To investigate whether progressive macular ganglion cell-inner plexiform layer (GCIPL) and peripapillary retinal nerve fiber layer (RNFL) thinning predict development of visual field (VF) defects in glaucoma suspects.
Retrospective cohort study.
This study included 541 eyes of 357 glaucoma suspects with a mean follow-up of 5.7 years. Progressive GCIPL and RNFL thinning were determined using Guided Progression Analysis (GPA) in optical coherence tomography (OCT). The development of VF defect was defined as the presence of three consecutive abnormal VFs. The risk of developing VF defect was evaluated using Cox proportional hazard models.
A total of 74 eyes (13.7%) and 87 eyes (16.1%) showed progressive GCIPL and RNFL thinning by OCT GPA, respectively, and 40 eyes (7.4%) developed VF defects. Eyes with progressive GCIPL (hazard ratio [HR], 7.130; 95% confidence interval [CI], 3.137-16.205) and RNFL (HR, 7.525; 95% CI, 3.272-17.311) thinning showed a significantly higher risk of developing VF defects. The rate of change in the average GCIPL and RNFL thickness was significantly higher in the eyes that developed VF defects (−0.71 and −1.13 μm/y, respectively) than the eyes that did not (−0.19 and −0.27 μm/y, respectively; all P < 0.05). Progressive GCIPL (43.1 vs. 63.1 months, respectively; P < 0.001) and RNFL (50.9 vs. 66.7 months, respectively; P < 0.001) thinning were detected significantly earlier than the development of VF defects.
Monitoring progressive change in GCIPL, as well as RNFL, effectively predicts the development of VF defects in glaucoma suspects.
Glaucoma is often a slowly progressive disease characterized by structural changes in the optic nerve head and retinal ganglion cell, as well as their axonal loss and accompanying visual field (VF) defects. Because the damage from glaucoma is irreversible, it is important to detect the disease at an early stage before significant VF loss has developed. Glaucoma suspects form the key group requiring early detection of progressive change. Progressive loss of the peripapillary retinal nerve fiber layer (RNFL) detected using optical coherence tomography (OCT) in glaucoma suspects is considered a risk factor for the development of VF defects. , Furthermore, previous studies have shown that macular damage is prevalent among patients with early glaucoma. In early glaucoma, progressive macular ganglion cell-inner plexiform layer (GCIPL) thinning is frequently detected before corresponding RNFL thinning. Therefore, macular imaging, in addition to peripapillary imaging, may be useful to monitor the progressive change in glaucoma suspects.
Guided Progression Analysis (GPA; Carl Zeiss Meditec, Dublin, California) was developed to facilitate the tracking of structural changes in OCT. It offers an effective approach to detect progressive GCIPL and RNFL thinning in patients with established glaucoma. In addition, integrating GCIPL and RNFL GPA in glaucoma monitoring could facilitate the early detection of progressive changes. However, little is known about longitudinal changes in GCIPL and RNFL thicknesses using OCT GPA in glaucoma suspects. Although VF examination has commonly been used to confirm the diagnosis of glaucoma, structural damage may occur in many patients before VF defects are detectable on standard automated perimetry (SAP). , Therefore, the purpose of this study was to investigate whether progressive GCIPL and RNFL thinning are predictive of the development of VF defects in glaucoma suspects.
Subjects and Methods
This study recruited subjects from the Asan Glaucoma Progression Study, an ongoing, longitudinal, retrospective cohort study, conducted at the Asan Medical Center (Seoul, Korea). The data were collected by reviewing medical records from April 2009 to February 2019. The Institutional Review Board of Asan Medical Center approved the present study, and all procedures were carried out in accordance with the principles of the Declaration of Helsinki. The requirement for informed consent was waived owing to the retrospective nature of the study.
At the baseline visit, all of the subjects underwent complete ophthalmologic examinations, including the measurement of best corrected visual acuity, intraocular pressure (IOP) by using Goldmann applanation tonometry, refractive error by using an autorefractor (KR-890; Topcon Corp, Tokyo, Japan), axial length (IOLMaster; Carl Zeiss Meditec), central corneal thickness (DGH-550; DGH Technology, Exton, Pennsylvania), slit-lamp biomicroscopy, and gonioscopy. At every 9-month follow-up visit (±3 months), the subjects underwent stereoscopic optic disc and red-free RNFL photography (AFC-210; Nidek, Aichi, Japan), RNFL and GCIPL imaging (Cirrus HD-OCT; Carl Zeiss Meditec), and VF testing (Humphrey field analyzer and the Swedish interactive threshold algorithm 24-2; Carl Zeiss Meditec).
