Interaction between slow wave sleep and elevated office blood pressure in non-hypertensive obstructive sleep apnea patients: a cross-sectional study

Abstract Purpose: Reduced slow wave sleep (SWS) has been linked to hypertension in some studies. The aim of the study is to investigate the association between SWS and office blood pressure (BP) in non-hypertensive obstructive sleep apnea (OSA). Methods: This is a retrospective study of 3350 patients who underwent polysomnography (PSG) in our hospital. Based on quartiles of percent SWS, participants were classified into four groups. BP was measured manually on the randomly chosen arm in a seated position with sphygmomanometer after PSG in the morning, and the average of the second and third measurements was used for this analysis. Elevated office BP was defined as a systolic BP≥140 mmHg or diastolic BP≥90 mmHg. Results: There were 1365 patients with OSA and 597 primary snorers included in our study. In OSA group, OSA patients with SWS <13.5% had a significant elevated risk with elevated office BP (OR,1.49[95%CI 1.05-2.10], P=0.025), compared to the highest quartile (percent SWS >39.2%). However, no significant relationship between decreased SWS and elevated office BP was found in primary snorers group. Conclusion: In non-hypertensive OSA patients, decreased SWS is associated with elevated office BP. Plain language summary This is the first study to investigate the association between decreased SWS and incident elevated office BP in non-hypertensive OSA patients. Our results found that in non-hypertensive OSA patients, decreased SWS is associated with elevated office BP. The relationship between decreased SWS and elevated office BP in OSA patients was evident especially in men and in those <60 years old.


Introduction
The interaction between sleep and hypertension has become apparent over past years.Sleep disorders have been confirmed to elevate blood pressure (BP) and develop hypertension [1][2][3], which is the most common prevalent risk factor for cardiovascular diseases [4].Sleep comprises of two states, including rapid eye movement sleep and non-rapid eye movement sleep [5].Slow wave sleep (SWS), one part of non-rapid eye movement sleep, increases vagal tone and reduces sympathetic tone, which consequently decreases heart rate and BP [6,7].Experimental studies showed that SWS deprivation significantly attenuated nocturnal blood pressure decline [8].Further prospective population-based studies supported the same conclusion [9,10].Fung et al. [9] found incident hypertension was associated with decreased SWS percentage after multiple adjustments in a cohort of elderly men.Consistently, another study observed lower levels of percentage SWS increased odds of incident hypertension in both men and women independent of potential confounders [10].
Obstructive sleep apnea (OSA) is a common sleep disorder characterized by recurrent episodes of apnea during sleep that leads to intermittent hypoxemia and arousals [11].There are accumulating evidences that OSA increased incidence of hypertension compared with individuals without OSA [2,[12][13][14], and the different measurements of BP had the same results [2,12,15].SWS is significantly reduced because of frequent respiratory events in OSA patients [16,17].Recently, in a cross-sectional study, Ren et al. [18] found decreased SWS was associated with higher risk for hypertension in OSA patients than primary snoring, especially in men and younger patients.Moreover, Zhang et al. [19] demonstrated the incidence of hypertension was increased in patients with lower SWS percentage and OSA.However, the subjects of two studies included patients with hypertension.Almost half of the patients were diagnosed with hypertension in the two studies, which may produce confusion on the relationship of SWS and BP.Up to now, the association between decreased SWS and incident elevated office BP in non-hypertensive OSA patients has not been determined.The presence of obstructive sleep apnea and chronic diseases can worsen the prognosis of sleep problems in non hypertensive people.Therefore, we sought to determine whether low proportion of SWS is associated with incident elevated office BP in a large cohort of non-hypertensive OSA patients.

