- Email: [email protected]
Retinal Nerve Fiber Layer Progression in Glaucoma
Retinal Nerve Fiber Layer Progression in Glaucoma A Comparison between Retinal Nerve Fiber Layer Thickness and Retardance Guihua Xu, BM,1 Robert N. Weinreb, MD,2 Christopher K.S. Leung, MD, MBChB1 Objective: To investigate the performance of spectral-domain optical coherence tomography (OCT) and scanning laser polarimetry to detect progressive retinal nerve fiber layer (RNFL) changes and to determine whether reduction of the RNFL retardance occurred before thinning of the RNFL in glaucoma. Design: Prospective, longitudinal study. Participants: One hundred eighty-four eyes of 116 glaucoma patients and 43 normal eyes of 23 healthy individuals. Methods: Patients were followed up every 4 months for at least 36 months with RNFL retardance (GDx Enhanced Corneal Compensation; Carl Zeiss Meditec) and RNFL thickness (Cirrus HD-OCT; Carl Zeiss Meditec) measured in the same visit. Progressive RNFL retardance and thickness changes were evaluated with eventbased analysis (Guided Progression Analysis; Carl Zeiss Meditec) with reference to the RNFL retardance change map and the RNFL thickness change map, respectively. The area and frequency distribution of RNFL changes were examined by overlaying the RNFL retardance change maps and the RNFL thickness change maps in the latest follow-up. The agreement of RNFL retardance and RNFL thickness progression was evaluated with k statistics. Main Outcome Measures: Number of eyes with progressive RNFL changes over time. Results: A total of 2472 OCT thickness maps and 2472 RNFL retardance maps were collected and reviewed with a mean follow-up of 55.1 months. Twenty-seven eyes (14.6%; 26 glaucoma patients) showed progressive RNFL thinning, whereas 8 eyes (4.3%; 8 glaucoma patients) showed progressive reduction of RNFL retardance. Seven eyes (3.8%; 7 glaucoma patients) had progression that was detected by both instruments, all with progressive RNFL thinning detected before progressive reduction of RNFL retardance became evident, and the mean lag time was 13.4 months (range, 4.0-37.6 months). The agreement between RNFL thickness and RNFL retardance progression was fair (k, 0.357). Progressive loss of RNFL thickness was observed most frequently at the inferotemporal 223 to 260 , whereas the inferotemporal 227 to 263 and superior 56 to 117 were observed most commonly for progressive loss of RNFL retardance. In the normal group, no eyes showed reduction in RNFL thickness or retardance. Conclusions: At a comparable level of specificity, progressive RNFL thinning was detected more often than progressive reduction of RNFL retardance. For eyes with progressive loss of RNFL thickness and RNFL retardance, the former preceded the latter. Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references. Ophthalmology 2013;-:1e8 ª 2013 by the American Academy of Ophthalmology.
Examination of retinal nerve fiber layer (RNFL) defects and monitoring of their progression are key components in glaucoma management. Although red-free RNFL photography remains a useful technique to visualize the RNFL, objective and quantitative assessment of the RNFL largely relies on digital imaging technologies, including scanning laser polarimetry (SLP) and optical coherence tomography (OCT). Scanning laser polarimetry measures the relative phase retardance of the RNFL as a consequence of its birefringent properties,1 whereas OCT measures the RNFL 2013 by the American Academy of Ophthalmology Published by Elsevier Inc.
thickness with reference to the tissue reflectivities of the retinal layers in cross-section.2 Both OCT and SLP underwent major modifications in the instrumentation and analysis algorithm over the past decade. The latest generations of OCT (spectral-domain OCT) and SLP (GDx Enhanced Corneal Compensation [ECC]; Carl Zeiss Meditec) have been shown to outperform previous instruments, time-domain OCT and GDx Variable Corneal Compensation, for detection of RNFL damage3e8 and progressive RNFL changes.9,10 ISSN 0161-6420/13/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2013.07.027
1
Ophthalmology Volume -, Number -, Month 2013 Although the testeretest variabilities of RNFL measurements obtained with the spectral-domain OCT and GDx ECC are low,7,11e14 their relative performance in monitoring progressive RNFL thinning in glaucoma patients is unclear. Specifically, it is unknown whether SLP can detect RNFL changes earlier than OCT in clinical practice. Scanning laser polarimetry has a theoretical advantage over OCT in the detection of change because reduction of RNFL retardance consequential to disruption of the microtubules could be evident before actual loss of the nerve fibers when the optic nerve is injured. In support of this, Fortune et al15 examined 33 rhesus macaques with experimental glaucoma induced by laser photocoagulation of the trabecular meshwork and showed that at the onset of optic nerve head change detected by a confocal scanning laser ophthalmoscope, RNFL retardance, but not RNFL thickness, decreased. A similar observation was noted in nonhuman primate eyes after optic nerve transection16 and intravitreal injection of colchicine,17 a microtubule polymerization inhibitor. These studies provide the foundation for this prospective, longitudinal study comparing OCT and SLP to detect progressive RNFL changes in glaucoma patients.
Methods Subjects A total of 116 glaucoma patients and 23 normal, healthy individuals were enrolled consecutively. They were followed up from June 2007 through December 2012 at the University Eye Center, the Chinese University of Hong Kong. All subjects underwent a full ophthalmic examination, including measurement of visual acuity, refraction, and intraocular pressure; gonioscopy; and fundus examination. Eyes were included if the visual acuity was at least 20/40. Eyes were excluded if there was evidence of macular disease, refractive or retinal surgery, or neurologic disease. Glaucoma patients were identified based on the presence of visual field defects (described below) with corresponding optic disc and RNFL changes in at least 1 eye independent of the level of intraocular pressure and the anterior chamber angle status. Ten eyes of 7 patients had primary angle-closure glaucoma. Normal individuals had no structural optic disc abnormalities, no history of intraocular pressure of more than 21 mmHg, no visual field defects, and no history of ocular disease, neurologic disease, or major systemic illness. Both eyes had undergone Cirrus HD-OCT (Carl Zeiss Meditec) and GDx ECC (Carl Zeiss Meditec) RNFL imaging and visual field testing (Humphrey Field Analyzer; Carl Zeiss Meditec) in the same visit every 4 months for at least 36 months. Eight eyes (4.3%) underwent trabeculectomy during the follow-up. The study was conducted in accordance with the ethical standards stated in the 1964 Declaration of Helsinki and was approved by local research ethics committee with informed consent obtained.
Visual Field Examination Visual fields were obtained using the Humphrey Field Analyzer II-i white-on-white Swedish interactive threshold algorithm standard 24-2 strategy (Carl Zeiss Meditec). All visual fields included in the study had fixation losses and false-positive and false-negative errors of less than 20%. Average visual field sensitivity was expressed in mean deviation (MD), as calculated by the perimetry software. A visual field defect was defined as having 3 or more significant (P < 0.05) nonedge contiguous points with at least 1 at
2
the P < 0.01 level on the same side of the horizontal meridian in the pattern deviation plot and confirmed with at least 2 consecutive examinations. Visual field progression was analyzed with event analysis using the Guided Progression Analysis (GPA; Carl Zeiss Meditec) according to the Early Manifest Glaucoma Trial criteria.18 Progression was defined when there were 3 or more points that showed significant change (more than the testeretest variabilities) compared with 2 baseline examinations (separated by 4 months in this study) for at least 2 consecutive tests (i.e., likely progression or possible progression were noted in the GPA printout in the latest follow-up visit). Twenty eyes from 18 patients were excluded because GPA was not performed in severely depressed fields.
Optical Coherence Tomography The details of the principles of spectral-domain OCT have been described elsewhere.19 The Cirrus HD-OCT was used to image the RNFL. The acquisition rate of the Cirrus HD-OCT was 27 000 Ascans per second, and the transverse and axial resolutions were 15 mm and 5 mm, respectively. An optic disc cube scan protocol was used to measure the RNFL thickness in a 66-mm2 region (200200 pixels) centered at the optic disc, and an RNFL thickness map was generated. Only images with signal strength of 7 or more were included in the analysis. Saccadic eye movement was detected with the line-scanning ophthalmoscope image overlaid with OCT en face image. Images with motion artifact, poor centration, poor focus, or missing data were detected by the operator at the time of imaging, with rescanning performed during the same visit. Serial RNFL thickness maps were analyzed for detection of change using the GPA (described below).