To be included, all participating glaucoma suspects were required to meet the following criteria: best corrected visual acuity ≥20/30; a spherical equivalent between −8.0 and +3.0 diopters (D), and a cylinder correction within +3D; and a normal anterior chamber and open angle on slit-lamp and gonioscopic examinations. A glaucoma suspect was defined as an individual with ocular hypertension (IOP >21 mm Hg) or an optic disc appearance suspicious of glaucoma (e.g., vertical cup-to-disc ratio of ≥0.7, focal neural rim thinning, or disc hemorrhage) as determined by 2 experienced graders (J.W.S. and K.R.S.), without evidence of repeatable glaucomatous VF damage. , Eyes with disagreement between graders were excluded. Subjects with secondary causes of elevated IOP, or any ophthalmic or neurological disease known to affect the optic nerve head, macular structure, or VF were excluded. All included subjects did not use IOP-lowering medications at baseline.
Optical Coherence Tomography Imaging
The GCIPL and RNFL images were respectively obtained by using macular and optic disc cube scans with the Cirrus HD-OCT system. The macular cube scan generated a GCIPL thickness map that covered a 6 × 6 mm 2 (512 × 128 pixels) area centered at the fovea. The average macular GCIPL thicknesses were measured within an annulus with inner vertical and horizontal diameters of 1 and 1.2 mm, respectively, and outer vertical and horizontal diameters of 4 and 4.8 mm, respectively. The optic disc cube scan generated the RNFL thickness map that covered a 6 × 6 mm 2 (200 × 200 pixels) area centered at the optic disc. The average peripapillary RNFL thicknesses were measured in a circle 3.46 mm in diameter. Only images with a signal strength ≥6 in both macular and optic disc cube scans were included. Images with motion artifacts, poor centering, or segmentation errors were checked and discarded by the operator, and re-scanning was performed during the same visit. Eyes were excluded if they had less than 6 pairs of GCIPL and RNFL images (20 eyes were excluded) or less than 3 years of follow-up duration (14 eyes were excluded). Forty-one OCT images from 23 eyes were excluded due to unmet signal strength, and 25 OCT images from 10 eyes were excluded due to uncorrectable segmentation errors. The average number of OCT examinations per eye was 8.5 (range, 6-17). In total, 4,532 OCT images were included in the final analysis.
Progressive Thinning in Guided Progression Analysis of the Ganglion Cell–Inner Plexiform Layer and Retinal Nerve Fiber Layer
The HD-OCT GPA (software version 10.0 Carl Zeiss Meditec) provides color-coded classification in order to facilitate the detection of progressive glaucomatous change in clinical practice. The built-in software automatically aligns, registers, and compares baseline and follow-up OCT images. Abnormal GCIPL or RNFL changes which exceed the range of the test-retest variability are presented using yellow and red codes in a 6 × 6 mm 2 (50 × 50 superpixels) map. The “possible loss,” with a yellow code, indicates the first detection of an abnormal GCIPL or RNFL change in the GCIPL or RNFL thickness change map. The “likely loss,” with a red code, indicates that an abnormal GCIPL or RNFL change is confirmed in a subsequent follow-up examination. In the current study, progressive thinning was defined as when at least 20 contiguous pixels, coded with red, in the GCIPL or RNFL thickness change map were detected during follow-up, and the same changes were observed in the latest follow-up visit. Trend analysis provides the rate of change in the GCIPL and RNFL thicknesses over time by using linear regression. The GPA requires 2 baseline tests and at least 4 tests to determine the “likely loss” and the rate of change in thickness.
Development of Visual Field Defect
All VF tests were performed using the Humphrey field analyzer with the 24-2 Swedish interactive threshold algorithm standard strategy. Only reliable tests with false positive or false negative errors less than 15% and fixation losses less than 20% were used in the study. The first VF test was excluded from the analysis in order to reduce the learning effect. The minimal abnormality for a glaucomatous VF defect included a cluster of three or more non–edge-contiguous points on a pattern deviation plot in a single hemifield (superior or inferior) with a P value of less than 5% (with at least 1 having a P value of less than 1%); a pattern standard deviation with a P value of less than 5%; or a glaucoma hemifield test result outside normal limits. The development of VF defect was defined when three consecutive VF tests met the same criterion for abnormality. The reference standard for glaucoma conversion was determined by the development of VF defect.
When the VF defect was confirmed in 3 consecutive VF tests, IOP-lowering agents were prescribed to the subjects. When preperimetic glaucomatous change was predicted, such as progressive GCIPL or RNFL thinning, the subjects were carefully followed without medication until they were confirmed to be perimetric glaucoma.
R version 3.5.1 software (R Foundation, Vienna, Austria) was used for statistical analyses. A linear mixed effect model was used to adjust for correlation between two eyes of the same patient. The demographics and clinical characteristics were compared between eyes with and without the development of VF defects. Univariate and multivariate Cox proportional hazard models were used to evaluate the risk of developing VF defects. A shared frailty model was employed to adjust for correlation between two eyes in the time-to-event data. Kaplan-Meier survival analysis and log-rank test were used to compare the cumulative probability of developing VF defect in eyes with and without progressive GCIPL or RNFL thinning.
