Measurements of OCT Angiography Complement OCT for Diagnosing Early Primary Open-Angle Glaucoma

PURPOSE

To compare measurements of global and regional circumpapillary capillary density (cpCD) with retinal nerve fiber layer (RNFL) thickness and characterize their relationship with visual function in early primary open-angle glaucoma (POAG).

DESIGN

Cross-sectional study.

PARTICIPANTS

Eighty healthy eyes, 64 preperimetric eyes, and 184 mild POAG eyes from the Diagnostic Innovations in Glaucoma Study.

METHODS

Global and regional RNFL thickness and cpCD measurements were obtained using OCT and OCT angiography (OCTA). For direct comparison at the individual and diagnostic group level, RNFL thickness and capillary density values were converted to a normalized relative loss scale.

MAIN OUTCOME MEASURES

Retinal nerve fiber layer thickness and cpCD normalized loss at the individual level and diagnostic group. Global and regional areas under the receiver operating characteristic curve (AUROC) for RNFL thickness and cpCD to detect preperimetric glaucoma and glaucoma, R² for the strength of associations between RNFL thickness function and capillary density function in diagnostic groups.

RESULTS

Both global and regional RNFL thickness and cpCD decreased progressively with increasing glaucoma severity (P < 0.05, except for temporal RNFL thickness). Global and regional cpCD relative loss values were higher than those of RNFL thickness (P < 0.05) in preperimetric glaucoma (except for the superonasal region) and glaucoma (except for the inferonasal and superonasal regions) eyes. Race, intraocular pressure (IOP), and cpCD were associated with greater cpCD than RNFL thickness loss in early glaucoma at the individual level (P < 0.05). Global measurements of capillary density (whole image capillary density and cpCD) had higher diagnostic accuracies than RNFL thickness in detecting preperimetric glaucoma and glaucoma (P < 0.05; except for cpCD/RNFL thickness comparison in glaucoma [P = 0.059]). Visual function was significantly associated with RNFL thickness and cpCD globally and in all regions (P < 0.05, except for temporal RNFL thickness-function association [P = 0.070]).

CONCLUSIONS

Associations between capillary density and visual function were found in the regions known to be at highest risk for damage in preperimetric glaucoma eyes and all regions of mild glaucoma eyes. In early glaucoma, capillary density loss was more pronounced than RNFL thickness loss. Individual characteristics influence the relative magnitudes of capillary density loss compared with RNFL thickness loss. Retinal nerve fiber layer thickness and microvascular assessments are complementary and yield valuable information for the detection of early damages seen in POAG.

Rates of Retinal Nerve Fiber Layer Thinning in Distinct Glaucomatous Optic Disc Phenotypes in Early Glaucoma

PURPOSE

To compare spectral-domain optical coherence tomography (SDOCT) measured circumpapillary retinal nerve fiber layer (cpRNFL) among 4 glaucomatous optic disc phenotypes in early glaucoma.

DESIGN

Clinical cohort study METHODS: In this study, 218 early glaucoma eyes that had at least 3 years of follow-up and a minimum of 4 SDOCT scans were recruited. The optic discs were classified into 4 types based on appearance: 76 generalized cup enlargement (GE), 53 focal ischemic (FI), 22 myopic glaucomatous (MY), and 67 senile sclerotic (SS). A linear mixed effects model was used to compare the rates of global and regional cpRNFL thinning among optic disc phenotypes.

RESULTS

After adjusting for confounders, the SS group (mean [95% CI]: -1.01 [-1.30, -0.73] µm/y) had the fastest mean rate of global cpRNFL thinning followed by FI (-0.77 [-0.97, -0.57] µm/y), MY (0.59 [-0.81, -0.36] µm/y), and GE (-0.58 [-0.75, -0.40] µm/y) at P < .001. The inferior temporal sector had the fastest rate of cpRNFL thinning among the regional measurements except for the MY group (-0.68 [-1.10, -0.26] µm/y, P = .002). In the multivariable analysis, GE (P = .002) and MY (P = .010) phenotypes were associated with significantly slower global rates of cpRNFL thinning compared with the SS phenotype.

CONCLUSIONS

Rates of cpRNFL thinning were different among the 4 glaucomatous optic disc phenotypes. Those patients with early glaucoma with SS phenotype have the fastest cpRNFL thinning. These patients may benefit from more frequent monitoring and the need to advance therapy if cpRNFL thinning is detected.

Diurnal Variation of Retinal Vessel Density in Healthy Human Eyes

PRECIS

A small increase in optic nerve head vessel density (VD), but not macular VD, in the evening compared with the morning is observed in healthy subjects.

PURPOSE

To evaluate the diurnal variation of the macular and optic nerve head (ONH) VD in healthy eyes as measured with optical coherence tomography angiography (OCT-A).

METHODS

In this prospective study of healthy individuals older than 18 years old, VD parameters, including macular whole image vessel density, parafoveal vessel density, ONH whole image vessel density, ONH whole image capillary density, circumpapillary vessel density (cpVD), and circumpapillary capillary density, were measured with OCT-A at 4 time points throughout the day (8 am, 12 pm, 4 pm, and 8 pm).

RESULTS

Twenty-nine healthy eyes were included from 15 subjects (mean age: 30.9 y). After adjustment for age and mean ocular perfusion pressure, a significant positive rate of change was found for cpVD (0.05%/h; P=0.027). In contrast, macular VD changes were not significantly different. When comparing morning (8 am and 12 pm) and evening (4 pm and 8 pm) measurements, there were small, but statistically significant, increases for all ONH measurements. The greatest increase was found for cpVD (0.58%; P=0.022). Significant but small increases in mean arterial pressure and mean ocular perfusion pressure were also observed.

CONCLUSIONS

There was a small increase in ONH VD, but not macular VD, in the evening compared with the morning in healthy human eyes. As the observed difference was within the measurement variability, our results suggest the timing of OCT-A examination does not affect retinal VD measurements.

Qualitative Evaluation of the 10-2 and 24-2 Visual Field Tests for Detecting Central Visual Field Abnormalities in Glaucoma

PURPOSE

To examine whether glaucomatous central visual field abnormalities can be more effectively detected using a qualitative, expert evaluation of the 10-2 test compared with the topographically corresponding central 12 locations of the 24-2 test (C24-2).

DESIGN

Cross-sectional study.

METHODS

Eyes with a glaucomatous optic nerve appearance or ocular hypertension (n = 523) and healthy eyes (n = 107) were included as cases and control subjects, respectively. The 10-2 and C24-2 visual field results of all eyes were graded by 4 glaucoma specialists for the probability that central visual field abnormalities were present.

RESULTS

The sensitivity of the 10-2 and C24-2 tests for detecting the cases at 95% specificity were not significantly different (e.g., 32.2% and 31.4%, respectively, for grader 1, P = .87; all graders P ≥ .25). At 95% specificity, the pattern standard deviation values from these tests had a similar sensitivity to the qualitative evaluation for the C24-2 test for all graders (P ≥ .083), but it had a significantly higher sensitivity than the qualitative evaluation for the 10-2 test for 3 graders (P ≤ .016).

CONCLUSIONS

The similarity in performance of the 10-2 and C24-2 test suggests that the increased sampling density of the former does not significantly improve the detection of central visual field abnormalities, even when based on expert assessment. These findings should not be taken to mean that the 10-2 test is not useful, but it underscores the need for its utility to be clearly established before incorporating it as routine glaucoma standard of care.

Individualized Glaucoma Change Detection Using Deep Learning Auto Encoder-Based Regions of Interest

Purpose

To compare change over time in eye-specific optical coherence tomography (OCT) retinal nerve fiber layer (RNFL)-based region-of-interest (ROI) maps developed using unsupervised deep-learning auto-encoders (DL-AE) to circumpapillary RNFL (cpRNFL) thickness for the detection of glaucomatous progression.

Methods

Forty-four progressing glaucoma eyes (by stereophotograph assessment), 189 nonprogressing glaucoma eyes (by stereophotograph assessment), and 109 healthy eyes were followed for ≥3 years with ≥4 visits using OCT. The San Diego Automated Layer Segmentation Algorithm was used to automatically segment the RNFL layer from raw three-dimensional OCT images. For each longitudinal series, DL-AEs were used to generate individualized eye-based ROI maps by identifying RNFL regions of likely progression and no change. Sensitivities and specificities for detecting change over time and rates of change over time were compared for the DL-AE ROI and global cpRNFL thickness measurements derived from a 2.22-mm to 3.45-mm annulus centered on the optic disc.

Results

The sensitivity for detecting change in progressing eyes was greater for DL-AE ROIs than for global cpRNFL annulus thicknesses (0.90 and 0.63, respectively). The specificity for detecting not likely progression in nonprogressing eyes was similar (0.92 and 0.93, respectively). The mean rates of change in DL-AE ROI were significantly faster than for cpRNFL annulus thickness in progressing eyes (-1.28 µm/y vs. -0.83 µm/y) and nonprogressing eyes (-1.03 µm/y vs. -0.78 µm/y).

Conclusions

Eye-specific ROIs identified using DL-AE analysis of OCT images show promise for improving assessment of glaucomatous progression.

Translational Relevance

The detection and monitoring of structural glaucomatous progression can be improved by considering eye-specific regions of likely progression identified using deep learning.

Optic Nerve Engraftment of Neural Stem Cells

Purpose

To evaluate the integrative potential of neural stem cells (NSCs) with the visual system and characterize effects on the survival and axonal regeneration of axotomized retinal ganglion cells (RGCs).

