Background
Obstructive sleep apnea (OSA) refers to a sleep-related disorder characterized by repetitive, incomplete or total obstruction of the upper respiratory tract combined with hypopnea and apnea during sleep, contributing to an intermittent decrease in the partial pressure of blood oxygen and blood oxygen saturation and hypercapnia [
1]. It is usually correlated with sleepiness, fatigue, inattention, memory loss, or even headaches during the day, seriously affecting quality of life and life expectancy [
2]. Currently, polysomnography (PSG) is the gold standard for the diagnosis of OSA [
3]. Patients with OSA have a greater risk of cardiovascular, cerebral, pulmonary vascular complications and even systemic multisystem pathophysiological changes [
4,
5]. As an independent risk factor for many systemic diseases, OSA is currently remarkably prevalent in Western countries; therefore, some countries have listed it as a major health problem. The prevalence rate of OSA in Western countries is 2%–5% and is gradually increasing annually; the rate reached 43.1% in Iceland in 2016 among those aged 40 to 65 [
6]. The incidence rate of OSA in adults in China is approximately 4%, according to an epidemiological survey [
7].
It is well known that vitamin D is a liposoluble vitamin present in two primary patterns: D2 (ergocalciferol, gained from dietary resources) and D3 (cholecalciferol, generated in the skin under irradiation by ultraviolet light, or 25(OH)D) [
8]. The main test to measure the serum vitamin D level is enzyme-linked immunosorbent assay (ELISA). Mass spectrometry (MS) is commonly used as well.
The relationship between 25-hydroxyvitamin D (25(OH)D) and OSA has been evaluated in several studies, including randomized controlled trials and observational studies [
9‐
11]. Kerley et al. [
9] discovered that 25(OH)D was markedly lower in OSA groups than in non-OSA groups, particularly in Caucasian populations. Mete et al. [
10] declared that those with severe OSA had an elevated exposure to vitamin D deficiency. However, Yassa et al. [
11] held that the severity of OSA was not related to serum vitamin D and that the vitamin D status did not alter the severity of OSA. While changes in the serum vitamin D level in patients with OSA may be relevant to inflammatory reactions, oxidative stress, energy metabolism, neuroendocrine regulation and so on, its specific pathogenesis is poorly understood.
Until now, the viewpoint that OSA patients have lower serum vitamin D has remained controversial. Thus, a meta-analysis in 2018 [
12] comparing the 25(OH)D serum level between controls and OSA patients was conducted. Whereas the main locations of the included studies are developed countries, such as Europe and North America; additionally, a former study did not compare the 25(OH)D serum level according to OSA severity. Therefore, this conclusion remains to be verified in observational studies. Recently, continuous research has been carried out to confirm whether low serum levels of vitamin D are pervasive in those with OSA, especially those with serious OSA. New research on OSA in Asians has emerged, but the results are contradictory. Thus, the objective of our updated meta-analysis is to include the latest observational studies in determining whether different types of OSA patients exhibit low serum vitamin D.
Materials and methods
Search strategy
Our meta-analysis was registered at the prospective register of systematic reviews (PROSPERO, the website is
https://www.crd.york.ac.uk/PROSPERO/, and the ID is CRD42020172659). The following eight electronic databases were searched individually from January 2000 to August 2020: Embase, Medline, Web of Science, PubMed, VIP, Wanfang, CNKI and SinoMed. The search terms were [(“vitamin D”) or (“vit D”) or (“ergocalciferol”) or (“cholecalciferol”)] and [(“sleep”) or (“obstructive sleep apnea”) or (“sleep apnea syndromes”) or (“obstructive sleep apnea syndromes”)].