Scanning Laser Polarimetry Scanning laser polarimetry was performed with GDx ECC. The general principles of SLP and the algorithm of GDx ECC have been described elsewhere.20,21 The SLP imaged an area of 40 horizontally20 vertically (256128 pixels) in the fundus and generated an RNFL retardance map (20 20 ; 128128 pixels) centered at the optic disc for RNFL assessment. The GDx ECC improved the signal-to-noise ratio of the SLP images by introducing a known bias retarder, which then was removed mathematically to determine the actual RNFL retardance.21 GDx ECC measured the retardation in nanometers and a fixed conversion factor (0.67 nm/mm) was used to calculate the RNFL thickness in micrometers. To ensure that the image quality was adequate for reliable measurement of the RNFL retardance, the software provided an image quality score and a typical scan score for each SLP image. The image quality score (1e10) was based on the correct alignment, fixation, and refraction of the scan, whereas the typical scan score (1e100) was a result of the support vector machine analysis generated based on the slope, standard deviation, and average magnitude of RNFL thickness measurements from the edge of the optic disc extending outward to 20 . In this study, an image quality score of 8 or more, a typical scan score of 40 or more, and a residual value of 4 nm or less were set as the minimum criteria for a good-quality scan as recommended by the manufacturer.
Analysis of Progressive Retinal Nerve Fiber Layer Thinning A total of 3455 OCT images and 2914 SLP images (including glaucomatous and normal eyes) were collected from the same OCT and SLP instruments from June 1, 2007, through December
Xu et al
RNFL Progression in Glaucoma
Table 1. Number of images excluded because of inadequate image quality, motion artifact, and image registration or alignment failure.
Image Quality *Motion Artifact Registration/ Alignment Failure
Cirrus HD-OCT
GDx ECC
Signal strength <7
Quality score <8 TSS <40 Residual value >4
277 14 3
388 2 39 d 13
TSS ¼ Typical scan score *There were no motion artifact images for GDx because these images would have a quality score<8.
Table 2. Demographics, Optical Coherence Tomography Findings, Scanning Laser Polarimetry Results, and Visual Field Measurements
No. of subjects No. of eyes Age (yrs) Spherical error (diopters) Baseline MD (dB) Cirrus HD-OCT Signal strength (mean of all visits) Baseline RNFL thickness (mm) GDx ECC Quality score (mean of all visits) Baseline RNFL thickness (mm)
Glaucoma
Normal
116 184 52.7614.95 2.884.06 4.255.00
23 43 56.79.17 0.252.04 0.821.35
7.810.83 78.7314.03
8.340.86 99.79.49
8.470.60 48.078.02
8.530.55 55.455.32
ECC ¼ Enhanced Corneal Compensation; MD ¼ mean deviation; RNFL ¼ retinal nerve fiber layer.
31, 2012 (the number of SLP images was smaller because the GDx underwent a major maintenance and software upgrade in Dublin, California, in February and March 2010). After excluding images with inadequate image quality, motion artifact, and failure to register with the baseline images for GPA (Table 1), 3161 OCT and 2472 SLP images were qualified for inclusion. To ensure a valid comparison of GPA between the instruments, we included only eyes that had both OCT and SLP qualified images available in the same visit. In total, 2472 Cirrus HD-OCT images and 2472 GDx ECC images from 184 glaucomatous eyes and 43 normal eyes were included for progression analysis. The Cirrus HD-OCT and the GDx ECC used the same algorithm, GPA, for detection of progressive RNFL change. The software automatically aligned and registered 2 baseline and follow-up images so that the same pixel locations could be measured for detection of change. The difference in RNFL measurement of an individual pixel between the baseline and the follow-up RNFL thickness and RNFL retardance maps was compared with an estimate of testeretest variability of that particular pixel (in-house proprietary database from Carl Zeiss Meditec). Pixels with RNFL measurement difference exceeding the testeretest variability between a follow-up and the first and the second baseline images would be coded in yellow in the OCT RNFL thickness change map (5050 pixels) and the SLP RNFL retardance change map (128128 pixels). If the same changes were evident in an additional follow-up image, the pixels would be coded in red. In this study, the 2 baseline images were separated by approximately 4 months and progression was defined when an area was detected in red in the RNFL thickness and retardance change map at the latest follow-up visit. The same definition was used for both instruments. Although the GPA also provided event analysis on the circumpapillary RNFL measurement and trend analysis on the average RNFL thicknesses, only the RNFL thickness and retardance maps were evaluated (the map analysis included all the available RNFL data obtained with the instruments, and thus provided a more informative and complete evaluation of progressive RNFL thinning). Because the optic disc region captured by the Cirrus HD-OCT (approximately 6.06.0 mm2) was greater than that captured by the GDx ECC (approximately 4.24.2 mm2), we considered only the central approximately 4.24.2 mm2 overlaid region of the RNFL thickness change map and the RNFL retardance change map for the comparison of GPA. To compare the GPA in the same region and to investigate the area of change and the frequency distribution of progressive RNFL changes, the RNFL thickness and retardance change maps of patients showing
progression at the latest follow-up were exported and overlaid using a computer program written in Matlab software version R2010a (The Math Works, Inc, Natick, MA).
Statistical Analyses Statistical analyses were performed using R software version 2.13.0 (R Foundation, Vienna, Austria). The survival probabilities of progressing eyes detected by OCT and SLP were compared with the log-rank test. The areas of progressive RNFL change between OCT and SLP were compared with the Wilcoxon signed-rank test. The agreement between progressive RNFL thinning and progressive reduction of RNFL retardance and between progressive visual field and RNFL changes were calculated with k statistics. A value between 0.0 and 0.2 indicates slight agreement, a value between 0.21 and 0.40 indicates fair agreement, a value between 0.41 and 0.60 indicates moderate agreement, a value between 0.61 and 0.80 indicates substantial agreement, and a value between 0.81 and 1 indicates almost perfect agreement.22 A P value < 0.05 was considered statistically significant.
Results A total of 2472 Cirrus HD-OCT OCT images and 2472 GDx ECC images obtained from the same visits, including 184 eyes of 116 glaucoma patients and 43 eyes from 23 normal individuals prospectively followed up for at least 36 months (mean follow-up time, 55.1 months; median, 55.6 months; range, 36.1e69.9 months), were analyzed. The mean interval of the imaging examinations was 4.7 months. At the baseline examination, there were 76% mild (MD 6 dB), 15% moderate (6 dB>MD>12 dB), and 9% advanced (MD 12 dB) glaucomatous eyes. The intraocular pressures (Goldmann applanation tonometry) recorded at the baseline and the latest follow-up visits were 18.64.6 mmHg (range, 7e32.5 mmHg) and 17.35.0 mmHg (8e34 mmHg), respectively. The demographics, visual fields, and RNFL measurements of the normal and glaucoma groups are presented in Table 2.
Detection of Progressive Retinal Nerve Fiber Layer and Visual Field Changes A total of 27 eyes (14.6%; 26 patients) had progressive RNFL thinning and 8 eyes (4.3%; 8 patients) had progressive reduction of
3
Ophthalmology Volume -, Number -, Month 2013 RNFL retardance detected by GPA (Fig 1). The survival probability of eyes evaluated with OCT was significantly worse compared with that of eyes evaluated with SLP (P ¼ 0.001; Fig 2). Seven eyes (3.8%; 7 patients) had progressive RNFL change detected by both instruments at the latest follow-up, with all having progressive RNFL thinning before progressive reduction of RNFL retardance (Figs 3 and 4). The mean lag time between RNFL thinning and reduction in RNFL retardance was 13.4 months (range, 4.0-37.6 months). The agreement between RNFL thickness and RNFL retardance progression was fair, and the k value was 0.357 (95% confidence interval, 0.254e0.460). Twenty-six eyes of 25 patients demonstrated visual field progression (Fig 5). For the 5 eyes whose progression was detected by both OCT and visual field examination and the 2 eyes whose progression was detected by both SLP and visual field examination, all of them showed correspondence in structure and function in the area where change was detected. However, the agreement with visual field progression was equally poor for both OCT (k, 0.052; 95% confidence interval, 0.025 to 0.129) and SLP (k, 0.055; 95% confidence interval, 0.017 to 0.127).
