Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Sep 1.
Published in final edited form as: Ophthalmol Glaucoma. 2021 Jan 30;4(5):541–549. doi: 10.1016/j.ogla.2021.01.003

Progressive Thinning of Retinal Nerve Fiber Layer and Ganglion Cell-Inner Plexiform Layer in Glaucoma Eyes with Disc Hemorrhage

Xiongfei Liu 1, Alicia Lau 1, Huiyuan Hou 1, Sasan Moghimi 1, James A Proudfoot 1, Eric Chan 1, Jiun Do 1, Andrew Camp 1, Derek Welsbie 1, Carlos Gustavo de Moraes 2, Christopher A Girkin 3, Jeffrey M Liebmann 4, Robert N Weinreb 1
PMCID: PMC8322143  NIHMSID: NIHMS1668443  PMID: 33529795

Abstract

Purpose:

To evaluate the thinning of circumpapillary retinal nerve fiber layer (cpRNFL) and macular ganglion cell-inner plexiform layer (mGCIPL) in primary open angle glaucoma (POAG) eyes with and without history of disc hemorrhage (DH).

Design:

Observational cohort study

Subjects:

39 eyes (34 subjects) with DH and 117 eyes (104 subjects) without DH from the Diagnostic Innovations in Glaucoma Study (DIGS) and the African Decent and Glaucoma Evaluation Study (ADAGES).

Methods:

All participants had at least 1.5-years of follow-up with a minimum of 3 visits with bi-annual spectral-domain optic coherence tomography (SD-OCT) cpRNFL and mGCIPL thickness measurements and visual fields (VF). The rates of cpRNFL and mGCIPL thinning were calculated using mixed effects models. The dynamic range-based normalized rates of cpRNFL and mGCIPL were calculated and compared between the DH and non-DH groups.

Main Outcome Measures:

Rates of cpRNFL and mGCIPL thinning

Results:

The rate of mGCIPL thinning was significantly faster in the DH group compared to the non-DH group (−0.62 vs. −0.38 μm/year, P=0.024). The rate of cpRNFL thinning in the DH quadrant and rate of mGCIPL thinning in inferotemporal sector in the DH group were faster than the corresponding regions in the non-DH group after adjusting for intraocular pressure (IOP) and race (−1.33 vs. −0.58 μm/year, P=0.053, and (−0.82 vs. −0.44 μm/year, P = 0.048, respectively). In the DH group, the percent rate of loss was significantly faster with mGCIPL than cpRNFL (−1.59 vs. −1.31 %/year, P=0.046). mGCIPL thinning rates were weakly associated with MD slope, VFI slope, and GPA (R2 = 3.6%(P=0.058), R2=2.4%(P=0.096), and R2= 5.2%(P=0.073), respectively). The area under operating receiver curves for VF progression based on Guided Progression Analysis was 0.75 for mGCIPL and 0.56 for cpRNFL in the DH group.

Conclusions:

In our study, the rate of mGCIPL and cpRNFL thinning was faster in DH eyes than non-DH eyes. mGCIPL showed higher proportional rates of thinning and greater association with functional progression compared to cpRNFL. In addition to cpRNFL, clinicians should consider incorporating mGCIPL imaging to monitor glaucoma progression, especially in glaucoma eyes with DH.

Precis

In glaucoma eyes with disc hemorrhage, the rate of macula ganglion cell-inner plexiform layer (mGCIPL) thinning is faster than that of circumpapillary retinal nerve fiber layer (cpRNFL), and also has greater association with functional deterioration.


Glaucoma is a progressive optic neuropathy characterized by loss of retinal ganglion cells (RGC) and their axons.1 Approximately 50% of RGC are found within 4.5mm of the fovea and the ganglion cell layer and retinal nerve fiber layer (RNFL) accounts for 30–35% of the retinal thickness in this area and are useful measures for detecting glaucomatous damage.2, 3 Spectral domain optical coherence tomography (SD-OCT) can provide quantitative assessment of the circumpapillary RNFL (cpRNFL) and the macular ganglion cell-inner plexiform layer (mGCIPL). Studies have found that both SD-OCT cpRNFL4, 5 and mGCIPL6, 7 have high reproducibility. Furthermore, studies have shown that OCT mGCIPL is comparable or better than OCT cpRNFL in its diagnostic performance to properly differentiate glaucoma and healthy subjects.8, 9 Disc hemorrhage (DH) is a significant risk factor for development and progression of glaucoma.1013 Gracitelli et al14 reported that DH eyes had faster RGC loss rates than eyes without DH. However, there are limited studies that have used SD-OCT to evaluate glaucoma progression with cpRNFL in DH eyes,1517 and there was only one study where Lee et al18 evaluated glaucoma progression and rate of change in DH eyes with both OCT cpRNFL and OCT mGCIPL. Little is known about whether ONH and macula changes in DH eyes follow the same pattern as non-DH eyes. Furthermore, it is unclear if structural progression of OCT cpRNFL and OCT mGCIPL can help predict functional progression in DH eyes. Therefore, evaluating the rates of change between cpRNFL and mGCIPL in DH and non-DH eyes and studying their relation to functional change should improve our understanding of SD-OCT and its role in detecting glaucoma progression in DH and non-DH eyes.

In this study, we evaluated the thinning of cpRNFL and mGCIPL in primary open angle glaucoma (POAG) eyes with and without a history of DH. In addition, we examined the association between these thinning rates and visual field (VF) progression.

Methods

Participants

Participants were enrolled in the longitudinal Diagnostic Innovations in Glaucoma Study (DIGS) and the African Descent and Glaucoma Evaluation Study (ADAGES). The protocols of the two studies are identical, and the methodological details have been described previously.18 All patients from the DIGS and ADAGES who met the inclusion criteria described below were enrolled in the present study. Informed consent was obtained from all participants. This prospectively designed study received institutional review board approval at each of the involved sites. The methodology adhered to the tenets of the Declaration of Helsinki and to the Health Insurance Portability and Accountability Act.