Study selection and exclusion criteria
The literature included met the following standards: (1) any observational study design, including case–control, cross-sectional and cohort studies; (2) literature reporting on the vitamin D level in OSA patients without restrictions on age, ethnicity, nationality or sex; (3) OSA diagnosed by polysomnography (PSG); and (4) search limited to full articles in English and Chinese. Studies meeting the following criteria were excluded: (1) the type of research was a case report, review, letter, commentary or editorial; and (2) the article did not contain adequate data, and it was difficult to contact the authors to obtain valid sources.
Study selection
The title and abstract of internationally relevant studies were screened by two review authors individually on the basis of our search strategy. The full text of the study was attained and confirmed if the abstract was eligible. Then, we reappraised potentially suitable studies by retrieving and assessing the full text. If there was conflict about the inclusion or exclusion of a study, the disagreement between reviewers was settled by analysis or consultation with the third reviewer.
Data extraction and management
Two reviewers (X.Y.L. and J.H.) obtained the data independently. A designed table was used to extract data from each included paper, as follows (1) basic information of the paper: study design, country, year of publication, first author, etc.; (2) baseline characteristics of the participants: age, number and sex; (3) OSA measurement method and its type; and (4) study quality. If a study eligible for our meta-analysis lacked essential content, the authors were contacted by telephone or e-mail no less than twice. Then, authors were requested to supply the missing data.
Methodological quality assessment
The methodological quality evaluation of all studies was conducted by two researchers independently. If a disagreement occurred in the process of the quality evaluation, it was negotiated by two reviewers or arbitrated by a specialist in the field. Cohort and case–control studies were evaluated through the Newcastle–Ottawa Scale (NOS) [
13], which included study population selection (4 items, full score 4), comparability (1 item, full score 2), and exposure or outcome (3 items, full score 3); scores of 0–3, 4–6 and 7–9 were considered to indicate low- (grade C), medium- (grade B) and high-quality (grade A) studies, respectively. Cross-sectional studies were evaluated using the quality evaluation criteria recommended by the Agency for Healthcare Research and Quality (AHRQ) [
14], which includes 11 items. Each item was rated as "yes", "no", or "unclear", with a score of 1 for "yes" and 0 for "no" or "unclear". Scores of 0–3, 4–7 and 8–11 were considered to indicate low- (grade C), medium- (grade B) and high-quality (grade A) studies, respectively.
Statistical analysis
A meta-analysis was performed to assess the summarized results of the studies with RevMan 5.3 software. Binary variables are expressed by odds ratios (ORs) and 95% confidence intervals (CIs), and continuous variables are expressed by standard mean differences (SMDs) and 95% CIs. Chi-square tests of Cochrane's Q statistic and I-squared were used to analyze the heterogeneity of the results, with heterogeneity thresholds of low (25%), moderate (50%), and high (75%). If P > 0.1 and I2 < 50% indicated homogeneity among studies, a fixed-effects model was adopted; if P ≤ 0.1 and I2 ≥ 50% revealed heterogeneity, a random-effects model was used. Obvious heterogeneity was analyzed by subgroup analysis or meta-regression; if more than one article existed in the subgroup, the analysis was performed by sex, country, number of samples, study design, study quality and severity of OSA (the apnea–hypopnea index (AHI) was applied in the classification; an AHI of 5–14.9, 15–29.9 and ≥ 30 events/hour was considered to indicate mild, moderate and severe OSA, respectively). Publication bias was examined in testing the results of the quantitative evaluation.
