Article Text

Download PDFPDF

Swept-source OCT angiography imaging of the macular capillary network in glaucoma
  1. Handan Akil1,2,
  2. Vikas Chopra1,2,
  3. Mayss Al-Sheikh1,
  4. Khalil Ghasemi Falavarjani1,3,
  5. Alex S Huang1,2,
  6. SriniVas R Sadda1,2,
  7. Brian A Francis1,2
  1. 1 Doheny Image Reading Center, Doheny Eye Institute, Los Angeles, California, USA
  2. 2 Department of Ophthalmology, David Geffen School of Medicine, Doheny Eye Centers UCLA, University of California, Los Angeles, California, USA
  3. 3 Eye Research Center, Rasoul Akram Hospital, Iran University of Medical Sciences, Tehran, Islamic Republic of Iran
  1. Correspondence to Dr Handan Akil, Doheny Image Reading Center, Doheny Eye Institute, Los Angeles, California, USA; handanakil84{at}


Purpose To evaluate the macular capillary network density of superficial and deep retinal layers (SRL/DRL) by swept-source optical coherence tomography angiography (OCTA) in patients with primary open angle glaucoma (POAG) and to compare the results with those of normal subjects.

Method In this prospective study, 24 eyes of 24 normal individuals and 24 eyes of 24 patients with mild to moderate POAG underwent fovea centred 6×6 mm cube macular OCTA imaging by a swept-source OCTA device (Triton, Topcon, Tokyo, Japan). Quantitative analysis of the retinal vasculature was performed by assessing vessel density (VD) as the ratio of the retinal area occupied by vessels at the SRL and DRL.

Results The mean VD (ratio) at the SRL and DRL was statistically significantly lower in patients with POAG (SRL, p<0.001; DRL, p<0.001). In the SRL, the mean±SD VD ratio was 0.34±0.05 in patients with POAG and 0.40±0.02 in normal individuals (p<0.001). In the DRL, the mean (SD) ratio was 0.37±0.05 in patients with POAG and 0.43±0.02 in normal individuals (p<0.001). The mean VD at the SRL was significantly correlated with ganglion cell inner plexiform layer thickness (r=0.42, p=0.04) but not with visual field mean deviation (r=0.4, p=0.06) and retinal nerve fibre layer thickness (r=0.5, p=0.06). The mean VD at the DRL did not show significant correlation with any other glaucoma parameter (p>0.05).

Conclusion The assessment of macular VD by swept-source OCTA may offer additional information for detection of glaucoma.

  • macula
  • glaucoma
  • retinal nerve fiber layer
  • ganglion cell inner plexiform layer
  • optical coherence tomography angiography

Statistics from


Glaucoma causes optic neuropathy in a progressive pattern with the ganglion cell and axon dysfunction at the posterior pole.1 2 It has been suggested that intraocular pressure may not be the only pathogenic factor in the disease process. The potential role of the retinal blood flow and microvasculature has been evaluated extensively to understand possible effects of ischaemia on the loss of retinal ganglion cells.1 2 The blood flow of the optic nerve head, retina, choroid and retrobulbar areas has been shown to be reduced in patients with glaucoma.3–5 To date, however, no studies evaluating the relationship of ocular blood flow and glaucoma have been able to show a reproducible quantitative assessment method.6

The clinical and epidemiological evidence has shown that inadequate ocular blood flow may play a central role in apoptotic retinal ganglion cell death; but current diagnosis and management of glaucoma have not taken these into consideration, yet.

A complete description of ocular blood flow in terms of microcirculation has not yet been established. Current assessment techniques only give information about certain parts of the ocular circulation.

It has been demonstrated that the macular microvasculature supplies blood to the inner retina through capillary plexuses: the superficial retinal layer (SRL) and deep retinal layer (DRL). The SRL is located in the retinal nerve fibre layer (RNFL), and the DRL consists of two plexuses, located at the inner and outer borders of the inner nuclear layer.7 The choroidal circulation plays a major role in the blood supply of the outer retina by diffusion.8 Glaucoma has been associated with morphological changes that affect some macular layers, particularly the nerve fibre and ganglion cell layers.9 Various alterations to the superficial and deep capillary plexus that cannot be seen by fundus angiography (FA) are speculated to have a role in the pathophysiology of glaucoma.