Eligible participants were aged > 18 years, had best-corrected visual acuity of 20/40 or better, spherical refraction within ±5.0 diopters (D), cylinder correction within ±3.0 D, and open angles on gonioscopy at study entry. Participants were excluded if they had a history of intraocular surgery except for uncomplicated cataract or glaucoma surgery. Complicated surgeries pertained to those that occurred either intraoperatively or post-operatively. Eyes with coexisting retinal disease except minimal non-proliferative diabetic retinopathy, uveitis, non-glaucomatous optic neuropathy, trauma, diagnosis of Parkinson’s disease, Alzheimer’s disease, dementia, history of stroke were excluded from the investigation. Each participant underwent a comprehensive ophthalmologic examination including review of medical and family history, best-corrected visual acuity testing, central corneal thickness (CCT) measurement using an ultrasound pachymeter (Pachette GDH 500; DGH Technology, Inc, Philadelphia, Pennsylvania, USA), slit-lamp biomicroscopy, gonioscopy, intraocular pressure (IOP) with Goldmann applanation tonometry, and dilated funduscopic examination. Stereoscopic optic disc photography (Kowa WX3D; Kowa Optimed, Inc, Torrance, California, USA or Nidek 3Dx; Nidek Inc, Fremont, California, USA), standard automated perimetry with the 24-2 Swedish Interactive Threshold Algorithm (SAP-SITA, Humphrey Field Analyzer; Carl Zeiss Meditec, Dublin, California, USA), and Cirrus HD-OCT cpRNFL and mGCIPL were obtained. The quality of VFs was reviewed by the VF Assessment Center (VisFACT) staff according to a standard protocol. Axial length (AL) was acquired with IOLMaster (Carl Zeiss Meditec, Dublin, California, USA).

Healthy subjects were defined as eyes with an IOP < 22 mmHg, without suspicious optic disc appearance for glaucoma, and no repeatable glaucomatous VF damage. Glaucoma suspects were defined as eyes with an IOP ≥22 mmHg and/or with an optic disc appearance that was suspicious for glaucoma but without evidence of repeatable glaucomatous VF damage. A suspicious appearing optic disc was defined as a disc with excavation, neuroretinal rim thinning or notching, or a localized or diffuse retinal nerve fiber layer defect suggestive of glaucoma with stereophotographs.18 POAG eyes showed reliable and repeatable glaucomatous VF damage. Glaucomatous VF damage was defined as a glaucoma hemifield test result outside normal limits and a Pattern Standard Deviation (PSD) outside 95% normal limits, and this was confirmed on at least two consecutive, reliable (≤33% fixation losses and false-negative results and ≤15% false-positive results) tests. The glaucoma disease stages were classified as early (mean deviation (MD) > −6 dB), moderate (−12dB ≤ MD ≤ −6dB), and severe (MD < −12 dB).

All participants were included who had at least one DH on stereophotography before OCT follow-up and had at least 3 visits over at least 1.5 years with OCT cpRNFL, OCT mGCIPL, and HVF completed at each visit of sufficient quality. When the eye had multiple DHs, the DH date closest to the baseline OCT date was determined as the baseline DH. The non-DHs were GS and POAG eyes enrolled from DIGS and ADAGES without history of DH. The non-DH group and the DH groups were matched based on age and baseline MD with a ratio of 3:1. For normalization of dynamic range, healthy subjects were included in addition to non-DH and DH groups.

Stereoscopic optic disc photography

All patients had stereoscopic optic disc photographs repeated at least every 12 months during follow-up. The images were reviewed with a stereoscopic viewer (Screen-VU stereoscope; PS Manufacturing, Portland, OR) by 2 or more experienced graders masked to the subjects’ identity and to other test results. The location of the DH was classified into four 90-degree sectors (superior, inferior, temporal, and nasal). The methodology used to grade optic disc photographs at the UCSD Optic Disc Reading Center has been provided elsewhere.19 Only photographs of adequate quality were included. Discrepancies between the 2 graders were resolved by consensus or adjudication by a third experienced grader. For this study, DHs were defined as located within 1/2 disc diameter from the optic disc border or within the cpRNFL as a splinter or flame-shaped hemorrhage and were not associated with optic disc edema, papillitis, diabetic retinopathy, central or branch retinal vein occlusion, or any other retinal disease.19, 20

Cirrus Spectral-Domain Optical Coherence Tomography

Measurements of cpRNFL and mGCIPL thicknesses were acquired using Cirrus HD-OCT (software v. 6.5; Carl Zeiss Meditec Inc, Dublin, California, USA). Cirrus HD-OCT uses a super luminescent diode scan with a center wavelength of 840 nm and an acquisition rate of 27,000 A-scans per second at an axial resolution of 5 mm. The protocol used for cpRNFL thickness measurement was the optic disc cube 200 × 200 protocol with cpRNFL thickness measurements calculated from a 3.46-mm-diameter circular scan automatically placed around the optic disc. The cpRNFL thickness measurements are divided into 4 sectors and separated into 90-degree intervals (superior, inferior, temporal, and nasal) using the same criteria as the DHs in the stereophotographs. The protocol used for the mGCIPL thickness measurement is the macular cube 200 × 200 protocol. This protocol is based on a 3-dimensional scan centered on the macula in which information from a 1024 (depth) × 200 × 200- point parallelepiped is collected. The ganglion cell analysis algorithm automatically segments the mGCIPL based on 3-dimensional data generated from the macular cube scan protocol. The algorithm identifies the outer boundary of cpRNFL and the outer boundary of inner plexiform layer at the macular region; this segmented layer yields mGCIPL thickness. The ganglion cell analysis algorithm reports the average of the mGCIPL thicknesses over six sectoral areas (superotemporal, superior, superonasal, inferonasal, inferior, and inferotemporal) that form an elliptical annulus around the fovea, as well as the overall average for the annulus. Image quality was evaluated by an experienced examiner from the UCSD Imaging Data Evaluation and Assessment Reading Center. To be included in the analysis, images were designated as good quality if the signal strength was greater than 6 and the image was centered on the optic disc or fovea for the optic disc and macular cube protocols, respectively. Only scans with adequate algorithm segmentation and without overt segmentation algorithm failure were included in the study.