Methods

For in vitro studies, primary, postnatal rat RGCs were directly cocultured with human NSCs or cultured in NSC-conditioned media before their survival and neurite outgrowth were assessed. For in vivo studies, human NSCs were transplanted into the transected rat optic nerve, and immunohistology of the retina and optic nerve was performed to evaluate RGC survival, RGC axon regeneration, and NSC integration with the injured visual system.

Results

Increased neurite outgrowth was observed in RGCs directly cocultured with NSCs. NSC-conditioned media demonstrated a dose-dependent effect on RGC survival and neurite outgrowth in culture. NSCs grafted into the lesioned optic nerve modestly improved RGC survival following an optic nerve transection (593 ± 164 RGCs/mm2 vs. 199 ± 58 RGCs/mm2; P < 0.01). Additionally, RGC axonal regeneration following an optic nerve transection was modestly enhanced by NSCs transplanted at the lesion site (61.6 ± 8.5 axons vs. 40.3 ± 9.1 axons, P < 0.05). Transplanted NSCs also differentiated into neurons, received synaptic inputs from regenerating RGC axons, and extended axons along the transected optic nerve to incorporate with the visual system.

Conclusions

Human NSCs promote the modest survival and axonal regeneration of axotomized RGCs that is partially mediated by diffusible NSC-derived factors. Additionally, NSCs integrate with the injured optic nerve and have the potential to form neuronal relays to restore retinofugal connections.

Reversal of a glaucomatous optic disc pit

Purpose

To report a patient in whom a glaucomatous optic disc pit (ODP) disappeared spontaneously.

Observations

A 59-year-old Korean woman presented with primary open-angle glaucoma, an ODP, and deep superior cecocentral scotomas. She was treated with topical ocular hypotensive medications and followed. Twenty-eight months later, the superior cecocentral scotomas were not detectable with repeated visual field testing. With repeated optical coherence tomography (OCT), the ODP was narrower and shallower; partially filled with prelaminar tissue, there was an increase in the minimal rim width. At the final examination, the cecocentral scotomas reappeared, although prelaminar tissue continued to fill the ODP.

Conclusions and importance

ODP can disappear spontaneously in glaucomatous eyes under ocular hypotensive treatment. However, this is not always associated with sustained visual field improvement.

Juxtapapillary Deep-Layer Microvasculature Dropout and Retinal Nerve Fiber Layer Thinning in Glaucoma

PURPOSE

We sought to characterize juxtapapillary (JP) and non-JP microvasculature dropout in patients with primary open-angle glaucoma and to compare their rate of retinal nerve fiber layer (RNFL) thinning.

DESIGN

Retrospective cohort study.

METHODS

A total of 141 eyes with primary open-angle glaucoma with ≥4 serial optical coherence tomography (OCT) images after initial OCT angiography for ≥2 years were included. Based on OCT angiography imaging, the 3 groups were matched by age and visual field mean deviation: JP group (parapapillary deep-layer microvasculature dropout in contact with the optic disc boundary, n = 47), non-JP group (dropout not reaching the optic disc boundary, n = 47), and no-dropout group (lacking the dropout, n = 47). The RNFL thinning rate was compared among the 3 groups.

RESULTS

The rate of RNFL thinning tended to be fastest in the JP group followed by the non-JP group and no-dropout group in all areas except the temporal and nasal sectors. Post hoc analysis revealed that the JP group had significantly faster RNFL thinning than did the no-dropout group in the global area and the inferotemporal and inferonasal sectors (P < .05). When subgroup analysis was performed for subjects in which the main sector of dropout was the inferotemporal sector, the JP group had significantly faster RNFL thinning than the other 2 groups in the corresponding inferotemporal sector (P < .001).

CONCLUSION

Eyes with JP microvasculature dropout showed faster RNFL thinning than eyes without dropout. These findings suggest that deep-layer microvasculature dropout, especially in contact with the optic disc boundary, is associated with rapid glaucoma progression.

Development and Validation of a Deep Learning System to Detect Glaucomatous Optic Neuropathy Using Fundus Photographs

Importance

A deep learning system (DLS) that could automatically detect glaucomatous optic neuropathy (GON) with high sensitivity and specificity could expedite screening for GON.

Objective

To establish a DLS for detection of GON using retinal fundus images and glaucoma diagnosis with convoluted neural networks (GD-CNN) that has the ability to be generalized across populations.

Design, Setting, and Participants

In this cross-sectional study, a DLS for the classification of GON was developed for automated classification of GON using retinal fundus images obtained from the Chinese Glaucoma Study Alliance, the Handan Eye Study, and online databases. The researchers selected 241 032 images were selected as the training data set. The images were entered into the databases on June 9, 2009, obtained on July 11, 2018, and analyses were performed on December 15, 2018. The generalization of the DLS was tested in several validation data sets, which allowed assessment of the DLS in a clinical setting without exclusions, testing against variable image quality based on fundus photographs obtained from websites, evaluation in a population-based study that reflects a natural distribution of patients with glaucoma within the cohort and an additive data set that has a diverse ethnic distribution. An online learning system was established to transfer the trained and validated DLS to generalize the results with fundus images from new sources. To better understand the DLS decision-making process, a prediction visualization test was performed that identified regions of the fundus images utilized by the DLS for diagnosis.

Exposures

Use of a deep learning system.

Main Outcomes and Measures

Area under the receiver operating characteristics curve (AUC), sensitivity and specificity for DLS with reference to professional graders.

Results

From a total of 274 413 fundus images initially obtained from CGSA, 269 601 images passed initial image quality review and were graded for GON. A total of 241 032 images (definite GON 29 865 [12.4%], probable GON 11 046 [4.6%], unlikely GON 200 121 [83%]) from 68 013 patients were selected using random sampling to train the GD-CNN model. Validation and evaluation of the GD-CNN model was assessed using the remaining 28 569 images from CGSA. The AUC of the GD-CNN model in primary local validation data sets was 0.996 (95% CI, 0.995-0.998), with sensitivity of 96.2% and specificity of 97.7%. The most common reason for both false-negative and false-positive grading by GD-CNN (51 of 119 [46.3%] and 191 of 588 [32.3%]) and manual grading (50 of 113 [44.2%] and 183 of 538 [34.0%]) was pathologic or high myopia.

Conclusions and Relevance

Application of GD-CNN to fundus images from different settings and varying image quality demonstrated a high sensitivity, specificity, and generalizability for detecting GON. These findings suggest that automated DLS could enhance current screening programs in a cost-effective and time-efficient manner.

Agreement Between 10-2 and 24-2C Visual Field Test Protocols for Detecting Glaucomatous Central Visual Field Defects

PRECIS

Moderate to substantial agreement between 10-2 and 24-2C perimetry for detecting central field defects suggests that adding central test points to the 24-2 protocol may improve efficiency of visual field (VF) testing for glaucoma management.

PURPOSE

The purpose of this study was to assess agreement between Humphrey Visual Field Analyzer 10-2 and 24-2C test protocols for detecting glaucomatous defects in the central 10 degrees of the visual field (CVFDs).

MATERIALS AND METHODS

VFs from 165 eyes of 18 healthy individuals, 12 glaucoma suspects and 62 glaucoma patients who completed 10-2 and 24-2C VF testing protocols within 6 months were included. CVFDs on 10-2 and 24-2C (within the central 22 points) test grids required a cluster of 3 contiguous points with P<5%, 5%, and 1% or <5%, 2%, and 2% within a hemifield on the total deviation (TD) or pattern deviation (PD) plot. Cohen kappa (k) was used to assess agreement between 10-2 and 24-2C test grids in identifying CVFDs. Specificity of each testing strategy was assessed in VFs from healthy eyes.

RESULTS

CVFDs in suspect and glaucoma eyes were combined and reported as localized to superior, inferior or both hemifields based on TD and PD plots for 10-2 and 24-2C test grids. Moderate to substantial agreement was observed between 10-2 and 24-2C grids for detecting any CVFD from PD (k=0.551) and TD (k=0.651) plots. Specificity was high in healthy eyes ranging from 0.94 to 1.0 for both test protocols.

CONCLUSION

Substantial agreement for identifying CVFDs using the 24-2C and 10-2 protocols suggests that combining tests by adding central test points to the 24-2 test grid may supplant the need for 2 perimetry regimens for detecting central and peripheral glaucomatous VF damage.

Superficial and Deep Macula Vessel Density in Healthy, Glaucoma Suspect, and Glaucoma Eyes

PRECIS

Macular superficial capillary plexus (SCP) vessel density is more informative than deep capillary plexus (DCP) vessel density for the detection of glaucoma.

PURPOSE

The purpose of this study was to characterize optical coherence tomography angiography macular SCP and projection-resolved DCP vessel densities and compare their diagnostic accuracies with ganglion cell complex (GCC) thickness in healthy, glaucoma suspect, and glaucoma eyes.

MATERIALS AND METHODS

Sixty-eight eyes of 44 healthy subjects, 26 eyes of 16 preperimetric glaucoma suspects, and 161 eyes of 124 glaucoma patients from the Diagnostics Innovations in Glaucoma Study with good quality high-density 6×6 mm2 macula optical coherence tomography angiography images were included. The diagnostic accuracy of SCP vessel density, projection-resolved DCP vessel density and GCC thickness were compared among groups.

RESULTS

Mean whole image vessel density (wiVD; % of area occupied by vessels containing flowing blood) in the SCP layer was highest in healthy eyes (49.7%), followed by glaucoma suspect eyes (46.0%), and glaucoma eyes (40.9%) (P<0.001). Mean wiVD in the DCP layer was similar in healthy (50.6%), glaucoma suspect (47.3%), and glaucoma eyes (45.7%) (P=0.925). Diagnostic accuracy of both GCC thickness and SCP wiVD was significantly higher than DCP wiVD for classifying healthy and glaucoma eyes [adjusted area under the receiver operating characteristic curve (95% confidence interval): GCC=0.86 (0.72, 0.94), SCP=0.80 (0.66, 0.91) and DCP=0.44 (0.30, 0.57)] (P<0.001).