Optical coherence tomography angiography (OCTA) is a newly developed imaging strategy that allows mapping of the microvasculature in various retinal layers,10 providing high-speed, three-dimensional evaluation of the microcirculation in the optic nerve head and peripapillary region.11–13

Although several studies have shown microvascular changes in the peripapillary area, quantitative comparisons of macular microvasculature between normal eyes and eyes with primary open angle glaucoma (POAG) have been limited.

We designed the current study to evaluate the macular capillary network density in the SRL and DRL in patients with POAG and to compare our results with those of healthy subjects using a swept-source OCTA.


The subjects for this cross-sectional and comparative study were patients with POAG who were examined at the glaucoma service at UCLA Doheny Eye Centers between January 2016 and June 2016. Healthy volunteers with normal ocular examinations (no eye pathology and intraocular pressure ≤21 mm Hg) served as controls. All examinations were performed according to a protocol that adhered to the tenets of the Declaration of Helsinki and were approved by the Institutional Review Board of the University of California, Los Angeles. Informed consent for study participation was obtained from all subjects.

All study participants underwent a regular ocular examination conducted by fellowship-trained glaucoma specialists, including refraction, intraocular pressure measurement, gonioscopy, anterior segment examination and a dilated fundus examination. In addition, fundus photography, standard automated perimetry (24-2 Swedish Interactive Threshold Algorithm (SITA) standard, Humphrey Field Analyzer; Carl Zeiss Meditec, Dublin, CA), peripapillary and macular optical coherence tomography (OCT, Cirrus HD-OCT; Carl Zeiss Meditec, Jena, Germany) were performed for all eyes.

Study participants with any kind of neurological disease or a history of diabetes or corticosteroid use were excluded. Eyes with a history of retinal disease, opaque media or ocular surgery (other than previous, uncomplicated cataract extraction surgery) were excluded. Eyes were also excluded if they had best corrected visual acuity worse than 20/25, had a refractive error outside the range −6.00 to +3.00 D, or had narrow angles (< grade 2 Schaffer grade) noted on gonioscopy.

Visual field testing

The patients underwent standard automated perimetry (Humphrey 24-2 SITA) at least twice within 6 months of the ocular examination and reliable visual field (VF) test results (fixation loss ≤20%, false-positive rate ≤15% and false-negative rate ≤33%) were used. VF defects were classified as glaucomatous by either an abnormal report on the Glaucoma Hemifield Test or a pattern standard deviation (PSD) of <5% of the normal reference (confirmed by two consecutive tests) and if the defects were clinically determined to be characteristic or compatible with glaucoma by the clinician.

Optical coherence tomography image acquisition and processing

Three-dimensional cube OCT data were acquired from eyes with the Cirrus HD-OCT device using the Macular Cube 200×200 scan protocol. This protocol, designed for retinal topography analysis, performs 200 horizontal B scans comprising 200 A scans per B scan over 1024 samplings within a cube measuring 6×6×2 mm.

Swept-source OCTA

A 1050 nm swept-source OCTA device (DRI OCT Triton, Topcon, Tokyo, Japan) was used to obtain the images. The device has an acquisition speed of 100 000 A scans per second. Scans were taken from 6×6 mm cubes centred on the fovea. The automated layer segmentation of the OCT instrument software helped generate en face images of the retinal vasculature from the SRL and DRL through en face slabs. The SRL and DRL extended from the internal limiting membrane to the inner border of the inner nuclear layer and from the inner border of the inner nuclear layer to the outer border of the inner nuclear layer, respectively.

Quantitative measurements and statistical analysis

The GNU Image Manipulation Program GIMP V.2.8.14 (available in public domain at was used to perform quantitative analysis.

The vessel density (VD) of the SRL and DRL slabs in the 6×6 macula scans was evaluated using the colour selection tool of the GIMP software in accordance with our previously described technique.14 Briefly, the colour threshold was changed to detect the entire vascular network in the scan area. VD was calculated as the proportion of the measured area occupied by blood vessels with flow, defined as pixels having decorrelation values above the threshold level.