Spatial and Frequency Distribution of Progressive Retinal Nerve Fiber Layer Changes The area of progressive RNFL change ranged between 0.34 and 9.40 mm2 (mean, 1.521.71 mm2) for the 27 progressing eyes detected by OCT, whereas it was between 0.47 and 3.85 mm2 (mean, 1.411.14 mm2) for the 8 progressing eyes detected by SLP. For the 7 eyes with progressive RNFL changes detected by both instruments, the area of progression measured from the RNFL thickness change map and the RNFL retardance change map ranged between 0.11 and 3.57 mm2 (mean, 0.911.20mm2) and between 0.49 and 2.52 mm2 (mean, 1.130.71 mm2), respectively, at the time of detection. At the latest follow-up visit, they were between 0.17 and 5.94 mm2 (mean, 1.861.91 mm2) and between 0.47 and 3.85 mm2 (mean, 1.351.22mm2), respectively. There was no significant difference in the area of progressive RNFL change between the instruments at the time of detection (P ¼ 0.398) or at the latest follow-up visit (P ¼ 0.128). The area of preexisting glaucomatous damage (i.e., area of RNFL measurements less than the lower 99% normal ranges in the baseline RNFL thickness and RNFL retardance deviation map) was greater for OCT than that for SLP (P ¼ 0.021, Wilcoxon signed-rank test). It was not correlated with the area of progressive RNFL thinning (P ¼ 0.431) or the area of progressive reduction of RNFL retardance (P ¼ 0.535, Spearman’s rank correlation). Taking all progressing eyes into consideration, the inferotemporal sector at 223 to 260 was the most frequent location where progressive RNFL thinning was evident (Fig 6A), whereas the inferotemporal sector at 227 to 263 as well as the superior sector at 56 to 117
Figure 1. Venn diagram showing the number of eyes (number of patients) with retinal nerve fiber layer (RNFL) thickness progression detected by optical coherence tomography and RNFL retardance progression detected by scanning laser polarimetry.
4
Figure 2. Graph showing survival probabilities of eyes detected with progression by optical coherence tomography and scanning laser polarimetry. Hash marks represent censored observation.
were the most common locations for detection of progressive reduction of RNFL retardance (Fig 6B).
Specificity of Guided Progression Analysis In the normal group (43 eyes of 23 subjects), no eyes showed progressive reduction of RNFL thickness or RNFL retardance during the follow-up period (mean, 49.19 months; range, 39.51e60.66 months), indicating that GPA of the RNFL thickness maps and RNFL retardance maps had a specificity of 100%. No eyes showed visual field progression at the latest follow-up visit.
Discussion To our knowledge, this is the first prospective, longitudinal study directly comparing progressive reduction of RNFL thickness and RNFL retardance in glaucoma patients. Analyzing progressive RNFL changes with the same eventbased algorithm, GPA, we found that more eyes (14.6%) showed progressive RNFL thinning than progressive reduction in RNFL retardance (4.3%) at a comparable level of specificity. Remarkably, for eyes with progression evident in both OCT and SLP, RNFL thinning was always evident before RNFL retardance became detectable, and the lag time could be up to 3 years (Fig 3). Monitoring RNFL thickness with OCT may be more effective in identifying progressive RNFL damage than monitoring RNFL retardance with SLP in glaucoma patients. Both RNFL retardance and RNFL thickness have been shown to be useful in monitoring glaucoma progression, although their relative abilities to detect change have not been compared.9,10,23e26 Previous studies were focused primarily on the average circumpapillary RNFL thickness, and only a few analyzed progression taking all data points in the RNFL thickness and retardance maps into consideration. With a median follow-up of 48 months, Alencar et al25 showed that the GDx Variable Corneal Compensation GPA of the RNFL retardance map had a specificity of
Xu et al
RNFL Progression in Glaucoma
Figure 3. A, Serial retinal nerve fiber layer (RNFL) retardance maps (upper panel) and RNFL retardance change maps (lower panel) of the same eye. B, Retinal nerve fiber layer retardance maps (upper panel) and RNFL retardance change maps (lower panel) of a glaucomatous eye. Progression was evident in the RNFL thickness change map (obtained on August 21, 2009) approximately 37 months before it became notable in the RNFL retardance change map (obtained on October 8, 2012).
100% in a group of 19 normal subjects and a sensitivity of 50% in 233 glaucoma suspects and glaucoma patients using visual field and optic disc photographs as the reference standard. We recently reported the use of the RNFL thickness map to visualize the patterns of progressive RNFL thinning.26 Following up 103 glaucoma patients over a median of 42 months, 24 patients
(23%) demonstrated RNFL thickness progression. Most demonstrated widening of RNFL defects, followed by development of new defects and deepening of pre-existing defects. It is conceivable that using all data points in a map would be more informative than just extracting the circumpapillary measurement for detection of change. We adopted the GPA for evaluation of glaucoma progression
Figure 4. A, Serial retinal nerve fiber layer (RNFL) retardance maps (upper panel) and RNFL retardance change maps (lower panel) of the same eye. B, Retinal nerve fiber layer retardance maps (upper panel) and RNFL retardance change maps (lower panel) of a glaucomatous eye. Progression was evident in the RNFL thickness change map (obtained on March 4, 2011) approximately 11 months before it became notable in the RNFL retardance change map (obtained on February 16, 2012).
5
Ophthalmology Volume -, Number -, Month 2013
Figure 5. Venn diagrams showing the number of eyes (number of patients) with progressive retinal nerve fiber layer (RNFL) damage detected by (A) optical coherence tomography, (B) scanning laser polarimetry, and perimetry.
because the same algorithm could be applied in both Cirrus HD-OCT and GDx ECC. We did not calculate the rates of change of circumpapillary RNFL thickness using trend analysis because the scales and the ranges of RNFL measurements between OCT and SLP are different, rendering the rates of change incomparable. The finding that RNFL thinning occurred before reduction in RNFL retardance is different from the previous experimental studies in nonhuman primates in which loss of RNFL thickness was evident only after reduction of RNFL retardance.15e17 In the experimental studies, RNFL changes were detected from a normal state to a condition in which the optic nerve was injured by nerve transection or high intraocular pressure. By contrast, the present study focused on patients with established glaucoma who were receiving treatment. The relationship between RNFL thinning and microtubule disruption thus could be different between the studies. It is worth noting that the RNFL thickness and retardance were measured with reference to a circular scan in the experimental studies and that there are fundamental
disparities between OCT and SLP for measurement of circumpapillary RNFL thickness and retardance. Although OCT measured circumpapillary RNFL thickness based on a ring with a diameter of approximately 3.46 mm, the SLP circumpapillary RNFL thickness was derived from an 8-pixel circular band with a width of approximately 4.84 corresponding to approximately 1.12 mm on the macaque retina. That is, the use of the circular band in SLP may provide additional RNFL data that may allow SLP to detect RNFL changes earlier than OCT. This is supported by the fact that glaucoma progression can be missed by a ring scan because some progression can be detected only beyond the ring.26 Although it is plausible to have disruption of microtubules resulting in reduction of RNFL retardance before actual loss of RNFL thickness, the greater measurement variability (it has been shown that the measurement noise of SLP RNFL retardance was approximately 50% larger than that for spectral-domain OCT RNFL thickness)15,27 and the narrower dynamic range of RNFL retardance (in this study, the range of OCT
Figure 6. Frequency distribution plots showing the spatial distribution of retinal nerve fiber layer (RNFL) progression constructed from overlying (A) the RNFL thickness change maps (n ¼ 27) and (B) the RNFL retardance change maps (n ¼ 8) obtained from the latest follow-up visit. The frequencies are color coded and presented in percentage (0%e100%). The dotted circles have a diameter of 3.46 mm centered at the optic disc.