Study design and subjects
This cross-sectional study consisted of 3350 patients from Sleep Medical Center, XuanWu hospital, who underwent polysomnography (PSG) from the beginning of 2002 to October of 2020.This study is a retrospective study and all participants' details are required to be anonymized.The study protocol was approved by the ethics committee of XuanWu hospital of Capital Medical University (protocol No. Clinical research 2021-185).
All participants were adults (age > 18 years) in our study.They first went to the respiratory outpatient department for clinical symptoms of snoring and then were evaluated at the Sleep Center for suspected OSA.They were interviewed using a comprehensive questionnaire to collect information, including age, sex, body weight, neck circumference, waist circumference, hip circumference, tobacco use, alcohol drinking and medical history.Diabetes mellitus (DM) [20] and coronary heart disease (CHD) [21] was determined by specialists according to clinical manifestations, auxiliary examinations and met standard guidelines for the diagnosis.
For the present study, 1260 (37.6%) participants had a history of hypertension or were taking anti-hypertensive medications were excluded.We also excluded subjects who were diagnosed or treated OSA before, were taking sleep-disrupting medical condition, were taking benzodiazepines, z-drugs or other medications that can affect NREM sleep, were diagnosed other comorbid sleep disorders [22], slept fewer than 3 h during PSG or missed BP data (see Figure 1 for details).Patients with an apnea hypopnea index (AHI) ≥5 events/hour were defined in OSA group [23], whereas subjects with an AHI <5 events/hour were regarded as primary snoring group.

Blood pressure measures
In the morning after PSG, BP was measured manually in triplicate with a 5-min interval before detachment of the polysomnographic sensors and electrodes, and the average of the second and third measurements was used for this analysis.All BP measurements were performed once on the randomly chosen arm in a seated position with sphygmomanometer (Yuwell, China) [24].Elevated office BP was defined as a systolic BP ≥ 140 mmHg or diastolic BP ≥ 90 mmHg.

Polysomnography
All subjects completed a full-night PSG monitoring in the sleep medical center of XuanWu Hospital.Under the monitoring, sleep data were automatically recorded on the computer.After the recording was completed, sleep parameters were scored following the American Academy of Sleep Medicine criteria [25].Apnea was defined as a reduction of airflow to less than 10% of the baseline for at least 10 s, while hypopnea was defined as a ≥ 50% reduction of airflow for at least 10 s, accompanied by 3% or greater decrease in arterial oxygen saturation (SaO 2 ).The total number of apnoas and hypopneas per hour was AHI during sleep.Sleep efficiency was defined as the ratio of sleep duration to total time in bed.SWS sleep was defined as the percentage of total sleep time and categorized into quartiles.The highest SWS percentage group was chosen as the referent in OSA and primary snoring group.

Statistical analysis
Differences in sample characteristics based on different SWS quartiles were assessed using ANOVA or Mann-Whitney U tests for normally distributed and skewed continuous variables in OSA and primary snoring group respectively.Chi-square tests were applied for categorical variables.
To account for potential nonlinear associations, we first examined the associations of SWS and elevated office BP in OSA and primary snoring group by logistic regression models and used the highest quartile of percentage SWS in both groups as a reference.Next, we assessed the association of SWS and OSA with elevated office BP by using primary snoring as a reference group.Covariates included age, sex, BMI, tobacco use, alcohol drinking, neck circumference, waist circumference, hip circumference, sleep efficiency, wake after sleep onset time, CHD, diabetes, nocturnal lowest oxygen desaturation (Lowest SpO 2 ) and AHI.
Because we found that quartile 3 of SWS had the lowest incidence of elevated office BP (OR 0.85) compared to highest SWS percentage group (quartile 4) in OSA patients, thus only the first three groups had potential linear associations.Therefore, we used linear regression models to explore the association between BP values and SWS percentages in only the first three groups (excluded quartile 4).Besides, we further examined the associations in men and women and in different age groups with OSA.Age, sex, BMI, tobacco use, alcohol drinking, neck circumference, waist circumference, hip circumference, sleep efficiency, wake after sleep onset time, CHD, diabetes, lowest SpO 2 and AHI were included as covariates.
Data were analyzed using SPSS 26.0, and comparisons with p values <.05 were considered statistically significant.

Basic characteristics and univariate analyses
This study finally included 1962 subjects for the final analysis, of which 1365 had OSA and 597 were primary snorers.Demographic, clinical and sleep characteristics of OSA and primary snorers stratified according to percentage SWS were shown in Tables 1 and 2. In univariate analysis, OSA patients with lower percentage of SWS were significantly older, more likely to be male, higher ratio of smoking and had higher BP.Those in the lowest quartile also had a higher AHI and total arousal events and lower sleep efficiency.In primary snoring group, subjects with lower percentage of SWS were older and had higher BP.