CONCLUSIONS

SCP vessel densities have better diagnostic accuracy for detecting glaucoma than DCP vessel densities. Although the diagnostic accuracy of the macula SCP is relatively modest, it is more informative than the DCP.

Implanted Microsensor Continuous IOP Telemetry Suggests Gaze and Eyelid Closure Effects on IOP-A Preliminary Study

Purpose

To explore the effect of gaze direction and eyelid closure on intraocular pressure (IOP).

Methods

Eleven patients with primary open-angle glaucoma previously implanted with a telemetric IOP sensor were instructed to view eight equally-spaced fixation targets each at three eccentricities (10°, 20°, and 25°). Nine patients also performed eyelid closure. IOP was recorded via an external antenna placed around the study eye. Differences of mean IOP between consecutive gaze positions were calculated. Furthermore, the effect of eyelid closure on gaze-dependent IOP was assessed.

Results

The maximum IOP increase was observed at 25° superior gaze (mean ± SD: 4.4 ± 4.9 mm Hg) and maximum decrease at 25° inferonasal gaze (-1.6 ± 0.8 mm Hg). There was a significant interaction between gaze direction and eccentricity (P = 0.003). Post-hoc tests confirmed significant decreases inferonasally for all eccentricities (mean ± SEM: 10°: -0.7 ± 0.2, P = 0.007; 20°: -1.1 ± 0.2, P = 0.006; and 25°: -1.6 ± 0.2, P = 0.006). Eight of 11 eyes showed significant IOP differences between superior and inferonasal gaze at 25°. IOP decreased during eyelid closure, which was significantly lower than downgaze at 25° (mean ± SEM: -2.1 ± 0.3 mm Hg vs. -0.7 ± 0.2 mm Hg, P = 0.014).

Conclusions

Our data suggest that IOP varies reproducibly with gaze direction, albeit with patient variability. IOP generally increased in upgaze but decreased in inferonasal gaze and on eyelid closure. Future studies should investigate the patient variability and IOP dynamics.

The influence of axial myopia on optic disc characteristics of glaucoma eyes

This study characterizes differences in glaucomatous eyes with and without high axial myopia using custom automated analysis of OCT images. 452 eyes of 277 glaucoma patients were stratified into non (n = 145 eyes), mild (n = 214 eyes), and high axial myopia (axial length (AL) > 26 mm, n = 93 eyes). Optic disc ovality index, tilt and rotation angle of Bruch´s membrane opening (BMO) and peripapillary choroidal thickness (PCT) were calculated using automated and deep learning strategies. High myopic optic discs were more oval and had larger BMO tilt than mild and non-myopic discs (both p < 0.001). Mean PCT was thinnest in high myopic eyes followed by mild and non-myopic eyes (p < 0.001). BMO rotation angle, global retinal nerve fiber layer (RNFL) thickness and BMO-minimum rim width (MRW) were similar among groups. Temporal RNFL was thicker and supranasal BMO-MRW was thinner in high myopic eyes. BMO tilt and PCT showed moderate and temporal RNFL and nasal BMO-MRW showed weak but significant associations with AL in multivariable analyses (all p < 0.05). Large BMO tilt angle and thin PCT are characteristics of highly myopic discs and were not associated with severity of glaucoma. Caution should be exercised when using sectoral BMO-MRW and RNFL thickness for glaucoma management decisions in myopic eyes.

OCT Angiography Artifacts in Glaucoma

PURPOSE

To determine the prevalence of different types of artifacts seen in OCT angiography (OCTA) images of healthy and glaucoma eyes and evaluate the characteristics associated with poor-quality images.

DESIGN

Retrospective study.

PARTICIPANTS

A total of 649 eyes of 368 healthy, glaucoma suspect, and glaucoma patients.

METHODS

Angiovue (Optovue Inc) high-density (HD) and non-HD optic nerve head and macula OCTA images of participants were evaluated by 4 expert reviewers for the presence of different artifacts, including eye movement, defocus, shadow, decentration, segmentation error, blink, and Z offset in the superficial vascular layer. Each OCTA scan was designated to have good or poor quality based on the presence of artifacts. The association of demographic and ocular characteristics with the likelihood of obtaining poor-quality OCTA images was evaluated.

MAIN OUTCOME MEASURES

The prevalence of OCTA artifacts and the factors associated with increased likelihood of capturing poor-quality OCTA images.

RESULTS

A total of 5263 OCTA images were evaluated. Overall, 33.9% of the OCTA images had poor quality. The majority of images with acceptable quality scores (QS ≥ 4) had no artifacts (76.6%). Other images had 1 (13.6%) or 2 or more artifacts (9.8%). Older age (P < 0.001), male gender (P = 0.045), worse visual field mean deviation (P < 0.001), absence of eye tracking (P < 0.001), and macular scan area (P < 0.001) were associated with a higher likelihood of obtaining poor-quality images. In images with acceptable QS, the commercially available quality measures including QS and signal strength index had the area under the receiver operating characteristic curves of 0.65 (95% confidence interval [CI], 0.62-0.69) and 0.70 (95% CI, 0.68-0.73) to detect good-quality images, respectively.

CONCLUSIONS

OCTA artifacts associated with poor-quality images are frequent, and their prevalence is affected by ocular and patient characteristics. One should not rely solely on the quantitative assessments that are provided automatically by OCTA instruments. A systematic scan review should be conducted to ensure appropriate interpretation of OCTA images. Given the high prevalence of poor-quality OCTA images, the images should be reacquired whenever an apparent and correctable artifact is present on a captured image.

Racial Differences in the Rate of Change in Anterior Lamina Cribrosa Surface Depth in the African Descent and Glaucoma Evaluation Study

Purpose

The purpose of this study was to determine if the rate of change in the depth of the surface of the lamina cribrosa due to glaucomatous remodeling differs between glaucoma patients of African descent (AD) and European descent (ED).

Methods

There were 1122 images taken longitudinally over an average of 3 years (range = 0.9-4.1 years) from 122 patients with glaucoma followed in the African Descent and Glaucoma Evaluation Study (ADAGES) and Diagnostic Intervention and Glaucoma Study (DIGS) were automatically segmented to compute anterior lamina cribrosa surface depth (ALCSD). The rate of ALCSD change was compared across racial groups after adjusting for baseline characteristics known to be associated with ALCSD or disease progression (visual field, ALCSD, corneal thickness, optic disk size, and age).

Results

After adjusting for all other covariates, the ED group had significantly greater ALCSD posterior migration (deepening) than the AD group (difference = 2.57 µm/year, P = 0.035). There was a wider range of ALCSD change in the ED compared with the AD group, and more individuals had greater magnitude of both deepening and shallowing. No other covariates measured at baseline had independent effects on the longitudinal changes in ALCSD (baseline visual field severity, baseline ALCSD, corneal thickness, Bruch's membrane opening [BMO] area, or age).

Conclusions

Glaucomatous remodeling of the lamina cribrosa differs between AD and ED patients with glaucoma. Unlike the cross-sectional associations seen with aging, in which a deeper ALCSD was seen with age in the ED group, glaucomatous remodeling in this longitudinal study resulted in more posterior migration of ALCSD in ED compared to AD patients.

The effect of daily life activities on intraocular pressure related variations in open-angle glaucoma

The recent advent of continuous intraocular pressure (IOP) telemetry has led to an increased awareness of the importance of IOP fluctuations, and theories have emerged that IOP variations could play as much a role in glaucoma progression as the mean level of IOP. The aim of the present study was to evaluate the direct effect of common daily activities on IOP-related profiles. Primary open-angle glaucoma and glaucoma suspect patients were prospectively enrolled from specialist clinics at the University of California San Diego (UCSD), USA. Patients were fitted with a SENSIMED Triggerfish (TF) contact lens sensor (CLS) and were instructed to return to their usual daily activities for 24 h. They were asked to record each specific activity or event in a diary. The protocol was repeated twice. The following events were recorded: "walking/cycling", "resistance training", "yoga/meditation", and "emotional stress". CLS measurements recorded 60-to-30 min prior to each event were used as a baseline reference, and all IOP-related fluctuations for 120 min after the start of each event were reported in relation to this reference. Forty relevant events from 22 CLS recordings in 14 patients were retrieved from the diaries. Walking/cycling (n = 10) caused a small but statistically significant elevation of the IOP-related profile during the activity (p = 0.018). Resistance training (n = 11) caused a persistent elevation of the IOP-related profile from the onset of the activity (p = 0.005) through 120 min after the activity was stopped (p = 0.007). Yoga/meditation (n = 4) caused a sustained drop in the IOP-related profiles through to 120 min, although this was not statistically significant (p > 0.380). Emotional stress (n = 13) was associated with a gradual elevation of the IOP-related profile from the start of the stressful stimulus. Both early and late variations were statistically significant (p = 0.038 and p = 0.021, respectively). The present study suggests that emotional stress and resistance training may be associated with persistent IOP-related profile elevation.