Two independent, masked, certified Doheny Image Reading Center graders measured the VD; the results coming from the main grader were used for the analysis. The intergrader agreement was calculated by using measurements made by both graders.

SPSS statistical software V.21 (SPSS, IBM) was used for statistical analysis. A p value of <0.05 was considered significant. Correlation of the VD with OCT and VF measurements was analysed using the Pearson correlation test. Intraclass correlation coefficients (ICC) with 95% CIs were calculated to assess intergrader agreement.


Overall, 48 eyes of 48 subjects including 24 eyes of 24 patients with POAG and 24 eyes of 24 healthy volunteers were included in this study. The mean age of the patients with POAG was 64.1±9.4 years; 13 (54.1%) were male and 11 (45.8%) were female. Eighteen (75%) were Caucasian and six (25%) were Hispanic.

The average age of control subjects was 62.8±13.7 years; 12 (50%) were male and 12 (50%) were female. Twenty (83.3%) were Caucasian and four (16.7%) were Hispanic.

There was significant difference between patients with glaucoma and age-matched healthy controls in terms of glaucoma parameters (p<0.05) (table 1).

Table 1

Characteristics of the study groups

Mean RNFL thickness was 69.3±11.9 µm, ganglion cell inner plexiform layer (GC-IPL) thickness was 63.8±11.04 µm, VF mean deviation (MD) was −6.4±5.2 dB and VF PSD was 7.0±4.5 dB. Nine patients were taking more than two glaucoma medications in the study group.

In the SRL, the mean±SD VD ratio was 0.34±0.05 in patients with POAG and 0.41±0.03 in normal individuals (figure 1) (p<0.001, table 2). In the DRL, the mean±SD ratio was 0.37±0.05 and 0.43±0.01 for patients with POAG and normal individuals, respectively (figure 2) (p<0.001, table 2). The mean vessel density was statistically significantly lower in both the SRL and DRL in patients with POAG (Figure 3).

Figure 1

Macular vascular network at superficial retinal layer in glaucoma and healthy eye.

Figure 2

Macular vascular network at deep retinal layer in glaucoma and healthy eye.

Figure 3

Histogram of the macular vessel density measurements at SRL and DRL in healthy and glaucoma eyes. DRL, deep retinal layer; POAG, primary open angle glaucoma; SRL, superficial retinal layer.

Table 2

Vessel density measurements of study groups.

The mean difference between the two graders was 0.01±0.02 mm2 for the SRL and 0.013±0.04 mm2 for the DRL. The ICC was 0.996 (95% CI 0.993 to 0.999) and 0.995 (95% CI 0.993 to 0.998) for the SRL and DRL, respectively.

In the POAG group, univariate regression analysis using the Pearson correlation coefficient showed that the mean VD at SRL was significantly correlated with macular GC-IPL thickness (r=0.42, p=0.04), but the significance was borderline with VF MD (r=0.4, p=0.06) and peripapillary RNFL thickness (r=0.5, p=0.06), and the difference was non-significant with VF PSD (r=0.16, p=0.4). The mean VD at DRL did not show statistically significant correlation with any other glaucoma parameter (p>0.05) (figure 4).

Figure 4

Scatterplot analysis of the correlation of the vessel density at SRL with visual field (MD) and GC-IPL. GC-IPL, ganglion cell inner plexiform layer; MD, mean deviation; PSD, pattern standard deviation; SRL, superficial retinal layer.

The area under the receiver operating characteristic (ROC) curve for differentiating normal from POAG was 0.903 (p<0.05) for the VD at the SRL and 0.91 (p<0.05) for the VD at the DRL. The ROC curves showed that the cut-off point was 36.5 for VD at the SRL (73% sensitivity to 95% specificity) and 41.1 for VD at the DRL (92% sensitivity to 99% specificity) between the controls and POAG eyes. ROC for differentiating normal from POAG was 0.87 (p<0.05) for the RNFL and 0.89 (p<0.05) for the GC-IPL.