Standard Automated Perimetry

All standard automated perimetry was performed using the 24–2 Swedish interactive thresholding algorithm (SITA) (Humphrey Field Analyzer; Carl Zeiss Meditec, Dubin, CA). Only repeatable and reliable tests were included. The HVFs used were within one year of all OCT scans. Guided Progression Analysis (GPA) was used to determine event-based VF progression, with likely progression counted as progression and possible progression not counted as progression. Mean Deviation (MD) and Visual Field Index (VFI) slopes were used for trend-based progression analysis.

Statistical Analysis

Continuous and categorical data are presented as mean (95% confidence interval, CI) and count (%). The statistical significance of differences in patient characteristics between the DH group and non-DH group was determined by two-sample t-tests for continuous variables and Fisher’s exact test for categorical variables. Eye characteristics were compared using linear mixed effects models with random intercepts to account for inter-subject variability. cpRNFL thickness and mGCIPL thickness trajectories were estimated using linear mixed effects models with random eye-within-patient intercepts and independent random slopes-within-eye. These models include fixed effects for DH history, time, and their interaction.

A forward-selection procedure was used to determine additional fixed covariates (mean IOP during follow up, race, and their interactions with time) for multivariable analysis. In order to perform inter-group comparisons of cpRNFL thickness and mGCIPL thickness change rates, dynamic-range based normalized coefficients were estimated, as previously described.21, 22 To come up with dynamic-range normalization, healthy eyes were included as well. In brief, the dynamic range of each measurement was estimated by calculating the mean value of the top and bottom 3% of eyes and the percent of dynamic range change was calculated as [(visit value - floor value)/dynamic range] ×100/year. The unit of the normalized coefficients (%/year) is annual percent change of the dynamic range. Comparisons of normalized slopes were conducted by fitting linear mixed effects models stratified by DH history for both normalized cpRNFL and mGCIPL thickness. A fixed indicator effect for outcome and its interaction with time was tested. DH history stratified mixed models were also fit to investigate the interaction of time and baseline age, race, gender, mean IOP during follow-up, baseline visual field MD, AL, and CCT on cpRNFL and mGCIPL thinning. Linear regressions and area under operating receiver curves (AUC) were used to evaluate the association between change rates of cpRNFL and mGCIPL thickness and VF progression parameters. P values less than 0.05 were considered statistically significant, and P values between 0.05 and 0.1 were considered borderline significant. Statistical analyses were performed using Stata 14.2 (StataCorp LLC, College Station, TX).

Results

34 subjects (39 eyes) in the DH group and 104 subjects (117 eyes) in the non-DH group were included. Baseline clinical characteristics are summarized in Table 1. The mean interval between DH and baseline OCT imaging was 2.69 (95% CI 1.7, 3.6) years. The average follow-up and number of visits with OCTs were similar between both groups, with 3.1 years and 5.4 visits for DH group, and 3.3 years and 5.7 visits for the non-DH group. A higher proportion of individuals of European descent were present in the DH group compared to the non-DH group (P = 0.003). In the DH group, 59% of DHs occurred in the inferior quadrant, 41% of the eyes in DH group had multiple DHs, and the average number of total DHs per eye was 1.8. Baseline IOP as well as mean IOP during follow-up periods were lower in the DH group compared to the non-DH group (P = 0.011 and P = 0.017 respectively). The mean number of IOP lowering medications was higher in the DH group (1.4) compared to the non-DH group (1.2) with borderline significance (P = 0.099). There were no significant differences between the groups in age, gender, distribution of glaucoma severity, CCT, baseline MD, baseline PSD, baseline cpRNFL, and baseline mGCIPL (all P > 0.1).

Table 1.

Demographic and Baseline Ophthalmic Characteristics of DH and Non-DH Groups During Spectral-Domain Optical Coherence Tomography Follow-up

DH group Non-DH group P value
By Subject (No.) 34 104
Gender (M/F, No.) 13/21 48/56 0.418
Age (years) 70.5 (67.1, 74.0) 69.9 (68.2, 71.6) 0.714
Race, No. (%) 0.003
 European Descent 28 (82.4%) 57 (54.8%)
 African Descent 4 (11.8%) 43 (41.4%)
 Asian 2 (5.9%) 4 (3.9%)
By Eye (No.) 39 117
Follow-up (years) 3.1 (2.8, 3.3) 3.3 (3.1, 3.4) 0.281
Visits of OCT 5.4 (4.7, 6.0) 5.7 (5.3, 6.2) 0.343
Diagnosis (glaucoma/glaucoma suspect), No. 31/8 87/30 0.519
Axial length (mm) 23.7 (23.4, 24.0) 23.8 (23.6, 24.0) 0.627
CCT (μm) 552.5 (538.6, 566.4) 538.8 (531, 546.5) 0.110
Baseline IOP (mmHg) 14.2 (13.0, 15.3) 15.8 (15.1, 16.5) 0.011
Mean IOP during follow-up (mmHg) 13.9 (12.9, 14.9) 15.3 (14.7, 15.8) 0.017
Mean total DH No. 1.79 N/A
 Multiple DH, Eye No. (%) 16 (41.0%) N/A
DH Location N/A
 Inferior, Eye No. (%) 23 (59.0%)
 Superior, Eye No. (%) 10 (25.6%)
 Nasal, Eye No. (%) 2 (5.1%)
 Temporal, Eye No. (%) 7 (17.9%)
Baseline MD (dB) −2.5 (−3.9, −1.2) −2.5 (−3.4, −1.7) 0.968
Severity, Eye No. (%) 0.784
 Early 34 (87.18%) 106 (90.60%)
 Moderate 3 (7.69%) 4 (3.42%)
 Advanced 2 (5.13%) 7 (5.98%)
Baseline PSD (dB) 4.2 (3.1, 5.4) 3.4 (2.9, 4.0) 0.205
Baseline cpRNFL thickness (μm) 76.0 (72.3, 80.0) 77.0 (74.5, 79.6) 0.667
Baseline mGCIPL thickness (μm) 72.2 (68.9, 75.5) 70.1 (68.3, 71.9) 0.282
Glaucoma Procedure (N/Y, No.) 22/17 81/36 0.187
IOP lowering medication during follow-up (No.) 1.4 (1.2, 1.7) 1.2 (1.0, 1.4) 0.099
Interval between DH and first OCT (years) 2.69 (1.73, 3.64) N/A
GPA with Likely Progression n (%) 10 (25.6%) 11 (9.4%)

DH=eyes with disc hemorrhage history; Non-DH=eyes without disc hemorrhage history; M=male; F=female; CI=confidence interval; CCT=central corneal thickness; IOP=intraocular pressure; MD=mean deviation; PSD=pattern standard deviation; cpRNFL=retinal nerve fiber layer; mGCIPL = ganglion cell-inner plexiform layer; N=no; Y=yes; GPA=guided progression analysis.