A hierarchical deep learning approach with transparency and interpretability based on small samples for glaucoma diagnosis

The application of deep learning algorithms for medical diagnosis in the real world faces challenges with transparency and interpretability. The labeling of large-scale samples leads to costly investment in developing deep learning algorithms. The application of human prior knowledge is an effective way to solve these problems. Previously, we developed a deep learning system for glaucoma diagnosis based on a large number of samples that had high sensitivity and specificity. However, it is a black box and the specific analytic methods cannot be elucidated. Here, we establish a hierarchical deep learning system based on a small number of samples that comprehensively simulates the diagnostic thinking of human experts. This system can extract the anatomical characteristics of the fundus images, including the optic disc, optic cup, and appearance of the retinal nerve fiber layer to realize automatic diagnosis of glaucoma. In addition, this system is transparent and interpretable, and the intermediate process of prediction can be visualized. Applying this system to three validation datasets of fundus images, we demonstrate performance comparable to that of human experts in diagnosing glaucoma. Moreover, it markedly improves the diagnostic accuracy of ophthalmologists. This system may expedite the screening and diagnosis of glaucoma, resulting in improved clinical outcomes.

Nocturnal Variability of Intraocular Pressure Monitored With Contact Lens Sensor Is Associated With Visual Field Loss in Glaucoma

PURPOSE

The aim was to determine whether 24-hour recording of intraocular pressure (IOP)-related ocular dimensional changes with a contact lens sensor (CLS, Triggerfish) is associated with the rate of visual field (VF) progression in primary open-angle glaucoma (POAG) patients.

DESIGN

This was a retrospective, observational cohort study.

PARTICIPANTS

Patients with POAG were included from the Glaucoma Clinic and Diagnostic Innovations in Glaucoma Study at the Hamilton Glaucoma Center at University of California, San Diego.

METHODS

A session of 24-hour CLS recording was acquired for 1 eye from each patient. The mean follow-up time was 9.9±4.0 years. The association between CLS variables and rate of change of mean deviation was determined by univariate and multivariate mixed linear regression models.

RESULTS

Thirty-two patients, aged 69.8±13.6 years were included, 50% were female. An average of 11.6±5.6 standard automated perimetry examinations was available with a mean rate of mean deviation progression of -0.2±0.4 dB/year. Mean IOP was 17.8±4.2 mm Hg. The mean number of IOP-lowering medications were 1.2±1.0. Each 10-unit larger nocturnal variability of IOP-related ocular dimensional changes measured by CLS recording was significantly associated with -0.25±0.11 dB faster VF loss in POAG patients (P=0.035).

CONCLUSIONS

Twenty-four-hour CLS recording of IOP-related ocular dimensional change was associated with faster VF progression. Such CLS recordings are useful to assess the risk of in progression in POAG patients.

Geometric Perfusion Deficits: A Novel OCT Angiography Biomarker for Diabetic Retinopathy Based on Oxygen Diffusion

PURPOSE

To develop geometric perfusion deficits (GPD), an optical coherence tomography angiography (OCTA) biomarker based on oxygen diffusion, and to evaluate its utility in a pilot study of healthy subjects and patients with diabetic retinopathy (DR).

DESIGN

Retrospective cross-sectional study.

METHODS

Commercial spectral-domain optical coherence tomography angiography (OCTA) instruments were used to acquire repeated 3 × 3-mm² and 6 × 6-mm² motion-corrected macular OCTA volumes. En face OCTA images corresponding to the superficial capillary plexus (SCP), deep capillary plexus (DCP), and full retinal projections were obtained using automatic segmentation. For each projection, the GPD percentage and the vessel density percentage, the control metric, were computed, and their values were compared between the normal and DR eyes. The repeated OCTA acquisitions were used to assess the test-retest repeatability of the GPD and vessel density percentages.

RESULTS

Repeated OCTA scans of 15 normal eyes and 12 DR eyes were obtained. For all en face projections, GPD percentages were significantly higher in DR eyes than in normal eyes; vessel density percentages were significantly lower in all but 1 projection (DCP). Large GPD areas were used to identify focal perfusion deficits. Test-retest analysis showed that the GPD percentage had superior repeatability than the vessel density percentage in most cases. A strong negative correlation between the GPD percentage and the vessel density percentage was also found.

CONCLUSIONS

Geometric perfusion deficits, an OCTA biomarker based on oxygen diffusion, provides a quantitative metric of macular microvascular remodeling with a strong physiological underpinning. The GPD percentage may serve as a useful biomarker for detecting and monitoring DR.

Use of Virtual Reality Simulation to Identify Vision-Related Disability in Patients With Glaucoma

Importance

Clinical assessment of vision-related disability is hampered by the lack of instruments to assess visual performance in real-world situations. Interactive virtual reality (VR) environments displayed in a binocular stereoscopic VR headset have been designed, presumably simulating day-to-day activities to evaluate vision-related disability.

Objective

To investigate the application of VR to identify vision-related disability in patients with glaucoma.

Design, Setting, and Participants

In a cross-sectional study, 98 patients with glaucoma and 50 healthy individuals were consecutively recruited from a university eye clinic; all participants were Chinese. The study was conducted between August 30, 2016, and July 31, 2017; data analysis was performed from December 1, 2017, to October 30, 2018.

Exposures

Measurements of visual acuity, contrast sensitivity, visual field (VF), National Eye Institute 25-item Visual Function Questionnaire Rasch score, and VR disability scores determined from 5 VR simulations: supermarket shopping, stair and city navigations in daytime, and stair and city navigations in nighttime. Duration required to complete the simulation, number of items incorrectly identified, and number of collisions were measured to compute task-specific and overall VR disability scores. Vision-related disability was identified when the VR disability score was outside the normal age-adjusted 95% confidence region.

Main Outcomes and Measures

Virtual reality disability score.

Results

In the 98 patients with glaucoma, mean (SD) age was 49.8 (11.6) years and 60 were men (61.2%); in the 50 healthy individuals, mean (SD) age was 48.3 (14.8) years and 16 were men (32.0%). The patients with glaucoma had different degrees of VF loss (122 eyes [62.2%] had moderate or advanced VF defects). The time required to complete the activities by patients with glaucoma vs healthy individuals was longer by 15.2 seconds (95% CI, 5.5-24.9 seconds) or 34.1% (95% CI, 12.4%-55.7%) for the shopping simulation, 72.8 seconds (95% CI, 23.0-122.6 seconds) or 33.8% (95% CI, 10.7%-56.9%) for the nighttime stair navigation, and 38.1 seconds (95% CI, 10.9-65.2 seconds) or 30.8% (95% CI, 8.8%-52.8%) for the nighttime city navigation. The mean (SD) duration was not significantly different between the glaucoma and healthy groups in daytime stair (203.7 [93.7] vs 192.9 [89.1] seconds, P = .52) and city (118.7 [41.5] vs 117.0 [52.3] seconds, P = .85) navigation. For each decibel decrease in binocular VF sensitivity, the risk of collision increased by 15% in nighttime stair (hazard ratio [HR], 1.15; 95% CI, 1.08-1.22) and city (HR, 1.15; 95% CI, 1.08-1.23) navigations. Fifty-eight patients (59.1%) with glaucoma had vision-related disability in at least 1 simulated daily task; a higher proportion of patients had vision-related disability in nighttime city (27 of 88 [30.7%]) and stair (27 of 90 [30.0%]) navigation than in daytime city (7 of 88 [8.0%]) and stair (19 of 96 [19.8%]) navigation. The overall VR disability score was associated with the National Eye Institute 25-item Visual Function Questionnaire Rasch score (R2 = 0.207).

Conclusions and Relevance

These findings suggest that vision-related disability is associated with lighting condition and task in patients with glaucoma. Virtual reality may allow eye care professionals to understand the patients' perspectives on how visual impairment imparts disability in daily living and provide a new paradigm to augment the assessment of vision-related disability.

Is Diabetes Mellitus a Blessing in Disguise for Primary Open-angle Glaucoma?

Although numbers of studies have addressed this question, the relationship between diabetes mellitus and primary open-angle glaucoma is still unclear. This article discusses progress in understanding the complex relationship between these 2 entities and recent shifts in perspective that challenge the traditional dogma regarding diabetes mellitus and primary open-angle glaucoma. There are still many unanswered questions.

The Effect of Inhaled Cannabis on Intraocular Pressure in Healthy Adult Subjects

Collapse

Intraocular Pressure Telemetry for Managing Glaucoma during the COVID-19 Pandemic

PURPOSE

To evaluate in glaucoma patients the feasibility and use of remote monitoring of intraocular pressure (IOP) with an implanted telemetry sensor during the coronavirus disease 2019 (COVID-19) lockdown.

DESIGN

Cross-sectional study.

PARTICIPANTS

Patients previously implanted with a telemetric IOP sensor (Eyemate; Implandata GmbH) were included.

METHODS

Intraocular pressure measurements acquired by the patients during the lockdown were collected by physicians who were located remotely. A questionnaire was sent to 10 participating study centers to evaluate the clinical impact of remote monitoring of IOP via the IOP sensor system.

MAIN OUTCOME MEASURES

Number of patients who obtained home IOP measurements.

RESULTS

Data were available from all centers and from 37 eyes of 37 patients (16 patients with a sulcus-based sensor and 21 patients with a suprachoroidal sensor). Thirty-four patients obtained IOP measurements during the lockdown. Mean age of the patients was 69.3 ± 9.6 years, and 48.6% were women. A total of 8415 IOP measurements from 370 measurement days were obtained. Based on remote IOP measurements, treatment was changed in 5 patients. In another 5 patients, treatment change was considered when physicians received the IOP measurements after the lockdown. Nine of the 10 study centers judged remote IOP measurements to have a clinical impact.

CONCLUSIONS

These results show the feasibility of patient-acquired measurement of IOP in conjunction with remote IOP monitoring by physicians with an implantable sensor. The data obtained impacted clinical decision making, including adjustment of ocular hypotensive therapy and avoiding unnecessary office visits during the COVID-19 pandemic.

