Abstract
There is increasing evidence that intermittent hypoxia resulting from obstructive sleep apnoea (OSA) is independently associated with dyslipidaemia. Currently, no data exist on potential links between OSA-related dyslipidaemia and susceptibility genes for dyslipidaemia in such patients. Our aim was to study the effects of the apolipoprotein E (APOE) genotype and sleep apnoea severity on atherogenic dyslipidaemia in patients with OSA.
519 clinically stable subjects prospectively recruited at a tertiary referral teaching hospital underwent full polysomnography. APOE gene polymorphisms were assessed using real-time PCR.
In all APOE genotype groups, serum triglycerides increased while high-density lipoprotein (HDL) cholesterol was reduced with increasing severity of OSA in each APOE genotype group, whereas the deleterious effects of OSA on serum apolipoprotein (Apo)B levels were observed in ϵ2 carriers and the ϵ3/ϵ3 genotype only. Nevertheless, the ϵ4 allele carriers had ApoB levels within the risk range, irrespective of nocturnal hypoxia. In addition, among patients with the high-risk ϵ4 genotype, those with the most severe nocturnal hypoxia had significantly higher triglyceride and lower HDL cholesterol levels compared with nonhypoxic ϵ4 subjects. APOE genotype and the oxygen desaturation index were both independent predictors of serum triglyceride levels (p=0.009 and p<0.001, respectively; R2=0.148) and ApoB levels (p=0.001 and p=0.003, respectively; R2=0.104).
Our findings suggest that OSA has adverse effects on several lipid parameters over and above the effects carried by APOE genotype. Further st1udies are needed to analyse the effects of high-risk genotypes on metabolic and cardiovascular outcomes in patients with OSA.
Abstract
OSA has adverse effects on several lipid parameters in addition to the effects carried by the APOE genotype http://ow.ly/sKOUW
Introduction
Atherosclerosis is one of the main causes of morbidity and mortality in patients with obstructive sleep apnoea (OSA) [1–3]. Atherogenic dyslipidaemia, characterised by an increase in plasma triglycerides, reduction in high-density lipoprotein (HDL) cholesterol and a predominance of small, dense, low-density lipoprotein (LDL) particles, is common in patients with OSA [4, 5]. Evidence from a murine model of OSA has demonstrated that chronic intermittent hypoxia induces upregulation of genes involved in lipid biosynthesis in the liver [6] and leads to decreases in plasma triglyceride clearance [7]. Therefore, the increased cardiovascular risk carried by OSA has been attributed, at least in part, to lipid abnormalities induced by intermittent hypoxia.
Genetic and environmental factors play an important role in the aetiology of atherosclerosis [8]. Although significant evidence links OSA to proatherogenic factors, such as the metabolic syndrome [9–12], systemic inflammation [13–15] and oxidative stress [16, 17], little is known about the role of genetic factors for increasing atherosclerotic risk in OSA. In particular, no data are available on potential links between OSA-related dyslipidaemia and susceptibility genes for dyslipidaemia in such patients.
Apolipoprotein E (APOE) is probably the most studied susceptibility gene for atherosclerosis and dyslipidaemia [18, 19]. There are three common alleles of the APOE gene (ϵ2, ϵ3 and ϵ4). Individuals carrying the ϵ4 allele have higher total and LDL cholesterol levels than individuals with the commonest ϵ3ϵ3 genotype and those carrying the ϵ2 allele have lower levels of total and LDL cholesterol [19–21]. In addition, associations between APOE genotype and triglycerides [18] and circulating apolipoprotein (Apo)E and ApoB concentrations have been reported in the general population [18, 22, 23]. Taken together, ϵ4 allele carriers represent the risky APOE genotype [19–23]. Nevertheless, the role of APOE gene polymorphisms in lipid metabolism disturbances in OSA, a condition known to be associated with increased cardiovascular risk, is unclear. APOE genotype is related to circulating ApoE levels in humans, with the lowest ApoE levels observed in ϵ4 allele carriers [18]. This finding, coupled with observations of increased serum lipids and profound atherosclerotic plaque growth in ApoE-deficient mice exposed to intermittent hypoxia [24], suggests that APOE genotype and hypoxia may both play a distinct role in OSA-related increased cardiovascular risk. Therefore, the purpose of our study was to investigate the effects of APOE genotype on atherogenic dyslipidaemia in patients with OSA.