SWS and elevated office blood pressure
In OSA group, patients with the lowest quartile of SWS had the highest proportion of elevated office BP.Compared with the highest quartile of percentage SWS (>39.2%, reference), patients with quartile 1(<13.5%)and quartile 2(13.5%-24.1%)were more likely to have a higher proportion of elevated office BP (Table 3 for details), whereas those in quartile 3(24.1%-39.2%)had a lower proportion of elevated office BP (OR,0.85[95%CI0.61-1.18]).After maximally adjusted analysis in model 3, only patients with OSA with <13.5% SWS had a significantly elevated risk elevated risk with elevated office BP(OR,1.49[95%CI1.05-2.10],p = .025).There was no significant relationship between SWS and elevated office BP in the primary snoring group (Table 4 for details).

Linear regression model for SWS and office blood pressure
We also applied linear regression models to examine the association between percentage SWS and office BP.There were some evidences that sleep duration and several cardiovascular outcomes had the 'j-shape' relationship [26][27][28], and in our study patients with OSA in quartile 3(24.1%-39.2%)had lower proportion of elevated office BP compared to quartile 4(>39.2%,reference), therefore linear regression was used only examine the relationship between percentage SWS and office BP in quartile 1, quartile 2 and quartile 3 SWS strata.SWS was only significantly associated with diastolic BP in OSA patients (p = .001,Table S1 in Supplemental Material).In subgroup, the percentage SWS was associated with higher BP only in men and in those <60 years old.There was no significant relationship between percentage SWS and systolic BP in OSA patients (p = .738,Table S2 in Supplemental Material).Moreover, no significant effect on systolic and diastolic BP was found in primary snoring group (p = .712,p = .064).

Discussion
This study showed that decreased SWS sleep (<13.5%) was associated with elevated office blood pressure in non-hypertensive OSA subjects, especially in diastolic BP.In a subgroup analysis, this relation was evident especially in men and in those <60 years old.Together, these results suggest that decreased SWS is likely to be an independent risk factor for increased office BP in OSA patients, possibly etiologically associated with hypertension.
Our results are consistent with the prior studies about the relationship between SWS and incident hypertension in OSA patients.An observational study reported that the percentage of SWS was significantly lower in patients with OSA with elevated BP [29] compared to the control group, and another study found [30] that percentage of SWS predicted approximately 10% of the nocturnal change in diastolic BP.However, the samples of these studies were too small.Ren et al. [18] conducted a large cross-sectional study to explore the relationship between SWS and incident hypertension.The results showed SWS was associated with elevated BP in OSA patients, and there was a significant interaction between SWS and OSA in hypertension when SWS was <4.9%.Moreover, the association was especially evident in men and in those <60 years old, which was same with our result.Prior study has already confirmed that men and younger OSA patients had higher risk for cardiovascular diseases [31].The difference between our study and Ren et al. is that the percentage of SWS on the blood pressure was different.In our study, OSA with <13.5% SWS sleep was associated with elevated office blood pressure, while SWS was <4.9% in Ren's study.The differences in the distributions of SWS across quartiles in two studies may be the reason.In our study, the percentage of SWS was much higher, because we exclude patients with hypertension which is confirmed to have lower SWS [9,10].
In our study, the protective effect of SWS sleep in OSA patients on the elevated office BP was highest in quartile 3(24.1%-39.2%)not in quartile 4(>39.2%,reference).Compared to quartile 4(>39.2%,reference), OSA participants in quartile 3 slightly decreased risk of elevated office BP, although not significant (OR, 0.85[95%CI 0.61-1.18]).This is the key difference between our study and Ren et al. [18].Prior studies have already found that both short and long sleep duration were associated with higher risk of mortality and cardiovascular events [32,33].It is extremely interesting the 'j-shape' relationship between sleep duration and cardiovascular outcomes.Similar to sleep duration, Javaheri et al. showed a short and high percentage of SWS sleep (>25.2%) were both associated with increased risk of hypertension [10], which was consistent with the result of our study.Larger clinical studies included patients with a high percentage of SWS sleep may be needed to evaluate the influence on increasing risk of hypertension.
In OSA patients, SWS is significantly reduced because of frequent respiratory events [16,17].The primary mechanism underlying the relationship between reduced SWS sleep and increased BP in OSA patients is related to alterations in autonomic nervous system activity during SWS.Compared to other stages, SWS is characterized by increasing in vagal tone and decreasing in sympathetic activity, leading to decrease heart rate (HR) and BP [6].An experimental study of selective SWS sleep suppression in healthy human subjects revealed attenuation in BP dipping during the first half of the night, when N3 sleep predominates [8].Tasali et al. [34] developed a model of SWS suppression over three nights in young, healthy individuals, and there was a reduction in vagal tone and an elevation in sympathetic tone by using spectral analysis of heart rate variability.It is possible that SWS reduction attenuates nocturnal sleep dipping though increasing sympathetic activity at night, and the cumulative effect of excessive sympathetic activation contributes to sustained daytime hypertension finally [35].The association between reduced SWS sleep and elevated office BP in OSA patients raises the important question of whether it is possible to reduce the risk of hypertension through enhancement of SWS sleep.Up to now, there were some evidences that SWS enhancement could improve cognitive function [36], memory deficits [37], alertness, attention [38] and executive function [39].In OSA patients, it seems to have a close relationship between SWS enhancement and the BP-lowering effects of continuous positive airway pressure therapy.Continuous positive airway pressure improved sleep quality with fewer arousal, decreased disturbed respiratory events and increased SWS sleep [40][41][42], which together contributed to decreasing nocturnal BP in OSA patients [43,44].In the future, new methods increasing SWS are needed to decrease nocturnal BP in patients with OSA.This is the first study to report the interaction between SWS and elevated office BP in non-hypertensive OSA patients.Moreover, use of a large sample increases generalizability in our study.However, there are still some limitations in our study.First, BP was measured only once per time point, and we cannot exclude the influence of white coat effects.Besides, we lack the 24-h ambulatory BP monitoring to describe BP changes accurately.Second, some patients included in our study were actually naïve-hypertensive in the non-hypertensive group according to the standard diagnostic criteria, which may produce the main bias of the result.Third, we also only measured SWS sleep using one night of PSG.It is possible that some individuals who had poorer sleep quality during PSG because they were sensitive to environmental changes.However, SWS sleep has been confirmed to have high night-to-night reproducibility [45].Fourth, in our study, we used the alternative definition of hypopnea.Compared to use of the recommended hypopnea definition, the alternative definition added a modest increase to the percentage diagnosed with OSA [46].This may increase the percentage of OSA diagnosis and severity.However, some studies also found the AHI based on ≥3% desaturation criterion had the some effect on predicting of adverse outcomes as an AHI based on ≥4% oxygen desaturation criterion [25,47].Finally, our study is retrospective, which may lack some details affected BP.Further prospective study is needed to evaluate the interaction between SWS and BP in OSA patients.

