nasal quadrants, SL was the most frequently identified structure (49/69 [71%] and 67/69 [97.1%], respectively), whereas in the inferior and temporal quadrants, SS was the most frequently identified structure (59/69 [85.5%] and 64/69 [92.8%], respectively) (Table 2). 

Table 2. 

 

Summary of Structures Identified on Cirrus and iVue SD-OCT Scans

Cirrus				iVue
Number of Images in Which Angle Structures Were Identified
Structure		n = 276	%		n = 276	%
SL		213	77.2		204	73.9
TM		95	34.4		187	67.8
SC		28	10.1		31	11.2
SS		227	82.2		183	66.3
Location Where Structure Was Most Frequently Identified
Structure	Quadrant	n = 69	%	Quadrant	n = 69	%
SL	Nasal	67	97.1	Nasal	66	95.7
TM	Nasal	34	49.3	Nasal and Temporal	60	87.0
SC	Temporal	11	15.9	Temporal	18	26.1
SS	Temporal	64	92.8	Nasal and Temporal	62	89.9
Most Frequently Identified Structure in Each Quadrant
	Structure	n = 69	%	Structure	n = 69	%
Superior	SL	49	71.0	SL	43	62.3
Inferior	SS	59	85.5	SS	31	44.9
Nasal	SL	67	97.1	SL	66	95.7
Temporal	SS	64	92.8	SL	64	92.8

iVue SD-OCT.

Using iVue SD-OCT, SL, TM, SC and SS were identified in 204/276 (73.9%), 187/276 (67.8%), 37/276 (13.4%), and 183/276 (66.3%) of images, respectively. Among quadrants, in the superior, nasal, and temporal quadrants SL was the most frequently identified structure (43/69 [62.3%], 66/69 [95.7%], and 64/69 [92.8%], respectively), whereas in the inferior quadrant, SS was the most frequently identified structure (31/69 [44.9%]) (Table 2). 

Comparing Cirrus with iVue, the most frequently identified structures were SS followed by SL for the Cirrus, whereas those for iVue were SL and TM. Both devices identified SL and SC equally frequently, but TM was more frequently identified in iVue images compared with Cirrus images (67.8% vs. 34.4%, respectively; P < 0.001), whereas SS was more frequently identified in Cirrus images compared with iVue images (82.2% vs. 66.3%, respectively; P < 0.001) (Table 2). 

The AC1 statistic for intraobserver agreement on identification of angle structures was 0.6–1 for Cirrus and 0.88–0.94 for iVue. The AC1 statistics for interobserver agreement on identification of angle structures are summarized in Table 3. 

Table 3. 

 

Interobserver Agreement on Identification of Angle Structures

Angle Structure	Quadrant	Cirrus	iVue
		AC1	AC1
SL	Superior	0.718	0.606
	Inferior	0.518	0.478
	Nasal	0.955	0.969
	Temporal	0.818	0.922
TM	Superior	0.231	0.363
	Inferior	0.595	0.097
	Nasal	0.168	0.818
	Temporal	0.031	0.564
SS	Superior	0.044	0.441
	Inferior	0.01	0.195
	Nasal	0.46	0.823
	Temporal	0.71	0.881
SC	Superior	0.899	0.809
	Inferior	0.897	0.97
	Nasal	0.528	0.522
	Temporal	0.75	0.595

Identification of Angle Closure

Cirrus SD-OCT.

Angle status was not determined in 28/276 (10.1%) of quadrants and 9/69 (13.0%) of eyes on Cirrus SD-OCT. This was due to an inability to clearly identify the scleral spur, especially so in the superior quadrant (15/69 [21.7%]). In the remaining 248 images, 206 (83.1%) were graded to be “closed” on Cirrus SD-OCT compared with 150 (60.5%) on gonioscopy (AC1 = 0.39). The disparity of proportion of images graded “closed” between the Cirrus SD-OCT versus gonioscopy was especially significant in the nasal and temporal quadrants (87.7% vs. 61.5% and 95.4% vs. 58.5%, respectively; P < 0.001 for both). In the 60 eyes that could be assessed for angle status, 12 (20.0%) were graded to be closed on Cirrus SD-OCT, compared with 26 (43.4%) on gonioscopy (AC1 = 0.35) (Table 4). 