Subjects and methods
Subjects
Clinically stable subjects without OSA (fewer than five obstructive apnoeas or hypopnoeas per hour of sleep), and patients with OSA (five or more obstructive apnoeas or hypopnoeas per hour of sleep and excessive daytime sleepiness) referred to the sleep unit at a tertiary referral teaching hospital (Dept of Respiratory Medicine, Faculty of Medicine, P.J. Šafárik University and L. Pasteur University Hospital, Košice, Slovakia) for a diagnostic sleep study were prospectively recruited between 2006 and 2012. Exclusion criteria were as follows. 1) Chronic respiratory diseases other than OSA, such as bronchial asthma, chronic obstructive pulmonary disease, restrictive pulmonary disorders or hypoventilation syndrome; 2) known hereditary metabolic disorders; 3) type 1 diabetes; 4) hypothyroidism; or 5) regular use of sedatives, antidepressant or antipsychotic medication or alcohol. The study was approved by the L. Pasteur University Hospital ethics committee, and all subjects provided written informed consent.
Sleep assessment
All participants underwent full attended diagnostic overnight polysomnography (Alice 4; Respironics Inc., Murrysville, PA, USA), comprising continuous recording of electroencephalography (EEG), electro-oculography, electromyography, ECG, thoracic and abdominal impedance belts, thermistor for nasal and oral airflow, pulse oximetry and microphone for snoring. All records were scored manually following the American Academy for Sleep Medicine (AASM) 2007 guidelines [25]. Apnoea was identified as a drop in airflow of >90% from the baseline excursion for ≥10 s; hypopnoea was defined as a reduction in airflow of ≥50% of baseline for ≥10 s accompanied either by a decrease in haemoglobin saturation for ≥3%, an EEG-recorded arousal, or both. The apnoea/hypopnoea index (AHI) was defined as the number of apnoea and hypopnoea events per hour of sleep. Oxygen desaturation index (ODI) was defined as the number of oxygen desaturations of haemoglobin of ≥3% per hour of sleep. In addition, the length of time with an arterial oxygen saturation measured by pulse oximetry (SpO2) <90% was used to assess the degree of nocturnal hypoxia. The classification of OSA severity was based on the AASM guidelines as follows [25]. Mild: AHI ≥5 and <15 episodes·h−1; moderate: AHI ≥15 and <30 episodes·h−1; and severe: AHI ≥30 episodes·h−1.
Biochemical measurements
A peripheral venous blood sample was drawn into EDTA tubes between 06:00 h and 07:00 h, after polysomnography and after 12–14 h of overnight fasting, and stored at -20˚C until analysis was undertaken. Fasting cholesterol, triglycerides, HDL cholesterol, fibrinogen, ApoA-1 and ApoB were measured by routine enzymatic methods. LDL cholesterol was calculated using the Friedewald equation.
Genetic analyses
DNA was extracted from peripheral blood leukocytes and amplified using real-time PCR using an air thermocycler (LightScanner32; Biofire Diagnostics Inc., Salt Lake City, UT, USA). To determine APOE polymorphisms, the Light-Mix Kit ApoE C112R R158C (TIB Molbiol, Berlin, Germany) was used. APOE genotypes were pooled in agreement with the method described in the recent meta-analysis [19] into three groups: ϵ3 homozygous group (ϵ3ϵ3 reference group, n=328), ϵ2 allele carriers (consisting of ϵ2ϵ2, ϵ2ϵ3 and ϵ2ϵ4 genotypes, n=62) and ϵ4 allele carriers (consisting of ϵ3ϵ4 and ϵ4ϵ4 genotypes, n=129).
Statistical analysis
Power calculations were performed based on published data on the effects of APOE genotype on serum triglyceride levels in the general population [16] and on the expected variability of triglycerides in OSA as observed by our group [12]. For a standard deviation of 0.72, a power calculation indicated that we would need a total cohort of ≥247 patients to detect a difference of 0.34 mmol·L−1 with a power of 80% at a 0.05 significance level.
