Introduction

Obstructive sleep apnea syndrome (OSAS) is a common chronic disorder characterized by repeated upper airway obstruction during sleep, causing periods of sympathetic activation, hypoxia, and hypercapnia. OSAS has been shown to be accompanied by a number of eye disorders, such as glaucoma (5.9–57 %) [1, 2], non-arteritic ischemic optic neuropathy (NAION, 28–71 %) [3–5], and floppy eyelid syndrome (20–32.6 %) [6]. Additionally, hypoxia may affect the endothelium and vascular resistance and further cause structural alterations of the choroid. The choroid is a highly vascularized tissue extending from the margins of the optic nerve head to the ora serrata. It plays an important role in the physiology of the eye. Nearly 90 % of the ophthalmic artery blood flow is distributed to the choroid. Besides providing oxygen and nutrients to the outer retina, the choroid regulates temperature and removes waste products from the retina [7]. It is also involved in aqueous circulation and ocular growth. Abnormal structure or function of the choroid is related to a variety of ocular and systemic diseases.

Optical coherence tomography (OCT) is a non-invasive, cross-sectional imaging method that is commonly used for retinal imaging. The newer OCT techniques have higher resolution, deeper tissue penetration, and faster acquisition speed. The emergence of enhanced depth imaging (EDI) with spectral domain OCT, polarization-sensitive OCT, and swept-source OCT enables better visualization of the choroid and evaluation of choroidal thickness in vivo. These possibilities have triggered a considerable scientific interest in the choroidal thickness in ocular and systemic diseases. Investigations focused on glaucoma and NAION have revealed reduction of choroidal thickness [8, 9].

Recently, several studies have reported measurements of the choroidal thickness with OCT in OSAS patients and normal controls. Some previous publications [10–13] reported that OSAS patients had abnormal choroidal structure alteration, while others [14, 15] did not. Results have been inconsistent and some studies are less convincing because of a small sample size. Therefore, a meta-analysis is useful to determine whether choroidal thickness (CT) is reduced in OSAS patients.

The aims of this study were to perform a meta-analysis and systemically evaluate CT measurements with OCT in a series of OSAS patients and in the normal control groups.

Materials and methods

This meta-analysis was performed according to the Cochrane Collaboration recommendations [16]. The results were reported in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) statement [17] .

Search strategy

Pertinent studies were retrieved by a systemic search of PubMed, Embase, Web of Science, and the Cochrane Library from inception to August 28, 2015. The search strategy was constructed using the following terms: obstructive sleep apnea syndrome, OSAS, choroidal thickness, and choroid. No language was restricted. Furthermore, references of reviews and identified studies were screened for potentially relevant studies.

Study selection

The following inclusion criteria were adopted in the present meta-analysis: (1) case-control or cross-sectional studies; (2) OSAS was diagnosed with international diagnostic criteria and graded according to apnea-hypopnea index (AHI). Patients with OSAS were divided into mild (5–15 events per hour), moderate (15–30), and severe (≥30); (3) patients with OSAS were compared with normal controls; (4) choroidal thickness in all subjects were obtained by OCT; (5) studies provided data of choroidal thickness of both OSAS patients and normal controls; (6) at least one of the following outcomes was used: subfoveal choroidal thickness (SFCT), choroidal thickness at 1 mm temporal to the fovea (T1), and choroidal thickness at 1 mm nasal to the fovea (N1). Exclusion criteria included the following: (1) studies without normal controls; (2) abstracts from conferences, case reports, comments, or reviews; (3) data were not retrievable; and (4) duplicate articles. In studies of the same population, only the latest or the most complete studies were included. The screening processes were performed independently by two reviewers, and discrepancies were settled by discussion.

Data extraction

Two reviewers extracted the necessary information from the eligible studies by using a standardized form. Extracted data included the following: first author, year of publication, location, OCT type, mean age, sex, sample size, body mass index (BMI), systemic diseases, AHI, spherical equivalent, axial length, intraocular pressure (IOP), the matched or adjusting factors, and choroidal thickness parameters. Discrepancies between the reviewers’ results were resolved by discussion, and a third independent reviewer was requested when there was a discrepancy noted.

