Association between serum/plasma levels of adiponectin and obstructive sleep apnea hypopnea syndrome: a meta-analysis

The relationship between obstructive sleep apnea hypopnea syndrome (OSAHS) and a variety of disease from obesity, type 2 diabetes mellitus and cardiovascular disease has been investigated previously. Reduced adiponectin levels are also associated with increased risk of these disease. However, whether serum/plasma adiponectin levels in OSAHS patients are lower than their counterparts remain controversial. Therefore, this study evaluated the association between serum/plasma adiponectin levels and OSAHS. We performed a comprehensive literature search to locate eligible articles published on electronic databases including PubMed, EMBASE, Cochrane Library, WANFANG (Chinese database), VIP (Chinese Database) and Chinese National Knowledge Infrastructure (CNKI). The methodological quality of included studies was evaluated using the Newcastle-Ottawa scale (NOS). Pooled standard mean difference (SMD) with 95% confidence interval (CI) was calculated as effect size. Heterogeneity test was performed by Cochrane Q test and I2 test. Subgroup analysis and meta-regression analysis were employed to detect the sources of the heterogeneity. RevMan 5.3 and Stata 12.0 software were used in this meta-analysis for data synthesis. A total of 20 eligible studies with 28 databases involving 1356 participants were included in this meta-analysis. Results revealed that serum/plasma adiponectin levels in OSAHS patients were significantly lower than that in controls [SMD = − 0.71, 95% CI = − 0.92 to − 0.49, p < 0.001]. Subgroup analysis indicated that the heterogeneity would decreased when subgroup analysis was stratified by race. In addition, meta-regression analysis also suggested that the adiponectin levels were only significantly correlated with race. The removal of any independent study did not affect the pooled SMD in the sensitivity analysis. The serum/plasma adiponectin levels were significantly lower in OSAHS patients than that in control subjects, suggesting a possible role of adiponectin in OSAHS pathogenesis, deserves further studies as a potential marker of OSAHS.


Introduction
Obstructive sleep apnea hypopnea syndrome (OSAHS), the most frequent sleep-related breathing disorder, is characterized by repetitive events of partial or complete collapse of upper airway during sleep. Clinically, it is characterized by snoring, witnessed apneic episodes, marked sleep fragmentation and daytime sleepiness, which would lead to metabolic disturbance, impaired quality of life, high morbidity as well as high mortality [1]. Approximately 10% of middle-aged men and 3% of middle-aged women are estimated to have moderate-severe OSAHS in the developed world [2]. The pathogenesis responsible for the syndrome is not completely elucidated, but there exist multiple potential etiologies. One of the most important risk factors is obesity, especially visceral obesity [3,4]. In obese population, the prevalence of OSAHS reaches up to 40-50% [5]. In addition, OSAHS is associated with insulin resistance which can account for a further increase in metabolic syndrome especially of type 2 diabetes mellitus and cardio-cerebrovascular diseases [6].
Adiponectin (Acrp30) is one of the common adipocytokines largely secreted by adipocytes and has insulinsensitizing, anti-inflammatory and anti-atherosclerosis properties [7]. Reduced levels of adiponectin are commonly observed in a variety of states associated with obesity and insulin resistance, such as type 2 diabetes mellitus. However, the relationship between OSAHS and serum/plasma adiponectin levels is complex and multidirectional. The complex relationship is due in part to the fact that obesity could be a cause, consequence, or confounding factor of OSAHS. In addition, some studies revealed that adiponectin levels in OSAHS patients were lower than that in non-OSAHS group [8][9][10] [12,13].
Thus, we sought to perform a meta-analysis using all available relevant studies to assess the association between adiponectin levels and OSAHS. Given that most of the previous findings were confounded by obesity, sex and age, we only included studies that found no statistically significant difference between OSAHS patients and controls in terms of age, gender and body mass index (BMI) to address these possible confounding factors.

Methods
This meta-analysis is being reported in accordance with Preferred Reporting items for Systematic Reviews and Meta-analysis (PRISMA) statement [14].

Search strategy
We performed a comprehensive literature search to locate eligible articles published on electronic databases including PubMed, EMBASE, Cochrane Library, WANFANG (Chinese database), VIP (Chinese Database) and Chinese National Knowledge Infrastructure (CNKI). Keywords and search strategy were as follows: "obstructive sleep apnea hypopnea syndrome" or "OSAHS" or "obstructive sleep apnea" or "OSA" or "obstructive sleep apnea syndrome" or "OSAS" or "obstructive sleep hypopnea" or "sleep apnea" combined with "adiponectin" or "ADPN" or "APN". The electronic databases were searched from inception through January 2018. Besides, the references cited in these articles were reviewed to identify additional publications. We only recruited data from fully published articles written in English and Chinese.

