Risk Factors for Non-arteritic Anterior Ischemic Optic Neuropathy: A Large Scale Meta-Analysis

Objective: We conducted a meta-analysis to explore all the potential risk factors for non-arteritic anterior ischemic optic neuropathy (NAION) based on the published literature. Methods: A comprehensive literature search through the online databases was performed to obtain studies concerning the risk factors of NAION up to June 2020. Pooled unadjusted odds ratios (ORs) or rate ratios (RRs) were calculated to evaluate the weight of risk factors. This study was registered in PROSPERO under the number CRD42018084960. Results: Our meta-analysis included 49 original studies comprising of more than 10 million patients. The following risk factors were proved to be significantly associated with NAION: male gender (OR = 1.67, 95% CI: 1.50–1.85, P < 0.00001), hypertension (RR = 1.28, 95% CI: 1.20–1.37, P < 0.00001), hyperlipidemia (RR = 1.43, 95% CI: 1.26–1.62, P < 0.00001), diabetes mellitus (DM) (RR = 1.53, 95% CI: 1.36–1.73, P < 0.00001), coronary heart disease (CHD) (RR = 1.68, 95% CI: 1.24–2.27, P = 0.0008), sleep apnea (RR = 3.28, 95% CI: 2.08–5.17, P < 0.00001), factor V Leiden heterozygous (RR = 2.21, 95% CI: 1.19–4.09, P = 0.01), and medication history of cardiovascular drugs. Conclusion: We concluded that the above risk factors were significantly related to NAION. Better understanding of these risk factors in NAION can help the direct therapeutic approaches.


INTRODUCTION
Non-arteritic anterior ischemic optic neuropathy (NAION) is the most common ischemic optic neuropathy. The incidence rate is 2.5-11.8 per 100,000 cases in men elder than 50 (1). Characterized by optic nerve ischemia mostly due to hypoperfusion of short posterior ciliary arteries (SPCAs) (2), NAION can lead to unilateral, sudden, and painless loss of vision among awake patients. Segmental or diffuse optic disc edema can be observed without evidence of arteritis (3). Although the detailed pathogenesis is unclear, NAION is probably related to systematic hypoperfusion, nocturnal hypotension, local autoregulation failure, and hypercoagulation (2). NAION is a naturally progressive disease and the contralateral eye involvement rate is 15-20% in the following 5 years (4). It has been proved that the medications, including corticosteroids, aspirin, and neurotrophic drugs, have limited and controversial efficacies (5,6). The risk factors should be taken into thorough consideration when providing precautions and treatments for NAION.
The two published meta-analyses reported influences of DM or sleep apnea on NAION, respectively (7,14). However, impacts of other factors differed in literature and no cumulative conclusions were reached. Therefore, we decided to perform a large-scale systematic review and meta-analysis on all the possible NAION risk factors identified in the published studies. To the best of our knowledge, this is the first meta-analysis concentrating on multiple factors of NAION. We expect to help the clinicians comprehensively understand the risk factors of NAION and provide more evidence for the preventions and treatments.

METHODS
We conducted this systematic review and meta-analysis in accordance with the meta-analysis of observational studies in epidemiology (MOOSE) guidelines. The protocol registration number of PROSPERO (https://www.crd.york.ac.uk/prospero/) was CRD42018084960.