Table 4. 

 

Comparison between Cirrus OCT, iVue OCT and Gonioscopy on the Ability to Detect Angle Closure in Each Quadrant and Eye

	Angle Status of Each Quadrant					Angle Status of Each Eye
	Undetermined		Closed		AC1	Undetermined		Closed		AC1
Cirrus vs. Gonioscopy
Cirrus	28/276	10.1%	206/248	83.1%	0.39	9/69	13.0%	12/60	20.0%	0.35
Gonioscopy	NA	NA	150/248	60.5%		NA	NA	26/60	43.3%
iVue vs. Gonioscopy
iVue	46/276	16.7%	189/231	81.8%	0.44	17/69	24.6%	10/52	19.2%	0.50
Gonioscopy	NA	NA	144/231	62.3%		NA	NA	21/52	40.4%
Cirrus vs. iVue
Cirrus	28/276	10.1%	182/212	85.8%	0.72	9/69	13.0%	9/47	19.1%	0.67
iVue	46/276	16.7%	175/212	82.5%		17/69	24.6%	8/47	17.0%

 NA, not applicable.

iVue SD-OCT.

Angle status was not determined in 46/276 (16.7%) of images and 17/69 (24.6%) of eyes on iVue SD-OCT. This was due to an inability to clearly identify the scleral spur, especially so in the superior quadrant (24/69 [34.8%]). In the remaining 231 images, 189 (81.8%) were graded to be “closed” on iVue SD-OCT compared with 144 (62.3%) on gonioscopy (AC1 = 0.44). The disparity of proportion of images graded “closed” between iVue SD-OCT versus gonioscopy was especially significant in the nasal and temporal quadrants (87.0% vs. 62.3% and 92.8% vs. 59.4%, respectively; P < 0.001 for both). In the 52 eyes that could be assessed for angle status, 10 (19.2%) were graded to be closed on iVue SD-OCT, compared with 21 (40.4%) on gonioscopy (AC1 = 0.50) (Table 4). 

The AC1 statistic for agreement between Cirrus and iVue SD-OCT on detection of angle closure in each quadrant and eye was 0.72 and 0.67, respectively (Table 4). The agreements between gonioscopy, Cirrus, and iVue SD-OCT in detecting closed angles in each quadrant and angle closure status of an eye are shown in Figures 1 and 2, respectively. The kappa statistic for interobserver agreement on identification of angle closure ranged from 0.20 to 0.40 on the Cirrus and 0.35 to 0.47 on iVue. 

Discussion

Our study demonstrated that with the Cirrus SD-OCT, SL, SS, TM, and SC were identified in 77%, 82%, 34%, and 10% of quadrants, respectively, whereas the corresponding figures for iVue were 74%, 66%, 68%, and 13% of quadrants, respectively. Both devices identified SL equally well, and SC equally poorly, but SS was more frequently identified in Cirrus compared with iVue images, whereas the TM was more frequently identified in iVue images compared with Cirrus. The variability in these figures could be explained by the inherent differences in quality of image acquisition and processing between the different machines, and the subjectivity of image interpretation between observers. In addition, the magnifications of the images are different. The magnification of the Cirrus machine is 2-fold that of the iVue. This results in the SC in the iVue image appearing compressed and lower in position compared with that in the Cirrus image. 

Currently available software analysis programs require the manual localization of the scleral spur, which can at times be difficult, especially in closed angles or where there is a smooth transition from cornea to sclera. With AS-OCT, Sakata et al. ¹⁶ found that the sclera spur could not be detected in approximately 30% of angle quadrants, a problem that was worse in the superior and inferior quadrants. In our study, we defined angle closure as contact between the iris and angle wall anterior to the scleral spur. We had difficulty determining angle status due to an inability to clearly discern the SS in some images. Although higher-resolution OCT devices improved the resolution of angle structures, in our study, angle status was not determined in 10% of quadrants and 13% of eyes using Cirrus SD-OCT, and 17% of quadrants and 25% of eyes using iVue SD-OCT, mainly due to an inability to identify the SS in these images. Inherent limitations of SD-OCT imaging technology could be possible reasons. The iVue uses markers for standardizing the image acquisition point, resulting in less penetration depth, attenuation of details at the scleral spur area, and motion artifacts that preclude distinct identification of landmarks. Possible solutions to these limitations include averaging of images and increasing the width to include both angles, and postimage acquisition processing similar to those of high-resolution retinal images. 