Values are shown as mean (95% confidence interval) unless otherwise indicated. Statistically significant and borderline significant P values are shown in bold. The first OCT exam was considered the baseline.

Table 2 summarizes the rates of cpRNFL thinning and mGCIPL thinning in the DH group and the non-DH group. Both the global cpRNFL thinning rate and the global mGCIPL thinning rate were significant in the DH group and the non-DH group (all P < 0.0001). For the DH group, the inferior quadrant (DH quadrant) for cpRNFL and the inferior sectors for mGCIPL were included in eyes with history of inferior DH. For comparison, the superior quadrant of the cpRNFL (non-DH quadrant) and superior sectors of the mGCIPL were included only in eyes without history of superior DH. These regions were then compared to the respective regions from eyes in the non-DH group. Whereas the rate of mGCIPL thinning was significantly faster in the DH group (−0.62μm/year) compared to that of the non-DH group (−0.38μm/year, P<0.05 in both univariable and multivariable models), no such significant difference was observed in the global RNFL thinning rates (P>0.1). When specifically comparing the cpRNFL rates in DH quadrants, a faster trend was observed in the DH group compared with the rate of the corresponding quadrant in the non-DH group and showed borderline significance (−1.33 μm/year vs. −0.58 μm/year, P = 0.053) after adjusting for IOP during follow-up and race. Similarly, the inferotemporal sector of the DH group had a significantly faster rate of thinning compared to corresponding sector in non-DH group (−0.82 μm/year vs. −0.44 μm/year, P = 0.048). No significant difference in the rates of RNFL thinning was observed in the non-DH quadrant for cpRNFL, or superior sectors for GCIPL between the DH and non-DH groups in both univariable and multivariable models (P>0.1). Table 3 summarizes the dynamic range-based normalized rates of cpRNFL and mGCIPL thickness change. Similar to original rates shown in Table 2, mGCIPL thinning of the DH group was significantly faster than the non-DH group (P=0.034). When comparing cpRNFL thinning to mGCIPL thinning within each group, notably, mGCIPL thinning (−1.59%/year) was faster than cpRNFL thinning (−1.31%/year) with significance (P=0.046) in the DH group. However, in the non-DH group, cpRNFL thinning and mGCIPL thinning had similar rates (P>0.1).

Table 2.

Rates of Retinal Nerve Fiber Layer and Ganglion Cell Inner Plexiform Layer Thinning in DH and Non-DH Eyes.

DH group Non-DH group P Value
Mean (95% CI) Mean (95% CI) Univariate Multivariate
ONH RNFL Thickness Change Rate (μm/year)
No. of Eye 39 117
Global −0.71 (−1.22, −0.19), −0.49 (−0.73, −0.25), 0.462 0.196
P* <.0001 <.0001
DH Quadrant RNFL Thickness Change Rate (μm/year)
No. of Eye 23 117
Inferior −1.33 (−2.16, −0.50), −0.58 (−1.06, −0.10), 0.136 0.053
P* 0.002 0.018
Quadrant without DH RNFL Thickness Change Rate (μm/year)
No. of Eye 27 117
Superior −0.37 (−1.27, 0.53) −0.72 (−1.12, −0.32) 0.454 0.616
P* 0.419 <.0001
Macula GCIPL Thickness Change
No. of Eye 39 117
Rate (μm/year) −0.62 (−0.87, −0.37), −0.38 (−0.57, −0.20), 0.034 0.024
<.0001 <.0001
GCIPL Thickness Change Rate (μm/year) Correspond to DH quadrant
No. of Eye 23 117
Inferonasal −0.66 (−1.14, −0.18) −0.43 (−0.64, −0.22) 0.169 0.130
P* 0.007 <.0001
Inferior −0.56 (−0.95, −0.18) −0.44 (−0.74, −0.14) 0.547 0.551
P* 0.004 0.004
Inferotemporal −0.82 (−1.27, −0.37) −0.44 (−0.69, −0.19) 0.089 0.048
P* <.0001 0.001
GCIPL Thickness Change Rate (μm/year) Correspond to Non-DH quadrant
No. of Eye 39 117
Superonasal −0.16 (−0.5, 0.18) −0.3 (−0.57, −0.03) 0.753 0.851
P* 0.360 0.028
Superior −0.41 (−0.72, −0.09) −0.46 (−0.78, −0.15)
P* 0.011 0.004 0.959 0.877
Superotemporal −0.36 (−0.71, −0.02) −0.26 (−0.58, 0.07)
P* 0.039 0.127 0.765 0.766

DH group= eyes with disc hemorrhage history; Non-DM group= eyes without disc hemorrhage history; CI= confidence interval; IOP=intraocular pressure. ONH = optic nerve head; RNFL = retinal nerve fiber layer; GCIPL = ganglion cell inner plexiform layer

Statistically significant and borderline significant P values are shown in bold. P* value represents comparison of rate from 0. Mean IOP during OCT follow-up and race were adjusted in the multivariate mixed effects model.

Table 3.