Search Strategy and Study Selection
Original literature was searched comprehensively through electronic Pubmed, Medline, Embase, and Cochrane Library databases. The related references were also screened, including gray literature (in the website http://graylit.osti.gov/). The language of included studies was restricted to English. The last search was on June 6, 2020. The search terms were applied as follows: "non-arteritic anterior ischemic optic neuropathy" OR "non-arteritic anterior ischaemic optic neuropathy" OR "NAION" OR "NA-AION" in combination with "risk" OR "factor" OR "risk factor." These terms were searched in all the fields of articles, not restricted to their abstracts.
The inclusion criteria were listed as follows. First, the clinical studies concerning the comparisons of risk factors between NAION and controls were taken into further consideration. Second, the risk factors should exist before the diagnosis of NAION, which was judged by carefully screening the abstracts and/or full texts. Third, the samples of case and control groups were provided directly, or odds ratios (ORs) or rate ratios (RRs) of risk factors were reported with 95% CIs. Accordingly, we excluded the animal experiments, case reports/series, abstracts, conference proceedings, repeated publications, non-published materials, reviews, and editorials.
Two investigators (B. L. and Y. Y.) did the literature search, study screening, data extraction, and eligible study quality assessment independently. The inconsistency was resolved by a third reviewer or via an open discussion.

Data Extraction and Study Quality Assessment
We collected the following data in a prepared standard form: first author, year of publication, country, ethnicity, study design, study duration, sample size, baseline information of the patient, and the number of patients (ORs or RRs with 95% CIs) in both the NAION and the control groups. The Newcastle-Ottawa Scale (NOS) (15) was applied for quality assessment of the nonrandomized controlled trials (RCTs). The studies achieving seven or more stars were regarded as high quality.

Statistics Analysis
We calculated the pooled unadjusted ORs or RRs on dichotomous variables to identify the association between the risk factors and NAION. Accordingly, the mean differences (MDs) were used on the continuous variables. Heterogeneity was determined through the chi-square test based on Q and I 2 values (16). No significant heterogeneity existed if the p-value was >0.10, and in this condition, the fixed-effect model was used. On the contrary, the random-effect model was applied. We also conducted the subgroup analyses according to different populations. The results were significant in our meta-analysis if a two-sided p-value was <0.05. Inverted funnel plot visual inspection was to assess the publication bias for all the comparisons, and the Egger's test was added when the number of studies was more than 10. The data analyses were performed in software RevMan (version 5.3; Cochrane Collaboration, Oxford, UK) and STATA (version 13.0; StataCorp, College Station, TX, USA).