Study selection
Two reviewers first independently reviewed the titles and abstracts to identify relevant articles. A second screening was based on full-text articles to further see whether they were eligible for inclusion. Any disagreement was resolved by discussion.

Inclusion and exclusion criteria of literature
The studies that satisfied the following criteria were included: 1) The study design was a case-control study that must have reported values in mean and standard deviation or median with range of adiponectin levels; 2) The study must have included at least two separate groups with one being a group consisting of individuals with OSAHS and the other consisting of individuals without OSAHS; 3) OSAHS was defined as apnea hypopnea index (AHI) ≥ 5; 4) All OSAHS patients were diagnosed for the first time, without receiving any form of treatment; 5) No statistically significant difference was found between OSAHS patients and controls in terms of age, gender and BMI; 6) All participants were adults (age > 18 years).
The exclusion criteria were: 1) Conference abstracts, reviews articles and case reports; 2) Original papers that did not contain precise data about serum/plasma levels of adiponectin in patients or controls; 3) Studies were not performed in humans; 4) Duplicate publication of articles.

Data extraction
The following information was recorded from each included study: first author's name, publication year, population country, total simple size, serum/plasma adiponectin levels in patients or controls, age, gender, BMI and assay approaches for adiponectin levels.

Quality assessment
The methodological quality of included studies was evaluated using the Newcastle-Ottawa scale (NOS) by two investigators independently. Discord was resolved by a third reviewer. The quality scale consists of three parts: selection, comparability and exposure assessment. The quality score ranges from 0 to 9. In our meta-analysis, we considered a study which is equal to or higher than 6 stars as a high-quality study.

Statistical analysis
Due to the inconsistency of measurement units and assay approaches, standardized mean difference (SMD) with 95% confidence intervals (CI) was chosen as effect size. Heterogeneity test was performed by Cochrane Q test and I 2 test. Generally, if P < 0.05 (Q-test) or I 2 > 50%, the heterogeneity was thought to exist and then the random-effect models would be used. Otherwise, fixed-effect models would be applied. An I 2 of 25 to 49% was considered to represent a low level of heterogeneity, 50 to 74% a moderate level, and 75 to 100% a high level. Subgroup analysis was performed to assess the impact of race, adiponectin source, assay approaches, average age, BMI and AHI. Sensitivity analysis was conducted to evaluate the stability of pooled results. Potential publication bias was assessed by using the funnel plots, Begg's rank correlation method and the Egger's linear regression method. The statistical analysis was performed with Revman 5.3 and Stata12.0 software. P < 0.05 were considered statistically significant.

Characteristics of the eligible studies
A total of 20 studies involving 1356 participants (OSAHS subjects [N = 878] and controls [N = 478]) were included in our meta-analysis. The information of the first author's name, publication year, population country, total simple size, assay approaches and NOS score of each study were showed in Table 1. The information of adiponectin levels, age, BMI and AHI are given in Table 2.

Pooled analysis
The value of I 2 was 73%, indicating that the studies were moderate heterogeneous. Therefore, the random effects model was used to combine effect size. Meta-analysis exhibited that serum/plasma adiponectin levels in OSAHS patients were significantly lower than that in controls (SMD = − 0.71, 95% CI = − 0.92 to − 0.49, p < 0.001) (Fig. 2).

Subgroup analysis
Subgroup analysis stratified by race, adiponectin source, assay approaches, average age, BMI and AHI were performed, and the results were shown in Table 3. Results exhibited that adiponectin levels were significantly lower in OSAHS patients among all subgroup. In addition, we find that I 2 would decrease when the subgroup analysis was stratified by race. Thus, race may be a potential source of heterogeneity.

Sensitivity analysis
Sensitivity analysis were performed to assess the stability of the results (Fig. 3). The removal of any independent study did not significantly change the pooled results, suggesting these results were stable (data not shown). Pooled analysis using random-effect model showed that pooled SMD was − 0.71 (95% CI: -0.92, − 0.49), P < 0.001). The fixed-effect model drew a similar result which pooled SMD was − 0.66 (95% CI: -0.77 to − 0.55, P < 0.001).