6
Xu et al
RNFL Progression in Glaucoma
average RNFL thickness was 50e114 mm, whereas the range of SLP average RNFL thickness was 26.37e65.08 mm) may have rendered SLP less sensitive to detect change compared with OCT. It is notable that both SLP and OCT GPA had an estimated specificity of 100%. This is consistent with the observation by Alencar et al25 that no eyes of 19 normal individuals had progressive reduction of RNFL retardance. Collectively, OCT may detect more eyes with progressive RNFL damage than SLP at a comparable level of specificity. The inferotemporal sector followed by the superotemporal sector were the most frequent locations where progressive RNFL thinning was evident. By contrast, progressive loss of RNFL retardance was noted at the superior and supranasal sectors, in addition to the infratemporal sector. Although this finding concurs with that of a previous study reporting that the rate of change of RNFL retardance was most rapid at the inferotemporal and superonasal meridians,23 it largely is unclear why there was a mismatch between RNFL thickness and RNFL retardance progression at the superior optic disc region (Fig 6). If progression of RNFL retardance at the superonasal sector represents a common sign of RNFL damage, development of superonasal RNFL thinning and inferotemporal visual field defects would have been observed more frequently. However, red-free RNFL photography study indicates that progressive RNFL thinning develops most often at the vicinity of inferotemporal RNFL defects,28 which corresponds well to the finding that visual sensitivity deteriorates most rapidly at the superonasal field.29 More investigation is needed to determine the validity and significance of RNFL retardance progression at the superonasal sector. Notably, the area of progressive RNFL damage detected by OCT and SLP may not be comparable directly because the area of pre-existing RNFL damage may differ. However, for the 7 eyes with progressive RNFL changes evident in both OCT and SLP, the finding that OCT was able to detect a greater area of pre-existing RNFL damage aligns well with its earlier recognition of progressive RNFL damage compared with SLP. We standardized not only the method of detection of change (GPA), but also the number of images (2472 images for both OCT and SLP) and the area of the optic disc region (approximately 4.24.2 mm2) for the comparison of progressive RNFL damage between OCT and SLP. In addition, stringent inclusion criteria were applied to the OCT and SLP images (Table 1) to ensure that the detection of change was reliable (the results of the study remained the same when the inclusion criterion of the typical scan score cutoff increased to 80, although an additional 81 GDx ECC images were excluded). While such standardization and high-quality image requirement minimized potential biases in the comparison between the instruments, a significant proportion of images were thereby discarded. This investigation also was hampered by the lack of an appropriate reference standard, which has been the intrinsic limitation in studies comparing the sensitivity of detection of change among the digital imaging instruments. Red-free RNFL photography may not serve as a reference standard because SLP and OCT may outperform RNFL photography in detecting RNFL defects.30,31 Moreover,
RNFL abnormality often is not detectable in red-free photographs until at least 50% of the RNFL is lost.32 Likewise, RNFL change can precede detectable reduction in visual sensitivity and neuroretinal rim loss detected by optic disc photographs. It therefore is not surprising to observe poor agreement between visual field and progressive RNFL changes (Fig 5). We currently are following up the same cohort of patients for a longer term to compare the predictive power of RNFL thickness and RNFL retardance progression for subsequent visual field change. This investigation will provide better insights into the clinical significance of the RNFL thickness and retardance progression. To summarize, analyzing serial RNFL thickness maps and RNFL retardance maps offers an efficient approach to detect and visualize progressive RNFL changes. For patients with progression detected by both OCT and SLP, progressive RNFL thinning was evident before progressive reduction of RNFL retardance. Having a higher detection rate and being able to detect RNFL damage earlier than SLP at a comparable specificity, spectral-domain OCT is a preferable imaging method for monitoring RNFL changes in glaucoma patients.
References 1. Weinreb RN, Dreher AW, Coleman A, et al. Histopathologic validation of Fourier-ellipsometry measurements of retinal nerve fiber layer thickness. Arch Ophthalmol 1990;108: 557–60. 2. Huang D, Swanson EA, Lin CP, et al. Optical coherence tomography. Science 1991;254:1178–81. 3. Mai TA, Reus NJ, Lemij HG. Diagnostic accuracy of scanning laser polarimetry with enhanced versus variable corneal compensation. Ophthalmology 2007;114:1988–93. 4. Bowd C, Tavares IM, Medeiros FA, et al. Retinal nerve fiber layer thickness and visual sensitivity using scanning laser polarimetry with variable and enhanced corneal compensation. Ophthalmology 2007;114:1259–65. 5. Sehi M, Guaqueta DC, Feuer WJ, Greenfield DS, Advanced Imaging in Glaucoma Study Group. Scanning laser polarimetry with variable and enhanced corneal compensation in normal and glaucomatous eyes. Am J Ophthalmol 2007;143: 272–9. 6. Sung KR, Kim DY, Park SB, Kook MS. Comparison of retinal nerve fiber layer thickness measured by Cirrus HD and Stratus optical coherence tomography. Ophthalmology 2009;116: 1264–70. 7. Leung CK, Cheung CY, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: a variability and diagnostic performance study. Ophthalmology 2009;116:1257–63. 8. Leung CK, Lam S, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: analysis of the retinal nerve fiber layer map for glaucoma detection. Ophthalmology 2010;117:1684–91. 9. Medeiros FA, Zangwill LM, Alencar LM, et al. Rates of progressive retinal nerve fiber layer loss in glaucoma measured by scanning laser polarimetry. Am J Ophthalmol 2010;149: 908–15. 10. Leung CK, Chiu V, Weinreb RN, et al. Evaluation of retinal nerve fiber layer progression in glaucoma: a comparison
7
Ophthalmology Volume -, Number -, Month 2013
11.
12.
13.
14.
15.
16.
17.
18. 19. 20.
between spectral-domain and time-domain optical coherence tomography. Ophthalmology 2011;118:1558–62. Mai TA, Reus NJ, Lemij HG. Retinal nerve fiber layer measurement repeatability in scanning laser polarimetry with enhanced corneal compensation. J Glaucoma 2008;17: 269–74. Mwanza JC, Gendy MG, Feuer WJ, et al. Effects of changing operators and instruments on time-domain and spectraldomain OCT measurements of retinal nerve fiber layer thickness. Ophthalmic Surg Lasers Imaging 2011;42:328–37. Grewal DS, Sehi M, Cook RJ, Greenfield DS, Advanced Imaging in Glaucoma Study Group. The impact of retardance pattern variability on nerve fiber layer measurements over time using GDx with variable and enhanced corneal compensation. Invest Ophthalmol Vis Sci 2011;52: 4516–24. Kim JS, Ishikawa H, Sung KR, et al. Retinal nerve fibre layer thickness measurement reproducibility improved with spectral domain optical coherence tomography. Br J Ophthalmol 2009;93:1057–63. Fortune B, Burgoyne CF, Cull GA, et al. Structural and functional abnormalities of retinal ganglion cells measured in vivo at the onset of optic nerve head surface change in experimental glaucoma. Invest Ophthalmol Vis Sci 2012;53: 3939–50. Fortune B, Cull GA, Burgoyne CF. Relative course of retinal nerve fiber layer birefringence and thickness and retinal function changes after optic nerve transection. Invest Ophthalmol Vis Sci 2008;49:4444–52. Fortune B, Wang L, Cull G, Cioffi GA. Intravitreal colchicine causes decreased RNFL birefringence without altering RNFL thickness. Invest Ophthalmol Vis Sci 2008;49: 255–61. Heijl A, Leske MC, Bengtsson B, et al; EMGT Group. Measuring visual field progression in the Early Manifest Glaucoma Trial. Acta Ophthalmol Scand 2003;81:286–93. Nassif N, Cense B, Park BH, et al. In vivo human retinal imaging by ultrahigh-speed spectral domain optical coherence tomography. Opt Lett 2004;29:480–2. Zhou Q, Weinreb RN. Individualized compensation of anterior segment birefringence during scanning laser polarimetry. Invest Ophthalmol Vis Sci 2002;43:2221–8.
21. Reus NJ, Zhou Q, Lemij HG. Enhanced imaging algorithm for scanning laser polarimetry with variable corneal compensation. Invest Ophthalmol Vis Sci 2006;47:3870–7. 22. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74. 23. Medeiros FA, Alencar LM, Zangwill LM, et al. Detection of progressive retinal nerve fiber layer loss in glaucoma using scanning laser polarimetry with variable corneal compensation. Invest Ophthalmol Vis Sci 2009;50:1675–81. 24. Grewal DS, Sehi M, Greenfield DS. Detecting glaucomatous progression using GDx with variable and enhanced corneal compensation using Guided Progression Analysis. Br J Ophthalmol 2011;95:502–8. 25. Alencar LM, Zangwill LM, Weinreb RN, et al. Agreement for detecting glaucoma progression with the GDx guided progression analysis, automated perimetry, and optic disc photography. Ophthalmology 2010;117:462–70. 26. Leung CK, Yu M, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: patterns of retinal nerve fiber layer progression. Ophthalmology 2012;119:1858–66. 27. Naghizadeh F, Garas A, Vargha P, Holló G. Comparison of long-term variability of retinal nerve fiber layer measurements made with the RTVue OCT and scanning laser polarimetry. Eur J Ophthalmol 2013;23:86–93. 28. Suh MH, Kim DM, Kim YK, et al. Patterns of progression of localized retinal nerve fibre layer defect on red-free fundus photographs in normal-tension glaucoma. Eye (Lond) 2010;24:857–63. 29. Pereira ML, Kim CS, Zimmerman MB, et al. Rate and pattern of visual field decline in primary open-angle glaucoma. Ophthalmology 2002;109:2232–40. 30. Medeiros FA, Zangwill LM, Bowd C, et al. Comparison of scanning laser polarimetry using variable corneal compensation and retinal nerve fiber layer photography for detection of glaucoma. Arch Ophthalmol 2004;122:698–704. 31. Ye C, To E, Weinreb RN, et al. Comparison of retinal nerve fiber layer imaging by spectral domain optical coherence tomography and scanning laser ophthalmoscopy. Ophthalmology 2011;118:2196–202. 32. Quigley HA, Addicks EM. Quantitative studies of retinal nerve fiber layer defects. Arch Ophthalmol 1982;100:807–14.
