Article Text
PDF
Abstract
Aims To evaluate the diagnostic ability of peripapillary vessel density measurements on optical coherence tomography angiography (OCTA) in primary open-angle glaucoma (POAG) and primary angle-closure glaucoma (PACG), and to compare these with peripapillary retinal nerve fibre layer (RNFL) thickness measurements.
Methods In a cross-sectional study, 48 eyes of 33 healthy control subjects, 63 eyes of 39 patients with POAG and 49 eyes of 32 patients with PACG underwent OCTA (RTVue-XR, Optovue, Fremont, California, USA) and RNFL imaging with spectral domain OCT. Diagnostic abilities of vessel density and RNFL parameters were evaluated using area under receiver operating characteristic curves (AUC) and sensitivities at fixed specificities.
Results AUCs of peripapillary vessel density ranged between 0.48 for the temporal sector and 0.88 for the inferotemporal sector in POAG. The same in PACG ranged between 0.57 and 0.86. Sensitivities at 95% specificity ranged from 13% to 70% in POAG, and from 10% to 67% in PACG. AUCs of peripapillary RNFL thickness ranged between 0.51 for the temporal sector and 0.91 for the inferonasal sector in POAG. The same in PACG ranged between 0.61 and 0.87. Sensitivities at 95% specificity ranged from 8% to 68% in POAG, and from 2% to 67% in PACG. AUCs of all peripapillary vessel density measurements were comparable (p>0.05) to the corresponding RNFL thickness measurements in both POAG and PACG.
Conclusions Diagnostic ability of peripapillary vessel density parameters of OCTA, especially the inferotemporal sector measurement, was good in POAG and PACG. Diagnostic abilities of vessel density measurements were comparable to RNFL measurements in both POAG and PACG.
- Glaucoma
- Imaging
Statistics from Altmetric.com
Introduction
Glaucoma is a chronic progressive optic neuropathy with characteristic optic disc and retinal nerve fibre layer (RNFL) changes. Although the exact pathogenic mechanisms of glaucoma are not fully understood, intraocular pressure (IOP) is a major causal factor with the risk of incident glaucoma and its progression increasing with higher IOP.1 It has also been proposed that reduced optic nerve head (ONH) perfusion is a cause of glaucoma in at least some individuals.2 ,3 Earlier studies have measured ONH blood flow using a variety of techniques and have shown reduction in ONH perfusion in patients with glaucoma. However, each of these techniques has certain limitations.4
Recently, optical coherence tomography (OCT) has been used to develop a three-dimensional angiography algorithm called split spectrum amplitude-decorrelation angiography (SSADA) for imaging the ONH microcirculation.5 Studies with this OCT angiography (OCTA) have demonstrated reduced ONH6–9 and peripapillary10–12 vessel density in patients with glaucoma. Most of the previous studies included patients with primary open-angle glaucoma (POAG) and there are no reports to date on the utility of peripapillary vessel density parameters of OCTA in eyes with primary angle-closure glaucoma (PACG). The purpose of this study was to evaluate the diagnostic ability of the peripapillary vessel density in different sectors on OCTA in POAG and PACG, and to compare these with peripapillary RNFL thickness as measured by spectral domain OCT (SD-OCT).
Methods
This was a cross-sectional study conducted at a tertiary eye care centre between June 2015 and August 2015. The methodology adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all participants.
Participants of the study included patients with POAG, patients with PACG and a group of control subjects. Control eyes in the study had IOP ≤21 mm Hg, no family history of glaucoma, normal anterior and posterior segment on clinical examination and non-glaucomatous optic discs, as assessed by experts on disc photographs. Both patients with PACG and patients with POAG had ONH changes characteristic of glaucoma (focal or diffuse neuroretinal rim thinning, localised notching or RNFL defects). Patients with PACG had occludable anterior chamber angles in three or more quadrants, with goniosynechiae and IOP >21 mm Hg at the time of diagnosis. Anterior chamber angle was examined using an indentation gonioscope and was considered occludable if the posterior trabecular meshwork was not seen in the primary position. Patients with POAG had open angles and IOP >21 mm Hg at the time of diagnosis. Inclusion criteria for all participants were age ≥18 years, corrected distance visual acuity (CDVA) of 20/40 or better and refractive error within ±5 D sphere and ±3 D cylinder. Exclusion criteria were presence of any media opacities, or retinal or neurological disease that could confound the examinations. Eyes with history of trauma or inflammation were also excluded. All participants underwent a detailed medical history, CDVA measurement, slit-lamp biomicroscopy, Goldmann applanation tonometry, gonioscopy, dilated fundus examination, visual field (VF) examination, stereoscopic optic disc photography, OCTA imaging with RTVue-XR SDOCT (Optovue, Fremont, California, USA) and RNFL imaging with Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, California, USA). All examinations were performed on the same day.