The current study used swept-source OCTA technology to evaluate the VD of the macular capillary network in patients with POAG. We demonstrated a statistically significant lower VD of the capillary network in both SRL and DRL of patients with POAG compared with healthy subjects.

Previous studies have shown varying evidence of ocular blood flow abnormalities in patients with POAG.4 5 In animal models of glaucoma, retinal ganglion cells have been shown to die via apoptosis.15 16 One reason for apoptosis in these cells might be ischaemia. Therefore, ocular blood flow measurement, especially by non-invasive methods, may be useful in the management of patients with glaucoma to determine the aetiology and severity of the disease process and to evaluate the outcome of medical and/or surgical therapy in the near future.

On the other hand, the structural changes caused by the destruction of the GC-IPL and nerve fiber layer in glaucoma may lead to secondary microvascular changes. The macular thickness and volume are significantly lower in glaucoma.2 Vascular abnormalities have been assumed to be associated with the structural changes. Many methods have been used to evaluate haemodynamics in glaucoma, but results of studies relating ocular microcirculation alterations to glaucoma are difficult to compare because of the various techniques applied, different parameters investigated and demographic differences of the glaucoma populations studied.

It should be noted that because of the limitations of the current OCTA technology, we did not actually measure functional blood flow and were limited to structural retinal vessel calibre measurements. A variety of methods have been used to document differences in ocular circulation between glaucomatous and normal eyes, but none of these have been found to be appropriate for routine clinical use.9 The low reproducibility and high variability of the measurements using those methods showed that they have limited ability to provide information on vascular structures which might contribute to the pathogenesis of the disease.17 18 Therefore, OCTA might be a useful tool to show the vascular changes through the measurement of retinal VD with excellent reproducibility. Our current data can be a useful starting point for studying the relation between macular vasculature and glaucoma.

Our OCTA results of attenuated retinal VD in glaucoma are corroborated by a previous study by Jonas et al that used fundus images to measure retinal vessel calibre. In that report, they obtained their vessel calibre assessments by projecting fundus images from slides onto a screen at a known magnification and measuring the vessel diameter using a sophisticated micrometric tool. They showed that the parapapillary retinal vessel calibre in the glaucoma group was significantly smaller than that of the normal eyes.19

Scanning laser ophthalmoscopy (SLO) has been described with fluorescein angiography (FA) and indocyanine green (ICG) angiography with their several limitations.20 SLO FA has brought valuable information on retinal haemodynamics of healthy and glaucoma subjects. Wolf et al used the SLO method to evaluate retinal blood velocities of patients with POAG and healthy controls.21 They demonstrated an 11% reduction in the mean dye velocity within major retinal arteries. They also found slower arteriovenous passage within the retina in POAG subjects. SLO ICG angiography has also been used to investigate ocular blood flow dynamics. In a prospective study, ICG angiograms with a new dilution analysis technique were recorded from patients with glaucoma and normal subjects.20 In contrast to the homogenous early choroidal filling pattern of normal subjects, the glaucoma group was found to have a heterogeneous pattern. In addition, peripapillary choroidal filling was delayed in glaucomatous eyes but not in normal subjects. They suggested that patients with glaucoma may suffer from reduced blood flow of the choroid. Although the choroidal vascular alterations have been claimed to play a role in the pathogenesis of glaucoma, many studies have not definitively established a relationship between macular choroidal vasculature and glaucoma.3 In our current study, a statistically significant decrease of VD was found at the SRL and DRL between patients with POAG and healthy individuals. Although both layers were found to be affected, a more consistent and severe decrease in VD was evaluated in the SRL. This observation could be compatible with the theory that the macular superficial capillary network in glaucoma has more vascular network changes than the DRL and choroid.22

Confocal scanning laser Doppler flowmetry has also shown blood flow alterations in glaucoma. A study using a new pixel-by-pixel analysis to compare age-matched controls with patients with glaucoma demonstrated that patients with glaucoma had significantly lower blood flow compared with healthy individuals.20

Doppler OCT strategy has been found to have a better level of precision than other Doppler methods for evaluating total retinal blood flow. Since the microvascular network of the retina plays a critical role in the pathophysiology of glaucoma, this method may not be useful due to the inability to assess the blood flow at those networks.18

OCTA has been shown to be a novel, non-invasive method that allows visualisation of the retinal blood flow at the microcirculation level in a three-dimensional, depth-resolved fashion.14 23–25 Using intensity and/or phase properties imaged over multiple B scans, OCTA is able to non-invasively measure the flow of red blood cells. Although it remains a promising non-invasive tool compared with conventional FA, limitations still exist, especially with regard to movement, projection artefact and segmentation issues.26

Previous studies showed that the more negative VF MD becomes, the lower the thickness of RNFL, meaning that eyes with more advanced glaucoma have a lower RNFL thickness. This relation also holds for the stage of glaucoma and ganglion cell layer thickness, although it is not as strong as with the RNFL. Our study showed that superior macular VD was significantly correlated with GC-IPL thickness. We speculate that vascular changes of superficial capillary plexus are prominent in glaucoma.

POAG eyes were found to have much variability in capillary density measurements compared with normal eyes. This may be caused by the different stages of vasculopathy within the study group. Patients with mild POAG seemed to have higher capillary density compared with patients with moderate glaucoma; however, the current study was not designed to evaluate these differences among glaucoma groups.

Although our study had a small sample size, the high repeatability of our measurements for the differences between the glaucomatous eyes and age-matched normal eyes was demonstrated reliably. The repeatability between graders for capillary density is a major strength of this study and it was found to be excellent with an ICC of 0.996 and 0.995 for the VD and a coefficient of variation of 0.01.

This prospective study is not without limitations. We enrolled a relatively small sample of patients with glaucoma, which may not be enough to reach a definite conclusion. It should be noted that antiglaucomatous eye drops and systemic medications might have an impact on qualitative measurements of the vascular structures. Current data are underpowered to evaluate this effect due to the cross-sectional design of the study. One might wonder whether the systemic diseases and medications could have an influence on VD measurements. Therefore, further studies are needed to consider the influence of these on qualitative measurements of microvascular networks. Test–retest variability and age-related macular changes may also limit the ability to use macular thickness measurements as a clinical endpoint for glaucoma progression. Lastly, it is important to emphasise that macular VD measurements have limited use for monitoring glaucoma in eyes with macular comorbidity. Thus, eyes with diabetic or age-related maculopathy are not candidates for monitoring macular vascular changes as a strategy for glaucoma diagnosis or detection of glaucomatous progression.

Based on the results of the current study, we hypothesise that OCTA provides a new imaging strategy for early diagnosis and management of glaucoma. Clearly, longitudinal data with larger sample sizes are needed to evaluate this possibility. Our future studies will look at OCTA over a greater range of glaucoma with more subjects at each severity to provide more information about the vascular pathophysiology of the disease.


OCTA is a new modality that may provide great assets to clinicians. Although OCTA technology provides high-speed, non-invasive, depth-resolved imaging of the retinal and choroidal vasculature using motion contrast instead of the dye used with fluorescein and ICG angiography, it still has limitations such as accuracy of segmentation, smaller field of view, background noise, artefacts caused by movement and need of special software to reduce variability.

OCTA technology should be further studied to assess whether it provides an accurate non-invasive option for detailed visualisation of the retinal microvasculature.

Clinically, the evaluation of macular blood flow by OCTA offers promise for detection, differentiation and diagnosis of various ocular diseases. The severity of specific microvascular alterations can be assessed to monitor disease progression. In addition, the technology may be useful for evaluating the vascular impact of common ocular treatments and surgical interventions.



  • Contributors HA, MA, KGF, BAF, ASH: the conception or design of the work, the acquisition, analysis and interpretation of data. HA, KGF, MA, BAF, VC, SS, ASH: drafting the work and revising it critically for important intellectual content. HA, MA, KGF, BAF, VC, ASH, SS: final approval of the version. HA, MA, KGF, BAF, VC, ASH, SS: agreement to be accountable for all aspects of the study.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval IRB of UCLA.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The additional data from the study can be requested from the corresponding author at

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.