Dynamic Range-based Normalized Rates of Retinal Nerve Fiber Layer and Ganglion Cell Inner Plexiform Layer Thickness in DH and Non-DH eyes

Dynamic range-based normalized rate (%/year) cpRNFL vs. mGCIPL
P value
cpRNFL mGCIPL
DH group −1.31 (−2.27, −0.35) −1.59 (−2.23, −0.94) 0.046
Non-DH group −0.91 (−1.36, −0.46) −0.98 (−1.45, −0.50) 0.727
DH vs. Non-DH P value 0.462 0.034

cpRNFL=circumpapillary retinal nerve fiber layer; mGCIPL=macular ganglion cell inner plexiform layer; DH group=eyes with history of disc hemorrhage; non-DH group=eyes without history of disc hemorrhage

Percent loss rate was calculated as [(visit value - floor value)/dynamic range] ×100/year, is shown as mean (95% confidence interval). Statistically significant and borderline significant P values are shown in bold.

Potential factors that could be associated with the rate of mGCIPL and cpRNFL thinning during follow-up, are summarized in Table 4. In the DH group, every 1 mmHg increase in mean IOP during follow-up was associated with 0.13 μm/year faster cpRNFL thinning rate (P= 0.020) in the DH group, and 0.04 μm/year faster mGCIPL thinning rate (P=0.028) in the non-DH group. Every 1 dB increase in baseline MD was associated with 0.13μm/year faster cpRNFL thinning rate in the DH group (P<0.001).

Table 4.

Results of Univariable Models Assessing the Effect of Each Putative Predictive Factor on Retinal Nerve Fiber Layer Thinning and Macula Ganglion Cell Inner Plexiform Thinning Over Time in DH and Non-DH Groups

DH Group (Coefficients) Non-DH Group (Coefficients)
Variables cpRNFL (95% CI), P mGCIPL (95% CI), P cpRNFL (95% CI), P mGCIPL (95% CI), P
Age at baseline, per year older −0.01 (−0.05, 0.02), 0.421 −0.02 (−0.05, 0.003), 0.096 0.009 (−0.02, 0.03), 0.475 0.007 (−0.02, 0.03), 0.630
Race, African descent vs. not 0.18 (−0.34, 0.69), 0.498 0.001 (−0.36, 0.36), 0.995 −0.25 (−0.71, 0.20), 0.277 0.05 (−0.25, 0.35), 0.740
Gender, Female −0.99 (−2.03, 0.05), 0.063 0.20 (−0.28, 0.67), 0.418 −0.09 (−0.56, 0.38), 0.703 0.13 (−0.22, 0.48), 0.465
Mean IOP, per 1mmHg higher −0.13 (0.24, −0.02), 0.020 −0.05 (−0.12, 0.01), 0.130 −0.05 (−0.12, 0.01), 0.119 −0.04 (−0.08, −0.005), 0.028
Baseline MD, per 1 dB better −0.13 (−0.18, −0.07), <.0001 −0.02 (−0.04, 0.005), 0.135 −0.003 (−0.03, 0.03), 0.876 −0.007 (−0.03, 0.01), 0.461
CCT, per 100 μm thicker 0.14 (−1.43, 1.73), 0.858 −0.06 (−0.55, 0.44), 0.825 0.61 (−0.14, 1.35), 0.111 0.22 (−0.35, 0.80), 0.449
AL, per 1 mm longer 0.22 (−0.59, 1.03), 0.595 0.12 (−0.13, 0.37), 0.357 0.14 (−0.04, 0.32), 0.134 0.04 (−0.07, 0.15), 0.522

IOP=intraocular pressure; MD= mean deviation; CCT= central corneal thickness; AL= axial length. cpRNFL = circumpapillary retinal nerve fiber layer; mGCIPL = ganglion cell inner plexiform layer, DH group = eyes with history of disc hemorrhage; non-DH group = eyes without history of disc hemorrhage

Values are shown as mean (95% confidence interval). Statistically significant and borderline significant P values are shown in bold.

Table 5 shows the performance of rates of cpRNFL and mGCIPL change and their ability to detect VF progression based on GPA criteria in the DH group and non-DH group. In both the DH group and the non-DH group, the AUC for discriminating VF progression was higher for mGCIPL (AUC =0.75; AUC = 0.68 respectively) than for cpRNFL (AUC =0.56 and 0.49, respectively).

Table 5.

Diagnostic Ability of Retinal Nerve Fiber Layer Thinning and Ganglion Cell Inner Plexiform Layer Thinning in Differentiating Visual Field Progression

AUC for VF Progression Based on GPA Likely Progression (95% CI)
cpRNFL Thinning Rate mGCIPL Thinning Rate
DH group 0.56 (0.36, 0.75) 0.75 (0.59, 0.91)
Non-DH group 0.49 (0.29, 0.69) 0.68 (0.47, 0.89)

DH group=eyes with history of disc hemorrhage; Non-DH group=eyes without history of disc hemorrhage; cpRNFL=circumpapillary retinal nerve fiber layer; mGCIPL=ganglion cell inner plexiform layer; AUC=area under receiver operating curves; GPA=guided progression analysis

Table 6 shows the association between cpRNFL and mGCIPL thinning rates with VF progression parameters. In the DH group, mGCIPL thinning rates were associated with MD slope (R2 = 3.6%, P = 0.058), VFI slope (R2=2.4%, P = 0.096), and GPA (R2= 5.2%, P = 0.073), respectively). Rates of global cpRNFL thinning were not significantly associated with MD slope, VFI slope, and GPA.

Table 6.

Association Between the Rates of Retinal Nerve Fiber Layer Thinning and Ganglion Cell Inner Plexiform Layer Thinning with the Rate of Visual Field Progression Parameters

Thinning MD Slope VFI Slope GPA (2 as stable)
Rate R2 (%) P value R2 (%) P value R2 (%) P value
DH group
cpRNFL 5.02 0.140 4.48 0.153 0.24 0.695
mGCIPL 3.57 0.058 2.42 0.096 5.24 0.073
Non-DH group
cpRNFL 0.15 0.802 0.10 0.854 2.66 0.258
mGCIPL 0.31 0.734 0.52 0.691 2.10 0.337

cpRNFL=circumpapillary retinal nerve fiber layer; mGCIPL=macular ganglion cell inner plexiform layer; MD=mean deviation; VFI=visual field index; GPA=guided progression analysis Statistically significant and borderline significant P values are shown in bold.