More recently, Cheung et al. ¹⁷ proposed the SL as a new anatomical landmark, independent of the SS location, for assessing ACA width quantitatively with Cirrus SD-OCT. They developed a computer-aided program to define two new quantitative parameters for assessing ACA width: Schwalbe's line–angle opening distance (SL-AOD) measured at the SL, and Schwalbe's line–trabecular-iris space area (SL-TISA) measured 500 μm from the SL. In their study, 20% of images had to be excluded because of poor image quality. In the remaining images, SL and SS could be identified in 95% and 85% of quadrants, respectively. SL-AOD and SL-TISA were significantly correlated with SS parameters (all r > 0.85) and gonioscopic grading (all r > 0.69). In eyes with closed angles (n = 36), SL parameters showed strong correlations with gonioscopic grading (r ranged from 0.43 to 0.44). These findings suggest that novel angle parameters, based on SL as a landmark, may be useful to quantify ACA width and to assess for risk of angle closure. ¹⁷ Both the devices used in our study identified SL in about 75% of images. A further study on these SL-derived quantitative parameters derived from images from both the Cirrus and iVue devices, and their correlation with gonioscopy could be performed to assess suitability of these devices for quantification of angle closure. 

We found that the agreement of Cirrus and iVue SD-OCT with gonioscopy for identification of angle closure in eyes was fair (AC1 = 0.35 and 0.50, respectively), with both the Cirrus and iVue OCT detecting angle closure at a lower rate compared with gonioscopy. On SD-OCT, a radiolucent gap is sometimes present between the cornea and sclera such that the iris does not seem to be in apposition with the angle, rendering the angle “open” on SD-OCT but “closed” on gonioscopy (Figure 3). In addition, although gonioscopy allows concurrent visualization of the entire angle quadrant, AS-OCT images are meriodional, and relate only to the particular cross-section of the angle scanned. Thus, they may not be representative of the whole quadrant. 

Figure 3. 

 

iVue SD-OCT image showing radiolucent gap between the cornea and sclera, rendering the angle “open” on SD-OCT but “closed” on gonioscopy.

In addition, scan acquisition with the patient looking at the side of the instrument may potentially change the image profile of the angle. Unfortunately, imaging of the angle with the current SD-OCT is technically not possible with the eye in the primary position, and we attempted to standardize this by using specified targets for each patient to fixate on. This may be an additional reason for the poor agreement between gonioscopy and these devices. 

Although intraobserver agreement on identification of angle structures was good for both devices (AC1 = 0.6–1 for Cirrus and 0.88–0.94 for iVue), the interobserver agreement on identification of angle closure was poor to fair on the Cirrus (kappa = 0.20–0.40) and fair on iVue (kappa = 0.35–0.47). In this study, we found that angle structure identification was generally easier in horizontal scans compared with vertical scans. As detailed in Table 2, angle structures were most frequently identified in either nasal or temporal quadrants, in both machines. Localization of angle structures was difficult in the superior and inferior quadrants, a finding consistent with that previously reported by Sakata et al. ¹⁶ As a result, interobserver agreement was generally poorer for superior and inferior quadrants than nasal and temporal quadrants. Significant interobserver variability exists for both machines, and angle closure status between observers may not be consistently reproducible. 

In conclusion, it was more difficult to determine angle closure status with the iVue SD-OCT compared with Cirrus SD-OCT. Although these new SD-OCT devices provide us with higher resolution and detailed imaging of angle structures, they are unable to determine angle status in all cases, have significant interobserver variability, and have only fair correlation with gonioscopy. 

Footnotes

 Supported in part by grants from the Biomedical Research Council, Singapore and the National Medical Research Council, Singapore.

Footnotes

 Disclosure: D.T. Quek, None; A.K. Narayanaswamy, None; T.A. Tun, None; H.M. Htoon, None; M. Baskaran, None; S.A. Perera, Carl Zeiss Meditec (R); T. Aung, Carl Zeiss Meditec (F, R)