Quality assessment

The methodological quality of included studies was assessed by two reviewers based on guidance in Cochrane Handbook for Systematic Reviews of Interventions [16]. It recommends the Newcastle-Ottawa Scale (NOS) system for grading the quality of case-control studies in meta-analysis. The NOS evaluates three dimensions (selection, comparability, and exposure), and the maximum scores were 4 for selection, 2 for comparability, and 3 for outcome or exposure. The overall NOS score was 9, and studies with ≥6 were regarded as having adequate quality.

Statistical analysis

The weighted mean differences (WMDs) were used for comparisons of choroidal thickness between OSAS patients and normal controls. In the present study, all summary estimates and their 95 % confidence interval (CI) were pooled by using the random effects model to obtain more conservative conclusions. Statistical heterogeneity between studies was quantified with χ 2 test and I 2 statistics. For the χ 2 test, a P value <0.10 indicates significant heterogeneity; for the I 2 statistics, values of 25, 50, and 75 % represent mild, moderate, and severe heterogeneity, respectively [18]. Sensitivity analyses were performed by assessing the influences of individual studies on the pooled results. Funnel plots and Egger’s test were used to evaluate potential publication bias. All statistical analyses were performed by Stata SE 12.0 (StataCorp, College Station, TX), and the level of significance was set at 0.05, two sided, except when otherwise specified.

Results

Eligible studies

Figure 1 shows the process of study selection. The search strategy initially identified 32 articles. Eleven reports were duplicate studies and removed. After title and abstract screening, seven reports were excluded because they did not study OSAS patients, and six did not evaluate choroidal thickness. After review of full texts, one report was excluded because the report was a letter [19]. Ultimately, seven eligible studies were included in the final meta-analysis [10–15, 20].

Fig. 1
figure 1

Literature search flow. OSAS obstructive sleep apnea syndrome

Full size image

Characteristics and quality of studies

All studies were case-control design including a total of 784 eyes (558 in the OSAS group and 226 in normal controls). Six studies [11–15, 20] were conducted in Turkey and one in China [10]. The mean age ranged from 42.3 to 58.6 years. Among the seven studies, six studies matched confounding factors in control subjects with OSAS patients: age, sex, and BMI were matched in four studies [11, 13, 15, 19]; age and IOP were matched in one study [12], and age and spherical equivalent (SE) were matched in another study [10]. Table 1 summarizes the main characteristics of the studies included. There were no significant differences in age, sex distribution, BMI, SE, axial length (AL), or IOP between patients with various degree of OSAS and normal controls. All studies set a strict inclusion and exclusion criteria, and patients with any systemic disease (such as hypertension, diabetes) that potentially affects CT were excluded. The inclusion and exclusion criteria of the included studies are summarized in Table 2. All studies had NOS scores ≥6 and the mean score was 7.3 (range 7 to 9), indicating the methodological quality was generally good.

Table 1 Characteristics of studies included in the meta-analysis
Full size table
Table 2 Inclusion and exclusion criteria of case and control in the included studies
Full size table

Data synthesis and meta-analysis

The meta-analysis of the SFCT comparison between OSAS patients and normal controls is depicted in Fig. 2. No significant heterogeneity was detected across articles comparing mild OSAS and controls (P = 0.845, I 2 = 0 %). The results showed that SFCT was reduced significantly compared with normal controls, with a pooled WMD of −8.06 (95 % CI −15.72 to −0.41) for mild OSAS, −29.99 (95 % CI −52.40 to −7.58) for moderate OSAS, and −53.72 (95 % CI −88.95 to −18.48) for severe OSAS. The SFCT reduction was more apparent as OSAS becomes more severe.