Footnotes and Financial Disclosures Originally received: February 11, 2013. Final revision: June 6, 2013. Accepted: July 18, 2013. Available online: ---.
Financial Disclosure(s): The author(s) have made the following disclosure(s): Manuscript no. 2013-235.
1
Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, China.
2
Hamilton Glaucoma Center and the Department of Ophthalmology, University of California, San Diego, La Jolla, California. Presented as a poster at: Association for Research in Vision and Ophthalmology Annual Meeting, May 2013, Seattle, Washington.
8
Christopher K. S. Leung: LecturerdCarl Zeiss Meditec, Heidelberg Engineering; Financial supportdCarl Zeiss Meditec, Optovue. Robert N. Weinreb: ConsultantdCarl Zeiss Meditec, Topcon; Financial supportdCarl Zeiss Meditec, Heidelberg Engineering, Optovue, Topcon, Nidek. Correspondence: Christopher K.S. Leung, MD, MBChB, Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, China. E-mail: [email protected].
Examination of retinal nerve fiber layer (RNFL) defects and monitoring of their progression are key components in glaucoma management. Although red-free RNFL photography remains a useful technique to visualize the RNFL, objective and quantitative assessment of the RNFL largely relies on digital imaging technologies, including scanning laser polarimetry (SLP) and optical coherence tomography (OCT). Scanning laser polarimetry measures the relative phase retardance of the RNFL as a consequence of its birefringent properties,1 whereas OCT measures the RNFL 2013 by the American Academy of Ophthalmology Published by Elsevier Inc.
thickness with reference to the tissue reflectivities of the retinal layers in cross-section.2 Both OCT and SLP underwent major modifications in the instrumentation and analysis algorithm over the past decade. The latest generations of OCT (spectral-domain OCT) and SLP (GDx Enhanced Corneal Compensation [ECC]; Carl Zeiss Meditec) have been shown to outperform previous instruments, time-domain OCT and GDx Variable Corneal Compensation, for detection of RNFL damage3e8 and progressive RNFL changes.9,10 ISSN 0161-6420/13/$ - see front matter http://dx.doi.org/10.1016/j.ophtha.2013.07.027
1
Ophthalmology Volume -, Number -, Month 2013 Although the testeretest variabilities of RNFL measurements obtained with the spectral-domain OCT and GDx ECC are low,7,11e14 their relative performance in monitoring progressive RNFL thinning in glaucoma patients is unclear. Specifically, it is unknown whether SLP can detect RNFL changes earlier than OCT in clinical practice. Scanning laser polarimetry has a theoretical advantage over OCT in the detection of change because reduction of RNFL retardance consequential to disruption of the microtubules could be evident before actual loss of the nerve fibers when the optic nerve is injured. In support of this, Fortune et al15 examined 33 rhesus macaques with experimental glaucoma induced by laser photocoagulation of the trabecular meshwork and showed that at the onset of optic nerve head change detected by a confocal scanning laser ophthalmoscope, RNFL retardance, but not RNFL thickness, decreased. A similar observation was noted in nonhuman primate eyes after optic nerve transection16 and intravitreal injection of colchicine,17 a microtubule polymerization inhibitor. These studies provide the foundation for this prospective, longitudinal study comparing OCT and SLP to detect progressive RNFL changes in glaucoma patients.
Methods Subjects A total of 116 glaucoma patients and 23 normal, healthy individuals were enrolled consecutively. They were followed up from June 2007 through December 2012 at the University Eye Center, the Chinese University of Hong Kong. All subjects underwent a full ophthalmic examination, including measurement of visual acuity, refraction, and intraocular pressure; gonioscopy; and fundus examination. Eyes were included if the visual acuity was at least 20/40. Eyes were excluded if there was evidence of macular disease, refractive or retinal surgery, or neurologic disease. Glaucoma patients were identified based on the presence of visual field defects (described below) with corresponding optic disc and RNFL changes in at least 1 eye independent of the level of intraocular pressure and the anterior chamber angle status. Ten eyes of 7 patients had primary angle-closure glaucoma. Normal individuals had no structural optic disc abnormalities, no history of intraocular pressure of more than 21 mmHg, no visual field defects, and no history of ocular disease, neurologic disease, or major systemic illness. Both eyes had undergone Cirrus HD-OCT (Carl Zeiss Meditec) and GDx ECC (Carl Zeiss Meditec) RNFL imaging and visual field testing (Humphrey Field Analyzer; Carl Zeiss Meditec) in the same visit every 4 months for at least 36 months. Eight eyes (4.3%) underwent trabeculectomy during the follow-up. The study was conducted in accordance with the ethical standards stated in the 1964 Declaration of Helsinki and was approved by local research ethics committee with informed consent obtained.
Visual Field Examination Visual fields were obtained using the Humphrey Field Analyzer II-i white-on-white Swedish interactive threshold algorithm standard 24-2 strategy (Carl Zeiss Meditec). All visual fields included in the study had fixation losses and false-positive and false-negative errors of less than 20%. Average visual field sensitivity was expressed in mean deviation (MD), as calculated by the perimetry software. A visual field defect was defined as having 3 or more significant (P < 0.05) nonedge contiguous points with at least 1 at
2
the P < 0.01 level on the same side of the horizontal meridian in the pattern deviation plot and confirmed with at least 2 consecutive examinations. Visual field progression was analyzed with event analysis using the Guided Progression Analysis (GPA; Carl Zeiss Meditec) according to the Early Manifest Glaucoma Trial criteria.18 Progression was defined when there were 3 or more points that showed significant change (more than the testeretest variabilities) compared with 2 baseline examinations (separated by 4 months in this study) for at least 2 consecutive tests (i.e., likely progression or possible progression were noted in the GPA printout in the latest follow-up visit). Twenty eyes from 18 patients were excluded because GPA was not performed in severely depressed fields.
Optical Coherence Tomography The details of the principles of spectral-domain OCT have been described elsewhere.19 The Cirrus HD-OCT was used to image the RNFL. The acquisition rate of the Cirrus HD-OCT was 27 000 Ascans per second, and the transverse and axial resolutions were 15 mm and 5 mm, respectively. An optic disc cube scan protocol was used to measure the RNFL thickness in a 66-mm2 region (200200 pixels) centered at the optic disc, and an RNFL thickness map was generated. Only images with signal strength of 7 or more were included in the analysis. Saccadic eye movement was detected with the line-scanning ophthalmoscope image overlaid with OCT en face image. Images with motion artifact, poor centration, poor focus, or missing data were detected by the operator at the time of imaging, with rescanning performed during the same visit. Serial RNFL thickness maps were analyzed for detection of change using the GPA (described below).
Scanning Laser Polarimetry Scanning laser polarimetry was performed with GDx ECC. The general principles of SLP and the algorithm of GDx ECC have been described elsewhere.20,21 The SLP imaged an area of 40 horizontally20 vertically (256128 pixels) in the fundus and generated an RNFL retardance map (20 20 ; 128128 pixels) centered at the optic disc for RNFL assessment. The GDx ECC improved the signal-to-noise ratio of the SLP images by introducing a known bias retarder, which then was removed mathematically to determine the actual RNFL retardance.21 GDx ECC measured the retardation in nanometers and a fixed conversion factor (0.67 nm/mm) was used to calculate the RNFL thickness in micrometers. To ensure that the image quality was adequate for reliable measurement of the RNFL retardance, the software provided an image quality score and a typical scan score for each SLP image. The image quality score (1e10) was based on the correct alignment, fixation, and refraction of the scan, whereas the typical scan score (1e100) was a result of the support vector machine analysis generated based on the slope, standard deviation, and average magnitude of RNFL thickness measurements from the edge of the optic disc extending outward to 20 . In this study, an image quality score of 8 or more, a typical scan score of 40 or more, and a residual value of 4 nm or less were set as the minimum criteria for a good-quality scan as recommended by the manufacturer.
Analysis of Progressive Retinal Nerve Fiber Layer Thinning A total of 3455 OCT images and 2914 SLP images (including glaucomatous and normal eyes) were collected from the same OCT and SLP instruments from June 1, 2007, through December
Xu et al
RNFL Progression in Glaucoma
Table 1. Number of images excluded because of inadequate image quality, motion artifact, and image registration or alignment failure.