Discussion

In our study, mGCIPL thinning was faster than cpRNFL thinning in eyes with DH history. In addition, the mGCIPL thinning rate had a greater association with functional progression then cpRNFL thinning rates. This finding is supported by the study done by Lee et al18 regarding DH eyes who reported a similar average rate of mGCIPL thickness change (−0.78μm/year) (95% CI −0.87 to −0.37μm/year) when compared to our study (−0.62μm/year) (95% CI −0.87 to −0.37μm/year). While baseline MD was not reported in their study, they included patients with MD > −6dB which was similar to the majority of our subjects.

Several studies have shown that the rate of cpRNFL thinning increases after DH.16, 17, 2325 This is also reflected in our study as both the global and DH quadrant had a faster cpRNFL thinning rate when compared to the non-DH group. One reason that the global cpRNFL thinning rate was not statistically significant compared to the non-DH group may be the fact that cpRNFL thinning is topographically related to DH location.23, 24 Given that 59% of the DHs were inferiorly located in our study, we specifically looked at the inferior quadrant in both groups, and then incorporated subjects without history of DH in the superior quadrant for comparison. The superior quadrant rate of change was actually lower in the DH group compared to the non-DH group, which in turn affected the global rate of cpRNFL change. The lower IOP and higher number of glaucoma medications in the DH group likely contributed to the lower rate of cpRNFL change in the DH group’s superior quadrant. Akagi et al16 showed that intensified glaucoma treatment in DH eyes decreased the rate of cpRNFL thinning in the global, DH quadrant, and non-DH quadrant when compared to rates of cpRNFL thinning before DH. Similarly, we found a negative association between mean IOP during follow up and the rate of cpRNFL thinning. After adjusting for IOP, the rate of cpRNFL loss in the inferior quadrant was borderline significant. Another reason for the smaller difference in cpRNFL thinning rate between the DH and non-DH group might be related to differing rates of change over time. As our baseline OCT on average occurred 2.69 years after the DH, the cpRNFL thinning rates might have initially been more accelerated and then slowed over time. Consequently, this may have decreased the difference in cpRNFL thinning rates between the two groups.

Given that cpRNFL and mGCIPL have different dynamic ranges, we normalized the measurements by calculating the percent loss from the mean value. This conversion facilitates meaningful comparison between the rates of cpRNFL and mGCIPL thinning.22, 26 In our study, while the percent loss rate was similar between cpRNFL and mGCIPL for the non-DH group, the mean loss rate for mGCIPL was faster than cpRNFL in the DH group. Several possibilities could explain this difference. First, the rates of thinning after DH occurrence for mGCIPL and cpRNFL could have changed over time. Given the fact that the mean interval between DH and baseline OCT imaging was 2.69 (95% CI 1.7, 3.6) years prior to the baseline OCT analysis in our DH group, cpRNFL thinning could have progressed faster initially and then leveled off, while the mGCIPL rate of thinning may have started later or been more sustained over time. This is plausible as cpRNFL thinning rate was more closely and inversely associated with IOP during follow-up compared to the mGCIPL thinning rate in the DH group, so the cpRNFL thinning rate may have been more affected than the mGCIPL thinning rate after IOP lowering treatment. It is also possible that treatment was intensified based solely on cpRNFL thinning data and not mGCIPL thinning data. Multiple reports have shown that the cpRNFL thinning rate increased within 1 year after DH.16, 17, 23, 24 However, to our knowledge, no studies have shown the annual rate of change over multiple years after DH. Therefore, future studies could focus on evaluating the rate of change for cpRNFL and mGCIPL over time. Secondly, disc hemorrhages can also occur within the macular vulnerability zone (MVZ), so they may be detectable earlier with mGCIPL scans.28. The central 8 degrees of the macula has over 30% of RGCs,2 and also contains the MVZ that is aligned with the inferotemporal cRNFL.27 Additionally, research has shown that MVZ damage can be detected earlier in the inferotemporal region of the mGCIPL scan compared to the topographically related cpRNFL scan.28 This was consistent with our finding that inferotemporal sector of GCIPL had significantly faster rate of thinning in DH eyes compared to non-DH eyes. A third possibility might be that in advanced glaucoma, mGCIPL tends to have faster rates of thinning and reaches the measurement floor later in the disease course than cpRNFL.2932 However, this was not the case in our study as the majority of our DH cohort had early glaucoma (87%).

Multiple studies have shown that eyes with DHs eventually develop VF progression and decreased cpRNFL thickness,15, 33 and some studies have made direct associations between VF and cpRNFL thickness.34 In our study, the mGCIPL rate of change had a higher AUC value for VF progression based on GPA analysis. Furthermore, the association between mGCIPL rate and MD slope, VFI slope, or GPA were borderline significant. cpRNFL rate was not associated with MD slope, VFI slope, or GPA. Shukla et al35 found that DH eyes were associated with more central damage in 24–2 and 10–2 VFs, so mGCIPL may be better than cpRNFL in detecting structural progression if there is central deterioration.

There are several limitations to our study. While the sample size is adequate for a longitudinal study, a larger sample size might affect several parameters that had borderline significance. The majority of the eyes in our cohort had early glaucoma. Therefore, our findings cannot be generalized to moderate or severe glaucoma where the rate of cpRNFL and mGCIPL thinning might be different. Moreover, optic disc photos are captured annually, therefore some DHs might have been missed in both groups. However, we primarily compared mGCIPL to cpRNFL within the same DH group, so the underestimation of the number of DHs as a potential confounding variable should have been minimized. Additionally, around 2.69 years on average elapsed between DH occurrence and baseline OCT, therefore the rates of cpRNFL and mGCIPL change should not be generalized to the time frame immediately following DH.