A total of 541 eyes of 357 glaucoma suspects (208 men and 149 women) were included in this study. The mean follow-up period was 5.7 ± 1.4 years (range, 3.1-9.4 years). At the baseline visit, the mean age, refractive error, axial length, central corneal thickness, IOP, average GCIPL and RNFL thicknesses, VF mean deviation (MD), and the proportion of ocular hypertension were 58.9 ± 13.6 years, −1.29 ± 3.16D, 24.38 ± 1.65 mm, 537.4 ± 41.7 μm, 16.2 ± 3.9 mm Hg, 78.6 ± 7.0 μm, 86.9 ± 10.5 μm, −0.79 ± 1.34 dB, and 15.3%, respectively.
Sensitivity and Specificity of Guided Progression Analysis for Detecting Development of Visual Field Defect in Glaucoma Suspects
The development of VF defect was detected in 40 eyes (7.4%) during the follow-up period. The clinical characteristics of eyes with and without the development of VF defect are summarized in Table 1 . Eyes that demonstrated the development of VF defects had a longer follow-up duration and thinner average GCIPL and RNFL thicknesses at baseline examination compared to eyes without. The GPA detected progressive GCIPL and RNFL thinning in 74 eyes (13.7%) and 87 eyes (16.1%), respectively ( Figure 1 ). Fifty-one eyes (9.4%) showed both progressive GCIPL and RNFL thinning. The sensitivity and specificity of progressive GCIPL thinning for detecting the development of VF defects were 72.5% (29 of 40 eyes) and 91.0% (456 of 501 eyes), respectively, and those of progressive RNFL thinning were 77.5% (31 of 40 eyes) and 88.8% (445 of 501 eyes), respectively. The rate of change in the average GCIPL and RNFL thickness was significantly higher in eyes that developed VF defects (−0.71 ± 0.57 μm/y and −1.13 ± 0.85 μm/y, respectively) than eyes that did not (−0.19 ± 0.32 μm/y and −0.27 ± 0.64 μm/y, respectively; all P < 0.05).
|All (N = 541)||Eyes with VFD (n = 40)||Eyes without VFD (n = 501)||P|
|Age (y)||58.9 ± 13.6||60.4 ± 13.4||58.8 ± 13.7||0.537|
|Refractive error (D)||−1.29 ± 3.16||−1.02 ± 3.14||−1.31 ± 3.16||0.601|
|Axial length (mm)||24.38 ± 1.65||24.30 ± 1.77||24.39 ± 1.65||0.941|
|Central corneal thickness (μm)||537.4 ± 41.7||530.4 ± 41.5||537.9 ± 41.7||0.425|
|Follow-up duration (y)||5.7 ± 1.4||7.7 ± 1.3||5.6 ± 1.3||<0.001|
|Ocular hypertension||83 (15.3)||10 (25.0)||73 (14.6)||0.078|
|IOP (mm Hg)|
|Baseline IOP||16.2 ± 3.9||16.7 ± 5.8||16.1 ± 3.7||0.295|
|Mean follow-up IOP||14.7 ± 2.5||14.9 ± 2.5||14.7 ± 2.5||0.388|
|Peak follow-up IOP||18.6 ± 4.9||19.5 ± 5.5||18.5 ± 4.8||0.098|
|Range of follow-up IOP||6.9 ± 4.7||7.9 ± 5.4||6.9 ± 4.7||0.077|
|Fluctuation of follow-up IOP||2.1 ± 1.2||2.3 ± 1.3||2.1 ± 1.2||0.118|
|Visual field measurement|
|Baseline MD (dB)||−0.79 ± 1.34||−0.93 ± 1.20||−0.79 ± 1.36||0.493|
|Baseline PSD (dB)||1.84 ± 0.88||1.84 ± 0.61||1.86 ± 1.06||0.772|
|Final MD (dB)||−0.88 ± 1.83||−3.05 ± 3.16||−0.71 ± 1.57||<0.001|
|Final PSD (dB)||2.02 ± 1.43||4.89 ± 3.42||1.84 ± 0.94||0.021|
|Rate of change in MD (dB/year)||0.09 ± 0.34||−0.31 ± 0.37||0.11 ± 0.32||<0.001|
|Rate of change in PSD (dB/year)||0.03 ± 0.22||0.40 ± 0.49||0.01 ± 0.16||<0.001|
|Macular GCIPL measurement|
|Baseline average GCIPL thickness (μm)||78.6 ± 7.0||74.8 ± 7.3||78.9 ± 6.9||<0.001|
|Final average GCIPL thickness (μm)||77.4 ± 7.5||71.0 ± 7.9||77.9 ± 7.2||<0.001|
|Rate of change in average GCIPL thickness (μm/y) a||−0.23 ± 0.37||−0.71 ± 0.57||−0.19 ± 0.32||<0.001|
|Peripapillary RNFL measurement|
|Baseline average RNFL thickness (μm)||86.9 ± 10.5||83.2 ± 8.8||87.2 ± 10.5||0.010|
|Final average RNFL thickness (μm)||85.1 ± 10.5||76.9 ± 9.9||85.7 ± 10.3||<0.001|
|Rate of change in average RNFL thickness (μm/y) a||−0.33 ± 0.70||−1.13 ± 0.85||−0.27 ± 0.64||<0.001|