Publication Bias
The funnel plot was not completely symmetrical, suggesting that the present study has some slight publication bias (Fig. 4). However, the Begg's tests (P = 0.06) and Egger's tests (P = 0.09) did not give sufficient evidence that the present study had publication bias.

Meta-regression analysis
In meta-regression analysis, the outcome variable was the SMD of adiponectin level and the covariates  included publication year, race, adiponectin source, assay approaches, average age, BMI and AHI. We can find that the adiponectin levels were significantly correlated with race (p = 0.002), but not significantly correlated with publication year (P = 0.56), adiponectin source(P = 0.09), assay approaches(P = 0.77), average age (P = 0.27), BMI (P = 0.12) and AHI (P = 0.28).

Discussion
The current meta-analysis sought to summarize all available studies on the serum/plasma adiponectin levels among OSAHS patients and control subjects. We found that patients with OSAHS had significantly lower serum/plasma adiponectin levels compared with control subjects, indicating that adiponectin may play a role in the development of OSAHS. However, previous studies have found uncertain results for the association between adiponectin levels and OSAHS. Zhang et al. [8], Ebru et al. [16], Ozturk et al. [18] and Lacedonia et al. [23] have found OSAHS to be a potential driver of decreased adiponectin levels independent of age, gender and BMI. This was similar to our results. Hypoxia induced by OSAHS has been shown to reduce adiponectin levels via disruption of mechanisms that regulate the secretion of adiponectin. Moreover, other factors such as insulin resistance and hypoxia-induced sympathetic activation may also play significant roles in reducing adiponectin levels. While Sharma et al. in a cross-sectional [15] to determine whether obesity or OSAHS is responsible for adiponectin levels in patients with sleep disordered breathing, they found that no significant difference was noted in the OSAHS group compared to obese controls. Wolk et al. [11] reported that higher adiponectin levels in OSAHS patients compared to controls, which suggested that OSAHS may stimulate compensatory mechanisms, which can be considered to be protective of the cardiovascular system. Moderate heterogeneity was observed among these studies. Therefore, subgroup analysis and meta-regression analysis were performed to detect the potential source of heterogeneity. In subgroup analysis, only stratification by race I 2 would result in a decrease of I 2 in both white and non-white groups. In addition, meta-regression analysis also suggested that the adiponectin levels were only significantly related with race. We can speculate race may be the potential source of heterogeneity in this present meta-analysis. Therefore, the relationship between serum/ plasma adiponectin levels and OSAHS from different races requires further investigation and especially the exact composition of the non-white group.
To our knowledge, this is the first meta-analysis conducted to assess the relationship between serum/plasma adiponectin levels in OSAHS patients and in control subjects. Most of clinical research studies have focused primarily on older obese males with apnea because OSAHS is more common in males and in overweight/obese populations [31,32]. However, in these studies the findings were confounded by obesity, gender and age. To address these confounded factors, we firstly excluded the impact of age, gender and BMI on serum/plasma adiponectin levels in our meta-analysis. Hence, a major strength of our meta-analysis is that no statistically significant difference was found between OSAHS patients and controls in terms of age, gender and BMI. Another strength of this meta-analysis is the large sample size among patients with OSAHS. Moreover, none of the patients included had obvious coronary heart disease or chronic respiratory diseases. Performing a subgroup analysis and meta-regression analysis to better understand the effect of race, adiponectin source, assay approaches, average age, BMI and AHI on the serum/plasma adiponectin levels is also another strength of this study. However, several limitations should be acknowledged. Even though we used a broad search strategy, we cannot claim to have been exhaustive in retrieving all studies. In addition, some nonsignificant findings from existing studies remain unpublished, which may have some important information. We also found some conference abstracts that might have been included, but they lacked substantial details on some important data. A major limitation of the present study is a new bias may be introduced because of the confounding factor adjustment, which would affect the accuracy of the combined effect value. Finally, meta-analysis remains retrospective research, which is impossible to avoid the methodological deficiencies of the included studies. Therefore, the relationship between serum/plasma adiponectin levels and OSAHS needs to be verified given the limited number of studies.
In conclusion, the serum/plasma adiponectin levels were significantly lower in OSAHS patients than that in control subjects, suggesting a potential role of adiponectin in OSAHS pathogenesis, deserves further studies as a potential marker of OSAHS.