Conclusions
In summary, after controlling potential confounders, lower SWS sleep (<13.5%) was associated with elevated office BP in non-hypertensive OSA subjects, especially in diastolic BP.Moreover, this relationship was evident especially in men and in those <60 years old.

Figure 1 .
Figure 1.Flow diagram representing process from initial participant enrollment to final analysis.

Table 1 .
demographic, clinical and sleep characteristics of osA patients stratified by sWs categories.

Table 2 .
demographic, clinical and sleep characteristics of primary snoring stratified by sWs categories.

Table 3 .
Adjusted ors and 95% CIs for the association between sWs and elevated office BP in osA.
Notes: Model 1 was adjusted for age, sex, BMI, neck circumference, waist circumference, hip circumference, tobacco use, alcohol drinking; Model 2 was adjusted for Model 1 and CHd, diabetes, sleep efficiency and wake after sleep onset; Model 3 was adjusted for Model 2 and lowest oxygen saturation during sleep and AHI.

Table 4 .
Adjusted ors and 95% CIs for the association between sWs and elevated office BP in primary snoring.Model 1 was adjusted for age, sex, BMI, neck circumference, waist circumference, hip circumference, tobacco use, alcohol drinking; Model 2 was adjusted for Model 1 and CHd, diabetes, sleep efficiency and wake after sleep onset; Model 3 was adjusted for Model 2 and lowest oxygen saturation during sleep and AHI.

Table 5 .
Adjusted ors and 95% CIs for the joint effect of osA and sWs on elevated office BP.Model 1 was adjusted for age, sex, BMI, neck circumference, waist circumference, hip circumference, tobacco use, alcohol drinking; Model 2 was adjusted for Model 1 and CHd, diabetes, sleep efficiency and wake after sleep onset; Model 3 was adjusted for Model 2 and lowest oxygen saturation during sleep and AHI.