Evaluating the neuroprotective impact of senolytic drugs on human vision

Glaucoma, a chronic neurodegenerative disease of retinal ganglion cells (RGCs), is a leading cause of irreversible blindness worldwide. Its management currently focuses on lowering intraocular pressure to slow disease progression. However, disease-modifying, neuroprotective treatments for glaucoma remain a major unmet need. Recently, senescent cells have been observed in glaucomatous eyes, exposing a potential pathway for alternative glaucoma therapies. Prior studies demonstrated that targeting senescent RGCs for removal (i.e., a senolytic approach) protected healthy RGCs and preserved visual function in a mouse ocular hypertension model. However, the effects of senolytic drugs on vision in human patients are unknown. Here, we used existing clinical data to conduct a retrospective cohort study in 28 human glaucoma patients who had been exposed to senolytics. Senolytic exposure was not associated with decreased visual acuity, elevated intraocular pressure, or documentation of senolytic-related adverse ocular effects by treating ophthalmologists. Additionally, patients exposed to senolytics (n = 9) did not exhibit faster progression of glaucomatous visual field damage compared to matched glaucoma patients (n = 26) without senolytic exposure. These results suggest that senolytic drugs do not carry significant ocular toxicity and provide further support for additional evaluation of the potential neuroprotective effects of senolytics on glaucoma and other neurodegenerative diseases.

A Randomized Controlled Trial Comparing Subconjunctival Injection to Direct Scleral Application of Mitomycin C in Trabeculectomy

PURPOSE

To compare the efficacy of intraoperative scleral application with subconjunctival injection of mitomycin C (MMC) in trabeculectomy.

DESIGN

Prospective, randomized, interventional study.

METHODS

This study took place in a single clinical practice in an academic setting. Patients had medically uncontrolled glaucoma as indicated by high intraocular pressure (IOP), worsening visual field, or optic nerve head changes in whom primary trabeculectomy was indicated. Patients were older than 18 years with medically uncontrolled glaucoma and no history of incisional glaucoma surgery. Patients were randomized to MMC delivered by preoperative subconjunctival injection or by intraoperative direct scleral application using surgical sponges during trabeculectomy. Comprehensive eye examinations were conducted at 1 day, 1 week, 6 weeks, 3 months, and 6 months postoperatively. Subconjunctival 5-fluorouracil injections were given postoperatively, as needed. The primary outcome was the proportion of patients who demonstrated IOP of <21 mm Hg and ≥30% reduction in IOP from baseline. Secondary outcome measures included the number of IOP-lowering medications, bleb morphology using the Indiana Bleb Appearance Grading Scale, and complication rates.

RESULTS

Participants (n = 100) were randomized into groups matched for baseline demographics, glaucoma status, and baseline IOP. At 6 months, there were no significant differences between the injection (n = 38) and sponge (n = 40) groups in surgical success (P = .357), mean IOP (P = .707), number of glaucoma medications (P = 1.000), bleb height (P = .625), bleb extension (P = .216), bleb vascularity (P = .672), or complications rates.

CONCLUSION

Both techniques of MMC delivery (subconjunctival injection and direct scleral application) resulted in comparable surgical outcomes and bleb morphologies.

OCT angiography measured changes in the foveal avascular zone area after glaucoma surgery

Background/Aims

To evaluate quantitative changes in the foveal avascular zone (FAZ) area after glaucoma surgery using swept-source optical coherence tomography angiography (SS-OCTA).

Methods

Fifty-four consecutive patients with primary open-angle glaucoma (POAG) who met the inclusion criteria and underwent unilateral glaucoma surgery to reduce intraocular pressure (IOP) between April 2018 and July 2019.

Eyes underwent IOP-lowering glaucoma surgery and their fellow (non-surgical) eyes were included. OCTA of the macula was performed in both eyes before glaucoma surgery and 3 months postoperatively. Two blinded examiners reviewed the image quality. Within- and between-group comparisons of the FAZ area and correlation of the FAZ area with age, IOP, central sensitivity and clinical variables.

Results

The mean (±SD) age was 66.7±11.3 years. After surgery, the IOP and FAZ area significantly decreased from 22.1±9.5 mmHg to 10.3±3.5 mmHg and from 0.485±0.193 mm² to 0.446±0.174 mm², respectively (both p<0.001). Conversely, in the non-surgery group, the preoperative and postoperative mean FAZ areas (0.398±0.119 mm² and 0.396±0.110 mm², respectively) did not significantly differ (p=0.469). Change in the FAZ area significantly correlated with the preoperative FAZ area, preoperative foveal sensitivity and change in IOP (all p<0.05).

Conclusions

The FAZ area is decreased with IOP-lowering surgery in patients with POAG, and change in the FAZ area was significantly correlated with both preoperative foveal sensitivity and change in IOP.

Capillary Density Measured by Optical Coherence Tomography Angiography in Glaucomatous Optic Disc Phenotypes

PURPOSE

To compare optical coherence tomography angiography (OCTA)-measured capillary density of the optic disc among 4 glaucomatous optic disc phenotypes.

DESIGN

Cross-sectional study.

METHODS

Circumpapillary capillary density (cpCD) of 4 glaucomatous optic disc phenotypes in 193 eyes of 141 glaucoma patients and cpCD in 92 eyes of 55 healthy subjects from the Diagnostic Innovations in Glaucoma Study (DIGS) were compared. Areas under the receiver operating characteristic (AUROC) curves were used to evaluate diagnostic accuracy among groups after adjusting for confounders.

RESULTS

Four glaucoma phenotypes were assessed: focal ischemic (n = 45), generalized cup enlargement (n = 60), myopic glaucoma (n = 38), and senile sclerotic (n = 50). Sex, mean ocular perfusion pressure, intraocular pressure, mean deviation, and the quality score did not differ among phenotypes. However, there were differences in age (P = .050), race (P = .039), axial length (P = .033), and retinal nerve fiber layer thickness (P < .001) among the groups. After adjusting for confounders, senile sclerotic discs had the lowest cpCD (37.1% [95% confidence interval, 35.3-38.8]), followed by focal ischemic (41.8% [40.0-43.6]), myopic glaucoma (42.1% [40-44.2]), and generalized cup enlargement (45.5% [44-47]) (P < .001) discs. The adjusted AUROC curves of cpCD for discriminating between healthy and glaucomatous eyes were highest in senile sclerotic eyes (0.928) and lowest in generalized cup enlargement eyes (0.704).

CONCLUSIONS

OCTA-measured vessel density differs among optic disc phenotypes. Clinicians should be aware that the performance of OCTA for glaucoma diagnosis may be influenced by the optic disc phenotype.

Sheath-Preserving Optic Nerve Transection in Rats to Assess Axon Regeneration and Interventions Targeting the Retinal Ganglion Cell Axon

Retinal ganglion cell (RGC) axons converge at the optic nerve head to convey visual information from the retina to the brain. Pathologies such as glaucoma, trauma, and ischemic optic neuropathies injure RGC axons, disrupt transmission of visual stimuli, and cause vision loss. Animal models simulating RGC axon injury include optic nerve crush and transection paradigms. Each of these models has inherent advantages and disadvantages. An optic nerve crush is generally less severe than a transection and can be used to assay axon regeneration across the lesion site. However, differences in crush force and duration can affect tissue responses, resulting in variable reproducibility and lesion completeness. With optic nerve transection, there is a severe and reproducible injury that completely lesions all axons. However, transecting the optic nerve dramatically alters the blood brain barrier by violating the optic nerve sheath, exposing the optic nerve to the peripheral environment. Moreover, regeneration beyond a transection site cannot be assessed without reapposing the cut nerve ends. Furthermore, distinct degenerative changes and cellular pathways are activated by either a crush or transection injury. The method described here incorporates the advantages of both optic nerve crush and transection models while mitigating the disadvantages. Hydrostatic pressure delivered into the optic nerve by microinjection completely transects the optic nerve while maintaining the integrity of the optic nerve sheath. The transected optic nerve ends are reapposed to allow for axon regeneration assays. A potential limitation of this method is the inability to visualize the complete transection, a potential source of variability. However, visual confirmation that the visible portion of the optic nerve has been transected is indicative of a complete optic nerve transection with 90-95% success. This method could be applied to assess axon regeneration promoting strategies in a transection model or investigate interventions that target the axonal compartments.

Gradient-Boosting Classifiers Combining Vessel Density and Tissue Thickness Measurements for Classifying Early to Moderate Glaucoma

PURPOSE

To compare gradient-boosting classifier (GBC) analysis of optical coherence tomography angiography (OCTA)-measured vessel density (VD) and OCT-measured tissue thickness to standard OCTA VD and OCT thickness parameters for classifying healthy eyes and eyes with early to moderate glaucoma.

DESIGN

Comparison of diagnostic tools.

METHODS

A total of 180 healthy eyes and 193 glaucomatous eyes with OCTA and OCT imaging of the macula and optic nerve head (ONH) were studied. Four GBCs were evaluated that combined 1) all macula VD and thickness measurements (Macula GBC), 2) all ONH VD and thickness measurements (ONH GBC), 3) all VD measurements from the macula and ONH (vessel density GBC), and 4) all thickness measurements from the macula and ONH (thickness GBC). ROC curve (AUROC) analyses compared the diagnostic accuracy of GBCs to that of standard instrument-provided parameters. A fifth GBC that combined all parameters (full GBC) also was investigated.

RESULTS

GBCs had better diagnostic accuracy than standard OCTA and OCT parameters with AUROCs ranging from 0.90 to 0.93 and 0.64 to 0.91, respectively. The full GBC (AUROC = 0.93) performed significantly better than the ONH GBC (AUROC = 0.91; P = .036) and the vessel density GBC (AUROC = 0.90; P = .010). All other GBCs performed similarly. The mean relative influence of each parameter included in the full GBC identified a combination of macular thickness and ONH VD measurements as the greatest contributors.