Image Quality *Motion Artifact Registration/ Alignment Failure
Cirrus HD-OCT
GDx ECC
Signal strength <7
Quality score <8 TSS <40 Residual value >4
277 14 3
388 2 39 d 13
TSS ¼ Typical scan score *There were no motion artifact images for GDx because these images would have a quality score<8.
Table 2. Demographics, Optical Coherence Tomography Findings, Scanning Laser Polarimetry Results, and Visual Field Measurements
No. of subjects No. of eyes Age (yrs) Spherical error (diopters) Baseline MD (dB) Cirrus HD-OCT Signal strength (mean of all visits) Baseline RNFL thickness (mm) GDx ECC Quality score (mean of all visits) Baseline RNFL thickness (mm)
Glaucoma
Normal
116 184 52.7614.95 2.884.06 4.255.00
23 43 56.79.17 0.252.04 0.821.35
7.810.83 78.7314.03
8.340.86 99.79.49
8.470.60 48.078.02
8.530.55 55.455.32
ECC ¼ Enhanced Corneal Compensation; MD ¼ mean deviation; RNFL ¼ retinal nerve fiber layer.
31, 2012 (the number of SLP images was smaller because the GDx underwent a major maintenance and software upgrade in Dublin, California, in February and March 2010). After excluding images with inadequate image quality, motion artifact, and failure to register with the baseline images for GPA (Table 1), 3161 OCT and 2472 SLP images were qualified for inclusion. To ensure a valid comparison of GPA between the instruments, we included only eyes that had both OCT and SLP qualified images available in the same visit. In total, 2472 Cirrus HD-OCT images and 2472 GDx ECC images from 184 glaucomatous eyes and 43 normal eyes were included for progression analysis. The Cirrus HD-OCT and the GDx ECC used the same algorithm, GPA, for detection of progressive RNFL change. The software automatically aligned and registered 2 baseline and follow-up images so that the same pixel locations could be measured for detection of change. The difference in RNFL measurement of an individual pixel between the baseline and the follow-up RNFL thickness and RNFL retardance maps was compared with an estimate of testeretest variability of that particular pixel (in-house proprietary database from Carl Zeiss Meditec). Pixels with RNFL measurement difference exceeding the testeretest variability between a follow-up and the first and the second baseline images would be coded in yellow in the OCT RNFL thickness change map (5050 pixels) and the SLP RNFL retardance change map (128128 pixels). If the same changes were evident in an additional follow-up image, the pixels would be coded in red. In this study, the 2 baseline images were separated by approximately 4 months and progression was defined when an area was detected in red in the RNFL thickness and retardance change map at the latest follow-up visit. The same definition was used for both instruments. Although the GPA also provided event analysis on the circumpapillary RNFL measurement and trend analysis on the average RNFL thicknesses, only the RNFL thickness and retardance maps were evaluated (the map analysis included all the available RNFL data obtained with the instruments, and thus provided a more informative and complete evaluation of progressive RNFL thinning). Because the optic disc region captured by the Cirrus HD-OCT (approximately 6.06.0 mm2) was greater than that captured by the GDx ECC (approximately 4.24.2 mm2), we considered only the central approximately 4.24.2 mm2 overlaid region of the RNFL thickness change map and the RNFL retardance change map for the comparison of GPA. To compare the GPA in the same region and to investigate the area of change and the frequency distribution of progressive RNFL changes, the RNFL thickness and retardance change maps of patients showing
progression at the latest follow-up were exported and overlaid using a computer program written in Matlab software version R2010a (The Math Works, Inc, Natick, MA).
Statistical Analyses Statistical analyses were performed using R software version 2.13.0 (R Foundation, Vienna, Austria). The survival probabilities of progressing eyes detected by OCT and SLP were compared with the log-rank test. The areas of progressive RNFL change between OCT and SLP were compared with the Wilcoxon signed-rank test. The agreement between progressive RNFL thinning and progressive reduction of RNFL retardance and between progressive visual field and RNFL changes were calculated with k statistics. A value between 0.0 and 0.2 indicates slight agreement, a value between 0.21 and 0.40 indicates fair agreement, a value between 0.41 and 0.60 indicates moderate agreement, a value between 0.61 and 0.80 indicates substantial agreement, and a value between 0.81 and 1 indicates almost perfect agreement.22 A P value < 0.05 was considered statistically significant.
Results A total of 2472 Cirrus HD-OCT OCT images and 2472 GDx ECC images obtained from the same visits, including 184 eyes of 116 glaucoma patients and 43 eyes from 23 normal individuals prospectively followed up for at least 36 months (mean follow-up time, 55.1 months; median, 55.6 months; range, 36.1e69.9 months), were analyzed. The mean interval of the imaging examinations was 4.7 months. At the baseline examination, there were 76% mild (MD 6 dB), 15% moderate (6 dB>MD>12 dB), and 9% advanced (MD 12 dB) glaucomatous eyes. The intraocular pressures (Goldmann applanation tonometry) recorded at the baseline and the latest follow-up visits were 18.64.6 mmHg (range, 7e32.5 mmHg) and 17.35.0 mmHg (8e34 mmHg), respectively. The demographics, visual fields, and RNFL measurements of the normal and glaucoma groups are presented in Table 2.
Detection of Progressive Retinal Nerve Fiber Layer and Visual Field Changes A total of 27 eyes (14.6%; 26 patients) had progressive RNFL thinning and 8 eyes (4.3%; 8 patients) had progressive reduction of
3
Ophthalmology Volume -, Number -, Month 2013 RNFL retardance detected by GPA (Fig 1). The survival probability of eyes evaluated with OCT was significantly worse compared with that of eyes evaluated with SLP (P ¼ 0.001; Fig 2). Seven eyes (3.8%; 7 patients) had progressive RNFL change detected by both instruments at the latest follow-up, with all having progressive RNFL thinning before progressive reduction of RNFL retardance (Figs 3 and 4). The mean lag time between RNFL thinning and reduction in RNFL retardance was 13.4 months (range, 4.0-37.6 months). The agreement between RNFL thickness and RNFL retardance progression was fair, and the k value was 0.357 (95% confidence interval, 0.254e0.460). Twenty-six eyes of 25 patients demonstrated visual field progression (Fig 5). For the 5 eyes whose progression was detected by both OCT and visual field examination and the 2 eyes whose progression was detected by both SLP and visual field examination, all of them showed correspondence in structure and function in the area where change was detected. However, the agreement with visual field progression was equally poor for both OCT (k, 0.052; 95% confidence interval, 0.025 to 0.129) and SLP (k, 0.055; 95% confidence interval, 0.017 to 0.127).
Spatial and Frequency Distribution of Progressive Retinal Nerve Fiber Layer Changes The area of progressive RNFL change ranged between 0.34 and 9.40 mm2 (mean, 1.521.71 mm2) for the 27 progressing eyes detected by OCT, whereas it was between 0.47 and 3.85 mm2 (mean, 1.411.14 mm2) for the 8 progressing eyes detected by SLP. For the 7 eyes with progressive RNFL changes detected by both instruments, the area of progression measured from the RNFL thickness change map and the RNFL retardance change map ranged between 0.11 and 3.57 mm2 (mean, 0.911.20mm2) and between 0.49 and 2.52 mm2 (mean, 1.130.71 mm2), respectively, at the time of detection. At the latest follow-up visit, they were between 0.17 and 5.94 mm2 (mean, 1.861.91 mm2) and between 0.47 and 3.85 mm2 (mean, 1.351.22mm2), respectively. There was no significant difference in the area of progressive RNFL change between the instruments at the time of detection (P ¼ 0.398) or at the latest follow-up visit (P ¼ 0.128). The area of preexisting glaucomatous damage (i.e., area of RNFL measurements less than the lower 99% normal ranges in the baseline RNFL thickness and RNFL retardance deviation map) was greater for OCT than that for SLP (P ¼ 0.021, Wilcoxon signed-rank test). It was not correlated with the area of progressive RNFL thinning (P ¼ 0.431) or the area of progressive reduction of RNFL retardance (P ¼ 0.535, Spearman’s rank correlation). Taking all progressing eyes into consideration, the inferotemporal sector at 223 to 260 was the most frequent location where progressive RNFL thinning was evident (Fig 6A), whereas the inferotemporal sector at 227 to 263 as well as the superior sector at 56 to 117
Figure 1. Venn diagram showing the number of eyes (number of patients) with retinal nerve fiber layer (RNFL) thickness progression detected by optical coherence tomography and RNFL retardance progression detected by scanning laser polarimetry.