The data are presented as mean±sd for all variables that were normally distributed, and as median (interquartile range) for variables that were not normally distributed. Differences between the groups were analysed using ANOVA for normally distributed variables and ANOVA on ranks for nonparametric variables. Discrete data were analysed using the Chi-squared test. Because the distributions of HDL cholesterol and triglycerides were skewed we log-transformed the variables for use in regression models. Least-squares linear regression analysis was used to assess the unadjusted relationships between parameters of OSA severity and serum lipid levels. For those lipid parameters that concurrently correlated with OSA severity and APOE genotype in univariate analyses we further assessed lipid levels in subgroups of subjects divided according to quartiles (Q) of ODI or SpO2 <90% within each genotype group. In multivariate analysis, multiple linear regression models were used, with lipid levels as dependent variables and age, sex, ODI or AHI, APOE genotype, body mass index (BMI), smoking, diabetes and statin use as independent variables. Analyses were conducted using SPSS for Windows software (version 14.0; IBM, Chicago, IL, USA). p<0.05 was considered statistically significant.
Results
Characteristics of the subjects
519 subjects participated in the study; 128 had no OSA, 126 suffered from mild OSA, 66 from moderate OSA and 199 from severe OSA. Basic demographic characteristics and polysomnographic findings in the study group are summarised in table 1. Male sex, higher age, higher BMI, arterial hypertension, type 2 diabetes and statin use were all associated with greater severity of OSA. Importantly, no differences in APOE genotype distribution were observed between patients grouped by OSA severity.
The effects of OSA on serum lipid levels
Serum HDL cholesterol levels were decreased, while triglyceride and ApoB levels were increased in those patients with mild, moderate and severe OSA compared with subjects without OSA (p=0.020, p<0.001 and p=0.030, respectively) (table 2). There was no observable effect of OSA severity on total or LDL cholesterol or ApoA-1. Importantly, adjustments for BMI or full adjustments (i.e. for sex, age, BMI, smoking, diabetes and statin use) had no effect on the observed differences in serum lipids between patients grouped by OSA severity.
HDL cholesterol levels were inversely correlated with AHI, ODI and arousal index (p<0.001 for all). In contrast, AHI, ODI and arousal index were directly correlated with triglyceride (p<0.001 for all) and ApoB levels (p<0.001, p=0.001 and p=0.002, respectively) (table 3).
The effects of APOE genotype on serum lipid levels
The effects of APOE genotype on serum lipids are displayed in table 4. There were significant differences between the ϵ2 allele carriers and the ϵ3ϵ3 homozygous group, as well as the ϵ4 allele carriers with respect to LDL cholesterol, triglycerides, ApoA-1 and ApoB levels, and these differences remained significant after adjustments for BMI or after adjustments for all covariates. Post hoc analyses revealed that compared with the ϵ3ϵ3 (reference) group, ϵ4 allele carriers had significantly higher triglyceride and ApoB levels (p<0.05 for both). APOE genotype had no effect on total or HDL cholesterol levels. The exclusion of subjects on hypolipidaemic therapy (n=99) or ϵ2ϵ4 heterozygous subjects (n=6) from the analyses made no difference to any of these findings (data not shown). Importantly, sex distribution, age, BMI, AHI and ODI were similar in the ϵ2 carriers, ϵ3ϵ3 homozygous group and as the ϵ4 allele carriers (p=0.944, p=0.733, p=0.352, p=0.299 and p=0.457, respectively).
Relationships between sleep apnoea severity and serum lipid levels across APOE genotype groups
Across all APOE genotype groups, there were significant increases in serum triglyceride levels with increased severity of nocturnal hypoxia, in association with reductions in serum HDL cholesterol (table 5). In contrast, serum ApoB levels increased from Q1 to Q4 ODI in ϵ2 carriers and the ϵ3ϵ3 homozygous group, but not in ϵ4 carriers. Post hoc analyses revealed that among the risky ϵ4 allele carriers those patients with Q4 ODI had significantly higher triglyceride levels and significantly lower HDL cholesterol levels compared with subjects with Q1 ODI. Similarly to the entire cohort, severity of OSA was not related to total or LDL cholesterol or to ApoA-1 levels in patients grouped by APOE genotype. To strengthen these analyses further, we performed analogical tests after dividing patients grouped by quartiles of AHI or time of SpO2 <90% within the respective genotype groups. Importantly, discrimination of sleep apnoea severity by AHI made no material difference to the observed effects of OSA on serum lipids within the respective genotype groups. Relationships between serum HDL cholesterol, triglycerides or ApoB and SpO2 <90% in univariate analyses were weaker compared with those with AHI or ODI (table 3); consequently, dividing the cohort by SpO2 <90% weakened the observed differences between OSA severity and serum lipids (data not shown).
