Optical Coherence Tomography Angiography Assessment of the Peripapillary Vessel Density and Structure in Patients with Nonarteritic Anterior Ischemic Optic Neuropathy: A Meta-Analysis

Background This meta-analysis is aimed at assessing the peripapillary vessel density (VD) and structural outcomes using optical coherence tomography angiography (OCTA) in patients with nonarteritic anterior ischemic optic neuropathy (NAION). Methods A comprehensive search of PubMed, Embase, Cochrane Library, and Web of Science databases for literature comparing VD and structural outcomes in patients with NAION and controls was performed. Mean differences (MDs) and its 95% confidence interval (CI) were calculated for continuous estimates. Review Manager (V5.30) was used for analysis. Results Fourteen published studies met the requirement. The radial peripapillary capillary (RPC) whole enface VD measured by OCTA was significantly lower in patients with NAION compared to that of the controls (MD = −10.51, P < 0.00001). The RPC inside disc VD was significantly decreased in the NAION group than that in the control group (MD = −8.47, P < 0.00001). For RPC peripapillary VD, there was a statistically significant difference between patients with NAION and the controls (MD = −12.48, P < 0.00001). The peripapillary retinal nerve fibre layer (p-RNFL) thickness was significantly lower in patients with NAION in comparison to the controls (MD = −22.18, P = 0.004). The ganglion cell complex (GCC) thickness in the macular zone of NAION patients was remarkably reduced compared to that in the controls (MD = −17.18, P = 0.0002). Conclusions The findings suggested that the peripapillary VD and RNFL thickness were attenuated, and the macular GCC thickness was reduced in patients with NAION. OCTA, in the future, may facilitate the diagnosis and monitoring of patients with NAION.


Introduction
Nonarteritic anterior ischemic optic neuropathy (NAION) is a visually threatening optic neuropathy characterised by sudden and rapidly painless vision loss, altitudinal visual field defect, and optic disc oedema [1]. NAION is estimated to occur in 2.3-10.3 per 100,000 individuals in the middleaged and elderly population [2,3]. Although the underlying mechanism of NAION remains unknown, the available evi-dence reveals that it may be associated with perfusion deficiency of the optic nerve head (ONH) microcirculation that is predominantly supplied by the short posterior ciliary arteries [4][5][6]. The probable risk factors that increase the onset of NAION include hypertension, hypercholesterolemia, diabetes mellitus, nocturnal hypotension, and obstructive sleep apnoea [7][8][9][10].
Currently, several imaging modalities have been employed to detect optic disc microvasculature in NAION [11]. Among these imaging techniques, fundus fluorescein angiography (FFA) is currently the standard procedure to recognize the impaired ONH blood flow in NAION [12]. However, FFA is an invasive procedure and does not quantitatively assess optic disc vascular structures at different layers [13]. As for laser Doppler velocimetry, delayed ONH blood flow has been found in patients with NAION; however, it cannot provide quantitative assessment of the vasculature in the ONH [14]. Optical coherence tomography angiography (OCTA) is a new, noninvasive imaging technology that can quantify the blood flow within the retina and choroid without needing fluorescein dye injection [15]. Several studies conducted by OCTA have demonstrated decreased vessel densities in the ONH and radial peripapillary capillary (RPC) in the eyes with NAION [16][17][18]. Prior studies also revealed that the reduced vessel density of the peripapillary retina was significantly correlated with the peripapillary retinal nerve fibre layer (p-RNFL) thickness and visual field mean defect [19,20]. However, to our knowledge, there has been no meta-analysis comprehensively investigating the peripapillary vessel density and structural features related to patients with NAION.
Therefore, the current meta-analysis was carried out to address this issue and provide robust evidence for ophthalmologists to properly manage patients with NAION.

Search Strategy.
This meta-analysis was performed following the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [21], and no ethical approval was needed. Two independent investigators comprehensively searched the electronic databases including PubMed, Embase, Cochrane Library, and Web of Science from inception through 31 May 2020 to identify relevant literature. To achieve maximum numbers of articles, the following search terms were used: ((((OCTA) OR (OCT angiography)) OR (optical coherence tomographic angiography)) OR (optical coherence tomography angiography)) AND (((Non-arteritic Anterior Ischemic Optic Neuropathy) OR (non-arteritic anterior ischemic optic neuropathy)) OR (NAION)). English-language articles were regarded as eligible. All discrepancies were resolved by deliberation with each other or discussion with the third author.

Study Selection.
Studies were included in this metaanalysis if they met the following criteria: (1) they are original articles; (2) they are studies comparing patients with NAION (including acute and nonacute (equivalent to chronic) NAION) with healthy controls; (3) OCTA data were provided as mean ± standard deviation (SD); and (4) primary outcomes included radial peripapillary capillary (RPC) whole enface vessel density (VD), RPC inside disc VD, RPC peripapillary VD, p-RNFL thickness, and macular ganglion cell complex (GCC) thickness.
The exclusion criteria were as follows: (1) case series, abstracts, posters, animal studies, reviews, comments, and meta-analyses; (2) study objective did not meet the inclusion criteria; (3) duplicate publication from the same study; (4) study outcomes could not be extracted; and (5) study results were unclearly reported.

Data Extraction and Quality
Assessment. Two independent authors (Ling Ling and Kaibao Ji) independently extracted the data from the included studies, and disagreements were resolved by discussion with the third author. The following data were collected from the included articles: first author, location, publication year, study design, sample size, mean age, sex ratio, type of OCTA, primary outcomes, and quality of a study. The quality assessment of the included studies was performed according to the Newcastle-Ottawa scale, which provided a score range of 0 to 9 points, with a higher score (≥5) indicating better quality [22].