4
Figure 2. Graph showing survival probabilities of eyes detected with progression by optical coherence tomography and scanning laser polarimetry. Hash marks represent censored observation.
were the most common locations for detection of progressive reduction of RNFL retardance (Fig 6B).
Specificity of Guided Progression Analysis In the normal group (43 eyes of 23 subjects), no eyes showed progressive reduction of RNFL thickness or RNFL retardance during the follow-up period (mean, 49.19 months; range, 39.51e60.66 months), indicating that GPA of the RNFL thickness maps and RNFL retardance maps had a specificity of 100%. No eyes showed visual field progression at the latest follow-up visit.
Discussion To our knowledge, this is the first prospective, longitudinal study directly comparing progressive reduction of RNFL thickness and RNFL retardance in glaucoma patients. Analyzing progressive RNFL changes with the same eventbased algorithm, GPA, we found that more eyes (14.6%) showed progressive RNFL thinning than progressive reduction in RNFL retardance (4.3%) at a comparable level of specificity. Remarkably, for eyes with progression evident in both OCT and SLP, RNFL thinning was always evident before RNFL retardance became detectable, and the lag time could be up to 3 years (Fig 3). Monitoring RNFL thickness with OCT may be more effective in identifying progressive RNFL damage than monitoring RNFL retardance with SLP in glaucoma patients. Both RNFL retardance and RNFL thickness have been shown to be useful in monitoring glaucoma progression, although their relative abilities to detect change have not been compared.9,10,23e26 Previous studies were focused primarily on the average circumpapillary RNFL thickness, and only a few analyzed progression taking all data points in the RNFL thickness and retardance maps into consideration. With a median follow-up of 48 months, Alencar et al25 showed that the GDx Variable Corneal Compensation GPA of the RNFL retardance map had a specificity of
Xu et al
RNFL Progression in Glaucoma
Figure 3. A, Serial retinal nerve fiber layer (RNFL) retardance maps (upper panel) and RNFL retardance change maps (lower panel) of the same eye. B, Retinal nerve fiber layer retardance maps (upper panel) and RNFL retardance change maps (lower panel) of a glaucomatous eye. Progression was evident in the RNFL thickness change map (obtained on August 21, 2009) approximately 37 months before it became notable in the RNFL retardance change map (obtained on October 8, 2012).
100% in a group of 19 normal subjects and a sensitivity of 50% in 233 glaucoma suspects and glaucoma patients using visual field and optic disc photographs as the reference standard. We recently reported the use of the RNFL thickness map to visualize the patterns of progressive RNFL thinning.26 Following up 103 glaucoma patients over a median of 42 months, 24 patients
(23%) demonstrated RNFL thickness progression. Most demonstrated widening of RNFL defects, followed by development of new defects and deepening of pre-existing defects. It is conceivable that using all data points in a map would be more informative than just extracting the circumpapillary measurement for detection of change. We adopted the GPA for evaluation of glaucoma progression
Figure 4. A, Serial retinal nerve fiber layer (RNFL) retardance maps (upper panel) and RNFL retardance change maps (lower panel) of the same eye. B, Retinal nerve fiber layer retardance maps (upper panel) and RNFL retardance change maps (lower panel) of a glaucomatous eye. Progression was evident in the RNFL thickness change map (obtained on March 4, 2011) approximately 11 months before it became notable in the RNFL retardance change map (obtained on February 16, 2012).
5
Ophthalmology Volume -, Number -, Month 2013
Figure 5. Venn diagrams showing the number of eyes (number of patients) with progressive retinal nerve fiber layer (RNFL) damage detected by (A) optical coherence tomography, (B) scanning laser polarimetry, and perimetry.
because the same algorithm could be applied in both Cirrus HD-OCT and GDx ECC. We did not calculate the rates of change of circumpapillary RNFL thickness using trend analysis because the scales and the ranges of RNFL measurements between OCT and SLP are different, rendering the rates of change incomparable. The finding that RNFL thinning occurred before reduction in RNFL retardance is different from the previous experimental studies in nonhuman primates in which loss of RNFL thickness was evident only after reduction of RNFL retardance.15e17 In the experimental studies, RNFL changes were detected from a normal state to a condition in which the optic nerve was injured by nerve transection or high intraocular pressure. By contrast, the present study focused on patients with established glaucoma who were receiving treatment. The relationship between RNFL thinning and microtubule disruption thus could be different between the studies. It is worth noting that the RNFL thickness and retardance were measured with reference to a circular scan in the experimental studies and that there are fundamental
disparities between OCT and SLP for measurement of circumpapillary RNFL thickness and retardance. Although OCT measured circumpapillary RNFL thickness based on a ring with a diameter of approximately 3.46 mm, the SLP circumpapillary RNFL thickness was derived from an 8-pixel circular band with a width of approximately 4.84 corresponding to approximately 1.12 mm on the macaque retina. That is, the use of the circular band in SLP may provide additional RNFL data that may allow SLP to detect RNFL changes earlier than OCT. This is supported by the fact that glaucoma progression can be missed by a ring scan because some progression can be detected only beyond the ring.26 Although it is plausible to have disruption of microtubules resulting in reduction of RNFL retardance before actual loss of RNFL thickness, the greater measurement variability (it has been shown that the measurement noise of SLP RNFL retardance was approximately 50% larger than that for spectral-domain OCT RNFL thickness)15,27 and the narrower dynamic range of RNFL retardance (in this study, the range of OCT
Figure 6. Frequency distribution plots showing the spatial distribution of retinal nerve fiber layer (RNFL) progression constructed from overlying (A) the RNFL thickness change maps (n ¼ 27) and (B) the RNFL retardance change maps (n ¼ 8) obtained from the latest follow-up visit. The frequencies are color coded and presented in percentage (0%e100%). The dotted circles have a diameter of 3.46 mm centered at the optic disc.
6
Xu et al
RNFL Progression in Glaucoma
average RNFL thickness was 50e114 mm, whereas the range of SLP average RNFL thickness was 26.37e65.08 mm) may have rendered SLP less sensitive to detect change compared with OCT. It is notable that both SLP and OCT GPA had an estimated specificity of 100%. This is consistent with the observation by Alencar et al25 that no eyes of 19 normal individuals had progressive reduction of RNFL retardance. Collectively, OCT may detect more eyes with progressive RNFL damage than SLP at a comparable level of specificity. The inferotemporal sector followed by the superotemporal sector were the most frequent locations where progressive RNFL thinning was evident. By contrast, progressive loss of RNFL retardance was noted at the superior and supranasal sectors, in addition to the infratemporal sector. Although this finding concurs with that of a previous study reporting that the rate of change of RNFL retardance was most rapid at the inferotemporal and superonasal meridians,23 it largely is unclear why there was a mismatch between RNFL thickness and RNFL retardance progression at the superior optic disc region (Fig 6). If progression of RNFL retardance at the superonasal sector represents a common sign of RNFL damage, development of superonasal RNFL thinning and inferotemporal visual field defects would have been observed more frequently. However, red-free RNFL photography study indicates that progressive RNFL thinning develops most often at the vicinity of inferotemporal RNFL defects,28 which corresponds well to the finding that visual sensitivity deteriorates most rapidly at the superonasal field.29 More investigation is needed to determine the validity and significance of RNFL retardance progression at the superonasal sector. Notably, the area of progressive RNFL damage detected by OCT and SLP may not be comparable directly because the area of pre-existing RNFL damage may differ. However, for the 7 eyes with progressive RNFL changes evident in both OCT and SLP, the finding that OCT was able to detect a greater area of pre-existing RNFL damage aligns well with its earlier recognition of progressive RNFL damage compared with SLP. We standardized not only the method of detection of change (GPA), but also the number of images (2472 images for both OCT and SLP) and the area of the optic disc region (approximately 4.24.2 mm2) for the comparison of progressive RNFL damage between OCT and SLP. In addition, stringent inclusion criteria were applied to the OCT and SLP images (Table 1) to ensure that the detection of change was reliable (the results of the study remained the same when the inclusion criterion of the typical scan score cutoff increased to 80, although an additional 81 GDx ECC images were excluded). While such standardization and high-quality image requirement minimized potential biases in the comparison between the instruments, a significant proportion of images were thereby discarded. This investigation also was hampered by the lack of an appropriate reference standard, which has been the intrinsic limitation in studies comparing the sensitivity of detection of change among the digital imaging instruments. Red-free RNFL photography may not serve as a reference standard because SLP and OCT may outperform RNFL photography in detecting RNFL defects.30,31 Moreover,
RNFL abnormality often is not detectable in red-free photographs until at least 50% of the RNFL is lost.32 Likewise, RNFL change can precede detectable reduction in visual sensitivity and neuroretinal rim loss detected by optic disc photographs. It therefore is not surprising to observe poor agreement between visual field and progressive RNFL changes (Fig 5). We currently are following up the same cohort of patients for a longer term to compare the predictive power of RNFL thickness and RNFL retardance progression for subsequent visual field change. This investigation will provide better insights into the clinical significance of the RNFL thickness and retardance progression. To summarize, analyzing serial RNFL thickness maps and RNFL retardance maps offers an efficient approach to detect and visualize progressive RNFL changes. For patients with progression detected by both OCT and SLP, progressive RNFL thinning was evident before progressive reduction of RNFL retardance. Having a higher detection rate and being able to detect RNFL damage earlier than SLP at a comparable specificity, spectral-domain OCT is a preferable imaging method for monitoring RNFL changes in glaucoma patients.