Furthermore, we did not breakdown OCT parameters into different sectors and included only the superior and inferior cpRNFL quadrants which were the quadrants most likely to incur DHs. Although we attempted to exclude patients with DH from causes other than glaucoma, such as hypertension or posterior vitreous detachment, it is possible that such patients were included. Finally, the ethnic makeup for the DH and non-DH groups were significantly different. Therefore, race was included as an additional fixed covariate in the statistical models to evaluate the cpRNFL and mGCIPL rates in the DH and non-DH groups.

In conclusion, we showed that the rate of mGCIPL and cpRNFL thinning was faster in DH eyes than non-DH eyes, and the rate of mGCIPL thinning was faster than the rate of cpRNFL thinning in DH eyes. In addition, mGCIPL showed greater association with functional progression compared to cpRNFL. Since mGCIPL potentially reflects a steadier and more prolonged thinning rate after DH compared to cpRNFL, clinicians should consider incorporating mGCIPL imaging in addition to cpRNFL imaging for monitoring glaucoma progression in patients with DH.

Financial Support:

National Institutes of Health/National Eye Institute Grants R01EY029058, R01EY011008, U10EY14267, R01EY026574, R01EY019869 and R01EY027510; Core Grant P30EY022589, an unrestricted grant from Research to Prevent Blindness (New York, NY); and grants for participants’ glaucoma medications from Alcon, Allergan, Pfizer, Merck, and Santen. The sponsor or funding organizations had no role in the design or conduct of this research.

Abbreviations and Acronyms:

DIGS

Diagnostic Innovations in Glaucoma Study

ADAGES

African Descent and Glaucoma Evaluation Study

UCSD

University of California - San Diego

CCT

central corneal thickness

SD

spectral-domain

OCT

optical coherence tomography

D

diopter

DH

disc hemorrhage

IOP

intraocular pressure

MD

mean deviation

PSD

pattern standard deviation

cpRNFL

circumpapillary retinal nerve fiber layer

mGCIPL

macular ganglion cell-inner plexiform layer

SITA

Swedish Interactive Thresholding Algorithm

VF

visual field

AL

axial length

POAG

primary open angle glaucoma

VFI

visual field index

GPA

glaucoma progression analysis

AUC

area under curve

MVZ

macular vulnerability zone

RGC

retinal ganglion cells

ONH

Optic nerve head

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Commercial Disclosures: Alicia Lau: none; Xiongfei Liu: none; Huiyuan Hou: none; Sasan Moghimi: none; James Proudfoot: none; Eric Chan: none; Jiun Do: none; Andrew Camp: none; Derek Welsbie: none; Carlos Gustavo De Moraes: none; Christopher A. Girkin: Research support - Carl Zeiss Meditec, EyeSight Foundation of Alabama, Heidelberg Engineering, National Eye Institute, Research to Prevent Blindness, SOLX; Jeffrey M Liebmann: Research support - Bausch & Lomb, Carl Zeiss Meditec, Heidelberg Engineering, National Eye Institute, Optovue, Reichert, Topcon; Consultant - Alcon, Allergan, Bausch & Lomb, Carl Zeiss Meditec, Valeant Pharmaceuticals, Reichert, Heidelberg Engineering; Robert N Weinreb, MD: Research support- Carl Zeiss Meditec, Centervue, Heidelberg Engineering, Konan, National Eye Institute, Optovue, Consultant- Aerie Pharmaceuticals, Allergan, Bausch& Lomb, Eyenovia, NiCox