CONCLUSIONS

GBCs that combine OCTA and OCT macula and ONH measurements can improve diagnostic accuracy for glaucoma detection compared to most but not all instrument provided parameters.

AIBP protects retinal ganglion cells against neuroinflammation and mitochondrial dysfunction in glaucomatous neurodegeneration

Glaucoma is a leading cause of blindness worldwide in individuals 60 years of age and older. Despite its high prevalence, the factors contributing to glaucoma progression are currently not well characterized. Glia-driven neuroinflammation and mitochondrial dysfunction play critical roles in glaucomatous neurodegeneration. Here, we demonstrated that elevated intraocular pressure (IOP) significantly decreased apolipoprotein A-I binding protein (AIBP; gene name Apoa1bp) in retinal ganglion cells (RGCs), but resulted in upregulation of TLR4 and IL-1β expression in Müller glia endfeet. Apoa1bp^-/- mice had impaired visual function and Müller glia characterized by upregulated TLR4 activity, impaired mitochondrial network and function, increased oxidative stress and induced inflammatory responses. We also found that AIBP deficiency compromised mitochondrial network and function in RGCs and exacerbated RGC vulnerability to elevated IOP. Administration of recombinant AIBP prevented RGC death and inhibited inflammatory responses and cytokine production in Müller glia in vivo. These findings indicate that AIBP protects RGCs against glia-driven neuroinflammation and mitochondrial dysfunction in glaucomatous neurodegeneration and suggest that recombinant AIBP may be a potential therapeutic agent for glaucoma.

Dynamic Scheimpflug Ocular Biomechanical Parameters in Untreated Primary Open Angle Glaucoma Eyes

Purpose

To characterize the corneal biomechanical properties of glaucoma eyes by comparing the dynamic Scheimpflug biomechanical parameters between untreated glaucoma and control eyes.

Methods

Cross-sectional observational data of dynamic Scheimpflug analyzer (Corvis ST) examinations were retrospectively collected from 35 eyes of 35 consecutive patients with untreated normal tension glaucoma and 35 eyes of 35 healthy patients matched on age and IOP. Ten biomechanical parameters were compared between the two groups using multivariable models adjusting for IOP, central corneal thickness, age, and axial length. The Benjamini-Hochberg method was used to correct for multiple comparison.

Results

In multivariable models, glaucoma was associated with smaller applanation 1 time (P < 0.001, coefficient = −0.5865), applanation 2 time (P = 0.012, coefficient = −0.1702), radius (P = 0.006, coefficient = −0.5447), larger peak distance (P = 0.011, coefficient = 0.1023), deformation amplitude ratio at 1 mm (P < 0.001, coefficient = 0.072), and integrated radius (P < 0.001, coefficient = 1.094). These associations consistently indicate greater compliance of the cornea in glaucoma eyes.

Conclusions

Untreated normal tension glaucoma eyes were more compliant than healthy eyes. The greater compliance (smaller stiffness) of normal tension glaucoma eyes may increase the risk of optic nerve damage. These results suggest the relevance of measuring biomechanical properties of glaucoma eyes.

Detection of Progression With 10-2 Standard Automated Perimetry: Development and Validation of an Event-Based Algorithm

PURPOSE

To describe the development of a new algorithm for detecting progressive changes in 10-2 visual field (VF) tests using event-based analysis and to test its validity in a second, independent glaucoma cohort.

DESIGN

Prospective cohort study.

METHODS

Patients with established open-angle glaucoma from the Macular Assessment and Progression Study (MAPS; development cohort, n = 151), and the African Descent and Glaucoma Evaluation Study (ADAGES; validation cohort, n = 52) were evaluated. The 10-2 VF results from MAPS were obtained during 4 test-retest sessions within a 4-month period. For the validation analysis, 10-2 VF results from ADAGES performed on at least 5 visits were used. The event-based pointwise changes on 10-2 tests in the validation cohort were determined using 2 progression criteria: at least 3 progressing VF locations on 2 or 3 consecutive tests ("possible" or "likely" progression). Linear mixed-effects models were used to evaluate VF progression.

RESULTS

In the validation cohort, the mean (SD) follow-up time was 2.3 (0.7) years. The number of eyes experiencing 10-2 VF progression based on "possible" and "likely" progression was 36 (54.5%) and 11 (16.6%), respectively. Eyes experiencing "possible" progression had MD changes (-0.60 dB/year [95% confidence interval (CI): -0.93 to -0.28]) faster than those not meeting this criterion (P < .001), whereas for those with "likely" progression the difference was -0.91 dB/year (95% CI: -1.26 to -0.56, P < .001).

CONCLUSIONS

A new event-based progression algorithm using the 10-2 VF can identify eyes experiencing more rapid MD progression and may be used as a tool to assess progressive macular functional changes in glaucoma.

Comparison of Peripapillary Capillary Density in Glaucoma Patients of African and European Descent

PURPOSE

To evaluate racial differences in optic nerve head peripapillary capillary density measured by OCT angiography (OCTA) in patients with open-angle glaucoma.

DESIGN

Observational, cross-sectional study.

PARTICIPANTS

Two hundred eighty-four eyes of 195 glaucoma patients and 108 eyes of 58 healthy participants from the Diagnostic Innovations in Glaucoma Study.

METHODS

Global and sectoral circumpapillary capillary density (cpCD) loss in participants of European descent (ED) and African descent (AD) were compared. Areas under the receiver operating characteristic curve (AUROCs) were used to evaluate diagnostic accuracy of cpCD and global circumpapillary retinal nerve fiber layer (cpRNFL) thickness in the 2 groups after adjusting for confounders.

MAIN OUTCOME MEASURES

Peripapillary capillary density and cpRNFL thickness measurements and their estimated loss.

RESULTS

Participants of AD and ED with glaucoma were of similar age and glaucoma severity. After adjusting for age, disc area, and other confounders, significantly lower cpCD was found in ED eyes compared with AD eyes in mild glaucoma (mean, 42.2% [95% confidence interval (CI), 41.2%-43.2%] and 46.5% [95% CI, 44.8%-48.1%], respectively; adjusted difference, 4.4 [95% CI, 2.6-6.2]; P < 0.001) and moderate to advanced glaucoma (mean, 34.7% and 38.5%, respectively; adjusted difference, 4.8 [95% CI, 1.6-8.1]; P = 0.005). Although capillary density loss was greater in all sectors in ED compared with AD participants, a similar sectoral pattern of density loss was observed in both racial groups. Lower mean deviation and older age were associated with lower cpCD in both races in multivariate models. The adjusted AUROC for discriminating between healthy and glaucomatous eyes for cpCD was higher for ED (0.95) compared with AD (0.68) patients (P < 0.001). Sensitivity at 95% specificity in AD participants was lower than in ED participants for cpCD (0.32 [95% CI, 0.11-0.64] vs. 0.83 [95% CI, 0.69-0.93], respectively; P < 0.001).

CONCLUSIONS

Although peripapillary capillary density parameters showed good diagnostic accuracy for detecting glaucoma in ED patients, their diagnostic accuracy was only modest in AD patients. Diagnostic performance of cpCD is race dependent, and clinicians should be aware that it has poorer performance in AD patients.

Short-Term and Long-Term Variability of Intraocular Pressure Measured with an Intraocular Telemetry Sensor in Patients with Glaucoma

PURPOSE

To evaluate the short-term and long-term variability of intraocular pressure (IOP) in eyes with primary open-angle glaucoma.

DESIGN

Prospective study.

PARTICIPANTS

Twenty-two patients previously implanted with a sulcus-based IOP sensor (EyeMate, Implandata GmbH, Germany).

METHODS

Twenty-two patients previously implanted with the EyeMate were requested to obtain at least 4 IOP measurements daily. Data were grouped according to the eye and the medication so that an eye treated with a particular medication was considered as one group, and the same eye treated with a different medication during the observation period was considered as a different group. A day was divided into 7 periods: night, midnight to 5:59 am; early, 6 am to 7:59 am; morning, 8 am to 10:59 am; noon, 11 am to 1:59 pm; afternoon, 2 pm to 5:59 pm; evening, 6 pm to 8:59 pm; and late, 9 pm to 11:59 pm. Short-term variability during a particular period was defined as the variability in IOP measurements obtained during that period on different days within 3 months of each other. Long-term variability was defined as the variability in IOP measurements obtained during a particular period on different days over a period of 1 year or more. Variability was assessed using intraclass correlation coefficients (ICCs).

RESULTS

The mean age of study participants was 67.8 ± 6.8 years and 36.4% were women. The mean follow-up duration of patients was 19.2 ± 21.3 months (median, 9 months; range, 1-58 months). Overall, 92 860 IOP measurements over 15 811 measurement days were obtained and analyzed during the study period. The number of measurements obtained from each eye ranged from 1 daily to 277 daily. Intraclass correlation coefficients for short-term variability among the 7 periods during the day ranged from 0.52 (morning) to 0.66 (early). Long-term ICCs ranged from 0.29 (night) to 0.51 (late).

CONCLUSIONS

Continual IOP monitoring showed that IOP has moderate short-term and high long-term variability in glaucoma patients. These findings demonstrate that single IOP measurements do not characterize day-to-day variations in IOP. Moreover, they show the importance of continual IOP monitoring in glaucoma patients.

Episcleral Venous Pressure and the Ocular Hypotensive Effects of Topical and Intracameral Prostaglandin Analogs

There is a limit beyond which increasing either the concentration of a prostaglandin analog (PGA) or its dosing frequency fails to produce increases in ocular hypotensive efficacy with topical dosing. Intracameral PGA dosing with a bimatoprost implant, however, does not exhibit the same intraocular pressure (IOP)-lowering plateau at studied concentrations, and the maximum-achievable ocular hypotensive effects are not yet known. This suggests that the bimatoprost intracameral implant may activate another mechanism of action in addition to the mechanism(s) activated by topical application. Episcleral venous pressure (EVP) is a key determinant of IOP, and experimental manipulation of the episcleral vasculature can change both EVP and IOP. The recent observation that topical and intracameral PGA drug delivery routes produce different patterns of conjunctival hyperemia suggested that the differences in the IOP-lowering profiles may be caused by differing effects on the episcleral vasculature. Recent experiments in animals have shown that topical PGAs increase EVP, while the bimatoprost intracameral implant causes a smaller, transient increase in EVP, followed by a sustained decrease. The increase in EVP could be limiting the IOP-lowering efficacy of topical PGAs. In contrast, the decrease in EVP associated with the bimatoprost implant could explain its enhanced IOP-lowering effects. Further research on EVP as a target for IOP lowering is indicated to improve our understanding of this potentially important pathway for treating patients with glaucoma.

Correlation Between Office-Hour and Peak Nocturnal Intraocular Pressure in Patients Treated with Prostaglandin Analogs

PURPOSE

To test the hypothesis that the correlation between office-hour intraocular pressure (IOP) and peak nocturnal IOP is weakened after using a prostaglandin analog.

DESIGN

Before-and-after study.

METHODS

Twenty-four-hour IOP data obtained in a sleep laboratory of 51 patients (22 patients with open-angle glaucoma and 29 patients with ocular hypertension) were reviewed. Patients had no IOP-lowering medication upon study entry and were then treated with prostaglandin monotherapy for 4 weeks. Measurements of IOP were taken every 2 hours in the sitting and supine positions during the diurnal/wake period (7:30 AM-9:30 PM) and in the supine position during the nocturnal/sleep period (11:30 PM-5:30 AM). Individual and average IOP readings during office hours (9:30 AM-3:30 PM) and peak IOP during the nocturnal/sleep hours were analyzed using the Pearson correlation coefficient and linear regression.

RESULTS

There were statistically significant correlations for all the paired variables for the analyses. Average office-hour IOP had a higher correlation with peak nocturnal IOP than individual office-hour IOP. After the treatment with prostaglandin analog, the correlation between average office-hour IOP and nocturnal peak IOP in the sitting position (r = 0.373) and the supine position (r = 0.386) were reduced from the sitting baseline (r = 0.517) and the supine baseline (r = 0.573) in right eyes. Similar change patterns appeared in left eyes.

CONCLUSION

There is a correlation between office-hour IOP reading and peak nocturnal IOP under no IOP-lowering treatment as well as under prostaglandin monotherapy. The strength of correlation was weaker under the treatment compared with baseline.

Characteristics of Focal Gamma Zone Parapapillary Atrophy

Purpose

The purpose of this study was to investigate the characteristics of focal γ-zone parapapillary atrophy (focal γPPA) in patients with primary open-angle glaucoma (POAG) using spectral-domain optical coherence tomography (SD-OCT).

Methods

Three groups of POAG eyes (n = 214) were defined according to the circumferential extent of Bruch's membrane (BM) within the β-zone PPA, as follows: (1) no γPPA (intact BM; n = 81), (2) conventional γPPA (γPPA involving the fovea-BM-opening axis; n = 89), and (3) focal γPPA (γPPA not involving the fovea-BM-opening axis; n = 44). Clinical and ocular characteristics, including age, axial length (AXL), and focal lamina cribrosa (LC) defects were compared among the three groups.

Results

The focal γPPA group was significantly older (60.6 ± 11.0 years) and had shorter AXL (24.10 ± 1.34 mm) than those of the conventional γPPA group (46.2 ± 13.8 years and 26.53 ± 1.61 mm, respectively; P < 0.001). These values of the focal γPPA group were similar to those of the no γPPA group (23.73 ± 0.97 mm for AXL and 64.0 ± 13.0 years for age). The focal γPPA group had a significantly higher prevalence of focal LC defects than did the other two groups (70.5% [31/44] for the focal γPPA group versus 46.1% [41/89] for the conventional γPPA group versus 37.0% [30/81] for the no γPPA group; P = 0.002).

Conclusions

Focal γPPA was differentiated from conventional γPPA by older age and shorter AXL. Further, focal γPPA was frequently accompanied by focal LC defects. Longitudinal studies elucidating whether focal LC defects and focal γPPA share common pathogenesis are warranted.

Phase 3, Randomized, 20-Month Study of Bimatoprost Implant in Open-Angle Glaucoma and Ocular Hypertension (ARTEMIS 1)

PURPOSE

To evaluate the intraocular pressure (IOP)-lowering efficacy and safety of 10- and 15-μg bimatoprost implant in subjects with open-angle glaucoma (OAG) and ocular hypertension (OHT) after initial and repeated administrations.

DESIGN

Randomized, 20-month, multicenter, subject- and efficacy evaluator-masked, parallel-group, phase 3 clinical study.

PARTICIPANTS

Adults with OAG or OHT in each eye, open iridocorneal angle inferiorly in the study eye, and study eye baseline IOP (hour 0; 8 am) of 22-32 mmHg after washout.

METHODS

Study eyes received bimatoprost implant 10 μg (n = 198) or 15 μg (n = 198) on day 1 with readministration at weeks 16 and 32, or twice-daily topical timolol maleate 0.5% (n = 198). Intraocular pressure was measured at hours 0 and 2 at each visit.

MAIN OUTCOME MEASURES

Primary end points were IOP and change from baseline IOP through week 12. Safety measures included treatment-emergent adverse events (TEAEs) and corneal endothelial cell density (CECD).

RESULTS

Both dose strengths of bimatoprost implant were noninferior to timolol in IOP lowering after each administration. Mean diurnal IOP was 24.0, 24.2, and 23.9 mmHg at baseline and from 16.5-17.2, 16.5-17.0, and 17.1-17.5 mmHg through week 12 in the 10-μg implant, 15-μg implant, and timolol groups, respectively. The incidence of corneal and inflammatory TEAEs of interest (e.g., corneal endothelial cell loss, iritis) was higher with bimatoprost implant than timolol and highest with the 15-μg dose strength. Incidence of corneal TEAEs increased after repeated treatment; with 3 administrations at fixed 16-week intervals, incidence of ≥20% CECD loss was 10.2% (10-μg implant) and 21.8% (15-μg implant). Mean best-corrected visual acuity (BCVA) was stable; 3 implant-treated subjects with corneal TEAEs had >2-line BCVA loss at their last visit.

CONCLUSIONS

Both dose strengths of bimatoprost implant met the primary end point of noninferiority to timolol through week 12. One year after 3 administrations, IOP was controlled in most subjects without additional treatment. The risk-benefit assessment favored the 10-μg implant over the 15-μg implant. Ongoing studies are evaluating other administration regimens to reduce the potential for CECD loss. The bimatoprost implant has potential to improve adherence and reduce treatment burden in glaucoma.

Weekly and seasonal changes of intraocular pressure measured with an implanted intraocular telemetry sensor

Background/Aims

To better understand seasonal and weekday intraocular pressure (IOP) variations, long-term daily IOP measurements were assessed in patients with glaucoma using an intraocular telemetric sensor.

Methods

This prospective, open-label, multicentre observational study analysed the IOP variation patterns in 22 eyes of 22 patients with primary open-angle glaucoma (67.8±6.8 years, 36.4% female) who had undergone placement of an intraocular telemetric sensor at the time of cataract surgery. The telemetric system combines an implantable IOP sensor with a hand-held reading device. Patients were instructed to self-measure their IOP as often as desired, but at least four times daily. Analysis of variance and Tukey multiple-comparison correction were used to assess the statistical significance of average and peak IOP variations between individual weekdays and months.

Results

Each enrolled patient recorded daily IOP measurements for an average duration of 721 days. On average, IOPs were highest on Wednesdays and lowest on Fridays (p=0.002). There were significant variations of IOP throughout the year, and IOP showed a seasonal pattern. Between mid-winter (December–January) and mid-summer months, there was a reduction in mean IOP of 8.1% (-1.55 mm Hg, p<0.05).

Conclusion

This study confirms previously observed seasonal variations of IOP. IOP was significantly higher in winter compared with summer months. Moreover, IOP was lower on Friday than on other days. The explanation for these results is not known.

Deep-layer Microvasculature Dropout in Preperimetric Glaucoma Patients

PURPOSE

To compare disease severity between preperimetric primary open-angle glaucoma (POAG) patients with and without deep-layer microvasculature dropout.

MATERIALS AND METHODS

Ninety-four eyes of 94 preperimetric POAG patients with β-zone parapapillary atrophy (βPPA) were categorized according to the presence of deep-layer microvasculature dropout defined as a complete loss of microvasculature within the choroid or scleral flange on optical coherence tomography angiography. Parameters representing disease severity, that is, visual field (VF) mean deviation (MD), global and sectoral (6-sector) retinal nerve fiber layer (RNFL) thickness, and other factors including age, focal lamina cribrosa (LC) defect, width of βPPA with and without Bruch membrane (BM) (βPPA+BM and βPPA-BM), and optic disc hemorrhage were compared between eyes with and without dropout.

RESULTS

Deep-layer microvasculature dropout was observed in 33 preperimetric POAG eyes (35.1%). Eyes with dropout had significantly thinner RNFL in all areas except the inferonasal sector, worse VF MD, and higher prevalence of focal LC defect, and larger βPPA-BM (P<0.05), whereas the 2 groups did not differ in age, disc hemorrhage, or βPPA+BM width (P>0.05). In the multivariable logistic regression, worse VF MD [odds ratio (OR), 1.485; P=0.045], thinner RNFL (OR, 1.141; P<0.001), and higher prevalence of focal LC defect (OR, 6.673; P<0.001) were significantly associated with dropout.

CONCLUSIONS

Deep-layer microvasculature dropout was observed in a considerable number of preperimetric POAG eyes, and worse disease severity was associated with dropout. Future studies elucidating the pathogenic role of deep-layer microvasculature dropout in the development and progression of glaucoma are warranted.

MicroRNA-19a-PTEN Axis Is Involved in the Developmental Decline of Axon Regenerative Capacity in Retinal Ganglion Cells

Irreversible blindness from glaucoma and optic neuropathies is attributed to retinal ganglion cells (RGCs) losing the ability to regenerate axons. While several transcription factors and proteins have demonstrated enhancement of axon regeneration after optic nerve injury, mechanisms contributing to the age-related decline in axon regenerative capacity remain elusive. In this study, we show that microRNAs are differentially expressed during RGC development and identify microRNA-19a (miR-19a) as a heterochronic marker; developmental decline of miR-19a relieves suppression of phosphatase and tensin homolog (PTEN), a key regulator of axon regeneration, and serves as a temporal indicator of decreasing axon regenerative capacity. Intravitreal injection of miR-19a promotes axon regeneration after optic nerve crush in adult mice, and it increases axon extension in RGCs isolated from aged human donors. This study uncovers a previously unrecognized involvement of the miR-19a-PTEN axis in RGC axon regeneration, and it demonstrates therapeutic potential of microRNA-mediated restoration of axon regenerative capacity in optic neuropathies.

The Relationship Between Intraocular Pressure and Rates of Central Versus Peripheral Visual Field Progression

PRECIS

In this longitudinal prospective cohort study of open-angle glaucoma patients, intraocular pressure (IOP) parameters (mean, fluctuation, and maximum) had a similar effect on glaucomatous progression in the central and peripheral visual field (VF) regions.

PURPOSE

To study the effects of IOP on rates of glaucomatous central versus peripheral VF progression.

METHODS

The African Descent and Glaucoma Evaluation Study (ADAGES) is a longitudinal prospective cohort study that recruited patients from 3 centers. A sample of those with established glaucoma were included in this study. The mean peripheral sensitivity (MPS) and the mean central sensitivity (MCS) were defined based upon the average total deviation of the peripheral and central (10 degrees) points of the 24-2 VF, respectively. Progression was based upon central and peripheral change from linear mixed-effects models. The relationships between VF progression and IOP mean, maximum, and fluctuation as continuous variables were also investigated. Main outcome measures were MPS and MCS progression rates.

RESULTS

A total of 452 eyes of 344 patients were studied. The mean number of VFs (SD) for each eye was 13.3 (6.4) over 9.1 (3.7) years. The mean baseline MD was -5.1 (3.9) dB and the mean rate of MD change was -0.26 dB/y [95% confidence interval (CI): -0.33 to -0.20, P<0.001]. Mean rates of MPS (-0.27 dB/y, 95% CI: -0.33 to -0.22, P<0.001) and MCS change (-0.26 dB/y, 95% CI: -0.31 to -0.21, P<0.001) were similar (P=0.351). Mean, fluctuation, and maximum IOP were significantly associated with MPS and MCS (all P<0.025).

CONCLUSIONS

The effect of IOP parameters on VF progression was statistically similar between central and peripheral VF regions.

Segmental differences found in aqueous angiographic-determined high - and low-flow regions of human trabecular meshwork

This work sought to compare aqueous angiographic segmental patterns with bead-based methods which directly visualize segmental trabecular meshwork (TM) tracer trapping. Additionally, segmental protein expression differences between aqueous angiographic-derived low- and high-outflow human TM regions were evaluated. Post-mortem human eyes (One Legacy and San Diego eye banks; n = 15) were perfused with fluorescent tracers (fluorescein [2.5%], indocyanine green [0.4%], and/or fluorescent microspheres). After angiographic imaging (Spectralis HRA+OCT; Heidelberg Engineering), peri-limbal low- and high-angiographic flow regions were marked. Aqueous angiographic segmental outflow patterns were similar to fluorescent microsphere TM trapping segmental patterns. TM was dissected from low- and high-flow areas and processed for immunofluorescence or Western blot and compared. Versican expression was relatively elevated in low-flow regions while MMP3 and collagen VI were relatively elevated in high-flow regions. TGF-β2, thrombospondin-1, TGF-β receptor1, and TGF-β downstream proteins such as α-smooth muscle actin were relatively elevated in low-flow regions. Additionally, fibronectin (FN) levels were unchanged, but the EDA isoform (FN-EDA) that is associated with fibrosis was relatively elevated in low-flow regions. These results show that segmental aqueous angiographic patterns are reflective of underlying TM molecular characteristics and demonstrate increased pro-fibrotic activation in low-flow regions. Thus, we provide evidence that aqueous angiography outflow visualization, the only tracer outflow imaging method available to clinicians, is in part representative of TM biology.

Matrix Metalloproteinases and Glaucoma Treatment

Matrix metalloproteinases (MMPs) are a family of proteolytic enzymes that degrade extracellular matrix (ECM) components such as collagen and have important roles in multiple biological processes, including development and tissue remodeling, both in health and disease. The activity of MMPs is influenced by the expression of MMPs and tissue inhibitors of metalloproteinase (TIMPs). In the eye, MMP-mediated ECM turnover in the juxtacanalicular region of the trabecular meshwork (TM) reduces outflow resistance in the conventional outflow pathway and helps maintain intraocular pressure (IOP) homeostasis. An imbalance in the MMP/TIMP ratio may be involved in the elevated IOP often associated with glaucoma. The prostaglandin analog/prostamide (PGA) class of topical ocular hypotensive medications used in glaucoma treatment reduces IOP by increasing outflow through both conventional and unconventional (uveoscleral) outflow pathways. Evidence from in vivo and in vitro studies using animal models and anterior segment explant and cell cultures indicates that the mechanism of IOP lowering by PGAs involves increased MMP expression in the TM and ciliary body, leading to tissue remodeling that enhances conventional and unconventional outflow. PGA effects on MMP expression are dependent on the identity and concentration of the PGA. An intracameral sustained-release PGA implant (Bimatoprost SR) in development for glaucoma treatment can reduce IOP for many months after expected intraocular drug bioavailability. We hypothesize that the higher concentrations of bimatoprost achieved in ocular outflow tissues with the implant produce greater MMP upregulation and more extensive, sustained MMP-mediated target tissue remodeling, providing an extended duration of effect.

Smart Electronic Eyedrop Bottle for Unobtrusive Monitoring of Glaucoma Medication Adherence

Glaucoma, the leading cause of irreversible blindness, affects >70 million people worldwide. Lowering intraocular pressure via topical administration of eye drops is the most common first-line therapy for glaucoma. This treatment paradigm has notoriously high non-adherence rates: ranging from 30% to 80%. The advent of smart phone enabled technologies creates promise for improving eyedrop adherence. However, previous eyedrop electronic monitoring solutions had awkward medication bottle adjuncts and crude software for monitoring the administration of a drop that adversely affected their ability to foster sustainable improvements in adherence. The current work begins to address this unmet need for wireless technology by creating a “smart drop” bottle. This medication bottle is instrumented with sensing electronics that enable detection of each eyedrop administered while maintaining the shape and size of the bottle. This is achieved by a thin electronic force sensor wrapped around the bottle and underneath the label, interfaced with a thin electronic circuit underneath the bottle that allows for detection and wireless transmission to a smart-phone application. We demonstrate 100% success rate of wireless communication over 75 feet with <1% false positive and false negative rates of single drop deliveries, thus providing a viable solution for eyedrop monitoring for glaucoma patients.

Effects of Study Population, Labeling and Training on Glaucoma Detection Using Deep Learning Algorithms

Purpose

To compare performance of independently developed deep learning algorithms for detecting glaucoma from fundus photographs and to evaluate strategies for incorporating new data into models.

Methods

Two fundus photograph datasets from the Diagnostic Innovations in Glaucoma Study/African Descent and Glaucoma Evaluation Study and Matsue Red Cross Hospital were used to independently develop deep learning algorithms for detection of glaucoma at the University of California, San Diego, and the University of Tokyo. We compared three versions of the University of California, San Diego, and University of Tokyo models: original (no retraining), sequential (retraining only on new data), and combined (training on combined data). Independent datasets were used to test the algorithms.

Results

The original University of California, San Diego and University of Tokyo models performed similarly (area under the receiver operating characteristic curve = 0.96 and 0.97, respectively) for detection of glaucoma in the Matsue Red Cross Hospital dataset, but not the Diagnostic Innovations in Glaucoma Study/African Descent and Glaucoma Evaluation Study data (0.79 and 0.92; P < .001), respectively. Model performance was higher when classifying moderate-to-severe compared with mild disease (area under the receiver operating characteristic curve = 0.98 and 0.91; P < .001), respectively. Models trained with the combined strategy generally had better performance across all datasets than the original strategy.

Conclusions

Deep learning glaucoma detection can achieve high accuracy across diverse datasets with appropriate training strategies. Because model performance was influenced by the severity of disease, labeling, training strategies, and population characteristics, reporting accuracy stratified by relevant covariates is important for cross study comparisons.

Translational Relevance

High sensitivity and specificity of deep learning algorithms for moderate-to-severe glaucoma across diverse populations suggest a role for artificial intelligence in the detection of glaucoma in primary care.