References 1. Weinreb RN, Dreher AW, Coleman A, et al. Histopathologic validation of Fourier-ellipsometry measurements of retinal nerve fiber layer thickness. Arch Ophthalmol 1990;108: 557–60. 2. Huang D, Swanson EA, Lin CP, et al. Optical coherence tomography. Science 1991;254:1178–81. 3. Mai TA, Reus NJ, Lemij HG. Diagnostic accuracy of scanning laser polarimetry with enhanced versus variable corneal compensation. Ophthalmology 2007;114:1988–93. 4. Bowd C, Tavares IM, Medeiros FA, et al. Retinal nerve fiber layer thickness and visual sensitivity using scanning laser polarimetry with variable and enhanced corneal compensation. Ophthalmology 2007;114:1259–65. 5. Sehi M, Guaqueta DC, Feuer WJ, Greenfield DS, Advanced Imaging in Glaucoma Study Group. Scanning laser polarimetry with variable and enhanced corneal compensation in normal and glaucomatous eyes. Am J Ophthalmol 2007;143: 272–9. 6. Sung KR, Kim DY, Park SB, Kook MS. Comparison of retinal nerve fiber layer thickness measured by Cirrus HD and Stratus optical coherence tomography. Ophthalmology 2009;116: 1264–70. 7. Leung CK, Cheung CY, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: a variability and diagnostic performance study. Ophthalmology 2009;116:1257–63. 8. Leung CK, Lam S, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: analysis of the retinal nerve fiber layer map for glaucoma detection. Ophthalmology 2010;117:1684–91. 9. Medeiros FA, Zangwill LM, Alencar LM, et al. Rates of progressive retinal nerve fiber layer loss in glaucoma measured by scanning laser polarimetry. Am J Ophthalmol 2010;149: 908–15. 10. Leung CK, Chiu V, Weinreb RN, et al. Evaluation of retinal nerve fiber layer progression in glaucoma: a comparison
7
Ophthalmology Volume -, Number -, Month 2013
11.
12.
13.
14.
15.
16.
17.
18. 19. 20.
between spectral-domain and time-domain optical coherence tomography. Ophthalmology 2011;118:1558–62. Mai TA, Reus NJ, Lemij HG. Retinal nerve fiber layer measurement repeatability in scanning laser polarimetry with enhanced corneal compensation. J Glaucoma 2008;17: 269–74. Mwanza JC, Gendy MG, Feuer WJ, et al. Effects of changing operators and instruments on time-domain and spectraldomain OCT measurements of retinal nerve fiber layer thickness. Ophthalmic Surg Lasers Imaging 2011;42:328–37. Grewal DS, Sehi M, Cook RJ, Greenfield DS, Advanced Imaging in Glaucoma Study Group. The impact of retardance pattern variability on nerve fiber layer measurements over time using GDx with variable and enhanced corneal compensation. Invest Ophthalmol Vis Sci 2011;52: 4516–24. Kim JS, Ishikawa H, Sung KR, et al. Retinal nerve fibre layer thickness measurement reproducibility improved with spectral domain optical coherence tomography. Br J Ophthalmol 2009;93:1057–63. Fortune B, Burgoyne CF, Cull GA, et al. Structural and functional abnormalities of retinal ganglion cells measured in vivo at the onset of optic nerve head surface change in experimental glaucoma. Invest Ophthalmol Vis Sci 2012;53: 3939–50. Fortune B, Cull GA, Burgoyne CF. Relative course of retinal nerve fiber layer birefringence and thickness and retinal function changes after optic nerve transection. Invest Ophthalmol Vis Sci 2008;49:4444–52. Fortune B, Wang L, Cull G, Cioffi GA. Intravitreal colchicine causes decreased RNFL birefringence without altering RNFL thickness. Invest Ophthalmol Vis Sci 2008;49: 255–61. Heijl A, Leske MC, Bengtsson B, et al; EMGT Group. Measuring visual field progression in the Early Manifest Glaucoma Trial. Acta Ophthalmol Scand 2003;81:286–93. Nassif N, Cense B, Park BH, et al. In vivo human retinal imaging by ultrahigh-speed spectral domain optical coherence tomography. Opt Lett 2004;29:480–2. Zhou Q, Weinreb RN. Individualized compensation of anterior segment birefringence during scanning laser polarimetry. Invest Ophthalmol Vis Sci 2002;43:2221–8.
21. Reus NJ, Zhou Q, Lemij HG. Enhanced imaging algorithm for scanning laser polarimetry with variable corneal compensation. Invest Ophthalmol Vis Sci 2006;47:3870–7. 22. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159–74. 23. Medeiros FA, Alencar LM, Zangwill LM, et al. Detection of progressive retinal nerve fiber layer loss in glaucoma using scanning laser polarimetry with variable corneal compensation. Invest Ophthalmol Vis Sci 2009;50:1675–81. 24. Grewal DS, Sehi M, Greenfield DS. Detecting glaucomatous progression using GDx with variable and enhanced corneal compensation using Guided Progression Analysis. Br J Ophthalmol 2011;95:502–8. 25. Alencar LM, Zangwill LM, Weinreb RN, et al. Agreement for detecting glaucoma progression with the GDx guided progression analysis, automated perimetry, and optic disc photography. Ophthalmology 2010;117:462–70. 26. Leung CK, Yu M, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain optical coherence tomography: patterns of retinal nerve fiber layer progression. Ophthalmology 2012;119:1858–66. 27. Naghizadeh F, Garas A, Vargha P, Holló G. Comparison of long-term variability of retinal nerve fiber layer measurements made with the RTVue OCT and scanning laser polarimetry. Eur J Ophthalmol 2013;23:86–93. 28. Suh MH, Kim DM, Kim YK, et al. Patterns of progression of localized retinal nerve fibre layer defect on red-free fundus photographs in normal-tension glaucoma. Eye (Lond) 2010;24:857–63. 29. Pereira ML, Kim CS, Zimmerman MB, et al. Rate and pattern of visual field decline in primary open-angle glaucoma. Ophthalmology 2002;109:2232–40. 30. Medeiros FA, Zangwill LM, Bowd C, et al. Comparison of scanning laser polarimetry using variable corneal compensation and retinal nerve fiber layer photography for detection of glaucoma. Arch Ophthalmol 2004;122:698–704. 31. Ye C, To E, Weinreb RN, et al. Comparison of retinal nerve fiber layer imaging by spectral domain optical coherence tomography and scanning laser ophthalmoscopy. Ophthalmology 2011;118:2196–202. 32. Quigley HA, Addicks EM. Quantitative studies of retinal nerve fiber layer defects. Arch Ophthalmol 1982;100:807–14.
Footnotes and Financial Disclosures Originally received: February 11, 2013. Final revision: June 6, 2013. Accepted: July 18, 2013. Available online: ---.
Financial Disclosure(s): The author(s) have made the following disclosure(s): Manuscript no. 2013-235.
1
Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, China.
2
Hamilton Glaucoma Center and the Department of Ophthalmology, University of California, San Diego, La Jolla, California. Presented as a poster at: Association for Research in Vision and Ophthalmology Annual Meeting, May 2013, Seattle, Washington.
8
Christopher K. S. Leung: LecturerdCarl Zeiss Meditec, Heidelberg Engineering; Financial supportdCarl Zeiss Meditec, Optovue. Robert N. Weinreb: ConsultantdCarl Zeiss Meditec, Topcon; Financial supportdCarl Zeiss Meditec, Heidelberg Engineering, Optovue, Topcon, Nidek. Correspondence: Christopher K.S. Leung, MD, MBChB, Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, China. E-mail: [email protected].