DISCUSSION
In our systematic review and meta-analysis, we included articles studying a variety of risk factors: age, gender, ethnicity, systematic diseases, ocular factors, genotypes, cardiovascular drugs, and so on. We finally concluded the following risk factors to be significantly associated with NAION: male gender, hypertension, hyperlipidemia, DM, CHD, sleep apnea, medication history of cardiovascular drugs, and factor V Leiden heterozygous. Some other systematic and ocular diseases were researched in <3 studies and did not seem to be significant risk factors. In the subgroup analyses based on ethnicities, we found that the influences of gender in Asians and CHD and sleep apnea in Europeans were not significant. Therefore, we should take notice when applying our results to different populations.
In our meta-analysis, the cardiovascular factors were highly associated with NAION and studied in most literature, including hypertension, hyperlipidemia, and DM. NAION probably results from topical and/or systematic hypoperfusion (2). Although this is not a thrombotic event, many predisposing factors related to thrombogenesis or hypercoagulative state can disturb the systematic blood circulation via different pathways (41). For example, hyperlipidemia is harmful to endothelial cells and accelerates the formation of atherosclerotic plaques, further leading to hypertension and CHD (27). Some biochemical markers (such as hyperhomocysteinemia) and genetic polymorphisms (such as factor V Leiden heterozygous) indicating hypercoagulative state were also significant risk factors in our meta-analysis. Therefore, the above factors are cofactors of NAION with similar mechanisms. Drugs treating cardiovascular diseases were proved to be significantly associated with NAION, too, such as antithrombotics, β-blockers, and statins. These drugs could not be considered as independent risk factors because they were only applied to treat the diseases.
Although the above major factors were reported to induce NAION, we summarized and reconfirmed these conclusions. In addition, we included the hypercoagulative biomarkers and possible risky genetic polymorphisms from the published literature. We first performed the meta-analyses on these factors, demonstrating that increased homocysteinemia, fibrinogen, lipoprotein(a), and factor V Leiden heterozygous were risk factors of NAION. Several diseases and biomarkers were identified in <3 studies, which are listed in Table 3. They might be the potential risk factors if more original studies were carried out. Specifically, we conducted the subgroup analyses based on ethnicities in seven risk factors. The cases were divided into Asians, Europeans, and mixed ethnicities from original publications. The ethnicity did not affect the association between smoking, hypertension, hyperlipidemia, DM, and NAION. Nevertheless, the incidence of NAION had no disparity in different gender among the Asians. CHD and sleep apnea were not significant risk factors in Europeans. This subgroup analysis expanded the application of our results to different peoples.
Diabetes and sleep apnea were proved to be important risk factors in the previously published meta-analyses (7,14). We reconducted meta-analyses on both factors with 11 new   studies included on diabetes and three on sleep apnea. Apart from the verification of their conclusions with larger sample sizes, we also performed the subgroup analyses according to ethnicities as stated above. For DM, its influence on NAION was not related to ethnicity in our pooled results, similar to the conclusions of Chen et al. (14). Furthermore, our conclusions were more powerful with three extra cohort studies compared with the previous meta-analysis (14), in which the included studies were all case-controlled because the cohort studies had a higher level of evidence. Sleep apnea was not a significant risk factor of NAION in the Europeans after our subgroup analysis, although a published study found that the Europeans were more likely to have NAION (14). However, only three studies were included in this subgroup, and we achieved marginal data (P = 0.06). We expected more valid original studies on different ethnicities, in case, a high-quality meta-analysis could be conducted.
We found no apparent association between the occurrence of NAION and 1-month use of PDE5-Is, which was published by us in 2018 (8). Although PDE5-Is mainly cause vasodilation and systematic hypotension (32), their influences on the NAION pathogenesis remain controversial. Because of the relatively low incidence of NAION and difficulties in diagnosis, it was hard to include adequate samples in the published literature, and confounders were not adjusted in several case-control studies. More clinical studies are necessarily needed to provide strong evidence on this point.
A crowded optic disc was often observed in NAION because hypoperfusion or ischemia of the optic nerve head is apparent in a tight optic disc structure (26). An optical coherence tomography (OCT) is a useful tool to measure the CDR. Several studies showed the association between the CDR and NAION. For example, González Martín-Moro et al. reported a smaller CDR to be a risk factor and poor prognostic marker  (50). Both pathogens disturb the function of endothelial cells, activate the secretion of inflammatory cytokines, and thus induce or promote atherosclerosis. In the cohort study by Chang et al., endstage renal disease was proved to increase the risk of NAION (31), probably because these patients had received repeated  (64), no consistent conclusions were reached due to the lack of well-planned studies with large sample sizes. All the above research were carried out in a small number of studies, so the studies with larger samples size are necessary to confirm these findings. Although our meta-analysis was conducted comprehensively and included multiple potential risk factors of NAION, it still has several limitations. First, all the included original articles were case-control or retrospective cohort studies. No RCTs or prospective cohort studies have been published yet. Recalled data might be incomplete and inaccurate, bringing biases to these studies. Second, most NAION possible factors have similar mechanisms and act as confounders. Independence of these factors may probably be examined by RCTs; however, no RCTs can be conducted on studying the risk factors. Third, the time span of the included studies lasts from 1991 to 2019, during which the changes in lifestyles and medical techniques influence the type of risk factors. Finally, some risk factors were examined in <3 studies and it was hard to confirm their association with NAION. The above limitations reduced the quality of our meta-analysis partly and restricted our results to be applied.
Consequently, our study concluded that the following risk factors were associated with NAION: male gender, hypertension, hyperlipidemia, DM, CHD, sleep apnea, medication history of cardiovascular drugs, and factor V Leiden heterozygous. Better understanding of these risk factors in NAION can direct future research and therapeutic approaches.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.