Statistical
Analysis. The Review Manager Software Version 5.30 (Cochrane Collaboration, Oxford, UK) was used for analysis. Mean differences (MDs) and its 95% confidence interval (CI) were calculated for continuous estimates. Heterogeneity among studies was performed using the chisquare statistic test and I 2 statistic test. I 2 values of 25%, 50%, and 75% represented mild, moderate, and high heterogeneity, respectively. A fixed-effect model was employed when no significant heterogeneity existed among studies; otherwise, the random-effect model was used. Potential publication bias was assessed by the funnel plot. P < 0:05 was considered a significant difference.

Results
3.1. Literature Search Results. The selection process of literature retrieval and screening is presented in Figure 1. A total of 265 potential articles were initially searched from the databases (PubMed: 110, Web of Science: 50, Embase: 105, and Cochrane Library: 0), of which 87 duplicated articles were excluded. In addition, 154 articles were excluded after reviewing the titles and abstracts. The residual 24 articles were in detail screened for the full text; three studies had unqualified data, three did not meet the inclusion criteria, and for the other four studies, data could not be extracted. Finally, fourteen articles [19,20,[23][24][25][26][27][28][29][30][31][32][33][34], with a total of 783 eyes (313 in the NAION group and 470 in the control group), were included in the meta-analysis.
The basic characteristics of the included articles are summarized in Table 1, and the quality assessment results of the included studies are reported in Table 2.

Publication Bias.
Funnel plots summarized the potential publication bias of RPC whole enface VD, RPC inside disc VD, RPC peripapillary VD, and macular GCC thickness among the studies. The results revealed that the distribution of studies was not an obvious asymmetry, indicating the absence of significant publication bias (Figures 12-15).

Discussion
To the best of our knowledge, this is the first meta-analysis to investigate the peripapillary microvascular and structural changes using OCTA in patients with NAION and control subjects. Fourteen eligible studies including 783 eyes were analysed in this meta-analysis. In the review, we pooled the mean RPC whole enface, RPC inside disc, and RPC peripapillary vessel densities of study participants, as well as their p-RNFL and macular GCC thicknesses. Our data figured out that there were significantly reduced peripapillary vessel density, p-RNFL thickness, and macular GCC thickness in patients with NAION compared to that of control subjects. And the funnel plots suggested no obvious publication bias.
Recently, some studies have used OCTA to assess the peripapillary microvascular in NAION eyes, suggesting reduced vessel densities of the RPC in the eyes with NAION, as well as reduced p-RNFL thickness [16][17][18][19]. Our current evidences reinforced the findings of the previous studies [35][36][37]. However, substantial heterogeneity was shown in our meta-analysis. Three studies contributed to the major     [32], and Song et al. [34]. The I 2 measurement for this item significantly declined from 93% to 44% (chi 2 = 17:97, P = 0:06), after the removal of the above three studies. In the acute-NAION subgroup, the heterogeneity also remarkably decreased from 96% to 46% (chi 2 = 5:58, P = 0:13), when the two studies were excluded [23,32]. Although these three studies were excluded from the analysis, the MDs for all were statistically significant. We also found that the other two studies [19,23] contributed maximum heterogeneity to the RPC inside disc VD, revealing that heterogeneity considerably decreased from 89% to 23% (chi 2 = 2:60, P = 0:27), when these two studies were excluded. The acute-NAION subgroup heterogeneity also subsequently decreased from 89% to 48% (chi 2 = 1:93, P = 0:16), after the study was removed [23]. There was only one study that reported RPC inside disc VD in the nonacute NAION subgroup; therefore, no heterogeneity was observed. The MDs were still significant after the removal of these two studies-Fard et al. [19] and Pierro et al. [32], but the sample numbers were relatively small within the groups. As to RPC peripapillary VD, two studies contributed majorly to the heterogeneity of this analysis-Mastropasqua et al. [20] and Song et al. [34]. The I 2 measurement for this parameter significantly declined from 85% to 34% (chi 2 = 6:10, P = 0:19) after removing the above two studies, subsequently resulting in decreased heterogeneity from 89% to 43% (chi 2 = 5:30, P = 0:15) in the nonacute NAION subgroup. No heterogeneity was found in the acute NAION subgroup, as merely one study reported the data. The MD estimates of this parameter were still significantly different after the two studies were excluded from analysis, suggesting results were robust.
There are several potential limitations to be considered in our meta-analysis. Firstly, the sample size included in the study was comparatively small and the quality of the evidence was relatively low. Secondly, the pooled estimates should be interpreted carefully, as high heterogeneity existed in this meta-analysis. Thirdly, the source of heterogeneity could not be fully determined as a result of insufficient data to perform comprehensive metaregression. Finally, although we did not register the protocol of our study in the PROS-PERO database, no corresponding protocols of this subject were found in the database. Future prospective and largescale cohort studies or intervention trials should be carried out to validate our results.
In conclusion, our findings showed that the peripapillary vessel density and RNFL thickness were attenuated in patients with NAION when compared with controls. Furthermore, we revealed that the macular GCC thickness was also reduced in patients with NAION. OCTA, in the future, may become a useful technique in the diagnosis and monitoring of patients with NAION.

Data Availability
All data are fully available without any restrictions.

Conflicts of Interest
The authors have no competing interests.