VF testing
VF examination was performed using a Humphrey Field analyzer II, model 720i (Zeiss Humphrey Systems, Dublin, California, USA), with the Swedish interactive threshold algorithm standard 24-2 program. VFs were considered reliable if the fixation losses, and false positive and false negative response rates were all ≤20%.
Optic disc photography
Stereoscopic optic disc photographs were obtained by trained technicians using a digital fundus camera (Kowa nonmyd WX, Kowa Company, Japan). Each optic disc photograph was evaluated independently by two of the three experts (HLR, ZSP and NKP) in a masked manner to determine the presence of glaucomatous changes. Discrepancies between any two experts were adjudicated by the third expert.
OCTA examination
OCTA imaging was performed using RTVue-XR SDOCT (AngioVue, v2015.100.0.33) after pupillary dilatation. RTVue-XR scans the optic disc using an 840 nm diode laser source, with an A-scan rate of 70 kHz/s. Optic disc imaging is performed using a set of two scans; one vertical priority and one horizontal priority raster volumetric scan covering 4.5×4.5 mm. An orthogonal registration algorithm is used to produce merged three-dimensional OCT angiograms.13 The SSADA algorithm compares the consecutive B-scans at the same location to detect flow using motion contrast.5 The software automatically fits an ellipse to the optic disc margin. The peripapillary region is defined as a 0.75 mm-wide elliptical annulus extending from the optic disc boundary (figure 1). An en face angiogram of the circulation is obtained by the maximum flow (decorrelation value) projection from the inner limiting membrane (ILM) to retinal epithelial pigment. In this study, the peripapillary vessels were analysed in the radial peripapillary capillary (RPC) zone. The RPC zone extends from the ILM to the nerve fibre layer. Peripapillary vessel density was defined as the percentage area occupied by the large vessels and microvasculature in the peripapillary region. The peripapillary region was divided into six sectors based on the Garway-Heath map (figure 1).14 Poor quality images with a Signal Strength Index (SSI) <40 or images with residual motion artefacts were excluded from the analysis.
Figure showing the peripapillary sectors where vessel densities are calculated. The vessel density is calculated over a 0.75 mm-wide elliptical annulus extending from the optic disc boundary.
RNFL imaging
RNFL thickness measurements were performed with Cirrus HD-OCT (software V.7.0.1.290) using the Optic Disc Cube 200×200 protocol.15 ,16 The 12 clock-hour RNFL thickness measurements (in the right eye format) were grouped to closely match the sectors of the OCTA parameters. Clock hours 8, 9 and 10 were considered as the temporal sector, 11 as the superotemporal sector, 12 and 1 as superonasal, 2, 3 and 4 as nasal, 5 and 6 as inferonasal and 7 as the inferotemporal sector. Only good quality scans with signal strength ≥6, absence of motion and blinking artefacts, and absence of segmentation failure were used for the analysis.
Statistical analysis
Receiver operating characteristic (ROC) curves and sensitivities at fixed specificities were used to describe the ability of OCTA and RNFL parameters to discriminate glaucomatous eyes from control eyes. To obtain CIs for area under the ROC curves (AUC) and sensitivities, a bootstrap resampling procedure was used (n=1000 resamples). As measurements from both eyes of the same subject are likely to be correlated, the cluster of data for the study subject were considered as the unit of resampling and bias corrected SEs were calculated. This procedure has been used to adjust for the presence of multiple correlated measurements from the same unit.17 ,18 The ROC regression modelling technique was used to evaluate the effect of glaucoma severity on the sensitivities of OCTA and RNFL parameters in diagnosing glaucoma.19 ,20
Statistical analyses were performed using commercial software (Stata V.12.1; StataCorp, College Station, Texas, USA). A p value of ≤0.05 was considered statistically significant.