Both APOE genotype and ODI, in addition to sex and BMI, independently predicted serum triglyceride levels in multivariate stepwise regression analysis (p=0.009, p<0.001, p=0.001 and p=0.010, respectively; R2=0.148). In addition, APOE genotype and ODI were both independent predictors of serum ApoB levels (p=0.001 and p=0.003, respectively; R2=0.104). To strengthen these analyses we further analysed independent effects of APOE genotype and OSA severity on lipid parameters in analogical models with AHI as an independent variable. Indeed, both APOE genotype and AHI independently predicted serum triglycerides (p=0.011 and p<0.001, respectively; R2=0.120). ApoB was independently predicted by AHI, while APOE genotype approached but did not reach statistical significance in this model (p=0.001 and p=0.061, respectively; R2=0.031). We also analysed an independent effect of an interaction term (APOE genotype*ODI) on serum triglyceride and ApoB levels. Nevertheless, this interaction term did not reach statistical significance in either of these models. Serum HDL cholesterol was independently predicted by ODI (p=0.008), but not by APOE genotype. No significant predictor of LDL cholesterol was identified in multivariate models.
Discussion
This observational study has demonstrated that serum triglycerides increased while HDL cholesterol was reduced with increasing severity of OSA in each APOE genotype group, whereas the deleterious effects of OSA on serum ApoB levels were observed in ϵ2 allele carriers and the ϵ3ϵ3 genotype only. Nevertheless, the ϵ4 allele carriers had ApoB levels within the high risk range irrespective of nocturnal hypoxia. In addition, among patients with the high-risk ϵ4 genotype those with the most severe nocturnal hypoxia had significantly higher triglyceride and lower HDL cholesterol levels compared with nonhypoxic ϵ4 subjects. Previously, potential links have been observed between APOE genotype and the risk of OSA in some, but not all, studies [26, 27]. In addition, some observations suggested a significant effect of APOE genotype on clinical outcomes in terms of cognitive impairment [28] and memory performance in OSA [29], without examining the effects on serum lipids. To our knowledge, our data are the first to examine lipid status within the various APOE genotype groups in patients with OSA, and to demonstrate that sleep apnoea has adverse effects on several lipid parameters over and above the effects carried by the APOE genotype.
ApoE is a multifunctional protein that is a component of several plasma lipoprotein–lipid particles and, as such, plays a key role in the metabolism of cholesterol and triglycerides [30]. In the population at large, the ϵ2 and the ϵ4 alleles have been linked to high triglyceride levels [19–21, 31]. The LDL receptor recognises ApoE and ApoB-100 and plays a crucial role in cholesterol homeostasis [32], whereas the structurally related very LDL receptor recognises ApoE, but not ApoB-100, and plays an important role in triglyceride metabolism [33]. In the present study, serum triglycerides were affected by both the severity of sleep disordered breathing and APOE genotype and, importantly, these differences remained significant after adjustments for BMI and other covariates. Similarly to reports in large population-based cohorts [19], the association of APOE genotype with triglycerides was nonlinear, with the highest levels among ϵ2 and ϵ4 allele carriers and the lowest levels in the common ϵ3ϵ3 reference group. The observation of significant independent effects of both OSA and APOE gene polymorphisms on triglycerides in the entire cohort led to an intriguing question: does hypoxia relate to lipid parameters irrespective of the underlying APOE genotype? Indeed, triglyceride levels increased with a rise in OSA severity in each APOE genotype group. Of interest, among the risky ϵ4 allele carriers, the difference in median triglyceride levels between the nonhypoxic and the most severely hypoxic subjects (Q1 versus Q4 ODI) was 1.02 mmol·L−1. Such an increase is clinically significant, while at the triglyceride level of 1.2 mmol·L−1 the vast majority of LDLs are less atherogenic large buoyant particles, at a triglyceride level >1.9 mmol·L−1 most of the LDL particles are small, dense proatherogenic LDLs [34].