References

  • 1.Weinreb RN, Aung T, Medeiros FA. The pathophysiology and treatment of glaucoma: a review. JAMA 2014;311(18):1901–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Curcio CA, Allen KA. Topography of ganglion cells in human retina. J Comp Neurol 1990;300(1):5–25. [DOI] [PubMed] [Google Scholar]
  • 3.Zeimer R, Asrani S, Zou S, et al. Quantitative detection of glaucomatous damage at the posterior pole by retinal thickness mapping. A pilot study. Ophthalmology 1998;105(2):224–31. [DOI] [PubMed] [Google Scholar]
  • 4.Mwanza JC, Chang RT, Budenz DL, et al. Reproducibility of peripapillary retinal nerve fiber layer thickness and optic nerve head parameters measured with cirrus HD-OCT in glaucomatous eyes. Invest Ophthalmol Vis Sci 2010;51(11):5724–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Roh KH, Jeoung JW, Park KH, et al. Long-term reproducibility of cirrus HD optical coherence tomography deviation map in clinically stable glaucomatous eyes. Ophthalmology 2013;120(5):969–77. [DOI] [PubMed] [Google Scholar]
  • 6.Mwanza JC, Oakley JD, Budenz DL, et al. Macular ganglion cell-inner plexiform layer: automated detection and thickness reproducibility with spectral domain-optical coherence tomography in glaucoma. Invest Ophthalmol Vis Sci 2011;52(11):8323–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Kim KE, Yoo BW, Jeoung JW, Park KH. Long-Term Reproducibility of Macular Ganglion Cell Analysis in Clinically Stable Glaucoma Patients. Invest Ophthalmol Vis Sci 2015;56(8):4857–64. [DOI] [PubMed] [Google Scholar]
  • 8.Mwanza JC, Durbin MK, Budenz DL, et al. Glaucoma diagnostic accuracy of ganglion cell-inner plexiform layer thickness: comparison with nerve fiber layer and optic nerve head. Ophthalmology 2012;119(6):1151–8. [DOI] [PubMed] [Google Scholar]
  • 9.Jeoung JW, Choi YJ, Park KH, Kim DM. Macular ganglion cell imaging study: glaucoma diagnostic accuracy of spectral-domain optical coherence tomography. Invest Ophthalmol Vis Sci 2013;54(7):4422–9. [DOI] [PubMed] [Google Scholar]
  • 10.Budenz DL, Anderson DR, Feuer WJ, et al. Detection and prognostic significance of optic disc hemorrhages during the Ocular Hypertension Treatment Study. Ophthalmology 2006;113(12):2137–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Drance S, Anderson DR, Schulzer M, Collaborative Normal-Tension Glaucoma Study G. Risk factors for progression of visual field abnormalities in normal-tension glaucoma. Am J Ophthalmol 2001;131(6):699–708. [DOI] [PubMed] [Google Scholar]
  • 12.Bengtsson B, Leske MC, Yang Z, et al. Disc hemorrhages and treatment in the early manifest glaucoma trial. Ophthalmology 2008;115(11):2044–8. [DOI] [PubMed] [Google Scholar]
  • 13.Medeiros FA, Alencar LM, Sample PA, et al. The relationship between intraocular pressure reduction and rates of progressive visual field loss in eyes with optic disc hemorrhage. Ophthalmology 2010;117(11):2061–6. [DOI] [PubMed] [Google Scholar]
  • 14.Gracitelli CP, Tatham AJ, Zangwill LM, et al. Estimated rates of retinal ganglion cell loss in glaucomatous eyes with and without optic disc hemorrhages. PLoS One 2014;9(8):e105611. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Chin YC, Perera SA, Tun TA, et al. Structural Differences in the Optic Nerve Head of Glaucoma Patients With and Without Disc Hemorrhages. J Glaucoma 2016;25(2):e76–81. [DOI] [PubMed] [Google Scholar]
  • 16.Akagi T, Zangwill LM, Saunders LJ, et al. Rates of Local Retinal Nerve Fiber Layer Thinning before and after Disc Hemorrhage in Glaucoma. Ophthalmology 2017;124(9):1403–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Hwang YH, Kim YY, Kim HK, Sohn YH. Changes in retinal nerve fiber layer thickness after optic disc hemorrhage in glaucomatous eyes. J Glaucoma 2014;23(8):547–52. [DOI] [PubMed] [Google Scholar]
  • 18.Sample PA, Girkin CA, Zangwill LM, et al. The African Descent and Glaucoma Evaluation Study (ADAGES): design and baseline data. Arch Ophthalmol 2009;127(9):1136–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Jonas JB, Iester M. Disc hemorrhage and glaucoma. Ophthalmology 1995;102(3):365–6. [DOI] [PubMed] [Google Scholar]
  • 20.Skaat A, De Moraes CG, Bowd C, et al. African Descent and Glaucoma Evaluation Study (ADAGES): Racial Differences in Optic Disc Hemorrhage and Beta-Zone Parapapillary Atrophy. Ophthalmology 2016;123(7):1476–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Hou H, Moghimi S, Proudfoot JA, et al. Ganglion Cell Complex Thickness and Macular Vessel Density Loss in Primary Open-Angle Glaucoma. Ophthalmology 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Hammel N, Belghith A, Weinreb RN, et al. Comparing the Rates of Retinal Nerve Fiber Layer and Ganglion Cell-Inner Plexiform Layer Loss in Healthy Eyes and in Glaucoma Eyes. Am J Ophthalmol 2017;178:38–50. [DOI] [PubMed] [Google Scholar]
  • 23.Kernstock C, Dietzsch J, Januschowski K, et al. Optical coherence tomography shows progressive local nerve fiber loss after disc hemorrhages in glaucoma patients. Graefes Arch Clin Exp Ophthalmol 2012;250(4):583–7. [DOI] [PubMed] [Google Scholar]
  • 24.Suh MH, Park KH, Kim H, et al. Glaucoma progression after the first-detected optic disc hemorrhage by optical coherence tomography. J Glaucoma 2012;21(6):358–66. [DOI] [PubMed] [Google Scholar]
  • 25.Park HL, Kim JW, Park CK. Choroidal Microvasculature Dropout Is Associated with Progressive Retinal Nerve Fiber Layer Thinning in Glaucoma with Disc Hemorrhage. Ophthalmology 2018;125(7):1003–13. [DOI] [PubMed] [Google Scholar]
  • 26.Zhang C, Tatham AJ, Abe RY, et al. Macular Ganglion Cell Inner Plexiform Layer Thickness in Glaucomatous Eyes with Localized Retinal Nerve Fiber Layer Defects. PLoS One 2016;11(8):e0160549. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Hood DC, Raza AS, de Moraes CG, et al. Glaucomatous damage of the macula. Prog Retin Eye Res 2013;32:1–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Kim YK, Jeoung JW, Park KH. Inferior Macular Damage in Glaucoma: Its Relationship to Retinal Nerve Fiber Layer Defect in Macular Vulnerability Zone. J Glaucoma 2017;26(2):126–32. [DOI] [PubMed] [Google Scholar]
  • 29.Belghith A, Medeiros FA, Bowd C, et al. Structural Change Can Be Detected in Advanced-Glaucoma Eyes. Invest Ophthalmol Vis Sci 2016;57(9):OCT511–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Bowd C, Zangwill LM, Berry CC, et al. Detecting early glaucoma by assessment of retinal nerve fiber layer thickness and visual function. Invest Ophthalmol Vis Sci 2001;42(9):1993–2003. [PubMed] [Google Scholar]
  • 31.Sung KR, Sun JH, Na JH, et al. Progression detection capability of macular thickness in advanced glaucomatous eyes. Ophthalmology 2012;119(2):308–13. [DOI] [PubMed] [Google Scholar]
  • 32.Lavinsky F, Wu M, Schuman JS, et al. Can Macula and Optic Nerve Head Parameters Detect Glaucoma Progression in Eyes with Advanced Circumpapillary Retinal Nerve Fiber Layer Damage? Ophthalmology 2018;125(12):1907–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Rasker MT, van den Enden A, Bakker D, Hoyng PF. Deterioration of visual fields in patients with glaucoma with and without optic disc hemorrhages. Arch Ophthalmol 1997;115(10):1257–62. [DOI] [PubMed] [Google Scholar]
  • 34.Rao HL, Pradhan ZS, Weinreb RN, et al. Optical Coherence Tomography Angiography Vessel Density Measurements in Eyes With Primary Open-Angle Glaucoma and Disc Hemorrhage. J Glaucoma 2017;26(10):888–95. [DOI] [PubMed] [Google Scholar]
  • 35.Shukla AG, Sirinek PE, De Moraes CG, et al. Disc Hemorrhages Are Associated With the Presence and Progression of Glaucomatous Central Visual Field Defects. J Glaucoma 2020;29(6):429–34. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES