Sleep Quality and Atherogenic Risk in Sleep Apnea Patients

Objective:Some studies have shown the atherogenic dyslipidemia as being sometimes independent of the obesity; however, with a close relation to the obstructive sleep apnea (OSA). The aim of this study was to investigate a relationship between parameters defining the quality of sleep and the standard lipid profile despite the obesity.Methods:The data of 211 patients with suspected OSA were analyzed prospectively. The following tests were performed: polysomnography, morphometric assessment and serum lipid levels.Results:Value of AHI showed a positive correlation with TG, TG/HDL ratio, BMI, plasma atherogenic index (API), circumference of neck, waist, and hip, and WHR. A negative correlation was found between AHI and HDL. The patients with severe OSA (AHI>30) differ significantly from the patients with moderate and mild OSA in terms of TG, and the ratio of TG /HDL and API and they are, therefore, in the group of much higher risk of cardiovascular disease. Significant predictors of the severity of sleep apnea are as follow: the ratio of TG/HDL, BMI and the neck circumference. The arousal index was the only parameter of the sleep quality significantly associated with the level of TG.Conclusion:The most important parameters of the quality of sleep that determine atherogenic risk are AHI and arousal index that are significantly correlated with TG/HDL ratio. Particularly, high risk of cardiologic problems applies to patients with AHI>30 who substantially differ in terms of lipid profile from the patients with mild or moderate OSA. Arousal index significantly distinguishes patients with AHI> 30 from the other groups of patients.


Subjects
The data of 211 patients (176 men and 35 women) with OSA diagnosed on the base of polysomnography (PSG) were analyzed. All the patients gave their informed consent to participate in the study. The Ethics Committee Review Board at the Medical University of Warsaw approved the study protocol. Patients were consecutively included in this prospective study according to the waiting list for a PSG test. In the morning, right after PSG was completed, every patient underwent lipid profile test. For this study, the inclusion criteria were age over 18 years old, no treatment for OSA (no oral medication, no CPAP, no surgical treatment). The exclusion criteria were pregnancy, cancer, diagnosed hyperlipidemia and/or treatment for hyperlipidemia, diabetes, previous tonsillectomy or septoplasty.

Polysomnography
Standardized attended 14-channel laboratory overnight PSG was performed, using an oronasal thermocouple (Grass, USA). Sleep stages were manually scored in 30 seconds sections using standard AASM 2005 criteria. The AHI, arousals, and desaturations were computed and then manually revised. PSG sleep time was obtained by summing the time spent in sections scored as any stage of sleep during the period from "lights off" (approximately 10:00 pm) to "lights on" (approximately 7:00 am). Obstructive sleep apnea is confirmed by AHI≥5. Mild OSA is determined by 5≤AHI≤15, moderate by 15≤AHI≤30, and severe by 30>AHI. Sleep quality was assessed based on proportions of the various stages, the number of EEG activations/ per hour of sleep (arousal index) and the number of awakenings to the state of wake per hour of sleep. The sleepless time (Wake After Sleep Onset -WASO) was assessed by time after the start of stage N2. The duration of the different phases of sleep: NREM (N1, N2, N3) and REM, was determined as a percentage of total sleep time.

INTRODUCTION
In patients with obstructive sleep apnea (OSA), most of which meet the criteria of the metabolic syndrome, dyslipidemia seems to be directly connected with overweight/ obesity (1)(2)(3). However, some studies demonstrated an independent association between OSA and atherogenic dyslipidemia in non-obese subjects with OSA: increased levels of triglycerides (TG), total cholesterol (TC), and LDL cholesterol (LDL), and lower levels of HDL cholesterol (HDL) (4)(5)(6). However still there is relatively small number of clinical trials dealing with the lipid profile in patients with OSA and their results are ambiguous (3,6,(7)(8)(9)(10). Two of the main polysomnographic parameters, desaturation index and Sat O 2 nadir are significantly associated with a reduced level of HDL (3). It was also found that AHI is strongly associated with the level of HDL, regardless of age, sex, body weight, BMI, diabetes or lipid-lowering drugs. (1). However, other studies did not show any relationship between AHI and the standard lipid profile (8)(9)(10). Indirect evidence for the association of OSA with dyslipidemia is the positive impact of CPAP therapy on one or more elements of the standard lipid profile (1,11,12). However, several other studies analyzing the lipid parameters (8,13,14) have not shown any effect from CPAP therapy. Perhaps therefore atherogenic dyslipidemia also depends on other factors, including the quality of sleep. In the literature, there is no data whether a pathological structure of sleep, despite respiratory events, might be associated with atherogenic lipid profile in OSA patients.
However, it is known that sleep fragmentation and, therefore, both the number of respiratory incidents and their adverse changes during the sleep reflected in the EEG, can reflect in biochemical system disorders. It is known that OSA influences in negative way the hypothalamicpituitary-adrenal axis by repeated arousals (15). It leads to permanent increase in the activity of the renin-angiotensin-aldosterone-cortisol axis (16,17). Moreover, all the published studies analyzed only the relationship between AHI or SO2nadir and the lipid profile without mentioning the atherogenic indices. It has been proven that the absolute values of the levels of TC, LDL-C, LDL-C, and TG do not always adequately reflect the risk of atherosclerosis. The atherogenic indices have a greater predictive value than do each of the lipid fractions individually (18).
This study aimed to determine whether there is any relationship independent of obesity between the quality of sleep and atherogenic lipid profile in patients with OSA.

Anthropometric analysis
Height and weight of participating subjects were obtained using the Tanita Corp. scale (BC 418 MA). Neck, waist, and hip circumferences were measured using a metallic anthropometric tape. All anthropometric measurements were rounded up and expressed in kilograms and centimeters. Waist to hip ratio (WHR) as well as body mass index (BMI) was calculated.

Statistical analysis
Statistical analysis was performed using Statistica software (StatSoft, Inc. 2010, data analysis software system, version 10). To analyzed the data the following tests were used: Student's t-test, the Mann-Whitney test, multivariate analysis of variance (ANOVA), Fisher's (LSD -least significant differences), Scheffe and Newman-Keulus tests, Spearman's rank order and the Pearson correlation coefficient tests. A P-value of <0.05 was considered statistically significant.
To determine the impact of OSA severity on the indi-vidual parameters, subjects were divided into subgroups according to AHI. Group I included subjects with AHI index 5≤AHI≤ 15, group II with index 15<AHI≤30, and group III those with index AHI>30 (Table 2). Significant differences between these subgroups were found in TG, TG/ HDL, BMI index, API SO2nadir, N1, arousal index, wake index, neck, waist, hip circumferences, and WHR. Group III was significantly different from both groups I and II in terms of TG levels (p=0.0324) and TG/HDL (p=0.0054). Furthermore, group III was significantly different from I and II in terms of BMI (p<0.0000) and API (p=0.0121). The parameter SO2nadir showed a significant difference be- To determine the influence of body weight on the examined parameters the subjects were divided into three groups relating to the BMI: Group A included subjects with BMI≤25, group B with index 25<BMI≤ 30, and group C with BMI>30 (Table 3). Significant differences in ANOVA were found between mentioned groups in terms of AHI, SO2nadir, fasting glucose, N2, N3, HDL, TG, Castellani index, TG/HDL ratio, API index, circumference of the neck, waist and hips. Post hoc analysis showed significant differences between the group C and A and B separately in terms of: AHI (for both p<0.0000), glucose (p=0.0332 and p=0.019 respectively) and HDL (p=0.001 and p=0.0068 respectively). Significant differences between groups A and B, A and C showed the Castellani ratio (p=0.0279, p=0.0011 respectively). Between groups A and B, A and C, and B and C separately, the API index (p=0.001, p<0.0000, p=0.0001 respectively), TG (p=0.0165, p<0.0000, p=0.0001 respectively), as well as TG/HDL ratio (p=0.042, p<0.0000, p=0.0002 respectively) were significantly different.
Significant differences in SO2nadir parameter were found between groups C and A (p=0.0001) and B (p=0.0001).
Relative duration of stage N2 in group I differed from group B (p=0.0044) and A from C (p=0.0023). As for relative duration of N3 the significant difference was noticed between groups A and B (p=0.0097) and between A and C (p=0.0054). Furthermore, the neck, waist and hip circumferences the differences were significant between each of the groups (p<0.0000). Correlation analysis showed many significant correlations in general, and between males and females (  Table 5.

DISCUSSION
A clinically significant correlations between AHI and the level of TG (0.19), and HDL (-0.21) were demonstrated, however this correlation was not strong. Atherogenic lipid profile is associated with a number of considerations than just the quality of sleep, therefore value of r>0.5 should not be expected. To support this statement it is worth mentioning that a similar relationship was found between AHI and the atherogenic plasma profile. Our rank correlation coefficients (Spearman) as well as linear (Pearson -not included in this paper) were comparable with the results of other authors. A significant correlation between the AHI count and the TG level (0.146) and a negative one with the level of HDL (-0.283) was found by Börgel et al. (1). Newman et al. (6) showed a negative correlation between AHI and HDL (-0.172) The group of patients with severe OSA, however, also had significantly higher BMI compared to patients with milder forms of OSA. This may suggest that the atherogenic dyslipidemia is caused mainly by overweight/obesity that is very common in patients with OSA. The significantly higher neck, waist, and hip circumferences, as well as WHR ratio in patients with severe OSA compared to the other three groups (without OSA, mild and moderate OSA) also support this hypothesis. The relationship between OSA and serum lipids is not simply linear. According to some authors, OSA besides body mass is associated in particular with atherogenic LDL-B phenotype, but only in non-obese patients (19). The results of some other studies showed that there is no relationship between the levels of the individual serum lipids and AHI. More important are the proportions of the individual components of the lipid profile (8)(9)(10). In this study, the significant parameters associated with the AHI were just TG/HDL ratio and API index, which are considered the most important for cardiovascular risk prediction.
Stepwise regression models constructed for AHI, HDL and TG showed that the values of these three parameters did not depend significantly on most of the sleep parameters (except arousal) but on the neck circumference, WHR ratio and BMI index. Our study did not confirm the initial assumptions about the relationship of atherogenic dys-lipidemia with relative shortening of the duration of SWS or the wake index, showed in particular in the linear correlation analysis and stepwise regression model where the relative duration of SWS was not a significant explanation for the AHI parameter or serum lipid levels as dependent variables. However, it is known that SWS is a recovery period of intense cardiovascular and hormonal secretion, and its quality determines the level of leptin. Both obesity and OSA are significantly associated with a resistance to leptin, which may generate (by an unspecified mechanism) a reduction of airway muscle tension (20,21). However, leptin levels did not significantly determine the value of the AHI alone, but only the association with BMI and percentage of fat in the body, which are the strongest predictors of OSA (22). Perhaps the calculated total time of slow-wave sleep, despite its considerable fragmentation induced by excitations, showed no differences large enough to reach a level of significance. This issue should be analyzed in larger number of EEG recordings.
WHR ratio, significantly higher in participants with severe OSA, puts these patients in the particularly high risk of cardiovascular disease group, because WHR is a strong predictor of these diseases (23, 24). Results of our study showed that patients with severe OSA differ significantly from patients with moderate and mild OSA in the level of TG and the ratio of TG/HDL and the API index. Thus, the limit of 30 apnea/hypopnea counts per hour of sleep is once again the rationale for the clinical classification of severe OSA. It should be noted that this group (III) had a significantly higher arousal index compared with other groups (I or II, or I and II together). Data in Table 4 show arousals incidents as not direct result of high AHI, since the coefficient of linear correlation between AHI and arousals was 0.46. It seems that the greatest impact on the quality of sleep is therefore the number of arousal incidents. Arousal index has a significant effect on the activity of the adrenergic system (25). Sleep fragmentation and repeated EEG excitation may also affect the ejection of free fatty acids (8). In this study, for analyzing all subjects as one group (but not for men and women separately) we identified a weak but statistically significant correlation between the arousal index and the level of TG (0.15). It should also be mentioned that in some subsets created during stepwise regression analysis, arousal index appeared as an equally strong explanatory variable for the TG levels as a dependent variable as BMI was. This may suggest the need for further clinical trials, in which the arousal index would be evaluated in conjunction with the secretion profile of selected hormones.
It seems that there are two distinct phenomena: incidents of arousal and incidents of apnea/hypopnea (measured with AHI and SO2nadir respectively), both causing sleep fragmentation. While respiratory incidents usually lead to arousals, the arousal incidents mainly lead to shallow sleep, interrupting a deep sleep, in particular REM and slow-wave. The correlation between the number of respiratory events with atherogenic lipid profile is well understood, but the fragmentation of sleep due to arousals is mentioned only as a possible cause of the activation of the hypothalamic-pituitary-adrenal axis. The authors did not find any data in the literature that would confirm a relationship between AHI and elevated TG and therefore increased atherogenic risk. This issue should be further investigated in a larger group of patients, and including sleep fragmentation index in each of its phases caused by arousals. The authors did not find any data in the literature analyzing patients with severe sleep apnea and proving them in particular vulnerable of developing cardiovascular diseases. In those patients, the lipid levels and atherogenic indexes are very abnormal. In our study, the arousal incidents in this group of patients were found to be significantly more frequent. Probably AHI> 30 is a certain limit, and above it, the sleep fragmentation is so serious that normal homeostatic functioning of the organism is no longer possible.
In conclusion, the most important parameters of the quality of sleep that determine atherogenic risk are AHI and arousal index that are significantly correlated with TG/HDL ratio. Particularly high risk of cardiologic problems applies to patients with AHI>30 who substantially differ in terms of lipid profile from the patients with mild or moderate OSA. Arousal index significantly distinguishes patients with AHI> 30 from the other groups of patients.