Results
Two hundred and twenty eyes of 139 subjects (62 eyes of 43 normal, 90 eyes of 53 POAG and 68 eyes of 43 PACG subjects) underwent OCTA imaging with SD-OCT. Among these, 60 eyes were excluded due to poor quality OCTA scans (27 eyes), RNFL scans (5 eyes), unreliable VF (19 eyes) or combinations of these (9 eyes), leaving 160 eyes (48 eyes of 33 control subjects, 63 eyes of 39 patients with POAG and 49 eyes of patients with 32 PACG) for the analysis. The pairwise agreement between the three experts for optic disc classification on stereo photographs ranged between 90.5% and 94.9%. Kappa ranged between 0.80 and 0.92. Twelve eyes each in the POAG and the PACG groups had normal VFs (pattern standard deviation with p>5% and/or within normal limits or borderline Glaucoma Hemifield Test result). Table 1 shows the characteristics of these three groups. Patients with POAG and PACG were significantly older than control subjects. The SSI of OCTA scans and signal strength of RNFL scans were significantly greater in the control compared with POAG and PACG subjects. AUCs and sensitivities at fixed specificities were therefore calculated after adjusting for the difference in age and signal strength between the groups using covariate adjustment as proposed by Pepe.21 All vessel density and RNFL measurements were significantly lesser in the glaucoma group. Though the median mean deviation (MD) was worse in PACG compared with POAG eyes, the difference was not statistically significant (p=0.49).
Demographic, visual field, vessel density and RNFL thickness characteristics of the participants
The AUCs and sensitivities at fixed specificities of the peripapillary vessel density and RNFL parameters are shown in tables 2 and 3, respectively. Inferior sector measurements of both vessel density and RNFL thickness showed the best AUCs and sensitivities at fixed specificities in POAG and PACG.
Diagnostic ability of peripapillary vessel density and RNFL parameters in differentiating primary open angle glaucoma from control eyes (figures in parenthesis represent 95% CIs)
Diagnostic ability of peripapillary vessel density and RNFL parameters in differentiating primary angle-closure glaucoma from control eyes (figures in parenthesis represent 95% CIs)
AUCs of the OCTA parameters were comparable (p>0.05) to the corresponding RNFL thickness parameters both in POAG and PACG groups. Sensitivities at fixed specificities of most of the OCTA parameters were also comparable to the corresponding RNFL thickness measurements (overlapping 95% CIs). The AUCs and sensitivities at fixed specificities of OCTA and RNFL thickness measurements in POAG were comparable to that in PACG (p>0.05 for all comparisons).
ROC regression analysis showed a statistically significant influence of glaucoma severity (as measured by mean deviation on VF) on the diagnostic abilities of both peripapillary vessel density and RNFL measurements (p<0.05 for all parameters). Figure 2 shows the effect of glaucoma severity on the sensitivity at 95% specificity of the inferotemporal peripapillary vessel density and RNFL measurements in (A) POAG and (B) PACG. Sensitivities of both the vessel density and RNFL thickness parameters increased significantly as the severity of glaucoma increased. Sensitivities of the inferotemporal peripapillary vessel density appeared to be better in POAG compared with PACG with increasing severity of the disease.
Sensitivity at 95% specificity of inferotemporal sector vessel density and retinal nerve fibre layer (RNFL) thickness measurements in (A) primary open-angle glaucoma (POAG) and (B) primary angle-closure glaucoma (PACG).
We ran the entire analysis considering one eye of subjects who contributed both eyes for our earlier analysis and found similar results. When considering the better eye of the patients with glaucoma for analysis (median MD: −5.3 dB both in the POAG and PACG groups), the inferotemporal sector vessel density showed the best AUC both in POAG (0.87) and PACG (0.86). The AUC of inferotemporal RNFL thickness was 0.86 and 0.85 in POAG and PACG, respectively. When considering the worse eye of the patients with glaucoma for analysis (median MD: −6.8 dB in POAG and −11.1 dB in PACG), inferotemporal sector vessel density showed the best AUC both in POAG (0.92) and PACG (0.87). The AUC of inferotemporal RNFL thickness was 0.89 and 0.88 in POAG and PACG, respectively. AUCs of the OCTA parameters were comparable (p>0.05) to the corresponding RNFL thickness parameters both in the POAG and PACG groups.
Discussion
There is limited literature on the diagnostic ability of peripapillary vessel densities of OCTA in POAG10–12 and none to our knowledge in PACG. Evaluating 12 glaucoma and 12 normal eyes, Liu et al reported a significant reduction of peripapillary vessel density in glaucoma eyes. The AUC, sensitivity and specificity of average peripapillary vessel density (0.94, 83.3% and 91.7%, respectively) were comparable to average RNFL thickness (0.97, 91.7% and 91.7%, respectively).10 Yarmohammadi et al compared the diagnostic ability of average peripapillary vessel density and RNFL thickness in 124 eyes with POAG (median MD: −3.9 dB). The AUC of peripapillary vessel density measurement (0.83) was lesser than that of the average RNFL thickness (0.92). This difference however was not statistically significant.11 Diagnostic abilities of average vessel density and RNFL thickness in POAG in our study were lesser than that reported by Liu et al but similar to that reported by Yarmohammadi et al. Sectorwise analysis of peripapillary vessel densities was not performed in both these previous studies.
We evaluated the diagnostic ability of peripapillary vessel density measurements in PACG and compared them with the RNFL measurements. We found them to be comparable. We analysed the diagnostic ability of OCTA parameters separately in patients with POAG and PACG because we hypothesised that the role of blood flow may not be the same in the pathogenesis of POAG and PACG, and expected the diagnostic ability of OCTA parameters to be different. However, we found that the AUCs of OCTA measurements in POAG were comparable to that in PACG. The severity of glaucoma in terms of mean deviation on VF, was greater in PACG (−9.2 dB) compared with POAG (−6.3 dB) eyes, in spite of it being statistically comparable. When we accounted for the effect of glaucoma severity on the diagnostic abilities of OCTA parameters, the sensitivity of the inferotemporal peripapillary vessel density appeared to be better in POAG compared with PACG with increasing severity of the disease (figure 2). This may indicate a lower prevalence of ocular perfusion abnormality in PACG compared with POAG. Future studies should evaluate this hypothesis.
Although RTVue-XR provides RNFL thickness measurements in addition to vessel densities, we considered the RNFL thickness parameters from Cirrus HD-OCT for the analysis because Cirrus HD-OCT provides the RNFL thickness in 12 clock-hour segments which can be combined into sectors that closely match the vessel density sectors provided by OCTA. Though a previous study has shown significant differences between RTVue and Cirrus OCT in RNFL thickness measurement,22 a study by Leite et al23 has shown no difference in their diagnostic abilities in glaucoma.
There are a few limitations of the study. Patients with glaucoma were significantly older than control subjects. Though we accounted for this difference during the calculation of diagnostic accuracies, it would have been ideal to match the groups during recruitment. The vessel density measurements evaluated in this study were the ones provided by the software automatically. We therefore could not exactly match the vessel density and the RNFL sectors for comparison. This may have affected our results and could also be a reason for the diagnostic ability of the nasal RNFL sectors to be higher than the temporal sectors. The software in its current form does not differentiate the changes in capillaries from that in large vessels. Another possible limitation of the current study was that we did not measure the blood pressure of the subjects or record their antihypertensive medication. However, previous studies have shown no relationship between blood pressure readings and peripapillary vessel densities on OCTA.10 Lastly, the sample size of our study, though was much larger than the previous studies on OCTA, was still small, as was evidenced by the wide CIs for the AUCs and sensitivities at fixed specificities.
Our study provides directions to future research with OCTA in glaucoma. Our study demonstrated that OCTA has the potential to provide useful information in glaucoma. Though the diagnostic abilities of vessel densities were not better than the tradition RNFL measurements, combining the information from vessel density measurements might enhance the diagnostic yield of other standard tests in glaucoma. Future studies should evaluate this. Being a cross-sectional study, we were unable to evaluate if vascular changes occur before structural (RNFL) changes in the development of glaucoma. Future studies should longitudinally evaluate if vascular changes on OCTA occur earlier than structural or functional changes in glaucoma. Future studies should also evaluate the utility of this new modality in detecting glaucoma progression.
In conclusion, the diagnostic abilities of peripapillary vessel density parameters of OCTA, especially the inferotemporal sector measurement, were good in POAG and PACG. Diagnostic abilities of vessel density measurements were comparable to RNFL thickness measurements both in POAG and PACG.
References
Footnotes
Contributors HLR was involved in (1) conception and design, acquisition of data, or analysis and interpretation of data; (2) drafting the article or revising it critically for important intellectual content; and (3) final approval of the version to be published. All the authors have contributed substantially to (1) conception and design, acquisition and interpretation of data; (2) revising it critically for important intellectual content; and (3) final approval of the version to be published.
Competing interests HLR is a consultant for Pfizer and Cipla, RNW is a consultant for Aerie Pharmaceuticals, Allergan, Alcon, Bausch & Lomb, Forsight Vision V, and Unity, and CABW is a consultant for Alcon, Allergan, MSD and Pfizer. RNW has received financial support in form of instruments or research funding from Topcon, Carl Zeiss, Neurovision, Optos, Heidelberg Engineering, Genentech and Quark.
Ethics approval Obtained from the Ethics Committee of Narayana Nethralaya.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- At a glance