In the present study, total cholesterol and LDL cholesterol were not increased in patients with OSA, in agreement with several previous reports [11, 35, 36]. In contrast, HDL cholesterol levels were inversely related to indices of OSA severity, and these relationships remained significant after adjustments for all relevant covariates. Our observations are in line with previous studies in cohorts with large sample sizes that have shown that OSA is associated with a decrease in HDL cholesterol [37, 38]. Importantly, adverse effects of sleep disordered breathing on HDL cholesterol were observed in each APOE genotype group, whereas APOE genotype was not linked to HDL cholesterol per se. The difference in median HDL cholesterol between the nonhypoxic and the most severely hypoxic subjects was ≥0.2 mmol·L−1, irrespective of APOE genotype, and patients in the Q3 and Q4 ODI groups had HDL cholesterol levels <1.1 mmol·L−1 in each genotype group. Of note, such levels are associated with increased cardiovascular risk, as suggested by the recent Prospective Cardiovascular Muenster (PROCAM) study, in which HDL cholesterol <1.15 mmol·L−1 was associated with a 2.6-fold increased risk of myocardial infarction [39].
Similarly to triglycerides, serum ApoB levels were independently predicted by both OSA severity and APOE genotype. ApoB reflects the number of all atherogenic triglyceride-rich and LDL particles, including the most atherogenic fraction of small, dense LDL particles, and thus represents an additional marker of increased cardiovascular risk [40]. Serum levels of ApoB >0.9 g·L−1 have previously been associated with an increased cardiovascular risk [34, 41]. Importantly, in nonhypoxic subjects within the ϵ2 and the ϵ3 genotypes the average ApoB was below this risky level, whereas patients with the most severe hypoxia had significantly higher mean ApoB levels, thus suggesting an adverse effect of OSA on ApoB in these genotype groups. Previous studies have demonstrated that in the general population, plasma ApoB concentrations are the highest in ϵ4 allele carriers [23], and our results parallel these findings. Of note, the lack of an additional deleterious effect of sleep disordered breathing on ApoB in the ϵ4 group may potentially represent a ceiling effect of APOE genotype on this lipid parameter, as among the ϵ4 allele carriers the ApoB levels were above the risky 0.9 g·L−1 level in all OSA severity groups.
Studying a well-defined cohort of subjects who all underwent full polysomnography represents one of the main strengths of this study. In addition, the proportion of ϵ2, ϵ3 and ϵ4 allele carriers in the present cohort of patients with OSA (12%, 63% and 25%, respectively) is similar to that observed in large epidemiological cohorts in the general population [19, 42], thus reinforcing the value of our observations on the effects of APOE gene polymorphisms on lipid phenotype in OSA. There are also some limitations to this study. First, the cross-sectional nature of the study design does not prove causation for the relationships between OSA, APOE genotype and serum lipids. Secondly, individuals with suspected OSA referred to the sleep laboratory are a discrete group and results obtained may potentially not be generalisable to the general population. Thirdly, the primary incentive of our study was to analyse, in patients with OSA, differences in serum lipids between the respective APOE genotype groups, and the study was powered as such. Therefore, the lack of a significant effect of the interaction term APOE genotype*hypoxia on serum lipids should be considered inconclusive and hypothesis generating. Further investigations in large cohorts are needed to approach this question.
In conclusion, we have demonstrated that OSA had a negative impact on serum triglycerides and HDL cholesterol irrespective of APOE genotype, thus suggesting deleterious metabolic consequences of sleep disordered breathing over and above the genetic effects. Increases in ApoB levels with increasing severity of OSA were confined to ϵ2 allele carriers and the ϵ3ϵ3 genotype only. Nevertheless, the ϵ4 allele carriers had ApoB in the risk range irrespective of the presence of OSA. Further studies are needed to analyse the effects of risky genotypes on metabolic and cardiovascular outcomes in patients with OSA.
Acknowledgments
The authors wish to express their gratitude to A. Schejbalova and Z. Lazarova (Sleep Laboratory, Dept of Respiratory Medicine, Faculty of Medicine, P.J. Šafárik University, Košice, Slovakia and L. Pasteur University Hospital, Košice).
Footnotes
Support statement: This work was supported by the Slovak Research and Development Agency under contract No. APVV-0134-11, VEGA 1/0112/11 and VEGA 1/0111/12 of the Ministry of Education, Slovakia.
Conflict of interest: None declared.
- Received June 11, 2013.
- Accepted October 24, 2013.
- ©ERS 2014