Discussion
Vitamin D deficiency has been noted in many metabolic diseases, and it is thought to be related to OSA progression [
21]. The present meta-analysis assessed the serum 25(OH)D level in OSA patients. The combined results show that the serum 25(OH)D level was noticeably lower in OSA patients than in the controls. Moreover, with increasing OSA severity, the serum 25(OH)D level decreased more obviously, suggesting that serum 25(OH)D might be a risk factor for OSA. Quite notably, the decrease in the serum 25(OH)D level was more obvious in Asians than in Caucasians. This phenomenon suggests that ethnicity may also affect the serum 25(OH)D level in OSA patients. Differences in the level of serum 25(OH)D among different ethnicities might be related to polymorphisms in vitamin D receptor genes and metabolic genes. Kirac et al. [
36] found that VDR (rs2228570) and VDBP (rs4588 and rs7041) mutations were highly related to OSA in Caucasian populations. Nie et al. [
41] indicated that OSA was associated with the angiotensin-converting enzyme gene, which plays a crucial role in vitamin D metabolism in Asian populations. Moreover, the serum 25(OH)D level decreased with increasing OSA severity. Both overweight OSA patients and obese OSA patients had low levels of vitamin D, indicating that BMI and OSA interact to influence the vitamin D level. A high BMI and OSA are causally interrelated. One study confirmed that low circulating levels of vitamin D were associated with obesity in humans, and obesity was thought to be one of the reasons for the reduction in the 25(OH)D level [
42]. Furthermore, subgroup analyses were conducted according to the PSG type, study quality and latitude, and the serum 25(OH)D level was still lower in the OSA patients. The degree of decline in the serum 25(OH)D level varied based on the above three factors. Of course, these differences may have been caused by individual differences among OSA patients and various detection instruments. However, overall, the results of the meta-analysis are genuine and reliable.
The underlying mechanism of the relationship between the serum 25(OH)D concentration and OSA remains unclear [
11]. According to the source, metabolism and various influencing factors of vitamin D [
26,
43,
44], we have summarized several seemingly reasonable biological explanations. Previous studies have proven that there is a significant correlation between the vitamin D level and obesity, and most OSA patients are obese [
21]. Most OSA patients included in our study had a BMI greater than 30, which meets the diagnostic criteria of obesity. Some obese patients usually do not like outdoor activities. Lack of outdoor activity might result in reduced vitamin D synthesis due to insufficient sun exposure. In addition, vitamin D, as a fat-soluble vitamin, is stored by adipose tissue [
45]. In obese patients, who have greater amounts of adipose tissue, the storage distribution volume of vitamin D is also remarkably increased [
45]. This effect of increased vitamin D storage in adipose tissue reduces the release of vitamin D into the circulation, resulting in a lower bioavailability of vitamin D. Fan et al. [
34] reported that vitamin D was negatively correlated with BMI and that there was an interaction between the vitamin D level and obesity. Moreover, 1 h of sleep disorder can reduce daytime activity by 3%, and with the loss of 1 h of sleep, the probability of obesity increases by 80%. Due to lack of sleep or poor sleep quality, OSA aggravates obesity, forming a vicious circle. Consequently, the vitamin D level becomes significantly lower in obese OSA patients.
Many factors can contribute to vitamin D deficiency. In addition to obesity, geographical location and solar exposure also have influences [
46]. Theoretically speaking, OSA patients at low latitudes near the equator should have higher levels of vitamin D, but people near the equator tend to have darker skin, and skin melanin reduces the ability of the skin to synthesize vitamin D. Neighbors et al. [
12] evaluated the latitude of the study and the vitamin D level in the study population and found no significant correlation between the latitude and serum 25(OH)D level. In this meta-analysis, the serum 25(OH)D level was lower in OSA patients than in the controls, regardless of whether the patients were from low- or mid-latitude areas. The degree of reduction in the serum 25(OH)D level was similar between low- and mid-latitude areas. Therefore, geographical location exerted little influence on the overall results of this study.
Some studies have reported that sleep fragmentation in patients with OSA due to nocturnal hypoxia can lead to daytime drowsiness, fatigue and other symptoms. Sleep fragmentation can lead to a decrease in outdoor activity and a decrease in vitamin D synthesis [
47]. In addition, due to the repeated aggravation of upper airway obstruction and hypopnea in patients with OSA, autonomic nerve function is mainly increased in terms of sympathetic nerve activity at night, and there is abnormal tension of the vagus nerve [
48]. Excitation of the sympathetic nervous system could partly inhibit vagus nerve activity, while abnormal vagus nerve activity affects gastrointestinal motility and the secretion of gastrointestinal hormones, in turn affecting the absorption and metabolism of vitamin D. At the same time, OSA patients suffer from sleep disturbances, intermittent hypoxia, upper airway obstruction and increased abdominal pressure for a long time [
49]. All of these disorders could result in gastroesophageal reflux and gastric ischemia, which also affect the absorption of vitamin D [
25]. Therefore, the level of serum vitamin D in patients with severe OSA is significantly lower than that in healthy controls.
Among the included studies, the study by Liu et al. [
30] demonstrated that tonsillar hypertrophy was related to low serum vitamin D and that tonsillar hypertrophy is one of the main causes of OSA in children. The mechanism may be that vitamin D could be related to immune regulation [
50‐
52]. A normal serum 25(OH)D level can enhance innate immunity, and a low level of vitamin D can significantly decrease the total number of peripheral T lymphocytes and the percentage of T cells, causing reduced cellular immunity and humoral immune function. Moreover, a low vitamin D level gives rise to dysfunctional upper airway immune regulation. Lower levels of vitamin D promote the production of inflammatory transmitters, such as tumor necrosis factor-α and interleukin-1, which increase the risk of respiratory tract infection [
49,
53,
54]. This leads to tonsillar hypertrophy, chronic rhinitis and nonspecific myopathy, which may aggravate OSA in children.
Generally, heterogeneity among studies in meta-analyses is related to the quality of the included research, population characteristics, experimental methods and other factors. High heterogeneity was found among the studies on the relation between serum vitamin D and OSA in our study. To explore the underlying source of heterogeneity, subgroup and meta-regression analyses were performed. Meta-regression showed no heterogeneity with respect to ethnicity, disease severity, PSG type, study quality, BMI or latitude. Regardless of the meta-regression results, we still conducted subgroup analyses by ethnicity, disease severity, PSG type, study quality, BMI and latitude. Unfortunately, no evident sources of heterogeneity were found. However, these factors could also increase the heterogeneity and reduce the reliability of this meta-analysis. In addition, the "leave-one-out" sensitivity study revealed no individual studies contributing to the high heterogeneity. Thus, we speculated that there are several other factors leading to the heterogeneity, including differences in blood sampling and storage methods, measurement methods and experimental conditions, light duration and climate in different regions, dietary intake and other underlying confounding factors.
This meta-analysis has several strengths in the examining the vitamin D serum level in relation to OSA. First, the overall results suggest that the serum 25(OH)D level may be a clinically useful biological indicator, which may help clinicians objectively evaluate the severity of OSA and provide an improved understanding of the potential pathophysiology involved in OSA. Kerley et al. [
55] pointed out that vitamin D3 supplementation could improve several physiological, biochemical and subjective features of OSA as well as decrease metabolic markers compared to a placebo. The effect size of the serum 25(OH)D level could suggest whether vitamin D supplementation is suitable in OSA patients. Nevertheless, the size of the study by Kerley et al. [
55] was very small, and the clinical significance of the serum 25(OH)D level needs further evaluation in larger trials. Second, this is the largest meta-analysis of the relevant literature, and subgroup analyses were carried out to provide more robust results. We also included the most recent published studies involving Chinese populations in this meta-analysis. Third, all of the included articles were of medium or high quality, making the assessment here more feasible. Fourth, no obvious publication bias was detected, suggesting that the combined results may be reliable.
However, there are still a number of underlying limitations to the study. First, the inconsistencies in the serum sample detection methods and the dietary diversity of the population are factors of heterogeneity, which could result in statistical error. Second, since this study lacked effective longitudinal cohort studies, we could not infer causality of the association between OSA and serum vitamin D. Third, due to data limitations, dose–response relationships between serum vitamin D and OSA risk were not obtained.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.