Fig. 2
figure 2

Meta-analysis of SFCT between OSAS patients and normal controls. SFCT subfoveal choroidal thickness, OSAS obstructive sleep apnea syndrome

Full size image

Three studies evaluated choroidal thickness at N1 between OSAS patients and normal controls. The meta-analysis of these data showed that there was no significant difference of N1 between mild OSAS and normal controls (WMD = −3.15, 95 % CI −17.58 to 11.28, P = 0.669). However, N1 in moderate OSAS (WMD = −19.13, 95 % CI −32.46 to −5.81, P = 0.005) and severe OSAS (WMD = −28.94, 95 % CI −43.78 to −14.10, P < 0.001) were significantly lower than the controls. The detailed results are presented in Fig. 3.

Fig. 3
figure 3

Meta-analysis of choroidal thickness at 1 mm nasal to the fovea (N1) between OSAS patients and normal controls. OSAS obstructive sleep apnea syndrome

Full size image

Analysis of T1 in three studies revealed mild between-study heterogeneity. There were no statistically significant differences in T1 between the patients with mild OSAS (WMD = 0.88, 95 % CI −14.51 to 16.27, P = 0.911) and moderate OSAS (WMD = −12.34, 95 % CI −26.00 to 1.32, P = 0.077) in comparison with controls. The detailed results are presented in Fig. 4.

Fig. 4
figure 4

Meta-analysis of choroidal thickness at 1 mm temporal to the fovea (T1) between OSAS patients and normal controls. OSAS obstructive sleep apnea syndrome

Full size image

Of the seven studies, two used Spectralis OCT (Heidelberg Engineering Inc.), three used RTVue-100 OCT (Optovue Inc.), and two used Topcan 3D-OCT 2000 (Topcon Inc.). Subgroup analyses according to the type of OCT were performed; the results showed that SFCT was thinner in patients with OSAS than in controls regardless of the type of OCT.

Five studies reported the relationship between choroidal thickness and AHI bivariate correlation analysis (Table 3). Only one study did not detect significant correlations between CT and AHI. Four studies (4/5) demonstrated a negative correlation between CT and AHI, which is consistent with the observation that CT was significantly decreased in OSAS patients compared with the normal controls.

Table 3 Summary of associations between choroidal thickness and AHI in the included studies
Full size table

Sensitivity analysis and publication bias

To further test the stability and reliability of our results, a sensitivity analysis was performed by sequentially excluding individual studies. Statistically similar results were obtained after omitting each single study, suggesting our primary results were robust. Funnel plots of SFCT were created to evaluate the potential publication bias, and the data points showed mild asymmetry, suggesting a publication bias (Fig. 5). In addition, Egger’s tests confirmed evidence of publication bias (P = 0.022 for mild OSAS vs control, P = 0.058 for moderate OSAS vs control, P = 0.038 for severe OSAS vs control).

Fig. 5
figure 5

Funnel plots for evaluating the potential publication bias: a mild OSAS vs control, b moderate OSAS vs control, and c severe OSAS vs control. OSAS obstructive sleep apnea syndrome

Full size image

Discussion

With the availability of EDI-OCT, we can image choroid non-invasively and precisely. Recent studies have demonstrated that CT measurements by OCT are highly reproducible and repeatable and have a very small range of variability [21–24]. Additionally, CT measurements obtained with different OCTs were highly correlated and could be used interchangeably [24]. It was suggested that variations in CT may indicate the presence of a disease, signal its progression, or provide insights into its prognosis. For example, CT was reported to decrease in NAION, glaucoma, and retinal vein occlusion, while it increased in carotid cavernous fistula, migraine, and central serous chorioretinopathy [25]. However, whether or not alternation of CT occurs in OSAS remains a subject of debate.

With the accumulation of a number of OCT studies, there is an opportunity to analyze the available data to resolve persistent clinical questions [18]. Based upon the published data from seven studies, our meta-analysis investigated how CT changed in different stages of OSAS. Our meta-analysis confirmed that CT changes can be detected using OCT in patients with OSAS. The pooled results demonstrated a significant reduction in SFCT in mild, moderate, and severe OSAS patients compared with normal controls. Further analysis in location at N1/T1 provided evidence that CT in severe or moderate OSAS was thinner than that in controls, while CT in mild OSAS was similar to that in controls. Thus, reduced CT could be an ocular marker of the severity of OSAS.

Reduction of CT in OSAS is biologically reasonable, though an exact mechanism remains unclear. It has been speculated that CT may be an indirect biomarker of choroidal circulation. The smooth muscles of the choroid vessel walls are innervated by sympathetic and parasympathetic nerves. Choroidal blood flow has been shown to have a capacity to autoregulate in order to maintain adequate oxygen saturation. Sogawa et al. [26] suggested that there were no significant correlations between the SFCT and the choroidal blood flow in healthy young subjects. But, Okamoto et al. [27] demonstrated that the reduction of SFCT after intravitreal bevacizumab injection in the eyes with central serous chorioretinopathy was caused by the reduction of choroidal blood flow. And several studies [28–30] have found that sildenafil citrate increases CT because of increased choroidal blood flow. Analysis of Doppler ultrasound variables and SFCT revealed an association between SFCT and internal carotid artery diameter and resistance index [31]. In OSAS, intermittent airway obstruction causes hypoxia and decreases the mean saturation of oxygen. Recurrent hypoxia and reperfusion episodes can lead to oxidative stress and inflammation, damage the vascular endothelium, decrease responsiveness to vasodilator agents, and activate the sympathetic system [32]. These serious conditions lead to vascular dysregulation, altered choroidal blood flow, and consequently reduced CT.

In this study, we did not find significant differences in T1/N1 between mild OSAS and controls. This may be explained by the autoregulation mechanism of the choroid. The sympathetic and parasympathetic fibers terminating on the choroid may remain balanced in some patients with mild OSAS. These results were consistent with the findings by Khayi et al. [33] that the regulation of choroidal blood flow was unaltered in men with OSAS who have no comorbidities. Tonini et al. [34] also reported that choroidal vascular reactivity to hyperoxia and hypercapnia in OSAS was similar to that in healthy subjects. The results of our study suggest that the choroidal blood flow is preserved in mild OSAS. On the other hand, patients with moderate or severe OSAS may have dominant sympathetic activity and decreased choroidal blood flow. More studies evaluating CT and choroidal blood flow in OSAS patients may be required to confirm this theory.

Difficulties in the meta-analysis are the heterogeneity of the existing studies and confounders. An important confounder in OCT measurement includes co-existing ocular conditions such as refractive error and glaucoma. Refractive error contributes to different retinal nerve fiber layer thickness and CT. The differences in IOP and SE between OSAS and controls were not statistically significant in four of seven studies, and the pooled measurements were similar when the combined data was restricted to these four studies. Hence, the results would not be biased by the IOP or SE. However, the potential effect of other confounding factors may have been present, as it is uncertain whether or not eyes with NAION, CCF, migraine, or retinal vein occlusion were truly excluded by these studies.

This meta-analysis has some limitations. First, CTs in the included studies were all obtained manually, a fact which may have introduced measurement bias. Although this is still the most commonly used method for determining CT, automated software will be required to obtain a more objective evaluation [35]. Second, the reported changes in CT as measured by the EDI-OCT are assumed to be generalized. It is possible that this might not be the case and that segmental changes in CT occurred and were overlooked. Third, this meta-analysis included studies published in indexed journals. We did not search for unpublished studies or original data. The funnel plot for publication bias was indeed significant, although this result may also indicate small-study effects. Further study may be warranted to analyze this. Fourth, OSAS duration, which has been shown to affect vascular autoregulation, may influence CT. Unfortunately, a subgroup analysis for this was unavailable because of lack of data. Finally, the included studies were performed in Turkey or China. The results of this study may not be generalizable to other populations. More studies with larger sample sizes and various ethnicities are needed.

In conclusion, the present meta-analysis showed that CT was significantly decreased in OSAS patients compared with the normal controls. The tendency of reduction of CT in severe OSAS was more obvious. OCT could be a useful tool for grading the severity of OSAS. These findings may be helpful for explaining the pathogenesis of OSAS and the mechanisms connecting OSAS and ocular diseases.